AGeneralized Rheological Modelfor Drilling Fluids With Cubic Splines

See discussions, stats, and author profiles for this publication at: https://www.researchgate.
net/publication/267461570
A Generalized Rheological Model for Drilling Fluids Using Cubic Splines
Article in SPE Drilling & Completion · April 2014

DOI: 10.2118/169527-MS
CITATIONS READS
10 979
2 authors:
Binh T. Bui Azra Tutuncu

PetroVietnam University Colorado School of Mines
36 PUBLICATIONS 290 CITATIONS 113 PUBLICATIONS 1,586 CITATIONS
SEE PROFILE SEE PROFILE
All content following this page was uploaded by Binh T. Bui on 04 December 2022.
The user has requested enhancement of the downloaded file.

DC169527 DOI: 10.2118/169527-PA Date: 17-February-16 Stage: Page: 26 Total Pages: 14
A Generalized Rheological Model for

Drilling Fluids With Cubic Splines
Binh T. Bui and Azra N. Tutuncu, Colorado School of Mines
Summary in a single direction is widely used. For simplicity, in the next sec-
In drilling-fluid rheological characterization and hydraulics mod- tion, the relation between shear stress and shear rate, or the flow-
eling, selecting a proper rheological model and obtaining rheologi- curve equation, is written in 1D form.
cal parameters with viscometers are critically important. Bingham In the early days, the linear correlation, the Newton model,
plastic, power law, and yield power law are the most commonly was the most popular model. In simple shear, the response of a
used standard models for drilling-fluid rheology because they are Newtonian fluid is described by a linear relationship between the
mathematically simple to use. However, because of the complex applied shear stress and the shear rate. After advancements of
nature of drilling fluids, these models do not often fit well to the modern viscometers and rheometers, researchers showed that
rheological data. This leads to a significant error in drilling-fluid complex fluids do not obey this linear Newtonian relationship.
hydraulics modeling. While searching for models that are based on These fluids are classified as non-Newtonian fluids. Non-Newto-
the physical interaction between the various components of the nian fluids are those for which the flow curves are not linear (i.e.,
drilling fluid, it is very practical to find simple mathematical func- the viscosity is a complex function of several factors including
tions that best fit to any experimental data obtained from field visc- shear rate and shear history). Non-Newtonian fluids are often sub-
ometers. This will improve the accuracy of hydraulics modeling— divided into three groups. These are time-independent, time-
in particular, for time-dependent drilling fluids. dependent, and viscoelastic fluids (Wilkinson 1960). Various cor-
In this research, cubic splines were used to fit the experimental relations were developed to model the rheological response of
data obtained from field viscometers. A generalized hydraulics non-Newtonian fluids. The most common models are the Bing-
model is presented to calculate pressure drop in pipe and concen- ham plastic, power law, Herschel-Bulkley, Carreau-Gahleitner,
tric annulus with detailed sample calculations. A numerical simu- Casson-Steiner, Casson, and Ellis models (Fig. 1). A statistical
lator was also developed to assist the calculations. The results review of these models for the petroleum-industry applications
obtained from the model were validated with experimental data to was published by Weir and Bailey (1996) and Bailey and Weir
verify the effectiveness of the proposed model. (1998). One of the common advantages of these models is that
One can implement the model introduced in this paper for any they have a simple mathematical form. In other words, the num-
fluids. The main advantage of the model is that it can capture all ber of model parameters is limited and can be obtained from a
the complex rheological response of the fluid. Hence, all readings small number of experiments. It is sometimes easier to obtain ana-
obtained from the viscometers can provide valuable input for lytical solutions in hydraulics problems. However, these models
the hydraulic model. Also, the numerical solution is stable and do not always give a good fit to the rheological data, and the
straightforward to implement. The model provides drilling engi- hydraulics calculations may be ill-fated.
neers a simple and powerful method to accurately predict the Because of the complex components of drilling fluids and the
pressure distribution along the wellbore. This is highly important diversity of drilling-fluid systems, drilling fluids are normally clas-
in drilling-fluid hydraulics-program optimization and well-control sified as non-Newtonian fluids. Also, drilling fluids often show
operation, especially in deepwater and Arctic environments. time-dependent properties and behave viscoelastically under cer-
tain conditions (Bui 2012). This explains why rheological charac-
terization of drilling fluids at different measurement conditions
Introduction and procedures gives different results. The time-dependent proper-
Accurate determination of the pressure along the wellbore is im- ties result from the interaction among various components of the
portant for maintaining the safety and efficiency of the drilling drilling fluid and the complex response of each component under
operations. It contributes to a better design of hydraulics pro- different shearing conditions. These components are often differ-
grams, suitable selection of mud pumps, and optimization of hole ent in size, chemical activity, specific gravity, and surface energy.
cleaning. Precise pressure-drop calculation also provides a better One more important phenomenon that often makes the rheological
control of the pressure in the wellbore and increases the rate of response of the drilling fluid complex is gelation. To fulfill its
penetration (Maglione et al. 2000). The accurate prediction of the complex functions, such as cuttings transport and solid suspension,
equivalent circulating density and annular velocity is critical for structure buildup or gelation is a desired property of the drilling
preventing kicks, cutting settling, and maintaining the continuity fluids. The gel formation, when the fluid is at rest, helps to keep
of drilling process (Bailey and Peden 2000). The selection of the the solid particles from settling. Gel-structure formation helps to
rheological model is one of the most important first steps in dril- prevent fluid invasion into the formation and lost-circulation prob-
ling hydraulics modeling. For fluids, the rheological model is of- lems (Bui 2012). Therefore, the response of drilling fluid in rheo-
ten the constitutive equation describing the relationship between logical measurements is very complex. Experimental data are
the stress and rate of deformation, shear rate. In literature, many repeatable with some degree of certainty if experimental procedure
theoretical and experimental studies were conducted to determine and all the conditions are the same (Bui 2012). Also, the flow
this constitutive relationship (Mezger 2006). The relationship curve of drilling fluids is very complex, and simple rheological
between stress and deformation (shear strain, shear rate) should models do not often fit well with experimental data.
be written in tensor form because these quantities are tensor quan- In the petroleum literature, many attempts were made to better
tities. However, in drilling-fluid applications, uniaxial shear, or characterize the rheological response of the drilling fluids. Most
simple shear, is commonly encountered. The constitutive equation of the rheological models that are currently used in petroleum en-
gineering are inherited from chemical and mechanical engineer-
ing. A summary of the common models used in literature is
Copyright V
C 2016 Society of Petroleum Engineers
presented in Table 1. One can obtain a complete list of available
This paper (SPE 169527) was accepted for presentation at the SPE Western North models in Mezger (2006). Weir and Bailey (1996) also provide
American and Rocky Mountain Joint Meeting, Denver, 17–18 April 2014, and revised for
publication. Original manuscript received for review 1 December 2014. Revised manuscript
their modification for models such as Colins-Graves, hyperbolic,
received for review 26 August 2015. Paper peer approved 7 October 2015. Robertson-Stiff, and Sisko models to account for the yield stress.
26 March 2016 SPE Drilling & Completion
ID: jaganm Time: 15:22 I Path: S:/DC##/Vol00000/150031/Comp/APPFile/SA-DC##150031

