You are on page 1of 14

Society of Petroleum Engineers

SPE 25456
A New Model for Laminar, Transitional, and Turbulent Flow
of Drilling Muds
T.O. Reed, Conoco Inc., and AA Pilehvari, U. of Tulsa
SPE Members
Copyright 1993, Society of Pf1.\roleum Engineers, Inc.
This peper was prepared for presentation at the Production Operations Symposium held In Oklahoma City, OK, U.S.A., March 21-23, 1993.
This paper was selected for presentation by an SPE Program Commillee 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 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. Permission to copy Is restricted to an abstract of not more than 300 words. illustrations may not be copied. The ebstrect should contain conspicuous acknowledgment
of and by whom the paper is presented. Write Librarian, SPE, P.O. Box 833836, Richardson, TX 75083-3838, U.S.A. Telex, 163245 SPEUT.
ABSTRACT
The concept of an "Effective" diameter is introduced for the flow of
drilling muds through annuli. This new diameter accounts for both
annular geometry and the effects of a non-Newtonian fluid. It
provides the link between Newtonian pipe flow and non-Newtonian
flow through concentric annuli. The method is valid in any flow
regime and can be used to determine whether a DOn-Newtonian flow
is laminar, transitional, or turbulent. An analytical procedure is
developed for computing frictional pressure gradients in all three flow
regimes. The analysis also quantifies bow flow transition is delayed
by increasing the yield stress of a fluid. In addition, it is shown that
transition in an annulus is delayed to higher pump rates as the ratio
of inner to outer diameter increases. Furthermore, the method
accounts for wall roughness and its affects on transition81 and
turbulent pressure gradients for non-Newtonian flow through pipes
and concentric annuli. Finally, the method runs on a 386 PC in only
a few seconds.
INTRODUCTION
The standard API methods for drilling hydraulics assume either a
Power Law or a Bingham Plastic rbeology modeL In reality, most
drilling muds correspond mucb more closely to the Modified Power
Law or Herschel-Bulkley rbeological model. This distinction is
particularly important for the annular geometries typical of normal
drilling conditions where shear rates are usually low and the Power
Law underestimates and Bingham Plastic model overestimates
frictional pressure drops. This paper also shows that these two
classical models, respectively, underestimate and overestimate pump
rates required for transition from laminat to turbulent flow.
Although there have been a number of papers pUblished on the
laminar flow of yield-pseudoplastics through annuli/-J none claim to
be uniformly valid for laminar, transitional, and turbulent flow. This
can be accomplished by introducing tbe concept of an "Effective"
diameter which accounts.for both annular geometry and the effects
of a non-Newtonian fluid. This diameter also enables the inclusion
of the effects of wall roughness on frictional pressure gradients and
the process of flow transition. Furthermore, the resulting model can
be used to determine whether flow of a non-Newtonian fluid through
References and illustrations at end of paper.
a pipe or concentric annulus is laminar, transitional, or fully turbu-
lent.
A derivation of the model is presented in the Appendix. Some
additional background and results from the model are given in the
fonowing discussions.
NEWTONIAN FLOW
RelatioDShip Between Pipe and Annular Flows. A considerable
number of "equivalent" diameters have been proposed over the years
for flow through conduits other than circular pipes.1O-1z The purpose
of defining such a diameter is to introduce definitions of friction
factor and Reynolds number that will enable application of the wen-
known relations for pipe flow to other geometries. In particular, the
objective is to be able to calculate what the frictional pressure drop
will be for a given fluid at a particular flow rate.
h shown in the Appendix, the friction factor for a concentric
annulus should be based on the Hydraulic Diameter, Dhy> which is
simply the difference between the inner and outer diameters. The
Reynolds number should be based on an Diameter equal
to the square of Lamb's Diameter, D
L
, divided by Dhyt see
Appendix Eqs. A.12, 13 and 14. When the friction factor and
. Reynolds number are defined in this manner, the classical relations
for Newtonian pipe flow, e.g., the Moody Diagram, can be applied
directly to concentric annuli. In fact, Jones and Leung
13
have proven
that these definitions also apply to fully-turbulent flow through an
annulus. This was demonstrated by showing that a variety of
experimental data for concentric annuli agree with the Colebrook
Equation for turbulent pipe flow, which is equivalent to the fully
turbulent and hydraulica1ly rough portions of the Moody Diagram.
Flow Transition in Annuli. In addition to being applicable in the
laminar and turbulent regimes, these same definitions can also be
used to predict the "critical" Reynolds number at which laminar flow
ends and transition to turbulence begins. Prof. Hanks and his
students have published experimental data on flow transition in
concentric annuli.
I
4-15 Some of their data is shown in Fi&- 1. The
dashed curve is the results of predictions from a transition theory
469
2
A NEW MODEL FOR LAMINAR, TRANSmONAL, AND
ruRBULENT FLOW OF DRILLINGMUDS SPE2S456
developed by Hanks in the early 1960s.7,1"17 The solid curve comes
from simply setting the Reynolds number, based on the Equivalent
Diameter of Jones and Leung, equal to 2100. A Reynolds number
based only on Lamb's Diameter is used for the ordinate in order to
display how the critical values vary with diameter ratio. It is readily
seen that the much simpler transition criterion provides better
agreement with Hanks' experimental data. The new criterion shows
that transition from laminar flow is delayed to progressively higher
velocities or pump rates as diameter ratio increases.
NON-NEWTONIAN FLOW
Pipe Flow. Prof. Metzner and his students reported some ~ i o n r i n
work in the 19508 on non-Newtonian flow through pipes.!l-Z1 Their
experiments with Power-Law fluids in pipes provided the first clear
definitions of how the friction factor varies with Reynolds number in
the laminar, transitional, and turbulent regimes. They found that a
decreasing power-law exponent delays transition to higher Reynolds
numbers and shifts the turbulent friction factors downward. This is
a direct result of increasing degrees of the phenomenon called wshear
thinning."
In addition to the test data, Metzner and his students also developed
a novel analysis that provided a way to generalize their results for
Power-Law fluids to all time-independent, non-Newtonian fluids.
They simply defined:
Wall Shear Stress = K' x (Newtonian Shear Rate)N (1a)
or using standard symbols:
, N
T
w
= K x (8 v / D) ..........uo (1b)
From this equation, it follows that the definition of the exponent WNW
is:
N =d(ln Twl / d(ln (8 v / D)] (1e)
They showed how the two parameters "K'w and WNW can be applied to
Power-Law (PL) fluids and Bingham Plastics (BP). The Appendix
provides the details on how this approach can also be applied to a
Herschel-Bulkley (BB) fluid, which is essentially a combination of the
PL and BP rheological models.
The Metzner group also developed the following relation for wall
shear rate in a pipe for a general fluid with time-independent
properties.
Generalized Pipe Shear Rate:
'Yw = (3N + 1) 8 v / (4N D) uo (2)
This expression led us to the following conclusion. An "Effective
w
diameter can be defined so that the generalized shear rate can be put
into the same form as the shear rate for Newtonian pipe flow, viz.,
(8 v/D). Using this logic, the Effective Diameter for generalized non-
Newtonian pipe flow is:
WElTective
w
Pipe Diameter =D
eff
=4N D/(3N + 1) .... (3)
The wEffective
w
diameter for non-Newtonian pipe flow is the
diameter of a circular pipe that would have the identical pres-
sure drop for flow ofa Newtonian fluid with viscosity equal to
the "apparent" viscosity and the same average velocity as the
non-Newtonian flow. Note that this diameter is less than the
physical diameter for pseudoplastics (N < 1) and is greater than the
physical diameter for di1atant fluids (N > 1). As will be shown, the
Effective Diameter makes it easier to relate non-Newtonian and
Newtonian flows. The advantage of such a connection is that the
well-established friction factor relations for Newtonian flows can then
