You are on page 1of 11

SPE 71074

Friction Pressure Correlation for Guar-Based Hydraulic Fracturing Fluids


Vibhas J. Pandey, SPE, Schlumberger Oilfield Services

Copyright 2001, Society of Petroleum Engineers Inc.


This paper was prepared for presentation at the SPE Rocky Mountain Petroleum Technology
Conference held in Keystone, Colorado, 2123 May 2001.
This paper was selected for presentation by an SPE Program Committee following review of
information contained in an abstract submitted by the author(s). Contents of the paper, as
presented, have not been reviewed by the Society of Petroleum Engineers and are subject to
correction by the author(s). The material, as presented, does not necessarily reflect any
position of the Society of Petroleum Engineers, its officers, or members. Papers presented at
SPE meetings are subject to publication review by Editorial Committees of the Society of
Petroleum Engineers. Electronic reproduction, distribution, or storage of any part of this paper
for commercial purposes without the written consent of the Society of Petroleum Engineers is
prohibited. Permission to reproduce in print is restricted to an abstract of not more than 300
words; illustrations may not be copied. The abstract must contain conspicuous
acknowledgment of where and by whom the paper was presented. Write Librarian, SPE, P.O.
Box 833836, Richardson, TX 75083-3836, U.S.A., fax 01-972-952-9435.

Abstract
Accurate prediction of tubular frictional pressure is an
important aspect in design, planning and execution of any
hydraulic fracturing operation. Majority of correlations
available to predict the tubular friction pressures are based on
laboratory generated data. The results from such tests have to
be scaled up to predict real time solutions. A novel way to
develop such correlation based on actual field data is
discussed in this paper. Real time data was obtained for over
300 hydraulic fracturing jobs that were pumped using the
Carboxymethylhydroxypropyl guar (CMHPG) fluid systems.
This data was analyzed using Fanning friction factor and
generalized Reynolds number. Both of these dimensionless
entities were modified to accommodate the effects induced by
polymeric fluids. Results obtained from the new correlation
were able to predict the average surface pressures within 4%
of the measured values for over 90% of the data. These results
were also compared with a few existing correlations and
encouraging conclusions were drawn. Pressure predictions
from the new correlation were routinely verified in the field
and it was observed that predicted pressures matched the
actual in a reasonable range of accuracy.
Introduction
Prediction of accurate surface pressure for a hydraulic
fracturing treatment is important from planning and execution
standpoint. The effect of tubular friction pressure is more
critical when a fracturing job has to be designed for relatively
smaller diameter tubing on a deep well. These pressures will
effect the parameters like selection of hardware required for
the job (e.g. tree-saver), hydraulic horsepower, maximum

permissible pump rate, and number of pumping units. Friction


pressures play a major role in determination of optimum pump
rate since the pressure rating of tubular cannot be exceeded.
Limitations on the rate affect the fracture geometry because of
the limits imposed on the bottomhole pressure. During the
execution phase, accurate estimation of tubular friction
pressure aids in better estimation of bottomhole pressure
trends, thus allowing on location engineers to have a better
idea of fracture propagation as the treatment progresses.
The purpose of this study was to develop friction
pressure correlations based on available surface pressure data
from the field. The study analyzed the data that was gathered
for over 300 hydraulic fracturing jobs pumped using delayedcross linking, CMHPG based gels, collected over a period of
one year. The data encompasses gel concentrations ranging
from 30 lbm/Mgal to 60 lbm/Mgal for tubular sizes ranging
from 2-3/8 inch to 4-1/2 inch.
Literature Review
A great number of studies are available on estimating friction
pressures losses for polymer based pseudoplastic power fluids.
Most of these studies are based on experimental data
generated by simulating field conditions. A typical
experimental setup for most of these studies involves a gel
mixer to mix the required gel concentration and a positive
displacement triplex pump. Fluid is discharged from the
triplex pump onto a straight or coiled tubular of known
diameter and pressure drop is recorded across several intervals
along the pipe.
Fluid flow in Smooth Pipes
Most of the fracturing fluids that are being used by the
industry today are non-Newtonian pseudoplastic fluids. The
rheological behavior of such fluids is described by the powerlaw equation defined as

= K (& ) ........................................................................ (1)


n

where n' is dimensionless flow behavior index, and K is


n
2
consistency index in lb-s /ft . is the shear stress and & is the
shear rate, given by following equations in oilfield units

VIBHAS J. PANDEY

w = 0.0208

dp
.............................................................(2)
L

5174.78q 3n + 1
& =

...................................................(3)
d 3 4 n
where d denotes the tubular internal diameter in inches, p is
the pressure drop in psi, q is the flow rate in bbl/min, and L is
the length in feet, over which the pressure drop is measured.
For laminar flow of such fluids in smooth pipe, Metzner and
Reed1 suggested the following relation

f =

16
...........................................................................(4)