200
Pa Newton
180 Bingham Plastic
170 Power Law
160 Herschel-Bulkley
150 Carreau-Gahleitner
140 Casson-Steiner
130
Casson
120
Ellis I
110
Shear Thickening
100
τ Shear Thinning
90
80
70
60
50
40
30
20
10
0
0 50 100 150 200 250 300 350 400 450 500 550 600 650 700 750 800 850 900 1/s 1,000
.
Shear Rate γ
Fig. 1—Commonly used rheological models in the petroleum industry.
The Prandtl-Eyring, Reiner, and Cross models are commonly able models for modeling of the drilling fluids. The Casson model
used to model the flow behavior of complex fluids. However, is recommended by Wang and Tang (1982), Houwen and Geehan
these models are not commonly used in drilling-fluid calculations. (1986), Tutuncu (1988), Hughes et al. (1993), and Davison et al.
The most commonly used models in drilling-fluid rheological (1999). The Carreau model is also used in the drilling industry to
characterization are Bingham plastic, power law, and yield power model the low shear viscosity, or zero shear viscosity of drilling
law. Bingham plastic model can capture the yield stress that is of- fluids, especially polymeric-fluids inverted emulsion muds. The
ten observed in drilling fluids, yet does not show the shear thin- modification of this model is also introduced by Elise (1995).
ning characteristic of drilling fluids. Bingham plastic model is Okafor and Evers (1992) provided a comparison of several mod-
often used for low-density and low-viscosity drilling fluids. Mel- els including Robertson-Stiff, power law, and Bingham plastic
rose and Lilienthal (1951), Briscoe et al. (1996), and other with the experimental data and suggest that Robertson-Stiff gives
researchers use Bingham plastic model to characterize the rheo- the best fit to their rheological and hydraulics data. Bailey and
logical behavior of bentonites. Power law model, on the other Peden (2000) showed the applicability of Sisko model whereas
hand, does not have the yield point, but shows the shear thinning Wang and Fan (2010) show that modified Sisko model, so-called
behavior and is often used for fluids with low yield stress (Wilkin- Tscheuschner model or four-parameter model, also fits well to the
son 1960). Drilling fluids often show both yield point and shear experimental data.
thinning properties. Hence, the yield power law model or Her- It is evident that most available rheological models were origi-
schel-Bulkley model has become the most commonly used model nally developed as empirical correlations. They are mathematical
in drilling-fluid calculation. equations that give a good fit to the experimental data. The models
There are also rheological models developed in the petroleum- with more parameters often fit the experimental data better than
engineering literature for characterizing the rheological properties the models with fewer parameters. There is also limited effort to
of drilling fluids. Savins (1958) proposes a “pseudoplastic general- derive constitutive equations that are based on the concentration
ized Newtonian” model by modifying Bingham Plastic model. The and the physical and chemical interactions between the different
model is later used by the author (Savins 1962) to show a good fit components of the fluids. One can use this approach for a simple
to the experimental data. Robertson and Stiff (1976) proposed a mixture in the absence of the experimental equipment for measure-
new rheological model that fit well to their experimental data from ments. However, it is not very practical for drilling-fluid applica-
Fann 35 and Fann 39 viscometers. Beirute and Flumerfelt (1977) tions because the drilling fluids have a very complex nature and
show that Robertson-Stiff model gives a better fit to their experi- also their rheological characteristics are strong functions of both
mental data for drilling fluids and cement slurries than the yield pressure and temperature. Hence, making a best use of the avail-
power law model. They also introduce a modified Robertson-Stiff able experimental data is a more practical approach. Although
model to match their experimental data. Graves and Collins (1978) many complex rheological models were proposed and applied for
presented a new rheological model to capture the plug-type flow of drilling fluids, no model provides a perfect match to all experimen-
non-Newtonian fluids. A modified Graves Collins model is also tal data points from the field viscometers and is applicable for all
proposed later by Aadnoy and Ravnoy (1991). Gucuyener (1983) kind of fluids. This often results in an inaccurate determination of
proposes a new model, often referred to as Gucuyener model, to the pressure loss and other issues related to pressure control and
model the flow curve of drilling fluids and cement slurries. Pileh- poor cuttings transport. Hence, determination of the rheological
vari et al. (2001) propose rational polynomial model and show a model that best fits to experimental data is critically important.
reasonable agreement to their experimental data. A five-parameter In this research, cubic splines were used to fit the rheological
model, called NE (novel equation), is presented by Nasiri and Ash- data from viscometer measurements to construct the flow curve.
rafizadeh (2010) to represent their experimental data and model the Cubic splines are splines composed of piecewise third-order poly-
flow of drilling fluids in annulus. Vipulanandan and Mohammed nomials which pass smoothly through a set of data points. One of
(2014) show that a hyperbolic model fitting well to their experi- the main advantages of the cubic spline model is that it can give a
mental data obtained from acrylamide polymer modified bentonite good agreement to any set of experimental data. None of the API
drilling-fluid experiment results. recommended standard rheological models such as Power law and
Besides the formulation of the rheological models for drilling yield power law models gives a perfect fit to any set of data. A
fluids, some research studies focus on the applicability of avail- perfect fit to all data points from the field viscometer is very
March 2016 SPE Drilling & Completion 27

Table 1—Common Non-Newtonian-fluid models.
important because the rheological response of the fluid at different applicability of the introduced model. The physics of how the
shear rates can be captured. Making good use of all measured data interaction between different components affects the rheological
points improves the accuracy of the hydraulic calculation. Even behavior of drilling fluids is not within the scope of this research.
though the number of equations and parameters of the cubic The main objective of this research is to reveal how to accurately
splines model is more than that of the other models, one of the model the flow curve of drilling fluids and to use it in predicting
main advantages of cubic splines is its polynomial form. One can pressure loss in the wellbore.
obtain both analytical and numerical solutions with the polyno-
mial form. Hence, the computational procedure is straightforward
and relatively easier than that of yield power law model and other Model Formulation
complex models. From the constructed flow curve, a hydraulics Cubic-Spline Determination. Natural Splines. A cubic spline
model is formulated with the Rabinowitsch-Mooney approach. is a sufficiently smooth cubic function that is piecewise defined,
The result is then validated with experimental data to verify the and possesses a high degree of smoothness at the points in which

the polynomial pieces connect. One can extrapolate the rheologi- The coefficients are determined as the following:
cal data from a rotational viscometer with cubic splines as 8 00 00
>
> si ðc_ i Þ si ðc_ i1 Þ
8 > c
> 3i ¼
s1 ðc_ Þ ¼ c01 þ c11 c_ þ c21 c_ 2 þ c31 c_ 3 ; c_ o c_ c_ 1 >
> 6ðc_ i c_ i1 Þ
>
> >
>
>
<s ðc_ Þ ¼ c þ c c_ þ c c_ 2 þ c c_ 3 ; >
> 00 00
2 12 12 22 32 c_ 1 c_ c_ 2 >
< c ¼ i i ðc_ i1 Þ c_ i1 si ðc_ i Þ
_
c s
s¼ : 2i
2ðc_ i c_ i1 Þ
>…
>
>
: >
>
>
> ½si ðc_ i Þ si1 ðc_ i Þ c2i c_ 2i c_ 2i1 c3i c_ 3i c_ 3i1
sn ðc_ Þ ¼ c0n þ c1n c_ þ c2n c_ 2 þ c3n c_ 3 ; c_ n1 c_ c_ n >
> c ¼
>
> 1i
ð1Þ >
> 2ðc_ i c_ i1 Þ
>
:
c0i ¼ si1 c1i c_ i1 c2i c_ 2i1 c3i c_ 3i1 ;
Eq. 1 contains 4n unknowns that one can determine from the ð7Þ
viscometer measurements. One can obtain the coefficients of each
equation with the following constraints: where the second-order derivatives are defined through the first-
• The splines have to pass through all points: order derivatives as
8 2
>
> @ si ðc_ i1 Þ 0 6 si si1 2½s0 ðc_ Þ þ 2s0i ðc_ i1 Þ
si ðc_ i Þ ¼ siþ1 ðc_ i Þ () c0i þ c1i c_ þ c2i c_ 2i þ c3i c_ 3i >
< @ c_ 2 ¼ s ð
i ic_ Þ ¼ 2
i i
ðc_ i c_ i1 Þ ðc_ i c_ i1 Þ
¼ c0iþ1 þ c1iþ1 c_ þ c2iþ1 c_ 2i þ c3iþ1 c_ 3i : . . . . . . . ð2Þ :
>
> @ 2
s ðc_ Þ 6 s s 2½s0 _
i i Þ þ 2si ðc
ð c 0 _
i1 Þ
>
:
i i1 0
¼ si ðc_ i Þ ¼ i1 i
þ
• The slope or the first derivative of the two consecutive cubic @ c_ 2 ðc_ i c_ i1 Þ2 ðc_ i c_ i1 Þ
functions at the common point must be equal, ð8Þ
@si ðc_ i Þ @siþ1 ðc_ i Þ From a set of experimental data ðc_ i ; si Þ, the first-order derivative
¼ () c1i þ 2c2i c_ i þ 3c3i c_ 2i can be evaluated with Eq. 6. The second-order derivative is evaluated
@ c_ @ c_ with Eq. 8. Then, spline coefficients can be determined with Eq. 7.
¼ c1iþ1 þ 2c2iþ1 c_ i þ 3c3iþ1 c_ 2i : . . . . . . . . . . . ð3Þ
Pressure-Drop Calculation. For non-Newtonian drilling fluids,
• The second-order derivative of two functions at the common turbulence is achieved only when the flow rate is significantly
point must be equal, high. The scope of this research is the laminar flow regime in pipe
and concentric annulus. The flow in eccentric annulus and turbu-
@ 2 si ðc_ i Þ @ 2 siþ1 ðc_ i Þ lent flow are not within the scope of this work.
¼ () c2i þ 3c3i c_ i
@ c_ 2 @ c_ 2 Pipe Flow. With force balance, we obtain the following corre-
lation between wall shear stress and pressure gradient:
¼ c2iþ1 þ 3c3iþ1 c_ i : . . . . . . . . . . . . . . . . . . ð4Þ
@P R @P 2
sw ¼ or ¼ sw : . . . . . . . . . . . . . . . ð9Þ
• For natural cubic splines, the second-order derivative at the @L 2 @L R
two points is zero, One can obtain flow rate from
8 2 2ðp ð
> @ s1 ðc_ 0 Þ R
>
< ¼ 2c21 þ 6c31 c_ o ¼ 0
@ c_ 2 Q¼ vðrÞdrdh: . . . . . . . . . . . . . . . . . . . . . . . . . ð10Þ
: . . . . . . . . . . . . . . ð5Þ
>
> @ 2 sn ðc_ n Þ 0 0
: ¼ 2c 2n þ 6c 3n _
c n ¼ 0
@ c_ 2 Assuming no wall slippage, and changing variables, we can
From Eqs. 2 through 5, we can obtain 4n equations to solve for arrive at the Rabinowitsch-Mooney equation correlating wall shear
the 4n unknown coefficients. Hence, one can construct the cubic stress and flow rate for pipe flow of any fluids, as shown in Eq. 11:
splines if rheological data are available. sðw
pR3
Constrained Splines. In the rheological data collected with Q¼ s2 c_ ds: . . . . . . . . . . . . . . . . . . . . . . . . . . ð11Þ
six-speed field viscometers, the shear-rate range between two con- s2w
0
secutive speeds is not uniform. There are more data at the lower
shear rates than that at the higher shear rates. Natural cubic Substituting the equation for shear stress (Eq. 1) into Eq. 11
splines often result in overshooting in the low shear-rate range. and using shear rate as the primary variable, one can calculate
We often observe the overshooting at a shear-rate range from flow rate from shear rate as
10.22 to 170.3 sec1 corresponding to rotational speed of 6 to 100
pR3
rev/min on field viscometers. To avoid overshooting, one should Q¼ 2
collect more data with measurement devices with more rotational c0i þ c1i c_ w þ c2i c_ 2w þ c3i c_ 3w
speeds. One simple method in the absence of these devices is to 2 c_ 3
ð1
use constrained cubic splines. To construct constrained cubic
3 2

