Professional Documents
Culture Documents
Chapter 1 Introduction
Developments in Hydraulic Fracturing .......................................................................1-3
Fracture Orientation: ..............................................................................................1-3
Fracturing Fluid: .....................................................................................................1-4
Proppants: ................................................................................................................1-5
Fracture Treatment: .................................................................................................1-6
Early Fracture Design ...................................................................................................1-8
9.10 Amoco Hydraulic Fracturing Course Outline .............................................................1-11
9.11 Nomenclature ..............................................................................................................1-14
9.12 References ...................................................................................................................1-17
9.13 The Continuity Equation ...............................................................................................2-1
Chapter 7 Proppants
What Causes A Proppant To Be Substandard? ............................................................ 7-3
Overview of Chap. 7 .................................................................................................... 7-3
9.40 Proppant Properties ..................................................................................................... 7-4
Sphericity and Roundness ........................................................................................... 7-4
Hardness ..................................................................................................................... 7-4
Size Distribution ......................................................................................................... 7-5
Crush Resistance ......................................................................................................... 7-9
Bulk and Grain Density ............................................................................................ 7-11
Acid Solubility .......................................................................................................... 7-11
Turbidity ................................................................................................................... 7-13
Resin-Coated Proppant ............................................................................................. 7-16
As we move towards the next century, we are challenged with applying this technology domesti-
cally in an attempt to offset large domestic trade deficits and declining production. In addition, as
our industrys focus moves internationally, methods of accelerating recovery, such as fracturing,
must be explored. Fig. 1.1 presents a world cross section of producing oil wells, their average pro-
duction and the total production of each country. This logarithmic plot shows that fracturing appli-
cations will continue to be important throughout North America, driven by the large number of
wells available and the corresponding low producing rates presently experienced by these wells.
100000 8
10000 6
1000 4
100 2
10 0
Saudi Arabia U. K. Nigeria Mexico China Canada U. S.
Country
# Oil Wells Well Rate Total Production
Excerpted DOE/FE-0139
The idea of hydraulically fracturing a formation to enhance the production of oil and gas was con-
ceived by Floyd Farris1 of Stanolind Oil and Gas Corporation (Amoco) after an extensive study of
the pressures encountered while squeezing cement, oil and water into formations. The first exper-
imental treatment intentionally performed to hydraulically fracture a well for stimulation was per-
formed by Stanolind in the Hugoton gas field in Grant County, Kansas, in 1947 as shown in
Fig. 1.2. A total of 1,000 gallons of napalm thickened gasoline was injected, followed by a gel
breaker, to stimulate a gas producing limestone formation at 2,400 ft. However, the deliverability
of the well was not changed appreciably. The hydraulic fracturing process was first introduced to
the industry in a paper written by J. B. Clark2 of Stanolind in 1948 and patented and licensed in
1949. These patents resulted in royalty income to Amoco in the 17 years following and essentially
funded the construction of the Amoco Production Research (APR) complex in Tulsa, Oklahoma
(i.e., APR is the house that fracturing built).
Halliburton Oil Well Cementing Company was given an exclusive license on the new process. The
first two commercial fracturing treatments were performed in Stephens County, Oklahoma, and
Archer County, Texas, on March 17, 1949, using lease crude oil or a blend of crude and gasoline,
and approximately 100 to 150 pounds of sand. Both wells were successful and thereafter applica-
tion of the fracturing process grew rapidly, peaking, as shown in Fig. 1.3, at an average of +3,000
wells per month by the mid-1950s and increasing the supply of oil in the United States far beyond
our early projections.3
The first one-half million pound fracturing job in the free world was performed in Stephens
County, Oklahoma, in October 1968, by Pan American Petroleum Corporation, now Amoco.
Hydraulic Fracturing Theory Manual 1-2 February 1993
History of Hydraulic Fracturing
Today, fracture treatments are performed regularly in all petroleum producing countries, including
the Soviet Union. It is estimated that at least 30% of the recoverable oil and gas reserves in the
United States can be attributed to the application of hydraulic fracturing.
4000
3000
2000
1000
Fig. 1.3 - Average Number of Fracturing Treatments per Month United States.
Significant technical advancements have been made during the four plus decades since the first
commercial treatments. After the first few jobs, the average fracture treatment consisted of about
750 gallons of fluid and 400 pounds of sand. Today, treatments average about 43,000 gallons of
fluid and 68,000 pounds of propping agent with the largest treatments exceeding one million gal-
lons of fluid and three million pounds of proppant. This reflects advancements made by the indus-
try in both theory and practice which have resulted in a better understanding of the fracturing
process. As this process evolved; cleaner and more suitable fluid systems were developed; sand
quality increased and higher concentrations were pumped; higher strength synthetic proppants
were developed for deep-well fracturing; pumping and monitoring equipment were improved and
computerized; and fracture design and evaluation techniques grew in sophistication.
Fracture Orientation:
The original, shallow fracture treatments were thought to be horizontal, even though some of the
deep wells that had been squeeze cemented showed cement in vertical fractures. The theory was
that the overburden was lifted and the fracture was inserted in a horizontal plane. Clark et al.4
reported on a method of forming a vertical fracture in 1953 by plastering the walls of the wellbore
to where it became a thick wall cylinder. Pressures were then applied to obtain vertical fractures,
otherwise it was theorized horizontal fractures were obtained. Huitt et al.5-7 extended the theories
in the late 1950s that the best fracture systems were horizontal and they could be obtained by
notching the formation. Hubbert and Willis8 with Shell Oil Company presented a paper in 1956
reporting on the work they had done in a gelatin model. This work indicated that all fractures were
vertical, creating quite a controversy. In spite of this, it was not until the mid-1960s that the indus-
try accepted the theory that practically all fractures were vertical and that only a few were horizon-
tal. Prior to this time, theories were advanced that all fractures with a treating gradient of over 0.8
or 0.9 psi per foot of depth were vertical. All those with treating gradients less than this were hor-
izontal. Work initiated by Cochran, Heck and Waters and reported on by Anderson and Stahl9
proved, without a doubt, that the majority of fractures were in fact vertical and it was a rare excep-
tion when a horizontal fracture was obtained.
100
90
AQUEOUS BASE FLUID
PERCENT OF TREATMENT
80
70
60
50
40
30
20
OIL BASE FLUID
10
1949 1953 1957 1961 1965 1969 1973 1977 1981 1985 1989
YEAR
Fracturing Fluid:
Hydraulic fracturing fluids have varied considerably over time as shown in Fig. 1.4. The first frac-
ture treatments were performed with gelled lease crude, later, gelled kerosene was used. In 1952,
refined and lease crude oils began to gain momentum, and by the latter part of 1952, a large portion
of all fracturing treatments were performed with refined and lease crude oils. These fluids were
inexpensive and safer, permitting greater volumes to be pumped at a lower cost. Their lower vis-
cosities exhibited less friction than the original viscous gel, thus injection rates could be obtained
at lower treating pressures. Higher injection rates, though, were necessary to transport the sand due
to the lower viscosity and high rates of leakoff for these fluids.
In 1953, with the advent of water as a fracturing fluid, a number of different gelling systems were
developed. Surfactants were added to minimize emulsions with the formation fluid and potassium
chloride was added to minimize the effect on clays and other water sensitive constituents of the
formation. Later, other clay stabilizing agents were developed that enhanced the potassium chlo-
ride and permitted the use of water in a greater number of formations. Other new innovations, such
as foams and addition of alcohol, have enhanced the use of water in a number of formations. Aque-
ous fluids such as acid, water and brines are now used as the base fluid in over 70% of all fracturing
treatments employing a propping agent. In the early 1970s, a major innovation in fracturing fluids
was to use crosslinking agents to enhance the viscosity of gelled water base fracturing fluids. Less
pounds of gelling agent were required to reach the desired pumping viscosity, thus reducing cost.
In many cases, however, too high a viscosity was obtained and pumping problems resulted. This
system was soon perfected by reducing the concentration of gelling agents and crosslinker, result-
ing in an economically satisfactory fracturing fluid system.
During the mid 1970s, fracture stimulations were designed for deeper formations. Gel stabilizers
were developed to maintain the properties of the fluid system at the higher temperatures at these
greater depths. The first of these temperature stabilizers was 5% methanol. Later chemical stabi-
lizers were developed that could be used alone, or with the methanol. There was a synergistic effect
obtained when the chemical and the methanol were used together as stabilizers.
Recently, a new innovation was introduced which gives even greater temperature stability. As the
gelled fluid reaches the bottom of the hole and the temperature is increasing, a secondary gelling
agent reacts giving a more uniform viscosity than previous surface crosslinked fluids. Improve-
ments in crosslinkers involve a delayed effect, thus permitting the fluid to reach the bottom of the
hole in high temperature wells prior to crosslinking. This system gives adequate viscosity for mov-
ing the propping agent through the surface equipment and into the tubing, reducing the shearing
effect caused by tubulars, and supplying a good fluid in the hydraulically created fracture to ensure
adequate proppant transport. These are only a few of the highlights of fracturing fluid develop-
ments. Many other developments have enhanced the performance of fracturing fluids.
Proppants:
To keep the artificially created hydraulic fractures open, proppants of many different kinds have
been used. The first fracturing treatment used a northern type sand for proppant; however, screened
river sand was also employed on many early treatments. In fact, on some of these treatments, con-
struction sand sieved through a window screen was employed as the propping agent. It was soon
realized, however, that a high quality sand was desirable and specifications were established on
the type of sand to be used. There have been a number of trends in the size of sand, from very large
down to small. From the very beginning a 20 to 40 U.S. standard mesh sand has been the most
popular and at the present time approximately 85% of the sand used is of this size. Numerous prop-
ping agents have been evaluated throughout the years, including plastic pellets, steel shot, Indian
glass beads, aluminum pellets, high strength glass beads, rounded nut shells, resin coated sands,
sintered bauxite and fused zirconium.
Fig. 1.5 shows that the amount of sand used per fracture treatment has steadily increased through
time. As shown, the concentration of sand (lb/fluid gal) remained low until the mid-1960s when
the use of viscous fluids, such as complexed water base gel and viscous refined oil were intro-
duced. At that time, large size propping agents were advocated to improve well deliverability.
Proppant design techniques at low sand concentration changed from the monolayer or partial
monolayer concept to pumping sand at multiple grain diameters and high concentrations. Over the
last decade, there has been another sharp increase in sand concentrations used corresponding with
improved hydraulic fracturing fluids and advanced pumping equipment.10 It is not infrequent to
1.8 90 90
1.6 80 80
1.0 50 50
Fluid/treatment
0.8 40 40
0.6 30 30
0.4 20 Sand/treatment 20
10 10
0 0
1949 1953 1957 1961 1965 1969 1973 1977 1981 1985 1989
Years
3000 30
25
HYDRAULIC HORSEPOWER
2500
2000 20
INJECTION
RATE, bbl/min
RATE
1500 15
HHP/JOB
1000 10
500 5
0 0
1949 1953 1957 1961 1965 1969 1973 1977 1981 1985 1989
YEARS
Fracture Treatment:
There are cases where as much as 15,000 hhp has been available on jobs with over 10,000 hhp
actually being utilized. Contrast this to some of the early jobs where only 10 to 15 hhp was
required. The initial jobs were performed at rates of two to three barrels per minute (bpm). Rates
Hydraulic Fracturing Theory Manual 1-6 February 1993
History of Hydraulic Fracturing
increased rapidly until the early 1960s where rates around 20 bpm became popular. Today, jobs
are performed at a low rate of about 5 bpm, to a high rate of over 100 bpm. At one time in the Hugo-
ton gas field, pumping rates of over 300 bpm were employed. Surface treating pressures sometimes
are less than 100 psi, yet others may approach 20,000 psi. Today, as treatment size, pressure and
pump rate increase, treatment costs have also increased, ranging from less than $10,000 to over
$1,000,000. The first two commercial treatments cost between $900 and $1,000.
Conventional cement and acid pumping equipment were utilized initially to execute fracturing
treatments. One to three units equipped with a jet mixer and one pressure pump delivering 75 to
125 hhp were adequate for the small volumes injected at the low rates. Amazingly, many of these
treatments gave phenomenal production increases. As the treating volumes increased, accompa-
nied with demand for greater injection rates, purpose built pumping and blending equipment was
developed to perform these specialized functions. Today, the development of fracturing equipment
continues, including intensifiers, high pressure manifolds, and computer control systems. Large,
massive hydraulic fracturing (MHF) treatments as illustrated in Fig. 1.7, were developed by
Amoco in the Hydraulic Fracturing Department, Amoco Production Research in Tulsa. The treat-
ments were developed to convert non-commercial, tight gas deposits found throughout North
America into viable, commercial properties. MHF treatments require several million dollars worth
of equipment, utilize in excess of one million gallons of fluid and have placed over 3.3 million
pounds of sand, injected in one continuous operation pumped over 10 hours at rates of approxi-
mately 40 bpm.
Sand and fluid are mixed in a piece of fracturing equipment called a blender. For the first few
years, sand was added to the fracturing fluid by pouring it into a tank or jet mixer containing frac-
turing fluid and connected to the pump suction. Later with less viscous fluid, a ribbon or paddle
type batch blender was employed. Finally, the continuous proportioner and blender was devel-
oped. Blending equipment has become very sophisticated to meet the need for proportioning a
large number of dry and liquid additives, then properly blending them into the base fluid with the
specified concentrations of sand or other propping agents. In order to handle large volumes of
propping agents required in large treatments, special storage facilities have been developed to
facilitate storing and moving the propping agents at the proper rate to the blender. Proportioning
and mixing of the gelling agents has become a very sophisticated procedure utilizing computer
control systems to step or ramp sand concentrations in the blender as shown in Fig. 1.8. It is nec-
essary to blend them in a uniform method to give the maximum yield viscosity. One procedure is
to use a concentrated gelling agent prepared prior to the treatment, then taken to the field where it
is proportioned into the base fluid in a semi-continuous method. A very uniform high yield viscos-
ity is obtained. With the advent of larger size treatments, it has become necessary to have a com-
puter control center (Fig. 1.9) to coordinate all of the activities that are transpiring simultaneously,
each of which is critical.
FRACTURING
FLUID
METERING
PUMP
PROPORTIONING
CONTROL
SAND
BULK OR SACK
AGITATOR
SAND - FLUID
MIXTURE TO
PUMP TRUCK
PRESSURIZER
Today, programs are capable of determining temperature profiles of the treating fluid during a
fracturing treatment. Such a profile can assist in designing the gel concentrations, gel stabilizer
concentrations, breaker concentrations and propping agent concentrations during the various
stages of the treatment. Models have been developed to simulate the way fluids move through the
fracture and how the propping agent is distributed. From these simulations, production increases
can be determined. Following a fracturing treatment, reservoir models and pressure transient anal-
ysis methods can then be used to history match the pressure and production performance to deter-
mine what type of treatment was actually achieved.
The history of fractured reservoir response analysis dates from the late 1960s. Tinsley et al.12 did
work on an electrolytic model to determine the effect fracture lengths and flow capacity would
have on the production increase obtained from wells with a different drainage radius. Several oth-
ers developed mathematical models for similar projections. Nolte and Smith13 developed proce-
dures to correlate between observations made during fracturing treatments and Britt14,15 and
Veatch16-18 presented methods to optimize the fracturing process. Several theories have been
advanced by this work which added considerably to the understanding of the hydraulic fracturing
process. This technology added considerably to the understanding of the hydraulic fracturing pro-
cess and is summarized in the SPE Monograph Volume 12.19
Marked advancements were achieved by Amoco and the industry during the 1970s and early
1980s. Much of what was learned during this period is now being applied to fracturing oil and gas
formations. The most notable contribution was field test procedures and data collection programs
developed to better estimate fracture design parameters. These include prefrac stress tests, minifrac
calibration treatments and the measurement of bottomhole treating pressures during fracturing.
Observations from these tests indicate lateral fracture extension rate, vertical growth behavior,
fracturing fluid leakoff rate, and general characteristics associated with defining fracture geome-
try. This information has led Amoco and the industry to a more precise and systematic approach
to fracture treatment design.
A question often asked today is, What can be changed to maximize profits? As shown in
Fig. 1.10, the optimum treatment results from balancing different parameters, i.e., fracture conduc-
tivity, fracture length and reservoir permeability, to achieve the maximum profit. Generally speak-
ing, the desired fracture length for optimal production is bigger for lower permeability formations
as shown in Fig. 1.11. Conversely, the desired fracture conductivity for optimal production is
greater for higher permeability reservoirs.
The optimum treatment will differ from field to field and from one area of a field to another based
on reservoir characteristics and treatment cost. Recognize that the amount of fluid and proppant
required to achieve a desired penetration will vary greatly from location to location as a function
of lithology, wellbore stresses and fracture containment. Therefore, it is very important for overall
financial optimization, that the optimization process be completed for each different situation and
that at least two or three different fluid and proppant systems be evaluated for each situation.
Fig. 1.12, illustrates a simplified schematic of the optimization process used in the design of
hydraulic fracture stimulations. The upper portion of Fig. 1.12 considers the reservoir response
resulting from fracturing and the revenue produced. The detailed aspects of reservoir behavior are
covered in other courses, however, a general discussion of how these topics relate to optimizing
revenue through fracture design is included in this manual in Chap. 3 and Chap. 9. The lower por-
2
Frac. 1/2 Length
1000s Feet
1
$ Revenue
Reservoir
Simulator
Years Length
$ Revenue
Less
$ Cost
Treatment Vol.
Fracture Length
$ Cost
Fracturing
Hydrafrac
Simulator
The topics detailed in this course include how a fracture is created, what proppants should be used
to hold it open and how the fluid flow in a reservoir is altered. The effect of fracture penetration,
the importance of fracture height development, the concepts of effective wellbore radius, dimen-
sionless fracture conductivity (FCD) and folds of increase (FOI) for steady-state conditions are dis-
cussed. The effect of early time transient production and bilinear flow, and the application of
economic analysis and revenue optimization are elements of coupled reservoir analysis and
The financial results obtained in fracturing can be significantly increased, over the standard prac-
tice of the industry, through a better understanding of the fracturing process, how to optimize a
treatment design, and the implementation of quality control in the field. The nomenclature which
follows on the next pages summarizes the most important and frequently used terms in the manual.
The SPE Monograph Volume 1219 provides a comprehensive review and list of references on
many of the aspects covered in this course.
1.3 Nomenclature
BHCP Bottomhole closure pressure in psi. It is equal to fracture pressure; it is also c.
BHTP Bottomhole treating pressure in psi. It is equal to surface treating pressure plus hy-
drostatic pressure minus friction pressure. It is also equal to BHCP plus PN.
CI Part of Ct. It is the effects of the frac fluid viscosity and relative permeability in
ft/ minute .
CII Part of Ct. It is the effects of the reservoir fluid viscosity and compressibility in
ft/ minute .
CIII Part of Ct. It is the effects of the wall building properties of the frac fluid in
ft/ minute .
Ct The total effects of the frac fluid leakoff coefficient in ft/ minute .
Ct It is the total compressibility factor of the reservoir and fluid in psi-1. It is used to
calculate part of CIII.
FOI Folds of Increase. It is the ratio of the stabilized production after fracturing to the
production before fracturing. It is equal to QFRAC/QUNFRAC.
Rock porosity in decimal percent.
Hp Permeability Height. That portion of the frac height, H, to which frac fluids may be
lost.
K' A property of gelled frac fluids called consistency index and is shear stress at a
strain rate of 1 sec-1. Data supplied by service companies.
L Hydraulic fracture length from tip to tip. It is equal to 2 times the hydraulic frac ra-
dius, xf, in feet.
Viscosity in cp.
n' A property of gelled frac fluids called Power Law Exponent or Flow Behavior In-
dex. Data supplied by service companies. Related to K'.
p The difference between the pressure in the fracture and reservoir pressure in psi,
used in CI and CII.
Pc Critical Pressure or Pressure Capacity. It is the net pressure above closure pressure
where a fracture may become unconfined.
PFCF Proppant Fall Correction Factor. It is a term used to tell the computer that a prop-
pant other than 20-40 mesh is being used, or that fall is through a crosslinked fluid.
PN Net Pressure. The pressure in the fracture above closure pressure. It is equal to
BHTP minus BHCP.
Q FRAC
---------------------
- Same as FOI. A measure of the results of the fracture stimulation.
Q UNFRAC
re Drainage radius in feet. Generally, it is one-half the distance to the next well.
r'w The stimulated wellbore radius effect due to the fracture in feet. It is the effective
or pseudo-wellbore flow radius resulting from the fracture.
t Time in minutes.
VIN Volume of frac fluid pumped into the well in cubic feet.
VLOST Volume of frac fluid leaked from the crack into the formation in cubic feet.
xf Fracture radius in feet (or fracture half-length). Measured from the center of the
wellbore to the end of the proppant on one wing of the fracture.
1.4 References
1. Farris, R. F.: U. S. Patent reissued Nov. 10, 1953, Re 23733.
2. Clark, J. B.: A Hydraulic Process for Increasing the Productivity of Oil Wells, Trans., AIME (1949) 186, 1-8.
3. Maly, J. W. and Morton, T. E.: Selection and Evaluation of Wells for Hydrafrac Treatment, Oil & Gas J, (May
3, 1951) No. 52, 126.
4. Clark, R. C. et al.: Application of Hydraulic Fracturing to the Stimulation of Oil and Gas Production, Drill. &
Prod. Prac., API (1953) 113-22.
5. Huitt, J. L. and McGlothin, B. B. Jr.: The Propping of Fractures in Formations Susceptible to Propping-Sand
Embedment, Drill. & Prod. Prac., API (1958) 115.
6. Huitt, J. L., McGlothin, B. B. Jr., and McDonald, J. F.: The Propping of Fractures in Formations in Which Prop-
ping Sand Crushes, Drill. & Prod. Prac., API (1958) 115.
7. Huitt, J. L.: Hydraulic Fracturing with Single Point Entry Technique, JPT, (March 1960) XII, No. 3, 11.
8. Hubbert, M. K. and Willis, D. G.: Mechanics of Hydraulic Fracturing, Trans., AIME (1957) 210, 153-66.
9. Anderson, T. O. and Stahl, E. J.: A Study of Induced Fracturing Using an Instrumental Approach, JPT (Feb.
1967) 261-67; Trans., AIME, 240.
10. Coulter, G. R. and Wells, R. D.: The Effect of Fluid pH on Clays and Resulting Formation Permeability, pre-
sented at the Southwestern Petroleum Short Course, Dept. of Petroleum Engineering, Texas Tech University,
Lubbock, Texas, April 17-18, 1975.
11. Howard G. C. and Fast, C. R.: Optimum Fluid Characteristics for Fracture Extension, Drill. & Prod. Prac.,
API (1957) 261-70.
12. Tinsley, J. M. et al.: Vertical Fracture Height--Its Effect on Steady-State Production Increase, JPT (May 1969)
633-38; Trans., AIME, 246.
13. Nolte, K. G. and Smith, M. B.: Interpretation of Fracturing Pressures, JPT, (Sept. 1981), 1767-75.
14. Britt, L. K.: Optimized Oil Well Fracturing, Phase I Report, Amoco Production Company Report F84-P-23
(May 25, 1984).
15. Britt, L. K.: Optimized Oil Well Fracturing, Phase II Report, Analysis of the Effects of Fracturing on the Sec-
ondary Recovery Process; Amoco Production Company Report F85-P-7 (Jan. 24, 1985).
16. Veatch, R. W. Jr.: Overview of Current Hydraulic Fracturing Design and Treatment Technology--Part 1, JPT
(April 1983) 677-87.
17. Veatch, R. W. Jr.: Overview of Current Hydraulic Fracturing Design and Treatment Technology--Part 2, JPT
(May 1983) 853-64.
18. Veatch, Ralph W. Jr.: Economics of Fracturing Some Methods and Case Study Examples, Amoco Production
Company Report F89-P-58 (Aug. 3, 1989).
19. Gidley, J. L., Holditch, D. E., Nierode, D. E., and Veatch, R. W., Jr.:, Monograph Series, SPE, Richardson, TX
(1989) 12.
Fracture design models attempt to simulate the natural phenomena associated with the hydraulic
fracturing process. They account for the total volume of fluid injected in the ground (continuity
equation) and estimate the fluid volume that leaks off in the formation and the fluid volume that
remains within the fracture; they relate fracture width to the applied hydraulic pressure (elasticity
equation); they account for pressure loss due to flow within the fracture (fluid flow equation); and
they predict fracture dimensions due to fluid pressure by satisfying a fracture propagation criterion
at the fracture tip.
In many cases, the consideration of continuity and elasticity equations provides insight into the
basic relationship between directly measured qualities of the fracturing process, such as injected
volume and treating pressure.
It states that the volume pumped into the fracture is equal to the volume lost to the formation by
fluid loss plus the volume remaining or stored in the fracture. The individual terms (for a constant
height fracture, pumped at a constant rate) are defined as follows:
Substituting Eqs. (2.2) - (2.4) into Eq. (2.1) , and solving for the tip to tip length, L, gives
Qt
L = ------------------------------------- (2.5)
3CH p t + wH
where Q = pump rate in cubic feet per minute (5.6 cu. ft. = 1bbl), t= pump time in minutes, C =
fluid loss coefficient in ft/ min , Hp = permeable fracture height in feet, w = average fracture width
in feet, and H = total fracture height in feet.
Eq. (2.5) determines the length which will result for a fracture treatment in terms of the other vari-
ables and compares within 10-15% of computer fracture models. Also this equation can be rear-
ranged to form a quadratic equation in terms of t . Solving this equation gives the pumping time
(i.e., VIN) to obtain a desired fracture length.
Inspection of Eq. (2.5) indicates that increasing any of the terms in the denominator (except time)
will decrease the fracture length. In particular, changing the height, H, and/or fluid loss coeffi-
cient, C, can have dramatic effects on fracture length. Fig. 2.1 shows an example of the relation-
ship between fracture height and length for a given treatment volume. Fig. 2.2 shows a similar
relationship between fluid loss coefficient and length.
600
500
400
Height - Feet
300
200
100
0
0 1000 2000 3000
Fracture Length - Feet
Fig. 2.1 - Fracture Height vs. Fracture Length 300,000 Gallon Treatment Design.
Length Polymer
Emulsion
2000
Height
Fracture Length (ft)
High Fluid Loss
1500
0
20 60 100 140 180 220 260
Volume (1000s Gallons)
Fig. 2.2 - Fracture Length vs. Volume Pumped for Low (emulsion) and High (base gels) Fluid Loss
Behavior.
A classical solution in the theory of elasticity predicts that, for an infinite, elastic slab, in plane-
strain (i.e., deformation restricted between parallel planes in the slab), with a pressurized slit
through the slab, the slit will deform into the shape of an ellipse. The ellipse will have a major axis
equal to the slit half-length and a minor axis proportional to the pressure and slit length, and
inversely proportional to the elastic modulus as seen in the upper portion of Fig. 2.3. This elastic
solution was applied to hydraulic fracturing, but in different directions as seen in the bottom por-
tion of Fig. 2.3. As shown, the ellipse in the PK model is vertical while the ellipse in the K model
is horizontal. As a result, a continuing debate has been waged during the last 30 years as to which
is correct. This debate is more than academic since the two models predict significantly different
fluid volumes to achieve a desired fracture length. In this regard, the K model requires greater vol-
ume per foot of length. Additionally, the K model implicitly assumes free slip between the frac-
tured bed and bounding beds which is physically improbable at depth.
ELLIPSE
ELASTICITY
D
W~ _ p P=S+p
E
The prevailing thought within Amoco is that the PKN model is most applicable for fractures which
are long when compared to their height and that the GDK model is more applicable for fractures
which are short compared to their height. In this latter scenario, a penny frac or a 3 Dimensional
model would be more appropriate.
Fig. 2.4 shows the resulting difference between the PKN and GDK models as a result of the dif-
ferent application of the elasticity relation. Note that their relationships for viscosity (for flow of a
Newtonian fluid), rate, and rock modulus are the same. However, the relationships for pressure and
width are very different as shown in Table 2.1.
Table 2.1 - Comparison of Perkins and Kern and Khristianovic Models.
1
Khristianovic WL p ~ ---------
L 1/2
-
W H--E- p L
_
I. Elasticity W~ p
P
W = -4 W
3/4 3/4
III. Combining I & II 1/4 1/4
p E--------
H
- ( QL ) p --------
E
1/2
- ( QL )
L
Fig. 2.4 - Comparison of Perkins and Kern and Khristianovic Models.
For the general case with length greater than height, the PKN model will predict less width; thus
from Eq. (2.5) , the PKN model will generally predict more length. Also, the PKN model predicts
that the net pressure (fluid pressure in fracture minus formation closure pressure) increases as
length, L, (or time, t,) increases, while the GDK model predicts net pressure decreases with length,
L, (or time, t,) as shown on Fig. 2.5.
Bottomhole pressure measurements indicate that, if height is relatively constant and significantly
smaller than fracture length, the pressure will increase as predicted by the PKN model. Also,
downhole televiewer pictures obtained by Amoco, which directly measured the fracture width in
an open hole completion, indicated that the pressure-width relationship of the PKN model was
most applicable.
log p
log p
log L log L
(TIME )
PKN
GDK
log L
Fig. 2.5 - Perkins & Kern (PKN) Model and Khristianovic (GDK) Model.5
The consequence of the different width assumptions in the models can be seen by a comparison of
service company designs based on exactly the same requested input. This comparison was made
by Amoco in 1980. The input variables supplied to the service companies are shown in Table 2.2.
Table 2.2 - Input Values - Service Company Designs.
Viscosity 100 CP
Table 2.3 shows the dramatic variations in the results because of the different schools of thought
in each company at that time. As shown, the Halliburton and Dowell Programs were based on the
GDK model, while the Western, Smith and Amoco programs were based on the PKN model. It is
noted that the BJ program set the leakoff height to 200 ft instead of 100 ft and the Western model
assumed that the fracture width down the complete length was the maximum value at the wellbore.
The large differences in the output indicate the impact of modeling assumptions associated with
comparing service company bids and highlight the importance of knowledgably designing your
own treatments. However, many oil companies still rely on the service companies for designs.
Table 2.3 - Results - Service Company Designs.
Average Volume,
Company Model Type Width Inches Sand, M lb M gal Pad, M gal
2.3 References
1. Perkins, T. K. Jr. and Kern, L. R.: Widths of Hydraulic Fractures, JPT (Sept. 1961) 937-49; Trans., AIME, 222.
2. Khristianovic, S. A. and Zheltov, Y. P.: Formation of Vertical Fractures by Means of Highly Viscous Fluids,
Proc., Fourth World Pet. Cong., Rome (1955) II, 579.
3. Nordgren, R. P.: Propagation of a Vertical Hydraulic Fracture, SPEJ (Aug. 1972) 306-14; Trans., AIME, 253.
4. Geertsma, J. and de Klerk, F.: A Rapid Method of Predicting Width and Extent of Hydraulically Induced Frac-
tures, JPT (Dec. 1969) 1571-81; Trans., AIME, 246.
5. Nolte, K. G. and Smith, M. B.: Interpretation of Fracturing Pressures, JPT (Sept. 1981) 1767-75.
kfw
F CD = ---------
- (3.1)
k xf
relates the fracture's ability to flow fluids to the wellbore to the reservoir's ability to flow fluids to
the fracture. If, for example, FCD is low (FCD 1.6) the fracture has finite conductivity and the res-
ervoir fluids would rather flow towards the wellbore than the fracture. It further indicates that
increasing fracture length would not result in improved reservoir response. Conversely, if FCD is
high (FCD 500), the fracture has infinite conductivity. As a result, increasing fracture conductiv-
ity would not improve reservoir response. For practical purposes, fractures having FCD > 30 act as
infinite conductivity fractures. The parameters used to define FCD are illustrated in Fig. 3.1.
k
kf
xf
Fracture Length
Fracture length or penetration generally has the greatest impact on low permeability reservoirs.
The following examples are from the Wattenberg Field, which is operated by Amoco Production
Company. This field is located north of Denver, Colorado, and has a permeability of about 0.005
md. Fig. 3.2 shows the effect of fracture half-length, xf, on cumulative gas production. As shown,
increasing fracture half length results in significant incremental gas recovery over a 25-year
period.
2000
1800
Cummulative Gas Production - MMCF
FRACTURE LENGTH
1600
1400
800 1000 ft
600
400 400 ft
RADIAL FLOW
200
0 2 4 6 8 10 12 14 16 18 20 22 24
Time (years)
Fig. 3.2 - Effect of Fracture Length Cumulative Gas Produced (25 Years).
Reservoir Permeability
Reservoir permeability, k, and its effect on fractured well performance is illustrated in Fig. 3.3 and
Fig. 3.4. Shown in the figures is the pressure distribution map for only one quadrant of a fractured
well. The pressure distribution map was obtained from a computer simulation after the well,
located in the upper left corner, was produced for a period of time. The simulated fracture in
Fig. 3.3 is located vertically on the left and has a high fracture flow capacity, kfw. The formation
permeability, k, in the computer simulator was very low at 0.005 md (5 micro darcies). Contours
of the pressure profile in psi were made and because gas flows perpendicular to these pressure con-
tour lines, streamlines which represent the path by which the gas travels to the well can be drawn.
Since the formation permeability is extremely low relative to the fracture flow capacity (kfw), the
flow is nearly linear and the fracture acts as an infinite conductivity fracture. As a result, the frac-
ture carries almost all the total gas flow to the well. The path of least resistance is the shortest dis-
tance to the fracture.
Fig. 3.4 shows a pressure distribution map for a fractured well with the same fracture flow capacity
as in Fig. 3.3, but this time the formation permeability is significantly higher at 100 md. Since the
formation permeability more nearly approximates the fracture flow capacity, equal pressure lines
become circular and the flow is nearly radial as can be seen by converging flow lines. In this case,
the fracture carries a relatively small fraction of the total gas flow which indicates that the benefit
realized from the fracture stimulation was minimal. In this case, the path of least resistance is pri-
marily via the reservoir.
psi
psi
1000psi
1200psi
psi
psi
psi
1200 psi
1000 psi
400 psi
600 psi
400 psi
600 psi
800 psi
800 psi
Well
1000
1200
400
600
800
Well
Pressure
Contour
PRESSU
Lines Pressure
Contour
F
Lines
R
F
A R
C A
T C
U Streamlines T
U Streamlines
R
E R
E
Fig. 3.3 - Pressure Distribution and Fig. 3.4 - Pressure Distribution and Approxi-
Approximate Streamlines, Reservoir K = mate Streamlines, Reservoir K = 100 md.
0.005 md.
The key difference in Fig. 3.3 and Fig. 3.4 is the ratio of the fracture flow capacity to the reservoir
permeability, k.
Fracture flow capacity is defined as the product of the permeability in the fracture, kf, and the frac-
ture width, w, with dimensions of md-ft. It is also referred to as fracture conductivity. Shown in
Fig. 3.5 are three types of fracture flow capacity. An infinite flow capacity fracture is a fracture
that acts similar to a large diameter pipeline where there is essentially no pressure drop from the
tip of the fracture to the wellbore. A finite flow capacity fracture has a pressure drop along the frac-
ture that is proportional to the fracture flow capacity, kfw. Nearly all created fractures have finite
capacity. The reservoir response associated with variable conductivity fractures is governed by the
arithmetic average flow capacity.
Estimates of kfw are available from the service companies and Amoco's Production Research
(APR) Department. The STIM-LAB data in Fig. 3.6 shows the effect of proppant type on liquid
permeability. The entire set of Stimlab data can be accessed in the Proppants Manual or from APR.
Fig. 3.6 shows that the manufactured proppants bauxite, intermediate density proppant and zirco-
nia have high permeability up to very high closure stresses.
INFINITE CAPACITY
kf
FINITE CAPACITY
k f1 k f2
The resin coated sand has intermediate permeability values, and the sands (Brady and Ottawa)
have the lowest values at higher stresses. Fig. 3.6 indicates that the Brady sand has higher per-
meability for closure pressure less than 5000 psi (i.e., nominally 6000 to 7000 ft) than the more
pure silica sand of the Ottawa type. This results because the Brady sand tends to be coarser (i.e.,
more toward 20 mesh) and more angular. At higher stresses the less pure and more angular sand
has less permeability (i.e., more crushing).
Fig. 3.7 shows laboratory values of conductivity, kfw for both Brady and Ottawa type sands. Note
that the Ottawa types are not available in the coarser sizes, while Brady is not available for the finer
sizes. Notice that at 4000 psi, the 8/16 Brady sand has about 5 times more conductivity or capacity
than the commonly used 20/40 Ottawa (i.e., 15,000 vs 2800 md ft).
Post treatment evaluation experience indicates that in-situ capacity is dramatically less than these
laboratory values. This results from gel residues, fluid loss additives and potentially rock debris.
Indicated values are about 1/3 - 1/10 of the lab values. In addition, Amoco's design program indi-
cates that propped widths of more than 1 lb/ft2 are difficult to achieve. It is noted that some service
companies claim they achieve 4 lb/ft2. Since the laboratory standard (i.e., Fig. 3.7 is 2 lb/ft2); a fur-
ther reduction for width must be made. The best method to determine in-situ capacity is to perform
well tests in the field and use the bilinear flow analysis techniques discussed in Section 3.3. If
actual in-situ values are not available, the following guideline for capacity should be used.
k f w lab data 2
expected k f w = 0.3 ---------------------------------lb/ft
2
expected (3.2)
lb/ft lab data
Most lab tests are run at 2 lb/ft2. However, your test data may be different. Proppant concentration
at which the tests were run should be available, or the data should not be used. The 0.30 factor is
a permeability reduction applied to the lab data to correct for inherent differences in in-situ fracture
conditions and idealized laboratory conditions. This is nothing but a fudge-factor and varies
widely. This correction may be used for scoping studies, but pressure transient testing is still the
preferred technique to obtain the actual in-situ value of kfw.
The importance of fracture conductivity and fracture length are illustrated in Fig. 3.8 through
Fig. 3.10. These figures show the results of simulations which combine variations of conductivity
and length with reservoir permeabilities of 0.005, 0.08, and 5.0 md, respectively. The results are
shown as the ratio of flow rate after fracturing to that before stimulation. This ratio is known as
Folds of Increase, FOI.
Proppant Concentration
2 20 8/16 2 20 2.0 lb/ft2
foot,DDxxftft
7
6 12/20
6 6 6
5
5
darcyxxfoot,
4 4 16/30
4
Wf,darcy
4
20/40
3 3
3 3 20/40
Note: No allowances have been made
for embedment or any form of
3-6
2 2 pack damage.
2 2
Conducitivty,klklxxWf,
30/50
1 1.0 Fracture Conductivity, kl x Wf, darcy x foot, D x ft
FractureConductivity,
1 1.0 9
Fracture
9 8 0.8
8 0.8 7 40/70
7 6 0.6
6 0.6 5
5
4 0.4
4 12JUL83
0.4 14JUL83 RWA
Note: No allowances have been made
for embedment or any form of RWA 3 0.3 70/140
3 0.3 pack damage.
Ottawa Sand
Brady Sand
July 1999
Reservoir Response To Fracture Stimulation
Fig. 3.8 shows for a formation permeability equal to 0.005 md that as the fracture flow capacity,
kfw, is reduced from 1000 md-ft to 1.0 md-ft, the effect of improved flow rate due to increased
fracture length is diminished. However, the effect becomes significant when kfw is increased from
1 to 10 and 100 md-ft. Beyond a kfw = 100 md-ft, the effect of increasing fracture flow capacity
has diminishing returns. Fig. 3.9 and Fig. 3.10 show that as formation permeability increases, the
effect of improved flow rate due to increasing the fracture length diminishes further.
12
12 11
11
K=0.05 MD
10
K=0.005 MD
10 9 Kfw=1000 Md-ft
kfw = 1000 Md-ft
9
Qfrac/Qunfrac
kfw = 100 Md-ft 8
Qfrac/Qunfrac
8
7
7 Kfw=100 Md-ft
kfw = 10 Md-ft 6
6
5
5
4
4
3 Kfw=10 Md-ft
3
2 2
kfw = 1 Md-ft Kfw=1 Md-ft
1 1 0
0 100 200 300 400 500 600 700 800 100 200 300 400 500 600 700 800
Fracture-Half Length Fracture-Half Length
Fig. 3.8 - Formation Permeability Equal to Fig. 3.9 - Formation Permeability Increased
0.005 md. to 0.05 md.
12
11
K=5.0 MD
10
8
Qfrac/Qunfrac
Fig. 3.9 shows a similar graph where formation permeability, k, is increased to 0.05 md. Notice
that increasing the flow capacity, kfw, above 100 md-ft will still have an effect on improving flow
rate. This was not the case when k was 0.005 md. Also note that for fracture flow capacities equal
to 10 md-ft or lower, there is little rate improvement as the fracture length increases.
Fig. 3.10 shows a similar plot with formation permeability, k, of 5.0 md. This plot shows that
increasing fracture length beyond 200 ft in a 5 md reservoir, has little productivity advantage.
Fig. 3.10 exposes the myth that fractures are only for low permeability wells. As reservoir perme-
ability increases, the Qfrac/Qunfrac ratio decreases for a given fracture length and conductivity. But
since for radial flow, the base rate is directly proportional to permeability, the base rate (Qunfrac) is
increasing. Would you invest in a frac for a 5 md well making 10 MMCFD? Fig. 3.10 indicates
that a 100 ft, 1000 md-ft frac would make it a 25 MMCFD well. When the importance of short,
high conductivity fractures is better understood, many high permeability wells will be fractured in
the future. In general, wells in high permeability reservoirs are the least expensive to stimulate and
often provide the greatest incremental benefit.
Fracture Orientation
As a reservoir's permeability decreases, the drainage pattern becomes more elliptical (i.e., smaller
aspect ratio) for an optimum fracture. This results because of two reasons: first, the drainage per-
pendicular to the fracture face decreases, and second, the optimum fracture length is longer.
Fig. 3.11 shows the effect of fracture orientation on reservoir drainage. This figure shows the ellip-
tical patterns after 10 and 25 years for Wattenberg reservoir conditions on 320 acre spacing. The
upper portion of Fig. 3.11, shows fractures placed properly with respect to the fracture orientation.
As shown, there is little interference and relatively complete drainage would occur. However on
the lower portion of Fig. 3.11, for a 45 azimuth, there is significant overlap of the patterns and
substantial areas of the reservoir that will not be drained. Also note that the contours are for a 300
psi drawdown at 10 and 25 years - very far from depletion.
If a similar contour map of the well configuration (unfavorably oriented) shown in the lower por-
tion of Fig. 3.11, was made after 100 years of production, it might show as complete a coverage or
drainage as the well configuration in the upper portion of Fig. 3.11 has shown in 25 years. It suf-
fices to say that fracture orientation can have a significant affect on both ultimate recovery and rate
acceleration benefits derived from fracturing.
It is obvious that to generally benefit from knowing the orientation, well placement must be
selected in a manner that differs from normal practices. The required spacing is with wells closer
in the direction perpendicular to the fracs and farther apart in the direction of the fracs. Also since
the orientation is likely not to be near 0 or 45 , the optimum well placement will be quite differ-
ent than normal patterns of subsequent quartering sections. An SPE paper by M. B. Smith1 gives
an excellent study of the effect of fracture azimuth, well spacing, and lost production for Watten-
berg.
DRAINAGE AREAS
INITIAL PRESSURE - 2800 PSI FORMATION PERMEABILITY = 0.004 md
2500 PSI
10 YEARS
5280'
25 YEARS
5280
DRAINAGE AREAS
INITIAL PRESSURE - 2800 PSI FORMATION PERMEABILITY = 0.004 md
2500 PSI
5280
10 YEARS
25 YEARS
5280
The pseudo steady-state radial flow for fractures in moderate-to-high permeability reservoirs per-
mits modeling by the effective wellbore concept. This concept was introduced by Prats2 along
with the term, FCD, discussed previously (page 3-1).
This powerful tool indicates that fracturing wells in moderate-to-high permeability reservoirs is
equivalent to increasing the area of the wellbore, i.e., a giant under-reaming job. Thus fracturing
in moderate-to-high permeability reservoirs is equivalent to enlarging the wellbore. Consequently
the relative benefits of fracturing are the same for heavy or light oils.
Taking the wellbore analog further and using the steady-state radial flow equation, the ratio of pro-
duction after and before fracturing is
q ln ( r e /r w )
-----f = FOI = -----------------------
- (3.4)
qo ln ( r e /r' w )
where FOI= folds of increase, qf = postfrac production rate, qo = prefrac production rate, re = exter-
nal drainage radius, rw = actual wellbore radius, and r'w = effective wellbore radius. When evalu-
ating the ratio of production in Eq. (3.4), the drawdown pressure, permeability and viscosity are
assumed the same before and after fracturing.
Prats also gave the theoretical relationship between r'w and dimensionless flow capacity. Fig. 3.12
gives this relationship in terms of FCD. The figure shows that for FCD > 30, that r'w = 0.5 xf; i.e.,
the fracture acts as an infinitely conductive fracture and there is no benefit from increasing FCD.
Fig. 3.12 also shows for small FCD (i.e., less than 0.3) that r'w is independent of the fracture length
and depends only on conductivity.
Studying Fig. 3.12 will reveal where the producer should be spending his money to increase the
results of a fracture stimulation. For example, if the reservoir permeability is 10 md, the fracture
has a conductivity of 1000 md-ft, the fracture half length is 500 ft, wells are 2000 ft apart, and bore-
hole diameter is 5.5 in:
F CD = 1000/10 500 = .2
r' w /x f = .048
Therefore,
x f = 1000 ft
F CD = 1000/10 1000 = .1
FOI = 2.04
Notice that the cost of the fracture stimulation would have more than doubled by going from xf =
500 ft to xf = 1000 ft with NO increase in r'w or FOI.
Suppose, instead of a longer frac, the decision is made to improve kfw. If kfw = 2000 md-ft instead
of 1000 md-ft.
F CD = 2000/10 500 = .4
r' w /x f = .09
( ln 1000/0.229 )
FOI = --------------------------------------
( ln 1000/45 )
= 7.6/3.1 = 2.45
Notice by doubling conductivity, a productivity increase of 20% has been accomplished. A review
of Fig. 3.7 indicates that conductivity could be doubled simply by changing from 20/40 to 16/30
mesh sand.
In summary, for FCD less than 0.5, increasing xf is a total waste of time and investment. The invest-
ment should be made on a higher conductivity proppant.
F CD = 1000/.02 1000 = 50
r' w /x f = .5
( ln 2000/0.229 )
FOI = --------------------------------------
( ln 2000/500 )
= 8.294/1.386
FOI = 5.98
The decision is made to improve fracture conductivity, kfw from 1000 to 2000.
r' w /x f = .5
( ln 2000/0.229 )
FOI = -------------------------------------- which is the same as before
( ln 2000/500 )
= 5.98
Notice, greatly improving fracture conductivity, kfw, had NO effect on increasing FOI.
F CD = 1000/.02 2000 = 25
r' w /x f = .48
( ln 2000/0.229 )
FOI = --------------------------------------
( ln 2000/960 )
= 8.294/.734
FOI = 11.3
It is evident from the above, that if FCD is greater than 25 to 30, improving fracture conductivity is
not helpful. The investment should be made to achieve more fracture length to increase FOI, if the
increased production offsets the increased cost of the treatment (i.e., economics, addressed in
Chap. 9). When FCD's are between 0.5 and 25, FOI will experience an increase if xf or kfw is
increased. Therefore when FCD's fall in the range of 0.5 to 25, economics must be used to determine
whether improving conductivity or creating longer fractures, or some combination of both, is the
most cost effective (i.e., profitable).
In using the FOI technique just shown, xf must be determined by trial and error for a design. That
is, once a FOI is selected, a r'w can be calculated that will be required to effect a given production
increase. However, since for finite conductivity fractures, xf affects both r'w and FCD, the xf is
required to yield the desired FOI.
Fig. 3.13 shows a modified version of Fig. 3.12 which includes the conversion of
( ln r e /r w )
FOI = -------------------------
( ln r e /r' w )
on the left vertical axis. On the right vertical axis are various xf /re curves. The horizontal axis is
kfw/kre. This parameter should be known for specific proppant size and concentration (i.e., kfw)
since the k and re should be known. Also from xf /re on Fig. 3.13, xf can then be determined from
the known re.
Fig. 3.14 shows the use of Fig. 3.13 for a case with a desired FOI = 5 (denoted by a), 160 acre
spacing (denoted by b), a horizontal line (denoted by c), the value of kfw/kre = 1.1 (denoted
by d), the intersection (denoted by e), and finally the indicated xf /re of 0.75 (denoted by f)
to achieve the FOI.
Fig. 3.13 can also be used in reverse; i.e., find the FOI for a given xf /re.
Another example, the objective is an FOI = 4, well spacing is 640 acres (re = 2640), k is 0.1 md
and the proppant selected will have a kfw of 1320 md-ft at the proposed concentration and closure
stress. This gives kfw/kre = 5
Enter Fig. 3.13 from the left vertical axis with FOI. Find the intersection for FOI of 4 and the well
spacing of 640 acres. This determines r'w/re. A horizontal line should be drawn from the intersec-
tion of the FOI and the spacing line, completely across the graph. Then enter Fig. 3.13 from the
bottom with kfw/kre of 5. Draw a vertical line up to intersect the r'w /re line. A curved line should
be drawn to the right vertical axis from the intersection of kfw/kre and r'w/re parallel to the xf /re
lines, xf/re is then determined to be 0.2.
Therefore,
x f /r e = 0.2
x f = 528 ft
Notice, that by varying kfw on the horizontal axis, xf /re and therefore xf will change.
Studying this graph will also show quickly where to invest time, effort and money. When the xf /re
curves become horizontal, increasing kfw will not result in an increase in FOI. Also, when kfw/kre
is very small, increasing xf has a minimal effect on FOI.
When designing any fracture stimulation, engineers must consider two primary factors:
(1) designing the treatment to yield the highest productivity or injectivity per dollar cost, and
(2) designing the treatment to minimize any loss in reserves. For moderate permeability wells
under primary recovery, fracture length should be optimized to reservoir permeability and fracture
conductivity. For reservoirs under secondary recovery, the fracture length must not only be eco-
nomically optimized as above, but other factors such as the impact of fracture length and fracture
orientation upon recovery must be addressed.
Two research reports by L. K. Britt,3,4 have been published which provide significant insight into
the importance of length and fracture orientation on secondary recovery projects.These reports
drew several conclusions that are pertinent to fracture stimulation design in waterfloods:
1. The older potentiometric reservoir response models, such as McGuire and Sikora are invalid.
2. Prats' effective wellbore radius concept (Fig. 3.12), whereby the effect of a fracture upon res-
ervoir response is modeled as an increased wellbore radius, is valid if frac lengths are less than
25% of the interwell distance.
3. Short fractures cause no loss in reserves, and can contribute significantly to rate acceleration.
4. Fracture length (radius) greater than 25% of the distance between injector and producer may
reduce reservoir recovery when the fracture orientation is unfavorable (injector or producer)
and improve recovery when the fracture orientation is favorable (injector to injector).
5. The economically optimum fracture stimulation for moderate permeability reservoirs (1-50
md) is short, with very high conductivity.
6. In-situ fracture proppant conductivity is on the order of 10-30% of published laboratory data.
To verify that Prats' results were correct using Amoco's reservoir simulators, the Coning model
was used to simulate primary recovery from a fractured moderate-permeability reservoir. Runs
were made comparing productivity by combining a radial model using Prats' effective wellbore
radius to simulate the effect of the fracture, and an areal gridded model using the Coning model
with actual fracture parameters. The results were found to be nearly identical.
This comparison was further evaluated for secondary recovery by using a model to compare a frac-
ture simulated in a radial mode using Prats' effective wellbore radius to an areal model for a five-
spot waterflood pattern with both injectors and producers stimulated with identical fractures
(Fig. 3.15).
80
PERCENT ERROR (%)
40
20
0
0 0.2 0.4 0.6 0.8 1
Increasing the fracture length on the gridded model provides the correct answer used as the basis
for the evaluation. Increasing the effective wellbore radius in the radial model to compare to that
in the areal model introduces about 10% error when the fracture length for each well reaches 25%
of the interwell distance, implying that Prats' radial flow curves are in error beyond this point.
The effect of fracture length on recovery was also evaluated for a five-spot moderate permeability
waterflood pattern. Fig. 3.16 shows the results of increasing fracture length on recovery. Recovery
is relatively unaffected for fracture lengths up to about 25% of the interwell distance. This data is
for the most unfavorable fracture orientation, where the producing well fracture is directly in line
with the injection well fracture.
50
40
PERCENT LOSS IN RECOVERY
30
20
10
0
0 10 20 30 40 50
FRAC RADIUS/INTERWELL DISTANCE, %
It should be noted that even though recovery is about the same for short fractures, the rate of recov-
ery can be significantly different. For moderate permeability, and a maximum fracture length of
25% of the interwell distance, 2 HCPV of water could be injected 20-30 years sooner than if the
well were unfractured, significantly increasing the economic viability of the project. Note also that
results of a study conducted by Connie Bargas5 indicate that unfavorable mobility recovery pro-
cesses (i.e., CO2 floods) are even more sensitive to fracture length and orientation.
When fracture stimulation is used to work over wells to restore lost injectivity or productivity, we
must ensure that the two goals stated at the beginning of this section are met. That is, fractures must
be designed to yield the maximum rate of return on investment, and must not reduce recovery due
to excessive length. In most cases, the economically optimum length will be less than the maxi-
mum to affect recovery.
To assure that secondary recovery is not affected by the placement of fractures in the reservoir, the
design fracture radius should not exceed the maximums shown in Table 3.1 unless wells are favor-
ably oriented.
In any situation where the potential to infill drill a field is high, some guidelines must be estab-
lished for the tightest well spacing that might be drilled. The maximum design frac length should
not be allowed to exceed 25% of that interwell distance. Once a hydraulic fracture is created, and
conductivity established either by proppant or by acidizing, we obviously cannot reduce that frac
length.
10 ac 165 ft
20 ac 233 ft
40 ac 330 ft
80 ac 466 ft
160 ac 660 ft
Class Problem
Find: xf
Given: k = 1 md, 160 acre spacing, Depth = 6000 ft (normal grad.), re = 1320 ft
Find: xf for 20-40, 12-20, 6-12 Brady sand to obtain 5-fold increase in production over non-
damaged or stimulated wellbore.
12-20 ____________
8-16 ____________
20-40 500
12-20
8-16
Explain:
Acid Fracturing
Fracturing with acid in carbonates creates a highly-conductive, etched fracture. Fig. 3.13 can be
used for predicting performance of an acid fracturing treatment by assuming FCD = (i.e., infinite)
or effectively greater than 30. The line shown on Fig. 3.17 represents an infinite conductivity frac-
ture (FCD > 30), and is equivalent to the vertical line for a specific kfw/kre for a propped fracture
(i.e., line d on Fig. 3.14). Equivalently for a given xf /re or FOI a horizontal line can be drawn
directly across Fig. 3.17 to determine the relationship between FOI and xf /re.
Many carbonate wells are initially acidized and later fractured with proppant. This causes a sand
production problem after the fracture treatment because any sand in an acid channel will not be
trapped and is eventually washed into the wellbore by production fluids. Therefore, if a propped
fracture would give a larger FOI, it would be desirable to conduct this fracture initially, thereby
saving the cost of an acid treatment, obtaining more production, and reducing sand production
problems.
For 40 acre spacing, maximum acid xf = 150 ft, maximum kfw = 1300 md-ft for
proppant, find if an acid frac or propped frac appears more optimum for k = 1 md
and k = 5 md.
Infinite Conductivity
Unstimulated
Fig. 3.18 also shows that fracture conductivity is even more important for transient flow than
pseudo steady-state flow. For the steady-state case of Prats (Fig. 3.12), there was little improve-
ment for FCD greater than 10; however, Fig. 3.18 shows that for tDf < 0.1, there is a dramatic reduc-
tion in qD (approximately proportional to the inverse of flow rate), or a dramatic increase in rate if
is increased from 1.004 to 1.234. This approximate doubling of flow rate is very significant for
fractures in very low permeability reservoirs which can stay in transient flow for a substantial por-
tion of their productive life.
The dimensionless time (tDf) on Fig. 3.18 is proportional to k/xf2. Therefore, low permeability res-
ervoirs which require large xf's tend to fall on the left side of Fig. 3.18, while higher permeability
reservoirs which require only short, but highly conductive fractures, tend to fall on the right side
where the much simpler steady-state analyses are applicable.
Fig. 3.18 also shows finite capacity fracture behavior (i.e., 1.045 1.234). In finite capacity
fractures, bilinear flow can occur. During bilinear flow, the pressure transient has not reached the
tip of the fracture; both linear flow from the reservoir to the fracture and linear flow down the
length of the fracture are occurring. The bilinear flow region, is very important for two reasons:
(1) unique fracture length cannot be found from the production response, and (2) the actual value
of conductivity in-situ, kfw can be determined. The log-log curves, either constant rate or pressure,
have a 1/4 slope for bilinear flow.
Fig. 3.19 shows a plot of pressure change vs. the fourth root of time for fractures with an FCD of
greater than 1.6, equal to 1.6, and less than 1.6, respectively. In addition, the lower portion of
Fig. 3.19 shows the effect of damage on the fourth root of time behavior. The upper plot on
Fig. 3.19 shows that a straight line should result on a pressure change vs. fourth root of time if the
fracture is in bilinear flow. It also shows how the data deviates from the straight line (bilinear flow)
is a qualitative indicator of FCD. If, for example, the data deviates up from the bilinear flow line
this indicates that FCD is greater than 1.6. Conversely, if the data deviates downward from the bilin-
ear flow line the FCD < 1.6. The lower plot on Fig. 3.19 indicates that if the bilinear flow line does
not go through the origin, the entrance to the fracture is damaged. This loss of production can result
from:
kill fluid was dumped into the fracture - let fracture clean up before conducting test.
Fig. 3.20 shows an example of these plots and the indicated kfw.
The data in Fig. 3.20 deviates downward from the bilinear flow line qualitatively indicating that
the FCD is less than 1.6. Since FCD is low, efforts should be made to either increase fracture con-
ductivity, reduce fracture length, or both. A more complete presentation of the transient response
of fractured wells is included in the Pressure Transient Analysis manual from the PTA course
given by the Training Center. Because of the importance of bilinear flow in the analysis of frac-
tured reservoirs and the improvement of treatment design, the section on bilinear flow from the
PTA course is included in this chapter for ease of reference.
SLOPE = mbf
FCD < 1.6
p, psi
BILINEAR
END OF FLOW
t1/4, hours1/4
DAMAGE OR
CHOKED FRACTURE
p, psi
ps IDEAL
0
0
t1/4, hours1/4
AMERADA BOMB
Downward Deviation
From Bilinear Flow
Line indicates FCD is
less than 1.6
Mbf = 134
Kfw = 1168/RcD = 1320 mdft
Fig. 3.21 depicts the various flow periods which are associated with finite conductivity vertical
fractures.
The Fracture Linear Flow, (a) on Fig. 3.21, is the first flow period which occurs in a fractured
system. Most of the fluid which enters the wellbore during this period of time is a result of expan-
sion within the fracture, i.e., there is negligible fluid coming from the formation. Flow within the
fracture during this time period is linear.
Equations which can be used to predict the following formation face pressure, pwf, during fracture
linear flow are presented by Cinco-Ley et al.,6 for the constant rate case. This reference also pre-
sents an equation which predicts the time when this flow period ends. Unfortunately, fracture lin-
ear flow occurs at a time which is too early to be of practical use in well test analysis.
Bilinear Flow
The next flow period to occur is called Bilinear Flow, (b) on Fig. 3.21, because two types of lin-
ear flow simultaneously occur. One flow is linear incompressible flow within the fracture and the
other is linear compressible flow in the formation. Most of the fluid which enters the wellbore dur-
ing this flow period comes from the formation. Fracture tip effects do not affect well behavior dur-
ing bilinear flow; accordingly, unless a well test is run sufficiently long for bilinear flow to end, it
will not be possible to determine fracture length from the data.
Bilinear flow was first recognized by Cinco-Ley et al.6 Since its introduction into literature, the
use of bilinear flow analysis to characterize both formation and fracture properties has been docu-
mented.7-11 The details of analyzing bilinear flow data will be detailed in subsequent discussions
beginning on page 3-35.
The analysis of Formation Linear Flow, (c) on Fig. 3.21, is covered in the Pressure Transient
Analysis course manual.
Pseudo-Radial Flow
The analysis of Pseudo-Radial Flow, (d) on Fig. 3.21, is covered in the Pressure Transient Anal-
ysis course manual.
WELL
WELL FRACTURE
FRACTURE
FRACTURE
FRACTURE
WELL
Dimensionless Pressure:
kh ( p i p wf ) khm ( p )
P D = ------------------------------- ( oil ) P D = ----------------------- ( gas ) (3.5)
141.2qB 1424T q
Dimensionless Time:
0.0002637kt
t Dxf = -----------------------------
2
- (3.6)
c t x f
kfw
F CD = --------- (3.7)
kx f
2.45 t Dx 1/4
PD -f
= ---------------------
1\/2
(3.8)
F CD
44.1qB 1/4
p i p wf = ------------------------------------------------------
1/2 1/4
- t (3.9)
h(k f w) ( c t k )
494qT
m bf = -------------------------------------------------
1/2 1/4
- (3.10)
h ( k f w ) ( kc t )
Constant Formation Face Pressure
Dimensionless Rate:
141.2qB 1424Tq
q D = ------------------------------- (oil) q D = ----------------------- (gas) (3.11)
kh ( p i p wf ) khm ( p )
2.72 t 1/4
1 Dx f
------ = -----------------------
- (3.12)
qD F 1/2
CD
1 48.9B 1/4
1/2 1/4
- = t ( oil )
--- = --------------------------------------------------------------------------
q ( p i p wf )h ( k f w ) ( c t k )
1 494T 1/4
--- = --------------------------------------------------------------------
1/2 1/4
- = t (gas)
q h ( k f w ) ( kc t ) m ( p ) (3.13)
48.9B
m bf = --------------------------------------------------------------------------
1/2 1/4
- (oil)
( p i p wf )h ( k f w ) ( c t k )
494T
m bf = -------------------------------------------------------------------
1/2 1/4
(gas)
h ( k fw ) ( kc t ) m ( p ) (3.14)
Note: The equations presented in this section are written specifically for pressure drawdown tests.
These equations can be modified for pressure buildup tests by replacing the pressure differ-
ential p = p i p wf , and the producing time, t, with appropriate values as shown in the fol-
lowing table:
Drawdown p = pi-pwf t
Buildup p = pws-pwf t or te
When the rate of a well is maintained constant, the pressure change at the formation face is
described by Eq. (3.9). This equation indicates that a plot of pi-pwf (pws-pwf) for buildup tests) vs.
t1/4 (t1/4 for buildup tests) will yield a straight line with slope, mbf, predicted by Eq. (3.10). The
plot of pressure change vs. fourth root of time is illustrated by Fig. 3.22. When bilinear flow ends,
the straight line will end and the plot will exhibit curvature which is concave upward or downward
depending upon the value of the dimensionless fracture conductivity, FCD. When FCD 1.6, the
curve will be concave downward; a value of FCD > 1.6 will cause the curve to be concave upward.
SLOPE = mbf
FCD < 1.6
p, psi
END OF
BILINEAR FLOW
t1/4, hours1/4
When FCD > 1.6, bilinear flow ends because the fracture tip begins to affect wellbore behavior. If
a pressure transient test is not run sufficiently long for bilinear flow to end when FCD > 1.6, it is not
possible to determine the length of the fracture. When FCD 1.6, bilinear flow in the reservoir
changes from predominately one-dimensional (linear) to a two-dimensional flow regime. In this
case, it is not possible to uniquely determine fracture length even if bilinear flow does end during
the test.
A more diagnostic plot to recognize the occurrence of bilinear flow is the log-log plot. From
Eq. (3.9),
44.1qB 1
log ( p i p wf ) = log -------------------------------------------------
1/2 1/4
- + --- log t . (3.15)
h ( k f w ) ( c t k ) 4
Eq. (3.15) indicates that a log-log plot of pi-pwf vs. t will yield a straight line with a one-fourth
slope; this is illustrated by Fig. 3.23.
SLOPE = 1/4
p, psi
t, hours
Fig. 3.23 - Log-log Plot Illustrating the Effect of Ideal Bilinear Flow for the Constant Rate Case.
When formation face pressure remains constant, the formation face rate will change with time as
described by Eq. (3.13). According to Eq. (3.13), a plot of 1/q vs. t1/4 should yield a straight line
with slope, mbf, defined by Eq. (3.14) this plot is depicted by Fig. 3.24. Following the end of the
bilinear flow period, the curve for F CD 2.8 will be concave downward and the curve for FCD >
2.8 will be concave upward. The straight line caused by bilinear flow ends for the same reasons as
described for the constant rate case.
Eq. (3.13) also indicates that a log-log plot of 1/q vs. t should yield a straight line with a slope of
one-fourth:
SLOPE = mbf
1/q Dp, psi FCD < 2.8
END OF
BILINEAR FLOW
t1/4, hours1/4
SLOPE = 1/4
Dp, psi
1/q
t, hours
Fig. 3.25 - Log-log Plot Showing Effect of Ideal Bilinear Flow for the Constant Rate Case.
The relationship between (tDxf)ebf and FCD is depicted graphically by Fig. 3.26. This relationship
can be approximated as:
0.1
F CD 3: ( t Dxf ) ebf ---------
2
- (3.17)
F CD
1.53
1.6 < F CD < 3: ( t Dxf ) ebf 0.0205 ( F CD 1.5 ) (3.18)
For the case where FCD 3, the dimensionless pressure at the end of bilinear flow is
1.38
( p D ) ebf = ---------- . (3.20)
F CD
Therefore,
1.38
F CD = ------------------ (3.21)
( p D ) ebf
and,
194.9qB
F CD = ------------------------------------- . (3.22)
kh ( p i p wf ) ebf
The relationship between (tDxf)ebf and FCD is presented graphically by Fig. 3.27. This relationship
can be approximated by the following equations:
2
6.94 10
F CD 5: ( t Dxf ) ebf = --------------------------
2
- (3.23)
F CD
10-1
10-2
(tDxf) ebf
10-3
10-4
10-5 -1
10 1 101 102
FCD
Fig. 3.26 - Dimensionless Time for the End of the Bilinear Flow Period vs. Dimensionless Fracture
Conductivity, Constant Rate Case.6
1 1.40
------------------ = ---------- . (3.25)
( q D ) ebf F CD
Therefore,
and,
197.7q ebf B
F CD = --------------------------------- . (3.27)
kh ( p i p wf )
The conventional analysis of bilinear flow data requires two plots - a log-log plot of the appropriate
rate or pressure function vs. t, and a cartesian plot of the appropriate rate or pressure function vs.
t1/4.
Liquid-Constant Rate
The following procedure can be used to analyze bilinear flow data for fracture conductivity and
fracture length when the production rate is constant:
3. If any data form a quarter slope in Step 2, plot pi-pwf vs. t1/4 on cartesian paper and identify the
data which form the bilinear flow straight line.