be applied to non-Newtonian flows. This greatly simplifies the task
of computing frictional pressure drops for such flows.
The remaining variable that is needed to define the Generalize
Reynolds Number is the "apparent" Newtonian viscosity. This
parameter is defined by:
App. Viscosity =Shear Stress/Shear Rate @ Wall (4a)
1J.w,app =K' (8 v / D)N/ 'Yw ..................................uo.... (4b)
With these definitions, the Generalized REynolds number (GRE) is
defined as:
NRe,O =p v Deffl 1J.w,app uo (5)
This expresses the Reynolds number for non-Newtonian pipe flow in
the same algebraic form as for a Newtonian fluid Metzner &Reed!g
showed that the GRE is related to laminar friction factor by the
same classical equation for Newtonian flow, viz,
F
c
=16/ NRe,o , Laminar Non-Newtonian Flow...... (6)
These authors showed that experimental pipe flow data for a variety
of non-Newtonian fluids followed this relationship. Later experi.
ments with Power-Law fluids in turbulent pipe flow led Dodge &
Metzner3l to modify the classical Colebrook equation for turbulent
friction factors in Newtonian pipe flow, see Eq. A-45. This was
necessary in order to correlate the data because the shear-rate exp0-
nent was found to be a significant parameter. In particular, friction
factors decrease with decreasing values of the Power-Law exponent.
Figure Z shows the correlations Dodge & Metzner developed for
laminar and turbulent flow. The transition zones were not correlated
by these authors, but their experimental data follows the indicated
curves between the end of laminar flow and the beginning of fully-
turbulent low. As described in the Appendix, a new correlation for
the transition zones bas been developed based on the experimental
data of Dodge & Metzner. The curves in the transition zones are
from this new correlation.
The generalized shear-rate exponent "Nil can be used to extend the
data of Dodge & Metzner for Power-Law fluids to other non
Newtonian fluids. In partiCUlar, Dodge & Metzner derived the
appropriate equations for the GRE, shear rate, and apparent
viscosity for flow of a BP through pipes. Their work is extended to
Herschel-Bulkley fluids and concentric annuli. in the Appendix.
However, before discussing annular flow, it is appropriate to first
discuss some results from the new transition criterion for fluids with
a yield stress.
Hanks and Pratt
J
proposed a model for flow of Bingham Plastics
through pipes. Their paper also includes a transition criterion. They
chose to base their criterion on the so-called Bingham Plastic
Reynolds number and the Hedstrom number. These two dimension-
less parameters are defined in the Appendix, see Eqs. A-50 and
A-51. Unfortunately, the BP Reynolds number is not a legitimate
measure of Reynolds number effects simply because the Plastic Vis
cosity (PV) is a constant and does not vary with shear rate as the
apparent viscosity does. Furthermore, the Hanks and Pratt criterion
does not agree with experimental data at higher values of the
Hedstrom number.
J
This disagreement is shown in Fig. 3. Other
investigators have also noted this disagreement.Z1-22 In contrast, the
solid curve in Fig. 3 is based on the new transition criterion which
requires the critical condition for any fluid to occur when the product
470
SPE 25456 T. REED AND A. PILEHVARI 3
of the Fanning friction factor and the GRE equals 16.1. (As may be
seen from Eq. 6, this product is 16 in Jaminar flow, and it exceeds this
value wben transiqon begins.) This generalized transition criterion is
designed to reduce to a critical Reynolds number of 2100 for
Newtonian pipe flow, see last section of Appendix.
The underprediction of the data by the Hanks and Pratt criterion is
even more apparent when critical values of the GRE are plotted as
a function. of Hedstrom number, Fig. 4. Here it becomes clear that
tbe Hanks and Pratt criterion predicts decreasing flow rates for
transition as the Yield Point (YP) increases. This certainly violates
tbe expected trend. The new generalized criterion show the expected
trend, i.e., higher pump rates, are required to achieve transition as
tbe yP of a BP increases.
It is instructive to plot the friction factors for BP pipe flow through
all three flow regimes with Hedstrom number as a parameter. This
is sbownin Fig. 5. This figure shows bow an increasing yP delays
transition to higher pump rates and also extends tbe transition zone
over a progressively wider range of pump rates. For example, wben
tbe Hedstrom number is 500,000, the flaw is not fully turbulent until
a Reynolds number of 100,OOO! This compares with a common value
of about 3,000 for Newtonian pipe flow.
It is particularly important to observe that all of the curves for
eventually merge into tbe turbulent curve for Newtonian
flow. This occurs because the yP becomes progressively less
significant as sbear rate increases. Hence, the curves have the
expected asymptotes. . There is one other point to notice from
comparing Figs. 5 and 3. A BP bas a sbear-rate exponent of 1;
wbereas, a PL fluid generally bas a nonunity exponent. This is the
basic reason wby the friction factor curves for different values of the
sbear-rate exponent do not merge and remain distinct for all
Reynolds numbers. In tum, this implies that a Herschel.Bulkley fluid
will exhibit somewbat different behavior since it combines tbe BP and
tbe PL rbeological models. In particular, one can infer that a yield
stress will delay and extend the transition zone, and tbe transitional
values of friction factor will eventually merge with the fully-turbulent
curve for a PL fluid with an exponent equal to whatever appears in
tbe Herscbel-Bulkley model, see Eq. 7 below.
At tbis point, it is appropriate to introduce some viscometer data for
a representative drilling mud. Some typical rotating-cup data are
shown in Fig. 6. Dial readings for six: rotary speeds are given. The
two parameters in the BP model (YP and PV) and the PL model (K
and "n") are determined according to standard API procedures by
fitting these two rheological models to the 300 and 600 rpm data. In
contrast, the three parameters of the HB model allow a good fit to
all six data points. Since annular flow shear rates are normally below
the 100 rpm (170 sec-I) viscometer data, it is important to have a
model which is valid at low shear rates. On the other hand, the HB
model can also be used to estimate frictional pressure drops and flow
regimes for flow through drill pipes where shear rates are high. The
HB parameters for the particular data of Fig. 6 are:
T = Yield Stress + K (Shear Rate)-
::: 14 +0.281 (,y)o.m Ibf/l00 ft2 [0.4788 Pal (7)
These parameters will be used for illustration purposes in Figs. 7 tbru
11. A weight of 12 ppg [1,438 kglmJ] will also be assumed for this
example mud.
Figure 7 presents friction factors through the transition zone for the
example mud flowing through pipes. The PL model leads to transi-
tion at lower pump rates. Initiation of transition is delayed to higher
pump rates based on the HB model, but fully-turbulent flow occurs
at about the same GRE as for tbe PL case. The HB curve for fully-
turbulent flow does not merge witb the PL curve at higher Reynolds
numbers because the sbear rate exponent for the PL model is only
0566. The results for the BP model is as expected, i.e., transition is
delayed tbe most by using this model. Furthermore, the extent of.
the transition zone is expanded significantly, and the GRE is an
order-of-magnitude larger before fully-turbulent flow is achieved.
In order to relate these results to a typical field application, consider
flow through a standard weight 5-incb drill pipe with an ID of
4.276 incbes. Fu1Iy-turbuient flow is achieved at a GRE of 3800 and
a pump rate Of 270 gpm [17.0 lis] using the PL model. The corre-
sponding results for the HB model occurs at a GRE of 4100 and a
pump rate of 310 gpm [19.6 lis]. In contrast, tbe BP model leads
to a prediction of fully-turbulent flow at a GRE of 22,300 and a
pump rate of 670 gpm [42.3 Urn]. (Note: The analysis does not
account for the effects of restrictions at tool joints whicb could cause
flow transition to begin at lower pump rates.)
It is also of interest to compare the frictional pressure gradients for
a realistic pump rate of 600 gpm [37.85 lis]. The results for the PI..,
HB, and BP models, respectively, are 0.085, 0.099, and 0.104 psilft
[22.62 kPa/m]. The relative close values for the HB and BP fluids