6 c01 þ c11 c_ þ c21 c_ þ c31 c_ c_ d c01 þ c11 c_ þ c21 c_ þ c31 c_
2 2 37
splines, the approach introduced by Kruger (2002) is used in this 6 7
60 7
paper. We eliminate the requirement of equal second-order deriv- 6 7
6 cð_ 2 7
atives and replace it with the specified first-order derivatives 6 7
6 þ c01 þ c11 c_ þ c21 c_ 2 þ c31 c_ 3 c_ d c01 þ c11 c_ þ c21 c_ 2 þ c31 c_ 3 7
2
defined numerically as 6 7
6 c_ 7
8 6 1 7
> 2 si si1 siþ1 si 6
6 cð_ 3 7
>
> s 0 _
ð c Þ ¼ 7
> i i
> 6 2 7
>
> c_ iþ1 c_ i si si1 þ ðc_ i c_ i1 Þ siþ1 si 6 þ c02 þ c12 c_ þ c22 c_ 2 þ c32 c_ 3 c_ d c02 þ c12 c_ þ c22 c_ 2 þ c32 c_ 3 7
>
< 6 7
3 s s0 s01 ðc_ 1 Þ 6 c2 7
s01 ðc_ 0 Þ ¼ 1 : 6… 7
> 2 c_ 1 c_ 0 2 6 7
>
> 6 c_ðw 7
>
> 6 7
> 0 4 þ c0i þ c1i c_ þ c2i c_ 2 þ c3i c_ 3 2 c_ dc0i þ c1i c_ þ c2i c_ 2 þ c3i c_ 3 5
> s0 ðc_ n Þ ¼ 3 sn sn1 sn ðc_ n Þ
>
: n
2 c_ n c_ n1 2 c_ i
ð6Þ ð12Þ

pR3 Similar to pipe flow, using force balance, we obtain the fol-
Q¼ 2 lowing correlation between the average wall shear stress and the
c0i þ c1i c_ w þ c2i c_ 2w þ c3i c_ 3w
2 3 pressure gradient:
_1
cð
6 c þ c c_ þ c c_ 2 þ c c_ 3 2 c c_ þ 2c c_ 2 þ 3c c_ 3 d_c 7 @P H @P 2
6 01 11 21 31 11 21 31 7 sw ¼ or ¼ s w : . . . . . . . . . . . . . . . . ð17Þ
6 7 @L 2 @L H
60 7
6 c_ 7
6 ð2 7
6 2 7 One can obtain flow rate from
6 þ c01 þ c11 c_ þ c21 c_ þ c31 c_
2 3
c _
c þ 2c _
c 2
þ 3c _
c 3
c
d_ 7
6 11 21 31 7
6 c_ 7 ð ðH
W W ð
ð H=2
6 1 7
6
6 cð_ 3 7:
7 Q¼ vðhÞdhdw ¼2 vðhÞdhdw: . . . . . . . . . . . . ð18Þ
6 2 7
6 þ c02 þ c12 c_ þ c22 c_ 2 þ c32 c_ 3 c12 c_ þ 2c22 c_ 2 þ 3c32 c_ 3 d_c 7 0 0 0 0
6 7
6 c_ 7
6 2 7 With the assumption of no wall slippage, the Rabinowitsch-
6… 7
6 c_ w 7 Mooney equation for the rectangular slot has the following form:
6 ð 7
6 2 3 2 2 3 7
4 þ c0i þ c1i c_ þ c2i c_ þ c3i c_ c1i c_ þ 2c2i c_ þ 3c3i c_ d_c 5 sðw
c_ i WH 2
Q¼ s_c ds: . . . . . . . . . . . . . . . . . . . . . . . . . . ð19Þ
ð13Þ 2s2w
0
Because all coefficients in Eq. 13 are determined from the

Substituting Eq. 1 for shear stress into Eq. 19 and using shear
rheological data, the integrand in the right-hand side is a polyno-
rate as the primary variable, one can obtain flow rate in Eqs. 20
mial. For an assumed wall shear rate, the integral can be evaluated
and 21:
analytically. Then, the calculated flow rate from Eq. 13 is com-
pared with the given flow rate. By varying the value of the wall 2
Do Do
shear rate, we can iteratively obtain a value of shear rate that p þ bDi bDi
b b
results in the calculated flow rate equal to the given flow rate. In Q¼
2 3 2
other words, for a given flow rate, Eq. 13 can be solved implicitly 16 c0i þ c1i c_ w þ c2i c_ w þ c3i c_ w
for the wall shear rate. Substituting the obtained wall shear rate 2 c_ 1 3
ð
into the equation for shear stress (Eq. 1), one can obtain wall shear
stress. The frictional pressure gradient is calculated from the cor- 6 c01 þ c11 c_ þ c21 c_ þ c31 c_ c_ d c01 þ c11 c_ þ c21 c_ 2 þ c31 c_ 3 7
2 3
6 7
relation between the frictional pressure gradient and the wall 60 7
6 7
shear stress (Eq. 9). 6 cð2
_ 7
6 7
Annular Flow. For a concentric annulus, with the narrow-slot 6 þ c01 þ c11 c_ þ c21 c_ 2 þ c31 c_ 3 c_ d c01 þ c11 c_ þ c21 c_ 2 þ c31 c_ 3 7
6 7
approximation approach, a concentric annulus can be represented 6 c_ 7
6 1 7
by a rectangular slot with the height, H, and the width, W, defined as 6
6 ð3
c_
7
7
6 7
1 p 6 þ c02 þ c12 c_ þ c22 c_ þ c32 c_ c_ d c02 þ c12 c_ þ c22 c_ þ c32 c_ 7
2 3 2 3
6 7
H ¼ ð Do Di Þ and W ¼ ðDo þ Di Þ: . . . . . . . ð14Þ 6 c_ 2 7
2 2 6… 7
6 7
6 c_ðw 7
The narrow-slot approach only accounts for the geometrical 6 7
4 þ c0i þ c1i c_ þ c2i c_ 2 þ c3i c_ 3 c_ d c0i þ c1i c_ þ c2i c_ 2 þ c3i c_ 3 5
correspondence between annular and rectangular slot, but the var-
iation of flow properties on the wall such as wall shear rate and c_ i
wall shear stress is not considered. For a real annular flow, the ð20Þ
wall shear stress and wall shear rate on two walls, outer and inner 2
walls, are not the same. However, in the narrow-slot approxima- Do Do
p þ bDi bDi
tion, the wall shear rate, and wall shear stress on two rectangular b b
walls are the same. For a very narrow slot, the wall shear stress Q¼
2 3 2
and wall shear rate on both inner and outer walls are relatively 16 c0i þ c1i c_ w þ c2i c_ w þ c3i c_ w
equal. In this case, the narrow-slot approximation is very effec- 2 c_ 3
ð1
tive. For an annulus with high diameter ratio (Do/Di), this
6 c01 þ c11 c_ þ c21 c_ þ c31 c_ c11 c_ þ 2c21 c_ þ 3c31 c_ d_c 7
2 3 2 3
approach leads to significant errors. To consider both geometrical 6 7
60 7
constraints and wall properties, we introduce a shape-correction 6 7
6 c_ 2 7
factor (b) and define the corresponding rectangular slot height and 6 ð
7
7
6 2 3 2 3
width as the following: 6 þ c01 þ c11 c_ þ c21 c_ þ c31 c_ c11 c_ þ 2c21 c_ þ 3c31 c_ d_c 7
6 7
6 c_ 1 7
1 Do p Do 6 7
H¼ bDi and W ¼ þ bDi ; . . . . ð15Þ 6
6 ð3 c
_ 7:
7
2 b 2 b 6 7
6 þ c02 þ c12 c_ þ c22 c_ þ c32 c_ c12 c_ þ 2c22 c_ þ 3c32 c_ d_c 7
2 3 2 3
6 7
where b is larger for higher diameter ratio (Do/Di) and approxi- 6 c_ 2 7
mately equal to unity for a very narrow slot. 6… 7
6 7
By measuring the pressure gradient in annular flow, one can cal- 6 c_ 7
6 ð w 7
culate average wall shear stress ðc w Þ. Also, from the measurement 6 7
4 þ c0i þ c1i c_ þ c2i c_ 2 þ c3i c_ 3 c1i c_ þ 2c2i c_ 2 þ 3c3i c_ 3 d_c 5
of wall shear stress at the outer and inner walls ðswo and swi Þ, one
can determine the shape-correction factor from the force-balance c_ i
equation (Eq. 15). ð21Þ

