SPE/IADC 57579

A New Hydraulics
*
Model for
*
Slim Hole Drilling* Applications.
Svein A. Hansen , Rolv Rommetveit , Njål Sterri and Bjarne Aas , RF-Rogaland Research, Antonino Merlo, ENI s.p.a.-
AGIP division.

*SPE Members.
Copyright 1999, Society of Petroleum Engineers Inc.
In ref. 7 a Reynolds number was defined from laminar flow
This paper was prepared for presentation at the 1999 SPE/IADC Middle East Drilling Technology considerations of Newtonian fluids in a concentric annulus. This
Conference held in Abu Dhabi, UAE, 8–10 November 1999.
Reynolds number could be used in the Coolebrook correlation,
This paper was selected for presentation by an SPE Program Committee following review of ref 14, to predict the frictional pressure loss for turbulent flow.
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 In refs. 8 and 9 a generalized Reynolds number was defined
correction by the author(s). The material, as presented, does not necessarily reflect any position
of the Society of Petroleum Engineers, its officers, or members. Papers presented at SPE for pipe flow with Non-Newtonian fluids. It was shown that
meetings are subject to publication review by Editorial Committees of the Society of Petroleum
Engineers. Electronic reproduction, distribution, or storage of any part of this paper for commercial
when using this definition, experimental friction factor, f, data
purposes without the written consent of the Society of Petroleum Engineers is prohibited. for a wide range of Non-Newtonian fluids could be reproduced.
Permission to reproduce in print is restricted to an abstract of not more than 300 words;
illustrations may not be copied. The abstract must contain conspicuous acknowledgment of where In ref. 10 a model was constructed for the flow of drilling
and by whom the paper was presented. Write Librarian, SPE, P.O. Box 833836, Richardson, TX fluids in concentric annuli, which is consistent with ref. 7 and
75083-3836, U.S.A., fax 01-972-952-9435.
ref. 9 in the appropriate limits.
The model described in the present work incorporates the
Abstract effects of eccentricity and drillstring rotation. The model is
A model to predict the pressure loss for slim hole drilling has consistent with refs. 7, 9 and 10 in the appropriate limits. The
been constructed. The model, which is constructed by theoretical model is formulated in a general way in order to facilitate its
and numerical analysis and experimental measurements, extension to other rheologies.
incorporates the effect of eccentricity, drillstring rotation and
rheology. Experimental work.
Experimental setup.
Introduction The fluids used in the experiments were water, different
Slim-hole drilling has been pursued in the last decade as a mean solutions of CMC (sodium carboxymethylcellulose) in water and
to drill exploration wells with a minimum of logistic support. different solutions of xanthan (AncoZan/Xanthan gum
Slimming down the well design will also give a cheaper well. polymers) in water. Over 1000 data points were collected during
However, the technique will bring new technical challenges, the experimental program.
such as the correct estimation of ECD and kick detection. Water is a Newtonian fluid. The solutions with water and
In a conventional well 90 % of the standpipe pressure CMC were well characterized by a two parameter power law
originates from friction in drillstring and drill bit. In a Slim-hole model. The solutions with water and xanthan almost have the
well 90 % of the standpipe pressure is due to friction in the character of Bingham fluid. Excluding the lower shear rates the
annulus. Correct estimation of the frictional pressure loss is water xanthan solutions are well described by the three-
crucial for estimating ECD, and hence also an important issue parameter Herschel-Bulkley model.
for well control. The flow loop consisted of a 4 m long vertical annulus, 2”
flexible hoses transporting fluid to and from the annulus, a
Motivation for the modeling approach. centrifugal pump delivering liquid to the annulus and a liquid
Over the last few years several works have addressed slim hole tank for the return from the annulus. The liquid flow rate to the
drilling hydraulics, refs. 1-6. The importance of the narrow gap, annulus was measured with an Endress & Hausser Flowtec DMI
the effect of eccentricity and the importance of rotation are 6531 Discomag electromagnetic flow meter. The maximum flow
generally acknowledged. Experimental data and field rate was 500 l/min. Fig. 1 shows a sketch of the slim hole flow
observations also show the importance of these effects. loop.
The modeling performed in these works has been restricted The annulus consisted of a transparent Plexiglas outer tube
to the flow of power law fluids and is generally not applicable to with inner radius 5.08-cm, and a steel inner tube with outer
other rheology models. radius 4.45-cm. The inner tube could be placed eccentric within
2 S. HANSEN, A MERLOAND R ROMMETVEIT SPE 57579

the outer tube. The eccentricity was controlled with a Nova 201 eccentric annulus. The model is valid for Newtonian, Power-
ultrasonic sensor. The maximum rotation rate of the inner tube Law, Bingham and Herschel-Bulkley fluids. Generally the
was 600 RPM. friction factore, f, is defined:
According to the information supplied by the producer of the
Plexiglas tube and steel tube the dimensions should be accurate τw D h DP (1)
within ±0.05 cm, see TABLE 1. f= = DL
The uncertainties in the annular dimensions are small O(1%),
1 ρV 2
2ρV 2
2
but due to the small annular gap, the uncertainty in the hydraulic
diameter is much larger, O(15%), see TABLE 2. When a generalized Reynolds number is used, the laminar
friction factor is calculated from:
Experimental observations.
The purpose of the experimental program was to provide data to 16 (2)
verify and complement the pressure loss model. In this f lam =
RE GEN
connection it is appropriate to discuss some of the observations
from the experiments.
In this publication is shown experimental results for four The onset of turbulence will occur at REGEN≈2100. For
fluids. The rheologies of the fluids are given in TABLE 3. transitional flow the friction factor is given by using Reed and
Eccentricity will reduce the frictional pressure loss relative to Pilehvari’s modification of Churchill’s friction factor, for
the concentric frictional pressure loss. This is clearly seen for generalized power law fluids, ref 10:
water in Fig. 2, where the data for the concentric annulus lies
9.4 × 10 −9
consistently above the data from the eccentric annulus. In Fig. 3 f tran = (RE GEN )2 (3)
the data for water for laminar and transitional flow is shown. For (4.767 − 2.167 × N ) 2