are a result of the HB being fully turbulent and the BP being in
transition at this pump rate. Again, if tool joints caused transition to
occur at lower pump rates and the BP were in fully-turbulent flow at
600 gpm [37.85 lis], then the difference between pressure drops for
the HB and BP would be larger.
Annular Flow. The flow model can be extended to concentric annuli
by altering tbe definition of the "Effective" diameter to include the
effects of the different geometry. The "Effective" diameter for a
concentric annuli is equal to DJIy IG. The "Effective" diluneter for
non-Newtonian flow through a concentric annulus is the dia-
meter of a circular pipe that would have the identical pressure
drop for flow ofa Newtonian fluid with a viscosity equal to the
"effective" viscosity, which is based on the average wall shear
rate in the annulus, and has a velocity equal to the non-
Newtonian annular flow velocity.
The "an function is based on a correlation of the analytical solution
by Fredrickson and Bird! for flow of a Power-Law fluid through a
concentric annulus. This function is dependent on both tbe ratio of
inner to outer diameters and the shear-rate exponent. The function
is defined in the Appendix by Eq. A-34. This correlation can be
applied to other fluids by using the generalized exponent "N", Eq. 1c,
in place of the usual PL exponent "n." The details are given in the
Appendix under tbe beading "Non-Newtonian Annular Flow."
The following discussion will focus on some results for a typical field-
size annulus. A 12-1/4" x 5" [31.12 emx 12.7 em] annulus is selected
for illustrative purposes. The same mud rbeology, Eq. 7, and a
weight of 12 ppg [1,438 kg! m
J
] will again be used for computing
frictional pressure gradients through the annulus. It is pertinent to
note that the weight of the mud does not affect laminar frictional
gradients, but weight does influence transitional and turbulent
gradients.
Figure 8 sbows bow the ratio of tbe generalized exponent "N" over
the shear-rate exponent Om" varies with shear rate for the example
HB fluid. As may be seen from Eq. 7, "m" bas the numerical value
of 0.792. Figure 8 shows bow "N" approacbes "m" as wall sbear rate
increases. This is the expected trend because the influence of the
Yield Stress (YS) on T
w
becomes progressively less with increasing
sbear rate, compare Eqs. 1b and 7. In the limit of infinite sbear rate,
N=m.
Figure 8 also identifies tbe initiation of transition at 1910 gpm
[1205 lis], and fully-turbUlent flow begins at 2010 gpm [126.8 lis].
This simply tells us that the flow is laminar over the range of pump
rates that are normally used in the field. Unfortunately, the
does not tell us bow drill string dynamics may influence flow
471
4
A NEW MODEL FOR LAMINAR, lRANsmONAL, AND
TURBULENT FLOW OF DRTI..LING MUDS SPE2S456
transition. It is reasonable to expect the energy added to the flow by
the agitation may cause transition at lower pump rates.
Another parameter in the model is the ratio of Effective diameter
over the Equivalent diameter. This ratio partia1Iy removes geometric
effects and highlights the non-Newtonian effects. Fipre' shows this
ratio approaches unity as "N" goes to one. This is the expected
limiting condition for Newtonian flow. Since Fig. 8 shows that "N"
increases with pump rate, the Effective diameter also increases with
pump rate. This has the net effect of reducing the magnitude of the
increase in frictional pressure gradient as pump rate is increased, i.e.,
the fluid is "shear thinning."
This leads to the question, what happens to the "apparent" viscosity
as pump rate increases? An "Effective" viscosity is defined in a man-
ner analogous to pipe flow. The primary difference for an annulus
is: the "Effective" viscosity is defined by the average wall shear stress
acting on the inner and outer walls and divided by the average wall
shear rate, see Eq. A-43. This viscosity is presented in Fig. 10 as a
function of pump rate. It continuously decreases with pump rate and
. has a noticeable drop at transition from laminar to turbulent flow.
This is caused by the rapidly increasing shear rate as transition occurs.
Figure 11 presents friction factors as a function of the GRB for the
example annulus and mud properties. For smooth walls, the increase
in friction factor through the transition zone is small. As indicated in
Fig. 8, the transition zone occurs between pump rates of 1910 and
2010 gpm [120.5-126.8 I.Js].
An example of the effects of rough walls is also included in Fig. 11.
It may be noticed that the initiation of transition occurs at the same
GRB' This is consistent with the experimental data of Nikuradse for
Newtonian pipe flow, see Schlichting.
24
As indicated in the figure,
transition to turbulent flow takes place over a smaller range of pump
rates, 1910 to 1970 gpm [120.5 to 124.3 I.Js], and the fully turbulent
pressure drops are approximately 70 percent larger than the corre-
sponding turbulent friction factors for smooth walls.
The "relative roughness" for this illustration is approximately 0.01.
This is based on dividing the absolute roughness height by the
"Equivalent" diameter, Eq. A-14. However, the correct definition of
relative roughness for non-Newtonian flows is the ratio of absolute
roughness over the "Effective" diameter. Hence, relative roughness
varies with the fluid and the shear rate.
Table 1 present a list of some of the relevant parameters through the
transition zone for the rough-wall case. It can be seen that the
relative roughness decreases as pump rate increases. The authors
believe these are the first computations for the effects of wall
roughness on non-Newtonian pressure gradients in the transitional
and turbulent regimes. (Note: pressure gradients are given in units
of inches of water per foot; these units are equal to 0.0361 psiIft or
0.8164kPa/m.)
Finally, Fig. 12 compares the model with test data for a Mixed Metal
Hydroxides (MMH) mud system flowing through a 5.023" x 2.375"
[12.76 cmx 6.03 em] annulus. The tests were conducted in Amoco's
Drilling Hydraulics Test Facility by personnel from the Petroleum
Engineering Department, Tulsa University. Pressure drops were
measured over a length of 50 ft [15.24 m]. Pump ratj:s were varied
from 50 to 500 gpm [3.16 to 31.55 I.Js]. Experimental data for a high
and a low rheology MMH fluid are included in Fig. 12 The
corresponding rotating-cup viscometer readings at the six standard
speeds are: 327,57.3,61.5,63.0,66.5 and 25,13.5,18.3,220,28.0.
These values are the average of readings for four different samples
taken during the time of a tesL
As indicated in Fig. 12, the high-rheology fluid appears to have been
in the laminar-flow regime for the entire range of pump rates. The
model predicts transition at higher pump rates and friction factors
that continuously decrease through the transition zone. According
to the model, the flow is not fully turbulent until a GRB of 18,000
and a pump rate of 77JJ gpm [45.42 I.Js]. The predicted values of
turbulent friction factors are nearly constant for a wall roughness of
0.00018 in. [0.0046 mm].
The above value of wall roughness gave good agreement between
predicted turbulent friction factors and experimental data for thelow-
rheology fluid. Unfortunately, water tests to determine wall
roughness for the annulus were not performed. However, Fig. 12
shows encouraging agreement between predictions for the low-
rheology fluid and the corresponding data through au three flow
regimes.
Additional tests wereconductedwith bentonitemuds flowing through
rough pipes. In these cases, wall roughness was determinedvia water
tests, and predictions from the model, using the experimental values
of wall roughness, agree very closely with the test data for au three
flow regimes. This data covers a GRB range of 500 to 270,000 and
includes relative roughness values up to 0.002. Because of space
limitations, these non-Newtonian pipe-flow data are not presented
here.
CONCLUSIONS
1. A new analysis has been developed for non-Newtonian flow
through pipes and concentric annuli. The method is based oil
relating non-Newtonian flows to Newtonian flows. The advan-
tage is that well-established results for Newtonian flows can be
applied to non-Newtonian flows.
2 The "Effective" diameter is a key concept of the method. It
accounts for both geometric and non-Newtonian fluid effects
on frictional pressure gradients in pipes and annuli.
3. Results agree with finite-difference computations for laminar