N RE

where, f is the Fanning friction factor defined as

f = 1238.63

w
.............................................................(5)
v 2

and NRE is the generalized Reynolds number defined as,

= 1.86
N RE

d n v (2 n )
.....................................................(6)
12 n 8 n K

in oilfield units. is the density of the fluid in lbm/gal, v is


the average fluid velocity in ft/sec, and d is the tubular internal
diameter in inches. n' and K' can either be obtained by
plotting the viscometer data on a log-log plot of shear stress
vs. shear rate or by using standard viscometer relations.
Viscometer consistency index, K'v can be converted to pipe
flow consistency index K'p as follows
n

3n + 1
K p = K v
..........................................................(7)
4n
Similar results can be obtained by using RabinowitchMooney2, 3 approach, based on which the frictional pressure
losses can be predicted by using the following general
equation

1 w
q
= 3 rx2 f& ( rx )d rx ................................................(8)
3
R
w 0
Eq. 8 can be used to predict frictional pressure losses for
laminar regime of Newtonian and non-Newtonian, time
independent fluids. For power-law fluids, Eq. 8 takes the
following form

SPE 71074

p 1293.7(3n + 1)q K
psi / 1000 ft .................. (9)
=

L
nd 3
3d
n

using standard oilfield notations.


Dodge and Metzner4 extended von Karmans5 work on
turbulent flow friction factor to include the power-law nonNewtonian fluid. An expression for the friction factor was
derived in terms of the generalized Reynolds number, NRE :

1
f

4.0
f
log N RE
(n )0.75

(1 n )2

] (n0.)4

1.2

......................... (10)

Dodge and Metzners correlations were based on the


experimental data gathered with CarbopolTM, a polymer that is
no longer used in the fracturing fluids.
Also, their
experimental data are very limited; no data exist for fluids
with n<0.4, and few data at higher Reynolds number
available. Further, this procedure requires trial-and-error
solutions for N'RE because the wall shear stress must be
evaluated from an unknown f.
Based on Dodge and Metzners classical approach, Shah6
developed correlations that have been used to predict the
friction pressures of fracturing gels. The general form of the
equation is given as follows

)
f = f + A(N RE

B (n )

..................................................... (11)

where f, A, and B are functions of n. Values of constants A


and B are provided in the literature6.
Fluid Flow in Rough Pipes
In case of rough pipes, the friction factor is a function of
Reynolds number and the relative roughness, h/d. Chen7
suggested an explicit correlation that relates friction factor,
pipe roughness parameter, pipe diameter, and Reynolds
number for transition and turbulent flow regimes. The relation
is given as

5.0452
h

3.7065d N

RE
1

= 4.0 log
....... (12)
1.1098

5.8506
f
log 1 h
+

2.8257 d
0.8981
N RE

The phenomenon of turbulent flow of non-Newtonian fluids in
rough pipes has not been explained well and the experimental
flow data are extremely rare. Dodge8 presented the following
equation

SPE 71074

FRICTION PRESSURE CORRELATION FOR GUAR-BASED HYDRAULIC FRACTURING FLUIDS

d
= n A log + B ......................................................(13)
2h
f

where A and B are unknown functions of n and require


experimental determination in pipes with various levels of
roughness. Based on the results of the study, it was suggested
that the effect of pipe roughness is independent of Reynolds
number. It is an established fact that the effect of pipe
roughness continues to diminish as the gel concentration
increases. This could be due to the fact that as the fluids get
more viscous, the viscous material embeds itself on the walls
of the pipe and forms an even surface.

where, j, A and s are empirically determined constants. Values


for j range from 0.0 to 0.266 for a variety of slurries and
polymer solutions flowing in tubes of different diameters11.
Eq. 17 can also be expressed in form of f and NRE. For
constant values of NRE, , n, and K'p, the effect of diameter
with respect to friction factor is shown as

n (n s )
+ n j

2 n

f d

................................................................. (18)

Depending on the values of n, j, and s, the friction factor


could either increase or decrease with diameter.

D.L. Lord and J.M. McGowan9, in 1986 proposed a laboratory


correlation that could be readily used for field applications.
This correlation relates tubing diameter, flow rate and
gel/proppant concentration to predict tubular frictional
pressure for polymer-laden fluids and also proppant-laden
slurries. Eq. 14 shows the relation

API adopted Eq. 17 for scaling up the turbulent flow of


fracturing fluids through tubing and tubing-casing annuli. API
RP 3912 recommends the addition of 0.036 to s (the smoothpipe flow-curve slope) to compensate for hydraulic roughness
in oilfield tubular goods. However, Shah13 in his study,
clearly demonstrated that the effect of roughness is not
independent of Reynolds number and addition of any numeric
value to s may not be adequate to define the fluid behavior.

ln(1 / ) = 2.38 8.024 / v 0.2365G / v

Drag Reduction of Polymeric Fluids

0.1639 ln G 0.028 Pe

1
G

......................(14)

where G is the gel concentration in lbm/Mgal, v is the


average tubular velocity ft/sec, P is the proppant concentration
in lbm/gal, and is the drag ratio defined as

= p G , P / p o ..................................................................(15)
where, po is the friction pressure of the Newtonian water
solvent, given in oilfield units by
p o = 0.40429d 4.8 q 1.8 L .....................................................(16)