5. Using the slope, mbf, from Step 4, compute the fracture conductivity, kfw, using Eq. (3.10):
2
44.1qB
k f w = -------------------------------------
1/4
- . (3.28)
m bf h ( c t k )
It should be noted that this calculation can only be made if k is known from a prefrac test.
6. If the bilinear flow straight line ends and the data rise above the straight line, determine the
value of p, i.e., pebf, at which the line ends. Then, from Eq. (3.24), FCD can be computed as
194.9qB
F CD = ------------------------------------- . (3.24)
kh ( p i p wf ) ebf
with FCD known, the fracture length can be computed using Eq. (3.7):
kfw
x f = ------------- . (3.29)
kF CD
It should be noted that Eq. (3.24) assumes FCD 3. If enough data is available beyond bilinear
flow, a type curve match should be attempted to verify that this is true.
10-1
10-2
FCD = 5
(tDxf)ebf
10-3
10-4
10-5 -1
10 1 2.8 10 10-2
FCD
Fig. 3.27 - Dimensionless Time to the End of Bilinear Flow for Constant Pressure Production.9
Liquid-Constant Pressure
When formation face pressure remains constant during a test, the following procedure can be used
to analyze the bilinear flow data for fracture conductivity and fracture length:
3. If any data in Step 2 form a quarter slope, plot 1/q vs. t1/4 on cartesian paper and determine the
slope, mbf, of the bilinear flow straight line.
4. Using the slope, mbf, from Step 3, compute the fracture conductivity, kfw, using Eq. (3.14)
2
48.9B
k f w = --------------------------------------------------------------
1/4
- . (3.30)
m bf ( p i p wf )h ( c t h )
5. If the bilinear flow line ends and the data rise above the straight line, determine the value of q
where the line ends, i.e., qebf. Then, from Eq. (3.27), FCD can be computed as
197.7q ebf B
F CD = ------------------------------- . (3.27)
kh ( p i p wf )
With FCD known, the fracture length can be computed using Eq. (3.24):
kfw
x f = --------------- . (3.29)
k F CD
Eq. (3.27) assumes FCD 5 ;accordingly, if enough data are available beyond bilinear flow, a
type curve match should be attempted to verify that this is true.
When a flow restriction exists in the formation adjacent to the fracture, or when a restriction occurs
in the fracture near the wellbore, the ideal bilinear flow behavior discussed previously, shown by
Fig. 3.22 and Fig. 3.24 will be altered. Ideal bilinear flow results in a straight line on a cartesian
plot of p (constant rate) or 1/q (constant pressure) vs. t; further, this line passes through the origin.
Bilinear flow still exists when a flow restriction is present; however, the restriction causes an extra
pressure drop, ps, in the system. This additional pressure loss does not alter the slope, mbf, of the
bilinear flow straight line; instead, rather than passing through the origin, the line will have an
intercept equal to ps for the constant rate case. This behavior is depicted by Fig. 3.28.
DAMAGE OR
CHOKED FRACTURE
p, psi
{
ps IDEAL
0
0
t1/4, hours1/4
Fig. 3.28 - Effect of a Flow Restriction on Bilinear Flow, Constant Rate Case.
A log-log plot of p (constant rate) or 1/q (constant pressure) vs. t will exhibit a straight line with
quarter slope for ideal bilinear flow. The slope of this line will be altered, however, when a flow
restriction is present. This non-ideal behavior is depicted by Fig. 3.25 for the constant rate case.
DAMAGE OR
CHOKED FRACTURE
p, psi
SLOPE = 1/4
t, hrs
Fig. 3.29 - Effect of a Flow Restriction on the Log-log Plot for the Constant Rate Case.
Wellbore storage will alter or completely mask the bilinear flow straight lines ideally expected on
the cartesian and log-log plots of p or 1/q vs. t1/4 and p or 1/q vs. time, respectively. Fig. 3.30
depicts the effect of storage on a plot of p vs. t1/4 for the constant rate case. The corresponding
effect of storage on the log-log plot is shown in Fig. 3.31. It has been reported by Cinco-Ley et al.,6
that the end of wellbore storage effects occurs approximately three log cycles after the end of the
unit slope line.
p, psi
IDEAL BILINEAR
FLOW
EFFECT OF
WELLBORE STORAGE
t, hrs
Fig. 3.30 - Effect of Wellbore Storage on a Plot of p vs. t1/4 for the Constant Rate Case.
SLOPE = 1/4
UNIT SLOPE
p, psi
= 3 LOG CYCLES
t, hrs
Fig. 3.31 - Effect of Wellbore Storage on the Log-log Plot for the Constant Rate Case.
Dimensionless Pressure:
kh [ m ( p i ) m ( p wf ) ]
P D = -------------------------------------------------
- (3.31)
1424qT
Dimensionless Time:
0.0002637kt
t Dxf = -----------------------------
2
- (3.6)
c t x f
kfw
F CD = --------- (3.7)
kx f
2.45 t Dx 1/4
PD -f
= ---------------------
1\/2
(3.8)
F CD
444.6qT 1/4
m ( p i ) m ( p wf ) = -------------------------------------------------
1/2 1/4
-t (3.32)
h ( k f w ) ( c t k )
444.6qT
m bf = -------------------------------------------------
1/2 1/4
- (3.33)
h ( k f w ) ( c t k )
Constant Formation Face Pressure
Dimensionless Rate:
1424qT
q D = -------------------------------------------------- (3.34)
kh [ m ( p i ) m ( p wf ) ]
2.72 t 1/4
1 Dx f
------ = -----------------------
- (3.12)
qD F 1/2
CD
1 493.6T 1/4
--- = ---------------------------------------------------------------------------------------------
1/2 1/4
- t (3.35)
q h ( k f w ) ( c t k ) [ m ( p i ) m ( p wf ) ]
493.6T
m bf = ---------------------------------------------------------------------------------------------
1/2 1/4
- (3.36)
h ( k f w ) ( c t k ) [ m ( p i ) m ( p wf ) ]
NOTE: The equations presented in this section are written specifically for pressure drawdown
tests. These equations can be modified for pressure buildup tests by replacing the
pseudopressure differential, m(p), and the producing time, t, with appropriate values as
shown in the following table:
When the rate of a gas well is maintained constant, the pressure change at the formation face is
described by Eq. (3.32). This equation indicates that a plot of m(pi)-m(pwf) vs. t1/4 for drawdown
tests, or m(pws)-m(pwf) for buildup tests, will yield a straight line with slope, mbf, predicted by
Eq. (3.33). This plot described by Eq. (3.32) is illustrated by Fig. 3.24. When bilinear flow ends,
the straight line will end and the data will exhibit curvature which is concave upward or downward
depending upon the value of the dimensionless fracture conductivity, FCD. When FCD 1.6, the
curve will be concave downward, a value of FCD > 1.6 will cause the curve to be concave upward .
When FCD > 1.6, bilinear flow ends because the fracture tip begins to affect wellbore behavior. If
a pressure transient test is not run sufficiently long for bilinear flow to end when FCD > 1.6, it is
not possible to determine the length of the fracture. When FCD 1.6, bilinear flow in the reservoir
changes from predominately one-dimensional (linear) to a two-dimensional flow regime. In this
case, it is not possible to uniquely determine fracture length even if bilinear flow does end during
the test.
A more diagnostic plot to recognize bilinear flow is the log-log plot. From Eq. (3.32)
END OF
BILINEAR FLOW
t1/4, hours1/4
444.6qT 1
log [ m ( p i ) m ( p wf ) ] = log -------------------------------------------------
1/2 1/4
- + --- log t . (3.37)
h ( k f w ) ( c t k ) 4
Eq. (3.37) indicates that a log-log plot of m(pi)-m(pwf) vs. t will yield a straight line with a one-
fourth slope; this is illustrated by Fig. 3.35.
When formation face pressure remains constant, the formation face rate will change with time as
described by Eq. (3.35). According to Eq. (3.35), a plot of 1/q vs. t1/4 should yield a straight line
with slope, mbf, defined by Eq. (3.36) this graph is depicted by Fig. 3.24. Following the end of the
bilinear flow period, the curve for FCD 2.8 will be concave downward and the curve for FCD > 2.8
will be concave upward. The straight line for bilinear flow ends for the same reasons presented for
the constant rate case on page 3-41. Eq. (3.35) also indicates that a log-log plot of 1/q vs. t should
yield a straight line with a slope of one-fourth:
493.6T 1
log ( 1 of q ) = log ----------------------------------------------------------------------------------------
1/2 1/4
- + --- log t . (3.38)
h ( k f w ) ( c t k ) m ( p i ) m ( p wf ) 4
The log-log plot of pressure change vs. time, illustrated by Fig. 3.35, is the primary diagnostic tool
by which bilinear flow can be recognized.
SLOPE = 1/4
p, psi
t, hours
Fig. 3.33 - Log-log Plot Showing Effect of Ideal Bilinear Flow for the Constant Gas Rate Well.
SLOPE = mbf
p, psi
END OF
BILINEAR FLOW
t1/4, hours1/4
The relationship between (tDxf)ebf and FCD for constant formation face rate is depicted graphically
by Fig. 3.37. This relationship can be approximated as:
(3.17)
SLOPE = 1/4
p, psi
t, hours
Fig. 3.35 - Log-log Plot Illustrating the Effect of Ideal Bilinear Flow for the Constant Pressure Case.
1.53
1.6 < F CD < 3: ( t Dxf ) ebf 0.0205 ( F CD 1.5 ) (3.19)
For the case where FCD 3, the dimensionless pressure at the end of bilinear flow is
1.38
( p D ) ebf = ---------- . (3.39)
F CD
Therefore,
1.38
F CD = ------------------ (3.40)
( p D ) ebf
and,
1965.1qT
F CD = --------------------------------------------------------- . (3.41)
kh [ m ( p i ) m ( p wf ) ] ebf
The relationship between (tDxf)ebf and FCD for constant formation face pressure is presented graph-
ically by Fig. 3.37. This relationship can be approximated by the following equations:
10-1
10-2
(tDxf) ebf
10-3
10-4
10-5 -1
10 1 101 102
FCD
Fig. 3.36 - Dimensionless Time for the End of the Bilinear Flow Period vs. Dimensionless Fracture
Conductivity, Constant Formation Face Rate Case.6
2
6.94 10
F CD 5: ( t Dxf ) ebf --------------------------
2
- (3.23)
F CD
(3.24)
1 1.40
------------------ = ---------- . (3.25)
( q D ) ebf F CD
10-1
10-2
FCD = 5
(tDxf)ebf
10-3
10-4
10-5 -1
10 1 2.8 10 10-2
FCD
Fig. 3.37 - Dimensionless Time to the End of the Bilinear Flow for Constant Pressure Production.9
Therefore,
and
1988T q ebf
F CD = -------------------------------------------------- . (3.42)
kh [ m ( p i ) m ( p wf ) ]
The conventional analysis of bilinear flow data requires two plots - a log-log plot of the appropriate
rate or pressure function vs. t, and a cartesian plot of the appropriate rate or pressure function vs.
t1/4.
Gas-Constant Rate
The following procedure can be used to analyze bilinear flow data for fracture conductivity and
fracture length. When rate is constant:
3. If any data in Step 2 form a quarter-slope, plot m(pi)-m(pwf) vs. t1/4 on cartesian paper and iden-
tify the data which form the bilinear flow straight line.
5. Using the slope, mbf, from Step 4, compute the fracture conductivity, kfw, using Eq. (3.33):
2
444.6qT
k f w = -------------------------------------
1/4
- (3.43)
m bf h ( c t k )
It should be noted that this calculation can only be made if k is known from a prefrac test.
6. If the bilinear flow straight line ends and the data rise above the straight line, determine the
value of m(p), i.e., [m(p)]ebf, at which the line ends. Then, from Eq. (3.42), FCD can be com-
puted as
1965.1qT
F CD = --------------------------------------------------------- . (3.42)
kh [ m ( p i ) m ( p wf ) ] ebf
With FCD known, the fracture length can be computed using Eq. (3.7):
kfw
x f = ------------- . (3.29)
kF CD
It should be noted that Eq. (3.43) assumes FCD 3. If enough data is available beyond bilinear
flow, a type curve match should be attempted to verify that this is true.
Gas-Constant Pressure
When formation face pressure remains constant during a test, the following procedure can be used
to analyze the bilinear flow data for fracture conductivity and fracture length:
3. If any data in Step 2 form a quarter-slope, plot 1/q vs. t1/4 on cartesian paper and determine the
slope, mbf, of the bilinear flow straight line.
4. Using the slope, mbf, from Step 3, compute the fracture conductivity, kfw, using Eq. (3.38):
2
493.6T
k f w = ---------------------------------------------------------------------------------
1/4
(3.44)
m bf h ( c t k ) [ m ( p i ) m ( p wi ) ]
5. If the bilinear flow line ends and the data rise above the straight line, determine the value of q
where the line ends, i.e., qebf. Then, from Eq. (3.43), FCD can be computed as
1988T q ebf
F CD = -------------------------------------------------- . (3.42)
kh [ m ( p i ) m ( p wf ) ]
With FCD known, the fracture length can be computed using Eq. (3.29):
kfw
x f = ------------- . (3.29)
kF CD
Eq. (3.29) assumes FCD 5; accordingly, if enough data are available beyond bilinear flow, a
type curve match should be attempted to verify that this is true.
3.6 References
1. Smith, M. B.: Effect of Fracture Azimuth on Production With Application to the Wattenberg Gas Field, paper
SPE 8298 presented at the 1979 SPE Annual Technical Conference and Exhibition, Las Vegas, Sept. 23-26
2. Prats, M.: Effect of Vertical Fractures on Reservoir Behavior - Incompressible Fluid Case, SPEJ (June 1961)
105-18; Trans., AIME, 222.
3. Britt, L. K.: Optimized Oil Well Fracturing, Phase I Report, Amoco Production Company Report F84-P-23
(May 25, 1984).
4. Britt, L. K.: Optimized Oil Well Fracturing, Phase II Report, Analysis of the Effects of Fracturing on the Sec-
ondary Recovery Process; Amoco Production Company Report F85-P-7 (Jan. 24, 1985).
5. Bargas, C. L.: The Effects of Vertical Fractures on the Areal Sweep Efficiency and Relative Injectivity of Ad-
verse Mobility Ratio Displacements, Amoco Production Company Report F89-P-13 (Feb. 13, 1989).
6. Cinco-Ley, H. and Samaniego-V., F.: Transient Pressure Analysis for Fractured Wells, JPT (Sept. 1981) 1749-
66.
7. Cinco-Ley, H. and Samaniego-V., F.: Transient Pressure Analysis: Finite Conductivity Fracture Case vs. Dam-
aged Fracture Case; paper SPE 10179, presented at the 1981 Annual Technical Conference and Exhibition, San
Antonio, Oct. 5-7.
8. Cinco-Ley, H.: Evaluation of Hydraulic Fracturing by Transient Pressure Analysis Methods, paper SPE 10043,
presented at the 1982 SPE Intl. Petroleum Exhibition and Technology Symposium, Beijing, March 19-22.
9. Bennett, C. O., Reynolds, A. C., and Raghavan, R.: Performance of Finite-Conductivity, Vertically Fractured
Wells in Single-Layer Reservoirs, SPEFE (Aug. 1986) 399-412; Trans., AIME, 281.
10. Guppy, K. H., Cinco-Ley, H., and Ramey, H. J. Jr.: Pressure Buildup Analysis of Fractured Wells Producing at
High Flow Rates, JPT (Nov. 1982) 2656-66.
11 Rodiquez, F., Horne, R. N., and Cinco-Ley, H.: Partially Penetrating Vertical Fractures: Pressure Transient Be-
havior of Finite Conductivity Fracture, paper SPE 13057, presented at the 1984 SPE Annual Technical Confer-
ence and Exhibition, Houston, Sept. 16-19.
The following mechanical properties are of interest in fracturing: (1) Elastic Properties of the For-
mation (i.e., Modulus of Elasticity and Poissons Ratio), (2) Fracture Toughness, and (3) Hardness.
Rock strength plays only a small role in the fracturing process and is not included in the fracture
design calculations.
Based on this simplifying assumption, formation properties can be characterized by two elastic
constants, the modulus of elasticity (or Youngs modulus), E, given in psi or units of pressure, and
Poissons ratio (in honor of the great French mathematician), , a dimensionless number as its
name implies. The modulus characterizes how stiff the formation is and quantifies how easily a
core is deformed by an axial stress (tension or compression). Poissons ratio quantifies how a core
bulges (expands or contracts laterally) by an axial compression or tension and it characterizes
(together with E) the transmittal of horizontal pressure due to the overburden.
Fracture design is greatly affected by how much the formation opens for a given pressure inside a
fracture. Fracture width depends on both fracture dimensions and formation stiffness. Fracture
width is inversely proportional to the formation plane strain modulus, E , given by
E
E = ------------------
-. (4.1)
( 1 2 )
D
W ---- p (4.2)
E
where, for simplicitys sake, E was used instead of E . This is usually a good approximation since
a rough estimate for the Poissons ratio for most rocks is between 0.20 to 0.35. Therefore, E is
expected to be about 4 to 12% larger than E. Note that the theoretically expected values for are
between 0 and 0.5 while moduli could have a much greater variability, from a few hundred thou-
sand psi to over 10 million psi.
Both of the elastic constants of a formation can be measured in the laboratory using a single com-
pression test. This test gives the modulus and the Poissons ratio under quasi static conditions.
These static properties characterize rock behavior under slowly varying loading, such as the one
resulting from the hydraulic fracturing process. Different values for the elastic constants can be
inferred (using elasticity relations) from the travel times of the compressional and shear sonic
waves (e.g. sonic logs) under dynamic conditions. The differences between dynamic and static
elastic constants are primarily of practical significance for the modulus. Dynamic moduli may be
much larger than static moduli and some correlation is usually needed to infer the static moduli
needed for fracturing design; in some cases the static moduli are 50 to 75% of the dynamic moduli.
Fig. 4.1 shows the typical result of a compression test (in this case, Bedford limestone). A small
core plug is jacketed and subjected to a confining pressure (usually equal to the overburden minus
reservoir pressure) in the triaxial cell; it is then loaded axially to produce plots of axial, lateral, and
volumetric strain vs. axial stress in excess of the confining pressure (Effective Axial Stress). Both
the axial and lateral strains are quantities calculated from measuring the decrease of the core length
and the increase of the core diameter using strain transducers that are mounted on the core.
12000
SECANT
MODULUS
TANGENT (DO NOT
MODULUS USE)
10000
AXIAL ULTIMATE
LOADS
EFFECTIVE AXIAL STRESS, PSI
8000
CONFINING
PRESSURE
6000
4000
0
-2.0 -1.0 0.0 1.0 2.0 3.0 4.0 5.0
STRAIN *10-3
Fig. 4.1 - Typical Stress-Strain Curve for Brittle Rocks.
The axial strain represents the ratio of the core shortening (length decrease) over its original
length and is a dimensionless number which is plotted positive for a length decrease (contraction).
The lateral strain represents the ratio of the core bulging (diameter increase) over its original
diameter and is a dimensionless number which is plotted negative for core diameter increase
(expansion). The volumetric strain, also shown on Fig. 4.1, is a calculated quantity given by the
following algebraic sum
It represents the ratio of the volume change over the original volume of the core, and is plotted
positive for contraction.
By definition, the initial slope of the axial strain curve is the modulus of elasticity or Youngs mod-
ulus, E, in psi. It is also called a tangent modulus because it is the slope of the dashed line drawn
tangent to the stress-strain curve at the origin (Fig. 4.1). Since the compression test was performed
at a confining stress approximating the in-situ conditions, this is the value that should be used for
modulus in a fracture design. By this, we mean that the formation in-situ is at a state of stress com-
parable to the one near the origin of the plot; any loading due to fracturing would make the stress
state go up or down on the curves near the origin. However, modulus depends on the confining
pressure, and some judgement should be exercised when data for the specific in-situ conditions are
not available. Modulus data should be used with a good understanding of what the testing condi-
tions represent because some labs draw, for example, a straight line from the origin to the point of
failure and report the slope of that line as modulus. This value is called, the Secant Modulus of
Elasticity and should not be used for fracture design calculations.
Poissons ratio, , represents the ratio of the lateral strain over the axial strain, both taken from the
linear behavior of the core near the origin, (i.e., over the range that the modulus straight line is
determined).
For the Bedford Limestone example in Fig. 4.2, at an effective axial stress of 4,000 psi the lateral
strain is -0.25 x 10-3 and the axial strain is 0.9 x 10-3. The Poissons ratio from Eq. (4.4) is 0.277.
Poissons Ratio quantifies the tendency of the material to bulge out for a given axial strain and
therefore how the material pushes laterally when it is subjected to an overburden pressure. The
theoretical range of Poissons ratio for uniform materials is between 0 and 0.5. Rocks which have
a competent structure (i.e. rocks with porosity that does not change significantly with loading) are
expected to have Poissons ratios in the same range. Good approximate values for Poissons ratio
for fracture width calculations are 0.25 for sandstone formations and 0.33 for carbonate forma-
tions.
However, Poissons ratio strongly affects how the closure stress is related to overburden pressure.
For example, a formation with 0 will develop almost no horizontal closure stress when sub-
jected to overburden; in contrast, a formation with 0.5 will develop a horizontal closure stress
almost equal to overburden, and will behave like a liquid! Real rocks fall somewhere between
those values, with the more ductile and plastic rocks having a higher Poissons ratio. Note that
rocks that have high porosity and low cementation (e.g. Valhall chalk) may have a close to zero.
This is because their porosity changes considerably with loading, and the bulging of the core is
accommodated by porosity reduction.
Example Data
H = 100 ft Fluid Loss H = 100 ft
C = .001 Spurt = 0
Q = 20 bpm
Viscosity = 100 cp (n' = 1)
Design Penetration (1/2 Length) = 500 ft
600 15
400 10
200 5
2 4 6 8
Youngs Modulus (106 psi)
10
Young's Modulus (million psi)
4
4 High Porosity
High Porosity (> 25%), Coarse Grained
2
2
5 10
5 10
Overburden-Pore Pressure (1000 psi)
Overburden-Pore Pressure (1000 psi)
Fig. 4.3 - Modulus of Elasticity for Sand- Fig. 4.4 - Modulus of Elasticity for Carbon-
stones. ates.
Also, the modulus values in Fig. 4.3 and Fig. 4.4 are for small samples. Many carbonate forma-
tions are naturally fractured; and in such a case, the modulus for the bulk in-situ rock would be
lower than a value for a small sample.
A similar chart for shales is not practical since Youngs Modulus for shales can vary from 500,000
psi for a high porosity, clay rich, shale to 6-8 million psi for a quartz cemented siltstone. If no core
is available for shales, sonic logs have been used to predict the modulus of the shales relative to
the modulus of the pay formation where core is available and modulus has been measured.
Table 4.1 lists typical modulus values for two special formation types.
Table 4.1 - Typical Modulus Values for Two Special Formation Types.
Modulus
Formation Porosity (106 psi)
Fig. 4.5 is a plot that allows the use of conventional Sonic Log data (compressional wave) to esti-
mate modulus. This dynamic modulus (i.e., estimated from correlation based on compressional
wave velocity in the formation) is greater than the static modulus needed for fracture design,
but, if laboratory tests are not available, the dynamic modulus sets an upper bound for modulus
and is preferable to Fig. 4.3 and Fig. 4.4. It can also be used to estimate the modulus in formations
where core is not available if lab data is available from other formations in the same well.
A better technique than conventional Sonic Logs is to calculate Youngs Modulus, E, from
Long Spaced Sonic Log data, using the compressional and shear wave velocities of the formation.
80
60
Sand
Dolomite
Lime
40
2 4 6 8 10 12 14 16
Ex 106 - psi
The best solution is to obtain core samples and have tangent modulus measured in a lab. If this is
impossible and E must be estimated, try to estimate on the high side. This will result in a design
with a narrower fracture width, higher net pressure and greater fracture height than should actually
occur, providing a conservative safe approach to fracture design.
Without getting too deep into theory, the fracture toughness concept comes from Griffiths1 work
on the fracture of brittle solids. The fracture toughness of a material represents its natural ability
to resist the propagation of a fracture. To quote an article by Srawley and Brown,2 In the simplest
terms, the fracture toughness of a material determines how big a crack the material is able to tol-
erate without fracturing when loaded to a level approaching that at which it would fail by excessive
plastic deformation. Fracture toughness can be quantified by lab experiments (such as the three
point loading of the Chevron notch) from which the loading vs. deformation curve is plotted until
failure, and the energy spent to fracture the specimen can be calculated from this diagram. It may
be noted that loading capacity of a specific specimen depends not only on crack size, but also on
crack shape, bulk of the specimen, crack orientation with respect to layering of material (e.g. for-
mation), temperature, rate of loading, etc. For this reason, it is very difficult to extrapolate labora-
tory results to the field, and an indirect assessment of apparent fracture toughness is done in the
field from treating pressure behavior using fracturing simulators, as described below.
The fracture toughness is quantified by either of two related parameters: (1) the critical strain
energy release rate, G, expressed in energy per area of created fracture (not the area of the fracture
faces) in units of force/length; and (2) the critical stress intensity factor, Kc, expressed in units of
pressure times square root of length. The relation between the two parameters for hydraulic frac-
turing problems (plane strain problems) is
2
G = Kc /E'. (4.5)
Typical laboratory range of Kc values are given by Thiercelin3 in Table 4.2. From Table 4.2 we see
that typical laboratory Kcs are of the order of 900 to 2000 psi in with a value of about 1500
psi in being a good rough estimate. A corresponding rough estimate of fracture energy is about 1
psi-in. Note that some simulators require Kc and some require G as input.
Fracture toughness relates the pressure required to propagate a fracture with the dimensions of the
fracture. Let us consider an example from the Wattenberg field,4 where fractures in the Muddy J
formation are highly confined by shale layers above and below the pay. Stress tests, minifrac and
fracturing treatments in the example well show that a fracture height of 90 ft is representative for
these type of calculations. Furthermore, net pressures, PN, on the order of 400 to 550 psi for mini-
frac treatments and 2100 psi for the main fracture treatments are typical. These observations indi-
cate the magnitudes of the formation toughness (i.e., critical stress intensity factor Kc), the
confining stress contrast c between layers, and other rock mechanics considerations. Consider-
Confining
Youngs Modulus Pressure KIc
ing the lateral propagation of the fracture tip of this highly confined fracture gives estimates of the
Muddy J pay toughness, or, better, its apparent toughness.
The fracture tip is essentially a penny shaped fracture that is subjected to the net treating pressure
PN. There is no stress contrast confining the fracture in the horizontal direction. Therefore, fracture
toughness is expected to be a dominant confining mechanism in the horizontal direction. From
fracturing mechanics,5 the stress intensity factor, K, in the opening mode of a penny shaped crack
under uniform pressure is given by
R
K = 2 P N --- (penny crack) (4.6)
where R is the radius and PN the uniform net pressure. The fracture propagates when K is equal to
the formation fracture toughness, Kc (which is a material property), and remains stationary when
K < Kc.
The fracture tip geometry of the Wattenberg fractures is characterized by R = 45 ft = 540 in and PN
= 500 psi. This value of net pressure is estimated from the minifrac treatment which does not have
the additional friction due to a proppant. With these values, Eq. (4.6) gives Kc = 13110 psi in . This
estimate is approximately 10 times greater than the fracture toughness of rocks measured in the lab
which have a typical toughness value of 1000 to 1500 psi in . Note that this discrepancy is a com-
mon phenomenon and consequently the calculated Kc is called an apparent formation toughness.
Several near fracture tip hypotheses contribute to an increased net fracturing pressure and could
contribute to an increased Kc. The most popular within the research community are (1) formation
plasticity, (2) non-penetrated (dry) zone near the tip, and (3) process zone (microfracture zone)
around the tip. Hypotheses (1) and (3) contribute to increased energy expenditure near the tip due
to plastic flow and intense microfracturing. Hypothesis (2) assumes a region where the hydraulic
pressure is not easily transmitted to the fracture tip due to asperities, gel plugging, increased gel
viscosity due to dehydration, and great frictional losses within very narrow crack opening. For all
the above reasons, it is quite common to input increased fracture toughness in the hydraulic frac-
turing simulators to match treating pressures and predict fracturing geometry.
4.3 Hardness
Rock hardness is important to fracture conductivity. The proppant may imbed into soft rocks, caus-
ing the fracture conductivity to decrease and the propped fracture to lose its effectiveness.
For most rock types, this is not a problem if a nominal design guideline of one pound of proppant
per square foot of fracture is achieved. For very soft formations (chalks are one example as seen
in Fig. 4.6), this is not sufficient and special fracture designs are required. If proppant embedment
is suspected due to productivity declines or pressure transient tests showing a loss of fracture
capacity with time, special lab tests are available to test core samples with various amounts and
types of proppant.
TEMP = 200F
2X2 DANIAN CHALK
1000
TOTAL CORE PERMEABILITY MD
100
Legend
0.4" PROPPED FRAC
10
0.25" PROPPED FRAC
0.1" PROPPED FRAC
MATRIX FLOW
1
10096-97
0.1
0 1000 2000 3000 4000 5000 6000 7000 8000 9000
4.4 References
1. Griffith, A. A.: The Phenomena of Rupture and Flow in Solids, Phil. Trans., Royal Soc. of London (1920) Ser.
A, 221, 163-98.
2. Srawley, John E., and Brown, William F., Jr.: Fracture Toughness Testing Methods, Fracture Toughness Test-
ing and Its Applications Symposium, 1964 Annual Meeting of ASTM, Chicago, June 21-26.
3. Thiercelin, M.: Fracture Toughness Under Confining Pressure Using the Modified Ring Test, Proceedings of
the 1987 US Symposium of Rock Mechanics, 149-56, June 29-July 1.
4. Moschovidis, Z.A., Broacha, E., and Gardner, D.: APR, Tectonic Correction of Closure Stress Profiles and Field
Data Analysis for Fracture Design for Wattenberg Gas Field, Colorado; Amoco Production Company Report
F91-P-59 (Nov. 1990).
5. Warpinski, N. R., and Smith, M. B.: Rock Mechanics and Fracture Geometry, Monograph Series, SPE, Rich-
ardson, TX (1989) 12, vi, 57-80.
Q tp
L = ----------------------------------------------
- (5.1)
3 C Hp tp + w H
where fracture height, H, and fluid loss height, Hp, appear in the denominator and have a great
effect on fracture length.
H is the total or gross fracture height which, of course, changes with time during a treatment. A
reasonable estimate of the initial fracture height, and of the variables governing height growth
is critical to an accurate solution for fracture length since, as seen in the relation above, length, L,
and height, H, are inversely proportional. It is usually desirable to maintain frac height within a
reasonable distance above and below the pay zone, to minimize useless fracture area (created
and propped fracture area which will not contribute to production) or to avoid fracturing into water
bearing layers. The fracture height obtained is largely controlled by formation properties. We have
some influence over the height obtained through controls on pump rate and fluid viscosity, but
must recognize the limits to which we can control height development.
Numerous oil field techniques and wellbore arrangements have been proposed in the past for lim-
iting fracture height:
Perforate a limited section and only frac where the perfs are
Perforate low in the wellbore, since everybody knows that you cannot frac below Total Depth
(TD)
Perforate high in the wellbore, so that you do not frac into water below.
Pump at high rate so that you get the job pumped before the fracture has a chance to grow out
of the zone.
Though these approaches may sound silly, we have all probably tried to use these or others in some
form or fashion. Some of them have limited application and may exert some influence over the
ultimate frac height obtained, but overall, they have minimal impact on frac height. Fig. 5.1 shows
a schematic of a fracture which basically grows where it wants to. The only wellbore condition
that can have a significant impact on frac height is the cement bond. A poor cement bond can
allow annular communication with another zone, and thus bypass a potential confining bed. Pump
rate and fluid viscosity do affect frac height through their indirect control on pressure, but to a very
small degree when compared to formation properties.
Fracture Height = ?
Pay Design
Water Actual
Vertical fracture growth and resulting fracture height is controlled by the interaction of hydraulic
pressure inside the fracture with mechanical properties of the rocks and in-situ stresses. The dom-
inant factors controlling frac height are listed below in order of decreasing importance.
Fracture pressure from high modulus (naturally high/low closure stress, etc.)
Ductility of bounding bed - may facilitate bedding plane slip, (small coal seams)
The most dominant controlling mechanism for frac height is vertical variations in closure stress
through strata of varying lithology and rock properties.1 Closure stress is the minimum, compres-
sive, in-situ stress. Pressure in the fracture must exceed this before a hydraulic fracture can open.
Fig. 5.2 shows a simplistic, idealized case of three zones of different stress. In this case, the bound-
ing beds (Zones 2 and 3) are assumed to be of infinite thickness and have the same closure stress.
The stress in the bounding beds is greater than that in the pay zone (Zone 1). Zone 1 is perforated
and a fracture is initiated. The fracture grows unrestricted to the height of Zone 1. At this point, the
relationship shown in Fig. 5.2 goes to work (Point A).
As injection continues, the fracture begins elongating and extending laterally from the wellbore.
Net fracture pressure, Pn, (bottomhole treating pressure outside the perforations minus the forma-
tion closure pressure, discussed in more detail in Chap. 8), begins to increase as the fracture
extends. During this period, the fracture is essentially acting as a pipeline carrying high viscosity
fluid from the wellbore to the fracture tip. As the pipeline grows longer, the pressure at the well-
bore must increase to overcome the increased friction drop along the ever lengthening fracture.
As net pressure, Pn, increases, the ratio of net pressure to the closure stress differential between the
pay zone and bounding beds begins to increase, moving one up the curve (Point B). When net
pressure has increased to about 50% of the stress differential between Zone 1 and Zones 2 and 3,
fracture height has increased to about 135% of the initial frac height (Zone 1). As net pressure in
the fracture increases, frac height continues to grow, until the frac height is twice the initial height
at a net pressure equal to 70% of the stress differential (Point D). The thickness of Zone 1 and the
absolute values of the stresses are independent of this relationship for a three zone system with infi-
nite bounding beds. Obviously, after net pressure reaches 70-80% of the stress differential between
the pay zone and bounding beds, small increases in net pressure (the net pressure to stress differ-
ence ratio) can add much additional frac height. The fracture height cannot be contained, and the
fracture grows uncontrollably out of zone. Note, however, that after this point is reached, fracture
length growth does not stop though it is slowed considerably. Thus, if no danger exists of the frac-
ture breaking into another (possibly undesirable) low stress zone - pumping may safely continue
in order to create a longer fracture. The economics of creating this additional fracture length will
be affected though, with significantly greater treatment volumes now being needed to create addi-
a 1.0
b
Gamma Ray Closure Stress
c
-------
Pn
0.8
c2
D
Shale Zone 2 C
0.6
Stress Difference
Sand c1 Hi Zone 1
0.4
0.2
Shale
c2 Zone 3 A
0
0 1 2 3 4 5
Ratio Frac Height:
c = c2 c1 Initial Frac Height ( ----
H
H
)
i
c d
Pressure
C
B
D
A B C D
Time
tional length. Similarly, sand is distributed over a greater and greater height, reducing the sand con-
centration per unit area.
This means that if we are to contain a fracture within zone, we must have some idea of closure
stress in the pay zone and bounding beds. If stress differences are only 700-800 psi, then we can
expect the fracture to grow uncontrollably out of zone at about 500-600 psi net fracture pressure.
Fracture treatments could be designed to stay within this net pressure limitation. On the other hand,
it may be difficult to achieve the length desired at these net pressures (since net pressure depends
on fracture length), and the treatment would have to be designed with this fracture height growth
in mind. Conversely, if the stress differential is on the order of 1500 psi, net pressure can be
allowed to rise to 1000-1100 psi before the fracture will begin to grow significantly out of zone.
This would allow an ample pressure limitation for designing most fracture treatments.
Obviously, an in-situ fracture closure stress profile, as seen in Fig. 5.3, is the major input data for
3-D or Pseudo 3-D fracture treatment design. The example in Fig. 5.3 illustrates a stress profile
generated by conducting multiple small volume, microfrac stress tests. Generally, such multiple
stress data are not available and some form of log-stress correlation will be required. However,
this example illustrates another important item - namely typical (or maximum) values for
in-situ stress differences. Consider data from the sandstone at 7500 ft showing a fracture closure
pressure (closure stress) of 6500 psi. Then consider the stress of 8000 psi at a depth of about
7650 ft in the Mancos Tongue Shale. This stress difference of 1500 psi at this depth represents a
stress difference of 0.2 psi/ft - and this is about the maximum stress difference which has been
recorded, verified, and published. Thus, assuming some lithology differences exist, an optimistic
estimate for in-situ stress differences might be:
8000
6000
9000
7000
00.0
50.0
100.0
150.0
200.0
0.3
0.2
0.1
0.0
0.1
m
ft
7300
PALUDAL
(2225m)
2250
7400 Estimated over-
(2255m) COAL burden stress
(1.05 psi/ft)
7500
(2286m)
ROLLINS
2300
7600
(2315m) SILT
7700 2350
MANCOS
TONGUE
(2347m)
SHALE
7800
(2377m)
2400
TONGUE COZZETTE
SAND
7900
(2408m)
8000
(2438m)
MANCOS
2450
8100
(2459m)
45 50 55 60 MPa
Fig. 5.3 - Variations in Fracture Closure Stress in a Sand/Shale Sequence.
The effects of lithology on in-situ stress (fracture closure stresses or closure pressure) along with
the effect of closure stress variations on fracture geometry may also be seen in Fig. 5.4, a set of
field data presented by Esso Canada.2 Fig. 5.4 compares two cases (within the same wellbore)
showing measured in-situ stresses along with pre and postfrac radioactivity logs for fracture height
growth. For Case 1, several stress tests (microfrac type stress tests) were conducted in zones with
(based on differing gamma ray readings) varying lithology. This stress data showed basically a
0.7 psi/ft (e.g. normal) stress gradient - and the postfrac logs suggest massive height growth
outof-zone. Case 2 shows stress data collected from two zones, both of which were perforated, and
a propped fracture treatment was conducted attempting to stimulate the two zones simultaneously.
The upper zone shows a significantly higher closure stress (associated with a different lithology)
and the postfrac logs indicate that the entire treatment entered the deeper, lower stress zone.
Thus we see examples - in the same wellbore - of lithology changes with and without associated
differences in fracture closure pressure. A guideline for interpreting stress profiles where no other
information exists might be:
There must be some change in lithology in order to expect some variations in closure pressure
- and thus some degree of fracture height confinement. However, do not try to quantify
lithology logs. That is, relatively minor apparent lithology changes could signify significant
stress differences, OR a major lithology change might have no associated stress differences.
As discussed in Chap. 4, the one exception to this would be for stress changes created by artificial
changes in reservoir pressure (e.g. depletion).
In addition to the stress difference in the beds, bed thickness is important. If the bounding beds are
not infinitely thick, then we must consider their thickness to determine if the fracture might grow
completely through the bounding beds and into zones of lower stress. A 2 ft shale bounding a 10
ft pay zone is obviously not going to stop a fracture from growing out of zone, nor will a 20 ft shale
bounding a 50 ft zone. A good rule for beds immediately bounding a zone to be fractured, is that
they should be at least as thick as the zone being stimulated to confine frac height; the basis
for this rule-of-thumb is discussed under Picking Fracture Height on page 5-12.
Consider the Pressure-Height Curve as seen in Fig. 5.2b. At the point where the fracture has tri-
pled in height (e.g., H/Hi = 1 and the fracture has grown upwards a distance equal to one initial
height and downwards one initial height), net pressure has reached 80 % of the in-situ stress
difference. Also at this point, pressure-height behavior is fairly flat, that is, relatively large
amounts of height growth begin to occur for small increases in bottomhole treating pressure. Thus,
even for infinite bounding beds, fracture height will begin to increase rapidly after an upward
or downward growth about equal to one original formation thickness.
Case 1
Apparent Lithology No Stress Difference
Collar
Locations
2675
2700
Depth (meters)
Base
Parts
Gamma Ray
0.7 psi/ft
gradient
2725
Post-Frac
7000
Gamma Ray 6000
2750
Increasing Gamma Activity Closure Stress (psi)
Case 2
Large Stress Differences
No FRAC in High Stress Interval
2060 Collar
Locations
Lower Zone Upper Zone
Perfs
Depth (meters)
2060
5000
4000
Perfs
Base Post-Frac GR
GR
2060
Increasing Gamma Activity Closure Stress (psi)
Fig. 5.4 - Examples of Lithology Changes, With and Without Associated Stress Differ-
ences.
The first important effect of bed thickness then is the thickness of bounding formations as illus-
trated by the four drawings in Fig. 5.5. This figure repeats the three-layer behavior discussed
above until point C is reached - e.g. the fracture has approximately tripled in height and the top
of the fracture has just reached the top of the barrier formation. At that point in time, the treating
pressure inside the fracture, near the wellbore, is considerably greater than the pressure needed to
propagate a fracture into the shallower low stress zone. Thus treating pressure will begin to drop
(sometimes fairly rapidly) as the fracture preferentially migrates into this new formation.
This can, in extreme cases, even lead to the main fracture beginning to grow shorter and can have
major (usually undesirable) effects on the ability to pump proppant, proppant placement, and on
stimulation effectiveness. Some of the treatment pumping problems which can arise from such
height growth behavior are discussed in Chap. 8. Also, some of the fracture modeling/fracture
design issues raised by such a fracture geometry are briefly discussed below.
The second major importance of bed thickness is thickness of the pay zone itself. The net pressure
which the stress and thickness of the bounding beds must counteract depends on the thickness of
the pay zone. Fig. 5.6 illustrates the net pressure required to create a 500 ft fracture for several pay
zone thicknesses. This figure shows that height growth would probably not be expected to be con-
fined to a 20 ft zone at 2000 psi, but height confinement could be expected for a 200 ft zone at 200
psi. While the actual net pressures tabulated in Fig. 5.6 are for a specific case, the figure can also
be used, in a general, qualitative, sense to estimate the potential for height confinement for partic-
ular zones.
Fig. 5.6 - Net Pressure Required to Create a 500 ft (1/2 Length) Fracture.
The actual net pressures tabulated in Fig. 5.6 are for a specific case. However, they might also be
viewed as typical values of net treating pressure for various gross zone thicknesses. Thus, if a
formation being considered for fracturing has a gross thickness on the order of 30 ft - then net treat-
ing pressure will probably be 1500 psi, and stress differences on the order of 1600 psi will be
needed to give reasonable height confinement. Assuming a formation depth of 6000 ft, the required
gradient of stress difference would be 0.27 psi/ft - good height confinement is unlikely and
extensive height growth would be expected. On the other hand, a typical net pressure for fracturing
a zone with a gross thickness of 60 ft might be on the order of 800 psi - with stress differences of
900 psi needed for reasonable height confinement. For a formation depth of 8000 ft, the required
gradient difference is only 0.1 psi/ft - and assuming some lithology differences exist - then fairly
good height confinement may be a reasonable possibility.
Probably the most important of the remaining variables which affects frac height (after the stress
and pressure behavior), are modulus contrasts (Fig. 5.7), and bedding plane slip (Fig. 5.8 and
Fig. 5.9).
Though not as strong a barrier as once thought, bounding beds with higher modulus than the pay
zone can retard height growth by causing fracture width in the bounding formations to be very nar-
row. However, as seen in Fig. 5.7, the maximum possible L to H ratios are fairly small - that is the
height confining effect of modulus contrasts is actually quite minimal.
For shallow depths, overpressured formations, or highly jointed formations such as coals, slip may
occur along bedding planes at the top or bottom tip of the fracture, Fig. 5.8, blunting the fracture
and arresting height growth. This would be a very strong barrier; however, it probably does not
occur often in oil and gas well fracturing except possibly at the interfaces with coal seams. Slip of
this type would be required for the Geerstma de Klerk model to be applicable for fractures with
lengths greater than their height (L/H > 1).
Fig. 5.9 presents the results of a series of lab tests conducted to determine the likelihood of a
hydraulic fracture stopping at an unbonded interface between two rock layers. As seen from these
Fig. 5.8 - Illustration of Fracture Interface Fig. 5.9 Interface Slip vs. Stress.
Slip.
results, for an effective vertical stress across the interface (e.g. overburden weight minus pore pres-
sure) of only 1000 psi the fracture crossed the interface for almost all rock types. Since an effec-
tive vertical stress of this magnitude would correspond to a depth of only about 2000 ft - it is clear
that interface slip will not be an effective barrier to vertical frac height growth for most oil and gas
well situations.
Fracture closure pressure or closure stress generally increases with depth, with a typical gradient
of 0.7 psi/ft - e.g. for each 100 ft increase in depth, closure pressure will increase by 70 psi. This
increase in closure stress is generally greater than the increase (with depth) in fluid pressure inside
the fracture due to the hydrostatic gradient of the fluid. As an example, consider a fracture 200 ft
in height which is filled with a water based fluid. Closure stress at the bottom of the fracture is
greater by about 140 psi than closure stress at the top; at the same time the driving fluid pressure
at the bottom is greater by 86 psi (assuming a hydrostatic gradient of 0.43 psi/ft for water). Thus
net pressure (e.g. driving fluid pressure minus closure pressure) is about 54 psi less at the bottom
of the fracture than at the top. Thus the fracture would have some tendency to grow upward rather
than downward.
However, for many (most?) fracturing cases net pressure may have a typical value on the order of
500 to 1000 psi - thus a difference (over the height) of 50 psi in net pressure is relatively insig-
nificant. Stress gradients, then, only become significant in affecting fracture height growth for
cases where significant height already exists (e.g. several hundred feet), or for cases of very low
net pressure (e.g. typically associated with low modulus formations and/or the pumping of very
low viscosity fluids).
Obviously, normal strata are not as simple as the idealized case described in Fig. 5.10, but the prin-
ciples are still applicable. If the bounding beds are not infinitely thick, then we must ensure that
they are of adequate thickness so the fracture does not grow completely through them and into a
zone of lower stress. A 2 ft shale bounding a 10 ft pay zone is obviously not going to stop a fracture
from growing out of zone. As discussed on page 5-6, a good rule for beds immediately bounding
a zone to be fractured is that they must be at least as thick as the zone being treated. Still, there will
be some height growth into the bounding layers with the final magnitude of fracture height being
predominantly determined by the stress difference between the pay and the bounding forma-
tions. Thus predicting or picking fracture height becomes an exercise in estimating (or measuring)
the in-situ closure stress for various zones.
There are tools which may, under some conditions, possibly aid in determining the in-situ stress
profile. However, in general, consideration of two dominant parameters will aid in constructing
reasonable estimates of in-situ stresses.
Lithology Changes
Pore Pressure
One minimum consideration for height confinement is significant lithology changes as seen with
a Gamma Ray log. Shales often have higher closure stresses than clean sands so thick boundary
shales can confine fractures. Such confinement is not always the case, but the lack of lithology
changes virtually ensures unrestricted height growth or radially shaped fractures. Thus a change in
lithology makes it possible for stress differences to exist. However, one should not try to quan-
tify a Gamma Ray log, e.g., if a lithology difference exists, then stress differences may exist and
fracturing pressure analysis (as discussed in Chap. 8) must be used to determine the magnitude of
the stress differences.
As discussed, closure stress is related to reservoir pressure. Therefore, a reservoir that has been
drawn down, as in a producing well, is likely to have a lower closure stress than normal in the pay
zone, and consequently a higher stress differential between pay and the bounding beds, improving
chances for height confinement. On the other hand, height confinement could be more difficult to
achieve in an injection well due to pressuring up of the pay zone. Thus pore pressure and pore pres-
sure differences between zones (e.g. due to partial depletion from offset production) is a major fac-
tor to consider in estimating in-situ stresses. Fracture closure stress is generally related to pore
pressure by3
c = ------------ ( OB p ) + p (5.2)
1
Inspection of Eq. (5.2) for a typical sandstone reservoir with a Poissons ratio, of 0.25 indi-
cates that for every psi change in reservoir pressure there is a corresponding 2/3 of a psi change in
closure pressure. Thus a depletion of 1500 psi in a sandstone will typically cause a reservoir clo-
sure pressure to decrease by about 1000 psi. Since there should presumably be no pore pressure
reduction in the surrounding impermeable shales, this 1000 psi decrease in the pay zone closure
pressure would be added to any naturally existing stress differences and very good height con-
finement can exist in depleted formations. Further inspection of Eq. (5.2) for a typical carbonate
reservoir would show a 1/2 psi change in closure pressure for every psi change in reservoir pres-
sure.
Special logs have been developed and marketed which may, sometimes be of value in determining
the in-situ stress profile (see Chap. 10). However, these logs are based on simple, elasticity
assumptions and should be treated with extreme caution. For sand/shale sequence geology, there
is often some relative truth in the logs and the actual stresses can frequently be successfully cal-
ibrated against the log derived stress values. Carbonate geology tends to be more complex and
the value of the logs is more questionable. In either case, however, the raw information from the
logs should never be used. If test procedures are not planned in order to calibrate the logs - then
the logs should not be run.
A measured/log stress correlation can be based on stresses actually measured in several zones in
the wellbore using closure stress tests as described in Chap. 8. This technique is the only one which
provides quantitative, in-situ data by which to determine the potential for height confinement, but
requires perforating and testing multiple zones in the well. If a number of fracture stimulations are
to be performed in the field, and height confinement is questionable and critical to the outcome of
the stimulations, then such testing coupled with running sonic logs as described above may be war-
ranted. Alternatively, a 3-D type fracture simulator may be used to infer the in-situ stresses through
history matching actual bottomhole treating pressure (BHTP) data. This is also discussed in
Chap. 8.
In summary, fracture height is the most critical variable to successful fracture treatment design,
and yet is one of the most difficult variables to measure. The three variables most strongly affecting
the ultimate frac height achieved during a treatment are: (1) closure stress differentials, (2) thick-
ness of the bounding beds, and (3) net fracture pressure. Several techniques exist by which to better
quantify frac height, involving everything from qualitative guesses to detailed quantitative mea-
surement. Finally, there is no substitute for experience in an area for picking fracture height or esti-
mating the in-situ closure stress profile, but whether an established field or a wildcat, there is
plenty of room for sound, engineering judgments.
Since fracture height and fracture height growth are the dominant variables affecting successful
propped fracture treatment design, fracture models which can account for height growth become
powerful, even indispensable, tools for modern job design or analysis. This is true in spite of the
common statement - We never have the data required to really use such fracture models. In fact,
one must realize that, in reality, 2-D fracture models are much harder to accurately use since there
is never, under any conditions, any way of accurately estimating fracture height in advance.
However, we can make reasonable estimates for the in-situ stress distribution. Also, since in many
cases the bottomhole pressure during a treatment is a strong function of the in-situ stresses and the
stress profile, we can use pressure data along with 3-D models to verify or modify these estimates,
finally arriving at a reasonably accurate description of the formation(s). This is most efficiently
done via a pressure history matching procedure as discussed in Chap. 8.
It is important to realize, however, that there are two types of 3-D fracture simulators.
Fully (or true) 3-D models calculate fracture width and fracture propagation at every point as a
function of the fluid pressure distribution everywhere inside the fracture. Among other things, this
ensures that the fully 2-dimensional flow field inside the fracture is used in calculating fluid pres-
sure and fracture width at each point. Models such as this are powerful tools and can be used for
analyzing quite complex geologic settings and complicated fracture geometry. Such models also
require extensive computer resources and are not usable for any type of routine well completion
designs. TerraFrac is one commercial fracture simulator of this type and this model is available in
Amoco. The TerraFrac model is discussed and some of its capabilities are briefly described in
Section 10.2 of this manual. Also, a few different fracture geometry cases are briefly reviewed
below along with some notes as to which geometry types require such fully 3-D modeling.
A more common, and usable, type of fracture simulator has been termed a pseudo 3-D model. Such
models are made more usable (in terms of time and required computer resources) through sev-
eral simplifying assumptions including:
1. The fracture length is at least equal to the fracture height, though through analytical approxi-
mations, such models can also give at least rough estimates of fracture geometry where vertical
height growth may be somewhat greater than fracture length.
2. Fracture height growth at any point along the length of the fracture is related only to the net
pressure at that point. Also, fracture width (and the vertical fracture width profile) at any point
along the fracture length is assumed to be related only to the net pressure at that point.
3. The greatest fracture penetration is occurring in the zone where the fracture initiates. Even with
these simplifying assumptions, however, pseudo 3-D models have proven in field practice and
through comparison with fully 3-D models, capable of handling many realistic and common
cases.
Schematically, a pseudo 3-D type fracture model proceeds as pictured in Fig. 5.12. Fracture length
propagation is calculated using calculations and assumptions similar to the traditional, 2-D, Per-
kins & Kern (PKN) fracture geometry. Along the fracture length the fracture is broken into indi-
vidual segments or cells, and the vertical fracture height growth for each cell is calculated as if
this cell represented a single Geertsma de Klerk (GDK) fracture geometry. As mentioned above,
the fracture width and width profile along with the height growth for each cell is assumed to be
related solely to the net pressure in that particular fracture segment or cell.
Geertsma deKlerk
Solution
While pseudo 3-D models are good, usable tools, it is important to realize that limitations do exist
and to recognize when the use of more sophisticated models is necessary. Fig. 5.13 illustrates sev-
eral possible fracture geometries and briefly comments on the applicability of Pseudo 3-D mod-
els to each case.
Radial Frac
Ideal P3D Geometry
OK for P3D Modeling
Stress
Profile
Just as it is difficult to pick a fracture height or to estimate the stress profile controlling height
growth, it is also difficult to measure fracture height after a job. However, several tools are avail-
able and these should be employed whenever possible to allow post-job evaluation and to improve
future jobs. The primary techniques for measuring height include temperature and Gamma Ray
logs (GR log); when conditions allow, an open hole completion; and, when the situation warrants
it, downhole televiewer logging. Procedures involved in running these logs are discussed in
Section 10.1 of this manual.
Temperature logs are, and will probably remain, the most widely used logs for measuring fracture
height. However, one significant restriction of the log should always be considered. Temperature
logs are very shallow investigative tools; and if the fracture deviates from the wellbore, it will
quickly become invisible. In general, a temperature log (or postfrac Gamma Ray log) showing
fracture height confined exactly to the perforated interval should be treated with extreme skepti-
cism.
The prediction of fluid loss height, Hp, is important for the design of a fracture treatment. The loss
height represents the net height in the fracture which will dominate the fluid lost to permeable
zones.
One method of selecting Hp, is illustrated in Fig. 5.14, where an Spontaneous Potential (SP) log is
used. For this procedure, the net section to the left of a line 1/3 the distance from the shale line
to the maximum sand deflection. This procedure neglects potential (if any) loss to shale and
dirty sands. A Gamma Ray Log might be used in a similar manner, with fluid loss height being
the net section to the left of a line 1/3 the distance from the shale line and the maximum sand
line. If adequate definition from a SP or GR log cannot be obtained, other cutoffs (porosity) can
be used.
Shale Line
Max.
Sand Line
Fluid Loss
Or
Permeable Line
Neglect Shales, ?
For a given field, the potentially arbitrary nature of this procedure is overcome if the procedure is
consistently used for fluid loss coefficients determined from minifrac pressure-decline analysis or
calibrated along with loss coefficients from the success or failure on past designs of offset wells.
This works because fluid loss height and fluid loss coefficient are multiplied together to arrive at
a fluid loss capacity analogous to reservoir flow capacity (kh). Doubling fluid loss height and
halving fluid loss coefficient yields exactly the same results as the base values. The fluid loss
height is commonly and wrongly confused with net pay height. Fluid loss height will always be
greater than the pay height. In many reservoirs where the net pay cutoffs from porosity logs are
well established, one should ensure that all net pay is included as fluid loss height.
The amount of fluid lost to the formation during a treatment is a primary design consideration. The
lost fluid is essentially wasted and represents a significant portion (i.e., generally 30 to 70%) of the
total fluid and cost of treatment.
C A
q l = ---------- (5.3)
t
where C is the fluid loss coefficient, A is the fracture wall area and t is the time since the area A
was exposed to fluid. The loss coefficient depends on three separate effects as shown on Fig. 5.15
and each of the three have the square root of time relationship given in Eq. (5.3). These effects and
how they are determined are discussed below.
The best estimate of fluid loss is obtained from the pressure decline analysis of a calibration treat-
ment (discussed in Chap. 8).
The composite fluid-loss coefficient depends on three separate linear flow mechanisms with the
separate coefficients, CI - fracturing fluid viscosity relative permeability effects, CII - reservoir
fluid viscosity-compressibility effects, and CIII - wall building effects. In any fracturing treatment,
each of these mechanisms acts simultaneously to varying extents and complements the other.
These mechanisms act analogously to a series of electrical conductors and their coefficients are
combined as shown in the following equation:
1 1 1 1
----- = ------ + ------- + --------- (5.4)
Ct C I C II C III
The fracturing fluid viscosity and relative permeability (i.e., filtrate) effect can be obtained from
the following equation:
k f p
C I = 0.0469 ------------------------ (5.5)
1000 f
Permeability, kf (md), to the fracturing fluid filtrate may be obtained by correcting pressure tran-
sient test derived permeabilities (ko, kw or kG) by reducing the value by a factor of about 5. How-
kc
CII = P ------
- (OIL)
2 p
(CARTER, SPE 1957)
Pk
CI = -------- (GAS)
2 f
k Pc
CIII = - (POLYMER, SOLIDS)
------------
2 f f c
fc FRACTION OF FLUID LOSS ON CAKE
Fracture
Reservoir
Three Components of Fluid Loss:
CI = Frac Fluid Effect
CIII CI CII CII = Reservoir Fluid Effect
CIII = Wall Building Effect
ever, if the filtrate from the frac fluid is similar to the reservoir fluid, than this reduction is not
necessary (i.e., water frac on a water injection well). The purpose of the reduction factor is to
account for relative permeability effects. If relative permeability curves are available they can be
used to determine kf.
Effective porosity should be obtained by correcting the formation porosity for in-place fluid satu-
rations. If, for example, a water based fluid is being used to frac a reservoir, the effective porosity
is reservoir porosity multiplied by (1-So-Sg). If a hydrocarbon based fluid is used; the effective
porosity is the reservoir porosity multiplied by (1-sw).
Pressure differential, p (psi), across the fracture face is the difference between bottomhole
treating pressure (i.e., p = BHCP + P N P R ) and reservoir pressure.
Since polymers are generally filtered from the base fluid by a low permeability matrix, the base
leakoff fluid viscosity, f , is usually that of 2% KCl water containing a slight amount of polymer.
A maximum value for f might be 5 cp with a minimum value of 0.5 cp, depending on formation
temperature.
The CI effect is primarily governed by the viscosity of the filtrate of the fracturing fluid. Since the
viscosity is generally very small (i.e., < 1 cp), the CI term is generally large for current fracturing
fluids and is not effective for fluid loss control. This is not the case for very viscous oils such as
those used during the 50s and 60s.
The reservoir fluid viscosity-compressibility (i.e., formation fluid) effect can be obtained from the
following equation:
k HC c t HC
C II = 0.374 p -------------------------------
- (5.6)
1000 HC
The CII effect is primarily governed by the compressibility, ct and therefore is very important for
liquid filled reservoirs such as oil wells or water injection wells. These generally have a very low
ct compared to gas reservoirs. However, the CII term has negligible control in gas reservoirs which
have a relatively high ct (ct gas = 1/pr gas).
Permeability to the reservoir fluids (kHC) (millidarcies) should be measured by a pressure transient
test. Viscosity and compressibility of the reservoir fluids should be determined as in a pressure
transient analysis (e.g., lab tests, tables, or calculations).
The wall building effect for the fluid loss coefficient is determined from data obtained experimen-
tally in a laboratory as shown in Fig. 5.16. A standard fluid loss test is conducted in a high pres-
sure-high temperature Baroid filter press containing core samples or filter paper. The fluid loss test
is run with a pressure differential of 1000 psi as standard, although p may be much larger, i.e.,
3000 psi. Additional work is required on the effect of p which is currently assumed to be neg-
ligible.
For very low k rocks (< .1 md), the tests should be run using filter paper instead of cores. Other-
wise, the data for CIII will be erroneous due p of the filtrate through the core during the early por-
tion of the test which has a high loss rate. The fluid loss in cubic centimeters is measured at time
intervals of 1, 4, 9, 16, 25, and 36 minutes; and these fluid loss values are then plotted on straight
coordinate paper against the square root of time in minutes (Fig. 5.16). The experimental fluid loss
coefficient is then calculated as follows:
where m is the slope of the plotted data (cc/ t ) and A is the cross sectional area (cm2) of the core
wafer.
Normally, CIII is furnished by the fracturing service company. For critical treatments, fluid loss
tests for the specific fluid and in-situ conditions should be requested.
Fig. 5.17 shows a qualitative comparison of CIII values for different fluids based on laboratory data
from low permeability cores. These test data were run at 150 F and one polymer loading. At
250 F, it has been found that the CIII values for most frac fluids increased by a factor of 1.5 to 2
because of the reduced viscosity of the filtrate through the wall (Fig. 5.15). Keep in mind that the
data in Fig. 5.17 is approximate and the wall building ability of a fracturing fluid depends on for-
mation temperature, and the fracturing fluid type and polymer loading under consideration.
The addition of 5% hydrocarbon to crosslinked water systems (Type III, on Fig. 5.17) can be a very
effective loss control additive for permeabilities less than 1 md and is generally recommended. The
addition of a hydrocarbon dispersion works primarily by reducing the relative permeability of the
polymer cake to water and by droplet plugging of pore throats. Adding the second (oil) phase
reduces the relative permeability to water. Since the hydrocarbon works primarily in the polymer
cake, this technique provides little benefit if most of the fluid loss is CI or CII controlled, as in high
permeability reservoirs. The effect of droplet plugging on a low permeability formation also makes
wall building fluid loss control important for emulsion and foam fluids.
Solid fluid loss additives are sometimes required for efficient fracturing in moderate to high per-
meability or naturally fractured reservoirs. These agents work by blocking the larger pore throats
(i.e., required to form wall building) and fractures. Fig. 5.18 shows the effect of silica flour (Hal-
liburton's WAC-9) on CIII. Such agents are silica flour, 100 mesh sand and manufactured mixtures.
These additives must be used with extreme caution if they are mixed with the proppant, since they
can plug the proppant, unless they are designed to dissolve in the produced fluid. Use of these
additives with proppant laden fluid is not recommended unless absolutely required and then such
that the total does not exceed 1% of the total proppant during the treatment. The addition of silica
flour to the pad at a loading of 15 lb/1000 gal has been used to seal off closed natural fractures.
Also, 100 mesh sand for the initial sand stage (1/4 to 1 lbm/gal) is effective for sealing open natural
fractures.
Spurt Loss
The total fluid lost when wall building dominates is a combination of the fluid lost before a filter
cake has begun to form (spurt loss) and the fluid lost through the filter cake during the treatment.
The point where the fluid loss curve intersects the ordinate on a fluid loss plot is known as the spurt
loss (see Fig. 5.16). For fluids that build effective wall cakes and low permeability formations, the
spurt loss is low. In this case, a value of zero (0) is used for spurt loss if the permeability is very
low (i.e., less than 0.05 md).
Generally, the service company supplies the spurt loss values for their various fluids. Table 5.1 is
an example from the Dowell Fracturing Fluids book showing CIII (i.e., Cw) and spurt for a
non-wall building fluid for various high permeability rocks (i.e., relatively high spurt) and amounts
of silica flour. Spurt loss can be significant for moderate to high permeability formations. For
example, assume a 500 ft fracture radius, 50 ft fluid loss height, and 5 md permeability. Table 5.1
shows 20 gals/100 ft2 spurt loss even with 20 lb/1000 gal silica flour. This equates to an additional
20,000 gal of fluid loss which must be included in the treatment design.
0.005
0.004
0.003
0.002
Permeability
CW -- ft/min1/2
0.1 - 150 md
0.001
0.0005
0.0004
0.0003
0.0002
0.0001
0 20 30 40 50 60 70 80 90 100
WAC-9 Concentration -- lb/1000 gal water
Pr = 2300 psi,
HC = 0.125 ;
Sg = 0.50;
g = 0.0174 cp ;
0.21 cp @ 250 F)
U L T R A F R A C 2 . 0
09:23:40 03/04/92 User ID: ZWXY01
Pressure Diff (psi) 1700. ((Clos Pres + 500 (psi) - Res Pres)
C-III 0.001 @ Test Temp (deg F) 150.
C-I = 0.004463
C-II = 0.026517
C-III = 0.001090 ft/min**.5 at 170. (deg F)
The viscosity values of fluids are determined by laboratory tests. The simplest, but idealized,
experiment of fluid flow is fluid being sheared between plates moving parallel and relative to each
other. The shear stress on the fluid is the shear force exerted on the plates divided by their area with
the units of pressure. The shear rate or velocity gradient is the relative velocity divided by distance
of separation and has the units of 1/time, usually in sec-1. The viscosity is defined as the shear
stress/shear rate.
The rotating cup/bob viscometer has been popularized in the industry by the Fann Instrument Co.
(now under the ownership of NL Baroid). As shown in the idealized drawing, Fig. 5.19, the shear
stress is the force exerted on the walls (sensed by the torque on the bob) divided by the surface
area, and the shear rate is the relative velocity of the stationary bob and the rotating cup divided by
the gap distance. For the standard system, i.e., the R1-B1 Rotor-Bob Geometry, a rotating speed
of 100 RPM represents a shear rate of 170 sec-1, and a speed of 300 RPM represents a shear rate
of 511 sec-1. Unfortunately, this device is not suited to some crosslinked polymer fluids, e.g. borate
crosslinked gels, because of their viscoelastic nature. Borate gels can crawl up and out of the cup.
In spite of this, most published data for borate gels are determined using cup and bob viscometers.
Viscosity is sufficient to characterize the stress-flow behavior, i.e., the rheological character, of
some simple fluids such as water and refined oils. These simple fluids have shear rate independent
viscosity. Most fracturing fluids, however, show shear-dependent viscosities, usually decreasing
with increasing shear rate, i.e., shear thinning, and thus more than one parameter is required to
characterize the rheology. Experimental shear stress and shear rate data are usually correlated by
some approximating rheological model.
The rheological models commonly used in the industry for many types of fluids are the Newtonian,
Bingham Plastic, and Power Law Models, as shown in Fig. 5.20. These models are selected
because they yield straight lines on linear or log-log graphs of shear stress vs. shear rate.
x d Fluid
(x)
Ri
Ro
= Shear Stress = F/A (Pressure)
wR
d
= Shear Rate = -- = ----
- ( --------
1
-) = --d ---------------
o
-
d Ro Ri
dx time
T /R
= F--A = ------------i - = ------------
T
-
2Ri H 2R i H
Rheological Models
I. Newtonian II. Bingham III. Power Law
log
Yp p n'
K'
log
1.0
= = Y p + p = K' n
Yp 0 n' 0
Newtonian Newtonian
p = K' =
A Newtonian type fluid has a linear relationship between the variables with a slope equal to vis-
cosity, i.e., water, brines, and oils.