flow of a Herscbel-Bullcley fluid through concentric annuli.
4. The analysis is valid for the laminar, transitional, and fully-
turbulent flow regimes. The method incorporates a new transi-
tion criteria that accounts for a delay in flow transition with
increasing ratio of the inner-to-outer diameters in concentric
annuli. This agrees with experimental data reported in the
literature.
5. Model predictions of critical Reynolds numbers agree with
published data for BP pipe flows.
6. The Herschel-Bulkley rheolOgical model is used which includes
Newtonian, Power-Law, and BinghamPlastics as special cases.
This IDOre accurate rheological model is equally valid for
predictions of pressure gradients in drill pipes and the large
annuli typical of full-scaJe
7. When using the same viscometer data, results from the
analyses show transition from laminar to turbulent flow occurs
at higher pump rates than for a Power-Law fluid, but signifi-
cantly lower than the corresponding Bingham Plastic fluid
8. The Colebrook equation for turbulent friction factors has been
extended so that it applies to non-Newtonian flows through
pipes and concentric annuli with smooth or rough walls.
9. The method accounts for the effects of wall roughness on
frictional. pressure gradients in transitional flow.
10. Model predictions of pressure gradients in au three flow
regimes have beenverified by experimental data for MMHand
bentonite flowing through pipes and annuli with varying
degrees of wall roughness.
472
SPE 25456 T. REED AND A PILEHVARI 5
7
w
= shear stress at walls of a pipe or annulus
11. The method runs on a 386 PC in only a.few seconds. This is
made possible by avoiding either or finite-
element numerical solutions. Furthermore, the capabilities of
these numerical techniques ha\ie not yet been developed to
model transitional nor turbulent flow of a Herschel-Bulkley
fluid.
NOMENCLATURE
JLw.eff
p
7
= "effective" Newtonian viscosity at walls of an annulus,
Eq. A-43
= fluid density, Ibm/gal
= shear stress, Ibf/l00 tt2
A = parameter used to define "N" for a Herschel-Bulkley
fluid, Eq. A-31
D = internal diameter of a circular pipe, in.
D
eff
= Effective diameter of either a pipe or annulus, Eq. A-37,
in.
Dhy = hydraulic diameter, Do - D
j
, in.
D
j = inner diameter of annulus = drill pipe OD, in.
D
L
= Lamb's diameter for a concentric annulus, Eq. A-l2, in.
Do = outer diameter of annulus = borehole diameter
F
c
= Fanning friction factor, 7..... /(1/2 p v2)
Flam = Fanning friction factor in laminar flow
Fir = Fanning friction factor in transition reginie
F
lUrb
= Fanning friction factor in turbulent flow
G = an ExIog correlation factor, Eq. A-34
He = Hedstrom number, Eq. A-51
K = consistency indexused for either a Power-Law, Eq. A-19,
or a Herschel-Bulkley fluid, Eq. A-29, Ibf-sec"'/l00 ft2
K'
= generalized index of Metzner and Reed,
Eq. Ib, Ibf-sec"/l00 ft '
L = length of pipe or annulus over which pressure drop is
measured
m = flow behavior index for Herschel-Bulkley fluids, Eqs. 7
andA-29
N = generalized flow behavior index of Metzner and Reed,
Eqs. Ib and lc
NRe,G
= Generalized REynolds number, Eq. A-44
n = flow behavior index for PL fluids, Eq. A-19
PV = plastic viscosity for a Bingham Plastic fluid
RCoL
=
Reynolds number based on D
L
v
= velocity of flow = flow rate/cross-sectional area
X = ratio of yP 1wall shear stress, Eq. A-54b
Y = an ExIog correlation factor, Eq. A-32
yP
=
yield point for a Bingham Plastic, Ibf/l00 tt2
YS
=
yield stress for a Herschel-Bulkley fluid, Eqs. 7 and A-29,
Ibf/l00 ft2 .
Z
=
an ExIog correlation factor, Eq. A-33
Greek Letters
e = absolute height of wall roughness, in.
.
'Yw,avg = average shear rate at walls, Eq. A-41, 1/see
JLw,app = "apparent" Newtonian viscosity, Eq. A-23, Ibf-sec/l00 ft2
ACKNOWLEDGEMENTS
We thank the management of Conoco, Inc. for permission to publish
this paper. We also thank Amoco, Inc. for allowing tests in their
research facilities by R Subramanian, a Tulsa University graduate
student
REFERENCES
1. Fredrickson, A G., and Bird, R B., "Non-Newtonian Flow in
Annuli," IntI. Engr. Chem. (March 1958) SO, No.3, 347-52.
2. Hanks, R W., "The Laminar-Turbulent Transition for Fluids
with a Yield Stress," AlChE1. (May 1963) 9, No.3, 306-09.
3. Hanks, R W. and Pratt, D. R, "On the Flow of Bingham
Plastic Slurries in Pipes and Between Parallel Plates," SPFJ
(Dec. 1967) 342-46.
4. Hanks, R W. and Ricks, B. L, "Laminar-Turbulent Transition
in Flow of Pseudoplastic Fluids with Yield Stresses," AlAA.
Jour. ofHydronaulics (Oct. 1974) 8, No.4, 163-66.
5. Hanks, R W., "The Axial Laminar FlowofYield-Pseudoplastic
Fluids in a Concentric Annulus,"IntI. Engr. CMm. Process Des.
Dev. (1979) 18, No.3, 488-93.
6. Haciislamoglu, M., and Langlinais, J., "Non-Newtonian Flowin
Eccentric Annuli," ASME J. Energy Resources Tech., (Sept
1990) lU, No.3, 163-169.
7. Oltafor, M. N. and Evers, J. F., "Experimental Comparison of
Rheology Models for Drilling Fluids," paper SPE 24086
presented at the 1992 Western Regional Meeting, BakerSfield,
CA, Mar. 3O-Apr. 1.
S. Gucuyencr, H. Land Mehmetoglu, T., "Flowof Yield-Pseudo-
plastic Fluids through a Concentric Annulus," AlChE 1.
(July 1992) 38, No.7, 1139-43.
9. Chin, W. c., Borehole Flow Modeling, Gulf Publishing Co.,
Houston (1992).
10. Miller, c., "Predicting Non-Newtonian FlowBehavior in Duets
of Unusual Cross Section," IntI. Eng. Chem. Pundom. (1972)
11, No.4, 524-28.
11. Hanks, R W., "On the Prediction of Non-Newtonian Flow
Behavior in Duets of Noncircular Cross Section," IntI. Eng.
Fundam.. (1974), No.1, 62-6.
12. Jensen, T. B. and Sharma, M. P. "Study of Friction Factor
Equivalent Diameter Correlations for Annular Flow of Non-
Newtonian Drilling Fluids," ASME 1. Energy Resources Tech.
(Dec. 1987) 109, 200-05.
13. Jones, O. C. Jr., and Leung, J. C. M., "An Improvement in the
Calculation of Turbulent Friction in Smooth Concentric
Annuli," ASME Jour. Fluids Engineering (Dec. 1981) 103,
615-23.
14. Hanks, R W., "Critical Reynolds Numbers, for Newtonian
Flow in Concentric Annuli," AlChEJ. (Jan. 1980) Z', No.1,
152-4.
473
6
A NEW MODEL FOR LAMINAR, TRANSmONAL, AND
TIJRBULENT FLOW OF DRILIJNG MUDS SPE 25456
15. Hanks, R. W., and Peterson, J. M., "Complex Transitional
Flows in Concentric Annuli," AlChE1. (Sept. 1982) 28, No.5,
SOO-06.
16. Hanks, R. W., "Tbe Laminar-Turbulent Transition in Noniso-
thermal Flow of Pseudoplastic Fluids in Tubes," AlChE 1.
(Sept. 1962) 8, No.4, 467-71.
17. Hanks, R. W., "Tbe Laminar-Turbulent Transition for Flow in
Pipes, Concentric Annuli, and Parallel Plates," AlChE lour.
(Jan. 1963) 9, No.1, 45-8.
18. Hanks, R. W., "ATheory of Laminar Flow Stability,"AlChE1.
(Jan. 1969) 15, No. I, 25-8.
19. Metzner, A. B., and Reed, J. C., "Flow of Non-Newtonian
Fluids - Correlation of the Laminar, Transition and Turbulent
flow Regions," AlChE 1. (Dec. 1955) I, No.4, 434-40.
20. Metzner, A. B., "Non-Newtonian Fluid Flow," [nil. Eng. Chem.
(Sept. 1957) 49, No.9, 1429-32.
21. Dodge, D. W., and Metzner, A. B., "Turbulent Flow of Non-
Newtonian Systems," AlChEl. (June 1959) 5, No. 2, 189-204.
22. Mishra, P., and Tripathi, G., "Transition from Laminar to
Turbulent Flow of Purely VISCOUS Non-Newtonian Fluids in
Tubes," Chem. Engt'. Sci. (1971) ZCi, 915-21.
23. Franco, V., and Verduzco, M. B., "Transition Critical Velocity
in Pipes Transporting Slurries with Non-Newtonian Behavior,"
Proc. 14th Miami Univ., Coral Gables, Multipbase Transport
&. Particulate Phenomena Int'l Symp., Vol. 4, 289-99, Miami
Beach, FL, 1988.
24. Schlichting, H., BoundtllY Layer Theory, 7th eel, McGraw-Hill
.Book Co. New York (1979), 616-20.
25. Bourgoyne, A. T. Jr., Chenevert, M. R, Millheim, K. K., and
Young, F. S. Jr., Applied Drilling Engineering, Society of
Pe.troleum Engineers, Richardson, 'IX, p. 140, 1986.
26. Govier, G. W., and Aziz, K., The Flow ofComplex MIXtures in
Pipes, Van Nostrand Reinhold Co., New York (1972) 201-2.
27. Hanks, R. W., "Low Reynolds Number Turbulent Pipeline
Flow of Pseudohomogeneous Slurries," Proc., Hydrotransport
5, Fifth Int'l Cont on the Hydraulic Transport of Solids in
Pipes, . British Hydromechanics Research Association
(May 1978) paper C2.
28. Theory and Application of Drilling Fluid Hydraulics, A.
Whittaker and EXLOG Staff (ed.), International Human
Resources Dev. Corp., Boston (1985), 102.
29. Hanks, R. W., "Tbe Not so "Generalized" ReynOlds Number,"
Proc. 4th Int'l Tech. Conf. on Slurry Transport, Slurry Trans-
port Association, Washington, D.C. (1979) 91-98.
30. Churchill, S. WOJ "Friction Factor Equation Spans All F1uid-