laminar flow the eccentric frictional pressure loss is lower than

the concentric frictional pressure loss. However, for transitional The turbulent friction factor is calculated by inserting the
flow the frictional pressure loss in the eccentric annulus generalized Reynolds number, REGEN, and the generalized power
approaches the frictional pressure loss in the concentric annulus. law index, N, into the modified Coolebroke expression:
This is caused by transition to turbulence at a lower flow rate in

( )
the eccentric annulus than in the concentric annulus. As the flow 1 4 (1− N 2 ) 0.4
= log 10 RE GEN f turb − (4)
rate increases further, both annuli become fully turbulent, and
the frictional pressure loss the concentric annulus again becomes f turb (N ) 0.75
(N )1.2
larger than for the eccentric one.
The same qualitative effect of eccentricity is seen in Fig. 6 Finally the method quoted in ref. 10 to calculate the friction
for the power law fluid NN03. factor for any flow regime can be used:
The effect of rotation is illustrated by the data for the power
law fluid NN01 in a concentric annulus, Fig. 4. The relative
effect of rotation is highest for the flow rates where the pure
((
−8
f 12 = f tran −8
+ f turb )) 8 12
+ f lam
12 (5)

axial flow close to the point of transition to turbulence. As the

axial flow increases further the relative effect of rotation
decreases. The effect of drill string rotation.
For the shear thinning fluids of this investigation, at low A complete frictional pressure loss model for slim hole
flowrates rotation will reduce the frictional pressure loss, and drilling should also take into account the effect of rotation of the
increase it for higher flow rates. This is shown in Fig. 5 and Fig. drill pipe. It is inherently difficult to take rotation into account
7. In particular, for the very low flow rates in Fig. 7, rotation by analytical or numerical methods. Due to coupling of the
reduces the frictional pressure loss dramatically. Since fluid different velocity components, the Navier-Stokes equations
NN04A exhibits a yield stress, rotation will help remove the cannot be solved analytically. Numerical solutions have been
unsheared plug and thus give a reduced frictional pressure loss. obtained only for a few special cases, ref. 23. In addition, to the
authors’ knowledge, there is no established model for the
The axial flow model. combined effect of rotation, turbulence and non-Newtonian
The crucial element of the axial flow model is the effects that can be used to perform large scale 3D numerical
generalized Reynolds number, which is defined on the basis of simulations. Therefore, in order to model the effect of rotation,
flow geometry and rheology data. The derivation of the we are to relying on tables based on experimental data.
generalized Reynolds number, REGEN, and the associated The effect of rotation is quantified in terms of the variable RFAC
generalized power law parameters K and N are given in the given as the ratio between the frictional pressure loss with
appendix. The calculation procedure described in the appendix is rotation and the frictional pressure loss without rotation:
similar for fluids flowing in a pipe or in a concentric or an
SPE 57579 A SLIM HOLE HYDRAULICS MODEL 3