Majority of fracturing fluids are drag reducing and nonNewtonian shear-thinning type. Drag reduction occurs when
small concentrations of polymer reduce friction pressure at a
given turbulent flow rate below that of the solvent alone. The
mechanism for drag reduction has not been clearly established,
although several ideas have been discussed. It is generally
theorized as somehow related to the viscoelasticity imparted to
the fluid by addition of the polymer. One parameter that
describes viscoelasticty is the relaxation time tp, which takes
into account the duration required for a polymer molecule to
regain its rest conformation after being deformed. Maxwells
14
simple viscoelastic model gives a possible relation for
relaxation time tp as follows

xx yy
N
= 1 .................................................... (19)
&
2 w
2 w&

where, d is the tubular diameter in inches, L is the length of


the tubular in feet, and q is the flow rate in bbl/min. This
correlation does not predict any dependence on diameter.

tp =

Turbulent Behavior of Fracturing Fluids

where, N1 is the measure of primary normal stress difference,


and w is the wall shear stress. Newtonian and viscous
inelastic liquids like power law fluids may invariantly yield
zero normal stress difference under similar flow conditions.
Another important dimensionless group that offers a
quantifiable indication of increase or decrease in drag
reduction is Deborah number. This number relates relaxation
time tp, with generalized Reynolds number NRE, based on the
15
relation proposed by Seyer and Metzner as follows

Most of the fluid flow pertaining to a typical hydraulic


fracturing job falls in the turbulent flow category. For such
flows, an empirical solution based on Bowens10 extension of
the Blasius equation has been shown to describe the turbulent
behavior of fluids, including Carboxymethylcellulose and guar
solutions. This relation takes the form
d (1+ j ) p f
4L

= A(8v ) .............................................................(17)
s

N De =

3
v
(N RE )4 t p ........................................................ (20)
d

VIBHAS J. PANDEY

Eq. 20 suggests that at constant N'RE and fluid velocity,


increasing the diameter reduces NDe15, 16. In other words, with
the increase of pipe diameter, the drag reduction decreases.
Similar phenomenon was observed when the analysis on
recorded data was carried out.
Drag reduction increases with the increase of polymer
concentration until a maximum value is reached. Based on his
study, Virk17 showed that the maximum drag reduction
(MDR) in turbulent pipe flow of dilute polymer solutions is
ultimately limited by a unique asymptote described by the
experimental correlation:

1
f

= 19.0 log10 N RE

f 32.4 ...................................(21)

In another study, Virk et al18 represented the MDR by an


asymptote on the curve of friction factor versus solution
Reynolds number as

f V = 0.42(N RE )s

0.55

.........................................................(22)

where (NRE)s is the Reynolds number based on the solvent


properties. It was observed that at concentrations higher than
the optimum, additional polymer will increase the viscosity
but may cause little additional increase in the relaxation time.
The net effect is a limited increase of Deborah number, and
hence a negligible increase of drag reduction.
Data Collection
Data collected for the study was obtained from the actual
hydraulic fracturing jobs carried out for several clients in
south Texas region. Due to their ability to remain stable at
higher temperature ranges for prolonged treatment times,
CMHPG based fluids are one of the most preferred fluid types
for carrying out hydraulic fracturing on the deep and hot wells
of south Texas.
Methodology
One of the prevalent ways to obtain frictional pressure data
during the fracturing treatment is to stop pumping
momentarily during the pad stage to obtain the instantaneous
shut in pressure (ISIP) at the wellhead. By adding the
hydrostatic pressure of the fluid column to the ISIP, the
bottomhole pressure can be calculated. If frictional pressures
through the perforations and fracture can be ignored, the
friction pressure in the tubular goods can be estimated from
the surface pressure, before the ISIP is taken, by the following
equation

p f = p s p bh + p H .....................................................(23)