Flow Regimes," Chemical Engineering, (1977) Nov. 7, 91-2.
31. Wilson, N. W., and Azad, R. S., "A Continuous Prediction
Method for FullyDeveloped Laminar, Transitional, and Turbu-
lent F1ows.in Pipcs,"ASMElour. ofAppliedMec1uJnics (1975)
March, 51-4.
APPENDIX-DERIVATION OF MODEL
The analysis begins with the question: Is there any way to relate
Newtonian pipe flow to non-Newtonian annular flow? If this is
possible, then all of the well-established knowledge about Newtonian
pipe flow can then be applied to the more difficult non-Newtonian
problem. This can be accomplished by first introducing an"Equiva-
lent" diameter to relate a concentric annular geometry to a Circular
pipe. Next, this concept can be extended by introducing an "Effec-
tive" diameter which properly accounts for non-Newtonian effects.
The development of the procedure begins with an analysiS of Iaminar
flow in pipes and annuli. Next, fully-turbulent flow is analyzed and
wall roughness is introduced. FJnally, a procedure is developed for
predicting when transition from laminar to turbulent flow begins and
when it ends.
LAMINAR NEWrONIAN FLOWS
Pipe Flow. The purpose of the following discussion is to introduce
the terminology that will be used. The solution forisothermaI fully-
developed viscous flow of a Newtonian fluid through a circUlar pipe
was derived by Hagen and Poiseuille in the 1800s. Their equation
for frictional pressure gradient is:
dP / elL =32,.,. v/0
2
(A-I)
Next, we note that a Newtonian fluid is defined to be one for which
shear stress is linearly proportional to shear rate, i.e., Shear Stress =
viscosity xShear rate.
T = ,.,. 'Y (A-2)
Equations A-I and A-2 can be related via a simple force balance
between the pressure gradient that drives flow through the pipe and
the viscous forces at the wall that oppose the motion. This provides
the following equation.
Tw =dP / elL X 0 /4 (A-3)
The subscript OW" refers to the shear stress and shear rate at the
inside wall of a pipe. An expression for shear rate at the waU can be
obtained by combining Eqs. A1, A-2, and A-3, viz.,
'Yw =8 v/ 0 (A.4)
The next step is to introduce the Fanning friction factor, which is
defined byEq. A-5.
F
f
= T
w
/ (Ia p v2) =dP / elL X 0 / (2 P v2) (A-S)
Since the friction factor is traditionally expressed as a function of
Reynolds number, this parameter is introduced next.
N
Re
=Pv D /,.,. (A-6)
Both the friction factor and the Reynolds number are dimensiOnless,
and the two can be related by combining Eqs. A-I, A-5 and A-6 to
obtain the classical relation for laminar pipe flow.
F
f
=16/ N
Re
(A-7)
Note that the friction factor and the Reynolds number both include
fluid density in their definitions. As may be seen from Eq. A.l, the
frictional pressure gradient in a laminar flow is independent of fluid
density. However, pressure drops in transitional and turbulent flows
are a function of the fluid density. Hence, these two dimensionless
parameters are useful in aU three flow regimes. This completes the
elementary definitions that are used sUbsequently in the analysis of
non-Newtonian annular flows.
Concentric Annular Flow. Lamtr' derived a solution for Newtonian
flow through a concentric annulus that is very similar to the Hagen.
Poiseuille solution for pipe flow. His equation for the frictional
prcasure gradient is:
474
SPE 25456 T. REED AND A PILEHVARI 7
As before, a force balance between the pressure gradient and the
opposing viscous shear stresses at the inner and outer boundaries of
an annulus can be constructed to obtain a relation between them.
The results are:
Tw,avg =(To Do + Tj Dj ) I (Do + Dj )
=dP I dL X (Do - D
j
) 14 (A-9)
This expression for average wall shear stress can now be used to
define the Fanning friction factor for annular flow.
F
c
=Tw,avg I (1/2 Py) =dP I dL x Dhy I (2 PY) (A-IO)
Next, we require this be the same function of Reynolds number as
for laminar pipe flow, Eq. A-7. This leads to the following equation
for Reynolds number.
N
Re
= P v D
L
2
1(p. Dhy) = Pv Deq I p. (A-U)
The various diameters are defined as:
Lamb's diameter:
D
L
= [00
2
+ Dl- (D
0
2
- Dl) Iln(D
o
I D
j
)]112,
see Eqs. A-I and A-S ..... (A-12)
Hydraulic diameter:
Dhy =Do - D
j
=Outer diameter-Inner diameter (A-l3)
Equivalent diameter:
Deq =D
L
2
1Dhy' Concentric Annulus (A-14)
An average wall shear rate can be defined that is consistent with Eq.
A-9 fOr the averagewall shear stress. For isothermal Newtonian flow,
the relation is:
Tw,avg = p. 'Yw,avg =p. (S v I Deq) (A-IS)
The term in parentheses for wall shear rate in an annulus is analo-
gous to Eq. A-4 for pipe flow.
The "Equivalent" diameter is the diameter of a circular pipe
that will have the identical frictional pressure drop as the
concentric annulus when the fluid is Newtonian and has the
same viscosity and the same average velocity (= flow ratel
cross-sectional area). The Equivalent diameter provides the
similarity link between pipe flow and annular flow. This concept is
applicable to transitional and turbulent flows if it is also stipulated
that the Newtonian fluid bas the same density. However, the next
step is to demonstrate how this diameter can be used when analyzing
non-Newtonian laminar flow in an annulus.
LAMINAR NON-NEWI'ONIAN FLOWS
Pipe Flow. Metzner and ReedIt and Dodge and Metznefl reported
some pioneering work in the 19S0s on non-Newtonian pipe flow.
Here we will briefly review their work and then extend their work to
the case of annular flows. Metzner and Reed developed the
following generalized expression for shear rate at the wall of a pipe
when the flow is laminar and the fluid has time-independent proper-
ties.
'Yw =[(3 N + I) 14 N] x (S v I D) (A-16)
Where "N" is defined as:
N = d[ln T
w
] I d[ln(Sv I D)]
= d[ln(dP I dL x D14)] Id[ln (Sv I D (A-17)
nNn is the slope of a log-log plot of the two variables (dP/dL D/4)
and (BvID). This slope varies with flow rate and rheological
properties. For a general time-independent non-Newtonian fluid,
Metzner & Reed used N to express wall shear stress as a function of
the Newtonian shear rate at the wall of a pipe, Eq. A-4.
T
w
=dP/dLxD/4=K'(Sv/D)N (A-IS)
In general, both K' and N are functions of Two In cases where the
slope of the log-log plot is not constant, N is the slope of a line that
is tangent at a particular point on the curve for In(dP/dL x D/4)
versus In(BvID). When N is a constant and equal to 1, K' reduces
to the Newtonian coefficient ofviscosity, and Eq. A-18 reduces to the
form of Eq. A-2 When N is greater than 1, the fluid is a dilatant,
and when N is less than 1, the fluid is a pseudoplastic. Hence, as
noted by Metzner & Reed, N is a measure of a fluid's non-Newtoni
an behavior. K' is a measure of the fluid's consistency and increases
as the fluid becomes more viscous.
In the case of a Power-Law fluid, Eq. A-18 becomes:
T
w
= dP I dLx D 14 = K' (Sv I Dt = K ('YW)B (A-19)
For this type of fluid, N =n =constant, and Metzner & Reed's
generalized consistency index becomes:
K' =K [(3n + I) 14n]B (A-20)
Here the AK" without a superscript is simply called the consistency
index. Note that Eqs. A-19 and A-20 are consistent with Eq. A16
for shear rate at the pipe wall. The expression for the generaIized
consistency index is more complex for a Bingham Plastic. The
derivation of K' for this type of fluid is presented in a clear and
straight-forward manner in a paper by Metzner.
Metzner & Reed defined a Generalized REynolds number (GRE)
by retaining the standard definition for friction factor and then
requiring it be related to Reynolds number by the classical relation,
Eq. A-7. They obtained the following equation for their GRE.
NRe,G =p ~ IK' SN-1 (A-21)
In the case of a Power-Law fluid, this equation becomes:
NRe,PL =Pv2- DB I {K [(3D + I) 14n]- S_1} .. (A-ZZ).
This equation can be put into a simpler form by introducing an
"apparent" Newtonian viscosity based on shear rate at the wall. The
apparent viscosity is defined by dividing wall shear-stress by wall
shear-rate. For a Power-Law fluid, the apparent viscosity bas the .
following form.
JLw,app =T
w
l'Yw =K [(3n + I) 14n x (S vI D)]"";1 (A-23)
. I"')'
When this expression for apparent viscosity is introduced uliO
Eq. A-22, the GRE for a Power-Law fluid reduces to:
475
8
A NEW MODEL FOR LAMINAR, TRANSmONAL, AND
TURBULENT FLOW OF DRILLING MUDS SPE25456
NRe,PL =P D V I #.&w" (3n + 1) 14n] (A-24)
Now, this equation for Reynolds number can be simplified even
further and put into the cIassical form for Newtonian flow by defining