Do Again, all coefficients in the above equation are known, the
p þ bDi DLs w ¼ ðp Do swo þ p Di swi ÞDL or
b average wall shear rate (_c w ) can be evaluated numerically.
In the laminar flow regime, the generalized Reynolds number
Do
þ bDi s w ¼ Do swo þ Di swi : ð16Þ is calculated through the wall shear stress and averaged velocity
b
from Eq. 22:

Rheological Data and generalized Reynolds number can be computed from the wall
shear rate with Eq. 22. If the assumed wall shear rate is too large,
the solution may not be obtained. Hence, it is recommended to
Obtain Cubic Spline
Coefficients
use a small initial wall shear rate, typically zero.
Sample Calculation. In this section, we present a detailed exam-

.
Start with γw = 0 ple to demonstrate the calculation procedure of the model intro-
duced. This example includes both pipe and annular flows. The pipe
diameter used is D ¼ 0.1143 m, and the annulus with inside diame-
Calculate Flow Rate Qcal ter (ID) is Di ¼ 0.127 m with an outside diameter (OD) of
(Eq. 13 and Eq. 21)
Do ¼ 0.254 m. The drilling-fluid density is 1200 kg/m3. The rheo-
logical data from laboratory measurement are presented in Table 2.
From the measured rheological data provided in Table 2, one
No . .
γw = γw + Δγw
. can construct the following natural cubic splines to model the
|Qcal–Qgiven| ≤ Tolerance
flow curve of the drilling fluid:
8
Yes >
> s1 ðc_ Þ ¼ c01 þ c11 c_ þ c21 c_ 2 þ c31 c_ 3 ; 5:11 c_ 10:22
>
>
>
> 2 3
< s2 ðc_ Þ ¼ c02 þ c12 c_ þ c12 c_ þ c32 c_ ; 10:22 c_ 170:3
Calculate τw (Eq. 1), dp/dL (Eq.9 and Eq. 17),
and Reynolds number (Eq. 22) s ¼ s3 ðc_ Þ ¼ c03 þ c13 c_ þ c13 c_ þ c33 c_ 3 ;
2
170:3 c_ 340:6 :
>
>
>
> s4 ðc_ Þ ¼ c04 þ c14 c_ þ c14 c_ 2 þ c34 c_ 3 ; 340:6 c_ 510:9
>
>
:
s5 ðc_ Þ ¼ c05 þ c15 c_ þ c15 c_ 2 þ c35 c_ 3 ; 510:9 c_ 1021:8
END
ð23Þ
Fig. 2—Numerical flow chart.
We need to evaluate 20 coefficients (cjk) of Eq. 23 with pre-
sented constrains. For the six rheological data points c_ i , si,
16 8qv 2 (i ¼ 1::6), one can use the following linear equation to solve for
Reg ¼ ¼ ; . . . . . . . . . . . . . . . . . . . . . . . . . ð22Þ 20 coefficients:
f sw
where f is friction factor; v is average flow velocity; and s w is av- c ¼ A1 s; . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ð24Þ
erage wall shear stress.
where
Computational Procedure. By assuming an initial value of the c ¼ ½c01 c11 c21 c31 c02 c12 c22 c32 c03 c13
wall shear rate, typically, a small value, one can evaluate the inte- c23 c33 c04 c14 c24 c34 c05 c15 c25 c35 T
gral on the right-hand side analytically. Then, the right-hand side ð25Þ
is compared with the given flow rate. If two values are equal
within a defined tolerance, then the calculation is ceased. Other- s ¼ ½s1 s2 s2 s3 s3 s4 s4 s5 s5 s6
wise, the wall shear rate is increased, and the right-hand side is 0 0 0 0 0 0 0 0 0 0T
evaluated again. This procedure is repeated until convergence is
achieved (Fig. 2). Then, the wall shear stress, pressure gradient ð26Þ
2 3
1 c_ 1 c_ 21 c_ 31 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
6 7
6 1 c_ 2 c_ 22 c_ 22 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 7
6 7
60 0 0 0 1 _c 2 _c 22 _c 32 0 0 0 0 0 0 0 0 0 0 0 0 7
6 7
6 7
60 0 0 0 1 c_ 3 c_ 23 c_ 33 0 0 0 0 0 0 0 0 0 0 0 0 7
6 7
60 0 0 0 0 0 0 0 1 c_ 3 c_ 23 c_ 33 0 0 0 0 0 0 0 0 7
6 7
6 7
60 0 0 0 0 0 0 0 1 c_ 4 c_ 24 c_ 24 0 0 0 0 0 0 0 0 7
6 7
60 0 0 0 0 0 0 0 0 0 0 0 1 c_ 4 c_ 24 c_ 34 0 0 0 0 7
6 7
6 7
60 0
6 0 0 0 0 0 0 0 0 0 0 1 c_ 5 c_ 25 c_ 35 0 0 0 0 7 7
60 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 c_ 5 c_ 25 c_ 35 7
6 7
6 7
60 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 c_ 6 c_ 26 c_ 36 7
A¼6
60 1
7:
6 2_c 2 3_c 22 0 1 2_c 2 3_c 22 0 0 0 0 0 0 0 0 0 0 0 0 7 7
6 7
60 0 0 0 0 1 2_c 3 3_c 23 0 1 2_c 3 3_c 23 0 0 0 0 0 0 0 0 7
6 7
60 0 0 0 0 0 0 0 0 1 2_c 4 3_c 24 0 1 2_c 4 3_c 24 0 0 0 0 7
6 7
6 7
60 0 0 0 0 0 0 0 0 0 0 0 0 1 2_c 5 3_c 25 0 1 2_c 5 3_c 24 7
6 7
6 7
60 0 1 3_c 2 0 0 1 3_c 2 0 0 0 0 0 0 0 0 0 0 0 0 7
6 7
60 0 0 0 0 0 1 3_c 3 0 0 1 3_c 3 0 0 0 0 0 0 0 0 7
6 7
6 7
60 0 0 0 0 0 0 0 0 0 1 3_c 4 0 0 1 3_c 4 0 0 0 0 7
6 7
60 0 0 0 0 0 0 0 0 0 0 0 0 0 1 3_c 5 0 0 1 3_c 5 7
6 7
6 7
40 0 1 3_c 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 5
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 3c6
ð27Þ

50
40
Shear Stress (Pa)

30
20
Experimental Data
Natural Cubic Splines
10 Bingham Plastic
Power Law
Yield Power Law
Constrained Cubic Splines
0
0 200 400 600 800 1,000
Shear Rate (1/sec)
Table 2—Fann 35 viscometer data for sample calculation. Fig. 3—Flow curve obtained from Fann 35 viscometer data.
The size of the matrix A is relatively small and can be inversed ent models, the relative error, the root-mean-square (RMS) error
with a spreadsheet. After solving Eq. 24, we obtain 20 coefficients (RMSE) and coefficient of determination (R2) are used. The RMSE
(cjk), and the following natural cubic splines are obtained to model and R2 are defined as
the flow curve. 8 vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
8 > uX
> s0 ðc_ Þ ¼ 3:066 þ 0:196046_c þ 0:001172_c 2 >
>
> u N c 2
>
> >
> u yi yi
>
> > t
>
> 7:6 105 c_ 3 ; 0 c_ 5:11 >
>
> RMSE ¼ i¼1
>
> >
>
>
> s ðc_ Þ ¼ 3:066 þ 0:196046_ c þ 0:001172_ c2 >
< N
>
> 1 2 32
>
> X N ;
>
> 7:6 105 c_ ;
3
5:11 c_ 10:22 > c ð29Þ
>
> >
> 6 yi yi c yi yi 7
>
> s2 ðc_ Þ ¼ 2:981557 þ 0:220839_c 0:00125_c 2 >
> 6 7
>
> > R2 ¼ 6sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
>
i¼1
ffisffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi7
>
> >
> 6 X 7
>
>
> þ 2:66 106 c_ 3 ; 10:22 c_ 170:3 >
> 4 N 2 X N 2 5
>
< s ðc_ Þ ¼ 17:41748 0:03346_c þ 0:000239_c 2 >
> y c
y c y y
3
: i i i i
s¼ ; i¼1 i¼1
>
> 2:6 107 c_ 3 ; 170:3 c_ 340:6
>
>
>
> s4 ðc_ Þ ¼ 5:19588 þ 0:074186_c þ 7:7 105 c_ 2 where yi and yci are the actual value and the calculated value; y i
>
>
>
> and yi c are the mean of actual values and calculated values; and N
>
> þ 4:94 108 c_ 3 ; 340:6 c_ 510:9
>
> is the number of data points.
>
>
>
> s 5 ðc_ Þ ¼ 11:6832 þ 0:036091_ c 2:2 106 c_ 2 Because the cubic splines are constructed from the rheological
>
>
>
> þ 7:6 1010 c_ ;
3
510:9 c_ 1021:8 data, both constrained and natural splines fit perfectly to the ex-
>
>
>
> perimental data. The Bingham plastic model results in an average
> s6 ðc_ Þ ¼ 11:6832 þ 0:036091_c 2:2 106 c_ 2
>
>
: error of nearly 53.9% and RMSE of 4.17. Power law and yield
þ 7:6 1010 c_ 3 ; c_ 1021:8 power law models provide a better approximation of the flow
ð28Þ curve than Bingham plastic model. The average error for power
law is approximately 24.1% and that of yield power law is 6%.
s0 ðc_ Þ and s6 ðc_ Þ are extrapolated from s1 ðc_ Þ and s5 ðc_ Þ, The highest shear rate from field viscometer, 1021.8 sec1, is of-
respectively. ten in the turbulent flow regime. Because yield power law model
From Eq. 28, the flow curve can be drawn. A comparison of often fits very well to the data in range of low shear rate, yield
the natural cubic-spline model and other models is presented in power law model will give better pressure-drop prediction than
Fig. 3 and listed in Table 3. To compare the prediction by differ- power law and Bingham plastic in laminar flow regime.
Table 3—Comparison between different rheological models.