SPE 71074

where, pf is the tubular friction pressure, pbh is the bottom hole


pressure, and pH is the hydrostatic pressure.
Stopping the pumps during the pad may not always be
possible and hence a slightly different approach was employed
for the study.
Essentially, three important parameters are
needed for the computation of average pressure drop in the
tubular goods. These are, mean perforation depth, ISIP, and
stabilized pad pressure. Fig. 1 shows a typical plot of
hydraulic fracturing job. In most of the cases before
commencing the treatment, original well fluid is displaced
with linear gel of known concentration, and pumping is halted
to record the ISIP. This is generally followed by an injection
test where a fixed volume of fluid is injected in the formation
at the design rate, and later the step rate test is carried out.
Once the results from the test are analyzed and corresponding
changes made to the design, the actual treatment in pumped.
In the region labeled Actual Job in Fig. 1, the landing of pad
on the perforations is marked by a distinct change in the
surface pressure trend, which in this case appears to be
decreasing with the onset of new fluid type. Such pressure
signatures were recorded in almost all jobs with a varying
degree. In some cases, the pressure appeared to level out
briefly before exhibiting substantial changes. This instant is
very important in the data collection procedure because of two
reasons, a) the tubing or casing is now loaded with a known
fluid, which is being pumped at a known rate, and b) not
enough net pressure to offset the results has been built at this
point. With only a few barrels of this fluid in the formation,
the surface pressure is now termed as stabilized pad pressure
and is recorded for every job. After deducting the last
observed ISIP from this value, the total frictional pressure is
obtained. Friction pressure gradient p/L, is obtained by
dividing the friction pressure by mid-perforation depth.
Attempts were not made to discern between tubular and
perforation friction at this point. Typically, the magnitude of
perforation-induced friction is very small compared to tubular
friction unless there is a case of massive near well bore
restriction due to hardware problem, scales or perforation
related problem. Fig. 2 shows the plot of computed friction
pressure versus the flow rate for 35-lbm/Mgal fluid for a
tubular diameter of 4-1/2 inch and 11.6 lb./ft. Since the data
was obtained from field, some scatter in the calculated results
is always expected. Factors that may affect the results include
perforation sizes and geometry, possible fluid viscosity
variations, temperature effects, and other reservoir related
parameters like near well bore depletion. Despite these
possibilities, it is clearly seen that majority of data seems to be
evenly distributed along the mean. One inference that can be
drawn from the distribution is the fact that the effects of near
well bore restrictions may be more pronounced in some wells
than others. The success of this procedure largely depends on

SPE 71074

FRICTION PRESSURE CORRELATION FOR GUAR-BASED HYDRAULIC FRACTURING FLUIDS

the volume of data obtained. Apparently, more number of


data points define the pressure drop trends more vividly.

diminishing drag reduction effect with the increase in


diameter.

Data Analysis

Modified Fanning Friction Factor, fm

Friction pressure data were collected for gel concentrations


ranging from 30 to 60 lbm/Mgal. Tubular outer diameters
varied from 2-3/8 to 4-1/2 inch. This data were the plotted in
plots similar to Fig. 2 and power-law type trend lines were
fitted on each data set. Fig. 3 shows one of such plots.

As suggested by Bowen in his earlier studies, the Fanning


friction factor in its current form may not be adequate to
define pressure drop in power-law fluids. It was seen from this
study that the friction factor for such fluids was predominantly
governed by the velocity and fluid behavior index n. It was
also found to be directly proportional to the tubular diameter,
but at a power lower than unity. When plotted with
generalized Reynolds number, the following proportionality of
Fanning friction factor appeared to be more applicable than
the one shown in Eq. 5

During an actual job, the polymeric gel is typically mixed on


the fly and the critical additives like the cross-linker and the
delaying agent are added as the fluid is being discharged from
the blender. It is customary to delay the cross-linking of the
CMHPGzirconate base cross linked fluid because of two
primary reasons. First, premature cross-linking on the surface
will cause an increase in surface pressures and may lead to a
shutdown. Secondly, once cross-linked, the fluids must not be
subjected to very high shear rates because they can cause
irreversible damage to the fluid stability in terms of their
proppant carrying ability. For the above reasons, these fluids
are usually delayed to a cross-linking temperature of around
95o F to 103o F, which could be close to the perforation
injection temperature for most of wells in south Texas.
For all practical purposes, the fluid continues to behave like a
linear gel until it reaches perforations, which is the point
where it should ideally begin cross-linking. This behavior
allows fluid rheology test on surface using Chan 35 or Fann 35
rheometer, without heating the sample. Table 1 shows the
rheological properties of the linear and cross-linked fluids. It
must be borne in mind that the properties mentioned under
cross-linked gel section are for the fluid where cross-linker
has been added but the system is not fully cross-linked
because the samples are not heated. This situation emulates
the field conditions where the additives are injected in the
stream but the fluid does not witness the cross-linking
temperatures on the surface, thus almost behaving like a linear
gel. Since the values given are averaged over a number of
repeatability checks, it is recommended to use them for
computing frictional pressure losses.
Friction Pressure Loss Correlation Development
Fig. 4 shows the plot of Fanning friction factor, f and
generalized Reynolds number, NRE for 35 lbm/Mgal for
available data for 2-7/8, 3-1/2, and 4-1/2 inch casing. As can
be seen from the plot, f and N'RE alone are not sufficient to
define the relationship. Data points representing individual
tubular diameters appear to display unique trends. Such
behavior was observed for the data sets pertaining to all the
gel concentrations under the study. It was also observed from
such plots that with the increase of the tubular internal
diameter the curves shifted to the higher side, indicating a

10

d n
v

........................................................................ (24)