an "Effective" pipe diameter which includes the remaining effects of
a Power-Law fluid. Hence, the obvious definition of an "Effective"
diameter for flow of a Power-Law fluid through a circular pipe with
diameter D is:
0eff =4n D I (3n + I), Power-Law pipe flow.. (A-25)
With this definition of the "Effective" diameter, the GRE for flow of
a Power-Law fluid through a pipe becomes:
NRe,PL = P V D
eff
IIJ.w,app (A-26)
By comparing Eqs. A-16 and A-25, it may be seen that the definition
for an "Effective" pipe diameter can be generalized to tbe case of any
time-independent fluid by simply replacing On" with N in Eq. A-25.
Hence, the generalized "Effective" pipe diameter is defined as follows.
0eff =4N D I (3N + I), Generalized Effective
pipediameter . (A-27)
Neither Metzner, his students, nor subsequent workers foresaw any
value in introducing the "Effective" diameter that is defined here. As
we shall see, it is indeed very helpful in extending their work on non-
Newtonian pipe flow to non-Newtonian annular flow.
Before continuing on to annular flow, an expression is needed for N
that is applicable to Herschel-BuIkIey (HB) fluids. Since N is defined
by Eq. A-17, a relation between wall shear stress and flow rate for an
HB fluid is necessary in order to evaluate N. The required relation-
Ship can be found in the book by Govier and Aziz.:IlI The resulting
equation for N is given below.
lIN = -3 + 'T
w
{(I + m)/[m ('T
w
- YS)] + 2('T
w
- YS)
[A(3m + 1)] + 2 YS/[A(2m + I)]} (A-28)
Where:
'T
w
=K' (8 v/D)N =YS +K <'Yw )IIl, HB fluid .. (A-29)
YS =Yield Stress
m =Exponent on shear rate is different from a -
Power-Lawfor clarity; Hanks%7... (A-30)
A = ('T
w
- YS)2/(3m + 1) + 2 YS ('T
w
- YS)/(2m + 1)
+ YS
2
/(m + 1) .. (A-31)
Concentric Annular Flow. Based on a cursory examination of the
foregoing discussions, it might appear that the work of Metzner &
Reed
l
' on non-Newtonian pipe flow could be applied to concentric
annuli by combining the "Equivalent" and the "Effective" diameters
and using the average wall shear rates and shear stresses in place of
the corresponding parameters at the wall of a pipe. However, a
solution for non-Newtonian annular flow is more involved. Fortu-
nately, a simple solution has been previously developed for laminar
flow of a Power-Law fluid through concentric annuli. The needed
solution appears in a book by ExIog's staff.
21
Their solution consists
of the following equations.
Y =0.37 (A-32)
Z = 1- [1- (Oil D
o
)Y]l/
Y
(A-33)
G =(1 + Z/2) [(3 - Z) n + 1] 1[(4 - Z) n] .. (A-34)
dP I dL =(4 KI DIIy) [8 v G I DIIy]Il (A-3S)
The authors state that this correlation is based on the analytical
solution originally obtained by Fredrickson and Bird.t This equation
can be related to Power-Law pipe flow by rearranging Eq. A-35 into
the form of Eq. A-19, viz.:
'T
w
=dP I dL x Oily 14 =K (8 V G I OIly)1l
=K (8 vI Oeff)1l (A-36)
Hence, the "Effective" diameter for laminar Power-Lawflow through
a concentric annulus is more complicated than simply inserting Deq
into Eq. A-25. It is defined by Eq. A-36 to be equal to:
=Oily I G =Effective Diameter for non-
Newtoman flow through Concentric Annuli ... (A-37)
We have designated Eq. A-37 as being valid for a general non-
Newtonian fluid because G can be generalized by simple replacing
On" with nNw in Eq. A-34. This Effective Dianleter is a function of
both the annular geometry and the rheology of the fluid. and it
provides the link between non-Newtonian annular flow and Newtoni-
an pipe flow.
The ExIog correlation can also be applied to flow of an HB fluid by
noting:
T. - YS =[dP/dL]HB X DIIy 14 - YS == K (8 v/Deff)8l (A-38)
or
[dP I dL]HB x DIIy 14 =YS + K (8 v I D
eff
)1Il ..... (A-39)
Hence, the utility of the ExIog correlation can be extended by simply
adding the Yield Stress to the average wall shear stress created by
movement of the fluid and (1) replacing On" with "N" in the definition
of G, Eqs. A-32 and A-34, and (2) replacing On" with Om" in
Eq. A-35.
A summary of the most relevant equations are given below for non-
Newtonian laminar flow of an HB fluid through an annulus.
N =d[In I d[In (Sv I Deq)]
= d[dP I dL x DIIy 14)] Id[In (Sv I Oeq)] (A-40)
'YWVI =(8 vI Dar> (A-41)
'TW,aYI = dP I dLx Oily I 4 = YS + K <'YW,aYl)1II .. (A-42)
1J.w.eff == 'Tw,aYl/-Yw,aYI (A-43)
NRe,G =P v Oeff/1J.w,eff (A-44)
The friction factor for annular flow is not affected and remains the
same, Eq. A-10. In addition, the relation between friction f8ctor and
the GeneralizedReynolds number, Eq. A-7, also remains unchanged.
In order to verif)' these results, an example calculation was done
using this model, and the predicted frictional pressure gradient was
compared with the corresponding result from the recent finite-
476
SPE 25456 T. REED AND A Pll..EHVARI 9
difference solution by Haciialamoglu and Langlinais.' They baYe
computed the frictional pressure gradient generated byflowofan lIB
fluid through a concentric annulus for the following conditions: Do
= 10 in. [25.4 em], D
i
= 7 in. [17.8 em], K = 250 eq. cP, "m" =
0.70, YS = 51bf/100 fi2, and a pump rate of 200 gpm [12.6 IJs].
For these conditions, the authors report a calculated pressure
gradient of 0.00870 psiIft. Using the method outlined herein, a
prcssure gradicnt of 0.0086 psiIft is obtained. This conclusively
disproves the claim by Hanks that thc Generalized Reynolds
Number cannot be applied to annular flows.
TURBULENT NEWTONIAN FLOW
Jones and Leuni' proved thatthc equivalent diameter, as defined by
Eq. A-14, could also be used in turbulent flow to relate concentric
annuli to circular pipes. They assembled a range of test data and
showed frictional pressure gradients in a smooth annulus could be
correlated by using thc "Equivalent" diameter in thc classical
Colcbrook for smooth pipes. This leads to the next
question: Can this be extended to non-Newtonian flows? In order
to answer this question, we next tum our attention to the work by
Dodge and Metzner on nOD-Newtonian pipe flows.
TURBULENT NON-NEWTONIAN FLOW
Pipe Flow. Dodgc and Metznefl were able to correlate turb\llent
pipe-flow data for a varicty of nOD-Newtonian fluids.
accomplished by using the GREof Metzner and Reed and modiJYing
the Colebrook Equation. They proposed the following form of
Colebrook's Equation for all time-independent fluids flowing through
smooth circular pipes.
1 / F
r
1l2
=(4 / Nl7S) log FP - Nfl)] - 0.40 / N1.2(A-4S)
In order to apply this equation correctly, the apparent viscosity must
be computed in a particular way. Dodge & Metzn,:r emphasized !b&t
the shear rate in Eq. A-23 must be based on an unagmary laminar
flow velocity that will generate the same wall shear stress as the
turbulent flow. Hence, an iteration is required in order to obtain the
correct apparcnt viscosity.
Concentric Annular Flow. This equation can also be applied to flow
through concentric aDDuli by simply using Deff, Eq. A-37, in the
GRE, Eq. A-44, and again defining the friction factor to be based on
the avcrage wall shear stress or frictional pressure gradient in an
annulus, Eq. A-10. Similarly, values of N are calculated by using thc
average wall shear stress in Eq. A-28. In. addition, the average wall
shear rate must be based on a laminar flowvelocity that will generate
the same wall shear stress as the turbulent flow. This imaginary
velocity can be computed by setting 1"w in Eq. A-38 equal to the
turbulent strcss and iterating. This vclocity is then used to calculate
the corresponding laminar shear rate, Eq. A-41, and an cffective
viscosity, Eq. A-43, which is used in the dcnominator of the General-
ized Reynolds Number. With these changes in definitions of
Reynolds number, friction factor and N, Eq. A-45 can be used to
compute turbulent pressure gradients for flow of a lIB fluid through
concentric annuli.
Effects of Wall Ro......ess. Colcbrook's original equation for
Newtonian turbulent flow also includes a term to account for the
effects of wall roughness.
2S
Colebrook's equation is equivalent to thc
turbulent portions of the well-mown Moody Diagram and can be
written in thc following form.
1 / F
r
1l2 =- 4 log [(0.27 E / D) + 1.26 / (N
Re
Fr)](A-46)
477
The relative wall roughness is specified by the ratio "E I D".
The corresponding equation ofDodge &Metzner for turbulent nOD-
Newtonian flow, Eq. A-45, can be rearranged into this same form.
This results in a ncw equation for turbulent friction factors which
combines nOD-Newtonian and wall-roughness effects.
l.2
1/ F
r
1l2 =- 4 log [(0.27 E / D
eff
) + 1.26
r
/