Frictional Pressure Gradient (Pa/m)

700 750
580 620
460 490
340 360
220 Generalized Reynolds Number 230 Generalized Reynolds Number
0 460 920 1380 1840 2300 0 460 920 1380 1840 2300
100 100
0 0.0044 0.0088 0.0132 0.0176 0.022 0.001 0.0186 0.0362 0.0538 0.0714 0.089
Flow Rate (m3/sec) Flow Rate (m3/sec)
Fig. 4—Frictional pressure gradient for pipe flow. Fig. 5—Frictional pressure gradient for annular flow.
21 23
Average Wall Shear Stress (Pa)

Wall Shear Stress (Pa)
17 19
13 15
9 11
5 Average Velocity (m/sec) 7 Average Velocity (m/sec)
0 .43 .86 1.29 1.72 2.15 0 .47 .94 1.41 1.88 2.35
1 3
0 46 92 138 184 230 0 65 130 195 260 325
Wall Shear Rate (1/sec) Average Wall Shear Rate (1/sec)
Fig. 6—Wall shear stress as a function of wall shear rate for pipe Fig. 7—Average wall shear stress as a function of average wall
flow. shear rate for annular flow.
It is evident that the coefficients corresponding to c_ 2 and c_ 3 stress is 17.2 sec1, using Eq. 15 and noting that the guessed shear
are relatively small. These coefficients become very small at very rate in the interval 10:22 c_ 170:3, we obtain the following
high shear rate, indicating that the nonlinear part is small and vis- equation to evaluate the calculated flow rate Qcal:
cosity is nearly constant at high shear rates. This agrees with the
observation that Newtonian behavior is often observed at very 0:1143 3
p
low and very high shear rates. In other words, the cubic spline 2
model can predict very well the Newtonian response at very low 2 5:11
ð
3
and high shear rates. 5 3 2

6 3:066 þ 0:196046_ c þ 0:001172_ c 2
7:6 10 _
c 7
6 7
After the flow curve is formulated, the aforementioned compu- 6
6 0
7
7
tational procedure is used to calculate the wall shear rate for a 6 2 5 3
7
6 0:196046_ c þ 2 0:001172_c 3 7:6 10 _ d_c
c 7
given value of the flow rate. For each value of flow rate, we can 6 7
6 10:22 7
6 ð 7
numerically obtain a value of wall shear rate from Eq. 13 for pipe 6 2 7
6þ 3:066 þ 0:196046_c þ 0:001172_c 7:6 105 c_
2 3
7
flow and Eq. 21 for annular flow. For example, for a given pipe 66 7
7
6 5:11 7
flow rate Qgiven ¼ 2 sec1, if the guessed value of wall shear 6
6 0:196046_c þ 2 0:001172_c 2 3 7:6 105 c_ 3 d_c 7
7
6 7
6 17:2 7
6 ð 7
6þ 2:981557 þ 0:220839_c 0:00125_ c 2
þ 2:66 10 6 3
c_
2 7
6 7
25 6 7
4 10:22 5

0:220839_c 2 0:00125_c 2 þ 3 2:66 106 c_ 3 d_c
Qcal ¼ 2
20 2:981557 þ 0:220839 17:2 0:00125 17:22 þ 2:66 106 17:23
Shear Stress (Pa)
ð30Þ
15
The integrant in the Eq. 30 is a polynomial and can be evaluated
manually or with a spreadsheet. After evaluating the right hand side
10
Experimental Data integral, we obtain Qcal ¼1.8 sec1. Because this calculated value
Power Law
Bingham Plastic of flow rate is smaller than the given value, a higher value of wall
5 Robertson Stiff shear stress is used for the next guess to calculate new value of flow
Sisko
Yield Power Law rate, as suggested by the numerical flow chart (Fig. 2). That new cal-
0
Constrained Cubic Splines culated flow rate is then compared with the given flow rate. This
0 200 400 600 800 1,000 process is repeated until the calculated flow rate is equal to the given
Shear Rate (1/sec) flow rate. The value of wall shear rate equal to 18.7 sec1 is
obtained through this process for Qgiven ¼ 2 sec1. For this wall
Fig. 8—Flow curve from Fann 35 viscometer data for Fluid Type shear stress, we used the third equation in Eq. 28 to get the corre-
A [Experimental data are from Okafor and Evers (1992)]. sponding wall shear stress. The shear stress corresponding to the

Table 4—Fann 35 viscometer data and comparison between different rheological models for Fluid Type A.
calculated wall shear rate is 6.69 Pa. The pressure gradient of 234.1 For Fluid Type A. From the rheological data (Table 4), the con-
Pa/m or 0.01035 psi/ft is calculated from Eq. 9 with the calculated strained cubic splines are constructed in Eq. 31:
wall shear stress. The Reynolds number obtained from Eq. 22
is 54.5. 8
>
> s0 ðc_ Þ ¼ 0:479 þ 0:045374_c þ 0:014279_c 2
Similarly, the wall shear stress and pressure gradient are eval- >
>
>
> 9:3 104 c_ 3 ; 0 c_ 5:11
uated for each value of flow rate. The frictional pressure gradient >
>
>
>
for different flow rates with corresponding Reynolds number is >
> s1 ðc_ Þ ¼ 0:479 þ 0:045374_c þ 0:014279_c 2
>
>
shown in Fig. 4 for pipe flow and in Fig. 5 for annular flow. >
> 9:3 104 c_ 3 ; 5:11 c_ 10:22
One more important observation is the overshooting of the nat- >
>
>
>
ural cubic-spline curve on the shear-rate ranges from 10.22 to >
> s ðc_ Þ ¼ 0:955176 þ 0:049154_c 1:9 104 c_ 2
> 2
>
170.3 sec1. This results in overshooting on the wall shear stress >
> þ 5:06 107 c_ 3 ; 10:22 c_ 170:3
>
>
vs. the wall shear-rate curve (Figs. 6 and 7), because the wall >
< s ðc_ Þ ¼ 1:552198 0:026436_c þ 1:2 105 c_ 2
3
shear stress vs. the wall shear-rate curve is the same as the flow s¼ :
curve on Fig. 3. In this case, using constrained cubic splines will >
> 3:6 108 c_ 3 ; 170:3 c_ 340:7
>
>
resolve this problem. >
> s4 ðc_ Þ ¼ 1:6389 þ 0:055469_c 7:5 105 c_ 2
>
>
For annular flow, with the diameter ratio, Do/Di ¼ 2, the >
>
>
> þ 5:06 108 c_ 3 ; 340:7 c_ 510:9
shape-correction factor b ¼ 1.085 is used. >
>
>
>
> s5 ðc_ Þ ¼ 3:420947 þ 0:02272_c 5:2 106 c_ 2
>
>
>
>
> þ 1:71 109 c_ 3 ; 510:9 c_ 1021:8
Model Validation >
>
>
> s ðc_ Þ ¼ 3:420947 þ 0:02272_c 5:2 106 c_ 2
The experimental data used in this section are from Okafor and >
> 6
>
:
Evers (1992). The pipe diameter is D ¼ 2 in., and the annulus has þ 1:71 109 c_ 3 ; c_ 1021:8
inside diameter (ID) of Di ¼ 1.5 in. with an outside diameter (OD)
of Do ¼ 3 in. Two drilling-fluid experiments (for Fluid Type A ð31Þ
and Fluid Type B) were used in their experiments. The density of
Fluid Type A is 1066.2 kg/m3, and that of Fluid Type B is A comparison of cubic-spline model and other common rheo-
1036.3 kg/m3. The rheological data are presented in Fig. 8 and logical models including Bingham plastic, power law, Sisko, Rob-
Table 4 for Fluid Type A and later in Fig. 11 and later in Table 8 ertson-Stiff, and yield power law is presented in Table 4. It is
for Fluid Type B. For annular flow, the shape-correction factor observed that all these models fit well to the experimental data,
b ¼ 1.085 is used for the diameter ratio Do/Di ¼ 2. whereas cubic spline model gives a perfect fit to the experimental
data. The Bingham Plastic shows the lowest match to the experi-
mental data compared with the other models. With the average
710
Cubic Splines error less than 4% and RMSE less than 1.7 Pa, Sisko, Robertson-
Herschel-Bulkley Stiff, and yield power law models also are good for rheological
Experimental Data
588 characterization of this fluid. Hence, Robertson-Stiff model is
originally recommended by Okafor and Evers (2001) to predict
466 the pressure drop. Bailey and Peden (2000) show that Sisko model
gives a very good prediction of pressure drop. Yield power law
model is also suggested by Founargiotakis et al. (2008) and Keles-
344
sidis et al. (Mitchell and Miska 2011).
The pressure gradient predicted by cubic-spline model is com-
222 Generalized Reynolds Number pared with the prediction by the yield power law model, as shown
in Fig. 9 and Table 5 for pipe flow and Fig. 10 and Table 6 for
0 500 1000 1500 2000 2500
100 annular flow. Because the rheological data fit very well to the
0 0.3 0.6 0.9 1.2 1.5 Sisko model, as shown in Fig. 8, this model also provides a rela-
Average Velocity (m/sec) tively good prediction compared with the cubic spline and yield
power law models. The pressure gradient for the pipe flow pre-
Fig. 9—Frictional pressure gradient for pipe flow for Fluid Type dicted by Bailey and Peden (2000) with Sisko model for pipe flow
A [Experimental data are from Okafor and Evers (1992)]. is also presented here for a comparison (Table 5).