Thus, for 0.38<n'<0.51 which forms the majority of the fluids


under the study, the tubular diameter would directly affect the
friction factor by a power range of 0.144 to 0.261, whereas the
velocity would inversely effect the same in a range of 3.24 to
2.8. This means that, for a constant generalized Reynolds
number, when compared to the original proportionality, the
effect of diameter on the friction factor diminishes by a factor
of nearly 4 to 7, whereas the effect of velocity increases by a
factor of 1.4 to 1.62. Mathematically this would cause the
effective value of calculated friction factor to decrease, but
would also cause the friction factors corresponding to lower
diameters and higher velocities, to increase. The net effect
would cause merging of all the data points in one curve. On
an individual basis, it is possible to normalize and hence
render the proportionality shown in Eq. 24 as dimensionless.
However, such a group may have little physical significance.
On the contrary, since the proportionality seemed to fit very
well in all the scenarios tried, it indeed appears to be
indicating some phenomenon that needs to be explored in
future studies.
Fanning friction factor f was now modified to incorporate the
diameter and velocity effects, and the following relation gave
the final form of equation,

fm =

0.0257632d n

......................................... (25)
L

where, fm is the modified form of Fanning friction factor, v is


in ft/sec, d is in inches, p/L is in psi/1000 ft and n is
dimensionless. Since it is possible to normalize the modified

VIBHAS J. PANDEY

entities, the end product given by fm still remains


dimensionless.
Fig. 5 shows the plot of square root of modified Fanning
friction factor versus generalized Reynolds number for 35lbm/Mgal fluid. Note that the scatter seen earlier, is now
represented by a singular curve. Thus, for one fluid type,
characterized by fluid behavior index n, a singular curve
could define the relation between tubular friction pressure
drop, average flow velocity, and tubular internal diameter.
Similar curves were plotted for other gel concentrations under
the study. Fig. 6 shows the plot for all gel concentrations
when plotted for square root of modified friction factor versus
generalized Reynolds number.
Modification of generalized Reynolds number, NREm
Plot of Fig. 6 indicates that some form of power law equation
can express the individual curves pertaining to various gel
concentrations. These equations can be grouped together and
coefficients corresponding to each correlation can be
presented in a tabular form to aid in friction pressure drop
calculations. However, it is a known fact that the usage of
coefficients often introduces computational difficulties and
makes the solution extremely cumbersome. To eliminate such
problems, further attempts were made to streamline the
process of calculating frictional pressure losses.
The slope of the curves in Fig. 6 is primarily dependent on the
fluid behavior index n, whereas the intercept is dependent on
the consistency index Kp, which is a function of n and the
apparent viscosity at a particular shear rate. Whereas the slope
of all the curves appears to be identical, the shift in the
intercept was because of the changing gel concentrations and
apparent viscosity. It was observed that with the introduction
of an additional parameter that linked the gel concentration,
fluid density, and the flow behavior index, it was possible to
define all the data points in one curve. This parameter was
defined as follows

Cp
1 =

1

n

................................................................(26)

where is the density of the fluid in lbm/gal, and Cp is the


polymer gel concentration in lbm/Mgal. Reynolds number
was now modified as

=
N REm

1.86d n v (2 n )
....................................................(27)
12 n 8 n K p 1

Plot of Fig. 7 shows the distribution of points along a straight


line on a logarithmic plot of square root of modified Fanning
friction factor and modified Reynolds number.
The

SPE 71074

correlation, with coefficient correlation R2=0.9984, is defined


as

fm =

33.14
.............................................................. (28)
)1.0458
(N REm

In terms of frictional pressure drop as p/1000 ft, the equation


can be stated as

v
p 5.7565 N
=
2
L
0.0257632 d n
0.5229 2
REm

.................................... (29)

Eq. 29 represents the relationship between frictional pressure


losses, average fluid velocity, tubular internal diameter, and
gel concentration.
Results and Discussion
Eq. 29 was used to calculate the frictional pressure drop for
various gel concentrations, tubular sizes and pumping rates.
Results obtained were compared with the measured values and
were found to be in very close agreement. General deviation
of the predicted results ranged from 0.10% in some cases to a
maximum of 6%. Over 90 percent of the data agreed to the
original results within a deviation of 4%. Generally, it was
seen that with the increase of fluid velocity, the deviation
increased. About 5% of the data was in the range of 4% to 6%
average deviation when the fluid velocity was higher than 65
ft/sec.
Most of the recorded data fell in the range of fully developed
turbulent flow regime. Pumping rates as high as 80 bbl/min
were recorded for an internal diameter of 4.0 inch, whereas
rates as low as 6 bbl/min were observed for internal diameter
of 1.995 inch. In most of the field applications these may be
the upper and lower limits of pumping rates, thus indicating
that the new correlation would efficiently cover majority of
scenarios.
Comparison with other correlations
Results obtained by using the new correlation shown by Eq.
29, were compared with the ones obtained using Lord &
12
McGowan9 correlation and Virk et. al MDR17. API RP39
procedure was not considered for this purpose because of the
complexity involved in the computational process.
Table 2 summarizes the results of comparison. Deviation
from the measured values was calculated for each rate
corresponding to a tubular type and then averaged for the
entire range. For example a deviation of 3.49% in 2-3/8 inch
4.7 lb/ft tubing, using Lord & McGowan relation, indicates the
mean deviation for all the rates available in that data set. As is

SPE 71074

FRICTION PRESSURE CORRELATION FOR GUAR-BASED HYDRAULIC FRACTURING FLUIDS

seen, the new correlation predicts the friction pressures more


closely than the others, for all fluid types. MDR by Virk et al
could predict more accurately for 3-1/2 inch and 4-1/2 inch
tubulars for 35-lbm/Mgal fluid. In contrast to that the
frictional pressure drop predictions by the new correlation is
shows a deviation of about 5% for 35-lbm/Mgal fluid. On an
overall basis however, the new correlation is capable of
predicting frictional pressure drops for CMHPG based fluids
over a wide range of application.