calculated values of REGEN, TAE, N and δ the model interpolates

[DP DL] (6) in the table to find the appropriate value of RFAC.
R FAC (RE GEN , TA , N, δ) =
f WR
[DP DL] =
WR
E
f NR Comparing the model with small-scale laboratory data.
NR
There is a relatively large uncertainty in the hydraulic diameter
RFAC is dependent on the generalized Reynolds number, of the flow loop due to uncertainties in the physical dimensions
REGEN, Eqn.( 47 ), the effective Taylor number, TAE, Eqn. ( 7 ), of the flow loop. In order to determine the physical dimensions
the dimensionless eccentricity, δ, and the generalized power law of the flow loop more accurately, the experimental data with
index N. The effective Taylor number is defined as, ref. 23: water was compared to the model data and the radii (R1 and R2)
of the annulus were adjusted to obtain a match. The physical
2 dimensions of the flow loop obtained this way are well within
 ρω   D h  κ
4
(7)
TA E =     the experimental error bounds, see TABLE 4.
 µE   2  1 − κ In Fig. 11 the friction factor f vs. Re is shown for flow of water
in both the concentric and the eccentric annulus. The Reynolds
The definitions of the effective viscosity, µE, and the radius number, Re, is conventionally defined with the hydraulic
ratio, κ, are given in the appendix. The rotational rate ω in rad/s diameter, Dh, as the length scale, i.e.:
is given by:
ρVD h (9)
Re =
2π (8) µ
ω= × RPM
60
The model reproduces the data well, except in the region of
The data accepted for determining, RFAC were the cases where transition to turbulence where the model underpredicts the data.
measurements were performed both with and without rotation. Using the generalized Reynolds number, Eqn. ( 26 ), the same
The data are shown in Fig. 8 and Fig. 9 where RFAC is given as data is shown in Fig. 12. The model fits the experimental data
functions of REGEN and TAE, respectively. Observing the well both for the concentric and the eccentric annulus.
dependency of RFAC on REGEN in Fig. 8, we note that for values In Fig. 13 the friction factor, f, is shown as a function of REGEN
of REGEN higher than 100 rotation will give a RFAC greater than for the power law fluid NN01 without rotation. For a Newtonian
one. fluid REGEN is given by:
The relative effect of rotation is highest for generalized
Reynolds numbers in the range 500 to 10000. For this range of ρVD eq ( 10 )
RE GEN =
the generalized Reynolds number, rotation increases the µ
frictional pressure loss strongly. If the generalized Reynolds
number is less than 100 rotation will reduce the frictional
pressure loss. The model predictions lies well within the experimental
In Fig. 9 where the dependency of RFAC on TAE is shown, a point uncertainty. The model reproduces the point of onset of
of transition can be identified at TAE∼20000. For values of TAE turbulence. When the flow is turbulent the experimental data and
higher than approx. 20000 the corresponding rotation rate will the model deviates. This is due to drag reduction, which is a
increase the frictional pressure loss relative to the non-rotating feature of turbulent flow of polymer solutions. The model does
case, whereas for TAE less than 10000 the corresponding rotation not account for drag reduction.
rate will reduce the frictional pressure loss. In Fig. 14 the f vs. REGEN relationship is shown for pure axial
The observed transition point is consistent with theoretical flow and two rotation rates for NN01. Again the model
predictions in ref. 22, and with numerical calculations in ref. 23. reproduces the data well until the onset to turbulence.
A detailed inspection of our data does not reveal any effect from
the axial flow rate on the point of transition in the pressure loss Comparison model with full-scale experimental data.
data. It is reported in ref. 22 that the onset of Taylor vortices is In 1994 a series of full-scale test related to slim hole drilling
delayed with increasing axial flow. If the increase in the were performed in an inclined onshore research well in
frictional pressure loss with rotation is due to the onset of Taylor Stavanger, Norway. These experiments are described in ref. 11.
vortices this effect should be observed also in our data. This is The test included hydraulics test, and details of the experimental
not evident from our data the way are presented. However, the setup can be found ref. 11. The hydraulics tests were performed
data are from several different fluids, and a detailed analysis for as circulation tests were the drilling fluid was circulated and the
each fluid could possibly lead to a different conclusion. pressure was measured at different depths in the well.
From the data in Fig. 8 and Fig. 9 a table is constructed for RFAC, One of the circulation tests is compared to the model
which is used by our pressure loss model. Based on the predictions in Fig. 15 and Fig. 16. The drilling fluid was water in
this case and the data were collected at two locations in the well.
4 S. HANSEN, A MERLOAND R ROMMETVEIT SPE 57579

In the upper part of the well the drillstring is concentric and in model NN, see Eqn.( 43 )
the lower part of the well the drillstring lies against the casing REGEN The generalized Reynolds Number, see Eqn. ( 47 ).
wall, i.e. positioned fully eccentric. RFAC The effect of rotation on the frictional pressure loss, see
In the upper part of the well the model predicts slightly lower Eqn. ( 6 ).
frictional pressure losses than what was measured, Fig. 15. The RPM Rotational rate, [rev./min].
tool-joints on the drill string will contribute to the TAE The effective Taylor number, see Eqn. ( 7 ).
experimentally measured pressure loss. In addition the annular V Mean velocity, [m/s].
gap is very small, 0.01 m and the frictional pressure loss is very αBH A parameter, see Eqn. ( 30 ).
sensitive to the uncertainties in the annular dimensions. αHB A parameter, see Eqn. ( 32 ).
In the lower part of the well the drill string was a flush pipe δ The dimensionless eccentricity, see Eqn. ( 18 ).
and the well was 63° inclined. The model predictions in Fig. 16 γ! The shear rate, [s-1]
for a fully eccentric annulus lie above the measured
ε The dimensional eccentricity, see Fig. 17, [m].
experimental data. However, if it is assumed that the annulus is
κ The radius ratio, see Eqn. ( 17 ).
concentric, the model will severely overestimate the
experimental data. In this well section the annular gap is small µ Viscosity, [Pas].
0.009 m and the model predictions in Fig. 16 are well within the µE The effective viscosity, see Eqn. ( 45 ), [Pa s].
experimental uncertainty. µp The plastic viscosity, see Eqn. ( 13 ), [Pa s].
ρ Density, [kg/m3]
Conclusions. τy Yield stress, [Pa].
A slim hole hydraulics model including rotation has been τw Wall shear stress, [Pa].
constructed using a combined numerical and experimental ω The rotational rate, [rad/s].
approach. ψ’ Tri-Gamma function, see ref. 24.
For the non-rotating case the model is also valid for non-slim ζ1 The inner radius in bipolar coordinates, ref. 18.
hole cases of pipe, concentric and eccentric annuli. The model ζ2 The outer radius in bipolar coordinates, ref. 18.
applies to Newtonian fluids, power law fluids, Bingham fluids Γ(κ,δ) A function, implicitly given in ref. 15.
and Herschel-Bulkley fluids. Ω(κ,δ) A function, see Eqn. ( 20 ).
Comparison with small-scale laboratory data and full-scale ΩPL(..) A function implicitly defined by Eqn.( 37 ).
data from a research drilling rig verifies the models
ΩBH(..) A function implicitly defined by Eqn.( 40 ).
performance.
ΩHB(..) A function implicitly defined by Eqn.( 42 )
Subscripts
Nomenclature
BN=Bingham Model
Bn The Bingham number, see Eqn. ( 39 ).
HB=Herschel-Bulkley Model.
c A parameter, c = τ y τ w . lam=laminar flow regime.
Deq The equivalent diameter, see Eqn ( 27 ) ,[m]. NN=any rheology model.
Dh The hydraulic diameter, see Eqn. ( 16 ),[m]. NR=No Rotation.
DP/DL Frictional pressure loss, [Pa/m]. PL=Power Law model
f The friction factor, see Eqn. ( 1 ). trans.=transitional flow regime.
k consistency index, [Pa sn] turb.=turbulent flow regime.
K Generalized power law parameter, ,[Pa sN] WR=With rotation.
n Power law flow behavior index.
m Herschel-Bulkley flow behavior index. Acknowledgements.
N Generalized power law index. Parts of this work were performed in a JIP founded by Agip,
R1 The outer radius of the drill pipe, [m]. Amoco and Norsk Hydro, and we will express our gratitude for
R2 The inner radius of the casing or open hole, [m]. the permission to publish this paper. We will also thank Dr.
Regen A generalized Reynolds Number, see ref. 9. Knut Bjørkevoll for valuable comments.
eq
Re PL A Reynolds number for power law fluids based on the
equivalent diameter, see Eqn. ( 36 ). References.
eq
1. Millheim K. K. and Walker S. H.:”An Innovatrive Approach to
Re BH A Reynolds number for Bingham fluids based on the Exploration Drilling: The Slim Hole High-speed Drilling System”;
equivalent diameter, see Eqn. ( 38 ). SPE19525, Paper presented at the 1989 Annual Technical
Conference and exhibition, San Antonio, TX, October8-11, 1989.
Re eq
HB A Reynolds number for power law fluids based on the
2. Bode D.J., Noffke R. B. and Nickens H. V.:”Well-Control
equivalent diameter, see Eqn.( 41 ). Methods and Practices in Small-Diameter Well Bores”, JPT,
RE GEN The generalized Reynolds number, defined for rheology November 1991, pp. 1380-1386
NN
SPE 57579 A SLIM HOLE HYDRAULICS MODEL 5