A Bingham Plastic type fluid differs from a Newtonian fluid by a non-zero stress (i.e., Plastic Yield
Value) at zero shear rate. The slope of the line is termed the Plastic Viscosity (not equal to apparent
viscosity). The initial work on these fluids was done by Bingham on paints and printer's ink - zero
flow (shear rate) on vertical surfaces (shear stress). This fluid model is used in the industry for
drilling muds and cements.
The Power Law fluid model is commonly used for representing frac fluids and predicts a straight
line on a log-log plot with the slope denoted as n' (generally < 1) and termed the Power Law Expo-
nent or Flow Behavior Index (n' = 1, Newtonian; n' > 1 shear thickening; n' < 1, shear thinning).
The stress at a shear rate of unity is denoted as K' and is termed the Consistency Index. This model
does not predict a yield value (no flow with stress, e.g., can form a stationary lip when poured,
remains as a glob on the table). The K' and n' values of real fluids change with increasing time and
temperature (generally K' decreases and n tends toward unity) and depend on their flow history.
Most service companies attempt to account for downhole flow conditioning in some manner when
testing crosslinked fluids.
Although the power law is the primary model used for fracturing fluids, it does not account for
other aspects of flow behavior exhibited by many fluid systems, such as nonhomogeneous flow,
e.g., slip or particle migration, or viscoelasticity. These factors can influence rheological scale-up
and proppant transport and are presently the subject of research. All fracturing simulators treat frac
fluids as if they were homogeneous power-law fluids.
Fig. 5.21 defines and gives an example of apparent viscosity for a power law fluid. The example
shows a realistic case for fracturing fluids. Different service companies have reported viscosity at
different shear rates (i.e., 170 sec-1 or 511 sec-1). The rate in a fracture can be 40/sec. The example
shows that the same fluid can be reported by one company to have 100 cp (at 170 sec-1), another
to have 58 cp (at 511 sec-1) and the fluid may have 206 cp (at 40 sec-1) in the fracture. Therefore,
in selecting fluids it is important to know what shear rate the data represents. Table 5.2 shows a
typical rheological data set presented by service companies for use in fracture design and/or anal-
ysis.
The viscosity of some fracturing fluids, can be very sensitive to their flow and thermal histories.
Fluids often encounter intense flow energies while being pumped downhole, ranging from 0.2
hp/ft3 to 8 hp/ft3. Delayed crosslinked gels are formulated to start crosslinking after the gel enters
the fracture and starts to heat up to avoid degradation of the crosslinks during high energy flow
condition. Foams and oil-base gels, on the other hand, may actually achieve better viscosities after
subjected to high-energy flow conditions. Thus, the viscosity of the frac fluid as it enters the frac-
ture is frac-fluid system dependent and is influenced by flow and thermal conditions.
Example
Power Law: Given: n' = .5
1 a = 100 cp, =170 sec-1
Find: K', a at 40 and 511 sec-1
K' = 100 x (170).5/ 4.8 x 104 = 0.27
a a (511) = (170/511).5 x 100 = 58 cp
a (40) = (170/40).5 x 100 = 206 cp
1
a = ---1- ( Depends on ) Find: K', a at 40 and 511 sec-1
For Power Law Model K' = 100 x (170).5 / 4,8 x 104 =. .027
4
4.8 10 K' 170 0.5
a = -------------------------- a (511) = ( -------- )
x = 58 cp
( 1 n ) 511
An entry temperature and corresponding wellbore n' and K' values are required to calculate the
entry viscosity of the frac fluid. As the fluid flows down the wellbore it acquires heat from the res-
ervoir and from conversion of flow energy to heat. As an estimate for fluid heat up for water-base
fluids and CO2 foams at typical fracturing flow rates, one can use 10F temperature increases at
7000 ft, 10,350 ft, 12,900 ft, 15,120 ft, and 16,780 ft. Thus, if pumping to 13,000 ft, one might
expect the fluid entry temperature to be about 30F higher than surface temperature. If pumping
oil-base gels, the fluid heats up roughly 25F at each of the above depths because of their smaller
heat capacities (e.g., 0.4 Btu/lbm -F vs. 1.0 for water).
As the fracturing fluid flows down the fracture it continues to heat to reservoir static bottomhole
temperature (BHT). Some fracturing design programs assume a bilinear temperature variation
based on the Perkins and Kern width model as shown in Fig. 5.22. The temperature increases lin-
early from the entry temperature to the reservoir temperature during the first one-quarter of the cur-
rent fracture wing length and remains constant for the remaining three-fourths of the wing. More
advanced programs calculate the fluid-temperature profile down the fracture using calculated or
assumed heat-transfer coefficients and material heat capacities. The resulting temperature profiles
are sensitive to fluid heat capacity and may vary significantly from Fig. 5.23.
As alluded to previously, the entry viscosity of the fluid depends on the type of fracturing fluid as
well as on the fluid and thermal histories at the surface and down the wellbore. Not all fluids have
maximum viscosities at the entry temperature. Some gelled oil systems, and most all delayed orga-
nometallic crosslinked gels show viscosities to increase as the fluid heats to reservoir temperature.
After attaining reservoir temperature, an eventual decline in viscosity will be observed. Fig. 5.23
shows typical viscosity trends for various fracturing fluids as a function of time at temperature. The
first point at 0 hours is the entry point and in this case it takes 1.3 hours to attain BHT. Note that
these viscosity trends are at different BHT.
Reservoir Temperatures
Reservoir temperature is a very important variable since the viscosity of the fluid will vary signif-
icantly depending on the amount of time the fluid has been at reservoir temperature (TR on
Fig. 5.22). Therefore, it is best to get a measured BHT. Notice - It is not the maximum log tem-
perature shown on the open hole logs. That value is much too low. The difference between 250F
and 270F can be significant.
Reservoir temperature should be determined by running a static temperature log in the well to be
fractured. This log can be run with a cement bond log. The well must be at static conditions for the
log to yield the temperature that we are interested in. It is suggested that the well be allowed to sit
idle, with no downhole operations of any kind, for at least 1 week prior to running the static tem-
perature log. After a number of such logs are run (5-10 wells) in a given field, the static bottom
hole temperatures measured can be plotted against depth to mid pay to determine a static temper-
ature gradient. Static temperature is expressed as T static = (T gradient (F/ft) * Depth (ft)) +
Ambient surface temperature. Once sufficient data has been obtained to determine this gradient,
the calculated static temperature can be used on future frac designs.
Note from Table 5.2 that for a YF400-5 fluid, a 25F error in temperature (285F vs. 260F) results
in only 2/3 the desired viscosity at 1 hr (228 cp vs. 342 cp) and only 1/2 the desired viscosity at 4
hr (108 cp vs. 205 cp).
When proppant is mixed into a fracturing fluid, the effect is an increase in apparent viscosity.
Recent experiments indicate that both K' and n' are changed when proppant is added to
uncrosslinked fluids, but there is no consensus on the best correlations to use on crosslinked frac-
turing fluids. The proppant effect on K' for the slurry can be approximated by:
n'
K ' slurry = K ' fluid ( C k )
with Ck = (1 - Cv /Cm)-2.5
Here Cv is the proppant volume fraction and Cm is the maximum possible proppant volume fraction
set to 0.6. This expression for K'slurry is supported by a limited amount of unpublished laboratory
data. Fig. 5.23 shows the effect of proppant concentration on slurry viscosity as developed by
Amoco and GRI, respectively.
RI
G
101
O CO
AM
100
0 2 4 6 8 10 12 14
Sand Concentration, lb/gal
zlkb02.038
Fluid viscosity is critical for the successful execution of pressure controlled treatments. Sufficient
viscosity is required for proppant transport, while excessive viscosity will proportionally reduce
the fracture penetration prior to the fluid pressure reaching the formation's pressure capacity (i.e.,
inefficient fracture extension).
For proppant transport, crosslinked gels are preferred over noncrosslinked gels. Studies show sub-
stantial reduction (e.g., 78%) in proppant fall rates through crosslinked gels, under shear, com-
pared to noncrosslinked gels with the same apparent viscosity. The fall rate through foams and
emulsions are also believed to be less than indicated by the apparent viscosity. Another consider-
ation is particle concentration which increases slurry viscosity and retards particle fall. The effect
of increased slurry viscosity due to proppant concentration is important for pressure controlled
designs and requires the base fluid's viscosity to be reduced as proppant concentration increases.
Also, the apparent viscosity for non-Newtonian fluids depends on the shear rate with lower rates
producing higher apparent viscosities. Generally, the shear rate in the fracture is lower than the 170
sec-1 normally used to characterize fluids.
The above considerations can significantly reduce the viscosity requirement over that indicated by
a direct use of Stokes Law. An example, illustrated in Fig. 5.3, show that if proppant fall were to
be limited to 10 ft in four hours, a direct application of Stokes Law would require a viscosity of
1500 cp for 20-40 mesh sand. Assume that under fracturing conditions the crosslink effect would
retard fall only by 50% in contrast to the 78% for ambient and laboratory conditions. In addition,
assume the slurry dehydrates from a low proppant concentration as it enters the fracture to 10
lbm/gal, Fig. 5.3, at the end of the treatment. For these conditions, the effect of hindered settling
would be equivalent to a multiple of 3.2 in the time-averaged value of viscosity. If the reference
viscosity is at 170 sec-1, the shear rate in the fracture is 40 sec-1 and the fluid can be characterized
by the power law with n = 0.6, the apparent viscosity would be 1.8 times greater in the fracture
than for the reference. If, during the time in the fracture and at reservoir temperature, the fluid vis-
cosity reduces by a factor of 10 with a log-viscosity vs. time relationship, the average value of vis-
cosity would be 4.3 times the final value. Combining these factors (2 x 3.2 x 1.8 x 4.3) results in a
multiple of 50, as shown in Table 5.3, and for the fluid considered, sufficient viscosity would be
achieved if it had a final viscosity of 1500/50 = 30 cp at the end of the treatment. Furthermore, this
estimate may be conservative since a reduction of the crosslink effect was used, the fluid does not
experience reservoir temperature for a portion of the fracture length, and suspended particles are
transported in the center portion of the channel (for viscoelastic fluids), where the shear rate is
lower and the apparent viscosity higher than the channel average. Consequently, the viscosity
requirements for proppant transport can be grossly overestimated and a reference value of 100 to
150 cp can provide significant transport.
The next chapter, Chap. 6, gives more background for selecting specific fluids and additives to
achieve the desired viscosities throughout a treatment.
Fracture Radius
The term radius implies one wing of the fracture or the fractures half length and is equivalent to
the reservoir notation xf. However, xf is the apparent productive length and may be smaller than the
design value of hydraulic fracture length as shown in Fig. 5.25. If the production is in bilinear flow,
the productive length is increasing with time, or if the conductivity is very low, (i.e., FCD < 1), the
productive length may be much larger than the apparent productive length, xf.
Fracture Length = ?
Pay
Hydraulic
Propped Length
Productive Length
Length
Consequently, the design radius should be larger than the desired productive length, xf, because of
the above discussion and for a safety margin. If the created fracture length is too small, a refrac
may be required, and there is some question if refracing can effectively increase the propped
length. Ideally, a calibration for each field should be made to determine the relationship between
design radius and productive length, xf.
Pump Rate
The consideration for pump rate has many facets and some fiction. Although pump rate increases
net pressure in the fracture, and hence, the potential for height growth, normally the significant
effect on height believed by some in the industry is more fiction than fact. If height growth is crit-
ical, reducing rate toward the end of the treatment will accomplish the required necessary reduc-
tion in net pressure and will facilitate the surface handling of the higher sand concentrations. Some
of the considerations for rate are discussed below.
Fluid Volume:
As shown on Fig. 5.26, the pump rate affects all three volume terms of the continuity equation, i.e.,
pump time, fluid loss time, and fracture volume (width). Increasing pump rate increases the vol-
ume of fluid stored in the fracture (increased p, w) and decreases the volume lost (less fluid loss
time). As a result, pump rate affects fluid volume required for a given length. The examples indi-
cate that the balancing point is for fluid efficiency of about 0.6-0.7. For treatments with higher effi-
ciencies, increasing rate will store more volume than is saved in fluid loss, while for lower
efficiencies the opposite occurs. Rate becomes most important for very low efficiency. As effi-
ciency goes to zero, the volume required for a given length is inversely proportional to rate, i.e.,
doubling rate reduces the required volume by one-half. The increase in volume for high efficiency
is generally not a consideration because the extra stored fluid will increase the fracture length after
shut-in, i.e., free extension will occur until the tip screens out.
Increased pump rate will significantly increase friction-loss pressures in the tubulars (and in the
perforations if inadequate number and size) and result in a small, but potentially critical, increase
in net fracture pressure, as shown in Fig. 5.27.
The increase in friction pressures also can dramatically increase horsepower requirements if fric-
tion-loss is a significant portion of the total surface pressure.
For cases where horsepower and pressure capacities of tubulars are an important consideration,
these considerations for rate become important.
FQ
1/4
1.75
( QL )
( TURBULENT )
MAY BE CRITICAL
TO HEIGHT CONFINEMENT
Increasing pump rate will increase proppant transport distance (per fall distance) by an amount
approximately proportional to the pump rate increase (as shown by the examples in Fig. 5.28.)
(Note that transport distance is independent of height.)
3/4 3/4
D
--H Q
-----------------
H
(D indep. of H)
The examples also indicate that increasing pump rate can reduce the fluid viscosity requirements.
These reduced requirements result from both the lower ultimate viscosity for proppant transport
needed and from the smaller residence times which reduce the initial viscosities required to allow
for time degradation. This can be very significant for large jobs in hot zones.
However, high pump rates down small tubulars (i.e., high friction pressures) may cause signifi-
cant fluid degradation for some fluid systems. These systems are nondelayed crosslinked systems
with metallic bonding (e.g., Titinate). Guidelines for these systems which will not result in signif-
icant degradation are:
For these degradable systems, pumping down the annulus can cause significant degradation at very
low rates due to the effect of the tool joints.
Degradation is not a consideration for fluids which rebuild their crosslink, i.e., borate crosslinker,
or fluids which benefit from shear, i.e., foams or emulsions. High pump rates can actually improve
the quality of foams and polyemulsion fluids.
a) BETTER TRANSPORT
b) REDUCE REQUIREMENTS
c) REDUCE TIME ENDURANCE FOR FLUID
Depth
The depth to mid point of perforations is used in the wellbore hydraulics equation to estimate sur-
face pressure. At the present time it is considered to be true vertical depth for hydrostatic calcula-
tions.
Friction Pressure
The pressure loss associated with the flow of fracturing fluid and proppant through tubulars. Gen-
erally the values to be entered are estimated for the fluid system in units of psi/100 ft.
The following table shows an example of data obtained from various fracturing service companies'
literature, measured at 20 bpm. Once the type of frac fluid and tubular size has been determined,
the base friction value from the service company for the required fluid system can be entered.
Table 5.4 - Turbulent Friction Pressures at 20 bpm (psi/100 ft).
Apollo Gel 20 30 40
A B C D E F GH I
1000
500 J
K
L
Friction Pressure - psi/1000 ft
N
100
50
10
1 5 10 50 100
Injection Rate - BPM
A - 1 1/4 in, 2.4 lb tubing H - 2 3/8 in x 5 1/2 in, 15.5 lb annulus
B - 2 3/8 in, 4.7 lb tubing I - 4 1/2 in, 9.5 lb casing
C - 2 7/8 in, 6.5 lb tubing J - 5 1/2 in, 15.5 lb casing
D - 2 3/8 in, 4 1/2 in, 9.5 annulus K - 2 7/8 in, 7 in, 23 lb annulus
E - 3 1/2 in, 9.3 lb tubing L - 2 3/8 in x 7 in, 23 lb annulus
F - 2 7/8 in, 5 1/2 in, 15.5 lb annulus M - 7 in, 23 lb casing
G - 4 in, 11 lb tubing N - 7 5/8 in, 29.7 lb casing
5.5 References
1. Warpinski, N. R., Schmidt, R. A., and Northrop, D. A.: In-Situ Stresses: The Predominant Influence on Hydrau-
lic Fracture Containment, JPT (March 1982), 653-64.
2. Kry, R. and Gronseth, M.: In-Situ Stresses and Hydraulic Fracturing in the Deep Basin, paper 82-3321 present-
ed at the 1982 Petroleum Soc. of CIM Annual Meeting, Calgary, Alta., June 6-9.
3. Hubbert, M. K. and Willis, D. G.: Mechanics of Hydraulic Fracturing, Trans., AIME (1957) 210, 153-66.
4. Harrington, L. J., Hannah, R. R., and Williams, D.: Dynamic Experiments on Proppant Settling in Crosslinked
Fracturing Fluids, paper SPE 8342 presented at the 1979 SPE Annual Technical Conference and Exhibition, Las
Vegas, Sept. 23-26.
Fluid Classification
Service companies offer fracturing fluids which can be categorized as either water-base or hydro-
carbon-base depending on the nature of their continuous phase. Fracturing fluids can be grouped
into the following classes:
1. continuous mixed (1/2 vol% hydrocarbon emulsion up to 5 vol% if liquid fluid loss additive
is used)
Polymer Emulsion:
Approximately 33% aqueous polymer solution as the external phase with 67% hydrocarbon
internal phase
Aqueous Foams:
N2, CO2, or 45%-CO2/25%-N2 in water, polymer solution, or gels with 65 - 85% gas internal
phase
Crosslinked Hydrocarbons:
Diesel, kerosene, or crude crosslinked with phosphate acid ester and aluminum, or fatty acid
and caustic
1. batch mixed
1. continuous mixed
Hydrocarbon Foams:
N2 or CO2 in diesel, kerosene, or crude oil with 65% - 85% gas internal phase
Within any of the above classes of fracturing fluids, the engineer is confronted with a list of mys-
terious sounding fluid system names (e.g. Saturn II, Water Frac, Versagel-HT, YF550-HT, YF-GO
III, Polyemulsion, etc.), and associated with each, an equally cryptic list of trade-name chemical
components and additives. As an example, the components for Versagel-HT (referenced on page
6.7) include WG-11, Cl-18, K-34, and HYG-3 with possible additives of GEL-STA, SP Breaker,
WAC-12L, CLA-STA, SEM-7, EnWaR-288, BE-3, ABF, etc.
To select the best fluid system for a particular hydraulic fracturing treatment, the engineer must
consider various criteria. The next section will discuss these criteria.
6
Viscosity Compatibility
Borate X-Link(<180F) 4 3 4 3 5 5 3 3 4 4 5 4
Lease Crude 4 3 2 3 3 5 5 4 2 3 2 2
Gelled Oil 2 4 4 4 4 3 5 4 4 3 2 2
Gelled Methanol 3 4 4 5 5 1 5 4 4 2 1
1 - BAD
5 - Excellent
(*) - Good loss control for permeability < 1md (+/-)
July 1999
Fluid Selection Criteria
Condensate < 32 F
Toluene 40 F
Methanol 52 F
It is easy to see why diesel No. 2 is so popular. In Canada FRAC OIL and methanol are used fre-
quently, perhaps partly because of colder weather which makes their use more safe. Special pre-
cautions are used when pumping flammable liquids such as brass hammers (to avoid sparks) when
tightening surface tubing, tarpaulins to cover surface hoses to protect personnel from spraying
hydrocarbons, spark arrestors, and the prohibition of smoking.1 Foamed fluids, which can be
hydrocarbon or water-base, are even more dangerous because of the expansion energies of leaking
foams.
There are varying degrees of toxicity associated with fracturing fluid components such as metha-
nol, FRAC-OIL, biocides, surfactants and crosslinkers. Breathing apparatus is required for blender
operators and anyone exposed to methanol vapors which can do irreversible brain damage. Oxi-
dizers, such as ammonium and sodium persulfate should not be allowed to contact fuel sources.
Corrosive acidic and basic additives should be handled with care. Material Safety Data (MSD)
sheets should be reviewed for all chemicals on location.
The recent emphasis on environmental awareness has limited the use of some additive such as cer-
tain extremely toxic biocides and crosslinkers (e.g. chromium-base). Service companies are sup-
posedly knowledgeable about environmentally acceptable frac fluid systems. Consult your Amoco
environmental representative if in doubt.
A fracturing fluid may damage a reservoir to various degrees. Ideally, core flow tests should be
done to evaluate the sensitivity of a particular rock to the fracturing fluid.
If a reservoir has swelling or migrating clays, the engineer should use adequate clay control addi-
tives when using water-base fluids or should use an oil-base system. Certain salts such as KCl or
ammonium chloride are effective to some extent in stabilizing swelling clays such as illite and
montmorillonite by replacing exchangeable cations in the clays which can cause expansion of the
stacked clay platelets when exposed to fresh water. Modified polyamines can reduce clay swelling
and clay migration by adsorbing on clay particles and locking them into place. Cationic poly-
meric clay stabilizers adsorb on to clay particles even more strongly but they have the potential of
plugging pore throats (because of their relatively large size) and can change the wettability of the
formation. Clay control additives, other than 1% KCl, are generally not recommended for tight for-
mation gas plays for this reason. 1% KCl is adequate for clay control in fluids with pH as high as
10. For higher permeability formations (where more conductive fractures are required) core tests
should be performed to assess the effectiveness of the prescribed clay stabilizers.
Water in fracturing fluids can actually dissolve the cementing material in some formations causing
the release of damaging fines and consolidation problems. Again, core flooding tests should be
conducted to evaluate this possibility.
The ionic nature of the components in a fracturing fluid system have the potential of changing the
wettability of formations. Ionic surfactants have ionic water soluble ends and nonionic hydrocar-
bon or fluorocarbon tails which are oil soluble. Sandstone formations, which are usually water wet
and negatively charged, will adsorb cationic surfactants and will thus become oil wet because of
their oil-soluble tail. Limestone formations, which usually are water-wet and positively charged,
will adsorb anionic surfactants and become oil wet. Since water-wet formations promote move-
ment of oil through the rock, anionic surfactants should be used in sandstone formations, and cat-
ionic surfactants should be used in limestone formations. For heterogeneous formations, e.g. sandy
limes, nonionic surfactants should be used. If surfactants do not improve fluid recovery they
should not be used unless they are required for foam or emulsion stabilization.
Another consideration is the introduction of anaerobic bacteria (i.e. sulfate reducing bacteria such
as Desulfovibrio) into the formation which can produced hydrogen sulfide and can turn the well
sour. This is particularly a concern in wells less than 170F. The fracturing fluid water should be
treated with an environmentally acceptable biocide. Thus, even in continuous mix operations, the
use of biocides should be considered. In very hot wells, the possibility of bacterial contamination
in the cooler regions of the wellbore exists.
When using water-base fluids, sometimes problems with water blocks occur. These can be miti-
gated by using fluorocarbon surfactants and/or methanol which have especially good surface ten-
sion lowering qualities. Water with reduced surface tension has lower capillary pressure and is
more easily displaced from pore throats through the rock matrix.
Fracturing fluid compatibility with the reservoir fluids is likewise important. Water-base fluids can
form emulsions with crudes or induce scaling with insitu water, e.g. CO3-- in sodium or potassium
carbonate buffers can form CaCO3 scale by reacting with CA++ in the formation water. Oil-base
fluids can induce sludging with crudes such as asphaltene or paraffin precipitation. The fracturing
fluid system should be mixed with the reservoir fluids prior to specifying the treatment to check
for incompatibilities, preferably at reservoir temperature and pressure. Fluorocarbon surfactants
should be used in dry gas wells where there is no danger of forming oil-water emulsions. API
RP392 describes a procedure conducted at ambient pressure, that tests for emulsion and precipita-
tion potential.
The compatibility of the fracturing fluid with its additives should be checked at location by per-
forming pilot tests before pumping. Sometimes incompatible additives can be brought on location
such as certain surfactants and biocides which can interfere with crosslinking. Some surfactants,
e.g. foams, may adsorb on silica surfaces, such as sand or silica flour and cause the foam to break.
Methanol is incompatible with guar at concentrations higher than 20 wt%. Most enzyme breakers
will not work when used at pH higher than 8.5 or at temperatures greater than 120F. Methanol
should not be used with breakers since it renders them useless unless very large concentrations are
used. Resin-coated proppants can interact with fluid additives. Some resin coatings can adsorb
breaker and crosslinker, and can lower fluid pH. Some encapsulated breaker are not compatible
with resin coated sands if the oxidizer breaker is released before the resin cures. For hydrocarbon-
base fluids, the effect of additives on the value of the recovered fluid after flow back should be
considered.
water loss to the formation; chemicals, such as SP Breaker (sodium persulfate), are used to enhance
polymer degradation; GEL-STA (sodium thiosulfate) is used to enhance polymer high-tempera-
ture stability; surfactants are used for better load recovery (EnWar-288), to aid in preventing
oil-water emulsions in the reservoir (LOSURF-251 nonemulsifier), and for emulsifying hydrocar-
bon in water (SEM-7); biocides are used to prevent biodegradation of the fracturing gel and to pre-
vent contamination of the well; etc. Service companies should be able to justify the use of every
fluid component and additive which they specify.
The fluid system should be easy to mix, with the polymer readily dispersed and hydrated. HPG
was developed in part to give better dispersibility and hydrating properties than guar. In the late
1980s, advances in fluid formulations, equipment control, and fluid property monitoring have
made continuous mixed fluid systems more popular. Continuous mixed water-base and hydrocar-
bon-base gels and foams are attractive because of reduced costs resulting from reductions in onsite
time and fluid waste.
A sample of frac fluid should be mixed before the treatment using all onsite chemical additives to
test for proper hydration, crosslinking, and additive incompatibilities. After noting the fluids tem-
perature and pH, viscosities of uncrosslinked gels should be checked with the Model 35 Fann vis-
cometer or equivalent. Qualitative checks on water-base gel crosslinking can be made using vortex
closure tests (e.g. 100 ml sample in a 250 ml beaker stirred at 450 rpm) and visually observing the
crosslink strength by pouring the gel from the cup. Todays delayed crosslinked systems have to
be heated somewhat to initiate crosslinking. Usually crosslinking begins between 90F and 140F.
Hydrocarbon gel viscosities can be checked using the Fann 35 or equivalent noting that mixing
intensity can effect extent and degree of crosslinking. Hydrocarbon gels are difficult to prepare and
require close quality control.1 The gellant and activator must be pilot mixed on location with the
particular hydrocarbon to determine the proper amounts of gellant and activator. Too little activa-
tor will yield no viscosity and too much will give gel degradation (activator is a base, e.g. sodium
aluminate, which can also serve as a breaker). The phosphate ester gellant must be added to the
hydrocarbon before the activator. If the activator and gellant are added together, a precipitant will
result. When the ester is uniformly distributed, the activator is added. Sometimes, the fluid has to
be circulated at the highest rate possible for 20 minutes to form the gel. Gelation can be stopped
by contaminants in the tanks such as residual surfactants, treating chemicals, or water. Improved
continuous mixed hydrocarbon gel systems are making preparation easier.
Gel quality control of continuous-mix jobs is possible if the service company has reliable real-time
pH and viscometer instrumentation on their preblenders.
Onsite rheological testing of foamed systems is not possible. However, the foaming potential and
halflife can be checked by putting the liquid with all its additives into a Waring blender about 1
inch above the impeller and mixing at maximum rate to entrain air. The time it takes for 1/2 the
solvent to return to its original state (drain) is the halflife.1 Foams with foaming agents have
halflives of 3-4 minutes, stabilized foams have halflives of 20-30 minutes, and crosslinked foams
have halflives over an hour.
For every type of fracturing fluid, proper additive metering and carrier vessel cleanliness are essen-
tial aspects of quality control. In the 1990s, chemical addition is automated so that complex sys-
tems, such as binary foams, can be successfully pumped. Quality control of fracturing materials
will be discussed in more detail in Chap. 11 of this manual.
Friction pressures for various types of fracturing fluids are shown in Fig. 6.2 compared with water
for flow in 3-1/2 inch tubing (3.00 inch ID). Polyemulsion (i.e., polymer emulsion) gives the high-
est friction pressure, even greater than water. There is some variation in reported friction pressures
by different service companies, the largest being for borate gels and foams. Today, special formu-
lations of delayed borate crosslinked gels are available which significantly lower friction pres-
sures. Friction pressure can cause a significant increase in wellhead pressure and horsepower
requirement and is an important consideration in design. Water-base solutions and gels formulated
with high-molecular weight polymers, e.g., guar, guar derivatives, cellulose derivatives, and poly-
acrylamide derivatives, are all drag reducing.
For overpressured reservoirs (i.e. those with reservoir pressures greater than hydrostatic water
pressure at that depth), either water-base fluids with hydrostatic gradients of 0.438 psi/ft for 2%
KCl fluids or hydrocarbon-base fluids with gradients ranging from 0.343 psi/ft for methanol to
0.379 psi/ft for 30API crude can be flowed back with natural pressure.
For underpressured reservoirs, lower density hydrocarbon-base fluids, energized fluids, or foamed
fluids can be used to assist flow back. Nitrogen foams at typical treatment pressures and tempera-
tures can have gradients less than 0.2 psi/ft. In addition, foams, by their composition of > 65 vol%
gas, only require about 1/3 as much liquid load return. CO2 foams can have gradients greater than
0.2 psi/ft at typical treating pressures and temperatures. However, the solubility of CO2 in fluids is
generally high compared to nitrogen (see Fig. 6.32), and can give added flow back assistance as it
comes out of solution.
When using less dense fluid, however, one must consider the higher surface treating pressures
required since up to 0.25 psi/ft hydrostatic head pressure can be lost relative to a water-base fluid.
Higher treating pressures can reduce the maximum allowable pump rates and/or increase pumping
horsepower (cost).
Appropriate Viscosity
Fracturing fluids are formulated to give sufficient viscosity to create wide fractures to prevent
pinch outs and to give sufficient width to create a conductive proppant pack. Widths sufficient
to prevent pinch outs are approximately equal to 2.5 times the maximum proppant diameter (e.g.
about 0.1 inch for 20-40 sand). Lower widths can conduct slurry if the fluid flow rate and viscosity
are high enough. Fracture width is generally not a strong function of viscosity (e.g. width 1 / 4
for the PKN model with a Newtonian fluid). Excessive fluid viscosity can increase the fracturing
pressure to the point where natural fractures open up giving additional fluid loss or the fracture can
break through confining zones and grow height. These conditions can lead to screen outs.
Fluid viscosities should be sufficient for adequate proppant transport. This is a rather ambiguous
criteria, however, since proppant has been pumped with very low viscosity fluids including slick
water. Even nitrogen gas has been used successfully to pump 20/40-mesh sand in the Devonian
shale when pumped at high rates. Generally speaking, however, larger quantities of fracturing fluid
can be pumped away at the higher viscosities. This may be a result of better proppant transport
and/or higher fracture pressures creating wider and higher fractures.
When using thin fluids to transport proppant, such as slick water or uncrosslinked polymer solu-
tions at elevated temperatures, it is probable that a settled bank forms along the bottom of the frac-
ture. Research by Biot and Medlin3 and Medlin, Sexton, and Zumwalt4 indicates that the formation
of an equilibrium bank (a settled bank of constant height above which all proppant is transported)
may not apply to field scale fractures although equilibrium banks have been observed in labora-
tory-scale slot flow devices. They believe that for thin fluids, proppant transport results primarily
from viscous drag where the suspension has nearly uniform proppant concentration. As the sus-
pension flows down the fracture, a relatively clear fluid layer forms at the top of the fracture as
proppant from the suspension falls to the settled bank. At any horizontal position x , down the frac-
1
ture, the clear-fluid height above the slurry top is given by:
x1
where U is the average fracture fluid velocity and vt is the terminal settling velocity of a particle.
Medlin et al. set forth the criteria that when vt /U is less than 0.1, suspension transport occurs.
When 0.1<vt /U<0.9, bed load transport occurs which is defined as the transport from a fluidized
layer of sand. However this fluidized layer is not very thick (less than a few inches) and is inde-
pendent of fracture size. When vt /U > 0.9, proppant will not be picked up. Values of vt depend on
the particle size and density and on the rheological nature of the fluid. For instance at room tem-
perature under static conditions, 20/40-mesh sand can settle in water at a rate of 1.7 ft/sec; in a 40
lb/1,000 gal HPG solution at 0.005 ft/sec; in a borate gel at 0.0007 ft/sec; and in titanium gels at 0
ft/sec. Under fracturing conditions, effective settling velocities of flowing slurries are not well
defined at this time.
For viscosities greater than 50 cp at 170 1/s, we will assume that the proppant is flowing as a sus-
pension with settling rate defined using some correlation relating the particle drag coefficient, CD,
to the generalized particle Reynolds number, N'Rep. These dimensionless groups are defined as:
n 2 n n 2 n
4 g dp p d p vt 0.695 d p v t
CD = ---------------
- --------------- ; N Rep = -----------------------
- = ----------------------------------------
-
3v t
2 3
n 1
K K
(in Oil-Field Units)
where g is the acceleration of gravity, dp is the particle diameter (inches), p is2the particle density
(lbm/gal), is the fluid density (lbm/gal), K' is the consistency index (lbf-sn'/ft ), and vt is terminal
particle velocity (ft/sec). N'Rep has been written using an effective particle shear rate such as:5
The actual shear rate on a particle surface settling in a power-law fluid can not be calculated explic-
itly, and thus p has been defined differently by various authors. If the relation of CD to N'Rep is
given, then vt can be solved. For instance, for laminar settling (Stokes Settling),
24 d p vt
C D = ------------ , Where N Rep = ----------------- = Newtonian particle Reynolds no.,
N Rep
and if this is assumed to apply to Non-Newtonian fluids, then the following relation is derived
using the expression for N'Rep:
n + 1 1/n
g d p ( p )
v t = ---------------------------------------
n 1
18 K (3)
in oil-field units:
n + 1 1/n
d p ( p )
= ---------------------------------
n
- , for N Rep < 2.0 .
9.626 ( 36 ) K
If the calculated vt, does not give a N'Rep less than 2., then the higher Reynolds number correlations
can be used.5 For 2 < N'Rep < 500, CD = 18.5/N'Rep0.6; and for 500 < N'Rep < 200,000, CD = 0.44.
Thus, this can involve a trial and error approach. Shah6 developed a method using empirically gen-
erated correlation constants which avoids trial and error.
Note that the above are relationships for single particle settling. As proppant concentration
increases, the particles may clump and give accelerated settling. Above a certain proppant concen-
tration, however, i.e. 4 lb/gal liquid, the separation between proppant particles becomes small
enough where hindered settling starts to reduce the settling velocity. Hindered settling can be
treated by increasing the K' to reflect an increase in the effective viscosity of the continuous
medium. Novotny7 performed static settling tests in simulated fractures using concentrated prop-
pant slurries in polyacrylamide solution and found the hindered settling velocity, vh, to be related
to proppant concentration as
ppg 5.5
v h = 1 ---------------------- vt ,
ppg + p
where ppg is the lbm proppant/gal liquid, and p is the proppant density in lbm proppant/gal prop-
pant (e.g. 22.1 for sand). Thus, for ppg = 8, and vt = 0.005 ft/sec, vh would be 0.0009 ft/sec. For a
fracture flow velocity, U, of 0.56 ft/sec, (40 BPM down a 50 ft high by 0.25 inch wide two-wing
fracture) this would give vh /U equal to 0.0016, which according to Medlins criteria would give
suspension flow. The clear fluid layer at the fracture top after 1000 ft would only be 1.6 ft. This of
course is a rough estimate of settling. The settling properties of flowing suspensions are not yet
well established.
The viscosity is affected by temperature and time and should be accounted for in fracturing design
since this can affect fracture width and proppant transport as stated above. There are high and low
temperature versions of water-base crosslinked and uncrosslinked gels, of hydrocarbon-base
crosslinked gels, and foams. Polyemulsion is usually restricted to temperatures less than 250F.
High temperature stabilizers, such as sodium thiosulfate or methanol, are used above 200F to
retard oxidative hydrolysis of water-base polymers. At pH less than 6., the sodium thiosulfate sta-
bilizer is not effective in some cases.
There can be a large variation of high temperature behavior for similar gel systems. For example
in Fig. 6.3, various service company high temperature gels are compared at 265F. All the gels
were tested by Amoco using the Amoco procedure for testing organometallic crosslinked gels. It
is apparent that high temperature stability is a function of pH and type of polymer, buffer, and
crosslinker. In some cases, service companies reported viscosities to be up to six times greater than
those measured in Amocos lab (e.g. those for the Saturn I, Apollo I, Gemini III DXL, and Titan
XL gels). The discrepancy is the result of many factors including conditioning method, viscometer
bob, viscometer procedure, and calculation method. At this time we feel our procedure gives the
most realistic data. As of 1991, the API is still a couple of years away from approving a recom-
mended practice for testing organometallic crosslinked gels.
Fluids with internal phases can have additional fluid loss control when used in low permeability
reservoirs ( < 1. md). This is true when the internal phase is a hydrocarbon, such as is the case when
diesel fluid loss additive is used. Aromatic hydrocarbons with surfactants which yield microemul-
sions are also used but to a lesser extent. Generally speaking 3% diesel gives about 80% of the fluid
loss reduction possible with hydrocarbon fluid loss additives. Accordingly, polyemulsion (i.e.,
polymer emulsion), with 67% hydrocarbon internal phase, gives the lowest values of wall building
fluid loss coefficient, Cw. Typical values for polyemulsion are < 0.0007 ft min< for permeabil-
ity < 25 md using guar or < 0.0015 ft min for permeability < 2 md using HEC. The dispersed
hydrocarbon acts to reduce the permeability to water in the polymer filter cake by relative perme-
ability effects. At permeabilities greater than 5 - 10 md, the hydrocarbon can penetrate pore throats
and the use of particulate fluid loss additive is also advisable. Cw generally decreases with increas-
ing polymer loading, except when using hydrocarbon fluid loss additive, in which case it is almost
independent of polymer concentration. Fig. 6.4 and Fig. 6.5 show Cw as a function of fluid-loss
additive type and concentration, and polymer concentration.8
0.01
Legend
Polymer-resin Mix
Silica Flour
Polymer-Silica-Clay Mix
Diesel (1-10 md)
Cw (ft/min1/2)
0.001
0.0004
0 10 20 30 40 50
Fluid Loss Additive Concentration (lb or gal/Mgal)
Fig. 6.4 - Wall-Building Coefficient vs. Fluid-Loss Additive Type and Concentration at 125 F.
Foams with gas internal phases can give fluid loss control comparable to gels with hydrocarbon
when the liquid external phase of the foam is stabilized with polymer. Foam Cws are also depen-
dent on formation permeability. Fig. 6.6 shows D-S fluid loss coefficient values vs permeability
for some of their foams at 150F. Oil-base gels exhibit similar fluid loss behavior in that the vol-
ume of fluid leaked off is proportional to the square root of time. It is not clear at this time what
kind of mechanism is responsible for this, i.e. polymer build up, pore throat plugging, or gell
invasion. Most oil-base fluids use some form of non-oil-soluble fluid loss additive.
At reservoir permeabilities less than 0.1 md, the total fluid loss coefficient, CT, starts to become
influenced by leakoff resistance in the reservoir rock. Reservoir-leakoff resistance is influenced by
the fracturing fluid filtrate and the formation fluids, the porosity of the reservoir, the compressibil-
ity of the formation fluids and the leakoff-driving pressure (the fracturing pressure minus the res-
ervoir pressure), as well as the reservoir permeability. Increasing the leakoff driving pressure from
0.005
Legend
0 lb Silica Flour/Mgal
0.004
25 lb Silica Flour/Mgal
50 lb Silica Flour/Mgal
100 lb Silica Flour/Mgal
0.003
Cw (ft/min1/2)
0.002
0.001
0 30 40 50 60 70 80 90 100
Fig. 6.5 - Wall-building Coefficient vs. Gelling Agent Concentration for Linear Gels at 125 F
Through 0.1- to 100-md Cores.
500 to 2000 psi can increase CT by a factor of 3. At a reservoir permeability less than 0.0001 md,
the reservoir resistance dominates leakoff and the fluid leakoff properties, (Cw), no longer are
important. Little added fluid-loss reduction is gained by using fluid loss additives. Most fracturing
simulators consider reservoir effects when calculating fluid loss.
For naturally fractured reservoirs, it is of primary importance not to allow polymer to leak off into
and plug the natural fractures which can be the primary source of permeability. Fluid loss additives
should be considered which prevent polymer from entering the natural fractures (particulate
agents) and which reduce the leakoff of the filtrate (hydrocarbon fluid loss additive). In the case
of naturally fractured reservoirs, Cw again will control fluid loss. Fig. 6.7 shows calculated CT as a
function of reservoir permeability for an East Texas Cotton Valley well with leakoff driving pres-
sure and diesel concentration as parameters for cases with and without natural fractures.9 In this
figure, particulate fluid loss additive is assumed necessary to seal natural fractures so that a filter
cake can form. Experimental tests have shown silica flour to be necessary to stop leakoff into
smaller natural fractures (i.e., 10 to 20 microns).8
psi
Fig. 6.7 - Calculated Total Fluid Loss Coefficient vs. Permeability ETCV at 275 F and 5000 psi
Reservoir Pressure. Leakoff Driving Pressure as a Parameter; With and W/O 3% Diesel
The spurt loss of a fluid, which can be defined as the fluid loss/area before the formation of a filter
cake, can be significant in naturally fractured reservoirs as well as reservoirs with permeabilities
greater than 1. md. Spurt values increase strongly with reservoir permeability and leakoff-driving
pressure and are affected by factors which affect fluid flow in reservoirs such as filtrate viscosities
(and thus temperature) and compressibility effects. Spurt values in excess of 1 gal/100 ft2 can be
expected for reservoirs with permeability greater than 10 md. Lab measurements show spurt values
are affected by fluid loss additive and polymer type, as well as permeability. Fig. 6.8 - Fig. 6.11
show the effects of fluid loss additive type and concentration, and polymer concentration on spurt
vs. permeability values.8
Fig. 6.8 - Spurt Loss vs. Permeability for Linear HPG Gels in Water at 125 F.
The preceding discussion dealt with fluid loss behavior of static fluids. Fluid loss can also be
affected by fluid flow in the fracture. For uncrosslinked polymer solutions, Cw is independent of
shear rate up to 300 1/s, but Cw for crosslinked gels can increase with shear rate. Leakoff driving
pressure can affect the functional form of fluid loss. Fluid loss can increase from being propor-
tional to t to proportional to t as the leakoff-driving pressure decreases from 1000 to 0 psi. Lab-
oratory tests have shown that flowing proppant does not change the dynamically measured Cw for
proppant concentrations up to 5 ppg.10 Dynamic fluid loss is a new technical concern which is not
considered in all fracturing simulators.
Fig. 6.9 - Spurt Loss vs. Permeability and Additive for 40 lbm Complexed HPG Fluids at 125 F.
Fig. 6.10 - Spurt Loss vs. Permeability and Gel Concentration For Complexed HPG Fluids at 125 F.
Fig. 6.11 - Spurt Loss vs. Permeability and Additive for Gelled Diesel at 125 F.
sions, scales, or water blocks will form. Methods described in the Compatibility With Formation,
Formation Fluids, and Chemical Additives section starting on page 6.6 can be used to reduce the
severity of these problems when using water-base fluids. If these are not effective, hydrocarbon
base fluids or foams should be considered.
Gel breakers oxidize the polymer backbone enabling the polymer to be flowed back out of the frac-
ture. Ammonium or sodium persulfate are commonly used at high temperatures > 150F or lower
temperatures with an activator. Enzyme breakers such as hemicellulose are used at temperatures
less than 120F and pH < 8.5. Western Company claims to have an enzyme breaker which is effec-
tive to pH 10. In 1991, service companies began offering encapsulated or crushable breakers
designed to release the oxidizer after pumping has stopped.
Cellulose polymers or synthetic polymers (e.g. polyacrylamide) leave negligible residue upon
breaking! However, these broken polymers can damage the rock matrix, apparently by adsorbing
on pore surfaces. Water-base fluids using guar or guar-derivatives, leave insoluble residue after
breaking, that can occupy from 1 - 3 vol% of the original fluid volume. This residue has been
shown to damage fracture conductivity. If the residue is dried, it loses more than 98% of its original
volume and, therefore, would no longer be a problem. However, most reservoirs are wet to some
degree, and this residue, which is not mobile,11 can permanently damage proppant packs. In addi-
tion, the effective polymer concentration is increased considerably by leakoff after shutin during
fracture closing. For instance, the pounds of polymer per gallon of proppant pack pore space, (Cp)eff
is:
p ( 1- p )C p
( C p ) eff = ------------------------------- ,
Cs p
where p is the proppant density (lbm/gal proppant), p is the proppant pack porosity, Cp is the
polymer concentration before closure (at shutin) in lbm/gal, and Cs is the pounds of proppant/gal
of liquid in the fracture at shutin.
For example, if at shutin Cs is 8 lbm proppant/gal liquid and Cp is 0.04 lbm HPG/ gal liquid, for
sand proppant with a density of 22.1 lbm/gal proppant and a proppant pack porosity of 0.35, the
effective polymer concentration after fracture closure would be 0.205 lbm HPG/gal liquid. It
would concentrate over 5 times. In addition to this, there is polymer already deposited on the frac-
ture wall before shutin which will contribute to the residue. Using a model based on the Kozeny
equation12 for permeability which accounts for polymer residue from leak off during and after shu-
tin, Fig. 6.12 and Fig. 6.13 can be derived which show the normalized impaired proppant pack per-
meability, k'/k, as a function of CT and Vrf for a position in the fracture with a width of 1/4 inch,
and a fracture age of 60 minutes. Vrf is defined as gal of residue/gal of fluid. Shown in these figures
are two hypothetical extremes. The first is where all the residue concentrates at the fracture wall,
(effectively reduces the fracture width--the most optimistic case) and the second is where the res-
idue is uniformly distributed--the most pessimistic case. Studies13 have shown that residue tends
to concentrate near the wall, so, the optimistic values are probably more realistic. However,
extreme permeability impairment can occur when Cs is less than 5 ppg at typical fracture widths
and leakoff rates, as shown in Fig. 6.12 and Fig. 6.13.
The preceding impairment discussion assumes that all the polymer breaks. In reality, using con-
ventional breaker loadings, this is probably not the case since solid breaker (e.g. sodium persulfate)
does not concentrate with the residue (it leaks off) and enzyme breakers are frequently destroyed
by temperature, high pH, and chemical additives. This realization has spawned the new genera-
tions of crushable and controlled release breakers which can be used at much higher concentrations
than before and which will accumulate with the polymer. Thus, it is now possible to achieve almost
complete polymer breaks. These new types of breakers are even being used at temperatures above
275F to maximize breaking, especially of high pH fluids. They also may be useful in breaking
methanol gels which are very stable and require high breaker loadings. Some kinds of encapsulated
breakers are incompatible with resin coated sands and can release varying amounts of breaker pre-
maturely, e.g., 5% during a 3 hr treatment.14
Encapsulated breakers are used very aggressively in the pad stage (e.g., 7 lb/1,000 gal)14 with the
concentration reduced somewhat during later proppant stages. Sometimes conventional breaker is
added at the final stages of the treatment to enhance near wellbore cleanup. The benefit of adding
large amounts (greater than 2 lb/1,000 gal) of conventional breaker is suspect. Tests have shown
Fig. 6.12 - Gel Residue Flow Impairment - Fluid Loss During & After Pumping; Residue Uniformly
Distributed or All on Wall.
Fig. 6.13 - Gel Residue Flow Impairment - Fluid Loss During & After Pumping; Residue Uniformly
Distributed or All on Wall.
that a frac fluid with 2 lb/1,000 gal breaker can be broken prematurely before it gets down to the
fracture.
Fewer problems with fracture conductivity impairment result when using hydrocarbon-base fluids
or foams, as long as they break properly. Hydrocarbon gels are broken with base additive. At lower
temperatures, e.g. < 120F, breaking hydrocarbon gels can be a problem. Foams break when the
liquid drains (half life), when the surfactant adsorbs onto the rock, and/or when the polymer in the
liquid phase breaks. Flow back with foams has the added advantage of the nitrogen or CO2 expan-
sion. Polymer emulsions break when the polymer in the continuous phase breaks and/or the sur-
factant adsorbs onto the rock.
Economics
After narrowing the list of possible fracturing fluid systems, the engineer should compare their rel-
ative costs. The costs of the base fluid and additives should be tallied along with disposal costs.
For hydrocarbon-base fluids and polymer emulsion, the value of recovered hydrocarbon should be
subtracted. In the past, hydrocarbon fluid and foam treatments were considerably more expensive
because of the added safety, equipment, and implementation costs. However, presently their costs
are becoming more comparable to water-base systems. Pumping cost is also a function of fluid
type through the effect on friction pressure and hydrostatic gradient. See Table 6.3 for typical frac-
turing chemical and hydraulic horsepower prices (ca. 1992).
In addition to the cost of materials and pumping, one should consider the net present value of
post-frac production. This is a function of the fracture geometry and conductivity which are both
affected by the fluid system through the fluid rheology, proppant transport, leakoff, and gel dam-
age. Making this evaluation is best done using an integrated design package including a fracturing
simulator, production simulator, and economic optimization program. See Chap. 9 of this manual
for information on economic optimization.
The term gelling agent for water is synonymous with the term water soluble polymer; how-
ever, only the latter properly describes the material. The family of natural water soluble poly-
mers consists of vegetable products, such as cellulose (although it is not water soluble, many of its
derivations are), starch, alginates and natural gums. Also included in this list of natural gelling
agents are animal products such as gelatin, glue and casein. Synthetic products fall into two major
categories: modified natural products and completely synthetic products. The modified natural
products are starch, natural gum or cellulose molecules which are modified with various chemical
side chain substitutions. Some completely synthetic products are polyalcohols, polyacids, poly-
ethers, and polyamides, made from a variety of synthetic monomers. Natural and synthetic water
soluble polymers are listed in Table 6.4.
The water soluble polymers most commonly used in fracturing fluids include guar gum and two
of its derivations, hydroxypropyl guar (HPG) and carboxymethyl hydroxypropyl guar (CMHPG)
whose chemical formuli are shown in Fig. 6.14 and Fig. 6.15.
Other types of water soluble polymers used for fracturing fluids include cellulose derivatives, the
most common of which are hydroxyethyl cellulose (HEC) and carboxymethyl cellulose
(CMHEC). HEC and CMHEC leave no residue when broken, but are more expensive than the
guar-based polymers. HEC cannot be crosslinked, but CMHEC can. Polyacrylamides are often
used as friction reducers although recently some companies have started using various forms of
crosslinked polyacrylamide. Although the large family of water soluble polymers has been
reduced to a relatively few that are commercially important, the interaction between these various
polymers, the ability to crosslink them and the possibilities of adding other materials to alter the
physical properties make the choice of fracturing fluid difficult at times. Table 6.5 lists the primary
types of water-soluble polymers used in fracturing today. Most service companies have the
guar-based polymers available as polymer concentrates for continuous mix application.
Fig. 6.17 and Fig. 6.18 show power law parameters for Halliburtons HPG (WG-11) solutions at
various concentrations as a function of temperature.
Guar, HPG, and CMHPG are the most commonly crosslinked fracturing polymers. CMHPG can
be crosslinked by aluminum and/or organic-zirconates because of its dual crosslinked functionality
(see Fig. 6.21). Titanate and zirconate crosslinkers are used at temperatures above 180 F because
of their high temperature stability. Refer to Table 6.6 for pH and temperature ranges for crosslink-
ers. Notably, the use of boric acid and borax is limited to temperatures less than 225 F. Slowly
solubilizing naturally occurring borate ores, such as Colemanite or Ulexite can be used at higher
concentrations, avoiding gel overcrosslinking at the surface but providing more boron at downhole
temperatures giving adequate viscosities to 275 F. Recently, B.J. Services developed an
organo-complexed borate crosslinker (OCB) which they claim is effective to 300 F.16
In the mid 1980s, service companies introduced the use of polymer (gel) concentrates which could
be continuously mixed during the treatment, rather than batch mixed the day before. This provides
both the service company and operating company with cost and time savings. However, close
monitoring of fluid streams and quality must be maintained during pumping. Typically, a polymer
concentrate is prepared by mixing the polymer (e.g. guar, HPG, or CMHPG) in diesel at concen-
trations of up to 5 lbm/gal diesel. Suspension stabilizers are used to keep the polymer dispersed.
When added to the mix water, the polymer in the gel concentrate hydrates rapidly and crosslinks
down the wellbore or in the fracture. This produces a gel with hydrocarbon (e.g. diesel) as a dis-
persed phase usually at a concentration near 0.5 vol%.
basic idea behind delayed crosslinked frac fluids is to retard the crosslinking reaction until the flu-
ids exit the very high shear regime occurring within the treating string, allowing crosslinking to
occur under relatively low shear conditions within the fracture. Crosslinking under this much less
severe shear regime results in a much higher in-situ viscosity with lower polymer concentrations.
There is some confusion as to whether delayed crosslinkers are time or temperature activated.
Crosslinking is a chemical reaction; therefore, chemical reaction kinetics apply. This infers that the
crosslinking rate (change in viscosity or molecular weight with time) is a function of concentration
of reactants, and temperature, as well as some sort of an effective activation energy.
The ideal delayed crosslinked fluid would undergo minimal crosslinking within the treating string
but quickly become crosslinked as soon as it left the perforations and entered the formation. This
is not likely to occur, unless there was a rapid and significant temperature change at the fracture
entrance (physically improbable due to heat transfer and subsequent temperature equilibration).
Practically, a sort of balancing act may be required. It is desirable to maximize in-situ fluid viscos-
ity as near the wellbore as possible to maintain adequate proppant transport in order to minimize
excessive proppant banking that can cover the lower perforations, thereby increasing the potential
for a near-wellbore screenout. This objective is weighed against the original objective of delayed
crosslinked frac fluids--increased fracture viscosity by not crosslinking within the treating string.
The practical solution may come by allowing a certain degree of sacrificial crosslinking to occur
within the treating string such that the reaction is proceeding as the fluid enters the fracture,
enhancing proppant transport early near the wellbore, and accepting loss of some long term vis-
cosity potential. In order to do this, variables such as treating string residence time and specific
flow energy, base fluid temperature, gel pH, polymer concentration, and heat-up rate within the
fracture need to be known or estimated. Service companies also have developed dual-crosslinker
systems (e.g. boron-delayed Ti or aluminum-delayed Zr) which provide viscosity at the wellbore
as the boron and aluminum crosslinks reheal. Later in the fracture, as the gel heats up, the Ti and
Zr crosslink under low shear and give good viscosity at high temperature.
Generally, full delay of crosslinking is desired throughout the pad volume with progressively less
delay as proppant is added and the fracture is cooled down. High temperature delayed crosslinked
frac fluids are not designed to be used below 200 F. They may not break completely at lower tem-
peratures or their crosslinkers may not react rapidly enough with cooled formations to provide ade-
quate near-wellbore proppant transport.
Fig. 6.22 shows Halliburtons n' and K' data for their delayed crosslinked Versagel HT at 250 F.
Fig. 6.3 shows the viscosity behavior at 265 F of Versagel HT compared to other service company
organometallic delayed-crosslinked gels formulated with 40 lbm/1,000 gal of various polymers, as
determined by Amoco.
Delayed crosslinked borate gel systems are also available. Delayed borate gel systems were devel-
oped to reduce the relatively high friction pressures of borate crosslinked gels (e.g., the BJ borate
gel in Fig. 6.2) and to provide high viscosities to 275 F. Crosslinking can be delayed by the use
of delayed pH-control additives or by using slowly solubilizing borate ores.
When mixing proppant into polymer emulsion, the proppant effectively adds to the internal-phase
volume fraction with a subsequent viscosity increase and, if in turbulent flow, an increase in fric-
tion pressure. The latter is responsible for friction-outs where pumping must be stopped because
of excessively high well-head treating pressures.17 To avoid this, polymer emulsions should be
pumped using the constant internal phase philosophy where the emulsion quality is reduced as
proppant is added to maintain a nearly constant viscosity. Table 6.7 shows how the emulsion qual-
ity is varied as proppant concentration is increased to maintain constant internal phase volume
fraction and constant viscosity.17 There is little difference in the two approaches which implies that
maintaining constant internal phase is a convenient means of maintaining viscosity. Too much
proppant can cause the emulsion to break, e.g. when the total internal phase approaches 0.85.
Fig. 6.24 shows the shear thinning behavior of polymer emulsion as a function of temperature for
a 0.67 quality fluid17 and Fig. 6.25 shows the effect of increasing the polymer concentration in the
water phase.18 Fig. 6.26 shows the effect of temperature and quality on viscosity at 511 1/s for
Western Co. diesel oil /0.57 wt% guar emulsions.
Polymer emulsion viscosity is also dependent on mixing energy, where viscosity increases with
increasing energy input. Fig. 6.27 shows the effect of emulsion droplet size on emulsion viscosity.
Viscosity doubles as droplet-size decreases by 50%. Thus, field mixing method can effect viscos-
ity significantly.17
tion, quicker fluid recovery due to gas expansion, less formation damage from invasion of foreign
liquids or additives, and good proppant transport due to yield stresses retarding proppant settling.
A typical composition of a foamed fracturing fluid is 70%, by total foam volume, of gaseous nitro-
gen, carbon dioxide, or 50%/20% CO2 /N2 (a binary foam) as microscopic bubbles in water con-
taining 1-6 gallons of a surface tension reducing surfactant (foamer), 40 pounds of HPG per 1000
gallons of liquid, and a breaker, if appropriate. The viscosity of [foam increases exponentially
above approximately 55 quality (percent dispersed gas volume to total foam volume) until an
unstable condition is attained around 95 quality.
Not only are rheological properties of foams very dependent upon foam quality, but they also
depend upon the viscosity of the constituent fluids, bubble size, and size distribution (foam tex-
ture). The smaller the bubble and the larger the fraction of these small bubbles, i.e., the finer the
texture, the higher the viscosity and stability of the foam at a given quality. Fig. 6.28 and Fig. 6.29
show n', K', and viscosity data for nitrogen foams as a function of temperature and/or shear rate
with quality and HPG concentration as parameters.15 Fig. 6.30 shows the effect of shearing time
on bubble size diameters.15 Fig. 6.2 shows some reported friction pressure data for foams and
Fig. 6.31 gives the friction pressure of Dowell Schlumbergers stabilized foam in 2 7/8 inch tub-
ing.
Texture depends upon the conditions under which the foam was generated. High shear conditions,
such that intimate liquid-gas contact can occur, enhances the generation of fine foam texture.
Increasing the viscosity of the aqueous phase via polymers also increases foam viscosity and sta-
bility. This is thought to occur by retarding the rate at which bubbles coalesce, as well as increasing
the resistance to bubbles slipping past one another.
The foam composition, amount of energy input during the generation of the foam, as well as foam
quality, may have a dramatic effect upon the resultant rheological properties. Typically, without
polymer, foams follow a Bingham plastic rheological model. With the addition of polymer, power-
law properties are introduced. Increasingly finer foam texture and higher polymer concentrations
result in increased non-Newtonian flow behavior. (See Chap. 5 of this manual for a discussion of
rheological models.)
There are questions to be answered about the feasibility of using foams for long-term, high tem-
perature, fracturing applications. There are limited data to verify that under typically low fracture
shear rates and for extended periods of time at high temperatures, foams retain sufficient viscosity
to ensure continued leakoff control, and sufficient proppant transport capabilities to make them
truly competitive with conventional fracturing fluids. This is particularly a concern after pumping
has stopped and flow near the wellbore almost ceases. It may be advisable to crosslink the last
stage of a foam job to give better wellbore stability. Disadvantages of foam are higher treating
pressures (for nitrogen foams) due to reduced head, and low proppant concentrations because of
the low fraction of water. This can be overcome by reducing the quality as higher sand concentra-
tions are required. As in the case of polymer emulsions, constant internal phase during proppant
addition maintains the foam viscosity approximately constant.19
Western Company has recently developed the use of binary foam composed of 50% CO2/20% N2
which they claim gives better well cleanup than pure CO2 foams. Fig. 6.32 shows the solubility of
CO2 and N2 in water,20 andFig. 6.33 shows the viscosity of a 60 lbm CO2 foam vs. a 50 lbm binary
foam.20
Foams are difficult to characterize because of their sensitivity to preparation techniques and test
conditions. There are no long-term foam data available where the foam was not circulated through
a pump (and perhaps restabilized). STIM-LAB testing has also indicated that foam stability is sen-
sitive to silica flour. Apparently, surfactant may adsorb onto the silica surface. The same may also
be true to some extent when using sand proppant.
Gelled Hydrocarbons
Gelled hydrocarbons were the first fluids used in hydraulic fracturing. In 1947, Stanolind Oil
pumped four stages of gelled gasoline followed by a gasoline flush down a well in the Hugoton
Field. Aqueous frac fluids were avoided until 1957, when it was found that clay control additives,
such as KCl, were effective in water sensitive formations. The earliest gelled hydrocarbons were
napalm-type fluids of aluminum octoate, and later in the 1950s, fatty-acid soaps composed of caus-
tic and tall oil fatty acids were successfully used.21 These gels usually provided adequate viscosity
to 150 F. In 1970, high-temperature gelled-hydrocarbon systems composed of aluminum
crosslinked orthophosphate esters were introduced, eventually leading to systems that are ther-
mally stable to 350 F.
The reaction of the ester and base (e.g., sodium aluminate) forms an association complex through-
out the hydrocarbon which increases its viscosity (see Fig. 6.34). The resulting gel is shear thin-
ning (n' typically lower than 0.25) and is capable of rehealing after seeing high shear conditions.
In fact, in preparing these gels, high shear conditions are required to form the association complex.
Hydrocarbons such as kerosene, diesel, and FRAC-OILTM are often used to prepare these gels. If
the produced crude has high enough gravity, e.g., > 35 (0.85 g/cm3), it can also be gelled.21 Use
of the produced crude is advantageous since it can reduce fluid incompatibility problems.
The primary advantages of gelled hydrocarbons are low damage to water sensitive formations and
low damage to proppant packs if the gel breaks properly. The disadvantages include the fire haz-
ards associated with pumping hydrocarbons, higher pumping pressures resulting from sometimes
higher friction pressures and lower specific gravity (less head), more fluid loss, sensitivities to
polar contaminates such as water, difficult quality control and mixing, and higher initial cost.
The viscosity of hydrocarbon gels appears less sensitive to temperature than water-base gels. This
is true when measured at higher shear rates. However, the low-shear (< 100 sec-1) viscosity may
be reduced significantly as temperature increases, and low-shear viscosities control proppant trans-
port. Unlike water base fluids, the viscosity of hydrocarbon gels increases with pressure (e.g.,
6%/1000 psi) giving an additional viscosity edge over published data which are generally obtained
at pressures less than 1000 psi. These trends are seen in Fig. 6.35.
Table 6.8 shows the results of a comparative evaluation of Halliburtons continuous mix
MY-T-OIL IV and batch mixed MY-T-OIL II using FRAC-OIL 200 and crude. Viscosities at
158 F (70 C) are a function of gellant, activator, and breaker concentrations. Table 6.9 gives data
for Western Companys MAXI-0-74 gel.
Gelled Methanol
In 1974, aqueous polymer solutions with up to 25% methanol in guar solutions and up to 60% in
HPG solutions started being used in water-sensitive formations. The maximum amount of metha-
nol is limited by precipitation of the polymer. Some polymers, such as hydroxypropylcellulose can
viscosify 100% methanol. In 1987 crosslinked forms of methanol became available, e.g. through
BJ Services. These methanol gels can be used with CO2, which is generally soluble in methanol at
all concentrations, forming a single phase.
Methanol gels are suited for water sensitive formations because of lower water concentration.
Methanol reduces surface tension which aids load recovery and the removal of water blocks. These
gels also have low fluid loss, low friction pressure and, when used with CO2, give energized flow-
back. Disadvantages include high cost, high flammability, toxic vapors, and large amounts of
breaker needed to break the polymer (methanol is a high temperature stabilizer). In water-sensitive
formations, gelled oils or diesel are generally preferred over gelled methanol.20
To further complicate matters, there is no standardized laboratory test for measuring fracturing
fluid viscosity. This implies that some of the data currently in the literature and used by service
and production company personnel are not directly comparable, let alone physically representative
of actual conditions of application. Currently, the API Committee on Well Completion Fluids has
a subcommittee investigating the possibility of developing a standardized testing technique for
crosslinked frac fluids. Such an API recommended procedure is not expected before 1993 and is
not expected to be applicable to borate crosslinked gels.
Amoco has issued a recommended test procedure for determining the rheology of titanium and zir-
conium gels (report F90-P-73).22 This procedure conditions the fluid in a bench-top mixer to sim-
ulate downhole pumping in casing and tubing before pumping the gel to the Fann 50C. Special
shear ramps are performed to check for slip flow, which can give anomalously low viscosities.
Test procedures which subject the fluid to simulated field mixing and turbulent down-hole flow
conditions before pumping into the Fann viscometer are preferred by Amoco, because of flow- and
thermal-history sensitivities of some fluids. Halliburton conditions its fluids by circulating through
a small loop using a Jabsco gear pump at a high flow rate for four minutes to simulate flow down
shallow wells and for ten minutes for flow down a deep well. DS conditions its gels by flowing
through capillary tubing at nominal shear rates matched to field nominal shear rates (2.5 minutes
at 675 sec-1 for the shallow well case and for five minutes at 1350 sec-1 for the deep well case).
However, the DS technique does not simulate turbulence, because capillary flow occurs at low
Reynolds numbers. It also does not match flowing energies. The Amoco method mentioned above
matches volumetric flow energy and achieves turbulence using a specially designed bench top
mixing device. The API will probably standardize testing using the DS capillary method of condi-
tioning. However, test data run using any form of conditioning is preferable to the old RP39
method which uses no fluid conditioning.
turing fluid. Halliburton has devised a four-digit designation associated with a Versagel fluid, e.g.,
Versagel 1400. The first digit of the Versagel designation indicates the use of a base polymer,
either WG-11 (HPG) or WG-12 (HPG with internal breaker). The second digit of the four-digit
Versagel designation is the polymer concentration of the base gel in 10's of pounds polymer per
1,000 gallons water. The third digit indicates the use of delayed hydration polymer, either HPG or
HEC. The HP guar is designated as WG-14, and HEC is designated WG-17. The last digit is the
secondary gelling agent polymer concentration in 10's of pounds per 1,000 gallons. Five percent
hydrocarbon can be added to Versagel for leakoff control.
More frequently, the delayed crosslinked Versagel HT is used, especially if fluid shear degradation
is anticipated (as is almost always the case). CL-18 (delay organo-metallic) as well as CL-11 (titan-
ate) are used to modulate the extent of initial crosslinking. The percentages of either constituent
will vary depending upon mix water, pH, temperature, treating string, residence time, etc. Some of
Halliburton's HPG systems (as of 1989) are given in the cross reference (Table 6.11).
Dowell Schlumberger also has a coding system for some of its water-base crosslinked gels. These
gels are labeled as YF for wide frac and have a three integer suffix. The first integer implies
both the polymer type and the crosslinker. If odd, it is a guar system, whereas if even, it is HPG.
This first integer is set to 1 or 2 if borate is the crosslinker (1 means a borate crosslinked guar and
2 a borate crosslinked HPG). Likewise, 3 and 4 refer to titanium systems and 5 and 6 refer to their
delayed zirconium gels. For their borate gel, they add the letter D to denote whether it is delayed.