Table 5—Comparison of cubic-spline model prediction with experimental data and other model predictions for pipe flow (Fluid Type A).
flow rate of 1.448 m/s for pipe flow may indicate that this data
3600
Cubic Splines
Herschel-Bulkley
point is out of the laminar flow regime. However, the critical
Experimental Data Reynolds number for non-Newtonian fluids is often higher than
2900
that of Newtonian fluids. Moreover, the critical Reynolds number
depends strongly on the pipe wall (Snyder and Castro 2002). The
2200 determination of the critical Reynolds number for non-Newto-
nian fluids is still an area for further research (Campos et al.
1500 1994; Trinh 2011). The original publication by Okafor and Evers
(1992) considers this last point to be in the laminar flow regime.
Bailey and Peden (2000) and Kelessidis et al. (cited in Mitchell
800 Generalized Reynolds Number and Miska 2011) also predicted that it is still in the laminar
0 420 840 1260 1680 2100 regime. Therefore, we assume laminar flow regime for this
100 data point.
0 0.5 1 1.5 2 2.5
Average Velocity (m/sec)
The narrow-slot approximation is used for annular flow. This
approach provides a good approximation if the gap is relatively
Fig. 10—Frictional pressure gradient for annular flow for Fluid
small and the error increases with the increasing of inside to OD
Type A [Experimental data are from Okafor and Evers (1992)]. ratio. The deviation of the yield power law model prediction for
annular flow (Fig. 10) could be the result of the narrow-slot
assumption because the difference between ID and OD is quite
high in this case (Do/Di ¼ 2).
One can observe that the cubic-spline model provides a very
good fit to experimental data. Sisko and yield power law models
give a very good prediction of the pressure drop, especially for For Fluid Type B. The constrained cubic-spline rheological
pipe flow (Fig. 9). The deviation of the last data point at the model for this experimental data is presented in Eq. 32:
Table 6—Comparison of the experimental data and model results for annular flow (Fluid Type A).

30 law, and Robertson Stiff models match well to the rheological

data (Table 7). Both yield power law and cubic-spline model give
25
a very accurate prediction of the frictional pressure gradient for
pipe flow with the average relative error, approximately 4%, as
Shear Stress (Pa)
shown in Table 8. Yield Power Law model estimates quite well

20 the pressure gradient at low flow rate, but it overestimates the
pressure gradient at high flow rate. Cubic spline gives a reasona-
15 ble estimation of the pressure gradient at higher flow rate, but it
Experimental Data
Power Law overestimates the pressure gradient at low flow rate (Fig. 12).
Bingham Plastic For annular flow, the cubic-spline model shows a better pre-
10 Robertson Stiff
Sisko diction of the friction pressure gradient than the yield power law
Yield Power Law model, as shown in Table 9 and Fig. 13. The relative error by
Constrained Cubic Splines
5 cubic-spline model is, in this case, 1.32% and that of Power Law
0 200 400 600 800 1,000 model is 4.75%, suggesting that both models can be used for this
Shear Rate (1/sec) fluids. The cubic-spline model gives a slightly better prediction.
Fig. 11—Flow curve from Fann 35 viscometer data for Fluid

Type B [Experimental data are from Okafor and Evers (1992)]. Summary and Remarks
In this research, we introduce a cubic-spline model and numerical
implementation for obtaining flow curves of drilling fluids. The
8
> s0 ðc_ Þ ¼ 8:61845 þ 0:04829_c þ 0:05483_c 2 main advantage of the cubic-spline model is that it can give a
>
> good agreement to any set of experimental data. The shear-rate-
>
>
>
> 3:58 103 c_ 3 ; 0 c_ 5:11 dependent behavior commonly observed in drilling fluids can be
>
>
>
> s1 ðc_ Þ ¼ 8:61845 þ 0:04829_c þ 0:05483_c 2 modeled with cubic splines. Hence, all measured data from field
>
>
>
> viscometers can be used to capture the flow properties of fluids.
>
> 3:58 103 c_ 3 ; 5:11 c_ 10:22
>
> The model presented in this paper provides an additional tool to
>
> s ðc_ Þ ¼ 10:4962 þ 0:05299_ c 2:4 104 c_ 2
>
> 2 rheological and hydraulic modeling.
>
>
>
>
> þ 5:5 107 c_ 3 ; 10:22 c_ 170:3 Both constrained and natural cubic splines are presented in
>
< s ðc_ Þ ¼ 9:61368 þ 0:06358_c 1:4 104 c_ 2 this research. The constrained splines are sometimes recom-
3
s¼ : mended to replace the natural splines to eliminate the overshoot-
>
> þ 1:7 107 c_ 3 ; 170:3 c_ 340:6
>
> ing often observed at low shear rates.
>
>
>
> s4 ðc_ Þ ¼ 22:8037 0:06385_c þ 2 104 c_ 2 A drilling-fluid hydraulics model that is based on the Rabino-
>
> witsch-Mooney approach for both pipe and annular flow was
>
> 1:7 107 c_ 3 ; 340:6 c_ 510:9
>
> derived. The computational procedure and field examples were
>
>
>
> s5 ðc_ Þ ¼ 11:7448 þ 0:024_c 1:5 105 c_ 2 presented. The results obtained from the new model introduced
>
>
>
> þ 4:8 109 c_ 3 ; 510:9 c_ 1021:8 here agree well with the measured data, indicating the accuracy
>
>
>
> and effectiveness of the model.
> s
> 6
> ðc_ Þ ¼ 11:7448 þ 0:024_ c 1:5 105 c_ 2
: Even though the number of equations and parameters involved
þ 4:8 109 c_ 3 ; c_ 1021:8 in the cubic-spline model is more than that of the other models,
ð32Þ the computational procedure is stable and straightforward. Com-
parison with other models indicates that cubic-spline approxima-
The flow curves predicted with other common rheological tion shows a very close fit to all rheological data points obtained
models are also presented here for comparison. Because Fluid from field viscometer (Fann 35). Hence, it results in a much better
Type B has higher yield stress than that of Fluid Type A, the prediction of the pressure drop.
power law model shows the lowest match to the experimental We demonstrate the determination of the cubic-spline coeffi-
data, as shown in Fig. 11. Both power law and Bingham plastic cient with a spreadsheet. The model can be implemented easily
models overpredict the shear stress at low shear rate and under- and should be used in field operations to improve the accuracy of
predict the shear stress at higher shear rate. Sisko, yield power drilling hydraulics modeling.
Table 7—Fann 35 viscometer data and comparison between different rheological models for Fluid Type B..

Table 8—Comparison of the measured data for pipe flow (Fluid Type B).

1700 4800
Cubic Splines Cubic Splines
Herschel-Bulkley Herschel-Bulkley
Experimental Data Experimental Data
1520 4180
1340 3560
1160 2940
980 Generalized Reynolds Number

2320 Generalized Reynolds Number
0 433 866 1299 1732 2165 0 433 866 1299 1732 2165
800 1700
0 0.44 0.88 1.32 1.76 2.2 0 0.6 1.2 1.8 2.4 3
Average Velocity (m/sec) Average Velocity (m/sec)
Fig. 12—Frictional pressure gradient for pipe flow for Fluid Fig. 13—Frictional pressure gradient for annular flow for Fluid
Type B [Experimental data are from Okafor and Evers (1992)]. Type B [Experimental data are from Okafor and Evers (1992)].
Nomenclature f ¼ friction factor

b ¼ shape-correction factor H ¼ rectangular slot height, m
ci ¼ spline coefficient Pi, Qi ¼ rational polynomial model parameters
dp/dL ¼ frictional pressure gradient, Pa/m r ¼ radial position, m
Di ¼ inside annular diameter, m R ¼ pipe radius, m
Do ¼ outside annular diameter, m Reg ¼ generalized Reynolds number
Table 9—Comparison of the measured data for annular flow (Fluid Type B).