3. Thonhauser G., Milheim K. K. and Spoeker H. F.:”The 22. DiPrima R. C.: “The stability of a viscous fluid between rotating
Applicability of Current Slimhole Hydaulics Algorithms Under cylinders with an axial flow”, Jou. Fluid. Mech., 9, p621, 1960.
Field Conditions – A Critical Review”, SPE30489, Paper 23. Lockett T. J., Richardson S. M. and Worraker W. J.: “The
presented at the 1995 Annual Technical Conference and Importance of Rotation Effects for Efficient Cuttings Removal
exhibition, Dallas, USA, October 22-25, 1995. During Drilling”, SPE 25768.
4. McCann R. C., Quigley M. S., Zamora M. and Slater K. 24. Abramowitz M. and Stengun I. A.: ”Handbook of mathematical
S.:”Effects of High-Speed Pipe Rotation on Pressures in Narrow functions”, Dover Publications, New York.
Annuli”, SPE 26343 , Paper presented at the 1993 Annual
Technical Conference and exhibition, Houston, TX, USA, October Appendix-1 Derivation of the model
3-6, 1993. Rheology Models.
5. Hansen S. A. and Sterri N.:”Drill Pipe Rotation Effects on The pressure loss model will be valid for Newtonian fluids,
Frictional Pressure Losses in Slim Annuli”, SPE30488, Paper
power law fluids, Bingham fluids and Herschel-Bulkley fluids.
presented at the 1995 Annual Technical Conference and
exhibition, Dallas, USA, October 22-25, 1995. For the Newtonian fluid model the shear stress is proportional to
6. Cartalos U., King I. Dupuis D. and Sagot A.:”Field Validated the shear rate:
Hydraulic Model Predicions Give Guidelines for Optimal Annular
Flow in Slim Hole Drilling”, SPE35131, Paper presented at the τ w = µγ! ( 11 )
1996 IADC/SPE Drilling Conference, New Orleans, Louisiana,
March 12-15, 1996.
7. Jones, O. C. and Leung J. C. M.:”An Improvement in the
Where µ is the viscosity.
Calculation of Turbulent Friction in Smooth Concentric Annuli”; Both the power law model and the Bingham model are two-
Jou. of Fluids Engineering, Vol. 103, Dec. 1981, pp. 615-623. parameter models. The power law model is given by:
8. Metzner A. B. and Reed J. C.:”Flow of Non-Newtonian Fluids-
Correlation of the Laminar, Transition and Turbulent-flow τ w = kγ! n ( 12 )
Regions”, A.I.Ch.E.Jou. Vol 1., pp434-440, 1955.
9. Dodge, D. W. and Metzner A. B.:”Turbulent Flow of Non-
Newtonian Systems”, A.I.Ch.E.Jou. Vol 5., pp189-440, 1955. where k is the consistency index and n is flow behavior index.
10. Reed, T. D. and Pilehvari, A.A.:”A New Model for Laminar, The Bingham model is given by:
Transitional and Turbulent Flow of Drilling Muds”; SPE25456,
Paper presented at Production Operations Symposium in τ w = τ y + µ p γ! ( 13 )
Oklahoma City, OK, USA March 21-23, 1993.
11. Steine O. G., Rommetveit R., Maglione R. and Sagot A.:”Well
Control Experiments related to Slim Hole Drilling”, SPE35121, where τY is the yield stress and µp is the plastic viscosity. Finally
Paper presented at the 1996 IADC/SPE Drilling Conference, New the three-parameter Hershel-Bulkley model is given as:
Orleans, Louisiana, March 12-15, 1996.
12. EXLOG staff: “Theory and Application of Drilling Fluid τ w = τ y + kγ! m ( 14 )
Hydraulics”; D. Reidel Publishing, Boston, 1985.
13. Bird R.B., Stewart W. E. and Lightfoot E. N.; ”Transport
Phenomena”, Wiley, New York, 1960 Where m is the flow behavior index.
14. Skelland A.H.P.: ”Non-Newtonian Flow and Heat Transfer”, All three non-Newtonian fluids are so called generalized
Wiley, New York, 1967. Newtonian fluids in the sense that all three models can be
15. Shah R.K. and London A. L.: “Laminar Flow Forced Convection written in the following form:
in Ducts”, Academic Press, New York, 1978.
16. Piercy N. A. V, Hooper M. S. and Winney H. F.: “Viscous Flow τ w = µ e γ! ( 15 )
through Pipes with Cores”, Proc. Roy. Soc. 1933.
17. Caldwell J.: ”The Hydraulic mean depth as a Basis for Form
Comparison in the flow of fluids in Pipes”, Jou. Roy. Tech. The effective viscosity µe is given in TABLE 5.
College (Glasgow), Vol. 2, part 2., Jan. 1930, pp 203-220.
18. Bird, R. B., Armstrong, R. C. and Hassager, O.: ”Dynamics of Laminar Flow of Newtonian fluids in pipes, concentric and
polymeric liquids”, Vol. 1, 2nd. edition, Wiley, 1987. eccentric annuli.
19. Haciislamoglu, M. and Langlinais, J.:”Non-Newtonian Flow in The eccentric annulus is characterized by three parameters R1, R2
Eccentric Annuli”, 13th Annual ASME Energy-Sources and ε, as given in Fig. 17. Using these quantities the hydraulic
Technology Conference, Drilling Technology Symposium
diameter is given by:
Proceedings, New Orleans, January 14-18, 1990, PD-27,115-123.
20. Hansen S. A.:”Laminar Non-Newtonian Flow in an Eccentric
Annulus, a numerical solution”, paper presented at the 2nd annual D h = 2(R 2 − R 1 ) = 2R 2 (1 − κ ) ( 16 )
meeting of the Nordic Rheology Society, Gothenburg, Sweden,
August 19-20,1993. Where we have introduced the dimensionless radius ratio:
21. Weast R. C. Editor.: ”CRC Handbook of Chemistry and Physics”,
66th. edition, page F-37, CRC press Inc., 1986.
6 S. HANSEN, A MERLOAND R ROMMETVEIT SPE 57579