The next two integers refer to the polymer concentration in lb/Mgal. For example, YF-140D is a
delayed crosslinked borate guar gel at a concentration of 40 lb/Mgal.
In 1991 Western Company changed their water-base crosslinked fluid naming system. Gels are
now referred to by a name that corresponds to the crosslinker type followed by a roman numeral
designation for the polymer type. Titanium, aluminum, zirconium, and borate gel systems are
referred to as APOLLO, GEMINI, SATURN, and VIKING respectively. Guar, HPG, CMHPG,
and CMHEC are indicated by I, II, III, IV respectively.
Generally, service company fluid trade names give little information about the nature or indicated
application of the fracturing fluid. Looking at the fluid cross reference (Table 6.11), it can be seen
that there are some exceptions. The most simple system, water with friction reducer (< 0.12 wt%
polyacrylamide) is given names such as Aqua Frac, Water Frac, and Slick Water. Some of the
CMHEC systems are given names like Kleen Gel or Krystal Frac XL which refer to the low residue
(essentially zero) of the CMHEC when it breaks. Some gelled oil systems are aptly named
MY-T-OIL or YF-GO III. Halliburtons new high temperature system (to 370 F) is called Ther-
magel (a high pH zirconium crosslinked CMHPG). Halliburton adds a suffix to some of its sys-
tems names referring to maximum intended temperature. Their LT designation implies
maximum intended use of 125 F, i.e., low temperature. HT refers to intended high temperature
use up to 300 F.
The cross reference (Table 6.11) provides a method for getting a general idea of generically similar
systems for different service companies. However, the exact chemical formulations may differ.
Some service companies, especially the big four take pride in their special formulations, which
according to their own testing, can give superior performance. Sometimes, however, the superior
performance may be a result of the particular test method used to evaluate it, rather than a superior
composition.
At this time, the viscosity guideline (discussed in Section 5.3 of this manual) will be used for fluid
scheduling. This guideline states that if a neat (proppant-free) fluid can maintain at least 50 cp
at 170 1/s shear rate during its lifetime in the fracture, then it is probable that adequate proppant
transport will result. This statement is based on the assumption that the effective viscosity acting
to reduce proppant settling is increased by the presence of proppant and by shear rates typically
lower than 170 1/s in the fracture. In addition, at times earlier than the total fluid exposure time in
the fracture, viscosity is usually greater since the fluid has had less exposure to degrading thermal
effects. Field experience has shown that fluids meeting this viscosity guideline can successfully
transport proppant. Whether this transport proceeds via perfect proppant transport, slow settling,
or an equilibrium banking process is presently the subject of research.
Fracturing design simulators require a knowledge of the viscosity of a fluid element at a given time
and position in the fracture. The fluid-element rheology is a function of exposure time at bottom-
hole temperature (BHT). However, the fluid-element exposure time is a function of the fracture
geometry and leakoff which are in turn functions of the fluid rheology and composition (as well as
the fracturing model used --PKN, GDK, etc.). Thus, the engineer does not know before the simu-
lation, what BHT exposure time each fluid stage is going to experience. Therefore, optimal sched-
uling of fluids with the appropriate viscosity is an iterative process. The following are two
approaches to fluid scheduling. The first uses a given fluid system with known rheology (n', and
K' as functions of time at temperature) and the second constrains the rheology of the fluid element
in the fracture to be between 200 cp and 50 cp. The first technique is more suited for smaller treat-
ments (less than three hours), whereas the second method is useful for larger treatments.
age fracture width and fluid efficiency are not known, values of 0.25 inches and 0.4 respectively
can be assumed as starting values.
The values of n' and K' at fracture entry temperature and at time at bottomhole temperature are
entered into the simulator. Also, a value of Cw for the fluid system or a total fluid loss coefficient,
CT, representative of the particular fluid system in a particular reservoir (ideally obtained from
minifrac testing) are input. The simulator is then run. If the desired fracture geometry and conduc-
tivity are not attained, the design engineer must examine the simulator results and adjust the fluid
system accordingly. For example if the job screens out prematurely, a new fluid could be chosen
which gives more viscosity; a fluid-loss additive could be added to lower the fluid loss coefficient;
or the maximum proppant concentration could be reduced. If the resulting conductivity is too
small, the engineer could try a more viscous fluid which would give wider fracture widths (if
height growth is not a problem); a cleaner fluid which gives less conductivity impairment; or a
larger maximum slurry proppant concentration. If the calculated pump time is substantially less
than the viscous life of the fluid (the time at bottomhole temperature during which the fluid vis-
cosity at 170 1/s exceeds 50 cp), the engineer may try repeating the simulation with a less viscous
fluid. This is particularly desirable when pumping water base gels where less polymer means less
fluid expense and better ultimate conductivities at a given fracture proppant concentration.
This method of fluid scheduling is essentially a trial and error approach, involving more or less
iterations depending on the engineers familiarity with the particular fracturing simulator and the
formation. The next method is more suited for jobs where multiple fluid types or stages are uti-
lized.
After the desired fracture length and fracture conductivity have been calculated by the simulator,
the engineer can schedule a fluid system. Fig. 6.36 shows the fluid-element time at temperature vs.
volume pumped calculated by a simulator using an assumed constant viscosity, such as recom-
mended above (i.e. 200 cp initially and 50 cp thereafter). The engineer then uses the rheological
data for the selected fluid system (e.g. the X-CEL gel system in Fig. 6.37) to mark the times at tem-
perature corresponding to when the particular fluid reaches 50 cp. These times are then marked off
on the ordinate of Fig. 6.36 and extended horizontally to intersect the time-at-temperature curve.
The fluid volumes corresponding to the intersection points define the stages of various polymer
loading for the fluid system. Fig. 6.36 shows the resulting gel schedule marked off on the abscissa.
The pad consists of 20,000 gal of a 50 lb crosslinked gel plus 60,000 gal of a stabilized 50 lb
crosslinked gel (X50S). The X50S gel continues for another 40,000 gal up to the 4 ppg proppant
stage. Then 40,000 gal of crosslinked 50 lb gel followed by 30,000 gal of crosslinked 40 lb gel and
50,000 gal of crosslinked 30 lb gel complete the pumping of proppant through the 10 ppg stage.
The middle sand stages on Fig. 6.36 have been reduced below the theoretical to account for addi-
tional slurry dehydration.
Using the specified gel schedule, the engineer inputs the rheology for the selected fluid system (n'
and K' as functions of time at temperature) plus the Cw appropriate for each stage and reruns the
simulator. If the simulated results meet the engineers specifications of length and conductivity,
then the design is completed. If not, the engineer must make appropriate rheology, fluid loss,
and/or maximum slurry proppant concentration modifications.
For a class problem, plot the data in Table 6.12 on the semilog graph paper in Fig. 6.38 to create
an exposure time plot similar to Fig. 6.37. Also, indicate the optimum time exposures for each of
the fluid stages.
Warning:
The testing of crosslinked gels is very difficult, with highly varying results from test to test. Some
service company data result from gels that were conditioned to simulate the flow history down the
tubular goods before testing on the viscometer. Viscosities of conditioned gels can be substantially
different from those of unconditioned gels. Also, if breakers are required for a fluid system (e.g.
for temperatures less than 250F), they should be added to all proppant laden portions of the fluid.
Breakers can significantly lower a fluids viscosity while pumping, and therefore, the viscosity vs
time at temperature plots (e.g. Fig. 6.37) should be adjusted accordingly. Encapsulated breakers
are now available which slow and/or delay the release of the breaker to avoid premature viscosity
breaking and to allow high breaker concentrations for better gel breaking.
The uncertainty in some of the data can be overcome by comparing similar systems for different
companies and using field experience. Fig. 6.39 - Fig. 6.41 provide guidelines for three common
fluid systems. These guidelines include the comparisons with various companies and have been
successfully applied in the field. Guidelines include the use of viscosity stabilizers at higher tem-
peratures and longer exposure times. The use of stabilizers allows higher viscosity to be main-
tained without using additional polymer.
Another guideline, which has been successfully applied, is for pad fluids. This guideline is shown
on Fig. 6.42 and was based on gel stability at high temperatures for an effective wall cake to con-
trol fluid loss. However, this guideline may be too conservative for the high temperature fluid sys-
tems currently available. Also, recent data for fluids with 5% hydrocarbon (generally used during
pad or first half of treatment with X-L gel and low k) indicate that polymer loading may not sig-
nificantly affect fluid loss. These reservations concerning Fig. 6.42 should be evaluated if fluid
schedules are developed which differ from the guideline.
Using the prior guidelines for fluid scheduling and the example plot of simulator results shown in
Fig. 6.43 develop a complete fluid and proppant schedule assuming the fluid system is Western's
APOLLO II/APOLLO II H system (Fig. 6.41). Assume the results in Fig. 6.43 are calculated
assuming the minimum viscosity requirements discussed previously.
Since frac tanks are generally 500 BBLS (20,000 gals), fluids should be selected in 20,000 gallon increments or
another increment which is convenient for the tank size actually used.
6.3 References
1. Ely, J. W.: Stimulation Treatment Handbook, An Engineers Guide to Quality Control, PennWell Publishing Co.,
Tulsa, OK (1985).
2. RP 39, Recommended Practice for Standard Procedures for the Evaluation of Hydraulic Fracturing Fluids,
API, Dallas (1983).
3. Biot, M.A. and Medlin, W.L.: Theory of Sand Transport in Thin Fluids, paper SPE 14468 presented at the
1985 SPE Annual Technical Conference and Exhibition, Las Vegas, Sept. 22-25.
4. Medlin, W.L., Sexton, J.H., and Zumwalt, G.L.: Sand Transport Experiments in Thin Fluids, paper SPE 14469
presented at the 1985 SPE Annual Technical Conference and Exhibition, Las Vegas, Sept. 22-25.
5. Daneshy, Ali: Proppant Transport, Monograph Series, SPE, Richardson, TX (1989) 12, vii, 210-22.
6. Shah, S. N.: Proppant Settling Correlations for Non-Newtonian Fluids Under Static and Dynamic Conditions,
SPEJ (April 1982) 164-70.
7. Novotny, E.J.: Proppant Transport, paper SPE 6813 presented at the 1977 SPE Annual Technical Conference
and Exhibition, Denver, Oct. 9-12.
8. Penny, G.S. and Conway, M.W.: Fluid Leakoff, Monograph Series, SPE, Richardson, TX (1989) 12, vii,
147-76.
9. Cameron, J.R.: Fluid Loss Testing on East Texas Cotton Valley Sand Cores to Determine the Effects of Diesel
and Polymer Concentration; With Consideration on Design Values of Spurt Loss and the Overall Fluid-Loss Co-
efficient, Amoco Production Company Report F88-P-21 (July, 1987).
10. McGowan, J.M. and McDaniel, B.W.: The Effects of Fluid Preconditioning and Test Cell Design on the Mea-
surement of Dynamic Fluid Loss Data, paper SPE 18212 presented at the 1988 SPE Annual Technical Confer-
ence and Exhibition, Houston, October 2-5.
11. Cooke, C.E., Jr.: Effect of Fracturing Fluids on Fracture Conductivity, JPT (Oct. 1975) 1273-82; Trans.,
AIME, 259.
12. Cameron, J.R.: Vol% Residue of HPG Vs. Guar in Borate Crosslinked Gels and Flow Impairment Models Based
on Vol% Residue and Fracturing Design Parameters, Amoco Production Company Report F90-P-41 (Feb.
1990).
13. Penny, G.S.: Evaluation of the Effects of Environmental Conditions and Fracturing Fluids on the Long-Term
Conductivity of Proppants, paper SPE 16900 presented at the 1987 SPE Annual Technical Conference and Ex-
hibition, Dallas, Sept. 27-30.
14. Small, J., et. al.: Improving Fracture Conductivities with a Delayed Breaker System: A Case History, paper
SPE 21497 presented at the 1991 SPE Gas Technology Symposium, Houston, Jan. 23-25.
15. Cameron, J.R. and Prudhomme, R.K.: Fracturing-Fluid Flow Behavior, Monograph Series, SPE, Richardson,
TX (1989) 12, vii, 177-209.
16. Brannon, H.D. and Ault, M.G.: New Delayed Borate-Crosslinked Fluid Provides Improved Fracture conductiv-
ity in High-Temperature Applications, paper SPE 22838 presented at the 1991 SPE Annual Technical Confer-
ence and Exhibition, Dallas, Oct. 6-9.
17. Roodhart, L.P. and Davies, D.R.: Polymer Emulsion: The revival of a Fracturing Fluid, paper SPE 16413 pre-
sented at the 1987 SPE/DOE Low Permeability Reservoirs Symposium, Denver (May 18-19).
18. Sinclair, A.R., Terry, W.M., and Kiel, O.M.: Polymer Emulsion Fracturing, JPT (July 1974) 731-38.
19. Harris, P.C., Klebenow, D.E., and Kundert, D.P.: Constant Internal Phase Design Improves Stimulation Re-
sults, paper SPE 17532 presented at the 1988 SPE Rocky Mountain Regional Meeting, Casper, WY, May 11-13.
21. Ely, J.W.: Fracturing Fluids and Additives, Monograph Series, SPE, Richardson, TX (1989) 12, vii, 131-146.
22. Cameron, J.R. and Gardner, D.C.: Suggested Amoco Procedure for Testing Titanium and Zirconium
Crosslinked Gels, Amoco Production Company Report F90-P-73 (Oct. 1990).
EQUIPMENT
ZONE A ZONE B
MILEAGE
33000 All units excluding sand and chemical delivery from
the nearest SES operating point, one way, per unit,
per mile . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.48 2.48
* Priced on Request
CHEMICAL ADDITIVES
ZONE A ZONE B
BACTERIA CONTROL
34000 BCS-1; per gallon . . . . . . . . . . . . . . . . . . . . . . . . . . . 33.83 33.83
34010 BCS-2; per gallon . . . . . . . . . . . . . . . . . . . . . . . . . . . 55.31 55.31
34020 BCS-3; per gallon . . . . . . . . . . . . . . . . . . . . . . . . . . . 32.23 32.23
34030 BCS-4; (Dryocide), per pound. . . . . . . . . . . . . . . . . . 16.50 16.50
34031 BCS-5; sodium hypochlorite, per gallon . . . . . . . . . . 6.48 6.48
34033 BCS-7; (X-CIDE 600), per pound . . . . . . . . . . . . . . . 52.80 52.80
BUFFERS
34080 BW-1; per pound . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.90 1.90
34090 BW-3; per pound . . . . . . . . . . . . . . . . . . . . . . . . . . . . .43 .43
34100 BW-4; per pound . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.40 2.40
34110 BW-5; per pound . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.95 1.95
34120 BW-6; per pound . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.68 2.68
34130 BW-9; per pound . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.12 1.12
34140 BW-10; ammonium chloride, per pound . . . . . . . . . . 1.18 1.18
34145 AA-11; caustic soda, per pound . . . . . . . . . . . . . . . . *P.O.R. *P.O.R.
* Priced on Request
CHEMICAL ADDITIVES
ZONE A ZONE B
CROSSLINKERS
34180 CX-1; aqueous gel, per gallon . . . . . . . . . . . . . . . . . . 40.99 40.99
34189 CX-4; low temperature, low pH, per gallon . . . . . . . . 45.00 45.00
34190 CX-5; low pH, per gallon . . . . . . . . . . . . . . . . . . . . . . 70.06 70.06
34200 CX-6; cold water, per gallon . . . . . . . . . . . . . . . . . . . 31.71 31.71
34210 CX-12; brine water, per gallon. . . . . . . . . . . . . . . . . . 34.00 34.00
34220 CX-13; high pH, per gallon . . . . . . . . . . . . . . . . . . . . 16.17 16.17
34230 CX-14; high temp., per gallon . . . . . . . . . . . . . . . . . . 24.23 24.23
34240 CX-15; high temp., per gallon . . . . . . . . . . . . . . . . . . 32.38 32.38
34245 CX-16; aqueous gel, per gallon . . . . . . . . . . . . . . . . . 51.81 51.81
34250 CX-91; aqueous gel, per gallon . . . . . . . . . . . . . . . . . 30.96 30.96
34251 CDA-2; crosslink delay additive, per gallon . . . . . . . . 47.85 47.85
34252 CX-DB2; high temperature delayed borate crosslink-
er, per gallon . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29.70 29.70
34253 DBX-1; delayed borate crosslinker, per gallon . . . . . 4.99 4.99
34254 RM-18; high pH boric acid, per pound. . . . . . . . . . . . *P.O.R. *P.O.R.
* Priced on Request
6
Fluid Selection and Scheduling
CHEMICAL ADDITIVES
ZONE A ZONE B
DIVERTING AGENTS
34255 DA-1; rock salt, course, per pound . . . . . . . . . . . . . . .20 .20
34260 DA-2; naphthalene, per pound . . . . . . . . . . . . . . . . . 2.82 2.82
34270 DA-3; benzoic acid flakes, per pound . . . . . . . . . . . . 2.11 2.11
34280 DA-4; rock salt, graded, per pound . . . . . . . . . . . . . . .20 .20
34290 DA-5; wax beads, per pound. . . . . . . . . . . . . . . . . . . 4.22 4.22
34300 DA-6; paraformaldehyde flakes, per pound. . . . . . . . 3.45 3.45
EMULSIFIERS
34330 PEM-1; water external, per gallon . . . . . . . . . . . . . . . 36.00 36.00
34340 PEM-3; 5% hydrocarbon systems, per gallon . . . . . . 28.14 28.14
* Priced on Request
6-57
CHEMICAL ADDITIVES
ZONE A ZONE B
FRICTION REDUCERS
Hydraulic Fracturing Theory Manual
References
DIESEL FUEL
Table 6.3 - Smith Energy Services Fracturing Services & Products.
Hydraulic Fracturing Theory Manual
6
Fluid Selection and Scheduling
34495 DIE-1; number one diesel, per gallon . . . . . . . . . . . . *P.O.R. *P.O.R.
34496 DIE-2; number two diesel, per gallon . . . . . . . . . . . . *P.O.R. *P.O.R.
* Priced on Request
CHEMICAL ADDITIVES
July 1999
July 1999
ZONE A ZONE B
SURFACTANTS
34580 FRS-1; fluid recovery surfactant, per gallon . . . . . . . 55.83 55.83
34585 FRS-2; fluid recovery surfactant, per gallon . . . . . . . 52.25 52.25
34590 FRS-3; fluid recovery surfactant, per gallon . . . . . . . 64.76 64.76
34591 FRS-4; fluid recovery surfactant, non-fluorocarbon,
per gallon. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20.79 20.79
34598 MCFRS;methanecoalfluidrecoverysurfactant,pergallon
SAA-1; per gallon . . . . . . . . . . . . . . . . . . . . . . . . . . . 35.97 35.97
34600 SAA-2; per gallon . . . . . . . . . . . . . . . . . . . . . . . . . . . 21.71 21.71
34610 SAA-3; per gallon . . . . . . . . . . . . . . . . . . . . . . . . . . . 18.22 18.22
34620 SAA-4; per gallon . . . . . . . . . . . . . . . . . . . . . . . . . . . 26.40 26.40
6-59
CHEMICAL DELIVERY
34660 For chemical delivery to location, per ton mile . . . . . .91 .91
34665 Minimum delivery charge for all chemicals . . . . . . . . 125.00 125.00
Hydraulic Fracturing Theory Manual
CHEMICAL RETURN
34666 For chemical return from location, per ton mile . . . . . .91 .91
34667 Minimum return charge . . . . . . . . . . . . . . . . . . . . . . . 125.00 125.00
References
34680 Liquid chemicals, per 55 gallon drum . . . . . . . . . . . . 121.00 121.00
34681 Liquid chemicals, per 5 gallon container . . . . . . . . . . 11.00 11.00
Table 6.3 - Smith Energy Services Fracturing Services & Products.
Hydraulic Fracturing Theory Manual
6
Fluid Selection and Scheduling
CHEMICALS NOT INCLUDED IN PRICE LIST
34999 Chemicals not included in Smith Energy Services
price list . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . *P.O.R. *P.O.R.
* Priced on Request
6-60
July 1999
July 1999
NATURAL
PLANT EXTRACTS
LARCH ARABINOGALACTAN
PECTIN
SYNTHETIC
ETHYCELLULOSE POLYVINYLPYRROLIDONE
CARBOXYMETHYL HYDROXYETHYL
CELLULOSE (CMHEC)* POLYACRYLIC ACIDS & SALT
ETHYLHYDROXYETHYLCELLULOSE POLYACRYLAMIDES*
References
STARCH AMYLOSE
Table 6.4 - Water Soluble Polymers.
Hydraulic Fracturing Theory Manual
6
NATURAL
STARCH AMYLODPECTIN
STARCH DEXTRINS
CARBOXYMETHYL HYDROXYPROPYL
GUAR (CMHPG)*
HYDROXYETHY GUAR
y
6-62
Guar Double
Pods Purified
Splits
Guar Single
Seeds Purified
Splits
References
Fig. 6.15 - Principal Guar Derivatives.,
Hydraulic Fracturing Theory Manual
6
Carboxymethyl Hydroxypropyl Guar
(Generalized Structure)
Guar Gum
HPG
CMHPG
Cellulose Derivatives
HEC
CMC
CMHEC
Polyacrylamides
July 1999
July 1999
References
Fig. 6.17 - Flow Behavior Index (n') vs. Temperature of Halliburtons HPG Solution.
6
6-66
Fig. 6.18 - Consistency Index (K'a) vs. Temperature of Halliburtons HPG solution.
July 1999
July 1999
6-67
Fig. 6.19 - Comparison of 40-lbm/1,000-gal Hpg Gels Crosslinked with Titanium Acetyl Acetonate
Subjected to Various Turbulent Flow and Temperature Histories.
Hydraulic Fracturing Theory Manual
References
6 Fluid Selection and Scheduling
Fig. 6.20 - Viscosity of Halliburtons Boragel (Borate Crosslinked Guar) as a Function of Time at
225 F.
Crosslinking Guar
D
0.8 B
C
A
n' 0.6
0.1
K'a
A
0.01
B
D
0.001
0 1 2 3 4 5 6
Time (hr)
Fig. 6.22 - Power Law Data for Halliburtons Versagel HT Fluid 250 F.
Proppant Loading Emulsion Quality Slurry Viscosity Emulsion Quality Slurry Viscosity
(lb/gal) (-) (mPa.s) (-) (mPa.s)
Proppant Loading Emulsion Quality Slurry Viscosity Emulsion Quality Slurry Viscosity
(lb/gal) (-) (mPa.s) (-) (mPa.s)
= 176 F
1000
75
80
70
(Viscosity, cp @ 511 sec-1)
80
60 75
50 70
100
60
50
10
70 90 110 130 150 170 190 210
Temperature (F)
Fig. 6.25 - The Effect of Shear Rate on Fig. 6.26 - Viscosity vs. Temperature for
Polymer Emulsion Viscosity. Western Super K-Frac (Polyemulsion).
Fig. 6.27 - Plot of Viscosity of a 0.67 Quality Emulsion Vs. Mean Droplet Size.
Fig. 6.28 - Power-law Data for a Water/N2 Foam Stabilized With 40 lbm Thickener/1,000 Gal Water.
Fig. 6.29 - Effect of HPG Concentration (lbm/1,000 gal) on the Viscosity of a 0.70-Quality Foam.
400 400
Friction Pressure - psi per 1000 ft
300 300
Foam Quality
200 200
0.85
0.80
100 100
90 90
80 0.75 80
70 70
60 60
50 0.70 50
40 40
0.65
30 30
20 20
0.60
0.55
10 10
1 2 3 4 5 6 7 8 9 10 20 30 40 50 60 70 80 90 100
Fig. 6.32 - Comparison of the Solubility of Carbon Dioxide and Nitrogen in Water.
Fig. 6.35 - Pressure Effect on a Partially Gelled Diesel at Ambient Temperature and at 180 F [data
by J. R. Cameron, courtesy Amoco Production Co. Research, Tulsa, OK (1986)].
(cp)
Viscosity (1bf-secn'/ft2
MY-T-OIL IV (Continuous-Mix System) @ 170 1/s n' K'
(cp)
Viscosity (1bf-secn'/ft2
MY-T-OIL II (Batch System) @ 170 1/s n' K'
Crude Oil 13 4 3.5 71 2.5 302 215 0.13 0.11 0.550 0.433
Frac Oil 200 7 2.4 2.0 71 2.2 102 81 0.25 0.12 0.102 0.150
Frac oil 200 8 2.7 3.0 70 2.2 155 81 0.19 0.13 0.289 0.149
Table 6.9 - Power-Law Rheology as a Function of Temperature, Western Maxi-0-74 Gelled Oil
System.
Temperature Viscosity
Maxi-0-74 Gel F Time n' K' 170 sec-1
Dowell
Composition BJ Services Schlumberger NOWSCO LTD Halliburton Smith Western
Water and Friction reducer Aqua Frac River Frac Friction Reducer Available Slick Water Water Frac
Gelled water Gelled Water Water Frac Gelled Water Water Frac Gelled Water Gelled Water
Aqua Frac WF-100 (guar) WGA-6
WF-200 (HPG) WGA-7
Gelled water with a fluid loss Gelled water plus Redifrac Gelled water & Water Frac plus Gelled Water (Maxi-Pad)
additive FLA Fluid Loss FLA with (Westpad A)
Logel 100 & FLA FL Additive available
Low residue gelled water (HPG) GW-8 WF200 LSR-1NB WG-11, 12, WGA-2 WGA-8 J-12
GW-32 HYG-5, (J-16) J-20
WG-20
No residue gelled water (HEC) GW-21 YFHC HEC Hygel 100,300 & WGA-3 Plus Gel
500, LOGEL 100 (J-5) J-6
WG-17, WG-21
Crosslinked guar system additive Ultra Frac YF-100 Hy Vis MY-T-GEL GWX-7 LT, VIKING I
Terra Frac YF100D MY-T-GEL LT GWX-7 HT VIKING I DHT
(Delayed) MY-T-GEL HT GWX-9 APOLLO I
100 - Borate Hybor Gel
YF-300 KO Gel
(Titanate) Thrifty Gel
YF-500-HT
(Zirconate Delayed)
Thin prepad with buoyant diverting agent Invertafrac Available
to control upward growth
Crosslinked HPG with 3-5% Terra Frac-D Stratafrac II Service Ultravis LPW Versagel GDX-7 APOLLO II H
hydrocarbon for fluid loss (Available with most +5% Diesel plus Diesel SATURN II H
systems)
Crosslinked HPG with high temperature Terra Frac YF400 ThermoVis Versagel HT GWX-7HT Saturn Gel
stabilizers RXL II YF600-HT
Crosslinked CMHEC Krystal Frac HyClean Kleen Gel Available APOLLO IV LPH
RXL
Krystal Frac XL
Krystal
Frac-D
(5% Diesel)
Dowell
Composition BJ Services Schlumberger NOWSCO LTD Halliburton Smith Western
Crosslinked CMHEC for high tempera- Super Krystal Kleen Gel II WZ-100018 Gel
ture Frac
Crosslinked guar or HPG with Borate Ultra Frac YF-100 (guar) HyVis GWX-9 VIKING II
YF-200 (HPG) Boragel Guar or VIKING II DHT
Hybor Gel HPG VIKING I
VIKING I DHT
CO2 compatible fracturing fluid Krystal Frac YFLPH PurGel II, &III GWX-4LT SATURN II LT
(CMHEC) (HPG) ACIDGEL Frac GWX-4HT
Super Terra Frac ACIDGEL Frac II
Versage LT
KleenGel,
MY-T-GEL LT
KlexenGel II
Alcogel I
MY-T-Oil I, II, &III
LOGEL 100
HYGEL 100 &
300
Economical, low residue cross-linked Terra Frac T YF-LPH Ultravis LPW Pur-Gel WGA-5 APOLLO I
system (Low pH system) GWX-5
Controllable delayed crosslink HPG sys- Terra Frac RXL YF-600-HT Ultravis H-T Versagel HT GWX-7 SATURN II
tem II & CL-18 APOLLO II
Hybor Gel
Controllable delayed crosslinked high BJ-Titan RXL YF600-HT (HPG) Thermo-Vis Thermagel GWX-7HT SATURN II
temperature system SpectraFrac G YF500-HT (Guar) Pur-Gel III APOLLO II
(CMHPG)
Crosslinked water-alcohol system Metho Frac Alcohol Ultravis LPW Alcogel II-X Available WZ-100013 Gel
(G-8) Waterfrac
(J-160)
Oil Systems
Oil without viscosifier Available Crude Frac Sandoil Available Oil Frac
Gelled Oil Oil Based Petrogel Hycar 2000 Viso-O-Frac Gelled Oil Low Friction
Ultra Frac V-O-Gel Frac
Crosslinked gelled oil for medium tem- Allo Frac YF-GO III HLG-1 My-T-Oil II PGO-1 Maxi-0-74 Gel
perature HLG-5
Crosslinked gelled oil for higher temper- Allo Frac HT YF-GO IV My-T-Oil III PGO-1 Maxi-0-86 HT
atures Gel
Water external emulsion developed by Polyemulsion Super Sand Polyemulsion Super WEP-1 Super K-Frac
Exxon Frac K-1 Emulsifrac
Dowell
Composition BJ Services Schlumberger NOWSCO LTD Halliburton Smith Western
Foamed Systems
Water and nitrogen foam with or without Aquafoam Foamfrac Foam Frac Foamfrac Foam West-Foam, N2
gel Stabilized AquaFoam Frac
Foam Soulution
(SFS)
Acid and nitrogen foam Etching foam Available Foamed Acid Available FAS-1 West-Foam, N2
Hydrocarbon and Nitrogen foam Available Available Foamed Hydro- N10 Frac Foamed Petro Foam
carbon Oil
Frac
Water and CO2 foam Available Available Poly-CO2 C-O-TWO Frac CDM-1 WestFoam, CO2
Pur-Gel II GWX-4LT
Pur-Gel III GWX-4HT
Powdered guar gum polymer. Delayed J111, J424 G-308WB WG-19 WGA-6 J-2, J-4
hydration, designed for batch mix appli- GW-27 J877 WG-22a
cations. WG-23
Powered Hydroxypropylguar gum. GW-32 J347 LSR-1NB WG-11 WGA-2 J-12 (J-18)
Delayed hydration polymer, designed for J362 WGA-8 J-20
batch mix applications. No internal J456
breaker. J876
Dowell
Composition BJ Services Schlumberger NOWSCO LTD Halliburton Smith Western
Chemically modified natural polymer for GW-20 J-271 G-317 MGA-1 WZ-100313
up to 80% methanol. GW-25
GW-35
Chemically modified natural polymer for GW-55 LSR-5 WG-20 MGA-1 Available
gelling 100% methanol
Liquid viscosifier for acid up to 15% AG-12 J425 DSGA SGA AGS-1 Acigel Lt
(15-28%) Liquid AGA-1 (low temp)
M33 AGA-2
AGA-4
AGA-5
Dowell
Composition BJ Services Schlumberger NOWSCO LTD Halliburton Smith Western
Friction Reducers
Liquid friction reducer for hydro-carbons J257 F-100 FR-5 OFR-1 FR-5AW
FRO-18 FR-7
Requires Requires
Activator Activator
Fluid Loss
Selectively graded fine mesh silica flour FLC-8 J84 Silica Flour WAC-9 WFL-2 F-11
used in water, oil and acid J418
Combination of graded oil soluble resin FLC-1 J238 WAC-11D AFL-2 Frac Seal M
and degradable low mole- cular weight
polymers. Non-damaging fluid loss addi-
tive used in water and acid
100 mesh benzoic acid used in water, FLC-1 J227 Available Available Flakes-DA-3 Available
acid or foam fracturing treatments. (Particulate)
100 mesh sand used in water, oil and 100 mesh FLA100 100 mesh 100 mesh 100 mesh 100 mesh
acid sand S100 sand sand sand sand
100 mesh oil soluble resin used water FLC-2 FLA10005 FL-30 OSR-100 AFL-3 FracSeal
and acid
100 mesh salt 100 mesh Available DA-4 100 mesh salt
sand AFL-2
Fluid loss additive for water and oil WAC-10 AFL-4 Aquaseal 2
Proprietary liquid fluid loss solution FLC-15 J-451 WAC-12L WFL-4 Aquaseal L
FLC-17 FLD-1
Fluid loss additive used in water and oil Adomite J110 Adomite Adomite WFL-1 Available
(Adomite Aqua) Aqua Aqua Aqua
Fluid loss additive used in oil base fluids Adomite J126 Adomite Adomite OFL-1 Adomite
(Adomite Mark II) Mark II Mark II Mark II Mark II
Dowell
Composition BJ Services Schlumberger NOWSCO LTD Halliburton Smith Western
Liquid fluid loss additives for use in oil Available Available Available Available Available Available
wells with water based fluids from
80-300 F (diesel or other hydrocarbon)
Breakers
Enzyme breaker for guar, guar deriva- GBW-10 J134 Breaker F GWV-3 WEB-2 B-11, B-11L
tives and cellulose derivatives GBW-30
Oxidizer breaker for guar, guar deriva- GBW-5 J218 Breaker S SP Breaker WCB-1 B-5
tives and cellulose derivatives
Acid breaker for guar, guar derivatives GBA-1 Breaker H MYF-5 P-4
and cellulose derivatives
Breaker for phosphate ester oil gels GBO-3 J318, YF 60 II Breaker MO II OXB-3 B-15
J-295 YF60 II K34 B-16
III B-25
J603 YF60 III
Gel breaker and filter cake degrader. Optikleen
Treatment follows water based fracturing
fluids. Used from 80-270 F.
Breaker for phosphate ester gels GBO-6 J295 K-34 OXB-3 Sodium
(YF-GO II, IV Bicarbonate,
J-603, J860, GO III B-25
Diverting Agents
Oil soluble resin in aqueous solution FLC-11 J237 L-12 Matriseal-0 AFL-1 ASP-530
Graded rock salt Rock Salt J66 Rock Salt TBA-110 DA-4 Westblock,
Salt-Trimix S-6
Dowell
Composition BJ Services Schlumberger NOWSCO LTD Halliburton Smith Western
Flake Benzoic Acid Benzoic Acid J227A Benzoic Acid TLC-80 DA-3 Westblock 3X
Super Flake &4
Regular
Oil soluble graded napthalene Moth Balls J116 TLC-15 DA-2 S-3
Diverting agent used in acid FLC-18 (Concentrate) Matriseal 0
(Solution) Matriseal
OSR-100
TBA-350
TLC-80
TBA-100
Matriseal OWG
TLC-155
Polymer Plugs
Emulsifiers
Oil external emulsifier for HCl and E U74 (D.A.D. acid), DL-22 AF-61 AAE-1 E-9
HCl-organic mixtures U60 (Super Sand
Frac), U80
Emulsifier for polyemulsion E-2, E-5 U78A (not for diesel) WS-50 SEM-5 PEM-1 Wellaid 266
SEM-6
SEM-7
Emulsifier for polyemulsion and CO2 FAW-16 U78E EF-10 SEM-5, ACO-1 (E-15), E-16
emulsion or CO2 foams. SEM-7
HC-2 PEM-1
AQF-1 FAA-2
AQF-2
AQF-4
Clay Stabilizers
Dowell
Composition BJ Services Schlumberger NOWSCO LTD Halliburton Smith Western
Cationic polymer for stabilizing clays Claytrol5 L53 (W-Winterized) CSA-6 Cla Sta II CCC-3 Claymaster 4
L42 Cla Sta 0
Cla Sta FS
Cla Sta XP
Surfactants
Nonionic fluorosurfactant for water and Inflow50 F-75N WS-70 SuperFlo II FRS-2 Flo Back 10
acid systems FRS-3 FS-2
USS-N
Cationic fluorosurfactant for water and Inflo 45 TEA-380 EnWaR-288 FRS-1 (FS-1)
acid systems Inflo 100
Nonemulsifiers
Nonionic nonemulsifier for water and NE-4 F40 W5-6 One LOSURF-251 EPS-4, EPS-5 Aqua Flow
acid NE-15 EZEPlo, W39 259, 300, 357 EPS-9 LT-17
NE-18 Pen-5, LOSURF SAA-2, SAA-8 Nine 40
S100 0
S200
S400
S600
Dowell
Composition BJ Services Schlumberger NOWSCO LTD Halliburton Smith Western
Anionic nonemulsifier NE-10 F78, M38 DL-22 TRI-S SAA-3 LT-5, AS-2
NE-31 W22, W27, Fracflo II SAA-7 LT-25, LT-31,
NE-32 W39, M38W MorFlo II F-Flow,
S-500 Parasol D
Cationic nonemulsifier for water and acid NE-1 AI-170 Cationic N SAA-4 I-5, LT-22
NE-7 Compounds SAA-1 LT-17, WK-1
NE-2 EPS-1 F-Flow,
NE-6 EPS-3 Parasol D
NE-9 F75N EPS-6
NE-11 (nonionic)
NE-12 marketed as
NE-13 Ezeflo F75
NE-20
NE-21
NE-22
Nonionic fluorosurfactant for water and Inflo50 Superflo USS-N FS-2 (FS-F)
acid
Nonionic surfactant and nonemulsifier D-4 F40 Pen-5 EPS-4, EPS-5 LT-21
for water and acid Also foaming EPS-9, SAA-8
agent for acid
Anionic nonemulsifier for water and DL-26 Fracflow, SAA-7 LT-5, LT-25
acid 3N
Fines Suspender
Anti-Sludge Agent
Anti-sludge agent for acid W35 DL-22 AS-5, AS-6 SPS-1 AS-2, LT-31
(W50) DL-26 AS-7, AS-8
pH Control
Strong base D-2 J465, M2, U28 Caustic Soda Caustic Soda Caustic Soda Caustic Soda
U28, J-221 (G-5,G-6)
(2% caustic)
Synergistic additive for extending inhibi- D-2 L6 Formic Acid MYF-2L Formic Acid WTI-25
tion times at elevated temperature L36 WTI-26
Dowell
Composition BJ Services Schlumberger NOWSCO LTD Halliburton Smith Western
Buffers (proprietary) BF-1, BF-5 M47 BA-10, BA-20 BW-1, BW-5 Buffer 1
BF-2 BA-30 BW-7, BW-10 Buffer 2
BF-3 BA-40 Buffer 4
BF-4
Crosslinkers
Proprietary crosslinking agent (Ti) XLW-39 (J352) ATX-25 CL-11 CX-1, CS-91, CL-9, T.I.C.,
CL-18 CS-6 CL-12
Proprietary crosslinking agent (Borate) XLW-1, XLW-2 L10 (Powder) BXL-1W CL-22 CS-13 (2-C, Powder),
(liquid) CL-2
Proprietary crosslinking agent Zr XLW-52 J366, J367 2R-XL CL-24 CX-7, CX-14 CL-14, CL-14W
(Activator) CL-15 CX11A, CS-15 CL-11
J444 (Temp. Acti- CL-21 CX-16
vated) CL-23
Foamers
Foaming agent for water and brine FAW-16 F52.1 TRI-S FAA-1 Foamex
(Water, Brine, Acid) FAA-2
Foaming agent for water and acids FAW-9 F78 (Foamer and SF-2 AQF-1, FS-2 (FS-F)
Fines Suspender) SGA-1, FAA-2
Pen-5
Foamer for hydrocarbons FAO-25 OFA-2 SNF-1 Petro Foam 1
Foaming agent for oil and conden- FAO-25 SF-3 OFA-2 SNF-1 Petro Foam 1
sates
Dowell
Composition BJ Services Schlumberger NOWSCO LTD Halliburton Smith Western
Foaming agent for water and meth- FAW-20 SF-8 ACO-1 SNF-4 Frac Foam 1
anol
Foaming agent for 100% methanol FAW-20 SF-8 ACO-1 SNF-4 Available
and methanol water mixtures
Scale Inhibitors
Scale Inhibitor ScaleTrol 4 L47, P-300 Phosphonate GSI-1 P-9
L49 Scale Inhib.
Scale Inhibitor ScaleTrol 6 L50 SST-245 Phosphonate GSI-1 P-8
Scale Inhib.
Scale Inhibitor ScaleTrol 8 L45 Phosphonate GSI-1 Ultra Sol II
LP-60
Scale Inhib.
Scale Inhibitor X-4 LP-55 GSI-1, GSI-2 P-7
GSI-3
Scale Inhibitor X-6 L35 Similar to GSI-2, GSI-1 P-2, P-3
LP-55 GSI-3
Dowell
Composition BJ Services Schlumberger NOWSCO LTD Halliburton Smith Western
Gel Stabilizer
Liquid stabilizer for high tempera- Methanol K46 Methanol Methanol, Methanol Methanol
ture Liquid
Gel-Sta
Powdered stabilizer for high temper- GS-1 J353 GS-1 Gel-Sta HTS-2 Gel Master
atures GS-2 HTS-2
GS-3
Stabilizer J59
Defoamer
Defoamer for aqueous fluids D-37L D47, AFA-Z AGD-2 DF-11
AntiFoamer-1 (Cold Water) AF-11
AF-11L
Defoamer for oil base fluids J291 RFP-1 DF-1
Biocides
Bactericide X-Cide X-cide BCS-2 Frac Cide 10,
102 M123 (Solid) 102W BCS-3 Frac Cide 2
BCS-4
Biocide X-Cide 207 BE-3 BCS-1 Frac Cide 20
Bactericide Adocide M155 Adocide Adocide Adocide
Biocide Adomall M76 Adomall Adomall Adomall
Biocide X-cide M-275 BE-4 Frac Cide 2,
207 Frac Cide 20
Dryocide
a. Especially for use in oil base slurry.
x40
Fig. 6.36 - Simulator Results of Fluid-Element Time at Temperature vs. Volume Pumped.
Super Gel System Time, Hours K' n' Viscosity at 170 sec-1
0 .00108 .95 40
X20 .25 .00017 1.0 8
.75
0 .02262 300
.80
.25 .00700 120
.5 .00339 .85 74
.75 .00188 .90 54
1.0 .00111 .93 37
X30SGS 1.5 .00049 .97 20
.65
0 .05294 420
.5 .01146 .75 152
1.0 .00320 .85 71
1.5 .00111 .93 37
X40SGS 2.0 .00051 .97 21
.65
0 .06932 .7 550
.5 .02194 225
.75
1 .00905 120
.8
1.5 .00496 85
2 .00311 .85 69
2.5 .00240 .87 59
3 .00178 .90 51
X50SGS 3.5 .00122 .95 45
.65
0 .08823 .7 700
.5 .03130 321
.75
1 .01486 197
.8
1.5 .00875 150
2 .00564 .85 125
2.5 .00428 .87 105
3 .00342 .89 93
X60SGS 3.5 .00286 .90 82
Fig. 6.38 - Class Example of Selecting Optimum Fluid for Time at Temperature.
10
7.1 Overview
The selection of a proppant for use in hydraulic fracturing is an economic decision requiring tech-
nical input. The purpose of this chapter is to provide the engineer with the technical capabilities to
make good economic decisions with respect to fracture design. This chapter is broken into several
sections.
First, the sources of the available fracture sands and commercial proppants are discussed. In addi-
tion, the size and quality of these materials is reviewed to provide the engineer with the technical
information required to make proppant decisions for fracture design. Next, the critical factors
which affect fracture conductivity are reviewed. Factors such as closure stress, size, concentration,
strength, shape, and gel residue effects can impact fracture conductivity and ultimately, well per-
formance. Finally, the economic aspects of proppants and/or fracture conductivity will be
reviewed.
7.2 Introduction
Historically, fracture stimulations have been performed for two reasons; to overcome the detri-
mental effects of wellbore damage and/or to stimulate the wells performance. The former reason
has been typically applied to wells in moderate to high permeability reservoirs and generally
resulted in the creation of short fractures. The latter generally resulted in the creation of long frac-
tures in wells in low permeability reservoirs. The success or failure of fracturing in either case
depended on whether or not the created fracture had adequate flow capacity so that the reservoir
fluids flowed to the fracture and then to the wellbore.1,2 If the flow capacity of the fracture was
large by comparison to the reservoir flow capacity, tremendous performance improvements would
be realized.
The purpose of the proppant is to keep the walls of the fracture propped apart so that a conductive
path to the wellbore is retained after pumping has stopped and fluid pressure has dropped below
that required to hold the fracture open. Ideally, the proppant will provide large enough flow capac-
ity to make negligible pressure losses in the fracture during fluid production. In practice, this ideal
might not be achieved because the selection of a proppant involves many compromises imposed
by economic and practical considerations.
The propped fracture must have a conductivity at least high enough to eliminate most of the radial
flow path that exists in an unfractured well and to allow linear flow from the reservoir into the frac-
ture. This requires relatively unimpeded linear flow within the fracture to the wellbore. To accom-
plish this, the proppant must enable the propped fracture to have a permeability several orders of
magnitude larger than that of the reservoir rock.
Figure 7.1 is a plot of steady state folds of increase versus fracture half length for a 40 acre drain-
age area. This plot shows the productivity improvement associated with a 1000 md-ft fracture ver-
sus an unstimulated well in a reservoir with permeabilities of 1 and 10 md. This figure clearly
indicates that there is no economic benefit associated with increasing fracture length beyond one
hundred feet in a 10 md reservoir while in a 1 md reservoir there is some economic benefit asso-
ciated with an increased fracture length. Figures 7.2 and 7.3 show similar plots for 2000 md-ft and
3000 md-ft fractures, respectively. Analysis of these figures indicates that, for any fracture length,
increases in fracture conductivity result in increased productivity. In a 10 md reservoir, for exam-
ple, a productivity improvement of 2.2 could be realized by creating a fracture of half length 100 ft
and conductivity of 1000 md-ft. A 2.6-fold increase could be realized by creating a fracture of the
same length with a 2000 md-ft conductivity. Creation of a 3000 md-ft fracture would result in a
2.7 fold production increase over an unstimulated well. Thus, increasing fracture conductivity
from 1000 md-ft to 3000 md-ft would result in an additional 23% production increase without sig-
nificantly increasing the treatment cost. It is this concept that underlies the importance of fracture
conductivity to fracturing. Performance improvements can be realized by improving conductivity
at little or not cost.
Steady State Folds of Increase
Fig. 7.1
Fig. 7.2
Steady State Folds of Increase
Fig. 7.3
Historical Perspective
One of the first proppants used in the early days of hydraulic fracturing during the late 1940s was
sand dredged from the Arkansas River. Initially, the sand was not cleaned and screened as todays
standards require, but as the need became evident, steps were taken to process the sand more thor-
oughly. During the mid-1950s, sand from the Saint Peter sandstone formation near Ottawa, Illi-
nois, entered the market.
As the need for a more economical and readily available fracturing sand grew, mines were opened
near Brady, Texas, in 1958, and production from the Hickory sandstone formation began to be
marketed. This sand, as well as most other high-quality sand used today, is mined from consoli-
dated sandstone formations. The mining process includes crushing, screening, and washing to sep-
arate the sandstone matrix into its individual sand grains. A wide range of particle sizes is found
in the deposits. Typically, only 20 to 30% of such deposits is found to be in a size range useful for
hydraulic fracturing applications.
The explosive growth of the hydraulic fracturing industry from the mid-1970s to the early 1980s
created shortages of fracturing sand. Supplies from the Saint Peter sandstone of Illinois were sup-
plemented by high-quality material from the Jordan, Ironton, and Galesville sandstones of Minne-
sota and Wisconsin. Similarly, sand from the Bidahochi formation in Arizona and aeolian dune
sand of Colorado augmented proppant production from the Hickory sandstone in Texas. Finally,
new sand-processing plants were constructed in Minnesota and Wisconsin specifically to produce
fracturing sand and to replace plants designed to supply sand for other applications. Table 7.1
highlights general information on available fracturing proppants. Figure 7.4 shows a plot of per-
meability versus stress for various 20/40 mesh proppants. As shown, the intermediate and high
strength proppants generally have greater retained permeabilities at higher stress levels than the
sands.
The subsequent sections will describe the physical properties of commercially available proppants
with the importance of these properties described in more detail in Section 7.5.
Brady-Type Sand
This rounded quartz sand, also known as brown or Texas sand, is mined from the Hickory sand-
stone in central Texas near the town of Brady. The Hickory sandstone was deposited during the
Upper Cambrian Age some 500 million years ago. The color of this sand results from small
amounts of iron oxide contamination in the crystal structure. Color variation has no bearing on the
strength of this sand or on any other sand discussed here.
As mined, the sand is polycrystalline; i. e., each whole grain is composed of more than one quartz
crystal bonded together, leaving cleavage planes in the whole grain. In terms of fines generated,
the API crush resistance test typically yields from <50 to as much as 85% of the API permissible
fines. The deposit yields acceptable fracturing sand in the 20/40 mesh size range and larger. Pro-
duction in sizes smaller than 20/40 mesh is not sized to meet API recommendations. Typical phys-
ical properties, fracture permeability, and pack-width data for this sand are presented in Table 7.1.
Table 7.1 Typical Physical Properties of Brady-Type Fracturing Sand*
30 minutes at 150 F, wt% 3.0 maximum 0.4 1.0 1.0 0.8 0.8
* All tests performed according to Reference 11 or 12. Sources include Saint Peter, Jordan, Galesville, and Ironton sandstones. Values shown are
averages of multiple production samples over a 4-year period.
The Bidahochi formation sand is mined from shallow, lightly consolidated lenses in eastern Ari-
zona. It was deposited during the Pliocene or Tertiary Age some 6 million years ago. This sand
contains grains of chert, which is stronger than quartz, along with rose and smoky quartz. Fractur-
ing sand from this formation is available in limited quantities in 12/20, 20/40, and 40/70 mesh
only.
The aeolian dune sand is mined in central Colorado from shallow, lightly consolidated lenses. This
sand was deposited during the Holocene Age less that 1 million years ago. The large sizes, 6/12
through 12/20 mesh, are as high in quality as those from the Hickory formation, but the small sizes,
16/30 through 70/140 mesh, contain so much feldspar that they produce excessive fines in the API
crush resistance test.
Ottawa-Type Sand
This well-rounded, very pure quartz sand exceeds API recommendations. In terms of fines gener-
ated, the API crush resistance test typically yields less than half of the maximum acceptable fines
on this sand. The sand also is monocrystalline. Crushed particles are primarily large chipped grains
rather than individual quartz crystals. Color variation is widespread in this sand, but has no impact
on its performance characteristics as a proppant. For the most part, the sand is well processed and
of high quality for fracturing applications. Typical physical properties, permeability, and pack-
width data of this sand are presented in Table 7.2.
Table 7.2 Typical Physical Properties of Ottawa-Type Fracturing Sand*
Top sieve 0.1 maximum 0.0 0.0 0.0 0.0 0.1 0.1
Between primary sieves 90.0 minimum 93.2 97.9 91.5 93.1 91.8 90.0
Second and sixth sieves 6.6 2.1 8.0 6.5 7.6 9.1
Pan 1.0 maximum 0.2 0.0 0.5 0.4 0.6 0.8
30 minutes at 150 F, wt% 3.0 maximum 1.5 1.0 1.0 0.9 1.2 2.5
* All tests performed according to Reference 11 or 12. Sources include Saint Peter, Jordan, Galesville, and Ironton sandstones. Values shown are
averages of multiple production samples over a 4-year period.
The Saint Peter sandstone, commonly known as Ottawa sand, was deposited in the Ottawa district
of Illinois during the Middle Ordovician Age some 460 million years ago. This sand is available
in 20/40 mesh and smaller sizes only. Color variation runs from white through gray-white to pale
yellow.
The Jordan sandstone was deposited in south central Minnesota and western Wisconsin during the
Upper Cambrian Age some 500 million years ago. Jordan fracturing sand is available only in
12/20 mesh and smaller sizes. The color varies from white through gray-white to pale yellow to
brown.
The Galesville and Ironton sandstones were deposited in south central Minnesota and western
Wisconsin during the Upper Cambrian Age some 500 million years ago. Ironton fracturing sand is
available in 12/20 mesh and smaller; the Galesville sand is available in 20/40 mesh and smaller
sizes only. Its color varies from white to light tan.
With the drilling of deeper wells, the shortcomings of glass beads and quartzitic materials as prop-
pants became apparent. Such materials are weakened by hot formation brines and tend to fail cat-
astrophically under high closure stress. These factors accelerated the search for improved
materials, and in the mid-1970s, a high-strength ceramic proppant, sintered bauxite, was intro-
duced. The inertness and strength of sintered bauxite are caused by its major constituent, corun-
dum, a form of aluminum oxide. Although expensive, sintered bauxite retains permeability under
very high stress and severe reservoir conditions better than any other proppant available today.
The expense of sintered bauxite motivated efforts to find less costly, but useful substitutes. Under
development at the same time as sintered bauxite, curable resin-coated sand was the first such
product to find application.
Research and development on other ceramic proppants during the early 1980s produced a less
expensive proppant containing mullite, another form of aluminum oxide, in addition to corundum.
It has helped to bridge the cost-performance gap between sand and bauxite. Because of its lower
cost and high performance, this material has enjoyed widespread use since its introduction.
Sintered Bauxite
As previously described, sintered bauxite is an inert, high-strength ceramic proppant. Patented by
Cooke et al., this high-density proppant is produced by the same manufacturing techniques as
refractory ceramics and metal-working abrasive grits. The raw material is primarily high-alumina
bauxite ore from South America. The ore is first ground to a particle size less than 15 m, shaped
into small ceramic pellets using water and a binder, and, after drying and screening, fired in a kiln
to bind the edges of the individual particles that make up each pellet. After the sintering process,
the color of the product varies from black to brown or gray. Typical physical properties, pack per-
meability, and width data for this proppant are presented in Table 7.3.
Table 7.3 Typical Physical Properties of Sintered Bauxite - High-Strength,
Sintered Ceramic Proppant17*
Silt and fine particle, FTU 250 maximum 80 100 100 120
Crush resistance, % fines generated Variable with size and stress
at 7500 psi 5.4 6.4 2.6 1.7
at 10,000 psi 10.6 12.2 4.3 3.0
at 12,500 psi 16.8 18.0 6.8 5.2
at 15,000 psi 22.5 23.2 10.7 7.3
Particle density, lbm/gal 28.4 maximum 30.88 30.88 30.88 30.88
3
Bulk density, lbm/ft 140.0 maximum 140.0 140.0 140.0 140.0
Clustering, wt% 1.0 maximum <1.0 0.0 0.0 0.0
* All tests performed according to Reference 11 or 12. Values shown are averages of multiple production samples over a 4-year period.
Sintered bauxite draws its strength from the unique manufacturing process and from the materials
present in the bauxite ore. Corundum, the major component of sintered bauxite, is one of the hard-
est materials known to man. It measures 9 on Mohs hardness scale. For comparison, quartz is 7
and diamond is 10. When crushed, bauxite does not shatter as completely as the sands; it simply
splits into large pieces that are still capable of providing flow capacity. This crush resistance is
caused partially by sintered bauxites elastic properties, which allow slight deformation before
failure under high stresses.
The first sintered bauxite proppants were angular in shape, which could cause increased abrasion
and failure of pumping equipment, treating lines, wellhead equipment, and chokes. Process
improvements have produced a material with roundness and sphericity values better than the best
fracturing sand and, thus, less abrasive than its predecessor. This proppant has become the standard
against which all other proppants are measured.
Intermediate-Density Proppant
Even though this material is often called intermediate-strength proppant, a more appropriate
term is intermediate-density proppant (IDP). The strength of this type of proppant is much closer
to that of sintered bauxite than to sand. While neither as strong nor as inert as sintered bauxite, this
material has an advantage over sintered bauxite in that it has a lower density (approaching that of
sand) than bauxite. The moderate density of these proppants makes them easier to transport and
place in the fracture than the denser sintered bauxite.
The search for a more economical replacement of sintered bauxite revealed that high-alumina,
domestic bauxitic ores could be used to produce a high-performance, sintered proppant with prop-
erties approaching those of sintered bauxite. In addition to corundum, this proppant contains mul-
lite, a less-dense mixed form of aluminum oxide. The result is a dark brown to tan proppant of
lower bulk density and lower specific gravity than bauxite. This new material is produced by man-
ufacturing techniques similar to those used for sintered bauxite. Typical physical properties, pack
permeability, and width data for this proppant are presented in Table 7.4.
Table 7.4 Typical Physical Properties of High-Strength, Intermediate-Density,
Sintered Ceramic Proppant17*
Silt and fine particle, FTU 250 maximum 100 100 100 120
Crush resistance, % fines generated Variable with size and stress
at 7500 psi 6.4 10.3 3.2 1.4
at 10,000 psi 13.6 19.4 6.0 2.7
at 12,500 psi 19.3 27.4 9.8 4.6
at 15,000 psi 26.9 33.9 14.3 7.4
Particle density, lbm/gal 28.4 maximum 26.29 25.95 25.62 26.12
3
Bulk density, lbm/ft 114.0 maximum 113.0 107.0 106.0 113.0
Clustering, wt% 1.0 maximum <1.0 <1.0 <1.0 <1.0
* All tests performed according to Reference 11 or 12. Values shown are averages of multiple production samples over a 4-year period.
** Currently available in limited quantities on special order only.
While the durability and strength of intermediate-density proppant are somewhat less than those
of sintered bauxite, performance is virtually equivalent in all but the deepest and hottest wells. At
high stress levels, the proppant breaks into large particles capable of providing good flow capacity.
The proppant particles have good resistance to corrosion by hot formation brines; their roundness
and sphericity are better than those of the best fracturing sands, while their bulk density is only
slightly higher.
Despite higher cost, intermediate-density proppants may replace sand at intermediate well depths
because of their improved performance. Within the next few years a variety of sources may be
developed to make this material widely available at lower costs.
Resin-Coated Proppants
The most commonly available resin-coated proppants are resin-coated sands. These low-density,
intermediate-strength proppants are available in two forms: curable and precured resin-coated
Ottawa-type fracturing sands. Both are manufactured by a process similar to that used to produce
coated sand for the foundry industry. Curable resin-coated sand was originally patented by Graham
et al., for use in gravel-packing operations. Precured resin-coated sand became available in 1982,
about 7 years after the first curable product was used in fracturing operations.
The emergence of a high-quality, curable resin-coated sand, along with the availability of a pre-
cured type, has led to a wide variety of fracturing applications. Although this proppant is not as
strong nor as tough as the ceramic proppants, it is a significant improvement over uncoated sand.
The plastic coating distributes point loads over a wider area on the sand grain and retards brittle
failure. As such, the product is useful at higher stress levels (e. g., in deeper wells) than conven-
tional fracturing sand.
The major application of the curable resin-coated sand is as a tail-in material to retain the sand in
producing zones that will not retain ordinary fracturing sand. The curable coating bonds the sand
grains together after they are in place in the fracture. This in-situ consolidation often prevents prop-
pant flowback, subsequent productivity loss, and damage to well equipment. Because of the con-
solidated nature of the proppant pack formed with resin-coated sand, compressive or tensile
strength is often used as the critical physical property to describe resin-coated sand rather than its
crush resistance. Typical physical properties, pack permeabilities, and width data for curable resin-
coated sand are presented in Table 7.5..
A curable resin coating can also be applied to proppants other than sand, and such materials as sin-
tered bauxite, intermediate-density proppant, and zirconia have all been coated and used in frac-
turing treatments. The use of a curable resin coating in these applications is largely the same as
with sand - to prevent proppant flowback.
The major application of precured resin-coated sand is to enhance the performance of sand at high
stress levels. This proppant is produced by heat curing the coating during the manufacturing pro-
cess rather than allowing curing to occur after the resin-coated sand has been pumped into place.
* All tests performed according to Reference 11 or 12. Values shown are averages of multiple production samples over a 4-year period.
** Currently available in limited quantities on special order only.
The resin coating also encapsulates the sand grains, thus, preventing the migration of crushed fines
during fluid production. It has also been shown to be resistant to destruction by hot formation
brines and crude oils at temperatures up to 300F [150C].
At low stress levels, the performance of this material is not materially different from that of sand.
At higher stress levels, however, performance of the resin-coated sand is improved considerably
over the original uncoated sand. Table 7.6 shows typical physical properties of this material.
* All tests performed according to Reference 11 or 12. Values shown are averages of multiple production samples over a 4-year
period.
** Currently available in limited quantities on special order only.
Closure Stress
The stress transmitted from the earth to the proppant during fracture closure causes crushing of the
proppant, reducing particle size and increasing surface area of the proppant, both of which reduce
permeability of the propped fracture. In addition to crushing, the stress applied to the proppant
pack serves to compact the particle bed, to reduce its porosity, and to reduce its permeability fur-
ther. The last effect occurs even at relatively low stress levels when breakage is not important.
Cycling of stress, as would occur with periodic shut-ins of a well, also reduces fracture conductiv-
ity irreversibly. Closure stress may also cause proppant particles to embed into the walls of a soft
formation, thus, decreasing fracture width and conductivity further.
An example of how closure stress affects permeability of different proppant materials can be seen
by comparing the permeability data for sand to sintered bauxite. Figure 7.5 shows a plot of perme-
ability versus stress for 20/40 mesh Hickory sand and Bauxite. As shown, Bauxite is clearly less
affected than sand within the stress levels tested.
The stress a proppant sees will depend on the overburden stress, the reservoir pressure, the bottom-
hole flowing pressure, the ability of the vertical stress to be transmitted to the horizontal direction
(related to Poissons ratio), tectonic stress (such as nearby mountain ranges) and to some extent,
the fracture geometry (usually a small contribution). A prefrac well test called a stress test is the
best method of estimating the stress on proppant. Or it can be estimated by the following equation:
Stress = k ( OB Pr ) + Pr Pf + Pt
where k = ratio of horizontal stress to vertical stress (k = (r/1-r)), OB = overburden stress (approx-
imately 1 psi/ft of depth), Pr = reservoir pressure, Pf = fluid pressure in the fracture and Pt = stress
due to tectonics (usually unknown and omitted).
A few observations can be made by studying this equation. First, as the reservoir pressure is
depleted, the stress on the proppant decreases. Second, as the well is drawn down further (Pf
becomes smaller at the wellbore), the stress on the proppant will increase. Also, since Pf increases
as one moves down the fracture (away from the wellbore), the maximum stress that a proppant will
see is early in the life of a well near the wellbore, assuming Pf does not change with time.
8/16 Hickory sand. As shown, the larger mesh proppants - e. g., 8/16 mesh - provide a greater con-
ductivity at lower stress levels than the more commonly used smaller sizes, such as 20/40 mesh.
As stress levels increase and particles are crushed, these differences in conductivity decrease
because particle size distribution, porosity, and surface areas become similar despite initial parti-
cle-size differences. At this point, other factors often play a more dominating role in proppant size
selection than conductivity considerations.
Consideration of proppant size is important in the design of fracturing treatments because a mini-
mum fracture width is needed to allow the proppant to enter the fracture. The generally accepted
values for this so-called admittance criterion require fracture widths in the range of two to three
times the largest grain diameter. An admittance criterion based on twice the largest grain diameter
requires fracture widths of 0.187, 0.066, and 0.033 in. for 8/16, 20/40, and 40/70 mesh proppants,
respectively. The largest of these values may be difficult to achieve in very deep wells with forma-
tions having high bottomhole fracturing pressures and usually requires the use of smaller proppant
for successful completion of the fracturing treatment.
Additionally, it should be thoroughly understood that proppant transport must be considered dur-
ing the selection of the size of the propping agent. Even though a 12/20 mesh proppant may be
much more conductive than a 20/40 mesh proppant, the smaller proppant is much easier to trans-
port deeply into a fracture than the larger proppant.
Proppant Concentration
The term proppant concentration refers to the amount of proppant per unit area of fracture wall
(measured on one side only). In customary units, it is expressed in pounds of proppant per square
foot of one wall of the fracture. If proppant settles to the bottom of a vertical fracture as it enters,
the concentration will be determined by the width of the fracture at the time of entry (i. e., during
pumping). If the proppant is suspended in the fracturing fluid until the fracture closes, concentra-
tion will be determined by both the width during pumping and the concentration of proppant in the
fluid.
Proppant Strength
The strength of proppants is of major concern in the design of propped fractures. Historically, this
strength has been expressed in terms of the load required to crush a single grain of proppant divided
by the diameter squared of its contact area at the point of crushing.
Another test, the API crush resistance test, was designed to determine the relative strength of prop-
pants in packs and has been tested and adopted by API for testing sands to be used in hydraulic
fracturing. The API test uses an apparatus for imposing a sustained load on a proppant pack. The
degree of size reduction sustained by the proppant is taken as an inverse measure of proppant
strength.
The API crush resistance test is a more complex measure of strength than that described above for
single particles. The values obtained are influenced by grain shape, particle-size distribution, pack-
ing arrangement, and other attributes of the particle pack. Although these factors are thought to
make the test more representative of proppant performance under field conditions than the single-
particle test, sensitivity of the measurement to several pack attributes makes the test more difficult
to reproduce, and small variations in results (e. g., 2 or 3%) are considered insignificant in critical
comparisons.
Figure 7.8 shows the relationship of closure stress to flow capacity of various proppants, which is
determined primarily by proppant strength. This figure shows that Hickory sand has greater per-
meability at low stress compared to Ottawa sand. This effect results from the fact that Hickory sand
has a larger sand distribution (20/30 mesh particles predominate) as compared to Jordan sand, as
well as the fact that Hickory sand is more angular. These attributes result in Hickory sand provid-
ing greater permeability than Jordan sand up to nearly 5000 psi stress.
200
100
60 Frac Sand
Ottawa Type
40
20 Precured Resin-
Coated Sand
10
6
Frac Sand
4 Brady Type
1
0 2 4 6 8 10 12 14
Closure Stress, psi in 1000's
Fig. 7.8 - All Proppants (non-Ultrafrac).
Above 5000 psi [34.5 MPa] closure stress, some of the largest grains break into smaller particles.
Thus, at higher stresses, Ottawa sand, which had not broken as much as Brady sand, is seen to have
the higher proppant-pack permeability. While the conductivity measurements on which these
results are based are very sensitive to proppant-pack attributes and difficult to reproduce, the com-
parison cited is from measurements made in the same laboratory and therefore are as comparable
as current measurement techniques permit.
Because the surface stresses are more uniform, a well-rounded, spherical particle is capable of car-
rying higher loads without crushing than a less-rounded particle. Therefore, at high stress levels, a
high degree of roundness and sphericity contribute to higher proppant particle does not pack as
well as a well-rounded particle and, thus, has more porosity and correspondingly greater perme-
ability. An example of this phenomenon was described previously. Hickory sand, which is some-
what more angular than Ottawa sand, has slightly better flow capacity below about 5000 psi
[34.5 MPa] than Ottawa sand, although the more rounded Ottawa sand is superior in proppant-
pack permeability at higher stress levels.
Embedment
If proppant particles penetrate the walls of the fracture, the effective width of the fracture, and
thereby the conductivity, is decreased. Not only is the width of the fracture decreased by embed-
ment, but fine particles are generated by failure of the formation rock. These fine particles may
also contribute to the loss of fracture conductivity.