3.

4.

The newly developed correlation is able to predict the


friction pressure losses with more accuracy than the
existing correlations.
Correlation developed at the end of study can be easily
programmed into a spreadsheet and does not involve any
coefficients or interpolation/extrapolation of results.
Since they are based on field gathered data, they can be
readily used in field.

Acknowledgements
How to use the equations
To use the correlation to compute friction pressure losses in
the tubulars, following simple steps are recommended.
1) Calculate n and Kv using standard relations of
rheometry. Convert, Kv to K'p using Eq. 7.
2) Calculate 1 using Eq. 26.
3) Calculate modified generalized Reynolds number using
Eq. 27.
4) Calculate friction pressure drop using Eq. 29.
The above steps can be programmed into a spreadsheet to
facilitate easy use in the field. Frictional pressure losses for
the laminar region can be computed using the Eq. 9 in oilfield
units. Results obtained from Eq. 9 and Eq. 27 are almost
identical for lower flow rates representing a laminar to
turbulent transition, for cases considered in this study. Thus
the typical low-pivot-high friction pressure plots pertaining to
fracturing fluids can be generated easily.
Comparison with field data
Frictional pressure predictions were verified on the field data
that were not used to generate the correlation. Results
obtained from the comparison were very good and showed
that the surface pressures, in most of the cases could be
predicted within 4% of the observed values.
Closer prediction of frictional pressures has helped in
estimating accurate bottomhole pressures.
During the
execution phase this has aided in better monitoring of the
fracture as it propagates in the formation. Better estimate of
bottomhole pressures has also led to a better post job analysis
with the help of which, more realistic results are being
summarized.
Conclusions
1.

2.

It is possible to develop frictional pressure correlations of


hydraulic fracturing fluids based on surface pressure data
acquired in the field.
The effect of tubular diameter and fluid velocity is
distinctly observed when analyzing the data for delayedcross linking CMHPG fluids.

The author wishes to thank Schlumberger for permission to


prepare and publish this study. Special thanks are due to Liz
Simon, Lab Engineer for preparing fluid samples and carrying
out rheology tests. Manuscript review and helpful suggestions
by Joel Robert are greatly appreciated.
Nomenclature
A, B = constants
Cp = Polymer concentration, lbm/Mgal [g/m3]
d = tubular internal diameter, inch [cm]
f = Fanning friction factor
fm = Modified Fanning friction factor
f = Infinite friction factor in Eq. 11
fV = Virks friction factor for max. drag reduction
G = Gel concentration, lbm/Mgal [g/m3]
h = Roughness parameter
j = constant
K' = consistency index - power-law,lbf-secn/ft2 [Pa-sn]
K'p = consistency index - pipe flow,lbf-secn/ft2 [Pa-sn]
K'v = consistency index viscometer, lbf-secn/ft2 [Pa-sn]
L = Tubular length, ft [m]
n' = Flow behavior index of power-law model
N1 = First normal stress difference, lb/ft2 [Pa]
NDe = Deborah number
NRE = Reynolds number
(NRE)s = Solvent Reynolds number
N'RE = Generalized Reynolds number
NREm = Modified generalized Reynolds number
3
P = Proppant concentration, lbm/gal [g/m ]
pbh = Bottomhole pressure, psi [kPa]
ps = Surface pressure, psi [kPa]
p = Pressure drop or friction pressure, psi [kPa]
pf = Friction pressure, psi [kPa]
po = Friction pressure of water solvent, psi [kPa]
pG,P = Pressure drop, psi [kPa]
pH = Hydrostatic pressure, psi [kPa]
p/L = Frictional pressure gradient, psi/ft [kPa/m]
q = Flow rate, bbl/min [m3/sec]
R = Tubular internal diameter, inch [cm]
s = constant
tp = Polymer relaxation time
v = Average fluid velocity, ft/s [m/s]

VIBHAS J. PANDEY

&

w
i,j

= Shear rate, seconds-1


= constant
= Factor used in Eq. 26
= Fluid density, lbm/gal [kg/m3]
= Drag ratio in Eq. 15
= Shear stress, lb/ft2 [Pa]
= Wall shear stress, lb/ft2 [Pa]
= Component of extra shear tensor where i,j=x,y or
z, lb/ft2 [Pa]

References
1.