where ψ’ is the so-called tri-gamma function, ref. 24.

κ = R1 R 2 ( 17 ) Eqns. ( 19 )-( 23 ) indicate that the expressions for the
pressure drop of Newtonian flow in pipes, concentric and
The dimensionless eccentricity is defined as: eccentric annuli are mathematically similar and can be expressed
within the same formalism.
ε ( 18 ) To obtain this we first reformulate Eqn. ( 19 ) to:
δ=
(R 2 − R 1 )
DP 32µV
= Ω(κ, δ ) ( 24 )
The dimensionless eccentricity may range from zero to one, DL D 2h
i.e. from concentric to fully eccentric.
Analytical expressions for the flow of a Newtonian fluid in Referring to Fig. 10, we note that in a fully eccentric annulus,
various conduits can be found in the literature, e.g. in refs 13-16. (δ=1), the frictional pressure loss is only approx. 40% of the
For the flow conduits of relevance in drilling applications, pipes value for a concentric annulus, (δ=0).
and concentric and eccentric annuli, the analytical expressions We use the friction factor, f, as defined in the conventional
relating the mean velocity V to the applied pressure gradient way:
DP/DL can be expressed in the following form: DP D h
D 2 DP ( 25 )
Ω(κ, δ ) DL 4
−1
V= h f=
32µ DL ( 19 ) 1 ρV 2
2
where the different flow conduits correspond to the different Requiring that the product of the Reynolds number and the
values of the parameters κ and δ in accordance with friction factor is equal to 16 provides consistency to the pipe
flow case. The generalized Reynolds number is then defined as:
ρVD eq
Re GEN = ( 26 )
µ
where the equivalent diameter, Deq, is defined as:

( 27 )
TABLE 6. Dh
D eq =
In our model the function Ω(κ,δ) is explicitly given as: Ω(κ, δ )
Ω(κ, δ ) =
(1 − κ )(1 − κ )
2 2

  ∞
ne − n ( ζ + ζ )  
1 − κ 4 + Γ(κ, δ ) 1 − 2∑ ( 20 )
2 1


n =1 sinh[n (ζ1 − ζ 2 )] 