An attempt to assess the severity of embedment has been made by ball-point penetrometer tests of
formation rock. These tests are not as important as was earlier thought because in most modern
fracture designs the proppant pack is many particles thick in the fracture. The intrusion of the prop-
pant into the fracture wall represents only a small fraction of the proppant-to-proppant interaction.
However, in soft formations such as North Sea Chalks, proppant embedment can be significant and
fracture designs are modified to increase fracture width and minimize the detrimental effects of its
occurrence.
Fracturing-Fluid Residues
The pore space of proppants packed in a fracture is sometimes decreased by the deposition of a
residue from water-based fracturing fluids. Such residue may cause a drastic decrease in fracture
conductivity under certain conditions. The problem is most pronounced when the volume of resi-
due from the polymer is higher, when the concentration of proppant in the closed fracture is lower,
and when stress on the fracture is higher which causes lower porosity. Figures 7.9 and 7.10 show
pictures of the residue from borate and zirconate crosslinked fluid systems, respectively. These fig-
ures show the proppant pack damage that occurs due to residue and also indicate that this damage
is minimized by using borate crosslinkers.
The most common residue is a product of the degradation of water-soluble polymers used to build
viscosity in fracturing fluids. Service companies have devoted much effort to reducing polymer
residues in fracturing fluids. Recent research has focused on developing more efficient thickeners
with more soluble degradation products. Some of the detrimental effects of residue deposition can
be alleviated by minimizing polymer concentrations, using higher proppant concentrations in flu-
ids that suspend the proppant, using foam or emulsion fluids, and avoiding conditions of extreme
proppant crushing. STIMLAB has developed a program PREDICTK (available from EPTG
Fracture Applications Team) which tabulates available data comparing retained permeabilities
after breaking and cleanup of various generic fracturing-fluid types. The data are presented as
retention factors that can be applied to API-type short-term permeability data to obtain a usable
value of proppant-pack flow capacity. The retained permeability includes the effects of time, tem-
perature, and fluid residues.
Inspection of this program reveals a direct comparison of the effects of increasing gellant loading
for a titanate cross-linked hydroxypropyl guar gum (HPG) type fluid. Increasing gellant from 40
to 50 lbm/1000 gal [4793 to 5991 b/m3] decreases retained permeability by an additional 15%.
Further reduction is encountered by a gellant increase from 50 to 60 lbm/1000 gal [5991 to
7190 g/m3] of about 15%.
Another comparison of fluid effects, i. e., guar gum vs HPG, shows little difference in retained per-
meability. Virtually no difference is seen between titanate cross-linked fluids and those linked with
zirconates.
Fracture closure, fluid leakoff, and viscosity breaking processes have a dramatic effect on cleanup
and regained permeability of the proppant pack. Breaking times of 2, 10, and 24 hours are com-
pared for a generic cross-linked fluid. Slow and fast breaks are compared for gelled oil. The trend
is the same: more rapid breaks tend to be more effective in terms of regained permeability.
In a comparison of the damaging effects of different types of generic fracturing fluids, one type
stands out as being the least damaging: foam fracturing fluids. These fluids, composed mainly of
a gas and minor amounts of gelled water, permit a proppant pack to regain 70 to 90% of its poten-
tial flow capacity.
Fines Movement
The fine particles created by grain failure at higher stress levels lead to lower proppant-pack per-
meability. The particle-size distributions resulting from such crushed particles have been investi-
gated by several authors and their effects on fracture conductivity reported. Fine particles have
been shown to migrate through the propped fracture and to plug the pore throats, thereby reducing
fracture conductivity. The long-term decreased permeability of sand proppant reported may be
caused at the least partially by movement of preexisting fines with continued flow through the
sand.
Non-Darcy Flow
For non-Darcy flow, the pressure drop in the fracture can be expressed by
2
p L f = v k f + ( ) ( )v (7.1)
where
p = pressure,
= viscosity,
v = velocity,
kf = permeability,
= fluid density.
The second term of the equation, with coefficient , expresses the increased pressure gradient as a
result of deviations from Darcys law. Values of have been measured for a variety of sand sizes
at different values of stress.
Non-Darcy effects can substantially reduce the effective fracture conductivity in high-flow-rate
gas wells. This reduction in conductivity will decrease the wells PI and can complicate the anal-
ysis of pressure-transient tests. To analyze wells properly where non-Darcy flow affects the pres-
sure distribution in and around the fracture, a reservoir simulator that includes non-Darcy flow
must be used by the analyst.
History
Shortly after the introduction of hydraulic fracturing and its acceptance by the industry, the impor-
tance of fracturing pressure data was recognized, as evidenced by a quotation from Godbey and
Hodges1
By obtaining the actual pressure on the formation during a fracture treatment, and if the
inherent tectonic stresses are known, it should be possible to determine the type of fracture
created.
Later, fracturing pressure and the relation between pressure and in-situ stresses were inherently
included in pioneering model development work of Khristianovic and Zheltov,2 Perkins and Kern,3
and Geertsma and de Klerk4 during the 1950s and 1960s. However, it was still several years later
before the analysis of fracturing pressure data started to become an accepted industry practice.
In 1978, Amoco Production Company initiated a coordinated program of field data collection5 and
analysis to improve the understanding of the mechanics of the fracturing process. Much of this
understanding had not changed since the early 1960s and was being severely tested by ever larger
and more expensive treatments. A series of papers at the annual meeting of SPE in 1979 presented
results from this program, including a paper by Nolte and Smith6 which first introduced a basis for
the interpretation of pressure behavior during a fracture treatment, and one by Nolte7 for interpret-
ing pressure decline after the treatment.
The paper by Nolte and Smith presented a means for inferring periods of confined-height exten-
sion, uncontrolled height growth, and, more importantly, identification of a critical pressure.
When a treatment reaches the critical pressure, fracture extension is reduced significantly and a
pressure (screenout) condition or undesired fracture height growth can follow. Nolte and Smith
demonstrated in the paper that a log-log plot of net fracturing pressure (above closure stress) vs.
treating time could be used to identify periods of unrestricted extension, confined height, excessive
height growth, and restricted penetration. This plot and technique has been used extensively since
its introduction by both operators and service companies to determine fracture characteristics and
geometry, and as an evaluation tool for optimizing treatment designs.
Nolte7 also presented analyses permitting some of the parameters that quantify a fracture and the
fracturing process to be estimated from the pressure decline following fracturing. At the time this
work was presented, there was no direct or simple procedure for evaluating the basic parameters
controlling a fracture treatment. Procedures were presented for quantifying fluid loss coefficient,
fracture length and width, fluid efficiency, and time for the fracture to close from the fracturing
pressure decline. The minifrac procedure was introduced for obtaining these parameters for use
in designing the actual fracture treatment.
The analysis procedures from these two papers6,7 have been used extensively by the industry to
evaluate fracture treatments related to tight gas massive hydraulic fracturing,8-10 waterflood
wells,11 moderate permeability oil wells,12 and geothermal formations.13 The work by Nolte and
Smith was extended to include analysis for determining proppant and fluid schedules from the
fluid efficiency when little or no information is available,14 e.g., wildcat area. In addition, theoret-
ical work has extended the analyses to cover the three popular 2-D fracture geometry models,15 to
cover more complex geometries involving fracture height growth,16 and to consider such phenom-
ena as pressure dependent fluid loss.17 In recent years, the service companies have built computer
treatment monitoring vehicles for use on-site in collecting and analyzing fracturing pressure data
using the analysis techniques presented by Nolte and Smith.18
Fig. 8.1 shows the first recording of bottomhole pressure during and after a fracture treatment. The
analogy to transient pressures in reservoirs can be seen in the figure with increasing pressure dur-
ing injection and the pressure falloff or decline after shutdown. The figure also shows that during
the first half of the treatment, the pressure was increasing, while during the last half of the treat-
ment, the pressure remained essentially constant, e.g., a critical pressure was reached. This period
might be interpreted that the increasing pressure indicates extension at essentially constant height,
with subsequent increasing height during the constant pressure period. During the treatment, the
rock was confining fluid at a pressure up to 1400 psi above the in-situ rock stress of the target for-
mation. During the initial portion of the decline (41-44 hours), the fracture is closing due to fluid
loss with the rate of loss proportional to the rate of pressure decline. The increased rate of pressure
decline after 44 hours is due to the increasing stiffness of the fracture closing on the proppant at
the wellbore. This time is significant for two reasons -- the propped width can be inferred from the
net pressure, and the well could be backflowed with minimum proppant production. Beyond
44 hours, the fracture is essentially closed on proppant and the pressure decline reflects reservoir
parameters as pressure declines back to initial conditions. At 56 hours, pressure has decayed back
to initial reservoir pressure.
Fracture
9000 Treatment Pressure Decline
Fracture Transient Reservoir
Closing Press. Near Wellbore
8000
Pe Frac. Closes Pressure From
Net Fracture on Prop Bottomhole Bomb
(50MPc) Pressure at Well, Inferred Pressure
Pe Propped
= Pbh-Dc
7000
Width
Closure Press, Pc
= Horiz. Rock Stress
6000
Reservoir Press.
5000
38 40 42 44 46 48 50 56 58
Clock Time (hrs)
The following discussion presents the basis for and examples of fracturing pressure analysis and
design. Also included are procedures for the successful field application of this technology.
Closure pressure is equal to or less than the breakdown pressure required to initiate a fracture and
less than the pressure required to extend an existing fracture (fracture extension pressure or frac-
ture parting pressure). An upper bound for closure pressure might be estimated from the initial
shut-in pressure (ISIP) after a small volume acid or prepad injection. An upper bound can also be
found from the breakpoint on a step-rate injection test (fracture parting pressure or fracture exten-
sion pressure). However, for quantitative analysis, a more definitive value is needed. While other
methods such as logs and core analysis, Chap. 10, exist to measure or estimate in-situ fracture clo-
sure stress, the only definitive data for pressure analysis comes from some type of injection test,
e.g., we must hydraulically fracture the rock in order to measure the data needed for hydraulic frac-
turing pressure analysis. For measuring closure stress, three basic types of tests are used: (1)
pump-in/decline tests, (2) step-rate injection tests (used to measure fracture extension pressure),
and (3) pump-in/flowback tests.
Microfrac Tests
Microfrac tests are a special type of pump-in/decline test used to measure closure stress in a small,
discrete zone. The test may be conducted in open-hole sections by isolating the test interval with
inflatable packers; however, for most commercial fracturing cases, testing is conducted by perfo-
rating a short (1 to 2 ft) interval of casing, typically at 4 to 6 shots per foot with a 60 or 90 perfo-
ration phasing.
These types of stress tests are discussed thoroughly by Warpinski19 and McLennan,20 and an ideal
test might appear as seen in Fig. 8.2. Tests typically might consist of injecting 20 gallons of water
at 5 gpm, the basic theory being that, after injecting a small volume (e.g., the term microfrac) of
low viscosity fluid at a low rate, the ISIP (instantaneous shut-in pressure) will be a very close
approximation to the actual closure pressure. In fact, where a clear ISIP exists as idealized in
Fig. 8.2, or seen for real data in Fig. 8.3, selecting closure pressure as equal to the ISIP may be an
acceptable approximation. However, as emphasized by Warpinski,19 tests must be repeated several
times in order to ensure that the value is repeatable. Generally, multiple repeat tests tend to reduce
any influence of rock strength since the fracture is no longer being extended but only reopened.
This will tend to make an ISIP more definitive and easier to pick, and, if the value is also repeat-
able, then a good value for closure stress has probably been found.
Second Cycle
First Cycle
Initial Breakdown (Pb ) Secondary Breakdown
1
Pressure (Pb2)
Propagation Pressure Propagation Pressure
Bottomhole Pressure
Time
7500.000
Bottomhole Pressure (psig)
6750.000
ISIP
6000.000
5250.000
4500.000
3750.000
3000.000
0.0000 2.500 5.000 7.500 10.000 12.500 15.000 17.500 20.000 22.500 25.000
Time (minutes)
However, it should be realized that the instantaneous shut-in pressure is always an upper bound for
closure pressure since a fracture cannot shut instantly when pumping is stopped. Therefore, pick-
ing an ISIP value and making use of this value must be done with care. Also in some instances, a
definitive ISIP is never realized and other analysis methods must be used to determine fracture clo-
sure pressure.
The most common analysis procedure, and the procedure recommended here, is to plot pressure
vs. the square root of shut-in time. A change in slope indicates a drastic change in the linear flow
behavior, and is taken to indicate the fracture closing. For example, Fig. 8.4 shows a cased hole
microfrac stress test conducted in the Mesaverde formation of the Rocky Mountains - and clearly
no definitive ISIP can be picked. However, plotting the pressure falloff vs. square root of shut-in
time shows a definitive change in slope, and closure pressure is chosen as identified in the figure.
For this Mesaverde well, closure stress was measured by Warpinski21 in several intervals, and this
data was reanalyzed as discussed by Miller and Smith22 using the reservoir type analysis of plot-
ting pressure vs. the square root of shut-in time. As seen in Fig. 8.5, agreement between the two
analysis methods was nearly perfect. Thus, picking an ISIP value does give a good value for clo-
sure stress. However, in many cases an ISIP could not be identified, whereas the square root plot
gave a definitive value and in virtually every case, the reservoir type, square root plot analysis
yielded a more subjective, definitive analysis.
Fig. 8.4 - Bottomhole Pressure, Square Root Time and Elapsed Time Since Shut-In.
0.9
0.8
0.7
0.7 0.8 0.9 1 1.1
ISIP Pick (psi/foot)
Fig. 8.5 - Comparison of ISIP vs. Root Time Analysis of Microfrac Stress Tests.
Pump-In/Decline Test
As discussed on page 8.4, microfrac tests are a special class of pump-in/decline tests used to mea-
sure stress in small, discrete formation intervals, and these micro tests typically use small vol-
umes of water injected at rates measured in gallons per minute. However, often it is more practical
for commercial fracturing applications to measure the closure stress over the entire intended com-
pletion interval. The basic test procedure is, of course, identical to a microfrac type test; however,
volumes are now measured in barrels and injection rate in bpm. For example, a typical test might
involve injecting 50 barrels of water at 20 bpm. The important, indeed critical, point is that the
injected volume and injection rate must be guaranteed to be sufficient to create and/or open a
hydraulic fracture. For this reason, it is often desirable to proceed the actual stress test with a
step-rate injection test as discussed on page 8.10.
For a pump-in/decline test, closure pressure is determined by injecting a volume of fluid at a rate
sufficient to create a fracture; then shutting in the well and allowing pressure to naturally decline
to below closure pressure (e.g., allow the fracture to close). For testing an entire completion inter-
val, this type of test is most useful in moderate to high permeability formations, where closure
occurs reasonably quickly. For very low permeability zones, e.g., tight reservoirs, closure time
may be so long that closure becomes difficult to identify. For these cases, pump-in/flowback tests
may be preferable as discussed on page 8.9.
Testing is usually conducted with the base fluid being used to prepare the fracturing fluid, e.g., KCl
water, diesel, produced formation fluid, etc. The pump-in portion of the test is performed at the
fracturing rate, and in most cases consists of 50 to 100 barrels of fluid. While a pump-in/decline
test over an entire completion interval may be procedurally similar to the microfrac tests discussed
earlier, a simple ISIP cannot be used to approximate closure pressure. Because of the larger vol-
umes and higher rates needed to ensure that the completion interval is fractured, the injection pres-
sure can easily be several hundred psi above closure pressure; thus, special analysis is mandatory
in order to identify fracture closure.
Since the test is, hopefully, being conducted in a porous, permeable formation, the first analysis is
to plot a Horner plot of the pressure decline as seen in Fig. 8.6. For this plot, pressure is plotted on
the y axis on a linear scale, and Horner time, [tp+ts]/ts, is plotted on the x axis on a logarith-
mic scale. If a semilog straight line is starting to develop as seen in Fig. 8.6, and if this line extrap-
olates to a reasonable value for reservoir pressure, then radial or pseudoradial flow may be
affecting the pressure decline behavior. In order for this pseudoradial flow to start developing, the
fracture must already be closed, thus pressure data falling on the semilog straight line is excluded
from the closure stress analysis. Next, the pressure falloff (prior to the point where pseudoradial
flow may be starting to affect the decline) is plotted vs. the square root of shut-in time as idealized
in Fig. 8.7. Initially, pressure should decline on a straight line indicating linear flow in the forma-
tion. The point where the fracture closes should cause a drastic change in the flow system and a
distinct change in slope on the square root plot. Note, however, that the change in slope may be
either up or down, depending on the relationship of the fracture's variables and those of the
reservoir. This implies a theoretical possibility that no change in slope may occur. Thus this anal-
ysis method should be treated with caution. In particular, this type of problem is most likely to
occur in low permeability formations where closure time is extended. In such situations, the
pump-in/flowback test, discussed below, should be utilized.
Bottomhole Pressure
POSSIBILITIES
t s (Shut-In Time)
Fig. 8.6 - Illustrative Horner Plot for Shut-In Fig. 8.7 - Illustrative Root Time Analysis for
Decline Test. Closure Stress.
Pump-In/Flowback Test
For low permeability, tight formations, the time-to-close for a pump-in/decline test may be quite
long, making identification of fracture closure (e.g., identifying two distinct slopes on a square root
plot) difficult. For these cases, a pump-in/flowback test (PI/FB) may be used to accelerate fracture
closure-- thus making closure pressure more identifiable.
Digital Readout
2 inch
Flowmeter
Flowback
Line
Disposal
Wellhead Pit
Digital Readout
For a PI/FB test, the injection is immediately followed by a flowback at a constant rate, typically
through a flowback manifold similar to that shown in Fig. 8.8. The constant flowback rate is main-
tained with an adjustable choke or valve and should be metered with a low-rate flowmeter. The
primary purpose of the flowback is to flow back at a rate on the order of the rate at which fluid is
leaking off to the formation. For this flowback rate, a characteristic reverse curvature occurs in the
pressure decline at closure pressure as shown by the middle curve in Fig. 8.9. The proper or ideal
flowback rate must be determined through trial and error, performing the first flowback at 1 to 2
bpm and changing the rate until the S-shaped character of the pressure decline is achieved. Once
the desired rate is achieved, at least one additional PI/FB test should be performed to ensure repeat-
ability.
The principle behind a pump-in/flowback test is illustrated in Fig. 8.10. During the early stages of
the flowback pressure decline (A), the fracture and formation are dominating behavior and the
pressure decline is normal. When pressure declines equal closure pressure at the wellbore, the
fracture begins to close in the near well region. However, this closure is over a limited distance and
since even a closed fracture possesses significant permeability, there will be no sudden, drastic
change in pressure decline behavior. Also, it should be noted that away from the well the fracture
is still open; driving fluid to the wellbore and preventing any sudden increase in the rate of pressure
decline.
With further pressure drawdown in the wellbore, the effective stress (e.g., fracture closure stress
minus pore pressure in the fracture) acting over the closed portion of the fracture increases,
Bottomhole Pressure
Time
Fig. 8.9 - Illustrative PI/FB Stress Test Analysis (linear p vs. t plot).
decreasing the permeability of the closed fracture. This begins to reduce flow into the wellbore and
the rate of pressure decline starts to accelerate as the flowback is increasingly coming from simply
the pressurized fluid in the well. The acceleration of the rate of pressure decline (B) creates the
characteristic reverse curvature behavior (C), and the point where this acceleration starts is
identified as fracture closure pressure, e.g., the point where the fracture first begins to close at the
wellbore.
An additional analysis procedure for PI/FB tests is a derivative plot such as seen in Fig. 8.11. For
this plot the change in pressure with respect to time, dP/dt, is plotted vs. time. Since closure is iden-
tified at the point where the rate of decline accelerates, closure would be identified with the max-
imum point on the derivative plot. For the example in the figure, the derivative is constant for a
fairly long period time, e.g., pressure is declining linearly with time. In such a case, closure should
probably be identified at the end of the constant derivative period, e.g., at the point where the rate
of pressure decline begins to accelerate. For this case, it would probably be advisable to run an
additional case with a higher flowback rate to achieve a more identifiable maximum on the deriv-
ative plot, and thus a more distinct value for closure pressure.
The SRT procedure is similar to that performed for reservoir flooding purposes. Fluid is pumped
at incrementally increasing rates and the final injection pressure recorded for each rate is plotted
vs. rate as seen in Fig. 8.12. A typical test may include rates ranging from 0.25 bpm to 20 bpm.
Leakoff
Pressure = Pc
k = Finite
Pressure < Pc
so frac is stressed
k = very small
The resultant pressures at each rate are plotted vs. rate and the breakpoint is identified as fracture
extension pressure. For best results each rate should be maintained for a fixed period of time (typ-
ically 2 to 5 minutes). Also, because of the very low rates at the beginning of the test, the proper
pumping equipment is required (e.g., low rate acid injection pump), equipped with a small ID
flowmeter for accurate metering. Conventional fracture pumping units have a difficult time main-
taining constant rates at less than 2 bpm.
E' 1/4
P net = ----- [ QL ]
H
where net pressure, pnet, is the total fluid pressure minus closure pressure, and the closure pressure
is equal to, and counteracts, the horizontal rock stress perpendicular to the fracture plane. Other
parameters in the equation are rock modulus, E', which can be obtained from laboratory core data;
fracturing fluid viscosity, ; injection rate, Q; and created fracture height and length, H and L. This
relation predicts that net pressure should increase with time as fracture length increases, provided
fracture height is near constant or restricted. However, variations from this prediction of increasing
pressure have been observed in numerous cases. The following discussion presents interpretation
techniques to interpret and analyze these pressure variations to aid in defining the fracturing pro-
cess for different situations.
For actual fluids used for fracturing, the exponent, e, can be bounded for cases of high and low
fluid loss, and where the fluids non-Newtonian power law exponent, n, varies from 1 for a New-
tonian fluid to n = 0.5 for a highly non-Newtonian fluid. For the Newtonian fluid with high fluid
loss, the exponent, e, would equal 1/8. For a highly non-Newtonian fluid with low fluid loss, e
would be 1/4. This defines the boundaries for Mode I fracture extension, as seen in Fig. 8.15, and
the following discussion centers on the four characteristic slopes shown in Fig. 8.16.
Mode I - A log-log net pressure to pump time slope of 1/8 to 1/4, as discussed above, implies that
the fracture is propagating with confined height, unrestricted extension, that fluid loss is
linear flow dominated, and that injection rate and fluid viscosity are reasonably constant.
These assumptions comply with the Perkins and Kern fracture growth model.
Fig. 8.16 shows the net treating pressure for three fracture treatments, the initial portion of each
treatment indicating confined height, unrestricted extension (Mode I). Beyond this portion,
though, the treating pressure deviates from the 1/8 to 1/4 slope, mentioned above - confined height,
unrestricted extension, linear flow fluid loss, and constant rate and viscosity. For cases 1 and 3 in
the figure, the slope is nearly flat, indicating near constant pressure which characterizes Mode II
behavior.
e
Pt
Log Time
Fig. 8.15 - Nolte-Smith Slope Limits for Mode I (Restricted Height, Unrestricted Extension).
To analyze what may cause this flattening of the pressure - time slope, the continuity or mass bal-
ance equation can be examined;
Q = q Loss + q Frac
where qFrac is the rate of fluid storage in fracture volume (e.g., w + H + L). Pressure is propor-
tional to fracture width, thus the equation can be rewritten as
Q = q Loss + K [ P + H + L ] ,
where K is a constant. For the Mode I behavior, injection rate Q is constant, height is constant (H
= 0), and qLoss increases with time as fracture area increases. Also, P and L are increasing with
time. If P goes to zero, then ql and/or H must increase to honor the equation. As a result, more
fluid is lost to the formation or stored in additional height. This leads into a discussion of Mode II
behavior on a log-log net pressure vs. time plot.
Mode II - A flat pressure:time slope indicates stable height growth or increased fluid loss which
negates the predicted pressure increase. The potential for height increase is shown in
Fig. 8.17, where the fracture penetrates a section of higher stress at a constant growth
rate. As additional height is generated, the cross-sectional area of the fracture increases,
thus reducing the flow velocity and frictional pressure drop down the fracture and
reducing the normal pressure increase. If height growth continues and reaches a low
stress zone, as seen in the figure, the pressure:time slope may become negative, indicat-
ing uncontrolled, rapid height growth (Mode IV). This type of behavior is discussed
later on page 8.19. The other variable that can change besides H, without violating the
continuity equation is qLoss (fluid loss). One mechanism for a higher fluid loss rate
would be opening of natural fissures intersected by the main fracture as shown in
Fig. 8.18. The opening of natural fissures increases fracture volume and fluid loss area,
and decreases the pressure in the fracture. When pressure declines below the stress
holding the fissures closed, the fissures re-close. Pressure then increases slightly and the
fissures reopen, etc. This opening-closing-opening of the fissures is like a pressure reg-
ulator, producing a constant pressure profile. Due to the increased fluid loss rate, Mode
II will normally be followed by undesired behavior such as a screenout.
Looking back at the continuity equation, if something occurs to stop fracture extension (i.e., L =
0), then either P or H must increase. As shown in Fig. 8.18, increased fluid loss to natural fis-
sures may dehydrate the slurry to the point that a proppant bridge forms in the fracture. If pumping
continues, no additional fracture penetration will occur. If the fracture is contained, pressure must
increase at a higher rate as seen in cases 1 and 2 on Fig. 8.16. If the fracture is not contained, the
rate of height growth will increase and pressure will decrease with time as shown by case 3 of
Fig. 8.16. In the case where the fracture is contained and the pressure increases, this rapid pressure
increase is characteristic of Mode III behavior on the log-log Nolte-Smith plot, Fig. 8.21.
Fig. 8.18 - Effect of Natural Fractures, Critical Pressure Causes Increased Fluid Loss.
Mode III - This behavior is characterized by a region of positive unit slope (i.e., 1:1 log-log
slope), indicating a flow restriction in the fracture. This implies that the pressure is pro-
portional to time or, more importantly, that the incremental pressure change is propor-
tional to the incremental injected fluid volume. This 1:1 slope is similar to the same
slope in Pressure Transient Analysis, indicating storage of fluid, in this case by swell-
ing or ballooning the fracture. Common causes of this behavior are pad depletion
where proppant reaches the fracture tip, slurry dehydration to natural fissures (dis-
cussed above), excessive height growth increasing fluid loss area, and/or proppant fall-
out due to poor gel quality.
Fig. 8.19 shows how excessive height growth can cause slurry depletion resulting in a premature
screenout. The fracture has grown through a shale section into a lower closure pressure sand. Due
to the higher stress in the shale, the fracture width is less than in the sands forming a pinch point
which will not allow sand to pass through, yet allows fluid to pass, dehydrating the slurry in the
target interval. As the slurry dehydrates it forms a plug which will eventually bridge in the fracture.
The approximate distance to the bridge can be calculated from:
Log Width
Top
Top
QE'
R max = x f max = 1.8 ---------------------
2
-
H p/t
where Q = pump rate (bpm), E' = modulus (psi), H = frac height (ft), and p/t = rate of pressure
increase (psi/min). This information can be useful in postanalysis and the design of future treat-
ments. A near-wellbore bridge would likely be caused by natural fissures, height growth, or a high
sand concentration slug; whereas a bridge some distance from the wellbore would more likely be
due to pad depletion, or sand fallout due to poor gel quality.
As noted previously on page 8.14, if fracture extension ceases and the fracture is not contained,
then rapid, unstable height growth will occur as pumping continues and the pressure:time slope
will become negative. This is Mode IV behavior as seen during case 3, Fig. 8.16.
Mode IV - A negative slope can be interpreted as rapid height growth into a lower closure stress
zone. Referring back to the continuity equation, discussed on page 8.17, a significant
decrease in pressure must be accompanied by a significant increase in one or more of
the other variables. A significant increase in fluid loss is possible from opening new
fractures or fissures, but is not likely with decreasing pressure. An increase in length
is not consistent with a decrease in pressure. The only change which is compatible
with a decrease in pressure is an increase in height.
The steepness of the negative slope would imply the rate of unstable growth. A high
rate of growth would exhibit a steep slope, while a low negative slope would imply a
low rate of growth. If the fracture grows into a much lower stress zone, the decrease
in pressure will be rapid. If the fracture grows into a slightly lower stress zone the neg-
ative slope will be shallower.
A negative slope observed from the beginning of the treatment indicates a lack of
height confinement. In this case the fracture will grow radially and future treatments
should be designed using a radial model.
While the observed pressure behavior on the net pressure vs. time plot is primarily a function of
fracture geometry, other parameters may interfere with interpretation. These parameters are shown
in Fig. 8.20, clearly showing that an increase in rate or viscosity will increase net pressure. As a
simple example of this, consider the plot in Fig. 8.21 for a gelled oil fracture treatment. The initial
declining pressure indicates unconfined fracture height, and then after 9 minutes, pressure begins
to increase. This might be interpreted as a change in fracture geometry but, for this simple case,
this is simply the time when gelled fluid is on the perforations. After 4 or 5 minutes, the fracture
is filled with this new, higher viscosity fluid, and pressure again begins to decline. Complete
records of treating parameters must be kept, and what was happening during a job borne in mind
when interpreting net pressure behavior.
Elasticity
W H--E P W R--E P
1/4 3/4
P E--------
H
- ( QL ) 1 / 4 Combining P E--------
R
- ( QR ) 1 / 4
Critical Pressure
As mentioned in the previous discussion on page 8.17, Mode II behavior on the net pressure vs.
time plot is usually followed by some undesirable behavior such as excessive height growth or a
screenout. For this reason, the net pressure where the pressure:time slope flattens is termed the crit-
ical pressure. For the case of height growth, critical pressure is roughly 70-80% of the differential
closure stress between the initial zone and bounding beds. When natural fissures exist, critical
pressure is approximately the net stress component (above closure pressure) acting normal to the
plane of the fissures, holding the fissures closed.
In fieldwide studies, critical pressure has been found to be reasonably constant. During the early
development of a field, strategic wells should be monitored to determine the critical pressure,
which can then be extrapolated to offset wells. Treatment designs can then be formulated to keep
net treating pressure below the critical pressure, possibly by reducing viscosity or rate. If it is
impossible to stay below critical pressure by these means, unconventional-type designs may pos-
sibly be developed to minimize height growth or screenout tendencies.
Three techniques which are recommended for measuring fracturing BHTP are seen in Fig. 8.22.
The first configuration uses a tubing string with an open annulus and a surface pressure recording.
The treatment is pumped down the tubing or casing, with the other side static. Pressures are mea-
sured on the static side and corrected for hydrostatic pressure to obtain BHTP. This configuration
is applicable if the fracturing pressure is greater than the hydrostatic head on the static side, which
is usually the case except in severely underpressured or depleted reservoirs. The second configu-
ration involves running the pressure sensor inside a side-pocket mandrel, above a packer, with
pressure transmitted to the surface via an electric line fastened to the outside of the tubing. The last
technique is running a downhole recording pressure gauge inside a tail pipe below a perforated
joint and packer. With this technique real-time access to the data is not possible. The data is
accessed after the treatment, when the pressure recorder is retrieved. With the first two techniques,
on-site computers can be used to manipulate and analyze the data for fracture treatment design or
to make on-site judgmental decisions during the treatment. The following describes in more detail
the procedures for using these three BHP measurement techniques:
1. Open-Ended Tubing - Run tubing (no packer) to within 100 ft of the perforations. Circulate
out any gas so as to leave a liquid filled static column, whether this is on the tubing or annular side.
Gas in the static column will reduce the hydrostatic head from which BHTP is calculated and
reduce the accuracy of the true BHTP due to gas compression and expansion during the injection
and shut-in periods. The density of the fluid used to circulate the hole should be measured period-
ically, so the hydrostatic head of the static column will be accurately known. During testing and
the actual fracture treatment, both tubing and annular pressure should be recorded continuously. If
injecting down the annulus, the tubing pressure will reflect bottomhole pressure and likewise, the
annular pressure will reflect BHTP when injecting down tubing.
2. Surface Recorded BHP Gauge - A side pocket mandrel containing the pressure gauge is run
above the packer. The wireline for the pressure gauge is strapped to the outside of the tubing as the
string is run in the hole, and a port from the mandrel to the inside of the tubing allows transmission
of pressure to the gauge. This type system is commercially available.
Pt Pa
Qt Qa Q
Q
WIRELINE
PACKER
SIDE POCKET MANDRIL
MANDRIL PORT
Qt - 0 PERFORATED
PRESSURE SUB
SENSOR (BLAST JOINT)
Pt - BHP-Pn
or PRESSURE
Qa - 0 BOMB
PACKER
SEATING
Pa- BHP-Pn NIPPLE
NO-GO NIPPLE
1 2 3
( ) (b) ( )
Fig. 8.22 - Well Configurations for Recording Bottomhole Treating Pressure.
3. BHTP Gauge Tail pipe Assembly - In this configuration, the pressure gauge is placed below
a perforated joint and packer in a tail pipe assembly. The complete assembly from bottom to top
would consist of a joint of tubing with a NO-GO nipple at the bottom, a seating nipple, a perfo-
rated sub, and a pup joint below the packer. The most reliable and least expensive way to prepare
the perforated sub would be to drill the holes in a machine shop. This would ensure all holes are
open, large, and properly spaced. The BHTP gauge would be run into the seating nipple on a slick
line, and the treatment pumped down the tubing and out the perforated sub. After the treatment,
the bomb could be retrieved with a slick line by latching into a fishing neck on top of the bomb or
by pulling the tubing string.
Accurate pressure measurements during prefrac testing and the actual fracture treatment are
required for useful analysis and evaluation. For prefrac tests, i.e., closure pressure, minifrac, etc.,
pressure resolution to nearest 2 psi and 10 second data acquisition is normally adequate. For the
main treatment pressures recorded to the nearest 5 psi, one data point every 20 to 30 seconds is
sufficient. In-line pressure transducers normally supplied by the fracturing service companies have
proven to be unreliable for this type work. Aside from the resolution of the transducers, they may
not be accurately calibrated. Given adequate notice, however, service companies can usually
obtain the precision-type transducers required.
1
Q = q Loss + ( L p net CH ) ( L/L + p net / p net + C p /C p + H /H ) ----- (8.1)
t
leaving the rate of pressure decline, pnet, proportional to the rate, qLoss. From pressure decline
analysis, values for fluid efficiency and fluid loss coefficient can be determined as will be shown
in this section. Note that while the equation above contains a term for changes in fluid compress-
ibility, Cp, this effect will not be included in this discussion. In general, this is not a major factor
for pressure decline analysis. The analysis of the pressure decline for fluid loss and fluid efficiency
is then combined with the Nolte-Smith analysis of treating pressures (for fracture geometry) to
give a complete description of the fracturing process. Note - neither analysis can truly stand
alone, they are complementary and must be used together to describe the process.
Fracture Stiffness
Fig. 8.24 shows equations used to describe fracture stiffness, S, for both a confined height fracture
and a fracture with radial geometry. For both geometries, S is proportional to the crack opening
modulus, E', and either fracture height, H, or fracture radius, R, or for a Geertsma fracture geom-
etry, to fracture length, L. As shown in the figure for the confined height case, if the fracture grows
into a bounding formation, the fracture stiffness, and thus pressure decline analysis is still prima-
rily dependent on the initial fracture height.
Knowing fracture stiffness, the fracture P-V relationship can be calculated from the expression
dV A
------- = --------- d p net /dt (8.2)
dt S
where dpnet is the change in average pressure in the fracture. Unfortunately, only wellbore pressure
can be measured, and even though the fracture has stopped extending, fluid will continue to flow
down the fracture and wellbore pressure will be higher than the average pressure in the fracture.
Thus, a term is defined which relates wellbore pressure to the average pressure in the fracture as
Fig. 8.25 provides graphs for determining for a confined height fracture. For a radial fracture or
a short Geertsma geometry fracture, will be approximately 1.
Since the rate of change in fluid volume, dV/dt, is equal to the fluid loss rate, qLoss, Eq. (8.2) can
be rewritten as
where qLoss is a volume loss term and thus, negative to the system.
Fig. 8.25 - Ratio of Average Pressure in Fracture to Wellbore Pressure, After Shut-In, for a
Confined Height Fracture.
C
v Loss = --------------------------- (8.5)
[t (a)]
where vLoss is the fluid loss velocity over an incremental area of the fracture, da; C is the fluid loss
coefficient; and (a) is the time when the area was created. The final relationship then between
fracture stiffness, S, rate of pressure decline, and C, the fluid loss coefficient, will be termed P*
as discussed on page 8.30. Note, despite the similarity in terminology with a Horner plot P*, the
P* value for fracture pressure decline analysis has no relation to reservoir pressure. Instead, P*
is simply related to the rate of pressure decline following an injection at fracturing rate.
Assuming linear flow or Carter type24 fluid loss, and referring back to Eq. (8.5), the total volume
rate of fluid loss, qLoss, can be found from
2Cda
q Loss = A --------------------------
[t (a)]
- (8.6)
e.g., integrating the fluid loss velocity over the entire fracture area with the factor of 2 occurring
since the fracture has two sides. Obviously, to reduce this to any usable form, the unknown, (a)
(e.g., the time when each element of the fracture area was created) must be known. In general, of
course, this is not a simple, or a known function, and if this is an important factor, then analysis of
the fracturing pressure decline data will become of limited usefulness. However, while this func-
tion is not known, it can be bounded and these bounds can be used to test the importance. For
example, as shown by Nordgren,23 Geertsma,4 and others, for very low fluid loss, fracture area will
grow approximately linearly with time,
while for very high fluid loss, fracture area will grow with the square root of time
Time
Fig. 8.26 - Fracture Growth with Time.
a
--- = ----
A tp
where A is the total fracture area created at the end of the pump time, tp, and 'a' is a small incre-
mental fracture area that was created or opened at time , < tp. This gives
a
s = --- t p
A
or
2Cda
q Loss = -----------------------------------
[ t ( A/a )t p ]
which can be integrated from area=0 to area = A to give the rate of fluid loss, qLoss, for times
greater than (or equal to) tp. This integration gives
2 AC
q Loss = -----------2 { t t t p }
tp
or
2 AC
q Loss = -----------2 { ( 1 + ) }
tp
where time, t, equals tp+ts (e.g., pump time + shut-in time) and = ts/tp.
Similarly, for high fluid loss, Eqs. (8.6) and (8.8) can be integrated to give
2CA 1 1
q Loss = ----------- sin ---------------------
tp (1 + )
2Cr p Af ( )
q Loss = ---------------------------
- (8.9)
tp
where
and a new parameter, rp, has been added for cases where only a fraction of the fracture area is lea-
koff area. That is, rp is the ratio of permeable area opened by the fracture to total fracture area,
and these two functions are plotted vs. dimensionless shut-in time, in Fig. 8.27. The similarity
between the two time functions seen in the figure indicates that an EXACT knowledge of how the
fracture grew with time is not necessary for the decline analysis - so long as the fracture was free
to extend, e.g., no screenout condition occurred. For example, consider the dashed curve in
Fig. 8.24, showing an ideal fracture area vs. time behavior for a treatment which screens out very
early. For this case, fracture area stops increasing early during the pumping. Thus, during the pres-
sure decline, all of the leakoff area is old, leading to lower than expected leakoff and application
of the pressure decline analysis to the postpumping pressure behavior would calculate an errone-
ously low fluid loss coefficient. Finally, note that Fig. 8.27 does not indicate that there is no behav-
ior difference between high and low fluid loss cases. Merely just that the exact time-rate-of-growth
of the fracture while pumping is not a dominant factor, and that postfrac fluid loss rate (and thus
pressure decline behavior) is a function of fluid loss coefficient, C, pump time, tp, and the total cre-
ated fracture area, A.
Fig. 8.27 - Bounds on Rate of Fluid Loss Function (bounds are less than 10% different after shut-in
time equal to 1/4 of pump time).
2 AC A
q Loss = ----------- r p f ( ) = -------d p net /dt (8.10)
tp S
or
2CS
d p net /dt = ------------ r p f ( ) (8.11)
tp
and this gives a definite relation between fracture stiffness, S, fluid loss coefficient, and postfrac
pressure decline. If pressure decline were a linear function of time (e.g., dp/dt = constant), then the
relation could be characterized with a simple psi/minute. For example, assume a case with a
pump time, tp, of 20 minutes. If 10 minutes after pumping is stopped, e.g., ts = 10 or = 0.5, the
rate of pressure decline, dp/dt was 5 psi/minute, then, from Fig. 8.25, f() 1. If the fracture stiff-
ness were known, then Eq. (8.11) could be solved for fluid loss coefficient. However, the behavior
is more complex than this, and a value, defined as P*, will be used to describe the pressure
decline behavior. Basically, P* is a single value which characterizes the rate of pressure decline.
A high value indicates a rapid pressure decline, which would usually correspond to high fluid loss,
however, it might also correspond to a very stiff formation. Thus we see that P* does not directly
describe fluid loss, but rather it will be seen to specify a relation between several variables.
Unfortunately, the rate of change of pressure, dpnet/dt, is hard to measure and use, making it con-
venient to integrate the pressure decline, dpnet/dt, to convert Eq. (8.11) into a pressure difference
form. Clearly integrating dp/dt from time = to to time = to + t
dp
-----
dt
- dt = p = p ( t=t o ) + p ( t o + t )
p ( o, ) = p ( o ) p ( )
where to (or o) is just a convenient marker time or starting time for calculating pressure differ-
ences.
Simultaneously, the right hand side of Eq. (8.11) is integrated from to to a later time, t giving
pCS
p ( o, ) = p ( o ) p ( ) = ----------- r p t p G ( o, )
2
where the G function, G(o,), is defined as
4
G ( o, ) = --- { g ( ) g ( o ) }
and arises from integrating the time function, f(), controlling the postfrac rate of fluid loss. For
example, for the low fluid loss (high efficiency) limit, g() is given by
4 3/2 3/2
g ( ) = --- { ( 1 + ) } ,
3
while, for the high fluid loss (low efficiency) limit,
1 1
g ( ) = ( 1 + )sin --------------------- +
(1 + )
Finally, redefining the variable group (C rp t p S)/(2) as P* gives
p ( o, ) = P* G ( o, ) , (8.12)
indicating that the variable P* is simply a multiplier which best matches the actual pressure
decline behavior to the theoretically perfect behavior defined by G,
CS
P* = ----------- r p t p .
2
Fig. 8.28 - Plot of G(,o), Master Curves for Matching Pressure Differences.
Consider a case where a minifrac (e.g., a volume of fracturing fluid pumped without proppant)
has been pumped down tubing while measuring surface annulus pressure. After shut in, the pres-
sure decline is measured as seen in Fig. 8.29 and tabulated in the table below.
The first step in any pressure decline analysis is to determine the fracture closure pressure and
closure time. For the example here, it is assumed that pre-minifrac stress tests indicated a (surface
equivalent) closure stress of 1500 psi. The minifrac pressure decline reaches this pressure after a
shut-in time, ts, of about 26 minutes - giving a closure time, tc, of 26 minutes. This gives a dimen-
sionless closure time, c, of 1.3, with, since no proppant was pumped, the fracture being com-
pletely closed at closure time.
0 1658
2 1.4 1642
4 2.0 1625
6 2.47 1610 1625-1610
= 15 psi
8 2.83 1595 30
10 3.16 1582 43
12 3.46 1569 56 13
14 3.74 1558 67 24
16 4.0 1544 81 38
18 4.24 1534 91 48
20 4.47 1525 100 57
22 4.69 1515 110 67 10
24 4.90 1507 75 18
26 5.10 1498 84 27
28 5.29 1493 89 32
30 5.48 1486 96 39
32 5.66 1481 101 44
34 5.83 1476 106 49
In selecting the start times for the pressure difference analysis, all start times must be less than
this dimensionless closure time since the analysis has no meaning for pressures below fracture clo-
sure pressure. Referring to the type curve of Fig. 8.28, one might select the 0.2, the 0.5, and
the 1.0 curves, since the dimensionless start times, o, for these curves all come prior to the
dimensionless closure time of c = 1.3. For the o = 0.2 curve, the corresponding real start time is
t o = o t p = 0.2 20 = 4 minutes .
Thus a column of pressure differences is created (as seen in Table 8.2) starting at a shut-in time of
four minutes. Similarly, a column of pressure differences is created corresponding to a real start
time of 10 minutes (to = o x tp = 0.5 x 20) and to a real start time of 20 minutes. These pressure
difference values are then plotted vs. shut-in-time (as three separate and independent curves) on
log-log scales identical to the type curve scales as seen in Fig. 8.30, and the data is matched to
the theoretical curve.
Note, however, that the theoretical type curves include two sets of curves: three dashed curves
for dimensionless start times of o = 0.05, 0.10, and 0.20; and solid curves for dimensionless start
times of o = 0.20, 0.50, 0.75, 1.0, and 2.0. The early time, dashed curves correspond to the
low efficiency solution, while the later time, solid curves correspond to the high efficiency,
e.g., low fluid loss, solution. Closure time, found by plotting the pressure decline vs. the
square-root of shut-in time, is used to determine which type curves to use. If the fracture closes at
a dimensionless time less than 0.5 (c < 0.5), e.g., a fracture closing in less than 30 minutes after a
1 hour pump time, then the high curves (dashed) should be used. For closure times greater than
pump time, the low curves (solid) should be used. For cases which fall into the gray area in
between these limits (e.g., maybe a closure time of 30 minutes after a pump time of 40 minutes)
the curves which best match the shape of the data should be used, and/or one might interpolate
between the two sets of theoretical type curves.
p ( 0, ) = ISIP p ( ) = P*G ( 0, )
p ( ) = ISIP P*G ( 0, )
or
P* = dp/dG .
That is, the slope of a linear plot of the shut-in pressure decline vs. 'G' (as defined earlier) gives
the match pressure P*.
Since this 'G' function is generally a complex function of the dimensionless shut-in time, d, the 'G'
Function Plot is clearly most amenable to computer generated analysis. Also, in several cases the
'G' function has been found to work better for very high fluid loss cases where closure time is on
the order of 20 to 30% of pump time or less. For cases with longer closure times, e.g., closure time
40% (or more) of pump time, the type curve approach discussed above often offers an easier anal-
ysis.
For the previous example, the pressure decline data is plotted vs. 'G' in Fig. 8.31, where, as before,
closure stress is assumed known from minifrac tests to be 1500 psi. (Actually, this would be a sur-
face equivalent closure pressure, with real closure pressure equal to 4530 psi, e.g., 1500 plus
the hydrostatic head of 7000 ft of water.) At any rate, in the 'G' Function Plot, the slope of the
data is taken just prior to closure pressure, though for this plot (which is an excellent example of a
'G' Plot) the slope is relatively constant from shut-in all the way down to fracture closure. Taking
the slope of the indicated line shows a slope of -98 psi, which gives P* = 98 psi, essentially per-
fect agreement with the earlier type curve match analysis.
This plot also shows a distinct slope change at a pressure of 1500 psi, e.g., just at closure pressure
and sometimes, a 'G' Function Plot can be used to determine fracture closure. The procedure is
similar to a root-shut-in-time analysis for closure, a distinct slope change is taken to indicate a dis-
tinct fracture behavior change, e.g., the fracture closing. Again, as with 'G' Function Plot analysis
in general, we have found this analysis procedure to be most useful in low efficiency (high fluid
loss) environments - though clearly this example shows a very clear 'G' Function analysis for a case
with closure time equal to 1.3 times pump time, e.g., c = tc/tp = 1.3.
A final note concerning 'G' Function Plots is - What 'G' Function should be used? For low effi-
ciency (high fluid loss) cases where c < 0.4 to 0.5, clearly the low efficiency function is correct.
Similarly, for longer closure time cases with c > 1, the high efficiency (low fluid loss) function as
used for Fig. 8.31 is probably most correct. However, for the gray area between these limits,
some distortion and error can be introduced by the lack of a purely applicable 'G' Function. In these
cases, type curve analysis often proves superior by allowing easy, manual interpolation between
the two limiting theoretical solutions.
Fluid Efficiency
Fluid efficiency is defined as the fracture volume (at the end of pumping, e.g., at time = tp) divided
by the total slurry volume pumped (e.g., fluid, sand, everything). As an aid in Pressure Decline
Analysis, the rate of pressure decline equation can be integrated to determine the volume of fluid
lost between shut-in, tp, and the time at which the fracture closes, tp + tc. For a minifrac treatment,
e.g., a small volume calibration treatment with no proppant, the volume lost between tp and tp+tc
equals the volume of the fracture at tp. Dividing this volume by the total volume injected gives effi-
ciency. Thus, a relationship between closure time and fluid efficiency exists as shown in Fig. 8.32.
The efficiency, ef, obtained from this figure is used to define a new variable, , which is used in
the type curve analysis and defined as
= V f /V L = e f / ( 1 e f ), or
e f = / ( 1 + ),
where Vf is fracture volume and VL is fluid loss volume during injection. can also be determined
directly from the type curve analysis in terms of the match pressure, P*, and the net fracturing
pressure at shut-in, ps (e.g., ISIP - closure pressure).
= p s /4K g o P* ,
where Go is the pressure difference function at = 0 (discussed on page 8.31) and equal to
1.57-0.238 ef (within 5%, Go = 1.45), and K is a correction to the fluid loss coefficient which
accounts for additional fluid loss only during pumping (e.g., spurt loss or opening of natural fis-
sures during injection). However, K cannot (at this time) be determined from any analytical pres-
sure decline analysis so should always be set equal to 1.
These two efficiency values supply a means of quality control for fracturing pressure decline
analysis. First, efficiency is determined from the dimensionless time-to-close, c and the graph in
Fig. 8.32. Next, the loss ratio, , is determined from the type curve match pressure, P*, and the
final net pressure, ps, as discussed above. This value for is then used to calculate an efficiency
In addition to this quality control procedure for the decline analysis, Section 8.6 presents a proce-
dure for determining a fracture treatment design schedule based solely on fluid efficiency. Also,
efficiency corrections are presented to account for proppant in the fracture at closure, so the pres-
sure decline after an actual propped fracture treatment can be used in a type curve analysis to cal-
culate fluid loss coefficient.
Example/Guidelines
The following will present some general guidelines for fracturing pressure decline analysis in the
context of reviewing an actual field example. The pressure data is the same as that presented and
discussed earlier in Fig. 8.29 and Table 8.2.
Lab Tests show the sand to have a Young's modulus of 4 to 5 million psi; the siltstones, 6-8
million; and the anhydrite, 8-10 million. Based on a simple volume percentage, a modulus of
6 million psi is assumed to be representative of the formation.
Before proceeding with the example, some general guidelines are given in Table 8.3, and these
guidelines will be followed (essentially step-by-step) for analyzing this data and calculating a fluid
loss coefficient.
Following the general guidelines, the first step is always to determine fracture closure pressure.
For this case, closure pressure was known as 1500 psi from pre-minifrac stress tests and one might
simply assume that the fracture closes when the pressure declines to this value. However, it is often
a good procedure to conduct a closure stress analysis with the decline data itself. This is particu-
larly appropriate since into a liquid saturated formation (remembering that this is an oil bearing
formation) can locally increase pore pressure and thus locally increase closure pressure, e.g., fluid
loss can generate what is often referred to as back stress. Since this is an oil zone, the pressure
decline is first plotted vs. root shut-in time as seen in Fig. 8.33. This shows a distinct slope change
at a pressure of 1500 psi, e.g., for this case the minifrac has not altered closure stress.
Table 8.3 - Guidelines for Analysis.
This plot also shows closure after about 26 minutes (also see Table 8.2) giving a dimensionless
closure time of
The dimensionless closure time of c = 1.3 is then used with the efficiency chart, Fig. 8.32, to get
a time-to-close efficiency of 47%. The match pressure of 100 psi along with the net pressure at
shut-in, ps, of 158 psi (as seen in Fig. 8.33) is used to calculate efficiency as
3.142 158
= p s /4K G o P* = -------------------------------------------- = 0.86
4 1 1.45 100
Note: If this calculated efficiency was significantly different from 50%, it would probably be best
to use this first calculated efficiency to recalculate go = 1.57 - 0.238 * ef, and then use this
new value of go to find a new efficiency. It is seldom worthwhile, however, to follow this
iteration for more than one time through.
This is clearly in excellent agreement with the time-to-close efficiency and thus the analysis can
proceed with confidence, e.g., there is no indication of unaccounted for fluid loss.
Note that up to this point, the analysis has been independent of fracture geometry, e.g., it made
no difference whether the fracture was radial, confined height, etc. However, once the match pres-
sure, P*, and efficiency have been determined and the efficiency checked, then it is necessary
to assume a fracture geometry in order to calculate a loss coefficient.
For this example, one might initially expect no height confinement based on: (1) no discrete beds
with sufficient thickness to contain a fracture, and (2) high modulus which leads to high treating
pressures and thus increases any tendencies for height growth. While it is not conclusive, the low
net pressure at shut-in of 160 psi reinforces this expectation since confined height fractures often
have higher net treating pressures than this. Equations from Table 8.5 can then be used as seen
below:
6 1/3
( 0.134 ) ( 20, 000 ) ( 6 10 )
x f = ----------------------------------------------------------------------------- = 211 ft
( 2 ) ( 3.14 ) ( 1 ) ( 100 ) ( 1.45 ) ( 1.86 )
and this radius is then used to calculate a fluid loss coefficient and fracture width
p p 6
C = [ ( P*x f )/ ( r E' t ) ] = ( 100 ) ( 211 )/ ( 1 ) ( 6 10 ) ( 20 ) = 0.0008 ft/ min
and
6
w = ( 6 p s x f )/E' = ( 6 ) ( 3.14 ) ( 158 ) ( 211 )/ ( 6 10 ) = 0.10 inches.
Taking a look at this problem from a slightly different view, assume that postminifrac logs were
available which gave indications of a gross fracture height of 350 to 400 ft. This value for H' might
then be used in the equations for a confined height fracture (e.g., a Perkins & Kern fracture geom-
etry) as seen below,
0.134 VG E'
x f = -------------------------------------------------------2-
4KP* x g o ( 1 + )H
6
( 0.134 ) ( 20, 000 ) ( 6 10 )
x f = ---------------------------------------------------------------------------------------------------------- = 163 ft .
( 4 ) ( 1 ) ( 100 ) ( 0.65 ) ( 1.45 ) ( 1.86 ) ( 375 375 )
However, it is immediately noted that this gives a tip-to-tip length of 326 ft which is less than the
approximate fracture height of 350 to 400 ft; thus, the Perkins & Kern model would not be appro-
priate, and the calculations should move on to the radial model (as discussed on page 8.26) or to
the Geertsma model calculations (which would be for a fracture with a tip-to-tip length less than
the height). For this example, the radial model shows a predicted radius of 211 ft which would give
a total, gross fracture height of H = 422 ft, and since this would be in fair agreement with the logs,
a radial model would probably be the most appropriate geometry model for describing the test.
It is important to note in these calculations that there are several uncertainties; in particular, the
final result for fluid loss coefficient (the usual goal for the decline analysis) is strongly dependent
on the value of modulus. If this value is not known from core analysis then the final result for 'C'
becomes uncertain. In many cases, however, the final analysis can be improved through a proce-
dure of pressure history matching as discussed in Section 8.3.
1. After a propped fracture treatment, fracture closure occurs when the fracture closes on prop-
pant. However, at this point, of course, the fracture is not completely closed, but is held par-
tially open by the proppant. Thus the time-to-close efficiency must be corrected as discussed
below.
2. The pressure decline analysis assumes that the fracture was free to propagate during the injec-
tion period. When proppant is included in a real stimulation there is, of course, always the pos-
sibility that due to slurry dehydration and/or proppant reaching the fracture tip, fracture
extension will be halted and a tip screenout will occur. This is usually evident from the net
pressure behavior and if such a condition occurs, then normal decline analysis is no longer ap-
plicable. Note, however, that pressure history matching as discussed below can still be used to
analyze the data with the time where the screenout starts (e.g., the beginning of the unit slope
on a Nolte-Smith plot, Fig. 8.16) being a good marker for history matching analysis.
The time-to-close expressions previously presented on page 8.35, assumed the fracture closed
completely, e.g., no proppant. Similar analysis can be performed from the postfrac pressure decline
1/2
xf
0.134 VG E '
= ------------------------------------------------
- [ 0.134 VG E ' ]
2 xf = ----------------------------------------------
4 K P * s g o ( 1 + ) H 8 K P * g ( 1 + ) H
s o
w = 12 s p s x f / E '
w = 6 s p s H / E '
C = ( P * s H ) / ( r p E ' t p) C = 2 P * s x f / r p t p E '
w = ( 6 p s x f ) / E '
C = (P *x f )/(r pE ' t p)
NOMENCLATURE
s - See discussion on reverse side of table
K - Correction factor for spurt loss, normally K = 1
T(F) 'n a s
if the propped volume of the fracture is taken into account. If proppant is considered, the effective
fracture volume that will close, Vf', can be written as
V f ' = V f V pr ,
where Vf is the total fracture volume created and Vpr is the volume of proppant including the
porosity of the proppant. In terms of an apparent fluid efficiency, ef', e.g., the efficiency that would
be calculated based on closure time and not corrected for the propped volume of the fracture, the
actual fluid efficiency can be expressed as
e f = 1 ( 1 f pr ) ( 1 e f ' ) ,
where fpr is the volume fraction of proppant pumped (including proppant porosity) relative to the
total slurry injected and defined as
f pr = V pr /V p = W / ( pr V p ( 1 ) ) .
W is the proppant weight, pr is the specific weight of the proppant material, e.g., 165 lb/ft3,
2.65 gm/cc, 22 lbs/gallon for sand, is the proppant porosity (typically on the order of 0.40 since
this refers to a proppant pack with essentially zero stress), and Vp = Vfl+W/pr. For example,
assume a fracture treatment containing 100,000 gallons of gel and 300,000 lbs of sand is pumped
at a rate of 30 bpm. After the end of injection, the pressure decline is monitored and fracture clo-
sure is detected at tc = 45 minutes. The total volume injected is
V p = 100, 000 gals + [ 300, 000 lbs/(22 lbs/gal) ] = 113, 636 gals .
Total pump time was 113,636 gallons/(42 gal/bbl)/(30 bpm) = 90.2 minutes and with a closure
time of tc = 45 minutes, the dimensionless time-to-close was
c = 45/90.2 = 0.50 .
This value of c = 0.50 is used with the time-to-close/efficiency relation to give an apparent effi-
ciency of 28%,
e f ' = 0.28 .
However, the actual efficiency must be greater than this since this apparent efficiency is based
on closure on proppant, and, of course, the fracture is not completely closed at this point. The
actual efficiency is then found from
e f = 1 ( 1 f pr ) ( 1 e f ' )
= ( 1 0.179 ) ( 1 0.28 ) = 0.41 .
to be equal to 41%.
This efficiency of 0.41 is now used with the pressure decline data (prior to closure on proppant)
to perform a type curve analysis using the same procedures discussed previously and outlined in
Table 8.3.
Simulator
Improved Designs
These uncertainties mainly arise since there are essentially more variables than there are equations.
The first of the two main equations can be represented by (from Section 8.3)
E' 3/4
p net = ----- [ QL ]
H
where the net treating pressure (and thus the value for ps used in the decline analysis) is mainly a
function of the modulus of the formation and the gross or total fracture height, H.
The second main equation is the pressure decline behavior which might be represented by the
P* value
CS
P* = ----------- r p t p .
2
where 'S' is the fracture stiffness which (for any fracture geometry) is primarily a function of frac-
ture height and the formation modulus. Thus there are three main variables or unknowns, modulus,
E, height, H, and fluid loss coefficient, C. The important point here is that since there are basically
three unknowns and only two equations, these equations and any solution for them is interde-
pendent. For example, simply solving the pressure decline equations for a loss coefficient gives no
assurance that the answer is meaningful; i.e., is the modulus and fracture height used to calculate
the fluid loss consistent with the net treating pressure. If these values are consistent, then the fluid
loss coefficient determined from P* will be a reasonable (though possibly still not unique) value.
This history matching process is illustrated in Fig. 8.34. For an example, consider the data in
Fig. 8.35. The Nolte-Smith plot of net treating pressure shows increasing pressure with a small
positive slope, indicating a confined height fracture and a numerical model was used to history
match this data and thus determine a height and modulus consistent with the actual treating pres-
sure behavior (with the modulus also being consistent with published industry data). This height
and modulus can then be used with some confidence to calculate a fluid loss coefficient from the
decline analysis. At this point, however, the calculated value for 'C' might be different from the
value used in the initial numerical modeling of the treating pressure, and if this difference is sig-
nificant (e.g., greater than 20 to 30% difference), the modeling should be redone with the new
value for 'C', modifying the height and/or modulus values as required. The new height and modu-
lus would be used to calculate a revised fluid loss coefficient, e.g., one would iterate. Note, how-
ever, that it is very seldom necessary more than one time since the net treating pressure is relatively
in- sensitive to a precise value for 'C'. Because of this relationship (that net pressure is relatively
insensitive to fluid loss), the history matching should always begin with matching the net pressure,
with the modulus and height thus determined then used to calculate a loss coefficient .
With this history match, then, one has a set of three main variables (H, E, & C) which yield a good
description of the minifrac test. These can then be used with some confidence to consider different
treatment designs, larger/smaller volumes, etc. Note, however, that even though the three values
may be consistent they are still not necessarily the correct values. External data is required to
fully determine the problem. For example, core data for the modulus might make this a fully deter-
mined problem. For the case in Fig. 8.35, postfrac temperature logs showed a height in fair agree-
ment with the history matching, making this a fully determined problem.
In particular, for a confined height fracture (e.g., a case where the net treating pressure increases
during a job as seen in Fig. 8.35), treating pressure is generally dominated by fluid flow consider-
ations and can often be reasonably predicted (e.g., maybe within 10%). For a confined height
fracture, net pressure can be approximated by the following equation
3 1/4
0.015 [ E Qx f ]
p net = --------------------------------------------
- (8.13)
H
0.134 VG E
x f = ------------------------------------------------------2- (8.14)
4KP* s g o ( 1 + )H
where is the average fluid viscosity (centipoise), 'VG' is the total fluid volume pumped in gallons,
'Q' is the pump rate in bpm, 'E' is the modulus in psi, xf is the fracture 1/2 length in feet, 'H' is the
gross fracture height in feet, and P*, , etc., are determined from the pressure decline analysis as
discussed earlier starting on page 8.30.
For other geometries such as an unconfined, radial fracture or a case where the fracture is initially
confined but then experiences significant height growth, rock mechanics considerations at the frac-
ture tip begin to play a more dominant role, often precluding the use of such simple, analytical
equations. However, such equations can be developed and may sometimes prove useful. For exam-
ple, for a radial fracture,
3 1/4
0.0078 [ QE ]
p net = -----------------------------------------
2/3
-
xf
x f = ----------------------------------------------
0.134 VG E' 1/3
.
2KP*g o ( 1 + )
As an example, consider the minifrac studied earlier in Section 8.4, with some of the relevant data
from that case listed in Table 8.5.
Table 8.5 - Minifrac Analysis Data.
Test Parameters
Volume=20,000 gallons tp = 20 minutes
Q = 24 bpm = 300 cp
Minifrac Analysis Parameters
K =1 ef = 0.46
DP* = 100 psi = 0.86
Pressure Decline Analysis Initial Results
(Calculations for Radial Fracture Geometry)
E' = Assumed equal to 6x106 psi
xf = Calculated as 211 ft
C = Calculated as 0.0008 ft/ minute
Using this data in the radial fracture geometry calculation for pnet gives a predicted net pressure at
shut-in (e.g., ps) of 240 psi, somewhat greater than the actual measured value of ps = 158 psi.
Remembering that the modulus was strictly an assumed value, one might then use a lower modu-
lus, say 4x106 psi to calculate (still using the initial value for xf) a final net pressure of 178 psi, in
fair agreement with the actual data. This new modulus is then used to revise the initial estimate of
fracture radius (xf), with a new calculated value of xf = 185 ft, and a new calculated loss coefficient
of 0.0010 ft/ minute . With this new fracture radius of 185 ft, and the new modulus of 4 million
psi, the new calculated ps is 195 psi, which is still about 20% greater than the actual data, thus one
more iteration might be in order with a modulus of maybe 3.5x106 psi. At the end of that final iter-
ation, a set of the three major variables (H, E, and C) would be determined which are compatible
with the minifrac data. In addition, since the calculated fracture radius of 190 ft (which gives a
gross fracture height at the wellbore of 380 ft) is consistent with fracture height logs, it is probable
that these values are a very good solution to the actual in-situ conditions.
The effect of natural fractures was discussed in Section 8.4, and this effect is often identifiable
from a constant net pressing pressure on a Nolte-Smith plot (e.g., a critical pressure) and some-
times by comparing the type curve match efficiency with the efficiency derived directly from the
time-to-close.
The possible effects of multiple formation(s) layers is more difficult to categorize since such
multi-layered geology can lead to gross distortions and changes with time of the basic fracture
geometry. As an example, consider the case pictured in Fig. 8.36, where a hydraulic fracture was
initiated in one zone, but then penetrated a barrier and broke into a zone with lower closure
stress. During the remainder of the pumping, the lower stress zone will accept most of the injected
fluid. That is the main fracture will not be in the zone where the fracture started. After shutdown,
however, one might expect the barrier between these two zones to close rather quickly - isolating
the perforated interval from the main fracture. Thus the pressure decline behavior will be dom-
inated by the characteristics of the perforated zone, and may give little or no information concern-
ing the redirection of the fracture geometry, or the characteristics of the lower stressed zone which
accepted most of the injection. Possibly, though, such behavior may be inferred through an obser-
vation of some decline in the net treating pressure indicating the height growth combined with dis-
crepancies between the P* derived efficiency and the efficiency derived from the time-to-close.
Another example of the effect of multiple layers might be seen in the Big pressure decline anal-
ysis problem. The problem as described and several parameters determined from the pressure
decline analysis are included in table Table 8.6.
Using the simple history match equations from page 8.48 (for a confined height, Perkins & Kern
geometry since the net pressure for the minifrac increased indicating height confinement),
3 1/4
0.015 [ E Qx f ]
p net = --------------------------------------------
- (8.13)
H
0.134 VG E
x f = ------------------------------------------------------2- (8.14)
4KP* s g o ( 1 + )H
and the problem definition data from Table 8.6, one calculates a final net treating pressure (e.g.,
net pressure at shut-in) of 688 psi, 20% less than the actual value of about 860 psi. Since net pres-
sure is most affected by fracture height and modulus, either the fracture height must be less than
the gross zone thickness (e.g., less than 150 ft), or the modulus of the formation(s) must be greater
than 7x106 psi, or ?. Since it might be unexpected (but not impossible) for the fracture height to
be less than the gross formation thickness, an initial approach to history matching this data would
probably be to increase the modulus. Doing this shows, after a couple of iterations, a modulus of
9x106 psi giving a calculated final net pressure, ps, of 885 psi, in near perfect agreement with the
actual data. The new calculated values for xf and 'C' are then 802 ft and 0.00075 ft/ minute ,
respectively.