Rec ¼ critical Reynolds number Gucuyener, I. H. 1983. A Rheological Model for Drilling Fluids and
W ¼ rectangular slot width, m Cement Slurries. Presented at the Middle East Oil Technology Confer-
c_ ¼ shear rate, 1/sec ence and Exhibition, Bahrain, 14–17 March. SPE-11487-MS. http://
c_ w ¼ wall shear rate, 1/sec dx.doi.org/10.2118/11487-MS.
c_ w ¼ average wall shear rate, 1/sec Herschel, W. H. and Bulkley, R. 1926. Measurement of Consistency as
D_c ¼ incremental shear rate, 1/sec Applied to Rubber Benzene Solutions. Proc., 29th Annual Meeting of
s ¼ shear stress, Pa the American Society of Testing Materials, Atlantic City, New Jersey,
sy ¼ yield stress, Pa USA, 21–25 June. Vol. 26 (82), pp. 621–629.
sw ¼ average wall shear stress, Pa Houwen, O. H. and Geehan, T. 1986. Rheology of Oil-Base Muds. Pre-
swi ; swo ¼ wall shear stress on the inner and outer wall, respec- sented at the SPE Annual Technical Conference and Exhibition, New
tively, Pa Orleans, USA, 5–8 October. SPE-15416-MS. http://dx.doi.org/
10.2118/15416-MS.
References Hughes, T. L., Jones, T. G., and Houwen, O. H. 1993. Chemical Charac-
terization of CMC and Its Relationship to Drilling-Mud Rheology and
Aadnoy, S. B. and Ravnoy, M. J. 1991. Improved Pressure Drop/Flow
Fluid Loss. SPE Drill & Compl 8 (3): 157–164. SPE-20000-PA. http://
Rate Equation for Non-Newtonian Fluids in Laminar Flow. SPE-
dx.doi.org/10.2118/20000-PA.
23474-MS. http://dx.doi.org/10.2118/23474-MS. Unsolicited.
Kruger, C. J. C. 2002. Constrained Cubic-Spline Interpolation for Chemical
Bailey, W. J. and Weir, I. S. 1998. Investigation of Methods for Direct Engineering Applications. http://www.korf.co.uk/spline.pdf (accessed
Rheological Model Parameter Estimation. J. Petrol. Sci. & Eng. 21 June 2013).
(1–2): 1–13. http://dx.doi.org/10.1016/S0920-4105(98)00040-0. Maglione, R., Robotti, G., and Romagnoli, R. 2000. In-situ Rheological
Bailey, W. J. and Peden, J. M. 2000. A Generalized and Consistent Pres- Characterization of Drilling Mud. SPE J. 5 (4): 377–386. SPE-66285-
sure Drop and Flow Regime Transition Model for Drilling Hydraulics. PA. http://dx.doi.org/10.2118/66285-PA.
SPE Drill & Compl 15 (1): 44–56. SPE-62167-PA. http://dx.doi.org/ Melrose, J. C. and Lilienthal, W. B. 1951. Plastic Flow Properties of Dril-
10.2118/62167-PA. ling Fluids—Measurement and Application. J Pet Technol 3 (6):
Beirute, R. M. and Flumerfelt, R. W. 1977. An Evaluation of the Robert- 159–164. SPE-951159-G-PA. http://dx.doi.org/10.2118/951159-G-PA.
son-Stiff Model Describing Rheological Properties of Drilling Fluids Mezger, G. T. 2006. The Rheology Handbook: For Users of Rotational
and Cement Slurries. SPE J. 17 (2): 97–100. SPE-6505-PA. http:// and Oscillatory Rheometers, second revision edition. Hannover, Ger-
dx.doi.org/10.2118/6505-PA. many: Vincentz Network GmBH & Co. KG.
Bingham, E. C. 1916. An Investigation of the Laws of Plastic Flow. Scien-
Mitchell, R. and Miska, S. 2011. Fundamental of Drilling Engineering.
tific Papers of Bureau of Standard 13, Department of Commerce,
SPE Textbook Series, Vol. 12. Richardson, Texas, USA: Society of
Washington, DC. Paper 278: 309–353.
Petroleum Engineers.
Briscoe, B. J., Luckham, P. F., and Ren, S. R. 1996. The Properties of
Nasiri, M. and Ashrafizadeh, S. N. 2010. Novel Equation for the Prediction
Drilling Muds at High Pressures and High Temperatures. Transaction
of Rheological Parameters of Drilling Fluids in an Annulus. Ind. Eng.
Royal Society 348 (1687). http://dx.doi.org/10.1098/rsta.1994.0088.
Chem. Res. 49 (7): 3374–3385. http://dx.doi.org/10.1021/ie9009233.
Bui, B. 2012. Determination of Viscoelastic Properties of Drilling Fluids.
Okafor, M. N. and Evers, J. F. 1992. Experimental Comparison of Rheol-
MSc thesis, University of Tulsa, USA.
ogy Models for Drilling Fluids. Presented at the Western Regional
Campos, T. D., Steffe, F. J., and Ofoli, Y. R. 1994. Statistical Method to
Meeting, Bakersfield, California, USA, 30 March–1 April. SPE 24086-
Evaluate the Critical Reynolds Number for Pseudoplastic Fluids in
MS. http://dx.doi.org/10.2118/24086-MS.
Tubes. J. Food Eng. 23 (1): 21–32. http://dx.doi.org/10.1016/0260-
Ostwald, W. 1925. Ueber die Geschwindigkeitsfunktion der Viskosität
8774(94)90121-X.
disperser Systeme. I. Kolloid-Zeitschrift 36 (2): 99–117. http://
Carreau, J. P. 1972. Rheological Equations From Molecular Network The-
dx.doi.org/10.1007/BF01431449.
ories. Trans., Society of Rheology 16 (1): 99–127. http://dx.doi.org/
10.1122/1.549276. Pilehvari, A., Serth, R., and Lagad, V. 2001. General Hydraulics Calcula-
Casson, N. 1959. A Flow Equation for Pigment-Oil Suspensions of the tion Method Using Rational Polynomial Model. Presented at the SPE
Printing Ink Type. In Rheology of Disperse Systems, ed. C. C. Mill, Annual Technical Conference and Exhibition, New Orleans, USA, 30
84–104. Oxford: Pergamon Press. September–3 October. SPE 71403-MS. http://dx.doi.org/10.2118/
Cross, M. M. 1965. Rheology of Non-Newtonian Fluids: A New Flow 71403-MS.
Equation for Pseudoplastic Systems. J. Colloid Science 20 (5): Prandtl, L. 1928. Ein Gedankenmodell zur kinetischen Theorie der festen
417–437. http://dx.doi.org/10.1016/0095-8522(65)90022-X. Körper. ZAMM—J. Applied Mathematics and Mechanics 8 (2):
Davison, J. M., Clary, S., Saasen, A. et al. 1999. Rheology of Various 85–106. http://dx.doi.org/10.1002/zamm.19280080202.
Drilling Fluid Systems Under Deepwater Drilling Conditions and the Reiner, M. 1930. In Search for a General Law of the Flow of Matter. J.
Importance of Accurate Predictions of Downhole Fluid Hydraulics. Rheology 1: 250–260. http://dx.doi.org/10.1122/1.2116315.
Presented at the 1999 SPE Annual Technical Conference and Exhibi- Robertson, R. E. and Stiff Jr., H. A. 1976. An Improved Mathematical
tion, Houston, USA, 3–6 October. SPE-56632-MS. http://dx.doi.org/ Model for Relating Shear Stress to Shear Rate in Drilling Fluids and
10.2118/56632-MS. Cement Slurries. SPE J. 16 (1): 31–36. SPE-5333-PA. http://
de Waele, A. 1923. Viscometry and Plastometry. J. Oil and Colour Chem- dx.doi.org/10.2118/5333-PA.
ists Association 6 (38): 33–69. Savins, J. G. 1958. Generalized Newtonian (Pseudoplastic) Flow in Sta-
Elise, A. 1995. An Improved Viscosity Equation to Characterize Shear- tionary Pipes and Annuli. Petroleum Trans., AIME 213: 325–332.
Thinning Fluids. Presented at the SPE International Symposium on SPE-1151-G-PA. http://dx.doi.org/10.2118/1151-G-PA.
Oilfield Chemistry, San Antonio, Texas, USA, 14–17 February. SPE- Savins, J. G. 1962. The Characterization of Non-Newtonian Systems by a
28973-MS. http://dx.doi.org/10.2118/28973-MS. Dual Differentiation-Integration Method. SPE J. 2(2): 111–119. SPE-
Eyring, H. 1936. Viscosity, Plasticity, and Diffusion as Examples of Abso- 163-PA. http://dx.doi.org/10.2118/163-PA.
lute Reaction Rates. J. Chemical Physics 4: 283–291. http:// Sisko, A. W. 1958. The Flow of Lubricating Greases. Industrial & Engi-
dx.doi.org/10.1063/1.1749836. neering Chemistry 50 (12): 1789–1792. http://dx.doi.org/10.1021/
Founargiotakis, K., Kelessidis, V. C., and Maglione, R. 2008. Laminar, ie50588a042.
Transitional, and Turbulent Flow of Herschel-Bulkley Fluids in Con- Snyder, H. W. and Castro, P. I. 2002. The Critical Reynolds Number for
centric Annulus. Can. J. Chem. Eng. 86: 676–683. http://dx.doi.org/ Rough-Wall Boundary Layers. J. Wind Engineering and Industrial
10.1002/cjce.20074. Aerodynamics 90 (1): 41–54. http://dx.doi.org/10.1016/S0167-6105(01)
Gahleitner, M., Bihlmayer, G., and Sobczak, R. 1991. Modified Carreau 00114-3.
Model. Kunststoffe, German Plastics 81 (7): 651–654. Steiner, E. H. 1958. A New Rheological Relationship to Express the Flow
Graves, G. W. and Collins, E. R. 1978. A New Rheological Model for Non- Properties of Melted Chocolate. International Chocolade Revue 13
Newtonian Fluids. SPE-7654-MS. http://dx.doi.org/10.2118/7654-MS. (7): 290–295.