  ζ 2 − ζ1 This way, the expression for the frictional pressure drop in
where the explicit expression for Γ(κ,δ) can deduced from ref. concentric and eccentric annuli can be formulated equally to the
15 and ζ are given in the nomenclature section. frictional pressure drop in a pipe with a diameter equal to the
The expression for Ω(κ,δ) in Eqn. ( 20 ) is quite complicated. equivalent diameter.
Ω(κ,δ) is evaluated as a function of κ for various values of δ in
Fig. 10. Explicit expressions of the Ω(κ,δ) can be found in some Pipe Flow of Non-Newtonian fluids.
cases. For the three Non-Newtonian rheology models considered,
For the case of pipe flow one has: analytical expressions exist for the relation between the
frictional pressure loss and the mean velocity.
Ω(0,0 ) = Ω(0,1) = 1 ( 21 ) For power law fluids the mean velocity in the pipe is given
as a function of the frictional pressure loss as:
1n
For the case of flow in a concentric annulus Ω(κ,δ) can be D h n  DP D h 
V=   ( 28 )
simplified to: 1 + 3n  DL 4k 
Ω(κ,0) =
(1 − κ)2 where k and n is given by Eqn ( 12 ).
1 − κ4 1− κ2  ( 22 ) For a Bingham fluid the mean velocity in the pipe is given as:
 − 
1− κ
2
ln(1 κ ) 
( 29 )
and finally for the case of flow in an annulus with unit Dh τy
eccentricity: V= α BH
2µ p
Ω (κ,1) =
(1 − κ )(1 − κ )
2 2

1 − κ 4 − 4 κ 2 ψ ' [ 1(1− κ )] ( 23 )
SPE 57579 A SLIM HOLE HYDRAULICS MODEL 7

where αBH is defined as:

c 4 − 4c + 3 Laminar flow of Non-Newtonian fluids in concentric and
α BH = ( 30 ) eccentric annuli.
12c
For the three rheological models considered her semi-analytical
and c is the ratio between the yield stress and the wall shear solutions are available for power-law and Bingham fluids in a
stress, τ y τ w . concentric annulus, refs. 12 and 13. If the annulus is eccentric no
Finally, for a Herschel-Bulkley fluid the mean velocity in the analytical solutions exist. In this case the governing equations
pipe vs. the frictional pressure loss is given as: must be solved numerically.
A numerical code was constructed which, based on input of
1
( 31 ) rheological data, geometry data and the frictional pressure loss,
D  τy n
V= h   α HB calculates the corresponding flow rate. Several such codes have
 k 
2   also been reported in the literature, e.g., refs. 19 and 20.
For each of the rheological models several hundred
where αHB is defined by the following relation: simulations were performed in order to account for every
combination of the input parameters.
(1 − c ) m +1   (1 − c ) 2c(1 − c )
m  1   2
c2  Power-law fluids.
α HB = + +  ( 32 )
1
 1 + 3m 1 + 2m 1 + m  For the flow of a power law fluid in an eccentric annulus the
c m

five dimensionless parameters κ, δ, n, Re eq PL and f uniquely

The parameter k and m are given from the definition of the describes the flow situation.
Herschel-Bulkley rheological model, Eqn. ( 14 ), and the Here we have introduced Re eq PL , a Reynolds number based
parameter c is defined in the same manner as in the Bingham on the equivalent diameter and defined as:
fluid case.
For all three rheological models a Reynolds number can be D neq V (2−n )ρ ( 36 )
PL =
Re eq
identified from the relation:
 3n + 1 
n

8(n −1) k 
f Re = 16 ( 33 )  4n 

This is obtained by manipulating Eqns. ( 28 ), ( 29 ) and ( 31 ). This definition ensures that Re eq PL is equal to Re N in the
This way a pipe flow Reynolds number for each of the three
rheological models is found. The expressions are given in Newtonian limit, n→1. Numerical simulations were performed
TABLE 7. for the relevant range of κ, δ and n. For each numerical
With some manipulation we can also express the Reynolds simulation the quantity Ω PL (κ, δ, n ) is evaluated. Ω PL (κ, δ, n ) is
numbers in terms of the generalized power law parameters K and defined through the relation:
N. Metzner and Reed, ref. 8, defined a dimensionless
generalized Reynolds number as: f Re eq
Ω PL (κ, δ, n ) = PL ( 37 )
16
D hN V ( 2− N ) ρ ( 34 )
Re gen =
8(N −1) K ΩPL(κ,δ,n) quantifies the effect of the power law rheology on the
flow in a concentric and eccentric annulus.
The generalized power law parameters for the various In Fig. 18 ΩPL(κ,δ,n) is shown for all values of κ and δ for
rheological models are given in TABLE 8. n=0.7. For the concentric annulus ΩPL(κ,δ,n) is fairly constant
For laminar flow it may seem superfluous to definine a for large values of κ. Increasing the eccentricity δ reduces
generalized Reynolds numbe, and generalized power law ΩPL(κ,δ,n) . For an annulus with unit eccentricity ΩPL(κ,δ,n)
parameters. The advantage is that these quantities derived for reduces as κ increases.
laminar flow can be used to predict the frictional pressure loss in
turbulent flow, Eqn ( 4 ):
Bingham fluids.
Defining a Reynolds number based on the equivalent
1
f
4
[
0.4
= 0.75 log 10 Re gen f (1− N 2 ) − 1.2
N N
] ( 35 ) diameter for the flow of a Bingham fluid in concentric and
eccentric annuli analogously to the power-law fluid case, we
write:
Using Eqn. ( 35 ) the friction factor for turbulent flow can be
calculated. A comparison of Eqn. ( 35 ) with experimental data
is given in ref. 9.
8 S. HANSEN, A MERLOAND R ROMMETVEIT SPE 57579

This function is evaluated in Fig. 20 for the case Bn=0.05 and

4ρVD eq ( 38 ) m=0.8. The dependence of ΩHB(κ, δ, Bn, m) on κ and δ is
BH =
Re eq cα BH
µp qualitatively similar as for the functions ΩPL and ΩBH shown in
Fig. 18 and Fig. 19, respectively.