Problem Definition
Volume Pumped = VG = 38,000 gallons
E = Modulus, estimated as 7 million psi
Gross formation thickness = H = 150 ft
Leakoff Height (= net height?) = 110 ft
Rate = 35 bpm
Pump Time = 25.5 minutes
Fluid Viscosity estimated at 300 cp
Pressure Decline Analysis Variables
P* = 260 psi
Final Net Treating Pressure = ps = 860 psi
Efficiency = 0.62
= 0.62 / (1 - 0.62) = 1.63
Initial Calculations
Fracture 1/2 Length = 624 ft
C = 0.00095 ft/ minute
Thus the pressure history matching gives a set of three major variables of H = 150 ft, E = 9x106
psi, and C = 0.00075 ft/ minute , which satisfy both the final net treating pressure of about 860 psi
and the pressure decline behavior of P* = 260 psi and efficiency = 62%. However, since core data
indicated a modulus on the order of 7 million psi, what might explain the higher apparent stiffness
of the formation(s)?
A possible answer to this might be seen in Fig. 8.37, which illustrates the geology of the forma-
tion, showing that the 110 ft net height (out of the 150 ft gross section) is actually composed of two
distinct sandstone layers with 30 ft of shale separating the two zones. Since the increasing pres-
sure behavior during the minifrac seems to indicate good height confinement (e.g., the over- and
underlying shales having higher closure stress than the sands), it might be reasonable to assume
that the separating shale might also be a barrier (e.g., have a higher closure stress) to fracture
growth. Thus this shale would pinch the fracture width (as seen Fig. 8.37), causing the fracture
to behave stiffer than a simple, 150 ft high fracture, thus explaining the need for an unusually
high modulus if the basic pressure analysis methods are to be used.
Given this more complex geology, a fracture simulator capable of treating multiple formation lay-
ers might be used to history match the actual data, as seen in Fig. 8.38 for the treating pressure
behavior. Once the model is successfully set up to history match the past, it can then, of course,
be used with some confidence to design future jobs. Or, in fact, where the dominant effect of the
multiple zones is to just stiffen the fracture, a simple Perkins & Kern type procedure might be
used for frac design by using the artificially high modulus value to account for the effect of the
shale layer on fracture width.
Fig. 8.38 - Nolte-Smith History Match, Pressure Decline Analysis Big Problem.
The above two brief examples have illustrated the extreme range of effects that multiple formation
layers can have on fracture pressure analysis - from the case of the frac growing totally out of zone
and almost invalidating the analysis methods; to a case where the basic analysis methods are fine,
but a slightly artificial modulus must be used in order to accurately describe the fracture width. In
general, it is this extreme range of effects that makes general statements about the effects of com-
plex geology difficult or impossible to make. However, while multiple formation layers clearly
create problems, two recent studies (Warpinski25 and Miller and Smith22) have shown that the
combination of pressure decline analysis with numerical modeling/history matching provides a
useful, powerful tool for analysis of such complex geologic cases.
2. The analysis requires relatively simple data collection and can generally be done from surface
pressure information. Also, the analysis can be completed in a short time making it an ideal
procedure for field use.
3. Final pumping schedule is not significantly affected by actual fracture geometry, thus efficien-
cy procedures can be used in formations (such as coal seams for one example) where actual
fracture geometry may be very complex. Also, this independence from fracture geometry
makes the procedure ideal for initial treatments in a new, wildcat area.
2. Efficiency procedure assumes no knowledge of actual fracture geometry, thus the pre-selected
treatment volumes used as a basis for developing the final pumping schedule may be insuffi-
cient for achieving required production, or the volumes may be excessive, incurring additional
costs and unnecessarily increasing the risks associated with completion operations.
The information generally needed for a stimulation are: (1) the fluid volume to be injected, (2) the
injection rate, (3) the proppant addition schedule, (4) the resulting propped fracture width and
length, and (5) the amount of time that fluids will be exposed to reservoir temperature. This expo-
sure time is needed for selecting the required fluid system along with the amount and type of fluid
additives. For a new area, the volume limitations may be determined from budget constraints, or,
for a more developed area, volumes may be specified based on the requirements to achieve a rel-
ative change in fracture length (or conductivity) from that achieved by prior treatments. Finally,
pump rate is often prescribed based on horsepower limitations or pressure limit constraints of the
wellhead and/or tubulars. While, as mentioned above, the efficiency procedure gives no informa-
tion on propped fracture length or width, it does give the final ingredient, that being the required
pad volume and proppant addition schedule.
While lack of knowledge of final propped fracture dimensions precludes any quantitative devel-
opment of the treatment design in terms of postfrac production; determining the required pad vol-
ume and pumping schedule still remains the most difficult and critical to obtain of any of the
necessary information. As an example, consider the final fracture conductivity distribution pic-
tured in Fig. 8.41. This is the results of a numerical simulation for a case which (purposefully)
included an excess pad volume. As seen in the figure, at shut-down (e.g., at the end of pumping)
the propped fracture 1/2 length is on the order of 500 ft, which was the design length. However,
due to the excess pad volume, the created length is nearly twice as long. Since the area of high fluid
loss is located near the fracture tip, fluid continues to flow from the wellbore region of the fracture
out toward the fracture tip after shutdown. This afterflow results in a proppant redistribution
leaving a relatively (undesirable) low fracture conductivity in the near well area - reducing future
production rates. Another example of the critical need for pad volume/proppant schedule informa-
tion is, of course, the case of inadequate pad volume. This will result in the slurry portions of a
treatment dehydrating and screening out, reducing the propped fracture length and possibly forc-
ing remedial wellbore cleanout operations. Thus, even for fixed treatment volume, either too
much, or too little pad volume is detrimental to final postfrac results.
However, an alternate method may be available when earlier propped fracture treatments have
been performed in the area, and where formation properties such as thickness and permeability do
not change radically from well-to-well. As an example, consider the ideal Nolte-Smith net pressure
plot in Fig. 8.42, and assume this is a field measured curve from an offset propped fracture stimu-
lation. At a pump time of 20 minutes, proppant is on the formation (e.g., pad was pumped for
twenty minutes) and one hour later (e.g., at a pump time of 80 minutes) pressure starts to increase
indicating that fracture growth has stopped. Probably this job would have been pumped to comple-
Fig. 8.41 - Fracture Conductivity Redistribution Resulting from Excess Pad Volume.
tion, since pressure only increases by 500 psi after the start of the screenout, with this relatively
small increase possibly not even being noted in normal surface pumping records. However, unless
this screenout was a planned occurrence, it is probable that fracture length is much less than
desired. While unfortunate for this well, the information can aid in future treatment designs by sim-
ply noting the pad percentage at the start of the pressure increase.
For this case, pressure starts increasing after 80 minutes, with a pad pump time of 20 minutes -
thus pad percentage for the first part of the job was 25%. For future treatments, the pad percent-
age should be increased in volume to at least equal 25% of the total pump time. More accurately,
since pad percentage is related to job size, the pad percentage of 25% could be used to back out
a fluid efficiency. The fluid efficiency thus measured for the first 80 minutes of the job is then used
to calculate an expected fluid efficiency for a larger treatment (as discussed below), and this
expected efficiency for the total job is used to determine the new, required pad percentage and pad
volume.
Pad Volume
Once an efficiency (or expected efficiency) has been determined for a proposed treatment, the
required pad percentage for the job is found from the simple relation
2
f p = (1 e f ) + f c (8.15)
where ef is the expected efficiency for the treatment, fp is the required pad fraction for the treat-
ment, and fC is a correction term.
In developing this, consider the curve shown in Fig. 8.43. This curve illustrates fracture area grow-
ing with time (or volume). Further, consider that at some time, ftp (where tp is the total pump time
and f is a fraction) a switch is made from pumping pad to pumping proppant laden slurry. Thus,
the initial fracture area created (e.g., the small element of fracture created just as pumping starts)
is exposed to fluid loss for the entire pump time tp, with this fluid loss coming out of the pad from
time '0' to time ftp, with subsequent fluid loss coming out of, and serving to dehydrate, the proppant
laden slurry.
Similarly, one might consider some later element of the created fracture area, da, which is created
at time = (e.g., before that time it did not exist since the fracture had not reached that point) and
has a total exposure time to fluid loss of = (tp - ). For some fraction of that total exposure time
( < tp), fluid loss from this increment of the fracture area will come from the pad volume. After
that point, the slurry front passes and subsequent fluid loss out of that element of the fracture
area will be coming out of the slurry. Assume then, that this point in time where the slurry front
passes an element of fracture area is similar for each element of the fracture. Then for some incre-
mental area, da, total fluid loss exposure time is . For a fraction of this total time, f, fluid loss is
from the pad while for the remainder of the exposure time, fluid loss is from slurry.
The volume of fluid lost during a fraction, f, of each incremental fracture area's fluid exposure
time, , can then be found by integrating14
A f
d
V Loss ( f ) = 2C ------- da
0 0
= f x V Loss
where VLoss is the total volume of fluid lost during the entire pump time. Thus the portion of fluid
lost for a (constant) fraction of the fluid exposure time of each incremental area of the fracture is
simply proportional to f . Also, if this assumption concerning the slurry front passing each ele-
ment of the fracture is correct, then this simple curve (dashed line in Fig. 8.43) defines the perfect
pad. That is, the slurry front reaches the fracture tip just as pumping stops, e.g., it neither reaches
the tip prematurely leading to proppant bridging (a screenout), nor does it fail to reach the tip, leav-
ing a portion of the fracture without proppant or allowing harmful afterflow proppant redistribu-
tion during fracture closure.
Clearly then this is a possible curve for the optimum pad volume, and based on this curve, the
desired fraction, f = fp, is readily found. As discussed above, the volume of fluid lost during a frac-
tion, f, of each fracture elements' fluid exposure time, equals f x VLoss, where VLoss is the total
loss volume during the treatment. For the ideal pad then this fractional lost volume exactly equals
the pad volume giving
f p xV p = f xV Los
where Vp is the total volume injected during the entire pump time tp. Since efficiency, ef, is defined
as fracture volume at the end of pumping divided by the total volume injected, then VLoss, must
equal
V Loss = ( 1 e f )xV p
However, reviewing the dashed (slurry front propagation) curve in Fig. 8.43 shows a vertical
slope at the beginning, e.g., implying an initially infinite velocity for the slurry front. This is clearly
an impossibility, and leads to a correction factor,14 fC, as shown in Fig. 8.44.
where fC = 0.05, efficiency, ef, > = 0.20, = ef/4, efficiency < 0.20 .
Using this (somewhat in reverse) with the ideal case shown in Fig. 8.42 where the pad percentage
(prior to start of screenout) was 0.25 gives an efficiency on the order of
2 2
0.25 = ( 1 e f ) + 0.05, ( 1 e f ) = 0.20
e f = 0.55
for the first 80 minutes pumping of that job. (Note that in this case, the final efficiency is greater
than 0.20, thus the initial estimate of fC = 0.05 was correct, otherwise it would have been necessary
to iterate on the correction term in order to find the actual efficiency.)
Of course, while the dashed curve in Fig. 8.43 represents the general character of an ideal pad
stage, the assumption that each incremental fracture area element is exposed to pad fluid loss and
slurry fluid loss in the same ratio (e.g., 'f' is a constant for each incremental element of the fracture)
is not proven. As one proof, or at least justification, for this assumption, pad percentage and
proppant addition schedules (as discussed in the following section) arising from the efficiency
analysis are compared to schedules developed from computer models in Fig. 8.45. This shows
actual treatment schedules from three separate areas, representing fluid efficiencies ranging from
18 to 70%. The low loss, high efficiency example is for a tight gas field in Colorado where height
confinement was virtually perfect; the middle curve comes from treatment histories from a gas
field in East Texas where some height growth generally occurred; and the third, high fluid loss
example, was for fracturing in a thick, moderate permeability, carbonate formation in the North
Sea. In each case, computer model designs were based on extensive data collection programs and
field experience, and, in each case, the final proppant schedule is seen to be quite accurately deter-
mined by fluid efficiency alone.
c avg = W /V p (8.16)
where W is the total weight of the proppant and Vp is the total slurry volume (fluid plus proppant)
injected. Note here that this definition of proppant concentration differs from the normal field
usage of pounds-of-proppant per gallon-or-fluid. Additionally, cf is defined as the final, maximum
proppant concentration pumped during a treatment, and due to fluid loss, cf must be greater than
cavg. One possible design goal for a propped fracture stimulation is to, at the end of pumping, have
a uniform proppant concentration, equal to cf, from the wellbore to the fracture tip. This will gen-
erate a fracture with reasonably uniform conductivity along the fracture length (assuming a single
type of proppant is used) and will maintain fairly uniform slurry viscosity throughout the fracture.
In terms of the fracture volume at the end of pumping, V = ef x Vp, this final proppant concentration
can be written as
c f = W /V = W / ( e f V p ) .
c D avg = c avg /c f = e f ,
where cD-avg is a normalized value for average concentration. Similarly, a normalized concen-
tration at any point in time during the treatment is defined by
c D = c/c f ,
and, for convenience a new time scale is defined, , where the new time scale starts at 0 when
proppant is started and reaches a value of 1 at the end of the job as illustrated in Fig. 8.46.
= ( t f t p )/ ( t p f t p ) .
In terms of this new time scale, certain fixed values for the normalized proppant schedule, cD, can
be stated
c D ( ) = 0 ( < = 0 ),
cD ( ) = 1 ( < = 1 )
c D avg = e f .
the exponent, , can be evaluated from the above limits on the function, cD, given above
= 1 ef .
or after incorporating a correction factor discussed on page 8.62 for the pad volume calculations
= 1 e f f C /e f .
and since this function satisfies the numerical end points for a proppant schedule as stated above,
satisfies the relation for the final average proppant concentration, and also provides a monotoni-
cally increasing schedule as commonly utilized in practice - it is expected to be a reasonable
approximation to an ideal schedule. As seen in Fig. 8.45, again for three cases covering a range of
conditions and fluid efficiency, this simple relation does indeed provide an acceptable pumping
schedule.
Fluid efficiency is related to pump time (e.g., volume and rate), fluid loss coefficient, C, and to the
fluid loss area, or rp, the ratio of loss area to total fracture area. While these are the primary vari-
ables governing efficiency, it is also slightly affected by fracture geometry (e.g., confined height
vs. radial fracture growth) and fluid rheology. For a general case there is no analytical solution for
fluid efficiency, however, as with the other fracturing pressure decline analyses discussed earlier,
it is possible to place certain bounds. For example, for efficiency approaching 0 (e.g., very high
fluid loss), fluid efficiency is proportional to time raised to a power14
( 2n + 1 )/ ( 4n + 4 ) "PK" Geometry
e f t** n / ( 2n + 2 ) "GdK" Geometry
( 5n + 2 )/ ( 82 + 8 ) "Radial" Geometry
where 'n' is the power law exponent for a non-Newtonian fluid. 'n' generally ranges between 0.5
and 1 for common fracturing fluids, and using n = 0.75 (a typical value for crosslink gels) gives
While the range between these various possible fracture geometries is possibly significant in some
cases, it is noted that the values above are for the limited case of very high fluid loss. As efficiency
approaches 1 (e.g., no fluid loss), then the fracture geometry does not effect efficiency, and, in
the above form, efficiency is proportional to time raised to the 0 power, e.g.,
0
e f t = constant = 1 .
Interpolating between these limits gives a ratio of efficiencies between two different pump times
(t2 and t1) as
but, generally, acceptable accuracy is obtained by simplifying the above ratio to a single relation-
ship
( 1 e1 ) 3
( e f 2 /e f 1 ) = ( t 2 /t 1 ) (8.18)
Example
As an example, consider a case where a minifrac test was pumped. The test consisted of a cross-
linked gel identical to the fluid planned for use during the propped fracture treatment. The test used
25,000 gallons (595 barrels) pumped at 25 bpm with a total pump time, tp, of 23.8 minutes. Frac-
ture closure was observed 28.6 minutes after shut-in, e.g., tc = 28.6 minutes. This gives a dimen-
sionless closure time of
c = t c /t p = 28.6/23.8 = 1.20
while the minifrac efficiency can be corrected for the larger volume, it cannot be corrected for
rate changes, thus in order to use simply the efficiency from the minifrac, the propped stimulation
treatment must be pumped at the same rate. This gives [using Eq. (8.18)] an expected efficiency
for the actual treatment of
( 1 0.45/3 )
e f 2 /0.45 = ( 4/1 )
0.18
e f 2 = ( 0.45 ) ( 4 ) = 0.35 = 35% .
and since the total expected treatment volume is 100,000 gallons, the pad stage should consist of
47,000 gallons.
and this equation is used to construct the simple table shown in Table 8.7, where the slurry vol-
umes shown are arbitrarily selected points which will be used to construct a curve of prop con-
centration vs. slurry volume. It is particularly important to note that the calculations are conducted
in terms of slurry volume and slurry concentration, e.g., pounds of proppant per slurry gallon, so
a conversion is necessary to the more common industry terminology of ppg (pounds of proppant
per fluid gallon).
These conversions from ppg (pounds of proppant per fluid gallon - Cfl) to pounds of proppant per
slurry gallon (Csl) have been made using the formulae
and
Finally, these calculated points might be plotted as shown in Fig. 8.47, and a smooth curve con-
necting the points constructed - with this curve then describing the ideal proppant addition sched-
ule. This curve might then be the final job input for a computer controlled ramp type treatment,
or the curve might be subdivided into discrete stages as seen by the dashed line in the figure, with
these discrete stages then being used for job control.
8
Eff, Mini-Frac = 0.45
7 Expected Eff, Main Frac = 0.40
Rate = 25 BPM
6
5
PPG
4
3
2
1
0
20 40 60 80 100
Slurry Volume (M-gallons)
Time/Temperature History
The efficiency can also be used to determine an approximate time-temperature history for the treat-
ment as illustrated in Fig. 8.40 as discussed by Nolte, in his paper Determination of Proppant and
Fluid Schedules from Fracturing Pressure Decline.14
8.7 Nomenclature
A Total Fracture Area created after pumping for tp minutes (ft2)
C Fluid Loss Coefficient (ft/ minute )
P Pressure Difference (psi)
P* Match pressure for pressure decline analysis (psi)
Dimensionless Shut-In Time, = ts/tp
c Dimensionless Closure Time, c = tc/tp
ef Fracture Fluid Efficiency = Fracture Volume at Shut-In (V)/Total Volume Pumped (Vp)
E Young's Modulus of Formation (psi), Typical Values - 2x106 psi to 8x106 psi
E' Crack Opening Modulus = E/(1-2) (psi)
f Fraction
fp Pad Fraction or Pad Percentage
fpr Proppant Fraction of Job, Vpr/Vp
H Total or Gross Fracture Height (ft)
Hp Permeable or Leakoff Height (ft)
pc Fracture Closure Pressure (psi)
pnet Net Fracturing Pressure (e.g., bottomhole treating pressure just outside the perforations
minus fracture closure pressure) (psi)
ps Net Pressure at Shut-In (e.g., ISIP - pc)
Porosity of Proppant Pack (typically on the order of 0.40)
Q Total Injection Rate (barrels/minute, bpm)
qLoss Fluid Loss Rate (bpm)
rp Ratio of permeable or leakoff area to total fracture area for P&K or Geertsma rp = Hp/ H;
for a radial geometry rp is more difficult to define and is normally set = 1
Loss Ratio = efficiency/(1 - efficiency)
pr Specific Gravity of Proppant (e.g., 2.65 gm/cc or 22 lb gal for sand)
S Fracture Stiffness for Pressure Decline Analysis
tc Closure Time, e.g., Shut-In Time to Fracture Closure (minutes)
tp Pump Time (minutes)
ts Shut-In Time (e.g., incremental time since pumping stopped) (minutes)
Time when an incremental element of fracture area is first exposed to fluid loss
V Fracture Volume (ft3)
VLoss Total Fluid Loss Volume During Pumping (ft3)
Vp Total Slurry Volume Pumped (ft3)
Vpr Total Proppant Volume Pumped (ft3), including porosity of proppant
Vfl Total Fluid Volume Pumped (ft3)
Dimensionless Shut-In Time, ts/tp or (t-tp)/tp
W Total weight of proppant pumped (pounds)
Poisson's Ratio for Formation (dimensionless), Typical Values - 0.15 to 0.25
Fluid Viscosity (centipoise)
8.8 References
1. Godbey, J. K. and Hodges, H. D.: Pressure Measurements During Fracturing Operations, Trans., AIME, (1958)
213, 65-69.
2. Khristianovic, S. A. and Zheltov, Y. P.: Formation of Vertical Fractures by Means of Highly Viscous Liquid,
Proc. Fourth World Pet. Cong., Rome (1955) Sec. II, 579-86.
3. Perkins, T. K. Jr. and Kern, L. R.: Widths of Hydraulic Fractures, JPT (Sept. 1961) 937-49; Trans., AIME 222.
4. Geertsma, J. and de Klerk, F.: A Rapid Method of Predicting Width and Extent of Hydraulic Induced Fractures,
JPT (Dec. 1969) 1571-81; Trans., AIME 246.
5. Veatch, R. W. and Crowell, R. F.: Joint Research/Operations Programs Accelerate Massive Hydraulic Fractur-
ing Technology, JPT (Dec. 1982), 2763-75.
6. Nolte, K. G. and Smith, M. G.: Interpretation of Fracturing Pressures, JPT (Sept. 1981), 1767-75.
7. Nolte, K. G.: Determination of Fracture Parameters from Fracturing Pressure Decline, paper SPE 8341, pre-
sented at the 1979 SPE Annual Technical Conference and Exhibition, Las Vegas, Sept. 23-26.
8. Schlottman, B. W., Miller, W. K. II, and Leuders, R. K.: Massive Hydraulic Fracture Design for the East Texas
Cotton Valley Sands, paper SPE 10133, presented at the 1981 SPE Annual Technical Conference and Exhibi-
tion, San Antonio, Oct. 4-7.
9. Elbel, J. L. et al.: Stimulation Study of Cottage Grove Formation, JPT (July 1984) 1199-1205.
10. Dobkins, T. A.: Procedures, Results, and Benefits of Detailed Fracture Treatment Analysis, paper SPE 10130,
presented at the 1981 SPE Annual Technical Conference and Exhibition, San Antonio, Oct. 4-7.
11. Smith, M. B.: Stimulation Design for Short, Precise Hydraulic Fractures SPEJ (June 1985) 371-79.
12. Smith, M. B., Miller, W. K. II, and Haga, J.: Tip Screenout Fracturing: A Technique for Soft, Unstable Forma-
tions, SPEFE (Feb. 1987) 95-103; Trans., AIME, 283.
13. Morris, C. W. and Sinclair, R. A.: Evaluation of Bottomhole Treatment Pressure for Geothermal Well Hydraulic
Fracture Stimulation, JPT (May 1984) 829-36.
14. Nolte, K. G.: Determination of Proppant and Fluid Schedules From Fracturing-Pressure Decline, SPEPE (July
1986) 255-65; Trans., AIME, 281.
15. Nolte, K. G.: A General Analysis of Fracturing Pressure Decline With Application to Three Models, SPEFE,
(Dec. 1986) 571-83.
16. Martins, J. P. and Harper, T. R.: Mini-frac Pressure Decline Analysis for Fractures Evolving From Long Perfo-
rated Intervals and Unaffected by Confining Strata, paper SPE 13869 presented at the 1985 SPE/DOE Low-Per-
meability Gas Reservoirs Symposium, Denver, May 19-22.
17. Castillo, J. L.: Modified Fracture Pressure Decline Analysis Including Pressure-Dependent Leakoff, paper
SPE 16417, presented at the 1987 SPE/DOE Low-Permeability Gas Reservoirs Symposium,.Denver, May 18-19.
18. Cooper, G. D., Nelson, S. G., and Schopper, M. D.: Comparison of Methods for Determining In-Situ Leakoff
Rate Based on Analysis With an On-Site Computer, paper SPE 13223 presented at the 1984 SPE Annual Tech-
nical Conference and Exhibition, Houston, Sept. 16-19.
19. Warpinski, N. R.: Investigation of the Accuracy and Reliability of In Situ Stress Measurements Using Hydraulic
Fracturing in Perforated, Cased Holes, Proc., 24th U.S. Symposium on Rock Mechanics, College Station, TX,
(June 1983) 773-86.
20. McLennan, J. D. and Rogiers, J. C.: How Instantaneous are Instantaneous Shut-In Pressures, paper SPE 11064,
presented at the 1982 Annual Meeting of SPE, New Orleans, Louisiana, Sept. 26-29.
21. Warpinski, N. R. and Teufel, L. W.: In-Situ Stresses in Low Permeability, Nonmarine Rocks, JPT, April, 1989.
22. Miller, W. K. II and Smith, M. B.: Reanalysis of the MWX-Fracture Stimulation Data from the Paludal Zone of
the Mesaverde Formation, paper SPE 19772, presented at 1989 Annual Fall Meeting of SPE, San Antonio, Tex-
as, Oct. 8-11.
23. Nordgren, R. P.: Propagation of a Vertical Hydraulic Fracture, SPEJ (Aug. 1972) 306-14; Trans., AIME, 253.
24. Carter, R. D.: Appendix I to paper by C. C. Howard and C. R. Fast, Optimum Fluid Characteristics for Fracture
Extension, presented at the 1957 ASME Spring Meeting, Mid-Continent District, Div. of Production, Tulsa,
OK, April.
25. Warpinski, N. R.: Dual Leakoff Behavior in Hydraulic Fracturing of Tight, Lenticular Gas Sands, SPE Pro-
duction Engineering (August 1990) 243.
9.1 Introduction
After 40 years of growth in income, we are now in a period where there will be little growth. We
have to continue to rationalize both staff and assets to reduce our operations to the size required
for expected level of (future) investment and to reduce costs so that cash flow can be maximized.
The fat, lazy days are over. We must continue to become leaner and meaner. We must improve our
efficiency. This is the charge made by the authors of a paper entitled Petroleum Reinvestment-
Is there a future for our Industry?
Doom and gloom or a challenge to be overcome? These statements bring home the importance of
properly maximizing cash flow in the management of our oil and gas properties and emphasize the
need to focus on immediate opportunities to bring about revenue improvement. Well stimulation,
either by acidizing or through hydraulic fracture stimulation, is one method available to generate,
virtually overnight, improved production revenues that will assist in our accomplishing this goal.
Well stimulation, however, is a business decision that can just as easily result in an investment loss
if not properly understood and applied.
Amoco Corporation has traditionally reinvested over 50% of it's total earnings in Amoco Produc-
tion Company (APC) for the sole purpose of developing reserves and the resulting production of
oil and gas. Over the last decade, APC has developed and applied hydraulic fracture stimulation
technology worldwide, an investment that today provides over 50% of all oil and gas produced in
our domestic U.S., Canadian and North Sea operations. Price declines in recent years have made
it increasingly difficult to justify investment in drilling, completing and stimulating wells. Low
prices have been compounded by an increased incidence of poor economic returns and project cost
overruns, as summarized in Table 9.1, suggesting better risk management procedures must be
included as a part of economic analysis and stimulation optimization. This section addresses the
methods to follow and the pitfalls to avoid when maximizing revenue from the implementation of
hydraulic fracture treatments.
Table 9.1 - Average of Gulf of Mexico Projects to 1988.1
Production: -10%
Reserves -9%
Economic optimization of a well stimulation treatment requires that the designer carefully balance
a large number of parameters describing the reservoir, including its fluid and rock properties, with
the inflow performance and associated cost of providing a man-made flow conduit that will pro-
duce the largest production increase at the least incremental cost. There are usually many solutions
to this problem because the different stimulation materials and their associated costs can be com-
bined in many ways to produce an optimum. The challenge facing us today is to consider all mate-
rials and sensitivities, and their associated risks, to arrive at the true optimum, a task that is by
no means trivial and is best suited to todays computer technology. Amoco Production Research
has developed an integrated fracture, reservoir, and economics program called ULTRAFRAC.
This program allows the user to assess the economic benefits and sensitivities of the fracturing pro-
cess. The following sections are some of the more important considerations to be evaluated when
optimizing stimulation treatments.
Amoco evaluates investment projects on the basis of several standards. The most important of
these will be discussed in this section, and the merits and shortcomings of each will be outlined.
As the discussion proceeds, it will become clear that no single measure is sufficient to adequately
analyze a project and that an evaluation utilizing a variety of measures is desirable. The measures
used within Amoco are defined as follows:
1. Net Present Worth or Value (PW or PV) The sum of all future cash flows discounted to the
initial time, at a stated discount rate.
2. Incremental Present Worth or Value of the Fracture (INCPVF) The Net Present Value of a
fracture case less the present value of the unfractured case.
3. Fracture Incremental Present Worth or Value (FINCPV) The Net Present Value of a fracture
case less the present value of the preceding case. Used to show diminishing returns.
4. Profitability Index (PI) The [continuous] compound interest rate whose discount factors make
the present worth of a projects net cash flows equal to zero.
5. Discounted Return on Investment (DROI) The ratio of a projects net present worth to the
present worth of the total investments discounted at a stated rate. (The denominator is calcu-
lated after tax and overhead and includes investment tax credits and the after-tax effect of de-
preciation.) In ULTRAFRAC, DROI includes capital expenses such as well costs in addition
to fracturing costs.
6. Fracture Discounted Return on Investment (FDROI) FDROI is defined as above only cap-
ital costs such as well costs are excluded. Only the AFIT (After Federal Income Tax) fracturing
costs are used in this economic analysis.
8. Payout (PO) The time for the cumulative undiscounted cash flow of a project to reach zero.
For example, a dollar received now would, at 5% annual interest, be worth $1.05 after one year.
Hence, to be indifferent between accepting a dollar now or a certain sum of money one year in the
future, that sum of money would have to be $1.05 (assuming 5% return is the highest return avail-
able to investors). The future worth (FW) of a dollar after one year at 5% is calculated as follows:
FW = 1.00 (1 + .05)
= 1.05
After two years, if the interest were left in the account, the future worth would be:
FW = 1.00 (1 + .05) (1 + .05)
= 1.00 (1.05)2
= 1.1025
Present worth is the value that, when invested at the given interest rate, will yield the given future
worth after the applicable number of periods. Using the previous example of $1.05 received after
a year, the present worth is $1.00 (since it would grow to the future worth of $1.05 when invested
at 5% for one year). Another way to think of present worth is the value in current dollars you would
require to make you indifferent between receiving that amount or the future worth.
The relationship of present and future worth can be stated generally as,
FW = PW (1 + i)n (2.1)
where FW = future worth, PW = present worth, i = interest rate (assumed constant), and n= number
of periods over which the interest rate applies.
In general terms, present worth is found by solving Eq. (2.1) for PW.
1
PW = FW -----------------n- (2.2)
(1 + i)
The quantity
1
-----------------n-
(1 + i)
The form of present worth discussed so far is known as end-of-period discrete (or periodic) dis-
counting. If one assumes that the time period over which compounding occurs is infinitesimally
short, the result is continuous discounting, the type employed Amoco. With continuous discount-
ing, the present worth is determined as follows:
FW
PW = --------
- (2.3)
e ni
The use of tables and computer programs simplifies the calculation of the discount factor 1/eni.
If more than one future amount, occurring at different times, is being discounted, it is necessary to
alter the equation to account for multiple cash flows. Eq. (2.4) illustrates the case of n cash flows,
each assumed to occur at year end.
PW = C o + C 1 ( DF 1 ) + C 2 ( DF 2 ) + ... + C n ( DF n ) (2.4)
where C0, C1, ..., Cn = annual point-in-time cash flows for years 1 through n and DF1, DF2, ..., DFn
= associated continuous discount factors for years 1 through n.
The discussion of present worth thus far has centered around cash flows which occur at a point in
time. More frequently, however, cash flows occur uniformly throughout a period, rather than at
year end. An example of a uniform cash flow is revenue from an oil well. The oil is not all produced
on December 31, 19xx; therefore end-of-year discounting is not appropriate. An example of a sit-
uation tailored to use end-of-period discounting might be annuity payments received at year end
for several years.
Table 9.2 summarizes the types of discounting and cash flows which exist and the applicable dis-
count factor tables, which are included, along with brief instructions, in a separate section of this
manual. Only the continuous form of discounting is utilized by Amoco and all future references to
discounting will be to that form.
Table 9.2 - Summary of Discounting and Cash Flows.
Annual continuous discount factors, the type normally used by Amoco, for point-in-time cash
flows are listed in Table 9.3, and factors for uniform cash flows are listed in Table 9.4. Examples
of present worth calculations for both uniform and point-in-time cash flows are also provided. For
anything other than the simplest of examples, computer programs such as ULTRAFRAC and
GEM handle the calculations.
Table 9.4 also shows an example of present worth calculation. The annual $75 M project net cash
flow streams are assumed to result from a $100 M investment. Discounted cash flows are obtained
by multiplying the annual net cash flows by the appropriate discount factors. The present worth of
the project is the sum of the discounted cash flows. Present worth has been calculated at 15% dis-
count rate for point-in-time and uniform cash flows.
Table 9.3 Calculation of Present Worth Using Continuous Discount Factors (Amoco).
0 -100 - -100
1 75 .8607 64.6
2 75 .7408 55.6
3 75 .6376 47.8
68.0 = PW15
(Point-in-Time)
0 -100 - -100
The significance of present worth is that, provided an investor has other investment opportunities
at the stated discount rate, he would be indifferent to accepting $81.1 M now or accepting the
undiscounted uniform cash flows over the three years of project life. In fact, the value of a firm is
frequently said to be the present worth of all of its cash flows from its various projects.
Present worth is helpful in ranking projects of the same size as illustrated by Table 9.5:
In examining these projects, it is clear that an investor would favor project A over B, because
Project B for the same investment ($1,000 M) yields $100 M less per year over the three-year
project life. Project A and Project C, however, each return a total of $500 M, and the concept of
present worth aids in differentiating between them. Project C is preferred because it returns more
of its cash earlier which leads to its having a higher present worth (the incoming cash can be rein-
vested). This once again emphasizes that both the timing and magnitude of investments have to
be considered. It is interesting to note that Project B, while returning all of its investment, still has
a negative present value at both 13% and 15% discount rates.
If this firms cost of capital is 13%, it would undertake all projects with a PW13 > 0, accepting
project A and C but rejecting B. However, if the firm were capital constrained, it would rank the
projects in order of economic attractiveness and choose those which maximize the value of the firm
within the imposed constraints.
Amoco has set a minimum investment criterion that those projects accepted must have a positive
PW15. Subject to the size of Amocos investment budget and manpower constraints, those
projects should be selected which maximize the present worth of the total package of projects
available.
Profitability Index
Profitability Index (PI) is defined as that [continuous] compound interest rate whose discount fac-
tors make the present worth of a projects net cash flows equal to zero. PI is also referred to as the
projects internal rate of return.
The PI may also be thought of as the discount rate which sets the sum of the discounted annual
cash inflows equal to the sum of the discounted annual cash outlays. Investments normally occur
at the commencement of a project, followed by a number of years of cash inflows. Where this pat-
tern is substantially altered, there may be multiple PIs, which is a serious limitation to the use of
this technique.
An example may be helpful in explaining PI. Suppose a firm is offered a project with annual end-
of-year point-in-time cash flows of $100 M for five years after an initial (time zero) investment
of $350 M. The calculation of PI for such a project is shown in Table 9.6.
Table 9.6 - Calculation of Profitability Index.
Discount Discount
Time (years) Cash Flow ($M) Factors Present Value Factors Present Value
+4.0 -15.0
Recall that the PI is that discount rate which sets the present worth of the project equal to zero.
Therefore, by interpolation,
4
PI = ------ x ( 14% 12% ) + 12%
19
PI = 12.4 approximately
Once the PI is calculated for a proposed project, it should be compared to the established standard.
In the current environment for Amoco, the minimum standard is 15 PI (or PW 15 0 ). Projects
which yield less than a 15 PI should not generally be accepted. However, other considerations,
such as an interrelationship with more profitable opportunities, may lead to their acceptance.
Should Amocos supply of projects returning at least 15 PI dwindle to the point where the available
monies exceed the investment requirements for such projects, the minimum PI standard would pre-
sumably be lowered, but never less than the cost of capital. Investors would prefer that Amoco pay
out the excess funds as dividends if they can earn higher return than can be realized by plowing
the funds back into Amocos operations. Amoco might also choose to invest the funds elsewhere
within the consolidated corporation if projects in other lines of business could yield a higher PI.
DROI is a measure of capital efficiency which may be viewed as the amount of after-tax present
worth generated per dollar of discounted investment. It is only used within Amoco Productions
domestic operations. Differing fiscal regimes in foreign countries make it difficult to define the
denominator of the expression on a consistent basis, so the measure is not useful to any subsidiary
having operations outside the United States. To understand how DROI is useful in economic eval-
uations, it may be worthwhile first to review other evaluation criteria, and the circumstances under
which they are useful. Some of their shortcomings will illustrate the utility of DROI.
When considering two mutually exclusive projects with the same investment, the one with the
higher present worth should be accepted. Likewise, when considering an entire collection of poten-
tial projects with different investment requirements (such as during budget preparation), the
present worth of the total package should be maximized. The decision as to which projects to
include and which to reject is complicated by the fact that not all projects offering a given present
value require an equal capital investment. DROI is a useful tool for dealing with this problem, as
illustrated by the following group, in Table 9.7, of potential projects available to a firm:
Table 9.7 - Utility of DROI in Project Ranking.
A 12 6 21 9 1.50
B 8 4 17 5 1.25
C 4 2 18 4 2.00
D 6 3 19 2 0.67
E 2 1 16 3 3.00
F 2 1 20 2 2.00
G 8 4 14 -2 -.50
* Assumes these are after-tax numbers and that no overhead, tax credits, or depreciation
credits exist.
Assume that this years capital budget allows $20 MM of expenditures. Since the projects return-
ing at least 15 PI exceed the available funds, some projects must be foregone. Under these condi-
tions, the firm should rank its projects in such a way as to maximize the present worth of the
package of projects. Ranking these projects on the basis of the highest PW15 results in Projects A
and B being selected with a combined PW15 of $14 MM. Ranking these projects on the basis of PI
results in the selection of projects A, F, and D with a combined PW15 of 9 + 2 + 2 = $13 MM for
the total $20 MM investment. Ranking on the basis of highest DROI15 yields projects E, C, F, and
A for a combined PW15 of 3 + 4 + 2 + 9 = $18 MM for the $20 MM investment, which is consistent
with the goal of maximizing PW15 of the package of projects given the spending limitations.
The PW method of ranking fails in the situation described above because of the different invest-
ments required to yield a given present worth. The PI method also fails to rank projects since it
implies an ability to reinvest cash thrown off by a project at the PI rate. Since this is not generally
the case, the PI method does not compare projects on a consistent basis.
In summary, DROI is of use in ranking projects of different investment magnitudes. It takes into
account the time value of money and it also measure a projects susceptibility to risk. In the above
example, a DROI15 of 1.50 is the minimum which would be accepted. Amoco in fact has no rigid
minimum DROI criterion. In general, where a 15 PI is Amocos minimum investment standard, a
DROI15 would be determined and used to rank the available investment projects. A DROI15 equal
to zero will indicate that the 15 PI standard has been met. While DROI provides a consistent
method of ranking projects, other factors such as payout, ROI, and maximum cash out-of-pocket
may be considered depending upon the investment climate.
Payout
Payout (PO) is defined as the length of time taken for the cumulative cash flow of a project to reach
zero. For some projects payout provides a rough measure of risk, by indicating how long the
investment capital is exposed. Amoco has no specific payout time criterion. When neither present
worth, PI nor DROI distinguishes between two mutually exclusive projects, the one with the
shorter payout is generally preferred.
The major shortcoming of the payout standard is that it fails to account for the timing of cash flows,
or to recognize cash flows after payout. If, for example, most of the project life occurs after payout,
later cash flows are not considered by the payout criterion. Table 9.8 summarizes a comparison of
two projects which have identical payouts but differ in present worth and illustrates how the timing
of cash flow is ignored by payout.
When used in combination with PI and present worth, payout does serve a useful purpose. Not only
does it indicate how long investment capital is at risk, but it also functions as a rough measure of
liquidity. For instance, if Amocos management decided that all available capital was to be needed
next year for a major expenditure, e.g., a large acquisition, then payout time could be the determin-
ing factor in ranking economically qualified projects.
0 -$2000 -$2000
1 1500 1000
2 500 1000
3 1000 1000
PI = 23.3 21.0
Return on Investment
Return on Investment (ROI) is defined as the ratio of the undiscounted cumulative net cash flow
of a project to the total investments (after tax and overhead and including investment and depreci-
ation tax credits). The ROI calculation is performed in the same manner as the DROI calculation
(shown on page 9-8) with the exception that all values are undiscounted in the ROI equation. When
comparing project with similar cash flow patterns, such as a number of individual development
drilling wells, ROI, in combination with payout, can provide an indication of project attractive-
ness.
Like payout, however, ROI does not account for the time value of money. This is illustrated by the
two projects in Table 9.9 which are identical with regard to ROI. When evaluated on a present
worth basis, which accounts for the time value of money, Project B is clearly preferred.
Another characteristic of ROI, which may be misleading, is that the measure increases dramati-
cally with an increase in project life. The example in Table 9.10 clearly demonstrates this effect
for five projects, each of which shows a 15 PI on a single $1,000 time zero investment. The cash
return is the total amount of cash to be returned to the investor at the end of the project.
All five projects are equally attractive assuming the ability to reinvest the cash in similar 15 PI
opportunities over the lives of the projects.
Amoco has no minimum ROI standard, for reasons which are apparent from the above example.
The high ROI, long-life project does have the advantage that the company does not have to go out
and find a 15% reinvestment opportunity quite as soon, but as long as it is assumed that such oppor-
tunity can be found, there is no need for a minimum ROI. Requiring minimum ROIs indicates that
the company does not have the ability to find reinvestment opportunities. As a result, ROI is not
included in ULTRAFRAC.
0 -200 -200
1 100 150
2 100 150
3 150 100
4 150 100
1 1,162 15 0.16
5 2,117 15 1.12
10 4,482 15 3.48
20 20,089 15 19.09
50 1,808,042 15 1,807.04
Incremental Economics
The PI standard should be employed to qualify projects for acceptance, but not to select between
mutually exclusive projects, i.e., projects such that either Project A or Project B may be under-
taken, but not both.
Incremental economics should be run in this case. If both projects return positive cash flows, there
is an opportunity cost in opting for one over the other. Hence, the benefit to the firm, in terms of
increased cash flow, is the difference (or increment) between the two cash flows.
An importance use of incremental economics is shown by the example below (Table 9.11). The
two alternatives represent the options of developing or dropping a certain lease. Note that because
Alternative A generates tax benefits with no cash expenditures, the resulting PI is infinite.
Examining either mutually exclusive option in isolation can result in an incorrect decision. In the
example, while Alternative A provides a positive PW15 due to the benefit of being able to write off
Alternative A Alternative B
(Drop) (Develop)
PI 19
the asset on current taxes, it is less than the PW15 of Alternative B. On the other hand, deciding on
Alternative B means foregoing the option of dropping the lease (an opportunity cost). The net ben-
efit to Amoco of developing would not be $3.5 MM, but rather $0.5 million.
When considering development of a lease, it is important to examine the drop alternative since
doing nothing is generally a poor alternative. Dropping the lease at least has the advantage of tax
write-offs. A development vs. drop analysis is ideally handled by incremental economics, as in the
above example. On occasion, the alternatives may both have negative (but different) PW15s, but
an incremental PW for one alternative over the other will always be positive.
Mutual exclusivity frequently gives rise to multiple PIs since the cumulative incremental cash flow
may have several sign reversals. In that case, the PW vs. discount rate profile would cross the hor-
izontal axis (PW=0) more than once (Table 9.12). The following example illustrates this situation.
Table 9.12 - Illustration of Multiple or Dual PI.
6-7 50 0 -50 50
7-8 25 0 -25 25
8-9 25 0 -25 0
The incremental cash flow in this case represents the benefit to the firm of selection Project B over
A. The cash flow of Project A becomes an opportunity cost which is subtracted from Project B to
determine the incremental cash flow. The present worth profile would be of the general shape
shown on Fig. 9.1. Points C and D indicate the discount rates for which the present worth is zero
(definition of PI).
This type of present worth profile is typical of most incremental projects. To avoid the problem of
multiple PIs, the present worth of the incremental cash flow stream (B-A) at the marginal reinvest-
ment rate should be examined. A positive PW15 would imply acceptance of Project B.
Sometimes the incremental cash flow approach is hard to apply. On some of the more complicated
scenarios which arise, the correct incremental cash flow stream is difficult to identify. However,
the importance of choosing the correct project alternatives and properly defining the problem can-
not be overstressed. Failure to do so may lead to a decision which does not maximize the present
worth of the total cash flows and, hence, of the corporation.
Situations in which present worth and PI may rank mutually exclusive projects differently occur
when the investment cost of one is larger than another, or when the timing of the projects cash
flows differs. Examples of mutually exclusive projects include the farm-out vs. drill decision and
the choice of 40-acre spacing vs. 20-acre spacing in the same field.
An example where PW and PI give different rankings to projects with dissimilar investments is
illustrated in Table 9.13. Project A calls for the investment of $100 and yields $150 after one year.
Its PI would be 40.6 with continuous discounting (point-in-time cash flow) and its PW15 would be
$29. Project B, in contrast, would require a $1 million investment and provide $1.25 million at the
end of a year. Its PI is only 22.4 but its PW15 is $75,884. The two methods rank the projects differ-
ently, as the PI of A is greater than the PI of B, but the PW15 of B is greater than the PW15 of A.
Obviously, you would prefer project B as it returns significantly more than the present worth.
Table 9.13 - Comparison of PW vs. PI for Ranking.
0 -100 -1,000,000
1 150 1,250,000
PI 40.6 22.4
PW15 29 75,884
An example of projects differing in the timing of their cash flows is shown Table 9.14. In ranking
Project C and Project D on the basis of PI, Project C would appear to be the better option. However,
a closer examination reveals that Project D has the higher PW15.
Table 9.14 - Timing of Cash Flow.
0 -25,000 -25,000
0-1 15,000 0
PI = 53 45
Fig. 9.2 is a plot of present worth vs. discount rate for two mutually exclusive projects such as the
40-or 20-acre spacing alternatives, which shows the curves crossing at some positive PW. Note
that the particular discount rate at which the decision is made (15% in this example) determines
the selection. At the intersection of the two curves one would be indifferent between 40- and 20-
acre spacing.
PI causes problems in reaching a decision when multiple (Dual PI) solutions occur, as shown in
the previous example. PI is defined as the intersection of the PW profile with the horizontal axis.
Note that in that example (Fig. 9.1), the profile has two points of intersection with the axis. In Dual
PI projects, PI should not be used as a ranking criterion. In this example, it is more appropriate to
utilize present worth and Discounted Return on Invement in the ranking process.
Why then use PI at all? There are several advantages to the PI method. One advantage is that it can
be compared directly with the cost of capital and anticipated rate of return. A second advantage is
that, unlike the PW method, PI abstracts from the size of a project. A PW15 of $50,000 can be
obtained on a $10 million investment as well as on an original outlay of $25,000. Accordingly, it
is possible to distinguish these two different sized projects on the basis of PI, but not on the basis
of present worth. A third advantage, and not an insignificant one, is Amoco managements famil-
iarity with PI. If management has a basic familiarity with the method, they can feel more confident
in their decision-making process. Despite these advantages, it is important to be aware of the short-
comings of PI, as well as those of each of the other investment criteria.
Timing enters into economic analysis in yet another way. The time of the analysis relative to the
life of the project must be established. Most of the discussion of investment decision-making so
far has centered around the timing and magnitude of cash flows produced by a project as viewed
at the present time. Fig. 9.3 indicates the cash flows and the point at which the analysis is under-
taken (time zero) for such a project. Note that the analysis and initial investment occur at time zero,
with cash flows received later in the project life.
0
1 2 3 4 Project Life
-1,000
Not all analyses are conducted before the initial investment is made. In the case of a develop vs.
drop decision on a well proposal, a reanalysis may be required after a considerable investment out-
lay has already occurred. Perhaps estimates of reserves have fallen or operating costs have soared.
Fig. 9.4 illustrates a well reassessment made after the initial investment spending occurred at time
t = -2. In this case, how should the economics be calculated?
Project
-1 0 1 2 3 4 Life
-1,000
The original investment of $1,000 represents a sunk cost and the $200 received at time t = -1 is a
benefit already received. No current decision can affect past expenditures, and conversely, no past
spending should be considered in a yet-to-spend decision. One qualifier to this statement exists.
Past spending, or sunk cost, may affect future economic decisions via an impact on future taxes.
Such effects must be considered in a yet-to-spend analysis.
The rationale of yet-to-spend economics, which evaluate only current and prospective cash flows
and disregard sunk costs, can best be illustrated Table 9.15. Assume that $500,000 (after tax) is
spent on exploration in a certain area and that two fields are found. The fields are subsequently
developed at a cost of $600,000 per field (after tax). One field is projected to have an operating
cash flow, after all operating costs, royalties, and local and federal taxes, of $2,000,000, and a PW15
of $560,000. The second field, of poorer quality, will have an operating cash flow of only $800,000
with a PW15 of $80,000. A yet-to-spend evaluation would show that both fields have positive
PW15s and PIs of 15 or better. Accordingly, both would be developed.
Table 9.15 - Rationale of Point Forward Economics.
If the sunk exploration costs ($250,000 per field) were considered when deciding whether or not
to develop the discoveries, the net cash flow and PW15 would differ, and the decision would differ.
Table 9.16 - Full Cycle Economics
Full-Cycle Economics
In fact, Field B would not be developed, and all the exploration costs would have to be assigned
to Field A. In this event, an analysis of the full-cycle economics shown as Table 9.16 of developing
Field A (including all sunk and anticipated cash flows over the life of a project) would show a final
net cash flow of $900,000 ($2,000,000 less $600,000 development cost and $500,000 total explo-
ration cost) and a PW15 of $60,000 ($310,000 less Field Bs $250,000 share of the exploration cost
at time zero). This answer is incorrect because by developing Field B, the total full-cycle net cash
flow would be $1,100,000, with a PW15 of $140,000, which is greater than that of developing Field
A only. Thus the analysis which considers sunk costs leads to an incorrect investment decision.
It must be remembered that past expenditures may have a substantial effect on the future tax con-
sequences. Previous costs may affect depreciation, cost depletion, and the gain or loss resulting
from sale or abandonment of the original project. As a result, future tax liabilities would be altered.
In analyzing future investments or other alternatives, considerations must be given to the cash
effects of the future tax consequences. Although sunk costs should be disregarded in a yet-to-spend
investment decision, except as to the resulting future tax consequences, they should be considered
in compiling a PIA. PIAs will be discussed in detail in Section IV.
Fracturing Pumping Equipment: Pump truck costs base minimum charges for all trucks except
pressure multiplier pumps, per well, for a period up to 4 hours continuous service, on location, per
hydraulic horsepower ordered. Prices are based on pumping pressure, and hydraulic horsepower
pumping charges increase with pumping pressure increment increases. Other costs include addi-
tional pumping time over 4 hours, nonpumping service time, minimum pump truck charges and
standby pumping equipment.
Propping Agent Pumping Charge: These charges apply when propping agents are pumped with
any fluid and are in addition to the fracturing pump truck charges. Prices per unit weight (usually
100 lbs (CWT)) are based on the type and size of the proppant.
Pressure Multiplier Pumps: These are usually required for pumping pressures in the 10,000 -
20,000 psi range. Charges include pressure multiplier pump base charges, per well, for up to 4
hours continuous service on location, per hydraulic horsepower ordered. Prices are based on
pumping pressure. Other costs include additional pumping time over 4 hours, nonpumping service
time, minimum charges, standby unit charges, and propping agent pumping charges.
Blender Services: Base charges for continuous proportioning and mixing of propping agent and
fracturing fluid, based on average injection rate, first 4 hours or fraction, per well. Other costs
include blender services time over four hours, based on pumping rate, nonpumping blender time;
blender standby; other blender and equipment charges such as paddle mixers, densitometers, etc.
Slurry Concentration Handling Service: These charges apply when propping agents are pumped
with any fluid and are in addition to blender charges and propping agent pumping charges. Prices
depend on propping agent concentration.
Auxiliary Stimulation Equipment: These items include sand handling equipment, radioactive
material for tagging sand, wellhead protective injection equipment (tree-savers, etc.), manifolds,
nitrogen, CO2 equipment, flow meters, fracturing support units, special equipment (tanks, transfer
pumps, valves, wellheads), ball sealer equipment, treating connections left on location, sand con-
centrators, etc.
9.4 References.
1. Campbell, J. M.,Analysis and Management of Petroleum Invests, Risk, Taxes and Time.
2. Prats, M.: Effect of Vertical Fractures on Reservoir Behavior - Incompressible Fluid Case, SPEJ (June 1961)
105-18;Trans., AIME, 222.
3. McGuire, W. J. and Sikora, V. J.: the Effect of Vertical Fractures on Well Productivity, Trans., AIME (1960)
219, 401-04.
4. Tinsley, J.M. et al.: Vertical Fracture Height - Its Effect on Steady-State Production Increase, JPT (May 1969)
633-38; Trans., AIME, 246.
5. Elkins, L.E.: Western Tight Sands Major Research Requirements, Proc., Gas Research Inst./American Gas
Assn./U. S. DOE Intl. Gas Research Conference, Chicago (June 9-12, 1980).
6. Petroleum Production Handbook, T. C. Frick (ed.), SPE, Richardson, TX (1962) Chap. 38.
7. Guerrero, E. T.: Practical Reservoir Engineering, The Petroleum Publishing Co., Tulsa, OK (1968) 72-75.
10.1Fracturing Tests
Introduction
The success of a fracture stimulation depends on the accuracy of the design theory, an understand-
ing of the propagation or growth of hydraulic fractures, and the accuracy of design parameters.
Many field and laboratory tests are available which allow a more accurate approximation of frac-
turing parameters. This section covers the more widely used tests; providing a description of the
test procedures and in some cases interpretation guidelines. Descriptions are included for core
tests, prefrac logs, perforation and permeability determination, bottomhole treating pressure mea-
surements, closure stress tests, minifracs, postfrac logs, and fracture azimuth determination.
Core Tests to Determine Mechanical Rock Properties and Fluid Loss Coefficient
Fluid Loss Coefficient Core can be analyzed to determine elastic modulus, Poisson's Ratio, and
fluid loss coefficient for use in fracture stimulation design. Core analysis is currently the best tech-
nique available for obtaining elastic rock properties.
Full diameter cores should be cut through the interval of interest, including both the pay zone and
adjacent formations, with coring of adjacent formations of sufficient thickness to obtain represen-
tative samples. In many cases, a gradation occurs from one bed to another; such as shale grading
into a sandstone forming a siltstone transition bed. In a case such as this, mechanical properties
tests performed on the transition core would not be representative of the adjacent shale formation.
When available, open hole logs from an offset well should be used to determine the required coring
interval.
Core for rock properties tests should have a minimum diameter of 2-1/2 inches, since the tests uti-
lize a 3/4-inch diameter by 1.5-inch long plug which is cut perpendicular to the long axis of the
core. The core should be peel-sealed on location. Peel-sealing the core prevents dehydration of the
samples, which provides a more accurate measure of elastic and mechanical properties at in-situ
conditions. Transporting the core back to a warehouse for peel-sealing allows excessive dehydra-
tion of samples. Past attempts to designate specific portions of the core interval to be peel-sealed
have led to confusion, and critical portions of the core have sometimes been left unsealed. Unless
personnel familiar with the selection of samples for the specific tests can be on location during the
entire coring operation, it is recommended that all of the core be sealed on-site and shipped to the
Amoco Research Department or outside laboratory for analysis. The core facility handling the
samples should be advised that the core is to be shipped straight to the Research Department or
laboratory with no whole-core or plug analysis to be performed. Routine core analysis can be per-
formed after samples have been collected for mechanical properties tests.
As discussed in Chap. 4, core is analyzed by triaxial stress-strain tests to yield modulus of elasticity
(E). The test is performed by applying a hydraulic pressure to the core plug, then loading it axially
and measuring the displacement or strain (). In determining modulus for fracturing calculations,
the applied hydraulic pressure is normally set equal to the mean effective stress () acting on the
reservoir rock, i.e., the confining stress. An additional stress is then applied which is representative
of the net pressure above confining pressure required to open a fracture. E is then determined from
the resultant stress strain curve as E = /. Fig. 10.1 shows stress strain curves for a sandstone
under several confining stresses to illustrate the sensitivity of E to confining stress. Care must be
taken to estimate the confining stress correctly.
3,000
24,000
18,000
Stress ()
psi
0
12,000
6,000
0
0.00 0.20 0.40 0.60 0.80 1.00
E-02
STRAIN () - Percent
Poisson's ratio () is also determined in the laboratory in a triaxial stress test. is the ratio of lateral
expansion to longitudinal contraction for a rock under a uniaxial stress condition. The ratio of the
measured lateral strain to the axial strain is . Fig. 10.2 shows an example of strain data and the
calculation of .
Cores are also used to perform static fluid loss tests to determine a fluid loss coefficient. An expla-
nation of the testing procedure and interpretation and use of the results is covered in the section on
fluid loss.
30,000
24,000
LATERAL
18,000
STRESS (s ) AXIAL
psi
12,000
6,000
0
-1.00 -0.80 -0.60 -0.40 -0.20 -0.00 0.20 0.40 0.60 0.80 1.00
E-02
Because alternative interpretations exist for breakouts, it should be emphasized that care must be
taken in utilizing this type of data to determine fracture azimuth. Although it may not be a good
technique as a primary indicator of azimuth, borehole ellipticity could serve as a powerful tool for
extrapolating data where more comprehensive azimuth measurements have been made.
for determining fracture height growth parameters. These logs measure shear and compressional
sonic velocities, which may be used to calculate dynamic elastic rock properties, and theoretical
closure stress in a given horizon.7 The stresses thus calculated should be calibrated to actual in-situ
stresses by measuring the in-situ closure stress in 3-4 zones in the wellbore, and shifting the cal-
culated stresses to match in-situ stresses.
Both Amoco and Schlumberger have developed a Digital Sonic Log, both of which have been used
successfully in this technique. This log is run routinely by both Amoco and Schlumberger.
Schlumberger charges only slightly more for their Long Spaced Sonic Log than for the standard
Borehole Compensated Sonic Log. The Long Spaced Sonic Log yields as good or better porosity
measurements as the Borehole Compensated Sonic, and yields information regarding stress pro-
files as described above, along with a qualitative indication of natural fractures.
The Borehole Televiewer (BHTV) is a sonic type tool, introduced by Zemanek et al.,9 which in
principal should be an excellent fracture identification tool. However, the tool has not always per-
formed up to its potential. The tool consists of a crystal which emits high frequency sonic pulses,
then receives and records the reflection of these pulses from the borehole wall - with the lack of
any reflection possibly indicating the existence of a fracture. One problem in using this tool is that
borehole ellipticity and/or wellbore deviation creates blind areas due to decentralization of the
tool.10 Also, at this time, fracture width cannot be defined with this logging method.
Temperature Logs
Base Temperature Logs: A base temperature log is run to determine geothermal gradient and
static bottomhole temperature. To obtain a valid static temperature survey, the well should have
been shut-in for at least one week prior to logging. Temperature disturbances caused by circulating
the well during clean-out operations, etc., require approximately 3-5 days to dissipate, depending
upon individual well conditions.
Preperforation Cold Water Circulation Temperature Surveys: This technique is used to iden-
tify zones in the wellbore which are apt to exhibit temperature anomalies on postfracturing tem-
perature surveys due to thermal conductivity and/or wellbore effects, such as shown in Fig. 10.4
and Fig. 10.5. These anomalies often are confusing and misleading and often complicate temper-
ature log interpretation for fracture height determination.
9600
FRACTURE TOP
9200 2710
FLUID MOVEMENT
EFFECTS
9800
9400 2670
10000
FRACTURE TOP PROFILES
SEPARATE
PERFS
10200
PERFS 9600 2630
10400
175 200 225 250 180 200 220 240 260 F
TEMPERATURE ( F) TEMPERATURE
82 93 103 116 126 C
Many anomalies are usually present on postfracturing temperature surveys but may not all be in-
dicative of the presence of a fracture. This technique provides a method to subtract out the non-
fracture related anomalies to improve the accuracy of postfrac temperature log interpretation. The
procedure for obtaining these surveys is as follows:
TEMPERATURE F
75 80 85 90 95 100 105 110 115
4300
INJECTION TIME
4400
4600 48 HR
SI
11 DIA
4700 HOLE
4800
INJECTION ZONE
4900
TEMPERATURE F
75 80 85 90 95 100 105 110 115
4300
4400
HOURS SHUT-IN
3 12 48
4500 INJECTION
CURVE
4600
CEMENT
14 DIA
HOLE
4700
4800
INJECTION ZONE
4900
1. Run static temperature log over interval to be fractured [approximately 1,000 ft above pay to
Plug Back Total Depth (PBTD)] at 20-30 ft/min.
2. Run tubing open-ended to 20-25 ft above PBTD.
3. Circulate water down tubing and up the annulus at maximum possible rate within pressure lim-
itations for at least 3-4 hours. Friction reducer may be added to the water to reduce pumping
pressure. The water may be recirculated if a significant temperature differential exists between
reservoir temperature and the outlet temperature of the water at the surface. Cold water should
be added to the inlet stream when the outlet temperature rises by 25% of the initial reservoir:
inlet temperature differential.
4. Trip in with temperature tool to 1,000 ft above the pay interval.
Perforating with many large holes will reduce perforation friction pressure and excessive shear on
the frac fluids. Perforating out of phase decreases the likelihood of the perforation being oriented
in a line at a high angle to the fracture azimuth, as shown in Fig. 10.6, and therefore reduces friction
pressure and shear between the wellbore and fracture. This method of perforating also affords a
better flow path to the wellbore during bottomhole pressure buildup and may reduce the need to
acidize the zone to attain an adequate flow rate for obtaining a buildup. If possible, do not stimulate
or breakdown the perforations prior to flow testing.
Narrow
Gap Vertical
Fracture
min
Cement
max
Fig. 10.6 - The Effect of Zero Degree Phasing Perforations on a Fracture Treatment.
Better results are obtained in the minifrac and fracture treatment analysis if only one pay zone is
perforated. The analysis of net pressure is complicated by fracturing multiple zones at the same
time, particularly if the zones are separated by sufficient thicknesses of confining beds to allow the
propagation of two or more fractures at the same time.
When closure stress tests are performed in shales to measure the closure stress of bounding layers,
experience has indicated that high density perforating with large charges could compress the shale
around the perforation tunnel. This added stress to the rock has made breakdown impossible in
some cases. Little is known at this time about the best method for perforating shales for stress test-
ing and further field research testing is required in this area.
A bottomhole pressure buildup test should be run to determine formation flow capacity. The for-
mation permeability is used to determine optimum fracture length, to set limits on the fluid loss
coefficient to be used for designing the fracture stimulation, for improving the accuracy of post-
fracturing performance prediction, and for analyzing postfrac buildup tests for fracture length and
conductivity.
Three techniques are recommended for measuring BHTP.12 Fig. 10.7 shows wellbore schematics
for executing these procedures. The first requires running tubing open-ended (without a packer)
and pumping down either the tubing or annulus. The other side is then static, and pressures at the
surface on the static side are a direct reflection of BHTP, corrected for hydrostatic pressure. The
second technique involves the use of a surface readout pressure gauge mounted in a side pocket
mandrel, strapping the electric line to the outside of the tubing. The third technique employs a
downhole recording pressure bomb placed into a simple mandrel below a packer. With this tech-
nique, actual BHTP are recorded, but the data cannot be accessed until after the treatment. For the
two procedures where BHTP is measured in real-time, the stimulation service companies can pro-
vide on-site computer vans which facilitate quick manipulation of the prefrac test and/or main
treatment data for plotting to make on-site judgmental decisions.
Pt Pa
Qt Qa Q
Q
WIRELINE
PACKER
SIDE POCKET MANDRIL
MANDRIL PORT
Qt - 0 PERFORATED
PRESSURE SUB
SENSOR (BLAST JOINT)
Pt - BHP-Pn
or PRESSURE
Qa - 0 BOMB
PACKER
SEATING
Pa- BHP-Pn NIPPLE
NO-GO NIPPLE
To ensure the mandrel assembly does not cause increased fluid shear during the treatment, (1) the
perforated subs should be prepared such that the perforation area is adequate to yield near zero per-
foration friction, and (2) the outside diameter of the assembly should not exceed the outer diameter
of the tubing to provide adequate annular space between the assembly and casing. Probably the
easiest and least expensive way to prepare the perforated sub is to have the holes drilled in a ma-
chine shop. This ensures all holes are open, large and properly spaced.
After the fracture treatment, the pressure bomb may be retrieved with a slick line by latching onto
a fishing neck on top of the bomb or by pulling the tubing string.
Closure stress is determined by pumping a volume of fluid at a rate sufficient to create a fracture,
and then allowing the fracture to close either by shutting-in the well and allowing pressure to de-
cline to below closure pressure, or by flowing the well back until pressure is reduced to below clo-
sure pressure.12 In either case, closure pressure is identified by a change in the pressure decline
characteristics as the fracture closes. Either test should be preceded by a step-rate test to determine
extension pressure, which should be within about 100 psi of closure pressure. The step-rate test
will also assure that a fracture exists before the closure test is attempted. Fig. 10.8 shows a typical
step-rate test plot. The time step at each rate should be constant, e.g., 2 minute intervals.
BOTTOMHOLE PRESS.
FRACTURE
INJECTION RATE
EXTENION PRESS
AT STEP END
TIME INJECTION RATE
To create the fracture requires that a sufficient volume of fluid be pumped at a sufficient rate. In
practically all cases, pumping for 10-20 minutes at 10 bpm has proven to be adequate; but, depend-
ing on the results of the step-rate test, these guidelines may be altered. In low permeability, low
leakoff formations 50 bbls at 5 bpm may be sufficient.
Any fluid, which is compatible with the formation rock and fluids, may be used for the tests. Gen-
erally whatever base fluid is to be used for the fracture stimulation is used for the closure stress
test: produced formation water, 2% KCl water, etc.
Determination of closure pressure from shut-in pressure declines is operationally very simple. The
well is left shut-in until pressure declines to a point at which closure pressure can be identified as
shown in Fig. 10.9. This method of determining closure pressure is most appropriate for high per-
meability formations which close quickly. In this type formation, closure would occur almost in-
stantly during a flowback test making identification of closure pressure difficult. The data, during
a shut-in decline test, should be plotted real-time, if possible, to determine the length of shut-in
time. The decline data can also be plotted on a Horner type plot, Fig. 10.9, to identify radial flow
and, thus, ensure the fracture has closed.13 Also, this plot can be used to estimate the near wellbore
reservoir pressure, p*. To ascertain the length of shut-in time may require a trial test, followed
by subsequent tests. The number of tests performed will depend on the agreement of closure pres-
sures picked. If good agreement is evident, only 2-3 tests may be required. It has been noted that
in liquid filled reservoirs closure pressure increases with each subsequent test due to an increase
in pore pressure. When this occurs, the earlier test results are probably most representative of for-
mation closure and should be used to calculate net pressure during the minifrac and fracture treat-
ment.