Trinh, K. T. 2011. On the Critical Reynolds Number for Transition From University of Mining and Geology (Vietnam) in 2008 and has
Laminar to Turbulent Flow. http://arxiv.org/ftp/arxiv/papers/1007/ been a faculty member there since then. He earned his MSc
1007.2464.pdf (accessed April 2015). degree from the University of Tulsa in 2012, focusing on visco-
Tutuncu, A. N. 1988. Rheological Evaluation of Invert Emulsion Muds elastic properties of drilling fluids. Bui’s research interests
Under Downhole Temperature and Pressure, MSc thesis. The Univer- include numerical modeling, rheology, mass transport in well-
bore and porous media, and geomechanics.
sity of Texas at Austin, Austin, Texas.
Vipulanandan, C. and Mohammed, A. S. 2014. Hyperbolic Rheological Azra N. Tutuncu is the Harry D. Campbell Chair and a professor
Model With Shear Stress Limit for Acrylamide Polymer Modified in petroleum engineering and the director of UNGI at the Colo-
Bentonite Drilling Muds. J. Petrol. Sci. & Eng. 122: 38–47. http:// rado School of Mines. She held various research and leadership
dx.doi.org/10.1016/j.petrol.2014.08.004. assignments in well engineering, rock physics, geomechanics,
and subsurface research groups at Shell International E&P and
Wang, Z. and Tang, S. 1982. Casson Rheological Model in Drilling Fluid Shell Oil Company. Tutuncu leads several multidisciplinary
Mechanics. Presented at the International Petroleum Exhibition and research consortia focused on reservoir characterization,
Technical Symposium, Beijing, 17–24 March. SPE-10564-MS. http:// coupled geomechanics, wave propagation, and fluid-flow
dx.doi.org/10.2118/10564-MS. measurements and modeling for conventional, deepwater,
Wang, G. and Fan, H. H. 2010. Utility Calculation Method of New Four- and unconventional reservoirs. She has more than 150 publica-
Parameter Rheological Model for Drilling Hydraulics. Presented at the tions in peer-reviewed journals and conference proceedings,
SPS/SPE International Oil and Gas Conference and Exhibition, Bei- in addition to more than 100 research reports and novel-tech-
jing, 8–10 June. SPE 130421-MS. http://dx.doi.org/10.2118/130421- nology applications in deepwater Gulf of Mexico, unconven-
tional gas-and-oil shales, heavy-oil sands and carbonates,
MS.
integrated borehole stability, in-situ-stress and rock-property
Weir, I. S. and Bailey, W. J. 1996. A Statistical Study of Rheological
determination, nonlinear rock deformation and failure, pore
Models for Drilling Fluids. SPE J. 1 (4): 473–486. SPE-36359-PA. pressure and fracture-gradient prediction, stimulation, drilling,
http://dx.doi.org/10.2118/36359-PA. drill-in and fracturing-fluid design, and compatibility analysis.
Wilkinson, W. L. 1960. Non-Newtonian Fluids. New York: Pergamon Tutuncu holds several patents on pore-pressure prediction,
Press. attenuation, and acoustic stimulation. She is a member of the
US Environmental Protection Agency’s Scientific Advisory
Binh T. Bui is a PhD degree candidate of the Petroleum Engi- Board on Hydraulic-Fracturing Research, chairs the American
neering Department at Colorado School of Mines. He is doing Rock Mechanics Association (ARMA) Induced Seismicity Com-
research on multiscale and multiphysics reservoir modeling for mittee, is a former president of ARMA, is the founder and the
enhanced oil recovery (EOR) in unconventional reservoirs at faculty adviser of the ARMA Mines Student Chapter, and Li-
Unconventional Natural Gas and Oil Institute (UNGI). Bui censed Professional Engineer and is a Licensed Geoscientist in
earned his BS degree in petroleum engineering from Hanoi the State of Texas.
View publication stats


AGeneralized Rheological Modelfor Drilling Fluids With Cubic Splines

Uploaded by

Copyright:

Available Formats

AGeneralized Rheological Modelfor Drilling Fluids With Cubic Splines

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

AGeneralized Rheological Modelfor Drilling Fluids With Cubic Splines

Uploaded by

Copyright:

Available Formats

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

A Generalized Rheological Model for Drilling Fluids Using Cubic Splines

Article in SPE Drilling & Completion · April 2014

Binh T. Bui Azra Tutuncu

SEE PROFILE SEE PROFILE

The user has requested enhancement of the downloaded file.

A Generalized Rheological Model for

26 March 2016 SPE Drilling & Completion

ID: jaganm Time: 15:22 I Path: S:/DC##/Vol00000/150031/Comp/APPFile/SA-DC##150031

Fig. 1—Commonly used rheological models in the petroleum industry.

March 2016 SPE Drilling & Completion 27

ID: jaganm Time: 15:22 I Path: S:/DC##/Vol00000/150031/Comp/APPFile/SA-DC##150031

Table 1—Common Non-Newtonian-fluid models.

28 March 2016 SPE Drilling & Completion

ID: jaganm Time: 15:22 I Path: S:/DC##/Vol00000/150031/Comp/APPFile/SA-DC##150031

March 2016 SPE Drilling & Completion 29

ID: jaganm Time: 15:22 I Path: S:/DC##/Vol00000/150031/Comp/APPFile/SA-DC##150031

Because all coefficients in Eq. 13 are determined from the

equation (Eq. 15). ð21Þ

30 March 2016 SPE Drilling & Completion

ID: jaganm Time: 15:23 I Path: S:/DC##/Vol00000/150031/Comp/APPFile/SA-DC##150031

Sample Calculation. In this section, we present a detailed exam-

March 2016 SPE Drilling & Completion 31

ID: jaganm Time: 15:23 I Path: S:/DC##/Vol00000/150031/Comp/APPFile/SA-DC##150031

Shear Stress (Pa)

Table 3—Comparison between different rheological models.

32 March 2016 SPE Drilling & Completion

ID: jaganm Time: 15:23 I Path: S:/DC##/Vol00000/150031/Comp/APPFile/SA-DC##150031

Frictional Pressure Gradient (Pa/m)

Frictional Pressure Gradient (Pa/m)

220 Generalized Reynolds Number 230 Generalized Reynolds Number

Average Wall Shear Stress (Pa)

5 Average Velocity (m/sec) 7 Average Velocity (m/sec)

March 2016 SPE Drilling & Completion 33

ID: jaganm Time: 15:23 I Path: S:/DC##/Vol00000/150031/Comp/APPFile/SA-DC##150031

34 March 2016 SPE Drilling & Completion

ID: jaganm Time: 15:23 I Path: S:/DC##/Vol00000/150031/Comp/APPFile/SA-DC##150031

March 2016 SPE Drilling & Completion 35

ID: jaganm Time: 15:23 I Path: S:/DC##/Vol00000/150031/Comp/APPFile/SA-DC##150031

30 law, and Robertson Stiff models match well to the rheological

shown in Table 8. Yield Power Law model estimates quite well

Fig. 11—Flow curve from Fann 35 viscometer data for Fluid

36 March 2016 SPE Drilling & Completion

ID: jaganm Time: 15:23 I Path: S:/DC##/Vol00000/150031/Comp/APPFile/SA-DC##150031

Frictional Pressure Gradient (Pa/m)

980 Generalized Reynolds Number

Nomenclature f ¼ friction factor

March 2016 SPE Drilling & Completion 37

ID: jaganm Time: 15:23 I Path: S:/DC##/Vol00000/150031/Comp/APPFile/SA-DC##150031

38 March 2016 SPE Drilling & Completion

ID: jaganm Time: 15:24 I Path: S:/DC##/Vol00000/150031/Comp/APPFile/SA-DC##150031

March 2016 SPE Drilling & Completion 39

View publication stats

You might also like