This definition of Re eq
BH , is equal to Re N in the Newtonian Features independent of rheology model.
limit, cαBH→1/4. The flow of a Bingham fluid in concentric or Using the subscripts NN to denote any rheology model, some
eccentric annuli is thus characterized by κ, δ, Re eq general considerations can be given. A generalized Reynolds
BH . One also
number can be defined by the requirement that its product with
needs to include the influence of the yield strength τY, this is the friction factor is equal to 16, as in the case of Newtonian
done through the introduction of the Bingham number, Bn, pipe flow:
which is defined as:
Re eq ( 43 )
2τ Y NN =
RE GEN NN
⇒ fRE GEN
NN = 16
Ω NN (...)
Bn = ( 39 )
DP
DL
(R o − R i )
An effective viscosity µe can be defined from the relation:
In the Newtonian limit τY →0 and Bn→0. The Bingham number
is not unique for a given geometry and rheology. The definition  8V  ( 44 )
of Eqn ( 39 ) shows that that the Bingham number is dependent τw = µe  
 D eq 
on the flow rate and the corresponding frictional pressure loss.  
The frictional pressure loss for a Bingham fluid in an annulus
is quantified by evaluating the product f Re eq
BH for the relevant
Deriving τW from Eqn. ( 43 ) and manipulating Eqn ( 44 ) an
explicit expression is obtained:
range of κ, δ and Bn. A function, ΩBH(κ,δ,Bn), can be identified
from this product through the relation:
ρVD eq ( 45 )
µe =
= 16Ω BH (κ, δ, Bn )
GEN
f Re eq
( 40 ) RE NN
BH

Finally we may define a generalized Reynolds number by

Because of the interdependence of f and Re eq BH with Bn, the
requiring that any fluid flowing in a pipe, a concentric and an
parameters will have to be calculated through an iterative eccentric annulus can be transformed into a generalized power
process. law fluid with the following relation between the wall shear
A number of simulations were performed for the relevant stress and the shear rate:
ranges of κ, δ and Bn in order to quantify ΩBH(κ,δ,Bn). In Fig.
19 ΩBH(κ,δ,0.15) is shown. For the concentric annulus  8V 
N
( 46 )
ΩBH(κ,δ,0.15) is fairly constant for large values of κ. Increasing τ W = K 
 D eq 
the eccentricity δ reduces ΩBH(κ,δ,Bn). For an annulus with unit  
eccentricity ΩBH(κ,δ,Bn) reduces as κ increases.
The generalized power law parameters K and N are not identical
Herschel-Bulkley fluids. to the power law parameter k and n. Requiring the formalism to
The development here is equivalent to the Power Law fluid be analogous to the Newtonian pipe flow case, a generalized
and Bingham fluid cases. A Herschel-Bulkley Reynolds number Reynolds number is derived by requiring that the product
based on the equivalent diameter is defined by: between the friction factor and the Reynolds number should be
equal to 16, i.e:
ρV (2−m )D meq 8 ( 41 )
HB =
Re eq cα mHB ρV (2− N )D eq
N
k 2m ( 47 )
fRE GEN = 16 ⇒ RE GEN = (N −1)
K8
The frictional pressure loss of a Herschel-Bulkley fluid can
again be quantified by evaluating the product f Re eq
HB for the
The generalized Reynolds number can be calculated for all
relevant ranges of κ, δ, Bn and m: rheology models considered here. The generalized power law
parameters are given by the expressions in
HB = 16Ω HB (κ, δ, Bn, m )
f Re eq ( 42 )
SPE 57579 A SLIM HOLE HYDRAULICS MODEL 9

TABLE 9.
For a power law fluid one can derive analytical expressions
for K and N since ΩPL(κ,δ,n) is independent of the flow rate. For
a Bingham fluid and a Herschel-Bulkley fluid one must use the
knowledge of the laminar wall shear stress to evaluate K and N
numerically through an iterative procedure, since ΩBH(κ,δ,Bn)
and ΩHB(κ,δ,Bn,m) are dependent on the flow rate through the
Bingham number. In the pipe flow limit, ΩPL(κ,δ,n)=1, we note
that the expression for K for a power law fluid is identical to the
value given in TABLE 8.
The frictional pressure loss in the turbulent regime can be
calculated with help of Eqn. ( 35 ). The generalized power law
parameter, N, as determined form the equations above for
laminar flow, is here to be used.
10 S. HANSEN, A MERLOAND R ROMMETVEIT SPE 57579

TABLE 1 THE UNCERTANITY IN THE DIMENSIONS OF THE FLOW

LOOP
TABLE 6 THE CORRESPONDENCE BETWEEN THE VALUES OF
R [m] ∆R [m] ∆R [%] Ω(κ,δ
δ) AND THE VARIOUS FLOW CONDUITS.