BOTTOMHOLE PRESS
BOTTOMHOLE PRESS
SHUT-IN DECLINE
START
CLOSURE RADIAL
PRESSURE P*
POSSIBILITIES
Closure stress determination from flowback pressures is only slightly more complicated than a
shut-in decline test and is more conducive for low to moderate permeability formations, which
would require extensive monitoring periods during a shut-in decline test. The flowback rate is de-
termined by the fluid loss characteristics of the formation and the surface pressure; the purpose of
the flowback being to flow back at a rate on the order of the rate at which fluid is being lost to the
formation. For this flow back rate, a characteristic reverse curvature occurs in the pressure decline
at closure pressure as shown on Curve b in Fig. 10.11. A suggested initial flowback rate is 1-2
bpm. The proper flowback rate is usually determined by trial and error on the first tests, flowing
back at different rates until the correct flow back rate is found and a good test is obtained.
BOTTOMHOLE PRESS
TIME
To control the flowback rate, a manifold similar to that shown in Fig. 10.12 is required. An adjust-
able choke, gate valve, or automatic constant flow regulator (e.g., manufactured by Oilmaster - se-
rial no. 280-390) should be installed downstream of a 1-inch and/or 2-inch flowmeter(s). When
selecting a flowmeter for measuring the flowback rate, one must keep in mind the rate range of the
meter used. Service companies tend to recommend, and will usually supply, a 2-inch turbine meter.
Experience has shown that it is difficult to impossible to measure flowback rates of 1-2 bpm with
meters of this size. The best choice seems to be a 1-1.5 inch turbine meter with digital readout in
bpm. Digital readout boxes, showing flowback rate, should be positioned near the valve or choke
for ease, accuracy, and quickness of adjustment. To minimize the adjustment of this valve or choke
from test to test, a full opening gate valve or Lo-Torque valve should also be placed between the
wellhead and flowmeter(s). This valve can be used to open and close the flowback system without
having to fully close the valve downstream of the flowmeter(s).
DIGITAL READOUT
2
FLOWMETER
FLOWBACK
LINE
DISPOSAL
WELLHEAD
PIT
DIGITAL READOUT
The following procedure is recommended for closure stress tests in low to moderate permeability
formations:
1. Since real-time data is necessary, either open-ended tubing or a downhole pressure recorder
with a surface readout is required to obtain BHP. In some cases, surface pressures may be suf-
ficient. Pressures and rates should be monitored and recorded continuously throughout the
tests.
2. Perform step-rate test to determine extension pressure and the minimum injection rate re-
quired to fracture the formation. Utilize the step-rate test as a pump-in/flowback test, flowing
the well back at a constant rate of 2 bpm. Note: In latter portion of pump-in, the injection rate
should be increased by an equivalent rate to the planned flowback rate. At the same time, the
flowback manifold should be opened and the flowback rate set prior to shutting down injection.
The shutdown should be slow, i.e., in 10-15 seconds be pumping at 0.5 bpm, then shutdown
completely. This will prevent fluid hammer effects in the wellbore, which could distort test
results.
3. Flowback at a constant rate until the BHP approaches reservoir pressure. To keep the flowback
rate constant will require constant adjustment to the valve as the surface pressure decreases.
4. Based on the required injection rate, perform pump-in/flowback test by injecting fluid for a
minimum of 10 minutes, e.g., if rate = 5 bpm, pump 50 bbls. Flowback using procedure in
Steps 2 and 3 above. Constant flowback rate may have to be increased or decreased from the
2 bpm in Step 2 depending on the results from Step 3. Fig. 10.11 shows examples of too high
and too low flowback rates.
5. Repeat Step 4 until a repeatable closure pressure is established.
6. Perform pump-in/shut-in decline using the same volume and rate determined above. Record
pressure decline until pressure falls well below the closure pressure determined above. Do not
flowback during this step.
Note: In formations with relatively high permeability (>0.1 md), acid ISIPs may closely approxi-
mate closure stress, if the acid jobs are small, pump rates are low (yet high enough to create a frac-
ture), and nitrogen or CO2 are not mixed with the acid.14 This will yield a first estimate of closure
stress in most cases and will set an upper limit for closure stress.
Minifracs
Minifracs or Calibration Treatments are pumped to obtain information on the mechanics of frac-
ture propagation during the small treatment (net fracture pressures, height growth or confinement,
etc.), and to collect data for determination of fracture geometry, time for the fracture to close, and
fluid loss coefficient.15 This test consists of pumping a relatively small volume of fluid, i.e.,
10-20% of the main fracture treatment depending on its size, using the main treatment fluid system
and pumping at the expected main treatment injection rate. During and after the minifrac, BHTP
and the shut-in pressure decline is monitored and recorded.
1. Batch mix the required amount of fracturing fluid. Batch mixing is required for gel consistency
and to minimize friction pressure variations throughout the test.
2. One of the BHP measurement techniques described previously on page 10-11 should be used
for measuring pumping and shut-in decline pressures. Tubing pressure and casing pressure
should be recorded by the fracturing service company. In addition, the wellhead should be
rigged with a lubricator as described under Temperature Profiles.
3. Pump minifrac at expected main treatment rate (constant rate throughout test). Record all pres-
sures and rates continuously throughout the job.
4. Shut down and record pressure decline for as long as required until the pressure bleeds off to
well below the closure stress value previously determined by the closure stress test.
Fracture geometry can be evaluated from a Nolte-Smith Log-Log plot of net fracturing pressure
(BHTP - closure pressure) vs. pump time as discussed previously in Chap 8. Design parameters,
including the fluid loss coefficient, can be determined using the pressure decline analysis which is
also presented in the Fracturing Pressure Analysis Section.
Temperature decay profile surveys should be run as soon as possible after a minifrac without in-
terfering with the collection of pressure decline data. If bottomhole pressure is measured via a stat-
ic tubing string, the lubricator can be rigged up on the wellhead ahead of time, and the closure
stress tests and minifrac can be pumped through a wing valve or T-connection below the lubricator.
The temperature tool is run in the lubricator before the job and isolated from the wellbore with a
valve while pumping.
If a wireline pressure gauge is run during the prefrac tests, the pressure decline data collection
should be completed and the pressure gauge removed prior to installing and running the tempera-
ture tool. If bottomhole pressure is measured via a static open-ended tubing string, the temperature
tool should not be run until after the pressure decline since running the tool will distort the pressure
data.
A minimum of three logging runs should be made at intervals of 45 minutes from the start of each
run. No backflow from the well should be allowed prior to or during temperature profiling. The
logs should be run from several hundred ft above the pay interval to several hundred ft below the
fracture bottom or plug back Total Depth (TD), logging down at a speed of about 20 ft/minute. It
is the Amoco engineer's responsibility to see that the logging company records the necessary data
on the log heading, including fluid type and volume pumped, total pump time, times minifrac start-
ed and ended, and fluid surface temperature.
This same procedure also applies to temperature decay profile surveys run after the main fracture
treatment.
As discussed earlier on page 10-8, a static base temperature log and cold water circulation survey
may be run to determine the temperature gradient and identify anomalies caused by formation
changes, the wellbore, and the completion. Fig. 10.13 shows the conductivity effects from differ-
ent formations on both pre and postfrac logs.11 Fig. 10.5, shown previously, shows how a washout
behind casing will create a cool anomaly which may be interpreted as a fractured zone. On the oth-
er hand, a washout completely filled with cement will insulate the wellbore and create a hot nose
on the log. Also, a change in tubular diameter, such as the bottom of tubing or casing can cause an
offset in the log. All of the above anomalies can be detected with the base temperature log and
subtracted out of the postfrac log interpretation.
8800 2690
TOP
9000
THERMAL PRE FRAC 2750 POST FRAC
CONDUCTIVITY PROFILE TEMP LOG
EFFECTS
9200 STATIC
LOG 2810 TOP?
12200 3720
GR
9600 2930
9800 2990
12300 3750
TOP?
10000 3050
FRACTURE TOP PROFILES
SEPARATE
10200 3110
PERFS
PERFS
10400 3170
175 200 225 250 275
F
TEMPERATURE 190 200 210 F
80 93 108 121 135 C 88
TEMPERATURE
93 98 C
Fig. 10.13 - Pre and Postfrac Temperature Fig. 10.14 - Temperature Log Showing
Logs Showing Thermal Conductivity Warm Anomaly Above Treatment Zone.
Effects.
Fig. 10.14 shows a warm anomaly or hot nose above the fractured zone and the obvious prob-
lems associated with picking the fracture top.11 It has been theorized that this is caused by fluid
movement after shut-in and that the hot nose is part of the fracture height.
Temperature crossovers are often seen below the perforated interval from one logging run to an-
other. Below the perforations, the wellbore is filled with stagnant, hot fluid; and any downward
fracture growth will place cooler fluid outside the casing than inside. Thus, heat flow will be in the
opposite direction from that across and above the fractured zone and the wellbore may cool down
with time. This often results in a temperature crossover, as seen in Fig. 10.15, which can be a
good indicator of the bottom of the created fracture.
Since temperature logs are shallow investigative tools, they only see the fracture at or near the
wellbore. If the created fracture is not vertical, but dipping at an angle somewhere between true
vertical and true horizontal, temperature logs will not provide a meaningful interpretation of the
fractured interval as illustrated in Fig. 10.16. This same problem occurs when the fracture is ver-
tical and the wellbore is deviated. Thus, under these circumstances temperature logs are, at best,
poor indicators of fracture growth.
GR
TEMP
LOG
#1 TOP
#4
4 1
Vertical Fracture
Straight Wellbore Fracture Communication
With Wellbore Dipping Fracture
Or Deviated Wellbore
In a well which goes on vacuum after a stimulation, the falling fluid level will continually carry
warm fluid down into the fractured zone, obscuring the temperature anomaly. This is possible in
injection well stimulations and on pumping wells with low reservoir pressure. In such cases, the
fluid level should be allowed to stabilize prior to running the logs.
Recommended
Tracer Half-Life Concentration
Noting the variation in half-lives, a postfrac gamma ray log should be run early in the half-life of
the tracer used. Also, for the most definitive results with regard to fracture height, the tagged ma-
terial should be added throughout the stimulation.
One advantage of gamma-ray over temperature logs is that they do not need to be run immediately
after a stimulation, allowing wellbore fill below perforations to be removed before logging. How-
ever, the other restrictions on the temperature logs apply equally to radioactivity logs - that is they
are shallow investigative tools (shallower, even, than temperature logs), the response is propor-
tional to fracture width, and the wellbore and completion can effect the resultant log profile. Thus
while the two logs are often used in combination, the potential exists for them to confirm one an-
other and still not yield reliable results.
One disadvantage of radioactivity logs is their inability to distinguish between a fracture and a
small channel behind casing. The temperature response due to a small amount of flow in a channel
or annular space behind casing may not alter the radial flow heat conduction around unfractured
portions of the wellbore and does not affect the temperature logs. However, any material deposited
in a channel is indistinguishable from tagged material in a fracture.
Fig. 10.17a shows a good example of pre and postfrac gamma ray logs.11 The radioactive material
indicates the top and bottom of the fracture and correlates well with the postfrac temperature log.
A second example, shown in Fig. 10.17b, utilized radioactive material in only the later pact of the
fracture treatment, thus radioactive material showed up only through a portion of the fracture.11 In
this same figure, radioactive material shows up across the hot nose indicating this to be, in fact,
part of the fracture height.
(a) (b)
9100 POST
FRAC 2780
TEMP
POST SP PROFILE
9300 WARM
PROFILE 2810
9200
BASE
9400 GR
POST FRAC
GAMMA RAY
9300 2840
FRAC ZONE
9500
9600
9500 2900
cepted techniques are tiltmeters and geophones, with increasing acceptance of oriented core anal-
ysis generated through recent consistent results from strain relaxation measurements.
Tiltmeters
Tiltmeters are highly sophisticated, extremely accurate bi-axial instruments which utilize bubble
sensors to measure the change in angle of a surface. These devices were originally developed to
aim intercontinental missiles, and were later employed by the U.S. Geological Survey for use in
the study of earth movements associated with earthquakes and volcanic activity. The use of tiltme-
ters to monitor hydraulic fractures, at depths up to 10,000 ft, is based on the assumption that the
earth will respond in a more or less elastic manner to deformations caused by opening a hydrau-
lic fracture. In that case, the surface of the earth will deform in a predictable manner and measure-
ments of this deformation can be interpreted to obtain data with respect to fracture geometry.17,18,19
Fig. 10.18 illustrates surface deformations associated with fractures of several orientations.
A typical tiltmeter array consists of 12-16 instruments evenly spaced radially around the well, at a
distance of about 0.4 times the depth of the zone to be fractured. Each instrument is installed in a
shallow cased hole, usually 10 to 20 ft deep, and packed into position using sand to insulate the
device from surface weather and noise effects.
The tiltmeter instruments are capable of measuring changes in tilt of a surface with accuracy on
the order of 1 x 10-7 radians. Due to the sensitivity of the measurements, changes in the level of the
earth's crust due to solid earth tides cause changes in the surface angle which are orders of magni-
DIP = 90
DIP = 60
DIP = 30
DIP = 0
tude greater than the fracture treatment. Fortunately, the period of the fracture event is much short-
er than the tidal noise and can be separated by post-analysis using frequency domain filtering
and/or tidal filtering. The residual from this filtering is then used to measure the tilt signal related
to hydraulic fracturing. The signals from both channels of a tiltmeter are combined to form a tilt
vector which embodies direction and magnitude of the tilt measured at that site. Fig. 10.19 shows
the recorded response for one channel from a single site.
To analyze the data, observed tilts are compared with theoretical values for many possible combi-
nations of fracture azimuth and dip; and thus, the azimuth and dip are determined which produce
the least error. An example shown in Fig. 10.20 shows theoretical tilt responses for vertical and
horizontal fractures and Fig. 10.21 shows a least error fit for observed vs. theoretical data.
Just as the pattern, or direction of the tilt vectors is related primarily to the fracture azimuth and
dip, the magnitude of the vectors is principally a function of fracture volume. Recent work has
been performed which combines fracturing pressure analysis with tilt vector magnitude to place
bounds on created fracture dimensions for wells shallower than 4000 ft, as seen in Fig. 10.22.
1.7537
1.7404
1.7270
1.7137
Tilt Signal
1.7004
1.6871
1.6738
1.6604
1.6471
1.6338
317.51 317.53 317.55 317.56 317.58 317.59 317.61 317.62 317.64
11:12:11:58:05 TO 11:12:15:40:31
Because extensive site preparation is required to install the tiltmeter array and a site aging period
is required, scheduling should begin far in advance of the hydraulic fracture treatment. Site prep-
aration should begin a minimum of three weeks prior to the treatment. District personnel involved
in this testing should work closely with the Research Department in setting up and executing these
tests.
Borehole Geophones
Borehole geophones measure the sonic energy, or noise, produced while a formation is being frac-
tured.21-,25 A set of three geophones is typically installed in the wellbore on a single conductor wire-
line prior to the well being fractured. Since a wireline is in the hole while fracturing, the treatment
is usually a small gelled-water minifrac without proppant. One geophone is vertical and the other
two are horizontal. The orientation of the geophone tool is determined using surface shots set off
in strategically located sites in an array with a radius equal to the depth of the tool. A minimum of
VERTICAL FRACTURE
(mirror symmetry
relative to the strike
of the fracture)
HORIZONTAL FRACTURE
(radial symmetry relative
to the wellbore)
Theoretical
Observed
R = 330 ft
R = 1000 ft
Error in Fil (%)
R = 570 ft
Fracture Volume
four shots are detonated, one at a time, using dynamite. The sites are 20 ft deep and located at equal
intervals of 45. The recorded arrival time of the shock wave indicates the direction of the source
with respect to the geophones.
Fracture azimuth is determined by analyzing the arrival times of sonic waves being propagated
through the formation as the rock cracks and the fracture extends in length. The variation in arrival
times between the three geophones is analyzed to determine the direction of the source of the sonic
waves (the tip of the fracture) from the wellbore. Fig. 10.23 shows an example of the type of results
obtained.
X
AMPLITUDE
SIGNAL
Y W E
Time
Typical microseismic event recorded on three
orthogonally mounted seismic detectors. The
time marks are 0.017 sec. apart.
S
Polarization of single-phase events recorded with
downhole three-axis geophone package.
The indirect oriented coring process uses a shoe on the bottom of the core barrel with three knives
to cut grooves in the core. One of these is the reference groove at a known orientation to an azimuth
lug attached to the inner core barrel. An orientation tool is mounted above the core barrel such that
the orientation lug is visible when the tool photographs the compass. The correction between the
reference knife and the orientation lug can be pre-set in the shop, but a preferred technique is to
hoist the barrel in the derrick and use an optical aligning device to determine their relative orien-
tation; this is then recorded for future calculations.27 Since this tying of orientation to depth is in-
direct, the biggest sources of error come from incomplete core recovery, breaks in core, or a
spiraling reference groove.
The technique of direct on-site measurement of strain relaxation from cores to determine fracture
azimuth is based on laboratory observations that the stress-strain behavior of rocks is not purely
elastic, but is a function of loading rate and time.28 In such a case, strains stored in the rock by the
in-situ stresses will not be released instantly when the core is cut, but will relax over many hours.
If the core can be recovered and instrumented during this time, the orientation of stresses can be
determined by measuring relaxation in different directions.
The strain relaxation process involves selecting several core samples as soon as possible after the
core reaches the surface. The samples should be selected from intact core sections to ensure good
orientation data, then removed to a reasonably constant temperature environment, sealed to pre-
vent moisture evaporation, and then tested by attaching the deformation gauges to the sample to
record strain relaxation (and temperature) data from 12 to 24 hours. These measurements are then
used to calculate the orientation of the in-situ stresses.29 Fig. 10.24 shows typical data taken from
strain relaxation measurements on a shale sample.30
Elastic Strain
STRAIN
C'
Time-Dependent Strain
Det1 - t2
C
to t1 t2
TIME
Core Recovered and Instrumented at C'
The strain relaxation technique has proven accurate in several tests where azimuth was also mea-
sured with other procedures.21,26,31,32 These include tests in a volcanic tuff in Nevada; a low perme-
ability Mesa Verde Sandstone; a low permeability gas sand in the Cotton Valley Formation; and a
high porosity, high permeability sandstone in Oklahoma.
Borehole Geometry
The geometry of the borehole (ellipticity) may be affected by the stresses in the earth in the near
wellbore region. The fracture azimuth is also affected by these stresses. 1-5 Therefore, a simple cor-
relation might be made between borehole ellipticity and azimuth if conclusive supporting data can
be obtained. As discussed earlier on page 10-5, borehole ellipticity measurements in two different
areas indicate that fracture azimuth is either parallel to or perpendicular to the long axis of the bore-
hole. By combining the results of the azimuth measurements discussed above with borehole geom-
etry, a correlation might be made for a given field which would greatly simplify fracture azimuth
determination. Borehole geometry must be obtained in open-hole, and can be measured with a
Borehole Geometry Log as previously discussed on page 10-5, or from the oriented caliper incor-
porated into the Dipmeter Log. The Dipmeter Log yields information useful in geologic interpre-
tation, whereas the Borehole Geometry Log describes only the orientation and dimensions of the
borehole.
10.2Introduction To TerraFrac
TerraFrac is a three dimensional fracturing simulator that is probably the most advanced commer-
cially available hydraulic fracturing simulator presently available. It has been in use by TRC since
1983 to address nonstandard fracture design problems. Fracturing design problems in wells in the
Valhall Field in the North Sea, as well as exploration wells all over the world, have been success-
fully addressed using the TerraFrac Simulator.
TerraFrac is installed on the IBM mainframe computer at the Research Center; however it is not
yet released for general use because of the complexity and time-consuming requirements of data
input, code execution, and the requirement of output analysis. The code is still undergoing devel-
opment and possesses very advanced capabilities such as thermal and poroelastic effects. It can
also be applied in fracture designs where the fracture may migrate considerably up or down from
the point of initiation, to study the effects of perforation placement on resulting fracture geometry.
TerraFrac solves the fracturing problem, in a general sense, i.e., it determines the fracture geome-
try as part of the solution process. A three-dimensional simulator is a simulator that can predict
fracture shape (width and height at any point along the fractures length). However, this is a nu-
merically demanding problem which is strongly nonlinear because of the coupling required be-
tween the fluid pressure distribution in the fracture with the stiffness of the opening fracture. The
solution of the problem may lead to fracture shapes that are complex, like the one at the top of
Fig. 10.25, which are relatively realistic even though they employ certain simplifying assump-
tions, e.g., planar fractures. The schematics in the lower part of Fig. 10.25 represent the simplest
models which are still used throughout the industry for simulating fracture treatment design. These
are idealistic versions of what may be happening downhole.
Actual?
R
wf
Penny
Area of Largest
xf Flow Resistance
Approximately Ellipitical
Share of Fracture
hf
xf
wf
wf
Perkins & Kern Geertsma & deKlerk
bore. It determines fracture geometry from the solution of a complex nonlinear interaction problem
of:
A short account of how the model works is as follows: TerraFrac determines the shape of the frac-
ture in an iterative way. It starts from an assumed fracture shape which is small relative to the frac-
ture dimensions after the treatment has ended. An initial pressure distribution is also assumed. It
is recommended to start the simulation with a small penny shaped fracture at the center of the per-
forations. If the perforation interval is large with varying closure stresses, one would probably
choose to initiate the fracture at a point where the closure stress is minimized. The fluid pressure
is assumed (handled internally) initially to be constant. The fracture width is dependent on fluid
pressure distribution and fracture shape, and can be calculated from an elastic 3-D rock deforma-
tion solution. TerraFrac has the capability to calculate fracture width for a general shaped fracture
with arbitrary fluid pressure distribution. The widths from this solution stage are used to solve the
fluid flow problem in the plane of the fracture. The fracturing fluid is assumed to behave like a
power law fluid in laminar flow between parallel plates. The widths determined from the elastic
solution are used as the distance between the parallel plates. The fluid pressure distribution can be
calculated by satisfying the momentum and continuity equations with appropriate conditions at the
boundaries. Then the fracture widths can be derived using this pressure distribution from the elas-
tic solution. In this way, an iteration can be performed to derive the pressures and widths which
are mutually consistent.
The tendency of the fracture to propagate can be quantified using the closure stress profile, elastic
constants, toughness, the fluid pressure distribution, and the pre-existing fracture shape. A Critical
Fracture Width is calculated internally (Fracture Propagation Criterion), and, if the width of the
fracture at some given distance behind the front exceeds the critical fracture width, the fracture
propagates. The distance of propagation is calculated from a combination of mass balance enforce-
ment and the amount by which the widths near the front exceed the critical fracture width.
During the propagation, leakoff is assumed to occur according to Carter's model. The enforcement
of the continuity equation dominates the propagation and is given priority. In this sense, the frac-
ture Propagation Criterion is satisfied within broad tolerances, while continuity near the fracture
front is satisfied more accurately.
Input To Terrafrac
The downhole schematic of Fracturing Configuration of Fig. 10.26 gives a pictorial definition of
the input to TerraFrac. For each formation layer, it is required to define reservoir (porosity, perme-
ability, thermal conductivity), and elastic (modulus, Poissons ratio, toughness) properties.
Input relative to model discretization, convergence limits, input, output, and plotting are also re-
quired. The model uses a combination of finite element and boundary element methods to solve
the coupled elastic-fluid flow problems. The fractures boundary is subdivided into quadrilaterals
which are further subdivided into four triangles. All calculations are performed on the triangles in
terms of the pressures and widths at the nodes. A typical plot of the mesh is shown on Fig. 10.28.
A detailed explanation of the input and the numerical techniques employed are beyond the scope
of this manual.
Note that the original TerraFrac formulation required the elastic properties to be uniform in all lay-
ers; however, an approximate way to account for the first order effects of modulus changes from
layer to layer has been recently implemented by TerraTek and has been installed on our IBM main-
frame computer.
The confined height example of Fig. 10.27 was devised to demonstrate the influence of leakoff and
closure stress gradient during the initial stages of fracture evolution. Furthermore, it acquaints the
reader with typical plots of the TerraFrac results produced by the plotting postprocessor developed
by the Frac Group. The mesh used for this analysis is shown on Fig. 10.28.
Depth-Feet
-7900
LAYER 1 C = 0.0
LAYER 4
C = 0.0005 ft/min
-8100
LAYER 5 C = 0.0
-8200
7200 7300 7400 7500 7600
CLOSURE PRESSURE - PSI
CLOSURE STRESS
FORMATION PROPERTIES
1-2
E = 1.26x106 psi
= 0.35 2-3
FLUID VISCOSITY 3-4
= 90 cp 4-5
PUMPING RATE PERFORATIONS
Q = 16 bbl/min
The fracture shape evolution gives an appreciation of the delicate balance of the in-situ parameters
and their influence on fracture shape. Note that steep closure stress gradients push the fracture
growth upwards, while low leakoff zones encourage fracture growth in them. This is clearly dem-
onstrated in Fig. 10.29 which shows fracture evolution until the fracture reaches the lower confin-
ing layer (layer 5). The fracture was initiated as a penny shaped fracture of 20 ft radius at 8025 ft
depth. The fracture initially propagates as a penny in layer 3. Later, the small leakoff of layer 4 is
attracting the fracture more than the closure stress gradient of 0.848 psi/ft of layer 2 and the frac-
ture grows downwards until it reaches layer 5. The remainder of the fracture evolution is shown in
Fig. 10.30. The fracture, being confined below, grows upwards until it reaches layer 1. From then
on, we have confined fracture growth and the TerraFrac analysis does not offer anything additional
to a PKN program analysis.
Y (FEET)
X (FEET)
X (FEET)
Y (FEET)
X (FEET)
Fig. 10.31 shows the plot of the step number vs. injected volume. From this plot we see that a great
amount of steps (and computing time) is spent during the initial propagation stages. During the first
40 steps only 23 barrels of treatment volume were injected. Consequently, a small amount of in-
jected volume propagates the fracture rapidly to a confined mode of fracture extension; therefore,
a PKN analysis is essentially applicable for the entire fracturing propagation process.
Fig. 10.32, Fig. 10.33, and Fig. 10.34 show plots of the evolution of leakoff volume, fracture vol-
ume, fracture width, and fracture dimensions. Fig. 10.35 shows the variation of fluid pressure dur-
ing the fracture treatment. The kinks in the pressure are due to numerical reasons and should be
smoothed out (see next paragraph). The maximum pressure reflects the slightly increasing pressure
trend of confined fracture extension. The pressure at the perforations (depths are plotted with ref-
erence to the center of perforations referred to as 0.0 ft) shows this increasing tendency to a lesser
degree. Note that hydrostatic head in the fracture forces the maximum pressure to occur below the
perforations.
Fig. 10.36 and Fig. 10.37 show the error distributions of the iteration scheme. Comparing these
figures with Fig. 10.35, we see that the pressure distribution is sensitive to these errors. This is ex-
pected due to the strong nonlinearity of the problem. Consequently, despite the stringent conver-
gence error of 1%, the TerraFrac user should be able to distinguish real behavior from spurious
numerical behavior of the solution. This is valid especially for pressures which are the most sen-
sitive.
80
60
STEP NUMBER
40
20
0
0 200 400 600 800 1000
TOTAL VOLUME INJECTED (bbl)
STEP NUMBER
Fig. 10.31 - Evolution of Leakoff Volume, Fracture Volume, Fracture Width, and Fracture
Dimensions.
600
BARRELS
400
200
0
0 200 400 600 800 1000
TOTAL VOLUME INJECTED (bbl) TOT. FRACTURE VOL (bbl)
TOT. LEAKOFF VOL (bbl)
Fig. 10.32 - Evolution of Leakoff Volume, Fracture Volume, Fracture Width, and Fracture
Dimensions.
0.25
0.20
INCHES
0.15
0.10
0.05
0.00
0 200 400 600 800 1000
TOTAL VOLUME INJECTED (bbl)
MAX FRAC WIDTH (in)
WIDTH (in) AT 0.0000C+00 ft
Fig. 10.33 - Evolution of Leakoff Volume, Fracture Volume, Fracture Width, and Fracture
Dimensions.
Fig. 10.38 shows the efficiency of the treatment. We see that the efficiency of the treatment drops
to approximately 20% while 80% of the volume injected leaks into the formation.
Two examples of unconfined fracture growth are briefly discussed in this section. They were taken
from a real case analysis of fracturing treatment for the Upper Hod formation of the 2/8A-17 well
in Valhall. These examples illustrate the capabilities offered by TerraFrac and the opportunity it
offers to enhance understanding of the fracturing process for complicated in-situ conditions.
Fig. 10.39 shows the two closure stress profiles considered; they were derived from our best esti-
mates of the in-situ conditions. Case A represents the base case; case B has a 200 psi lower closure
stress in the Tor relative to case A (due to reduced reservoir pressure after production) and a 50 psi
higher closure stress in the Dense zone to account for its higher confining capacity. The perfo-
rations are located directly below the dense zone. A constant 15 bbl/min pumping rate and a 90 cp
downhole viscosity fracturing fluid were assumed. The reservoir pressure was taken as 6275 psi.
Completion experience in Valhall has established that the Tor should not be directly perforated be-
cause it produces solids and plugs the well. The Upper Hod is perforated instead. Upper Hod treat-
ments have the dual purpose of stimulating the poorer Hod formation and communicating with the
rich Tor formation. Fracture height growth is not confined and fracture shapes may be complex
dependent on the in-situ conditions. It has been the practice to design such fracture treatments as
500
400
300
FEET
200
100
0
0 200 400 600 800 1000
TOTAL VOLUME INJECTED (bbl)
MAX FRAC LENGTH (ft)
MAX FRAC HEIGHT (ft)
MAX HEIGHT ABOVE CNTR (ft)
MAX DEPTH BELOW CNTR (ft)
Fig. 10.34 - Evolution of Leakoff Volume, Fracture Volume, Fracture Width, and Fracture
Dimensions.
300
250
PSI
200
150
100
50
0 200 400 600 800 1000
TOTAL VOLUME INJECTED (bbl)
MAX PRESSURE (psi)
PRES (psi) AT 0.0000F+00 ft
0.6
0.4
0.2
0 200 400 600 800 1000
TOTAL VOLUME INJECTED (bbl)
CONVERGENCE ERROR (%)
penny shaped fractures for lack of a better alternative. However, using TerraFrac we can deter-
mine fracture shape and study the effects of closure stress profile, actual closure stress gradient,
leakoff variation, and position of perforations. It is this capability that makes the TerraFrac simu-
lator so useful for Valhall field and other fields where no significant confining barriers exist.
Fig. 10.40 shows the fracture evolution for case A. The fracture was initiated (for both A and B
cases) as a small penny (of 10 ft radius) located at the center of the perforations, which is the origin
of the Y-axis. Note that in case A the fracture essentially remains approximately a penny, although
some confinement can be observed at the shale-Tor interface.
Fig. 10.41 shows the fracture evolution for case B. For this case the shape is drastically different.
It grows mainly in the Tor where closure stress is low. The lower part of the fracture simply con-
nects the perforations. This type of behavior can only be quantified by numerical simulation and
represents a delicate balance of the in-situ values of closure stress, closure gradients, and leakoff
as well as the location of the perforations and fluid rheology.
Fig. 10.42 compares the fracture width profiles along the wellbore for both A and B cases. In case
A, the maximum fracture width occurs close to the perforations (the origin of the Y-axis). In case
B, the fracture grows unsymmetrical with respect to the perforations and a pinching point devel-
ops. Width pinching near the perforations may cause a screen-out during the early stages of the
treatment.
10
6
PERCENT
-2
-4
Fig. 10.43 shows the fracture width history for both cases. The maximum fracture width and the
fracture width at the perforations (i.e., at 0.0 ft) are plotted vs. the total injected volume. In case A,
we see no significant difference between these two values, both of which increase with the volume
of the fracturing treatment. In case B, the max width occurs in Tor and increases with the volume
injected as expected. However, the width at the perforations initially increases (while the fracture
is still a penny) and subsequently decreases at about 200 bbl, to remain constant at approximately
0.10 inches for the remaining of the treatment. This pinching effect may be the reason for prema-
ture screen-out. For such a case, an increased pad volume does not diminish the danger of
screen-out. More viscous fluid and small proppant may be required to pump the fracturing treat-
ment successfully. Note that the width at perforations can actually decrease during pumping of the
treatment, especially when unconfined nonsymmetric fracture growth occurs. The width history
plot may by used to estimate the volumes of the pad and the total volume of the treatment, so that
proppant is introduced when the fracture attains sufficient width. The maximum proppant size may
also be estimated. For example, case B allows a 20/40 proppant to be pumped with a maximum
proppant diameter of 0.0331 inches.
The character of the pumping pressure behavior for the two cases is also different as shown in
Fig. 10.44. These pressure histories are sufficiently smooth to represent real pressure behavior.
The maximum pressure and the pressure at the perforations are plotted vs. the total injected vol-
ume. Note that the pressures plotted are in addition to the reference pressure of 7084 psi. Due to
0.8
0.6
0.4
0.2
0
0 200 400 600 800 1000
TOTAL VOLUME INJECTED (bbl)
TOTAL FRAC VOL/VOL INJ
STEP LEAK VOL/INJ VOL
hydrostatic pressure the maximum pressure occurs below the perforations. Case A demonstrates a
typical pressure decrease during pumping which is characteristic of unconfined fracture growth of
a penny shaped fracture. Case B shows a complicated pressure behavior at the early pumping stag-
es. This is due to the presence of the pressure barrier in the dense zone which temporarily confines
the fracture.
In some cases the pressure plot may be used as a closure stress diagnostic tool by comparing the
simulated pressure with the actual pumping pressure during a minifrac test.
Fig. 10.45 shows the evolution of the fracture dimensions. Maximum fracture length, fracture
height above perforations, fracture depth below perforations, and maximum fracture height are
plotted vs. the total volume injected. In case A, the fracture propagates in both the horizontal and
vertical directions. In case B, the fracture is essentially confined height-wise and grows
length-wise in the Tor formation. An estimate of the total fracture treatment volume may be made
from this plot, based on the desired dimensions of the fracture.
Summary
TerraFrac is be a valuable simulation tool both for research and design of hydraulic fractures.
1. It can be used to determine the fracture shape for given in-situ and pumping conditions.
-8200
SHALE C=0 0.75 psi/ft
-8500
0.64 psi/ft
L. HOD C=0.002 ft/ min
-8600
6700 6800 6900 7000 7100 7200
FORMATION PROP.
CLOSURE PRESSURE (psi)
E = 1.26 X 106 psia
= 0.4
FLUID VISCOSITY
= 90 cp
PUMPING RATE CLOSURE STRESS A
Q = 15 bbl/min CLOSURE STRESS B
2. It can be used to study the effect of the location of the perforations and the associated problems
of width pinching.
3. It may be used to diagnose in-situ closure stress features by comparing the actual minifrac pres-
sure with simulated pressure.
It is possible, however, to make some overall proppant scheduling judgements using the history
plots. For example, the proppant volume at screen-out conditions should be less than the fracture
volume at any instant, and this leads to an upper limit for proppant loading per fluid gallon.
150
1338 bbl SHALE
125
896 bbl
100
570 bbl
75 TOR
346 bbl
50
201 bbl
Y FEET
25 DENSE ZONE
0
113 bbl
-25
-50
-75 U HOD
-100
-125
-150
0 50 100 150 200 250 300
X FEET
140 SHALE
120
631 bbl
100
442 bbl
80
298 bbl
60
128 bbl
Y FEET
40
TOR
20 DENSE ZONE
0
-20
U HOD
-40
138 bbl
-60 87 bbl
-80
0 40 80 120 160 200
X FEET
175
150 SHALE
125
100 1424 bbl
75 1012 bbl
TOR
Y FEET
50 751 bbl
25 DENSE ZONE
0
-25
U HOD
-50
-75
-100
0 50 100 150 200 250
X FEET
150
A17A
125
1338 bbl
100
75
50
25
Y FEET
0
-25
-50
-75
-100
-125
-150
0.00 0.05 0.10 0.15 0.20
WF IN
A17B
140
120 631 bbl
100
80
60
40
Y FEET
20
0
-20
-40
-60
-80
0.00 0.05 0.10 0.15 0.20 0.25
WF IN
0.25
A17A
0.20
0.15
INCHES
0.10
0.05
0.00
0 500 1000 1500 2000
TOTAL VOLUME INJECTED (bbl)
0.30
A17B
0.25
0.20
INCHES
0.15
0.10
0.05
0.00
0 400 800 1200 1600
TOTAL VOLUME INJECTED (bbl)
7.084000E+03
200
100
0
0 500 1000 1500 2000
TOTAL VOLUME INJECTED (bbl)
400
300
PSI
200
100
400
A17A
300
FEET
200
100
0
0 500 1000 1500 2000
TOTAL VOLUME INJECTED (bbl)
600
A17B
400
FEET
200
MAX FRAC LENGTH (ft)
X MAX FRAC HEIGHT (ft)
MAX HEIGHT ABOVE CNTR (ft) 0
0 400 800 1200 1600
X MAX DEPTH BELOW CNTR (ft) TOTAL VOLUME INJECTED (bbl)
10.3References
1. Gough, D. I. and Bell, J. S.: Stress Orientations from Oil Well Fractures in Alberta and Texas, Cdn. J. Earth
Sci. (1981) 18, 638.
2. Thorpe, R. and Springer, J.: Relationship Between Borehole Elongation and In Situ Stress Orientation at the Ne-
vada Test Site, paper presented at the 1982 U.S. Rock Mechanics Symposium, Berkley, CA, Aug. 25-27.
3. Babcock, E. A.: Measurement of Subsurface Fractures from Dipmeter Logs, AAPG Bull. (July 1978) 62, 1111.
4. Brown, R. O., Forgotson, J. M., and Forgotson, J. M. Jr.: Predicting the Orientation of Hydraulically Created
Fractures in the Cotton Valley Formation of East Texas, paper SPE 9269 presented at the 1980 SPE Technical
Conference and Exhibition, Dallas, TX, Sept. 21-24.
5. Bell, J. S. and Gough, D. I.: Northeast-Southwest Compressive Stress in Alberta: Evidence from Oil Wells,
Earth and Planetary Sci. Letters, 45, 475-82.
6. Dutton, R. E., Nolte, K. G., and Smith, M. G.: Use of the Long-Spaced-Digital-Sonic Log to Determine Rela-
tionships of Fracturing Pressure and Fracture Height for Wells in the East Texas, Cotton Valley Tight Gas Play,
Amoco Production Company Report F82-P-12 (February 15, 1982).
7. Beaudoin, G. J.: Interpretation and Use of 3-D Sonic Data: A Preliminary Study, Amoco Production Company
Report F80-E-13 (September 1980).
8. Smith, M. G., Rosenberg, R. J., and Bowen, J. F.: Fracture Width: Design vs. Measurement, paper SPE 10965,
presented at the 1982 SPE Annual Technical Conference and Exhibition, New Orleans, Sept. 26-29.
9. Zamenek, J. et al.: The Borehole Televiewer - A New Logging Concept for Fracture Location and Other Types
of Borehole Inspection, JPT (June 1969) 762-74; Trans., AIME, 246.
10. Bredehoeft, J. D., et al.: Hydraulic Fracturing to Determine the Regional In Situ Stress Field, Piceance Basin,
Colorado, Bull., GSA (Feb. 1976) 87, 250-58.
11. Dobkins, T. A.: Improved Methods To Determine Hydraulic Fracture Height, JPT (April 1981) 719-26.
12. Nolte, K. G.: Fracture Design Considerations Based on Pressure Analysis, paper SPE 10911 presented at the
1982 SPE Cotton Valley Symposium, Tyler, TX, May 20.
13. Nolte, K. G.: Analysis of Pump-In/Shut-In Tests for Closure Pressure, Amoco Document.
14. Rosepiler, J. M.: Determination of Principal Stresses and Confinement of Hydraulic Fractures in Cotton Val-
ley, paper SPE 8405 presented at the 1979 SPE Annual Technical Conference and Exhibition, Las Vegas,
Sept. 23-26.
15. Nolte, K. G.: Determination of Fracture Parameters from Fracturing Pressure Decline, paper SPE 8341 present-
ed at the 1979 SPE Annual Technical Conference and Exhibition, Las Vegas, Sept. 23-26.
16. Heidt, J. H., Nolte, K. G., and Smith, M. B.: Fracturing Field Research Programs, unpublished Amoco Re-
search document, September 1981.
17. Wood, M. D., Pollard, D. D., and Raleigh, C. B.: Determination of In-Situ Geometry of Hydraulically Generated
Fractures Using Tiltmeters, paper SPE 6091 presented at the 1976 SPE Annual Technical Conference and Ex-
hibition, New Orleans, Oct. 3-6.
18. Wood, W. D.: Method of Determining Change in the Subsurface Structure Due to Application of Fluid Pressure
to the Earth, U.S. Patent No. 4,272,696, (1981).
19. Davis, P. M.: Surface Deformation Associated with Dipping Hydrofracture, J. Geophysical Res. (1983) 881,
No. 87, 5826.
20. Pollard, P. O. and Holzhausen, G.: On the Mechanical Interaction Between a Fluid-Filled Fracture and the Earth
Surface, Tectonophysics (1979) 53I, 27.
21. Lacy, L. L.: Comparison of Hydraulic-Fracture Orientation Techniques, SPEFE (March 1987) 66-76; Trans.,
AIME, 283.
22. Schuster, C. L.: Detection Within the Wellbore of Seismic Signals Created by Hydraulic Fracturing, paper SPE
7448 presented at the 1978 SPE Annual Technical Conference and Exhibition, Houston, Oct. 1-3.
23. Pearson, C.: The Relationship Between Microseismicity and High Pore Pressure During Hydraulic Stimulation
Experiments in Low Permeability Granite Rock, J. Geophysical Res. (Sept. 1981) 86, 7855-64.
24. Albright, J. N. and Pearson, C. F.: Acoustic Emissions as a Tool for Hydraulic Fracture Location: Experience
at the Fenton Hill Hot Dry Rock Site, SPEJ (Aug. 1982) 523-30.
25. Dobecki, T. L.: Hydraulic Fracture Orientation by Use of Passive Borehole Seismics, paper SPE 12110 pre-
sented at the 1983 SPE Annual Technical Conference and Exhibition, San Francisco, Oct. 5-8.
26. Teufel, L. W.: Prediction of Hydraulic Fracture Azimuth from Anelastic Strain Recovery Measurements of Ori-
ented Core, Proc., 23rd U.S. National Rock Mechanics Symposium (1982) 238-46.
27. Rowley, D. S., Burk, C. A., and Manual, T.: Oriented Cores, Christensen Technical Report, Christensen Dia-
mond Products (Feb. 1981).
28. Robertson, E. C.: Viscoelasticity of Rocks in State of Stress in the Earths Crust, W. Judd (ed.), (1964) 181-224.
29. Blanton, T. L.: The Relation Between Recovery Deformation and In-Situ Stress Magnitudes, paper SPE 11624
presented at the 1983 SPE/DOE Low-Permeability Gas Reservoirs Symposium, Denver, March 14-16.
30. Blanton, T. L. and Teufel, L. W.: A Field Test of the Strain Recovery Method of Stress Determination in Devo-
nian Shales, paper SPE 12304 presented at the 1983 SPE Eastern Regional Meeting, Champion, PA, Nov. 9-11.
31. Teufel, L. W. et al.: Determination of Hydraulic Fracture Azimuth by Geophysical, Geological, and Orient-
ed-Core Methods at the Multiwell Experiment Site, Rifle, Colorado, paper SPE 13226 presented at the 1984 An-
nual Technical Conference and Exhibition, Houston, Sept. 16-19.
32. Smith, M. B., Ren, N. K., Sorrels, G. G., and Teufel, L. W.: A Comprehensive Fracture Diagnostic Experiment.
Part II. Comparison of Seven Fracture Azimuth Measurements, paper SPE 13894 presented at the 1985 Sympo-
sium on Low-Permeability, Denver, May.
Hole Diameter
Perforation hole diameter directly affects the proppant size and maximum concentration that can
be pumped during a fracturing treatment. Perforations must be large enough relative to the maxi-
mum proppant diameter to prevent bridging. Fig. 11.1 shows the minimum recommended perfo-
ration size necessary to inject various size proppants at different concentrations. For example, to
pump 20/40 mesh sand at 10 ppg, a minimum perforation diameter of 0.20 in. is recommended.
casing grade
casing thickness.
API charges are tested in casing from K-55 to L-80. When using P-110 and harder casing, the
entrance hole size will be reduced by as much as 20%.
The ideal stand-off to obtain maximum performance from a perforating gun is approximately
1/ 4 in. to 3/4 in., depending on gun size and charge design. If stand-off is significantly greater
than this, hole diameter and penetration will be reduced. Also, if the jet charges do not exit the
port plugs of the gun through the near center of the plug, perforation performance can be
dramatically reduced. Following a perforation job, all guns should be inspected to determine
what percent of charges fired and any misalighned firing through the port plugs.
Table 11.1 provides a very approximate chart of gun type/size, casing/tubing size, and weight
charge versus perforation entry hole diameter. These diameters were generated by various service
companies using the API recommended cement target. Results from different service companies
can vary dramatically; thus, this chart should only be used as a rough reference. When
determining the most appropriate perforating gun and weight charge, the service company should
be consulted to obtain the most recent data and recommendations.
Table 11.1 - Approximate Chart of Gun and Casing/Tubing Sizes Versus Charge Size and Entry Hole
Diameter for Various Type Perforating Guns.
NOTE: Entry hole diameters generated with API Concrete Target test.
Number of Perforations
In addition to perforation size, the number of holes open affects the injection rate at which a frac-
ture treatment can be pumped. To determine the number of perforations required for a specific
treatment design, the following equation can be used
( P pf ) ( d pf ) 4 ( ) 2 1/2
i pf = -------------------------------------- (11.1)
0.2369 ( )
where, i pf is the specific injection rate per perforation (bpm/perf), P pf is perforation friction
(psi), dpf is perforation diameter (in.), is the perforation coefficient (usually 0.9), and is the
maximum fracturing fluid (slurry) density (lbs/gal). is an efficiency number that corrects for
the fact that all perforations are not perfectly circular or smooth orifices. Assuming minimal per-
foration friction, a value of 100 psi is normally used in the equation.
If a well has already been perforated before the fracture treatment design is formulated, which is
usually the case; Eq. (11.1) can be used to determine the maximum injection rate through the
available perforations or decide if additional perforations are required. An example of this is
shown in Table 11.2 for a well that was perforated and tested and found to have much higher per-
meability and skin than anticipated. Initially, the well was shot 2 spf over the 20 ft pay interval with
a hole diameter of 0.38 in. From testing, the well appeared to have a permeability of 200 md and
a skin of +20. Based on fracture modeling, a treatment rate of 40 bpm was desired, limited by the
workstring, and to obtain good conductivity through the damaged region, a maximum proppant
concentration of 10 ppg 20/40 mesh sand was required. As seen in Table 11.2, the minimum num-
ber of perforations required for this treatment was about 110 or 70 more than available. Thus, the
well had to be either reperforated prior to fracturing or the maximum injection rate reduced to 15-
20 bpm, the later probably not feasible given the expected high fluid leak-off.
Perforation Phasing
When perforating for a fracture treatment smaller phasing angles are better, i.e., 90 or 120 phas-
ing better than 180 or 360 (same as 0) phasing. As shown in Fig. 11.2, if enough of the perfora-
tions are not in the near direction of the preferred fracture azimuth, the fracture must traverse
around the outside of the cement to reach this orientation. Since the fracture will propagate perpen-
dicular to the least principle stress, the portion of the fracture which travels around the wellbore
will be subjected to higher stress, resulting in a narrower width or pinch-point. This causes a
high fluid shear environment and can result in fluid degradation and proppant bridging and an
ensuing screenout. This type environment is the most common cause of tortuosity or a tortuous
fracture path caused by some near-wellbore restriction such as described above. Most cases of tor-
tuosity can be cured with proper perforating to insure good communication between the wellbore
and main fracture body. This will also, typically, result in reduced treating pressures (lower HHP
costs) and better post-frac performance.
Table 11.2 - Example Calculation of Number of Perforations Required or Maximum Rate Obtainable
for Fracturing Treatment Design.
Determine the number of perforations required to inject at 40 bpm or the maximum injection rate
possible with the existing 40, 0.38 in. holes. Assume a perforation friction of 100 psi. The max-
imum planned slurry design is 14.59 lbs/gal.
1. To safely inject at 40 bpm:
( P pf ) ( d pf ) 4 ( ) 2 1/2
i pf = --------------------------------------
0.2369 ( )
i pf = 0.7 bpm/perf
patterns can play a dominant role in fracturing from non-vertical wellbores. To better under-
stand this, the following briefly describes possible fracture to wellbore patterns in deviated wells.
While current state-of-the-art does not allow complete quantification of the effects of well devi-
ation on fracturing, it is clear that these effects are related to two angles: (1) the well deviation from
vertical, , (assuming a vertical fracture) and (2) the difference in direction between the wellbore
and the preferred fracture azimuth, , as shown in Fig. 11.3. Best communication will exist when
these two angles are minimized. Basically, there are five possible patterns of wellbore to frac-
ture communication for deviated wells (and vertical fractures). First is when the wellbore is
parallel to the maximum horizontal stress direction, i.e., parallel to the preferred fracture azimuth,
and the fracture follows the wellbore. This is the only good scenario and fracture behavior can
be expected to be similar to behavior for a vertical well. The remaining four patterns are illustrated
in Fig. 11.4, and in order of increasing badness, include (1) a single fracture along the wellbore
turning gradually to the preferred orientation, (2) a single fracture parallel with the well but then
turning sharply to follow the preferred azimuth, (3) a single fracture crossing the well, and (4) mul-
tiple fractures crossing the well. In each of these cases, high apparent downhole friction may
be caused by near-wellbore fracture width restrictions (tortuosity).
For the most awful case, i.e., multiple fractures crossing the wellbore, a small clustered group
of perforations is often used as shown in Fig. 11.5, though this may not totally eliminate multiple
fractures. To totally eliminate the possibility of multiple fractures, a single plane of perfora-
tions is desired, or even better a notched casing using abrasive techniques. Some perforation
patterns may maximize the chances of creating the preferred single fracture along the wellbore. In
particular, two lines of perforations (0-180 phasing), properly oriented, with a minimum
spacing between holes, should maximize the chances of this occurring. The fracture, though, may
then still have to turn to follow the preferred azimuth. Any real calculation of an optimal perfo-
Fig. 11.4 - Bad to Awful Patterns of Wellbore to Fracture Communication for Deviated Wells
(and Vertical Fractures).
Over-Pressured Perforating
Another procedure first introduced in Prudhoe Bay, is the combination of in-line perforations
with super over-pressure perforating. ARCO has shown that a rapidly propagating fracture
turns much more slowly and smoothly to follow its preferred direction than a hydraulic fracture
propagating at a normal speed. The following perforation procedure is followed: (1) a small
volume of water is placed in the bottom of the well, (2) the perforation guns are then positioned
and the remainder of the well filled with nitrogen at relatively low pressure, (3) water is then
injected into the top of the well to compress the gas and increase bottomhole pressure to a level far
beyond the fracture closure stress, and (4) the perforating guns are fired, opening perforations in
the pipe and creating and rapidly propagating a fracture (downhole injection rates on the order of
100s of bpm have been measured during the initial breakdown following perforating). Since the
fracture is being created with pressure greater that the other in-situ stresses, a fracture can open
and propagate at unfavorable angles. This high pressure, combined with dynamic effects of rapid
propagation cause a smooth, slow turning to the favorable fracture orientation, e.g., case 1 in
Fig. 11.4. Thus, in principal, this procedure should produce the least non-ideal deviated well
fracture, though of course for certain combinations of wellbore orientation and in-situ stresses,
the same procedure could cause the very undesirable, multiple, crossing fractures, with the critical
conditions where this might occur again being related to the differences in the three directional in-
situ stresses. Since the directions and magnitudes of all in-situ stresses is usually not known, deter-
mining the proper conditions for this type of completion becomes subject to field experiments.
Other Considerations
The location and length of the perforated interval needs to also be considered under certain cir-
cumstances. For example, if a large pay zone is bounded above by a zone of similar stress, it may
be more conducive to perforate only the lower half of the pay to obtain more complete vertical cov-
erage. With the entire pay perforated, the fracture would tend to initiate in the top half and might
grow more in an upward direction and place a large portion of the treatment in non-pay. Similar
perforating strategy might also be appropriate when an oil-water or gas-oil contact is in the near
proximity to the pay zone.
Performing a fracture treatment down casing can be quite beneficial, this configuration allowing
higher injection rates and lower surface treating pressures and, in turn, requiring less fluid and
hydraulic horsepower to perform the treatment. In certain situations, though, it may be necessary
to pump down tubing with a packer to isolate the annulus, i.e., when the casing is not strong enough
to withstand fracturing pressure or shallower perforations exist. The third configuration, i.e.,
pumping down open-ended tubing, allows fracturing BHP to be obtained via the open annulus and
this can be a very valuable tool in determine fracturing behavior, especially on early wells in a
development program. The disadvantages to this configuration, however, are that the casing must
be strong enough to withstand fracturing pressure and pumping down tubing lowers the injection
rate and/or increases the surface treating pressure. This and alternative methods of measuring frac-
turing BHP are presented later. First, though, the following discusses the pressure limitations of
each configuration and briefly how to determine them.
p s = BHTP p h + p f + p pf (11.2)
where, BHTP is the expected bottomhole fracturing pressure, p h is hydrostatic pressure, p f is pipe
friction, and p pf is perforation friction. While p h and p f are easily calculated or data exists, often
times BHTP and p pf are unknown at onset of a treatment. In an exploratory or new development
well, a minifrac may be in order to determine these values, along with in-situ stress and fluid leak-
off data.
Casing burst values can be found in most service company or casing design handbooks. To deter-
mine a safe surface treating pressure, a burst safety factor of 1.1 is recommended for fairly new
casing. For older casing, this should be increased. Assuming a 2000 psi sudden increase in pres-
sure if a screenout occurs, the design treating pressure should not exceed the safety factor
reduced casing burst pressure minus 2000 psi.
Pop-offs or pressure relief valve should always be installed on the injection line(s) and set/tested
to just below the predetermined maximum surface treating pressure.
When the tubing is latched or anchored in the packer, disallowing tubing movement, forces on
the packer should be calculated to select a packer strong enough to withstand these forces. Tubing
pressure will cause an upward-acting force below the packer, and the annular pressure will
cause a downward-acting force above the packer. This can be computed by the equation
F a = [ ( A p Ao ) po ] [ ( A p Ai ) pi ] (11.3)
When a locator seal assembly is used, allowing tubing movement, the forces and length changes
on the tubing must be calculated to determine the length of seals to run and slack-off when landing
the tubing. The four different effects that cause these forces and length changes are
1. piston effect,
2. buckling effect,
4. temperature effect.
The first three are caused by pressure changes and the last by temperature changes in the well-
bore. Table 11.3 includes the equations used to calculate these to determine the length of seals
required and tubing slack-off for fracturing conditions. Again, screenout conditions need to always
be considered in designing the tubing/packer configuration.
This configuration also allows pumping of a fracture treatment down the annulus and monitoring
the BHTP on the tubing side. When pumping down the annulus, a blast joint should be used at the
top of the tubing string to prevent erosion. Again, the maximum surface treating pressure should
not exceed the safety factor reduced burst pressure of the casing. Also, screenout conditions should
always be considered in determining the maximum treating pressure and pressure relief valves set
to just below this pressure on all injection lines.
open-ended tubing,
0.2L
dLm = --------------------
- -------------------------------- (11.9)
1.0 10**7 R**2 1 L = length of tubing or casing, in. [cm]
dLm = length change due to mechanical force, in. [cm]
TEMPERATURE EFFECT: dLt = total length change due to changes in pres. & temp., in. [cm]
F4 = 207 ( As ) ( dT ) (11.10)
dL1 = length change due to piston force, in. [cm]
dL4 = 0.0000069 ( L ) ( dT ) (11.11) dL2 = length change due to buckling, in. [cm]
Hydraulic Fracturing Theory Manual
SLACKOFF EFFECT: dL3 = length change due to ballooning force, in. [cm]
dL4 = length change due to temperature force, in. [cm]
( Fm ) ( L ) ( dr **2 ) ( Fm**2 )
WELLBORE CONFIGURATION
dLm = ------------------ + ---------------------------------------- (11.12) dpi = change in pressure in tubing at packer, psi [kPa]
( E ) ( As ) 8EI ( Ws + Wi Wo )
dpo = change in pressure in annulus at packer, psi [kPa]
TOTAL EFFECT: dr = clearance between casing ID and tubing OD, in. [cm]
dT = change in average temperature, deg F [deg C]
= F 1 + F 3 + F 4 + Fm
Fp (11.13)
Wi = weight of fluid inside tubing, lbm/in. [kg/cm]
dLt = dL1 + dL2 + dL2 + dL3 + dL4 + dLm (11.14)
Wo = weight of fluid in annulus displaced by tubing, lbm/in. [kg/cm]
ACTUAL FORCE: Ws = weight of steel, lbm/in. [kg/cm]
Fa = [ ( Ap Ao ) po ] [ ( Ap Ai ) pi ] (11.15)
11
gauges in a tail-pipe below a perforated joint below the packer,
To obtain fracturing BHP with open-ended tubing, the annulus must be kept full of a known den-
sity fluid and the annulus surface pressure measured. The main advantage of this system is that can
be monitored real-time. To insure that the annulus is full of the same density fluid, the wellbore
should be circulated prior to the fracture treatment and the density of the fluid measured.
Gauges are often times placed in a nipple in a tail-pipe configuration below a perforated tubing
joint below the packer. This is a good method for obtaining fracturing BHP when the annulus
must be isolated. While proppant may fall down around the gauges, making retrieval with wire-
line difficult, the data can still be retrieved by pulling the tubing string. When possible, a fishing
neck should be installed on top of the gauges and the location of the gauge landing nipple placed
so that the fishing neck extends into the perforated joint. Proppant is less likely to pack around the
fishing neck, making wireline retrieval of the gauges more likely.
Gauges can be placed in the rathole, but this requires going in and washing out proppant settled
from the under-flush to retrieve them.
Measurement of fracturing BHP can be a valuable tool in evaluating fracturing behavior, espe-
cially on early development wells in an area. The best method of retrieving this data must be deter-
mined as a function of the wellbore configuration and the requirements of the treatment.
size and number have to be considered when determining the design injection rate to prevent
tool erosion and shear-thinning of the fracturing fluid.
The diameter of the tool ports must also be large enough to prevent proppant bridging and even-
tual treatment screenout in the gravel-pack tool. The previous section discussed proppant
bridging in perforations and the same applies here. Depending on the number and port diameter,
the design maximum proppant concentration may have to be limited.
Another consideration deals with the blank pipe normally used to extend from the top joint of
screen to the bottom of the gravel-pack tool assembly. This must be of a sufficiently high enough
grade to withstand maximum collapse forces during a frac-pack operation. This needs to be
determined under screenout conditions.
permeability (logs/core),
In an early development well these would need to be measured or determined directly. In later
development wells, though, it might be possible to extract reasonable estimates from offset wells.
This again would depend, to a great degree, on the spacing of the wells, the complexity of the local-
ized geology, and the number and behavior of previous treatments in the area.
Fracture Geometry. After it has been established that a fracture stimulation will provide suffi-
cient economic recovery, certain data and parameters are required to ascertain what size treatment
is required to optimize recovery. For fracture length, width, and height determination, the follow-
ing data is required.
Again, the extent to which this data is collected should depend on when the well is drilled in a
development program, the availability of data from previous wells, the complexity of geology, and
the number and behavior of previous treatments. For example, injection/decline tests and minifracs
may only be required on a select few wells early in a development program to ascertain formation
stresses and leak-off. Also, overburden stress and Young's modulus values should only be required
on early wells, unless the geology varies significantly from one area to another in the development
region.
Treatment Fluid and Proppant Evaluation. The areas that need to be addressed when optimiz-
ing treatment fluid and proppant requirements are
ability of the fluid to carry the proppant the desired distance out in the fracture,
the strength and size of the proppant to provide the necessary fracture conductivity.
Laboratory testing by the service company may be required early in the life of a development pro-
gram to choose the most appropriate fluid and proppant.
fluid/chemical/proppant amounts,
on-site storage,
equipment,
The expected treatment schedule should be reviewed by the service company at the earliest pos-
sible date to insure that these requirements can be met. Often times, in remote areas, chemicals or
proppant may have to be ordered weeks or even months in advance. Also, additional equipment
may have to be brought in from other areas or scheduled.
effectively acquired. Regardless which party takes on the bulk of responsibility for the design, all
designs should be reviewed by the other partner to insure there are no differences or questions
before proceeding to the field. When prepared, the frac brief should also be reviewed and
approved by both parties since, ultimately, safety issues are still the responsibility of Amoco.
Once the design and procedure have been prepared, under the alliance arrangement it is the service
company's responsibility to implement the treatment. This includes making sure the location size
is adequate, that adequate storage tanks are provided, and that sufficient fluid, chemicals, proppant,
equipment, and personnel are available to fulfill the requirements of the treatment. Certain phases
of this will require interaction with the operator representative, e.g., enlargement or grading of the
location.
During the treatment it is the ultimate responsibility of the service company to insure that the
materials pumped meet design specifications and that proper quality control procedures have been
implemented to insure this. Periodically, the service company should be called upon to demon-
strate to the operator the quality of the fluids and proppants on-site. It is also the service company's
responsibility to see that the treatment is pumped as close to design as possible, adhering to safe
practices as dictated by both parties. An operator representative should be present for the pre-treat-
ment safety meeting and treatment execution and should be allowed to interject comments or make
changes to the treatment if deemed necessary to insure completion of the treatment or to prevent
an unsafe situation.
Post-treatment appraisal should include both parties. Any deviations from the design and prob-
lems encountered with equipment or fluids should be documented and contingencies formulated
to help prevent reoccurrences.
Sufficient viscosity to develop the necessary fracture width and transport proppants the
desired distance into the fracture.
An efficient (i.e., low fluid loss) fluid to minimize the amount of fluid required.
Easy to remove from the formation with minimal damage to the formation and proppant pack.
Choosing a fracturing fluid will require compatibility testing by the service company with forma-
tion core/fluids and rheology testing with the actual base (source) mixing fluid. In a new area or
formation, where no historical data exists, these tests should always be performed. Starting with
this step, a proper fluids quality control program should include the following:
Choosing the appropriate gel system and familiarity with this system.
Sampling.
Since most treatment today are performed with water-based, oil-based, or foam-based fracturing
fluids, the following focuses on quality control measures for these.
When a water-based gel is used, it is imperative that a fresh, clean water be used and that the ser-
vice company check the ion content and bacteria count. Certain ions, bacteria, and other foreign
materials can interfere with the proper building of a quality fracturing fluid. One example of this
is a source water used in Australia where the natural borate content is high and causes a weak
crosslink of the base gel if the pH is lowered. This was discovered through laboratory testing and
prevented a potentially nasty situation, i.e., crosslinking of the base gel in the frac tanks. Instead
these jobs are successfully pumped by adding the pH reducing chemicals on-the-fly.
Table 11.4 is an example form for testing the base mixing water. The three most important com-
ponents of a base mixing water are the iron content, pH, and bacteria count. For most water-
based gel systems, the total iron content should be less than 25 mg/liter. Excess iron can
reduce the temperature stability of the gel as well as causing the gel to be more shear-sensitive
when crosslinked. Excessive iron is usually introduced into the system through rusty frac tanks
or transports delivering the water to location.
The pH of the base mixing water should be in the 6 to 8 range. A pH higher than 8 can cause
poor gel hydration and a ph less than 6 can cause gel lumping and fish eyes. It is desirable to
start with a base mixing water pH close to a 7. A pH buffer, acid, or base can be used to bring the
pH into this desired range.
One of the more common sources of gelation problems is bacterial contamination of the base
water. Certain types of bacteria thrive on gel as a food source, destroying the gel structure by
bacterial enzymes. Sulfate reducing bacteria are most common, converting sulfates in the fluid
and reservoir to sulfide, a detrimental formation blocking agent. This type of bacteria is
characterized by a blackening of the water and a strong hydrogen sulfide odor.
Bacteria presence is most common during summer months when temperatures exceed 80F.
During hot periods, bacteria growth accelerates in stagnant water such as that stored in frac tanks
for an extended period (as little as a few days). As a preventative, bactericide should always be
added to the frac tanks prior to filling. This measure is more effective and less expensive than
combating the bacteria after it has flourished. If the water becomes contaminated, dispose of the
water, re-clean the tanks, and re-fill with bactericide treated water. Adding bactericide to a
contaminated tank will not solve the problem. This will only kill the bacteria, but the bodies
and enzymes will remain.
In many cases, the water should be filtered through a 10 micron filtering system. While this
is more costly and time consuming, it can prevent much larger costs incurred if the gel cannot be
properly mixed and has to be disposed of. When city water is routinely used, filtering should not
be required. The service company/operator must make sure, though, that the source water deliv-
ered to location is the same as that requested and not contaminated in transit. Visual inspection
of the water in the frac tanks should be a routine step prior to mixing any gel. This may help
detect contaminants in the water and, possibly, the improper filtration of the water.
When oil is the base fluid, usually lease crude or diesel, compatibility/stability tests should be
performed with the gel system chemicals (preferably those to be used during the treatment). Most
systems will not gel as easily with crudes having an API gravity of 30 or higher. Also, some
diesels may contain detrimental components. The content of diesel will vary with supplier,
refinery, and seasonal changes. Special additives are included in extreme cold regions to prevent
diesel freezing and these can be detrimental to the gelling and gel breakage process. Oil-based
fluids are more difficult to mix than water-based gels and by knowing the properties of the oil
and performing early pilot tests, the first step has been taken to insure the fluid can be prepared
on-site with the least amount of difficulty.
WATER-BASE FLUID:
Iron Content <20 mg Fe/liter
- Reduce Temp. Stab. of Gel
- Transports / Frac Tanks
pH between 6.0-8.0
- >8.0 Poor Gel Hydration
- <6.0 Lumping or Fish-Eyes
Bacteria
- Most Common Above 80 deg F
- Will Destroy Gel Structure
- Hydrogen Sulfide Odor
OIL-BASE FLUID:
Lease Crude or Diesel
deg API Gravity or Lower Best
Diesel Content Change with Supplier, Refinery, and Seasonal Change
Compatibility Test is a Must
If frac tanks are showing signs of excessive rust and wear, the valves do not operate freely,
and/or the tanks are not thoroughly clean they should be rejected. This will require an inspection
by the service company and/or operator. Ultimate care should be taken to insure that the transport
and on-site storage of the base-mixing fluids results in a clean fluid to start with in mixing the
fracturing fluid. Of all the quality control measures, this is one of the most important to
preventing gelation problems. Prevention far exceeds the cure.
Water-based fluids can be broken into two categories, i.e., linear and crosslinked gels. Linear gel
is water thickened with a viscosifier, the more common viscosifiers being guar, hydroxypropal
guar (HPG), hydroxyethylcellulose (HEC), and carboxymethyl HPG (CMHPG). With linear gels
the only means of increasing the viscosity is to increase the polymer loading. Crosslinked fluids
on the other hand start with a linear gel and a borate or metal (zirconate or titinate) crosslinker is
added which ties or bonds the polymer molecules together, resulting in a pseudoplastic fluid with
much higher viscosity than obtainable with simple linear gel systems.