FP - Nf2) ] ... (A-47)
This extended form of the Colebrook equation can be applied to
non-Newtonian flow through pipes and concentric annuli by simply
using the correct forms for the frictiOD factor, the Effective Diame-
ter, and the Generalized Reynolds number.
There is a legitimate question about whether the coefficient for thc
roughness term (0.27) should be a function of N or not. Note that
the Effective Diameter is a function of N, and in the case of
pseudoplastics (N < I), a given wall roughness, Eo will cause a greater
percent increase in the frictional pressure gradient than occurs in
Newtonian flow. It is uncertain whether this is sufficient to properly
account for the influence of non-Newtonian fluids on wall roughness
cffects. Additional tests with bentonitc muds in rough pipes (E =
0.0047 in. [0.119 mm] from water tests) show good agreement with
Eq. A-47 up to Reynolds numbers of 270,000. However, additional
tests with systematic variations of mud properties and relative rough-
ness are needed before Eq. A-47 can be verified conclusively. In the
meantime, this equation is proposed for calculation of frictional
pressure gradients in turbulent non-Newtonian flow through pipes
and concentric annuli with either smooth or rough walls.
TRANSmONAL FLOW
Geometric Effect&. Hanks
1
and Hanks and Pctcrson
lS
havc rcported
experimental measurements of transitional flow through concentric
annuli for Newtonian fluids. Reference 15 presents experimental
data for concentric annuli with four different ratios of iDDer-to-outer
diameter (aspect ratio). The reported values of Reynolds number at
transition are plotted in Fig. 1. Hanks also included the results from
a transition theory that he first began developing back in the early
1960s."'6,17 His predictions for the complete range of aspect ratios is
shown in Fig. 1 as a dashed line. Hanks' theory shows a maximum
in transition Reynolds number at an aspect ratio of about 0.15.
However, thc experimental data docs NOT corroborate the trend of
Hanks' theoretical predictions.
The solid curve in Fig. 1 is based on the "Equivalent" diameter of
Jones & Leung
13
, Eq. A-14. If the critical Reynolds number is based
on this diameter and set equal to 2100, the following equation is
obtained.
(NRe,Eq)c:rit =DEq P V / ,.,. =2100 (A-48)
This equation can be rearranged into a Reynolds number based on
Lamb's diameter, Eq. A-l2.
(N&e,0c:rit =D
L
P v / ,.,. =2100 x Dhy / D
L
......... (A-49)
This provides a transition ReynOlds number that varies with aspect
ratio and reduces to the correct relation for Simple pipe flow. This
equation is plotted as the solid curve in Fig. 1. It should be noticed
that the Reynolds number based on Deq, Eq. 54, does exactly what
it is supposed to do in that it successfully combines the critical
Reynolds number for all concentric annuli into an equation that has
the same form as the classical relation for pipe flow. Hcnce, the
10
A NEW MODEL FOR LAMINAR, TRANSmONAL, AND
TIJRBULENT FLOW OF DRllJ.JNG MUDS SPE2S456
similarity analysis properly accounts for geometric effects. As before,
this idea can be extended to include non-Newtonian effects by using
the "Effective" diameter, Eq. A-37.
Non-Newtonian Etl'eet&. In Hanks' 1963 -paper, he used the same
theory as he used for Newtonian flow through concentric annuli to
predict transition Reynolds numbers for Bingham Plastics in pipes.
He plotted his results in terms of a critical Bingham Plastic Reynolds
number. The BinghamPlastic Reynolds number for a pipe is defined
as:
NRe,BP = D Pv / (Plastic Viscosity) (A-50)
Note that this is not a correct measure of Reynolds number effects
because the Plastic VISCOSity (PV) is coiJstant for a given fluid, and,
in reality, the apparent viscosity will vary with shear rate, Eq. A-43.
Hanks' critical values for the Bingham Plastic Reynolds number are
shown in Fig. 3 as a function of Hedstrom number. The "critical"
Reynolds number is a measure of the flow rate at which laminar flow
ends and transition begins. The Hedstrom number is a measure of
the influence of a fluid's yield stress on the flow. For the case of a
Bingham Plastic flowing through a pipe, it is defined as:
-He = Hedstrom No. = p yp l)2 / (pV)2 ...... (A-51)
Hanks' theoretical values ofcritical BinghamPlastic Reynolds number
appear to agree with measured data up to a Hedstrom number of
about 5,000 as plotted in his original 1963 paper.
z
At higher values,
his theory diverges from the data, see Fig. 3. In contrast, the solid
curve in this figure passes through the experimental data. This curve
was generated by utilizing the Generalized Effective diameter for a
pipe, Eq. A-27, and the correct "apparent" viscosity to define a
Generalized Reynolds number, Eq. A-26.
Next, the new transition criterion requires the critical condition for
any fluid to occur when the product of the Fanning friction factor and
the GREequals 16.1. This generalized transition criterion is designed
to reduce to a critical Reynolds number of 2100 for Newtonian pipe
flow. The GRE for pipe flow is related to the BP Reynolds number
via the equation below.
Where the apparent viscosity for BinghamPlastic pipe flow is defined
by:
J.I.w,app =(YP +PVx'Yw) / 'Yw (A-53)
The wall shear rate is defined by Eq. A-16 for pipe flow, and N, given
by Eq. A-28, reduces to the following expression for a Bingham
Plastic.
N
BP
= (3 - 4X + X
4
) /3(1 - X
4
), (A-54a)
with: X = yP / T
w
(A-54b)
This agrees with the results of Metzner' for BP fluids and, again,
demonstrates the consistency of the similarity analysis.
Friction Factors through the Transition Regime. In a 1977 paper,
Churchill
30
developed a simple method for calculating friction factors
through the transition zone in Newtonian pipe flow. Churchill
devised the following procedure for combining laminar and turbulent
friction factors in order to calculate friction factors through the
transition zone. He first defined an intermediate term based on the
transitional and fully-turbulent friction factors; this termis designated
F
1
and is defined by:
(F
1
)-8 =(F1r)-8 + (F
tudl
)-8 (A-55)
F
1
is then used in a similar equation involving the laminar friction
factor to compute friction factor through the transition zone.
(Ff)12 = (F
1
)12 + (F
Iam
)12 (A-56)
This equation can then be solved for friction factor at !!!!! Reynolds
number or flow regime.
Churchill selected the following functional form for friction factor in
the transition zone.
Fir =Const. x N
Re
2
(A-57)
Based on the numerical computations of Wilson and Azad
31
Churchill chose a value of 1.42x for the constant when the fluid
is Newtonian. We have extended Churchill's analysis to non-
Newtonian fluids by replacing the constant coefficient with a function
of "N". An analysis of the friction factor curves of Dodge and
MetzneeZ1 for "n" =I, 0.8, 0.6 and 0.4, Fig. 2, and using the generally
accepted criterion of 2100 for transition in Newtonian pipe flows led
to the following equation for this coefficient.
C(N) = 9.4 x 10-
9
/ [4.767 - 2.167 N] 2 (A-58)
This equation should be used in place of the constant coefficient in
Eq. A-57. In addition, the Reynolds number can be replaced with
the GRE. This leads to the following equation for transitional
friction factors in pipe and concentric annuli.
Fir = C(N) x (NRe,G) 2 (A-59)
This is a new relationship for transitional friction factors in circular
pipes and concentric annuli. Note from Table 1 that "N" varies
through the transition zone. Furthermore, by incorporating "N" in
Eq. A-58, the equation applies equally well to Newtonian, Power-
Law, Bingham Plastics, and Herschel-Bulkley fluids.
When this definition of friction factor for the transition zone is used
in Churchill's procedure, Eqs. A-55 and A-56, the product of friction
factor and GRE is equal to 16.1 when the GRE = 2100 for
Newtonian pipe flow. This requirement produces increasingly higher
pump rates for the initiation of transition as yield stress of a fluid
increases. Transition is assumed to be complete when the friction
factor from Eq. A-56 is within 1 percent of the turbulent friction
factor from the extended Colebrook equation, Eq. A47.
478
I Fig. 3 Transition in Bingham Plastic plpeflows. I
Thousands
I
en