Metzner, A.B. and Reed, J.C.: "Flow of non-Newtonian


Fluids-Correlation of the Laminar, Transition and
Turbulent Flow Regions," AIChE J. (1955) 1, No. 4, 43440.
2. Rabinowitch, B.; Ze Physik Chemie. Vol. A145 (1929),
pp 1-26.
3. Mooney, M.: Journal of Rheology. Vol. 2 (1937), pp
210.
4. Dodge, D.W. and Metzner, A.B.: "Turbulent Flow of nonNewtonian systems," AIChE J. (1959) 5, No. 2, 189-204.
5. von Karman, T.: National Advisory Committee for
Aeronautic Technical Memo No. 611, Washington
(1931).
6. Shah, S.N.: "Correlations Predict Friction Pressures of
Fracturing Gels," Oil & Gas J. (Jan. 16, 1984) 92-98.
7. Chen, N.H,: "An Explicit Equation for Friction Factor in
Pipe," Ind. Eng. Chem. Fund. (1979) 18, No. 3, 296-97.
8. Dodge, D.W.: "Turbulent Flow of non-Newtonian Fluids
in Smooth, Round Tubes," PhD dissertation, U. of
Delaware, Newark (1957).
9. Lord, D.L. and McGowan, J.M.,: "Real-Time Treating
Pressure Analysis Aided by New Correlation," SPE
15367 presented at the 61st Annual Technical Conference
and Exhibition of the SPE held in New Orleans, LA, Oct.
5-8, 1986.
10. Bowen, R.L. Jr.,: "Designing Turbulent-Flow Systems,"
Chem. Eng. (July 24, 1961) 143-50.

Fluid Type
lbm/Mgal
10
20
25
30
35
40
45
50
60

SPE 71074

11. Gidley, J.L., Holditch, S.A., Nierode, D.E., and Veatch,


R.W. Jr.,: Recent Advances in Hydraulic Fracturing,
Ch. 9 Fracturing Fluid Behavior, SPE Monograph, Vol
12. Henry L. Doherty Series, pp 189, 192.
12. RP 39, Recommended Practices for Standard Procedures
for Evaluation of Hydraulic Fracturing Fluids, second
edition, API, Dallas (Jan. 1983).
13. Shah, S.N.; "Effects of Pipe Roughness on Friction
Pressures of Fracturing Fluids," SPE Production
Engineering, May 1990, pp 151-156.
14. Papanastasiou, T.C.,: Applied Fluid Mechanics, Ch. 9
Rheology and Flows of Non-Newtonian Liquids,
Prentice Hall International Series in the Physical and
Chemical Engineering Sciences, 1994.
15. Seyer, F.A. and Metzner, A.B.,: "Turbulent Flow
Properties of Viscoelastic Fluids," Cdn. J. Chem. Engg
(June 1967) 45, 121-26.
16. Darby, R. and Chang, H.D.: Generalized Correlation for
Friction Loss in Drag Reducing Polymer Solutions,
AIChE J. (March 1984) 30, No. 2, 274-80.
17. Virk, P.S.,: "Drag Reduction Fundamentals," ," AIChE J.
(July, 1975) 24. No. 4, 625-646.
18. Virk, P.S., Mickley, H.S., and Smith, K.A.: "The Ultimate
Asymptote and Mean Flow Structure in Tom's
Phenomenon," Transactions of the ASME. Vol. 24. June
1970, 488-493.
SI Metric Conversion Factors
cp x 1.0*
ft x 3.048*
2
ft x 9.9290 304*
in. x 2.54*
Dynes/cm2 x 1.0*
lbf/ft2 x 4.788 026
lbm/gal x 1.198 264
3
ft x 2.831 685
o
F (oF-32)/1.8

E-03 = Pa.s
E-01 = m
2
E-02 = m
E+00= cm
E-01 = Pa
E-02 = kPa
E+02= kg/cm2
3
E-02 = m
o
= C

*Conversion Factor is exact

TABLE 1RHEOLOGICAL PROPERTIES OF FLUIDS


Linear Gel
Cross-linked Gel (non-heated)
n
K'v, lbsn/ft2 K'p, lbsn/ft2
n
K'v, lbsn/ft2 K'p, lbsn/ft2
0.736966 0.000645 0.000687
0.662965 0.001024 0.001109
0.602665 0.003355 0.003679
0.584962 0.003609 0.003971
0.557995 0.005912 0.006539
0.543142 0.006306 0.006995
0.517848 0.009281 0.010344
0.514573 0.009043 0.010084
0.485427 0.014202 0.015919
0.489385 0.0131
0.014673
0.459432 0.017005 0.019144
0.457207 0.018166 0.020459
0.424498 0.028696 0.032481
0.434403 0.026267 0.029688
0.40691
0.037500 0.042554
0.415037 0.034846 0.039496
0.376721 0.063071 0.071856
0.382146 0.05999
0.052547