Quality control procedures for linear gels are fairly straight forward and include checking the
following:
pH of the fluid.
Prior to mixing any gel in the frac tanks, pilot tests should first be performed with the base-
mixing water from each tank. Minimum equipment requirements to perform these tests, which
should be supplied by the service company, include:
Fann 35 viscometer.
pH meter.
Thermometer.
All pilot tests should be performed using samples of the chemicals supplied on-location for the
treatment. The following procedure should be followed and all phases recorded:
Mix the gel sample to include all chemicals planned for the treatment, including the breaker.
Measure the viscosity of the gel, taking Fann readings at 100, 300, and 600 rpms.
Measure and record the viscosity of the fluid in the heat bath every 30 minutes or so to
determine viscosity degradation for use in the design model and to evaluate breakage of the
fluid. (At a minimum, the fluid should be heated and sufficient breaker added to insure that
the breaker on-site works.)
Fig. 11.1 and Fig. 11.2 are Fann viscometer readings for various linear HPG gel loadings and
temperatures at a shear-rate of 511 sec-1 (corresponds to Fann 35 - 300 rpm reading). The same
are provided for linear HEC gel in Fig. 11.3 and Fig. 11.4 . These can be used as a guide in
checking the initial gel viscosity at surface temperature. Some tolerance should be allowed, i.e.,
a few cp either side of the values shown in the plots, since there will be some variance in the
mixing of each tank, e.g., for a 50 lb HPG gel loading an acceptable range of Fann readings at
70F might be 45-53 cp. If the gel viscosity falls much outside this range, another sample should
be caught from the tank and re-checked. If it is still outside this range, then corrective measures
must be taken to bring it into spec if it can not be used in the tail-end of the treatment when the
formation is coolest
When performing break tests, it is usually necessary to mix several samples with different
breaker concentrations to determine the final breaker loading. The breaker loading should be
tailored so that the maximum effective loading is added throughout the treatment to insure
Fig. 11-9 - Hydroxpropylguar (HPG) - Fann Viscosity v. Temperature for Various Polymer Loadings
in 2% KCI Water.
eventual complete degradation of the gel. When the expected BHT exceeds the achievable
temperature of field heat baths, the breaker scheduling will have to be determined in the
laboratory (also required when go into a new area or formation or where significant changes are
made to the treatment size and/or the fracturing fluid or water source is changed).
Different breaker concentrations may be required for different fluid systems. Also, the
reservoir BHT will dictate the amount of breaker that can be added so the gel degrades in the
time required. Two general types of breakers are available for use, i.e., raw (oxidizing or
enzyme) breaker and encapsulated (delayed) breaker. It has been shown in industry studies that
up to a point more breaker is better from the standpoint of degrading the gel filter-cake
formed on the fracture walls and removing the gel residue from the proppant pack. A rough
rule-of-thumb is to design the breaker schedule so the pad fluid is completely broken in
twice the expected pump time and the tail-end fluid is broken in about an hour after shut-
down. For example, as shown in Table 11.3, on a 1-hour or less treatment, sufficient breaker
should be added to the pad to break the fluid in approximately 2 hours and the breaker schedule
increased so the tail-end fluid breaks within 1 hour after shut-down. Depending on the reservoir
temperature and gel loading, only encapsulated (delayed) breaker may be required in the pad,
whereas in the later stages the raw breaker concentration may be increased and the encapsulated
breaker concentration decreased. When feasible, the breaker schedule should be
Fig. 11.10 - Hydroxyethlcellulose (HEC) - Fann Viscosity v. Polymer Concentration in 2% KCI Water
@ 60F.
Fig. 11.11 - Hydroxyethylcellulose (HEC) - Fann Viscosity v. Temperature for Various Polymer
Loadings in 2% KCI Water.
designed/evaluated on-site based on pilot test ing with the actual mix fluid and breaker
stock provided for the treatment. Ineffective batches of breaker have been known to exist!
Table 11.6 - Sample Fracture Treatment Schedule with Raw and Encapsulated Breaker Ramps.
Fluid Type Fluid Vol Raw Brkr Enc. Brkr Prop Conc Rate
(gals) (#/Mgal) (#/Mgal) (ppg) (bpm)
Quality control procedures for crosslinked gels are much the same as for linear gels with the
exception of checking the ability of the fluid to crosslink, it's crosslinked consistency, and
crosslink time. The same procedures as above should be followed to prepare and test the base
linear gel prior to crosslinking. Crosslinked gel systems generally fall in one of two categories,
i.e., instantaneous crosslink and delayed crosslink. The crosslink time can be controlled by a
number of methods. Some companies control crosslinking by adjusting pH, others vary crosslink
time by changing the crosslinker concentration or by blending crosslinkers, while still others use
retarders or accelerators to control the time to crosslink without changing the crosslinker or pH.
Again, pilot testing is very important in determining if the gel is properly crosslinking and how
long it takes to crosslink. Generally, the best way to test this is to obtain a sample of the
crosslinker on-site and observe the crosslinking of the linear gel in a blender. The speed of the
blender should be set just high enough so a vortex forms in the center of the linear gel sample.
The crosslink time is then measured from the time the crosslinker is added until the vortex closes
and a mushroom forms on top of the sample - this termed the Vortex closure time. For
instantaneous crosslink systems, the gel should form a bonded structure very quick. For a
delayed system, though, this may take some time, depending on the temperature and pH of the
fluid. When testing a delayed crosslinked gel, it is best to heat the base gel to the expected
average wellbore temperature during the treatment to perform the pilot tests. The delay time for
the crosslink is determined by the expected residence time in the pipe, i.e., the gel should ideally
be crosslinking just outside the perforations. This will minimize pipe friction pressure and
minimize the shear on the gel before it enters the fracture.
A good crosslinked gel should exhibit a strong bonding with a smooth texture that can be lipped
out of the sample container and returned as a whole unit. If a gel is under-crosslinked it will
exhibit a weak, slimy, runny appearance absent of strong bonding. At the other end of the
spectrum an over-crosslinked gel will exhibit a chunky, rough, brittle appearance. This type
fluid, while viscous in appearance will have poor temperature stability.
The crosslinker itself. Catch another sample from a different drum on-site.
In special cases where the gel simply will not crosslink properly, it may be due to contaminants
in the base gel. This, however, can be prevented if proper laboratory and pilot testing (including
Again, as for the linear gel systems, breaker scheduling is very important for crosslinked gel sys-
tems. There is nothing worse than pumping a crosslinked gel into a reservoir that might not break.
Undoubtedly many wells have been ruined this way. If possible, breaker tests should be performed
in a heat bath on-site to determine the optimum breaker schedule. If the reservoir temperature
exceeds 200F, the upper limit for most conventional heat baths, then extensive laboratory testing
should be performed by the service company with the actual mixing water and preferably the
chemicals from the treatment stock to determine the best breaker schedule for the desired time
period.
When utilizing resin-coated proppants, these can have a dramatic affect on the crosslink and
break time of crosslinked gel systems. Some of the crosslinker and breaker are neutralized by the
resin, requiring that additional amounts of these chemicals be added to achieve the same gel.
Again, this requires extensive testing by the service company to determine the adjustments
required. This is generally a hard thing to quantify and impossible to determine on-site. A new
encapsulated curable resin-coated proppant has just recently been introduced on the market which
still bonds in the fracture with closure stress yet is inert to fracturing fluids and chemicals. If it
proves to work as advertised, it should eliminate the problems associated with the interaction with
crosslink gel chemicals and breakers.
All pilot test results done on-site should be recorded on a form similar to Table 11.7.
Most crude oil from 28F API and higher can be gelled with this system; however, the particular
crude must be tested for gelation prior to a decision being made on its use. Testing should also be
performed with each diesel source to determine its suitability. The content of diesel may vary with
supplier, refinery, and seasonal changes. Special additives included to prevent diesel freezing can
have a detrimental affect on the gelation and breakage of a gelled-oil system.
The concentration of gelling agent (aluminum phosphate) is the controlling factor in determining
the viscosity of the gel. This concentration will depend on the desired viscosity at BHT. Typical
viscosities achievable with this system are 50-300 cp (170 sec-1) at 80-190F and 50-150 cp at
200-250F. The activator (sodium aluminate or other base) is normally held to a constant ratio
with the amount of gelling agent, e.g., if 8 gals/Mgals gelling agent is used, 3 gals/Mgals of acti-
The concentration of pH control additive is dependent on the pH of the particular oil. Lab testing,
through trial and error, is required to determine the correct concentration of pH control additive.
In lab testing the gelling agent and activator are added first and then the pH control added drop
wise until gel viscosity is observed.
Breakers used for gelled oil systems include sodium bicarbonate (baking soda) and slaked lime or
calcium hydroxide. The sodium bicarbonate is used on most treatments unless the BHT is very low
or a short break time is desired. Typical concentrations of breaker range from 10-
75 lbs/1000 Mgals depending on the BHT, pump time, and the desired break time. As noted for
the water-based gel systems, it is preferable to add as much breaker as lab tests indicate possible
while still maintaining sufficient viscosity to safely complete the treatment. Cases have been sited
where no or an insufficient amount of breaker were added and the gel did not break, causing the
treatment to be ineffective and plugging of the formation.
In a gelled-oil system, the sodium bicarbonate breaker also acts as a fluid loss agent, this being in
a free flowing powder form. Other additives such as Adomite Aqua and silica flour can be used,
however, these are not recommended unless absolutely necessary.
On-site quality control test procedures for gelled-oil systems should include the following:
Add the gelling agent and activator to the sample(s) at the prescribed concentrations
recommended by the service company and previous lab testing.
Determine the amount of pH control additive by adding drop-wise until viscosity develops.
Test the viscosity of the fluid with a Fann 35 viscometer. If the viscosity is within the desired
range, proceed with the next step. If it is too high or too low, prepare another sample, adjusting
the gelling agent and activator concentration and going through the pH additive test again.
Continue to retest until the desired viscosity is obtained or the best gel obtainable is achieved.
After the gelling agent, activator, and pH additive concentrations have been fine-tuned, mix
several more samples with varying concentrations of breaker and immerse these in a heat bath
to monitor the gel degradation and break time. The gel should have sufficient viscosity to
complete the treatment safely and then break back to less than 10 cp at a Fann 300 rpm reading.
An example of the results of this type testing are shown in Fig. 11.12. In this case, it was
determined that 40 lbs/Mgals of breaker was optimum for completing the short pump time
treatment and getting a good break within a reasonable time after the treatment.
These tests should be performed for each tank of oil as some may react differently than others
requiring slight alterations to the chemical additive combination.
Fig. 11.12 - Results of Field Break Test for Oil-Base Gel, Using a Fann 35 Viscometer and Heat Bath
to Determine Optimum Breaker Concentration.
Wearing rubber gloves and safety goggles when handling all chemicals. Some are acidic
and some alkaline and can cause severe burning.
Take the necessary precautions associated with a highly flammable fluid, i.e., ground the
blending and pumping equipment, no smoking on location, use of shrouds on high pressure
discharge lines, and proper fire fighting equipment.
Since most foam treatments use a water-based linear gel for the liquid phase, the same quality
control procedures outlined previously for this type gel should be applied here. Generally, these
would again include checking the base gel viscosity, pH, and temperature to make sure the gel
meets design specifications. Little can be done in regard to quality controlling the gas phase, aside
from checking to make sure sufficient quantity is on-site to conduct the desired treatment.
Because foam fracturing treatments are more complex to perform than single-phase treatments, it
is important that the service company treater and engineer fully understand the surface proppant
schedule and liquid/gas rates to obtain the desired concentrations and rate downhole. A plan should
be carefully laid out with a table such as shown in Table 11.8 for all to follow during the treatment.
Table 11.8 - Sample Schedule Prepared for Constant Clean Side and Nitrogen-Rate Foam Fracture
Treatment.
35,000 10,500 0.0 0.0 0 35,000 10,500 13,820 4.5 0.0 15.0 55:33
25,000 7,500 1.0 3.3 25,000 26,130 8,630 13,820 4.5 0.7 15.7 39:37
30,000 9,000 2.0 6.7 60,000 32,712 11,712 13,820 4.5 1.4 16.4 47:29
12,500 3,750 3.0 10.0 37,500 14,195 5,445 13,820 4.5 2.0 17.0 19:53
10,000 3,000 4.0 13.3 40,000 11,808 4,808 13,820 4.5 2.7 17.7 15:53
7,500 2,250 5.0 16.7 37,500 9,195 3,945 13,820 4.5 3.4 18.4 11:54
1,800 540 0.0 0.0 0 1,800 540 13,820 4.5 0.0 15.0 2:51
TOTALS:
121,800 36,540 200,000 130,840 45,570 193:10
Perform pilot tests as soon as possible on-site to insure that the available chemicals all work
and that the base mixing fluid is not contaminated.
Take tank dips before and after the treatment to help in evaluating the accuracy of metering
during the treatment and to access what was actually pumped.
Set up a sampling program during the treatment to determine that the system is acting as pilot
testing predicted it would. On a relatively small treatment, this might include catching 1-2
samples during the pad and 2-4 samples during the proppant stages. On a larger treatment, with
a pump time of several hours, several samples should be caught during the pad and, preferably,
one sample per proppant stage. In the most severe case on a crosslinked gel treatment, where
the gel is not properly crosslinking and this can be detected before proppant is started, the
treatment should be aborted and the problem remedied before a reattempt of the treatment.
Immerse half of the treatment samples in a heat bath to determine the gel break time and to
determine the earliest possible time at which the well can be flowed back. The remaining
samples should be retained for a period of time after the treatment until the well has cleaned
up. Also, if gelation problems occur during the treatment, these samples may help determine
the cause.
Record all phases of the gel pilot testing, inventory, mixing, and results of treatment sampling.
Proppant concentration.
Proppant grain shape - roundness/sphericity.
Many of these variables can be controlled to varying degrees through proper quality control prac-
tices.
Fig. 11.13 illustrates the affect closure stress has on several types of proppants, showing that at
higher stresses, higher strength proppants will be required to provide adequate fracture conductiv-
ity. If closure stress is unknown, this parameter should be measured through injection/decline test-
ing. A list of currently available proppants and their recommended stress limitations are shown in
Table 11.9, including sand, intermediate-strength, high-strength, and resin-coated proppants.
Industry studies over the past 10-15 years have shown that when proppants are subjected to stress
and temperature for longer periods of time, conductivity decays with most of this decay occurring
over the first 100 hours. Fig. 11.14 shows examples of this for several different type proppants. In
designing the fracture treatment, long-term conductivity data should be obtained from the ser-
vice company and used instead of the more typically published short-term data.
White Sand 2.65 Used to prop open created Low closure stress.
fracture to conduct
hydrocarbons to the
wellbore. Closure pressure
to 4,500 psi. Ranked 1st
among sands.
Carbo ISP-1 Carbo Ceramics 3.16 Intermediate strength 20/40 mesh only.
bauxite. Closure stress up Competes with
to 15,000 psi. Less InterProp 1.
expensive than Carbo-Prop
HC. Broader size
distribution.
InterProp 1 RCP Norton-Alcoa 3.06 Same as above with resin Same as above.
coating for flowback control.
AcFrac SB Acme Borden 2.56 Partially cured white sand, More compatible
ULTRA requires stress for bonding. with oxidizing
For flowback control and breakers.
strength. Will not set up in
wellbore in screenout.
Closure stress to 8,000 psi.
Brown Sand 2.65 Brady sand. Used to prop Low closure stress.
open created fracture to
conduct hydrocarbons to
the wellbore. Closure
pressure to 4,500 psi.
Ranked 2nd among sands.
Colorado Sand 2.65 Used to prop open created Does not meet a
fracture to conduct number of API RP-
hydrocarbons to the 56 guidelines.
wellbore. Closure pressure
to 4,500 psi. Ranked 3rd
among sands.
Arizona Sand 2.70 Used to prop open created Does not meet a
fracture to conduct number of API RP-
hydrocarbons to the 56 guidelines.
wellbore. Closure pressure
to 4,500 psi. Ranked 4th
among sands.
16/20, or 12/20 mesh proppant is used. Particle size distribution can have a measurable affect on
fracture conductivity. For example, as shown in Fig. 11.15, one sand having 90% of the grains fall-
ing between the designated 20/40 mesh sieve screen sizes, as compared to another having only
60% of the grains within the required range, will exhibit much higher conductivity and the contrast
will increase as the closure stress increases.
Testing of proppant size distribution requires a sieve analysis and should be routinely performed
as a quality control practice on most fracture treatments. The American Petroleum Institute (API)
provides several publications detailing tests for sands and intermediate- and high-strength prop-
pants. Shown in Table 11.10 is an excerpt from API RP 56 showing the range of various frac sands
(6/12 to 70/140) and the nest of sieve screens recommended for testing. A minimum of 90% of the
tested sand sample (also generally applies to other proppant types) should fall between the
designated sieve sizes correlative to the indicated mesh size, i.e., 6/12, 12/20, 20/40, etc. Not over
Fig. 11.14 - Example of Long-Term Permeability (Conductivity) Data for Various 20/40 Mesh
Proppants Tested @ 275F and a Proppant Concentration of 2 lbm/sq. ft.
0.1% of the total proppant sample should be larger than the largest sieve screen mesh and not over
1.0% should be smaller than the smallest sieve screen mesh.
Fig. 11.15 - Effect of Proppant Particle Size Distribution on Fracture Conductivity for 20/40 Mesh
Sand at Proppant Stresses of 5,000-10,000 psi.
Fig. 11.17 shows the effect on fracture conductivity for two sands at stresses of 5000 and
10,000 psi, one exceeding API specs and other falling below the recommended roundness/spheric-
ity. From this, it is obvious that the rounder, more spherical proppant results in much higher con-
ductivity, i.e., 61% higher at 5000 psi and 300% higher at 10,000 psi stress.
Proppants routinely used and/or new proppants being considered for use should be subjected to
grain shape testing under a microscope by the service company and/or operator.
Proppant Fines
The presence of silts, clays, and other fine particles in the proppant can also reduce fracture con-
ductivity. Fine particles can be detected through an API recommended turbidity test on-site. The
recommended procedure for this test is:
Using a black marking pen, record the proppant sample identification on one side of a clear
glass 4-ounce prescription bottle (100 ml in 10 ml increments) in characters approximately
1/2 high.
Place the proppant sample in the container and fill to the 20 ml mark, gently tapping and
leveling the sand and further fill to bring to but not exceed the 20 ml mark.
Hold the bottle at arm's length toward a moderate light source with the side of the bottle with
the identification mark facing the light source.
Table 11.10 - Recommended API Proppant Specifications for Frac Sand and Manufactured
Proppants.
Recommended Mesh Size Requirements
A minimum of 90% of the tested sample should fall between the designated sieve size. Not more than
0.1% of the tested sample should be larger than the first sieve size. Not over 1% should be smaller than
the last sieve size.
Recognized Proppant Mesh Sizes
Frac Sand Size + + * + * + * +
Designation 6/12 8/16 12/20 16/30 20/40 30/50 40/70 70/140
4 6 8 12 16 20 30 50
6 8 12 16 20 30 40 70
8 12 16 20 30 40 50 100
USA Sievesa
10 14 18 25 35 45 60 120
Required for Testing
12 16 20 30 40 50 70 140
16 20 30 40 50 70 100 200
Pan Pan Pan Pan Pan Pan Pan Pan
* Primary Mesh Size
+ Alternate Mesh Size
a USA Sieve Series as defined in ASTM E 11-70
Interpretation
If the sample ID can be read through the water phase, the proppant should be judged clean
and suitable for use.
If the sample ID can not be read, the proppant sample should be judged dirty and unsuitable
for use.
Fig. 11.16 - Chart for Visual Estimation of Roundness and Sphericity for Proppants Used in
Hydraulic Fracturing (from Krumbein and Sloss, 1955).
Fig. 11.17 - Effect of Proppant Roundness and Sphericity on Fracture Conductivity for 20/40 Mesh
Sand at Proppant Stresses of 5,000-10,000 psi.
If the sample ID can be read but with some difficulty, let the sample stand for 10 minutes and
repeat the test. If the ID still cannot be read, the sample should be judged unsuitable for use.
The most suitable time to perform the turbidity test is when the proppant bins or trucks are
loaded in order to get a representative sample from a moving stream. Also, performing the test at
this early time in the operation will allow for corrective measures to be taken if necessary without
undue delay to the treatment.
Proppant fines can also be generated in the fracture if the proppant strength is not high enough to
withstand formation stresses. Crush resistance testing should be performed periodically on rou-
tinely used proppants and all new proppants being considered for use, using the recommended pro-
cedure in API RP 56. Testing of this nature can be performed by the service company, the Amoco
Tulsa Technology Center, or by commercial core laboratories. Table 11.11 includes the recom-
mended test cell loads and maximum allowable fines for frac sand.
Table 11.11 - Stress to be Applied and Suggested Maximum Fines for Frac and Sand Crush
Resistance Tests (API RP 56).
If using more than one type or size of proppant on a treatment, know where each proppant is
loaded in the proppant storage bins on location and discuss with the frac operator in what order
these are to be run.
Catch samples of the proppant during various stages of the treatment and label these according
to size/type proppant and when they were caught.
Key personnel from the operator and service co. should have a final review meeting to go over
the treatment pump schedule, maximum treating pressures, lines of authority, contingency
plans in case of mechanical/pressure problems, and safety issues.
Service co. should supply properly working radios and headsets to all personnel at key
equipment focal points, i.e., blender, pumps, sand delivery, frac tanks, valve operator, and frac
operator in control van.
Prior to pressure testing, the service co. should perform a radio check to make sure all radios
and headsets are fully functional. If an insufficient number of working radios are not
available, the treatment should not be performed until this is remedied.
Safety Meeting
The safety meeting should be conducted by the frac operator with all personnel on-site and should
include the following:
Pressure testing.
Operator responsibilities.
Operator safety gear, including safety glasses, hard hats, safety boots, and proper clothing.
Chemical hazards.
Location and use of fire extinguishers, eye wash facilities, first aid kits, and other safety
equipment.
Smoking restrictions.
Pressure Testing
Prior to every fracture treatment, all injection lines and valves should be pressure tested and pop-
offs or pressure relief valves set. All lines should be properly staked and all personnel not
directly involved in the pressure test should move well clear of the area surrounding the lines, well-
head, and fracturing equipment. All lines should be tested to just above the determined maximum
treating pressure set by the operator, with this pressure being held and recorded for a minimum of
5 minutes. If treating down tubing, with the plan to hold back-pressure on the annulus, both tubing
and annulus lines must be tested. All leaks should be eliminated prior to proceeding with the treat-
ment.
Pressure relief valves should always be installed on injection lines and set and tested to just below
the determined maximum treating pressure.
Treating Problems
Oftentimes, treating problems arise during the job that must be dealt with, the most common be
equipment failures, gel not properly crosslinking, proppant delivery problems, and abnormal pres-
sures. These can be very disruptive to the successful completion of a treatment if proper planning
and contingencies have not been put in place. Some of the more common problems are as follows:
Blender failure due to mechanical problem - In most cases, it is advisable to have a standby
blender rigged-up and operational. While blender failures are rare, the cost of standby is
minimal compared to the costs that might be incurred if the treatment has to be prematurely
aborted. Hoses should be run from the standby to the frac tank manifold, the standby blender
tub filled with gel, and sufficient chemicals installed on this blender to complete the job if need
be. Also, when a standby blender is installed, provisions need to be made to change the sand
delivery over to this second blender.
On small treatments, where leak-off is expected to be high, fracture closure time may not
allow sufficient time to switch to a standby blender. If this is the case, a standby serves no
purpose.
Sanding-up the blender tub - This is caused when fluid cannot be transferred to the blender
at a fast enough rate, blender operator error (i.e., letting tub fluid level get too low), and/or
the proppant feed rate into the tub is too high. If this occurs, switch immediately to the
standby blender and resume the treatment. If no provision has been made for a standby, the
treatment may have to be terminated. With a single blender, opening multiple fluid tanks may
increase the feed rate to the blender. If problems are encountered early in the treatment in
Inadequate fluid transfer from tanks to blender - This can be caused by the inability of the
centrifugal pump on the suction side of the blender to draw fluid from the tanks, especially
those a significant distance from the blender; partially closed valves; and/or plugged or kinked
suction hoses. Adequate suction lines of proper sizing must be installed to achieve the desired
rate. On extremely large treatments, transfer blenders may be required to transfer fluid from
remote tanks to the primary blender.
Proppant delivery system failure - This is usually the weak link on any treatment. On smaller
treatments, there is seldom any way to provide backup for the proppant delivery system. If this
fails, the treatment should be flushed. On larger treatments, however, dual-belt proppant
conveyor systems are available, which allow the treatment to be continued if one belt bogs
down or the hydraulics on one side fail. It is important that the proppant delivery system used
be capable of delivering the desired pounds per minute with adequate safety margin.
Pump failures - This is one of the more common occurrences and should be dealt with by
having adequate standby. The amount of standby is usually determined by the nature of the
treatment. Long, high-pressure treatments, where an abrasive proppant such as bauxite is
pumped, should have a minimum of 100% standby. On other treatments, 50% standby is
usually sufficient.
Inaccurate metering - Flowmeters, densimeters, and pressure transducers can sometimes fail.
It is advisable to have at least two of each type meter/gauge in the main injection line. All
should be properly calibrated prior to the treatment, and early in the treatment, they should be
checked against other gauging methods, e.g., the flowmeter checked against tank dips and the
densimeter checked against sand screw RPM's. Prior to and immediately following the
treatment, the fluid, chemicals, and proppant should be inventoried to determine what was
pumped and how this compares with the treatment metering.
Loss of power or electronics to treatment van - This can present a very dangerous
situation if not dealt with properly. A proper contingency plan is required to avoid this. This
would include good communication with the operators and material gaugers - the volume and
rate obtained from the tank gaugers, the sand concentration determined from the blender sand
screw RPM's, and the pressure monitored on the pump trucks.
Fluctuations in treating pressure can often signal quality control problems. Pressure changes can
be caused by mechanical problems, a change in gel properties, varying proppant concentration, and
formation responses.
Abnormal pressures from wellbore conditions - The more common pressure problems
caused by wellbore conditions are excessive pipe or perforation friction, and downhole
equipment failures or leaks. An ISIP early in the treatment or during pre-frac testing can detect
whether or not pipe and/or perforation friction is a problem. Without bottomhole pressure data,
though, it can be difficult to tell the difference between perforation friction and excessive
pipe friction due to improperly mixed frac fluids. One common reason for excessive pipe
friction on crosslinked gel treatments is that the fluid is crosslinking too quick or the
crosslinker addition is incorrect. Also, different tanks of base gel may result in slightly
different crosslinked gel frictions. This should be checked as a first line of action. If the fluid
mixing is not a problem and the treating pressure is excessively high, the treatment should be
aborted. Re-design of the treatment at a lower rate may be required or the well re-perforated or
an acid ball-out performed. Proper pre-frac testing with BHP, including a gel minifrac, can
usually detect these types of problems so they can be remedied prior to performing the
treatment.
When the treatment is pumped down tubing, below a packer, positive annulus pressure should
always be held to immediately detect any communication through a tubing leak or packer
failure. If this occurs, the treatment should be terminated.
Abnormal pressures from formation response - The two most common abnormal pressure
responses caused by the formation are usually a result of fracturing out of zone or pressuring
out (screening-out). If the fracture grows out of zone into a lower stressed interval, a sudden
drop in pressure may be apparent. This, however, is often hard to detect at the surface due to
changing friction and hydrostatic pressures. Pressure increases preceding a screenout may
serve as an early warning signal to flush the treatment. Often, though, this is also masked by
the changing friction and hydrostatic pressures and it is not until a complete screenout occurs
that it is apparent at the surface.
The service companies have developed means of calculating BHTP from surface pressure and
friction correlations to use in detecting downhole pressure changes. Due to the oftentimes
inaccuracy of these correlations, though, the calculated BHTP can be misleading and has
resulted in premature flushing of treatments when, in fact, the rising pressures were nothing
more than poor gel quality and increased friction pressure. As a rule-of-thumb, calculated
BHTP's can not be relied on and should not be used to make real-time decisions during a
fracturing treatment. The only time when they may provide valuable information is when
pumping down large tubing or casing where friction is not a factor and the proppant stages are
large enough that the pipe contains all one slurry.
The flush counter should be immediately zeroed at the time the proppant concentration from the
wellhead densimeter starts to fall off from the maximum slurry concentration. This will generally
result in leaving the maximum proppant concentration in the fracture at the wellbore. Performing
the flush this way, though, will also result in a tail-off of proppant left in the wellbore above
the point of flushing, this tail-off being that in the lines and blender tub behind the wellhead den-
simeter. If adequate rat-hole is not available to accommodate this, the wellbore may have to be
cleaned out.
Another method to prevent some of the proppant tail-off is to by-pass the blender tub. While this
prevents the blender contents from being pumped into the well, there are certain problems that can
occur in switching to the by-pass, the most common of these being a loss of prime on the pumps.
While this method is attractive, it is not recommended.
It is not advisable to flush Foam fracturing treatments with foam. To accurately determine a
foam flush volume, the bottomhole conditions (temperature and pressure) must be accurately
known and this is seldom the case. After a foam fracturing treatment, the reservoir should be
charged up enough to unload a column of water, provided the flowback is initiated in a timely fash-
ion.
When to Flowback
Following a non-foam type fracturing treatment, the well should remain shut-in long enough for
the fracture to close and the tail-end gel to fully break. If closure time is expected to be short,
this can be monitored from surface pressure in the frac control van. In tighter reservoirs, the shut-
in time may be as much as 24 hours. Samples caught during the treatment, should be placed in a
heat-bath immediately after the treatment to monitor the break time. Depending on the size of treat-
ment, it will take some time for the reservoir in the proximity of the fracture to recover to original
BHT. This needs to be taken into consideration to allow some safety margin prior to initiating
flowback.
Where a foam fracturing treatment has been performed, the primary objective is usually to flow
the well back in a relatively short time frame to aid in cleanup from under-pressured reservoirs.
Typically, this is done within 1-2 hours following the treatment and, in most cases, this is sufficient
time for the foam to degrade, i.e., foam half-life generally less than 1 hour. Initiating flowback
immediately after a non-foam type treatment, i.e., forced closure, is not recommended. During
flowback, the fracture will try to close in the near-wellbore region and, if the proppant is still sus-
pended in viscous fluid, much of the proppant in this region of the fracture may be pulled back into
the wellbore. If this happens, as suspected, the result could be poorer near-wellbore fracture com-
munication. Only in cases where the closure time is very long and proppant may tend to settle
beneath the primary pay zone, should forced closure even be considered.
Temperature Logs
Post-frac temperature logs are the most common method of measuring fracture height due to their
operational ease of use. They are easy to run, can be run in cased- or open-hole, and have minimal
impact on operations since the well is typically shut-in for several hours after a stimulation.
Procedure. The most reliable procedure for running post-frac temperature logs is to first run a base
(pre-fracture) log to determine the undisturbed temperature gradient of the formations, then two to
three additional logs following the fracture treatment as shown in Fig. 11.18. The best results are
obtained by logging down, so the temperature sensor is always entering undisturbed fluid, at a
speed of 20-30 fpm. The best time to obtain the base log is prior to any other completion phase,
e.g., perforating, running tubing, etc. This, however, might not always be possible or cost-effective
and the next best method is to run the base log the day before the fracture treatment. The post-frac
logs should be run shortly (1-3 hours) after the treatment and multiple passes made with a mini-
mum of 3/4 to 1 hour between logging starts. No flowback from the well should be allowed prior
to logging, but if this is necessary, it is usually possible to obtain a good log by allowing a couple
of hours for the temperature to re-stabilize following the flow back. In the case of an under-pres-
sured reservoir where the fluid level might continue to fall after the treatment, the fluid level should
be allowed to nearly stabilize prior to running the post-frac logs.
When To Log. Post-frac temperature logs are typically run when there is a question about the
degree of fracture height containment that occurred as compared to model predictions. Fracture
height determination is generally more applicable when stimulating low permeability zones,
where the objective is to achieve long fracture half-lengths. In particular, in new areas with wells
having virgin reservoir pressure where little is known of the boundary stresses, it is usually appro-
priate to conduct a minifrac and run post-minifrac logs to calibrate the model stress profile for
final design determination. Temperature logs may then also be run after the main treatment to con-
firm or verify model predictions. In moderate to high permeability zones usually the main objec-
tive is to by-pass near-wellbore damage and fracture height is not as critical. For these type zones,
though, temperature logs might still be appropriate following a minifrac if the objective is to stay
out of alternative pay zones or water-bearing intervals in close proximity to the target zone.
Also, when fracturing thick intervals of moderate to high permeability, where multiple layers
exist with varying permeabilities and stresses, temperature logs might be appropriate following
a minifrac and/or treatment to evaluate how much of the interval was treated.
Fig. 11.18 - Base and Post-Fracture Temperature Logs: Runs #1 - 8 hrs after Injection and Run #2 -
18 hrs after Injection.
Limitations. Ideal logs, such as shown in Fig. 11.18, are rare and may be in error when they do
occur. To help in evaluating when temperature logs should be run and how best to interpret them,
it is important to understand their limitations. The two primary factors affecting temperature log
interpretation are the created fracture width and wellbore conditions. Low-stress and/or low-
modulus zones can have significantly greater fracture width and will accept the majority of fluid.
This results in more cooling across this region(s) and the largest temperature anomaly. This is both
a strength and weakness: a strength because the log is indicating where the bulk of fluid went and
a weakness because the larger anomaly can mask height growth into higher stress/modulus zones,
leading to misinterpretation of fracture height and geometry.
Wellbore conditions can also affect temperature log interpretation, these including placement of
the downhole assembly and wellbore deviation. Pumping down tubing will create a temperature
anomaly immediately below the tubing because of the difference in radial heat flow rates for a tub-
ing/casing configuration compared with fluid flow just inside the casing.
that portion of the fracture not in direct contact with the wellbore. Thus, in situations such as this,
temperature logs have limited application.
Temperature logs have very little to no application in deviated well fracturing, unless they are
used to determine whether vertical or horizontal fracturing is occurring. Even for this
application, their interpretation may be questionable.
Interpretation. While temperature logs, when used as a stand-alone tool, can be difficult to inter-
pret, they can yield valuable information when interpreted correctly. Several analysis techniques
have been developed to aid in this. One of these is a cold-water circulating test to assist in anal-
ysis. This involves circulating down tubing and up the annulus to cool the wellbore without creat-
ing a fracture. Post-circulation logs then indicate perturbations caused by thermal conductivity
changes and wellbore effects, such as washouts. Post-frac logs can then be compared to this pre-
frac log to identify fluid movement outside the pipe and thus the presence of fracture height
growth. An example of this is shown in Fig. 11.19. For the particular case of a warm nose above
perforations, further evidence of the correctness of this interpretation was given by comparing
post-frac temperature and gamma-ray logs, as shown in Fig. 11.20.
Fig. 11.20 - Comparison of Post-Fracture Temperature Log and Post-Fracture Gamma-Ray Log.
Running of multiple post-frac logs also improves the temperature log interpretation. Fig. 11.21
shows an example where the temperature profile changed with later logs, giving a much easier
interpretation.
Downward fracture growth is difficult to determine with a temperature log. Typically, at the con-
clusion of a fracture treatment, the wellbore fluid below the perforated interval is very near static
reservoir temperature. Thus, the temperature log will show a sharp break at this point and this is
often misconstrued as the fracture bottom when, in fact, this is only indicating stagnant fluid in the
rathole. If the fracture has grown downward, the fluid outside the wellbore will be cooler than that
inside the rathole and post-fracture cooling below the perforated interval may be observed, result-
ing in a temperature cross-over as shown in Fig. 11.22. This is a clear indication of downward
fracture growth and the point of cross-over is interpreted as the fracture bottom.
Gamma-Ray Logs
Gamma-ray logs are another common method of measuring fracture height. These are conducted
by inducing artificial radioactivity in the fracture by including tagged proppant or tagged liquid in
the normal fracturing proppant or fluid, followed by post-treatment gamma-ray logs. One advan-
tage of gamma-ray logs over temperature logs is that they need not be run immediately after
stimulation, allowing wellbore fill below perforations to be cleaned out before logging. The other
restrictions on temperature logs, however, apply equally to radioactive logs, i.e., they are shal-
Fig. 11.21 - Example of Temperature Profile Changing with Time, Later Log Showing a Clearer
Interpretation.
low investigative tools (shallower even than temperature logs) and the response is proportional to
fracture width. Thus, while the two logs are often used in combination, the potential exists for them
to confirm one another and still not yield totally reliable results.
Procedure. Radioactive materials are added to the fracturing slurry stream with proper injection
and metering equipment supplied by the radioactive tracer company. Both high-pressure and low-
pressure equipment is available for adding the radioactive material either upstream or downstream
of the high pressure pumps. As mentioned earlier, tracers can be added either in a solid (proppant)
form or a liquid form. Zero-wash tracers, patented by ProTechnics, should be used whenever pos-
sible. This minimizes the residual radioactive material left in the wellbore and eases the interpre-
tation of post-frac gamma-ray logs. Table 11.12 lists the available types of tracers, their
recommended application, the more commonly used isotopes, mesh sizes for solid tracers, and
crush resistance limitations. Typically, for a fracturing treatment, proppant embedded with Sc-46
(Scandium), Sb-124 (Antimony), and/or Ir-192 (Iridium) are used as solid tracers. These same iso-
topes can also be used in liquid form. Typical tracer concentrations are 0.15-0.8 mCi per
1000 gals of fluid or pounds of proppant, depending on the gamma-ray tool size planned for use.
For a 1-11/16 in. tool, the recommended concentration is 0.35-0.8 mCi/1000 gals or lbs and for a
3-5/8 in. tool, it is 0.15-0.30 mCi/1000 gals or lbs. The types and concentrations of isotopes used
is dependent on the program objectives and the time before post-frac logs will be run, some iso-
topes having longer half-lifes than others, i.e., Au-198 (gold) - 3 days, Sb-124 - 60 days, Ir-192 -
74 days, and Sc-46 - 84 days. The tracer service company should be consulted to obtain recom-
mendations for a specific application.
A proper gamma-ray logging program for fracture treatment evaluation should always include a
pre-treatment log to identify naturally occurring isotopes in formation layers and to provide a base
log for comparison to the post-frac log. Comparing these two greatly enhances the interpretation
of post-frac logs. Spectral gamma-ray tools are available for use in distinquishing multiple iso-
tope tracing.
Application. Some of the more common applications of tracer logs are:
Minimum fracture height. Radioactive tracers can identify the minimum height of the
medium pumped - either hydraulic height (liquid tracers) or propped height (proppant tracers).
In many cases, this minimum height may be equal to or very close to the created height; but,
Table 11.12 - General Description and Properties of ProTechnics Patented ZERO WASH Tracer
Products.
SOLIDS
Custom irradiated High Volume Zero Wash products are available by special order.
LIQUIDS
Most tracers can be made both oil and water soluble. Specific chemical additives can be used
to balance the pH of the fluid to enhance adsorption on the formation or to reduce plating out
on tubulars.
MINI-TOOL (1-11/16)
Proppant settling. The effects of proppant settling to the lower part of the perforated
interval or out of the desired zone can be seen with radioactive logs.
Staging efficiency. In many cases, the need to separate a fracture treatment into multiple stages
is apparent from tracer analysis. There may be a larger stress or pore pressure contrast between
layers than assumed causing inefficient stimulation in a single stage treatment. Conversely,
multiple stage treatments may be determined to be unnecessary with tracer analysis. Fig. 11.24
shows an example where very little proppant was placed in the upper three sets of perforations
with the initial treatment and post-frac performance was disappointing. A second treatment
(refrac) was performed after isolating the bottom perforations with a bridge plug. As shown
from the second log, after the refrac which was tagged with a different isotope than used on the
first treatment, the upper sets of perforations were stimulated. Post-frac performance
doubled.
When Applicable to Run. Table 11.13 lists some of the more common circumstances under
which tracer logs might have an application in helping to define fracture treatment effectiveness.
Fig. 11.23 - Example Gamma-Ray Log Where Three Isotopes Were Used to Tag Proppant - Sb-124
(1-3 ppg), Sc-46 (4-6 ppg), and Ir-192 (6 ppg). Note Bottom Zone - Lower Concentrations Only, with
High Concentration in Top Zone Only.
Fig. 11.24 - Example Gamma-Ray Logs Where (1) the Initial Fracture Treatment Only Propped the
Bottom Zone and (2) Successful Propping of the Upper Zones with a Second Treatment.
Table 11.13 - Some Criteria for Tracer Addition and Gamma-Ray Logging.
If Stress Contrast Between Zone and Barrier is < an Amount That Might Allow Growth Out
of Zone.
If the Permeability Varies by a Factor of Some Significant Percentage or More Per Stage.
If Stress Contrast is > Some Significant Pressure Psi Between Perforated Layers.
Slurry
Cost Vol Net Average Fluid
M$ (mhsld) Pressure Width Efficiency FCD FOI
xf = 250
xf = 500 *
xf = 750
H = 50'
H = 100' *
H = 150'
C = .005
C = .0015 *
C = .0025
Q = 20
Q = 35 *
Q = 40
E=4
E=8 *
E = 12
Abstract
The Workshop No 2. was drilled as an infill producing well to a deeper horizon. Repeat Formation
Test (RFT) pressure data and dipmeter log, however, indicate the planned completion interval is
in a fault block already being depleted by at least two offset wells. As a result, plans are being made
to complete a porous and permeable gas sand at 5300 ft. This sand has been seen in a number of
wells in the area and appears to be fairly continuous over at least an entire section. Because the
sand is in a known regulatory horizon, it will be completed as a development well.
Purpose
This problem illustrates the benefits associated with the application of fracture pressure analysis
techniques to the design and optimization of fracture stimulations. The techniques utilized include
the analysis of closure stress, step rate, flowback, and minifrac tests. Analysis of this data will be
used to develop an understanding of the fracture characteristics. This will be done through building
an ULTRAFRAC data file and history matching minifrac data. Once the fracture is characterized,
ULTRAFRAC will be run in DESIGN Mode to determine the optimum treatment design for the
new horizon in the well in question.
Description
A calculated log section over the potential completion shown in Fig. P.1 highlights the pay sand.
In addition, a Long-Spaced-Digital-Sonic (LSDS) log over the section showed the sand to have a
compressional wave sonic travel time of approximately 65 micro-seconds per foot. Fig. P.2 shows
a plot of acoustic travel time vs. Youngs Modulus, E, for various rock types. Use this figure to
formulate initial estimates. The maximum bottomhole temperature recorded over the sand interval
was 170 F.
Reservoir pressure in the sand is unknown, though, the deeper target sands have been found to be
normally pressured (e.g., 0.433 psi/ft of depth). The interval in question was drilled with 9.5 lb/gal
mud with no significant gas shows. Assuming a normal pressure gradient indicates a reservoir pore
pressure of 2295 psi at 5300 ft. The well was completed in 5.5 inch casing with 2 7/8 inch tubing
landed at 5130 ft. The well was perforated with 2 shots/ft and produced for 24 hours at a rate of
520 mscfd. The production stream was dry gas. A build up test was run but bottomhole pressure
gauge failure precluded the build up interpretation. Surface pressure data indicated fairly stable
wellhead pressure of 1200 psi.
Some core data exists from a nearby exploration well which indicated the interval in question was
a clean sandstone with a porosity between 11.2 and 14.7%. Core permeability tests showed perme-
ability to air (no confining stress) of 0.44 to 0.55 md. Generally, core permeability is reduced by a
factor of approximately 5 to represent in-situ gas permeability.
Sand
Dolomite
Lime
E x 106 - psi
Production plans include compression which will allow a surface producing pressure of 100 psi
with a flowing bottomhole pressure of approximately 450 psi. The Gas Marketing Business Unit
indicates that it will cost nearly 0.15 $/mcf to transport the gas to the delivery point. Annual oper-
ating costs are $6,000 per well in the area.
Because of the limited rock property data a series of injection flowback tests were conducted to
determine formation closure pressure. During these tests, bottomhole treatment pressure was mea-
sured with a gauge set at 5375 ft. The closure stress tests were conducted with 2% KCl water with
friction reducer. This data is used to find the fracture closure stress, and possibly to give an esti-
mate of reservoir pressure.
Reservoir fluid data assuming a gas gravity of 0.65, a reservoir pressure of 2295 psi, and 170
degrees is calculated in the fluid property section of ULTRAFRAC to be:
Z-Factor: 0.85696
A. Pumpin/Shutin Test
An injection/shutin pressure decline test conducted by injecting 25 bbl of KCl water at 10 bpm
(tp = 2.5 minutes) and then recording the pressure decline as shown in Fig. P.3. Fig. P.4 shows
a plot of pressure vs. horner log time extrapolated to give a rough approximation of reservoir
pressure.
Injection/Decline Test
4250.000
4000.000
3750.000
Pws psia
3500.000
Closure Pressure = psi
3250.000
3000.000
2750.000
0.0000 0.5000 1.000 1.500 2.000 2.500 3.000 3.500 4.000 4.500 5.00
4000.000
Pws psia
3750.000
3500.000
3250.000
Start of Pseudo-Radial Flow Effects
3000.000
A step rate injection test was then performed immediately following the pressure decline.
the step rate test consisted of pumping KCl water at rates of 0.5, 1, 2, 3,4, 5, 7, 10, 12, and
15 bpm for two minutes at each rate. The rates and pressures at the end of each two minute
step are shown in Fig. P.5.
4500
4000
Bottomhole Press (psig)
3500
2500
2000
0 5 10 15 20
Inj Rate (bpm)
C. Pumpin/Flowback Test
At the end of the step rate test the rate was increased to 17 bpm, the well was then flowed back
at a rate of 2 bpm. The resulting pressure decline is shown in Fig. P.6.
D. Minifrac Test
A gel minifrac was pumped by injecting 38,000 gallons of 40 lb/1000 gals crosslinked frac flu-
id down tubing at an average rate of 35 bpm (injection time approximately 25.5 minutes), and
then flushing the well with 2% KCl water. Analysis of the pumping decline data is used to cal-
culate the fluid loss coefficient. The minifrac data both injection and pressure decline are
shown in Fig. P.7. A plot of net pressure versus time for the minifrac injection period is shown
in Fig. P.8.
4000.000
Pressure psia
3800.000
3600.000
3400.000
3200.000
3000.000
30.000 32.000 34.000 36.000 38.000 40.000 42.000 44.000 46.000 48.000 50.000
Clock Time (minutes)
Procedure:
Step 1: Update economic and reservoir data.
Step 2: Develop geomechanical input from porosity and sonic log data.
Step 3: Evaluate injection/decline, step rate test, and pumpin flowback test data for closure
pressure.
Step 5: Evaluate minifrac test for fluid efficiency and leakoff coefficient.
Step 7: Enter Framode in Analysis and enter minifrac volume of 38,000 gallons.
Step 8: Save file.
Step 9: Enter Quick Worksheet and history match net pressure, P*, and efficiency by alter-
ing fluid loss coefficient and Youngs Modulus.
Step 10: Once history match obtained with Quick Worksheet, execute fracture model and
match net pressure data.
Step 11: Once matched, change FRACMODE to design and conduct an optimization study.
Note both optimum length and conductivity.
Workshop Problem 3
Understanding the reservoir response to fracturing through the evaluation of a Tight Gas Well.
Abstract
The Nowayds No. 1 was drilled as a 640 acre location in the East Texas Cotton Valley Field. Con-
ventional fracture stimulations in the Cotton Valley have consisted of 415,000 gallons of 40 lb
crosslinked frac fluid and 1 MMlb of 20/40 sand. A service company representative showed up one
Monday morning recommending a significantly larger fracture treatment using carbo prop plus.
This problem illustrates the importance of understanding the reservoir response to hydraulic frac-
turing (i.e., the importance of FCD). The problem also highlights the benefits of optimizing fracture
treatments in the East Texas Cotton Valley.
Description
The Cotton Valley Sand Formation is upper Jurassic in age and is bounded by the Bossier Shale
below and the Travis Peak Formation above. The formation is approximately 1,400 ft thick and is
typically found at depths ranging from 8,000 to 10,800 ft and covers nearly all of Panola, Rusk,
and Harrison Counties in East Texas. The Cotton Valley is basically a transgressive-regressive
marine sequence. The Taylor Zone, the lowermost sand member in the Cotton Valley is an offshore
bar-shoreface transition and consists of a series of small bars and shales. The Taylor Sand is nearly
250 ft in gross thickness in the Nowayds No. 1. Above the Taylor interval lies 200 ft of shale which
generally confines a 2,500 ft fracture.
The Taylor sand, in the Cotton Valley trend, has an average permeability of approximately
0.005 md. Buildup tests in the vicinity of the Nowayds No. 1 indicated an average reservoir pres-
sure of 4,300 psi. Production from the Taylor is a dry gas at about 265F. Table P.1 summarizes
treatment and design data important to this analysis, while Table P.2 shows a conventional pump
schedule used in this area. Average porosity and water saturation for the Taylor Sand is 6% and
55%, respectively. Fig. P.1 shows a log section of the Nowayds No. 1. Fig. P.2 shows plots of net
pressure during a fracture treatment performed in offset wells using the conventional treatment
design. Fig. P.3 shows a dimensionless closure time to injection time plot to be used in this analysis
to determine fluid efficiency.
Objective
Your job is to compare the conventional Cotton Valley treatment design (Table P.1.) to the service
company recommendation shown in Table P.3. Develop an ULTRAFRAC data set and evaluate
the two designs in the ANALYSIS Mode. Compare and contrast the economic results of the two
treatments. How do these results compare to your optimum design (DESIGN Mode).
Procedure:
Step 1: Using the default file in the ULTRAFRAC database, Table P.1 and Figure P.1,
update the geomechanical data, calculate leakoff, set fracmode to analysis and input
conventional schedule.
Step 6: Modify leakoff and rerun trying to match efficiency, repeat until matched.
Step 13: Run and determine optimum (optional) fracture length and conductivity.
Table P.1 - Data for Example Well Analysis Table P.2 - Conventional/Schedule
Sand
Treatment Volume (MGALS) 360
Stage Volume lbs/gal
Treatment Rate (bpm) 75
Pad 72,000
Treatment Time (min.) 178
1 31,200 1
Closure Time (min.) 356
2 4,875 2
Fluid Type Crosslink
3 12,000 3
Reservoir Depth (ft) 10000
4 28,000 4
Temperature (F) 265
5 35,000 5
Permeability, (md) .005
6 40,000 6
Fracture Height (ft) 300
7 42,000 7
Fluid-Loss Height (ft) 280
8 50,000 8
Formation Modulus (106 psi) 8
9 45,000 9
Proppant Concentration (lbs/gal) 9
Workshop Problem 4
Application of fracture optimization to a multilayer reservoir, North Cowden Unit, Ector County,
Texas.
Abstract
Fracture treatment optimization techniques have been developed using Long-Spaced-Digital-
Sonic (LSDS) logs, pumpin-flowback, pumpin-shutin, minifrac, and downhole treating pressure
data. These analysis techniques have been successfully applied to massive hydraulic fracturing
(MHF) of tight gas wells and short highly conductive fractures in moderate permeability reservoirs
alike.
Purpose
This problem illustrates the application of fracture analysis techniques to a moderate permeability
reservoir. These techniques will be used to develop an ULTRAFRAC data set and identify large
zonal variations in rock properties and pore pressure which result from the complex carbonate
geology. The inclusion of geologic factors in fracture treatment design and their resulting effects
on fracture geometry will be highlighted.
Geologic Setting
The North Cowden Unit produces from the Permian Age Grayburg Formation at an average depth
of 4,300 ft. The Grayburg, which is bounded by the Queen and San Andres Formations, consists
of varying percentages of dolomite, anhydrite, and sand as shown in Figure P.11. Gross pay thick-
ness is 450 ft with a net pay thickness of approximately 200 ft. The average porosity is 9% and
permeability ranges from 0.1 to 50 md with an average of 2 md the average water saturation is
50%. The Grayburg Formation is informally subdivided into nine zones. Zones that are primarily
sand are identified as S-1 through S-3. Figure P.12 shows a CNL/FDC typelog of the Grayburg
sequence with an additional S-4 interval, which is present in portions of the field.
The Grayburg Formation exhibits a great diversity of carbonate lithologies ranging from supratidal
mudstones to fusulinid wackestones representing subtidal conditions. Interfingered with these car-
bonates are terrigeneous felspathic quartzarenites. The carbonates and sands have been extensively
dolomitized with all sediments having varying degrees of anhydrite infilling and/or anhydrite pore
filling and cementation. Although there is a lack of any distinct trend, localized areas of relatively
high pore volume and flow capacity do exist and generally correspond to sand development areas,
most notable the S-2 horizon as shown in the velocity profile, Figure P.13.
Description
The North Cowden Unit was in the early stages of an extensive infield development program
designed to densify from 80 acre to 40 acre development. One hundred and fifty infield producing
wells have been identified for drilling at a capital investment of $250 M per well with an additional
$50,000 per well to complete. The North Cowden Unit budget for 1992 totals $15 MM (i.e., drill
and complete 50 wells).
The production engineer, while reviewing historical NCU performance, found that NCU producers
did not perform as well as their offset competitors. It was also determined that treatments in the
Unit consistently exhibited increasing net pressure during stimulations (see Figure P.14) which
was inconsistent with the presumed radial fracture propagation. An ivnestigation of NCU fracture
treatment designs showed that, historically, 30,000 gallons of 30# crosslinked gelled water and
30,000 lbs of 8/12 mesh sand were pumped down 2-7/8 inch tubing at rates of 15 bpm as shown
in Table P.4. Prefrac production rates average 150 BOPD x 150 BWPD with a 6% production
decline. The economic limit is 5 BOPD.
Additional data available includes a long-spaced-digital-sonic log from a nearby well in the field
which was used to determine Youngs Modulus and Poissons Ratio on a foot by foot basis as
shown in Figure P.15. Table P.5 shows the comparison of LSDS derived rock properties to phys-
ical core measurements.
Further in-situ test data available includes closure test data from the S-2 and D-2 horizons. Anal-
ysis of these test are included as Figure P.16 and Figure P.17, respectively. Figure P.18 shows a
dimensionless pressure match of post-frac pressure decline data which indicates a D-5 horizon
fluid leakoff coefficient of 0.00025. The sand intervals in the unit should be expected to have an
even higher fluid loss coefficient.
Also shown for completeness (Figure P.19 - Figure P.21) are net pressure plots of stimulations
conducted in the D-2, S-2, and D-5 intervals, respectively. The production engineer was charged
with evaluating this treatment schedule to ensure that it was the optimum treatment design to max-
imize the present value of the infield development program.
Abstract
The NS25 was drilled and completed as a horizontal well. Production from the 5000 ft horizontal
section averages 500 BOPD well below expectations. In addition, the well produced slugs of chalk
(formation). Wellbore stability is reasoned to be the cause for the under performance of the lateral
section. Therefore, to minimize both proppant or formation flowback problems resin coated prop-
pants are used in fracture stimulations. To aid in the design and execution of the fracture stimula-
tion, a minifrac was conducted utilizing 56,000 gallons of a water based crosslinked borate system
pumped at 35 BPM which is the maximum achievable rate from the stimulation vessel.
This problem illustrates the benefits of utilizing minifrac data to design treatments at the well site.
This will be accomplished by first analyzing the minifrac data and then utilizing design techniques
captured in an EXCEL spreadsheet program to design and implement a treatment.
Description
The NS25 produces from the TOR Formation as shown in Figure 1. The Tor, as shown, is some
75 ft thick in the NS25. Rock properties testing has indicated that the chalk formation has a
Youngs modulus of 500,000 psi and a poissons ratio of 0.3. Because of the softness of the chalk,
the resistance to the creation of a fracture is great and as a result, fracture toughness is on the order
of 15,000. The reservoir is normally pressured and has a permeability of 10 md and a porosity of
30%.
The stimulation vessel from which the fracture stimulation was to be performed was fully loaded
with 1 million pounds of resin coated proppant and 329,000 gallons of 30# crosslinked borate frac-
turing fluid. In additional to conventional land based materials costs an additional $300,000 boat
service charge is required to stimulate this well.
Objective:
Your job is to utilize the results of the minifrac test to optimally place all of the fluid and proppant
loaded on the boat. Secondly, test this design with ULTRAFRAC and determine the fracture length
and conductivity of the fracture stimulation. Also note the in-situ pounds per square foot of prop-
pant in the fracture. Finally, utilize this information to determine the post-frac production rate and
payout of the stimulation.
Procedure:
Step 1: Interpret the minifrac test and determine fluid efficiency, closure pressure, and clo-
sure time.
Step 2: Build an ULTRAFRAC file by first adding the boat service charge to miscellaneous
costs in the economic section.
Step 4: Estimate leakoff and check with respect to the minifrac analysis.
Step 5: Simulate minifrac and match final pressure and fluid efficiency
Step 6: In design mode, determine the optimum fracture length and conductivity.
Step 7: In analysis mode, select optimum design schedule and execute fracture simulation
to determine length, conductivity, and in-situ pounds per square foot.
Step 8: Use Quick Worksheet or Prats curve to estimate Post-Frac Folds of Increase, FOI.
A candidate well for the first stimulation has been selected. The well was shut in and the pressure
falloff measured; three closure stress tests were run; then 200 bbl of fracturing fluid were injected
down tubing with the surface annulus pressure measured during injection and during the pressure
decline after injection (Note: Due to possible problems with old casing, all the stimulations must
be pumped down tubing.) The data from these tests, along with some open hole logs are included
as attachments.
Also included is the present worth of increased injection. This was developed based on model runs
for various fracture lengths and conductivities. The chart is the present worth of increased oil pro-
duction only, and does not include the fracturing costs. Based on service company price books,
stimulation costs should be about $1.00/gal for fluid, $0.08/lb for sand, and $2500.00 for mileage,
workover rig time, etc.
Your assignment - should you decide to accept it - is to design a stimulation program, in sufficient
detail to allow field execution. Since there will be no time to evaluate long-term effectiveness of
different procedures, the job should be designed to maximize the discounted return on investment
(DROI). Also, fracture length should not interfere with reservoir sweep, since hydraulic fracture
azimuth is totally unknown in this field.
BHST = 100F
Test Parameters -
0.5 400
1.0 403
2.0 418
3.0 425
4.0 430
5.0 435
6.0 438
7.0 443
8.0 449
10.0 458
12.0 463
15.0 468
17.0 471
20.0 479
22.0 480
24.0 482
25.0 483
Shut-in at 25.0 minutes
26.0 1 1 434
27.0 2 1.41 406
28.0 3 1.73 392
29.0 4 2.00 380
30.0 5 2.24 369
31.0 6 2.45 359
32.0 7 2.65 350
33.0 8 2.83 341
34.0 9 3.00 332
35.0 10 3.16 324
37.0 12 3.46 307
39.0 14 3.74 290
41.0 16 4.00 274
43.0 18 4.24 258
280
250
250
Prepare: Optimum Fracture Design with sufficient detail that an Experienced Field Fore-
man could execute the treatment with the results you expected.
The cost of drilling and setting casing on this well is estimated to be $5 MM. Based on production
from other ells in the field, the unstimulated, damaged PW(15) is estimated to be $2.5 MM. Des-
perate Energy management wants you, the engineer, to evaluate data and make a recommendation
whether the well should be drilled and what the optimum fracture design would be. Conventional
fracture treatment designs, provided by the service company for other wells in this field, have been
unsuccessful in obtaining economic production rates. Unfortunately, the records for these treat-
ments are not available. The wellbore configuration normally used in this field is shown in
Attachment 1. Data from pre-frac tests performed on the offset are included in Attachments 4 and
6.
2. Due to proppant embedment in the soft chalk, an in-situ proppant concentration of 2 lbs/sq ft
must be achieved.
3. If a fracture treatment is performed, the cost for doing the treatment will be $3.50/slurry gal for
gel Ottawa sand or $.25/slurry gal for gel + intermediate proppant. Equipment and pumping
charges are figures in as part of the material costs.
Parameters -
rw = 0.40 (ft)
Test Data -
Flow Time - 6 (hrs) , Rate = 582 BOPD
Final Flowing Pressure (BHFP) = 2334 (psia)
Using the plot and the following reservoir properties, find fracture conductivity and length, Fcd,
and steady-state folds of increase (FOI) resulting from this stimulation.
Reservoir Properties -
Calculations -
Fcd =
Steady-State FOI =
Based on extrapolating from 3 months production, it has been determined that the PW(15) of the
increased production resulting from the stimulations is $390,500. The cost of the fracture job was
$35,500. Calculate the discounted return on investment for this stimulation.
DROI =
Fracture design calculations were done based on analysis of post-frac pressure data from the first
well. These were used to develop a price table for changes in the stimulation design:
(Costs in $1,000.00)
xf = kfw 0.5 * base base case 2 * base
0.5 * base 24 34 48
base case 25 35.5 51.5
2 * base 27 39 59
Using these prices and steady-state calculations for folds of increase (FOI), calculate the DROI for
the various stimulation designs. Should a change in design be recommended? Should more cases
be considered?
DROI
xf = kfw 0.5 * base base case 2 * base
0.5 * base
base case
2 * base
Halliburtons 6-52
Western Company 6-53
D
Cold water circulation temperature surveys 10-8 Delayed crosslinked systems 6-8
Compatibility Density log 10-5
formation, formation fluids and chemical additives 6-6 Depth 5-40
safety and environmental 6-5 Design package, integrated 6-23
Compression test 4-2, 4-3 Desired fracture half-lengths 1-12
Computer control console 1-9 Devonian shale 6-11
Concentrates, polymer 6-30 Diesel fluid loss additive 6-14
Concentration, effective polymer 6-20 Dimensionless fracture
Conditions capacity (FCD) 3-1
dynamic 4-2 conductivity (FCD) 1-12
quasi static 4-2 Disadvantages
Conductivity of proppant 7-19 foamed frac fluids 6-46
Consistency Index 5-29 gelled hydrocarbons 6-48
Constant methanol gels 6-49
formation face polymer emulsion fluid 6-40
pressure 3-29, 3-42, 3-44 Discounted return on investment (DROI) 9-3, 9-8
rate 3-28, 3-41 Dowell Schlumberger coding system 6-53
internal phase 6-40 Downhole
Continuity equation 2-1, 8-17, 8-19, 8-25 flow, turbulent conditions 6-51
Continuous television 10-7
mixed fluid systems 6-8 Drag
proportioner 1-8 coefficient, particle 6-12
Controlling fracture height 5-2 reducing 6-9
Core DROI (discounted return on investment) 9-3
bulging 4-2, 4-4 Droplet-size 6-41
flow tests 6-6 Dynamic
tests 10-3 conditions 4-2
Cost(s) fluid loss 6-18
pumping 6-23 moduli 4-2
pumping hp 6-11 modulus 4-5
Criteria, fluid selection 6-3
Critical E
concentration 6-9 Economics 6-23
pressure 8-1, 8-20 Effect of
strain energy release rate 4-7 flow restrictions 3-37
stress intensity factor 4-7 wellbore storage 3-37
Cross reference of similar additives 6-54 Effective
Crosslinked particle shear rate 6-12
aluminum orthophosphate esters 6-47 porosity 5-21
delayed fluids 6-38 wellbore radius (rw) 3-10
delayed systems 6-8 Elastic modulus 10-3
dual functionality 6-34 Elasticity equation 2-1
gels 6-14 Emulsifying 6-8
hydrocarbon 6-2 Emulsions 6-7
ideal delayed fluid 6-38 Encapsulated
polymer solutions (gels) 6-1 breaker(s) 6-7, 6-20
Crosslinking Environmental and safety compatibility 6-5
agents 1-5 Enzyme breakers 6-7, 6-20
fast 6-32 Equation
fast, water-base gels 6-32 continuity 2-1
Crushable breakers 6-20 elasticity 2-1
fluid flow 2-1
Equilibrium bank(s) 6-11
F Foam(s) 6-15
aqueous 6-1
Fatty-acid soaps 6-47 binary 6-43, 6-46
FCD CO2 6-9
dimensionless fracture capacity 3-1 friction pressure data 6-43
dimensionless fracture conductivity 1-12 hydrocarbon 6-2
FDROI (fracture discounted return on investment) 9-3 hydrocarbon-base 6-23
Filtrate 5-20 nitrogen 6-9
FINCPV (fracture incremental present worth or value) 9-3 viscosity data 6-43
Finite capacity 3-4 texture 6-43
Flash points 6-5 Foamed frac fluids
Flow advantages 6-41
back and cleanup 6-18 disadvantages 6-46
behavior index 5-29 Foaming potential 6-8
Fluid loss 6-14 FOI (folds of increase) 1-12, 3-5, 3-10
addtives 5-23 Folds of increase (FOI) 1-12, 3-5, 3-10
coefficient 5-20, 10-3 Formation
foams 6-15 elastic properties 4-1
oil base gels 6-15 fluid 5-22
rate 8-27 linear flow 3-27
test 5-22 permeability 3-1
Fluid(s) wettability of 6-6
affected by fluid flow loss 6-18 Frac Height
aqueous 1-4 variables affecting 5-15
bingham plastic 5-29 Fracture 5-36
classification 6-1 closure pressure 5-11
crosslinked delayed 6-38 closure stress 8-4
degradation 5-39 determining fluid efficiency 8-58
description of fracturing types 6-30 discounted return on investment (FDROI) 9-3
dynamic loss 6-18 early design 1-8
efficiency 5-37, 8-36, 8-44, 8-55 extension pressure 8-4
flow equation 2-1 flow capacity 3-1, 3-2, 3-3
foamed frac 6-41 geometry 5-16
hydrocarbon-base 6-23 half-length 3-2, 5-36
ideal delayed crosslinked 6-38 height 5-1, 5-3, 5-15, 5-16
low loss 6-14 controlling 5-2
napalm-type 6-47 growth 5-1, 5-3
optimal scheduling for 6-70 incremental present worth or value (FINCPV) 9-3
polymer emulsion 6-40 initial height 5-3
power law 5-29 length 3-1, 5-16
pressure calculating 5-15 linear flow 3-27
rheological testing of fracturing 6-49 orientation 1-3, 3-8
scheduling 6-70 radius 5-36
scheduling given the fluid rheology 6-70 stiffness 8-26
scheduling using contained rheology 6-71 stimulation
selection 6-1 critical factors to optimum 1-11
selection criteria 6-3 design 1-12
viscosities, proppant transport 6-11 toughness 4-7
viscosity 5-27 treatment 1-6
volume 5-37 design 5-15
Fluid-element width 5-15, 5-16
exposure time 6-70 Fracturing
rheology 6-70 effect of modulus on 4-4
time at temperature vs. volume pumped 6-72 fluid(s) 1-4
Fluidized layer of sand 6-12 compatibility with its additives 6-7
Fluorocarbon, surfactants 6-7
W
Wall building 5-23
effect 5-22
Water
blocks 6-7
slick 6-1, 6-11
soluble polymers, natural 6-30
Water-base