!""l

i>

.....
.....
10
8
1.0
0.8
0.8
....... 0.4
.......:.... :... ; ..:..:.:.:.:.
' ' ..'. ,'. , .'. '.
. . . . . . . .
; : :..:. .:.:.
. . .
..:..1..
10
4
10
6
HEDSTROM NUMBER
..1.
....... ; ...
1:21 ... "-\.
Fig. 4 - Critical Generalized Reynolds Number for
Btngham Plastic pipeflows
l-
f-'
..--
I--
I
1,.-1-'
V
I---
FFORE .. 18.1
l/
I

i-o'f:
----

I
..
..
..
..
..
I
..
..

HANKS (1883) I
..
..
..
I
1800
1500
1200 3
10
2400
2100
2700
3000
I Fig. 2 - Fiction factor. for Power-Law plpeflows. I
0.1 c: :.. ,.; :..:.: .:.. 1 :. L: :.1 : J
a:
t
if
z
o
! ...
!
z
z
if
' ' -. 0' _._.
.. ....
0.001' I I I I 1 I III I I I I I I II' I I I I I! II!
10
2
10
3
10
4
10&
GENERALIZED REYNOLDS NUMBER
.......:.... ;... ; ..:..:.:.:.:.
o
z
>=
W
II:
o
W
N
::::i
<
II:
W
Z
W
CJ
oJ

t=
it
o
2285
2672
10
8
<> HANKSl19801
o HANKSPETERSONI19821
10
4
10
6
HEDSTROM NUMBER
............................... 1 I
j
TEST DATA RANGE ... .-




HANKS (1963)
.....--
I . I I I
1600 I i I I J I I I I
o M U U U
INNER DIAMETER I OUTER DIAMETER
1800
22
20
o 18
z
16
...
o 14
z
ti 12
a:
g; 10
... 8
:J
i= 6
5 4
2
o
10
3
Fig. 1 - Transition in Newtonian Annular flow.
""
a
.=
g 2700 ; : : : : :- .
rn 0:<>
Q :
I ...,.:.: :.,.- --; --" ,.. "C, '.'.':'.'.' ,.,., ,.:.:.
z 2100' _.. '. .. HANKSf1980t :
2 : : : ; .
to-
iii
I
to-
.j:l.

......
tv
;I>


t::Dt;;





.... - .
....................................
12-1/4" x 5" Annulus
Example Mud Properties
300 < Pumprate < 2100 gpm
............ : : : .
- - -
. . . .
...................................................
I I
..........................................................
. . .. - . .. . .. .
'-:._..
...... ---1 n-ensltlon. 1910 -_.:-- ..
100 200 300 400 500 800

ROTATIONAL SPEED OF VISCOMETER RPM


Fig. 6 - Comparison of Viscometer data for typical
water-benntonlte mud with three rheological models.
0.6( I I I I I I
f!!
Z 0.5

2
0.4
II..
o
20.3

.' 0.2
E

-
Z0.1

80
'"
70
...
-
0
80
0
,..
.....
HED
-
10
3
:e &0
10
4
C!)
10&
Z
a
5x10& c(
W
a:
-I
c(
a
10&
................ \ ............
. . .
.- - - .
... ; .....
...

::1i ansition : - - - - - :.
_.
. . .
.......... : .
........ :. :-:.:.
". ," t : .' - .: .... r ",.,"
..:.... :... ..:. -: . : -:.:-
10
3
10
4
GENERALIZED REYNOLDS NUMBER
................ ... '., . . .. . .. '.' ." .........
Fig. 6 - Effects of Hedstrom Number on
transition of Bingham Plastics In plpeflovv.
... -..... ..... -
0.01
cr:
0
l-
e,)

Z
0
j::
e,)
iii:
II..
"
Z
Z
z

0.1
a:

z
0
0.01
it
IL
CJ

Z
Z

0.001
10
2

o
0.001 3
10 10
4
10
5
GENERALIZED REYNOLDS NUMBER
Fig. 7 - Effects of three rheological models
on flow transition In pipes .
0
'
I I I I I I
o 50 100 150 200 260 300
AVERAGE ANNULAR WALL SHEAR RATE (1Isec)
IFIG. 8 EFFECTS OF SHEAR RATE ON EXPONENT "N"I

0\
1.2, , 260. i iii ii' i
en
til

0\
121/4" x 6" Annulus
Example Mud Properties
.................. .................
........ 1
.............

fI)
0
160
(J
!!
>

ti

tb 601- .. ,
............................
. . .
121/4" x 5" Annulus
Example Mud Properties
.....: : ; : : ; : : : .
1
0.2
0.8
I
-0.6

0.4
Fig. 9 - EFFECTS'OF SHEAR ItATE EXPONENT
ON EFFECTIVE DIAMETER OF ANNULUS

i>
'"d

....
21
...
19
t'
6" 23/8" Annulus
MMH Mud System
, to
7 9 11 13 16 17
PUMPRATE 1 100 IGPM)
5
0 "
..................... , , , .
. . . . . .
......;. .. -;":- .:. ; .. -: :- .:. :- .. .. ;..; -:.; .:-:.,
0' , I I I , , , I ,
3
Fig. 10 - EFFECTS OF PUMPRATE ON
EFFECTIVE VISCOSITV IN ANNULUS
0.1\ : : : :'S;:: ::i: :; :;:;: i :;:;:;: :::::; :::;: ::;: :;: >:;;; ::::::;: ::;: :;:; :;: :::;1
c:J
z
i
z

a:
e

IL
Z
o
ti 0.01
iii!
Yo
1
.; :- .;. -:-:.:-;. :- .. -:.. : -:. : .:-:.,
12-1/4" x 6" Annulus
Example Mud Properties
t
..... .:. .:-:-: ;. ',:::: ::: ':--:'
0.11'" .?S.;., '. " '..' "'l
:::::-,: . : :::::::: :::::::: :::::: ::::: :::: ::: :::: ::::::::::: ::::::::::: : ::
0' , I J I II I I I I
o 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
"N" GENERALIZED SHEAR RATE EXPONENT
a:

z
0
ti 0.01
iii!
Yo
c:J
z
i
z

it
.....
I I I ) 5
0.001 2 3 104 10
10 10
I I , 6
0.001 ' 2 3 104 10
10 10
GENERALIZED REYNOLDS NUMBER GENERALIZED REYNOLDS NUMBER
FIG. 11 - FRICTION FACTORS FOR ANNULI
WITH SMOOTH AND ROUGH WALLS
FIG. 12 - COMPARISON OF MODEL WITH TEST
DATA FOR MMH IN A ROUGH ANNULUS
......
!.

TABLE 1
PARAMETERS FOR TRANSITION ZONE IN ROUGH 12-Yl X 5" ANNULUS
------- --- - - - - -- ----
MUD FANNING

VELOCITY FRICTION GENERAL.

RELATIVE FPGRAD
(GPM (FfIMIN)- "N" FACI'OR REY. NO. W"GRE FLOW TYPE (l/SEe ROUGHNESS (In.H,llIFt)
1860. 364.5 .3609 .6151E-02 26042E+03 16.02 LAMINAR 186.2 .015627 .4034
1870. 366.5 .3617 .6098E-02 26284E+03 16.03 LAMINAR 187.1 .015607 .4043
1880. 368.4 .3625 .6046E-02 26S29E+03 16.04 LAMINAR 188.0 .015588 .4051
1890. 370.4 .3634 .5996E-02 26781E+03 16.06 LAMINAR 188.9 .015567 .4060
1900. 3724 .3643 .5947E-02 27042E+03 16.08 LAMINAR 189.9 .015545 .4070
1910. 374.3 .3653 .5901E-02 27314E+03 16.12 lRANSIT 191.1 .015522 .4081
1920. 376.3 .3664 .5857E-02 27603E+03 16.17 lRANSIT 1924 .015497 .4093
1930. 378.2 .3679 .5819E-02 27947E+03 16.26 lRANSIT 194.1 .015462 .4110
1940. 380.2 .3699 .5790E-02 28359E+03 16.42 lRANSIT 196.4 .015417 .4131
1950. 3822 .3741 .5797E-02 29083E+03 16.86 lRANSIT 201.4 .015321 .4179
1960. 384.1 .5574 .1061E-Ol 6.0752E+03 64.46 lRANSIT 634.7 .012561 .7728
1970. 386.1 .5595 .1060E-Ol 6.1540E+03 65.25 TIJRBULENT 644.6 .012539 .7802
1980. 388.0 .5617 .1060E-Ol 6.2329E+03 66.05 lURBULENT 654.7 .012517 .7876
1990. 390.0 .5637 .1059E-Ol 6.3121E+03 66.84 lURBULENT 664.8 .012496 .7950
2000. 3920 .5658 .1058E-Ol 6.3915E+03 67.63 lURBULENT 674.9 .012476 .8024
2010. 393.9 .5678 .1057E-Ol 6.4711E+03 68.43 lURBULENT 685.2 .012456 .8099
2020. 395.9 .5699 .1057E-Ol 6.5509E+03 69.23 lURBULENT 695.5 .012436 .8175
2030. 397.8 .5718 .1056E-Ol 6.6310E+03 70.03 lURBULENT 706.0 .012417 .8251
2040. 399.8 .5738 .1055E-Ol 6.7112E+03 70.83 1URBULENT 716.5 .012398 .8327
2050. 401.8 .5757 .1055E-Ol 6.7917E+03 71.63 TIJRBULENT 727.1 .012379 .8403
2060. 403.7 .5776 .1054E-Ol 6.8724E+03 7244 lURBULENT 737.7 .012361 .8480
2070. 405.7 .5795 .1053E-Ol 6.9533E+03 73.25 lURBULENT 748.5 .012343 .8557
2080. 407.6 .5814 .1053E-Ol 7.0344E+03 74.06 lURBULENT 759.4 .012326 .8635
2090. 409.6 .5832 .1052E-Ol 7.1157E+03 74.87 lURBULENT 770.3 .012309 .8713
2100. 411.6 .5850 .1052E-Ol 7.1972E+03 75.68 lURBULENT 781.3 .012292 .8791
......

>-


tl:i
Cr.1
hjt'"'




tJ-
n
OS

en
til

You might also like