SPE 71074

FRICTION PRESSURE CORRELATION FOR GUAR-BASED HYDRAULIC FRACTURING FLUIDS

Fluid Type
30 lbm/Mgal

35 lbm/Mgal

40 lbm/Mgal

45 lbm/Mgal

50 lbm/Mgal
60 lbm/Mgal

8000
Load
Hole

7000

TABLE 2SUMMARY OF COMPARISION


Method
Tubular
McGowan
Virk
OD & lb/ft
2-3/8 4.7#
3.49 %
18.24 %
2-7/8 8.3#
6.34 %
8.47 %
4-1/2 16.6#
-22.84 %
-14.54 %
2-7/8 6.5#
5.44 %
14.25 %
3-1/2 9.3#
-11.19 %
-0.87 %
4-1/2 13.5#
-22.22 %
3.62 %
4-1/2 11.6#
-19.06 %
-7.63 %
2-7/8 6.5#
13.09 %
23.87 %
4-1/2 13.5#
-10.62 %
8.73 %
4-1/2 11.6#
-14.23 %
4.25 %
2-7/8 6.5#
3.36 %
18.81 %
4-1/2 13.5#
-11.32 %
14.85 %
4-1/2 11.6#
14.46 %
9.39 %
4-1/2 13.5#
-11.36 %
23.21 %
4-1/2 11.6#
-7.31 %
20.67 %
4-1/2 11.6#
4.38 %
44.56 %

New
Correlation
1.30 %
-2.50 %
-0.80 %
0.01 %
-5.05 %
5.22 %
-0.38 %
2.82 %
4.92 %
1.74 %
-2.18 %
0.69 %
0.70 %
-1.09 %
-1.42 %
-1.26 %

100

Pump-in
&
Step down

90

Actual Job

80
6000
70

Stabilized pad pressure

5000

60
Slurry Rate(bbl/min)

4000

50
40

3000

Total Slurry(bbl)

30
2000
20

Proppant Conc, lbm/gal

1000
0

10
0

20

40

60

80

100

120

140

160

Treatment Time (min)


Fig. 1--Typical plot generated during the hydraulic fracturing treatment. Treating pressure and total slurry
pumped can be read from the left axis whereas the slurry rate and proppant concentration can be read from
right axis. Stabilized pad pressure is generally observed after the well is displaced with the pad fluid.

10

VIBHAS J. PANDEY

SPE 71074

Plot of Friction pressure vs. Flow rate


35-lbm/Mgal 4-1/2" 11.6 lb./ft

Plot of Friction pressure vs. Flow rate


35-lbm/Mgal 4-1/2" 11.6 lb./ft
1000
Friction Pressure, psi/1000 ft

Friction Pressure, psi/1000 ft

1000

100

100

10

10
10

10

100

Fig. 2--Plot showing the distribution of measured tubular


friction pressures observed for several flow rates for 35lbm/Mgal fluid. Internal diameter of casing is 4.0 inch.

Fig. 3--Power-law best-fit curve is used here to define the


trend of friction pressure data distribution along different
2
value of flow rate. R value observed was 0.9285.

f m 0.5 vs.N' RE
35 lbm/Mgal. 2-7/8" to 4-1/2"

f vs.N' RE
35 lbm/Mgal. 2-7/8" to 4-1/2"
0.01
2.875" 6.5 lb/ft
3.5" 9.2 lb/ft
4.5" 13.5 lb/ft

Square root of Modified


Friction Factor, f m0.5

Fanning Friction Factor, f

0.01

4.5" 11.6 lb/ft

0.001

0.0001
10000

100
Flow Rate, bbl/min

Flow Rate, bbl/min

100000
Generalized Reynolds Number, N' RE

1000000

Fig. 4--Distinct curves representing different tubular internal


diameters can be seen in the plot. It is thus observed that
the conventional plot of f vs. NRE may not be adequate to
define fluid behavior.

0.001
10000

2.875" 6.5 lb/ft


3.5" 9.2 lb/ft
4.5" 13.5 lb/ft
4.5" 11.6 lb/ft

100000

1000000

Generalized Reynolds Number, N' RE

Fig. 5Modification of Fanning friction factor results in


data to fall in a singular curve. By plotting square root of
modified friction factor the data remains within one log
cycle.

SPE 71074

FRICTION PRESSURE CORRELATION FOR GUAR-BASED HYDRAULIC FRACTURING FLUIDS

f m 0.5 vs.N' RE

f m 0.5 vs.N' REm

Square of Modified Fanning


Friction, fm0.5

30 lbm/Mgal n'=0.51
35 lbm/Mgal n'=0.49
40 lbm/Mgal n'=0.46
45 lbm/Mgal n'=0.43
50 lbm/Mgal n'=0.42
60 lbm/Mgal n'=0.38

Modified Friction Factor, f m0.5

0.01

0.01

0.001
10000

11

0.001
100000

100000
Generalized Reynolds Number, N' RE

fm =

33.14
)1.0458
(N REm

1000000

10000000

1000000

Fig. 6Plot of square root of modified friction factor and


generalized Reynolds number for various fluid types.
Separation of curves is very distinct because of the
modification applied on the friction factor.

Modified Reynolds Number, N' REm

Fig. 7Plotting the data after taking into consideration


the effect of gel concentration, leads to the above plot.
Equation shown above represents the relation. Observed
2
R = 0.9984.

You might also like