Ri. 0.0445 ±0.0005 ±1.12 Ω(0,0) Pipe

Ro. 0.0508 ±0.0005 ±0.984 Ω(0,1) Pipe

Ω(κ,0) Concentric Annulus
TABLE 2 THE UNCERTANITY IN THE HYDRAULIC DIAMETER OF
THE FLOW LOOP Ω(κ,1) Annulus with unit eccentricity.
Dh [m] Dh,min [m] Dh,max [m] ∆Dh [%] Ω(κ,δ) Eccentric annulus.
0.0126 0.0106 0.0146 16 TABLE 7 THE GENERALISED PIPE FLOW REYNOLDSNUMBER FOR
THE VARIOUS RHEOLOGY MODELS.
TABLE 3 THE RHEOLOGY OF THE TEST FLUIDS
Rheology Model Pipe flow Reynolds number
Fluid Rheology τy [Pa] K [Pa sn] Index n or m
Power Law D nh V (2− n )ρ
Water N 1.0 ×10-3 1 Re PL = n
 3n + 1 
8(n −1) k 
NN01 PL 9.75×10-3 0.955  4n 
NN02 PL 7.15×10-1 0.5605 Bingham Model 4ρVD h
Re BH = cα BH
µp
NN03 PL 9.5 ×10-3 0.965
Herscel-Bulkley D mh V (2− m )ρ 8
NN04A HB 2.45 0.14 0.60 Re HB = cα mHB
k 2m
NN04B HB 1.91 0.23 0.54
TABLE 8 THE EXPLICIT ANALYTICAL EXPRESSIONS FOR THE
TABLE 4 THE NOMINAL AND CORRECTED PHYSICAL DIMENSION GENERALISED POWER LAW PARAMETERS N AND K FOR PIPE
OF THE FLOW LOOP FLOW.
Nominal Error [m] Error [%] Corrected Change Change
[m] [m] [%] [m] Fluid K N
Ri 0.0445 ±0.0005 ±1.12 0.04488 0.854 0.00038
Ro 0.0508 ±0.0005 ±0.984 0.05044 -0.708 -0.00036 Newtonian K=µ N=1
n
Power Law  3n + 1  N=n
TABLE 5 THE EFFECTIVE VISCOSITY FOR THE VARIOUS K = k 
RHEOLOGY MODELS.  4n 
Rheology Model Effective Viscosity Bingham τw 8V ∂τw
K= N=
µe = µ  8V 
N
τWDh ∂8V 
Newtonian  D   D
 h   h

Power Law µ e = kγ! n −1

τY Herscel-Bulkley τw 8V ∂τw
Bingham K= N=
µe = + µp  8V 
N
τWDh ∂8V 
γ!  D   D
 h
 h 
τY
Herschel Bulkley µe = + kγ! m−1
γ!
SPE 57579 A SLIM HOLE HYDRAULICS MODEL 11

5000
Concentric, δ= 0.0
4500 Eccentric, δ=0.7
TABLE 9 The explicit analytical expressions for the generalized
power law parameters K and N. 4000

3500
Fluid K N
3000

DP/DL [Pa/m]
Newtonian K=µ N=1
2500
 3N + 1 
N

K = kΩ PL (κ, δ, N )
Power Law N=n
 2000
 4N 
1500

1000
τW 8V ∂τ W
Bingham
K= N= 500
 8V 
N
τ W D eq  8V 
Fluid
 D  ∂ 
 D 
0
 eq  eq 0 20 40 60 80 100 120 140 160 180 200
Q [l/min]
Fig. 2 THE FRICTIONAL PRESSURE LOSS DP/DL, VS. FLOWRATE
τW 8V ∂τ W Q, FOR WATER.
Herschel-
K= N=
 8V 
N
τ W D eq  8V  300
Bulkley
 D  ∂  Concentric, δ= 0.0
Fluid  eq   D eq  Eccentric, δ=0.7
250

200
DP/DL [Pa/m]

Valve

150
2" Pressure sensor

100
Sparkplugg

50
102.5 cm
Axial
62.5 cm
∆ P-sensor Radial
∆ P-sensor 0
0 5 10 15 20 25 30 35 40
Axial Sparkplugg 10 cm
62.5 cm Sparkplugg Q [l/min]
∆ P-sensor Radial 4m
∆ P-sensor Fig. 3 THE FRICTIONAL PRESSURE LOSS DP/DL, VS. FLOWRATE
Axial
Q, FOR WATER IN THE REGION OF TRANSITION TO TURBULENCE.
62.5 cm
∆ P-sensor
228 cm 7000
No Rotation
122.5 cm 100 RPM
6000
200 RPM
300 RPM
Liquid Flowmeter
400 RPM
5000
500 RPM
Pump
600 RPM
Valve
DP/DL [Pa/m]

2"
Liquid Tank Motor 4000

3000
Fig. 1 The slim hole flow loop.

2000

1000

0
0 50 100 150 200 250
Q [l/min]
Fig. 4 DP/DL vs. Q FOR VARIOUS ROTATION RATES, δ=0.0,
FLUID NN01.
12 S. HANSEN, A MERLOAND R ROMMETVEIT SPE 57579

15000
No Rotation
200 RPM
400 RPM 4
600 RPM

3.5
10000
DP/DL [Pa/m]

2.5

FAC
5000 2

R
1.5

0
0 20 40 60 80 100 120 140 160 180 200 0.5
Q [l/min]
Fig. 5 DP/DL vs. Q FOR VARIOUS ROTATION RATES, δ=0.7, FLUID 0
NN02. 10
0 1
10 10
2 3
10
4
10
5
10
REGEN

8000
Fig. 8 RFAC as a function of the generalised Reynolds number,
δ=0.0
δ=0.7 REGEN.
7000

6000
4

5000
DP/DL [Pa/m]

3.5

4000
3

3000
2.5

2000
FAC

2
R

1000
1.5

0
0 50 100 150 200 250 300 350 1
Q [l/min]
Fig. 6 DP/DL vs. Q FOR FLUID NN03 IN A CONCENTRIC ANNULUS 0.5
δ=0.0, AND AN ECCENTRIC ANNULUS δ=0.7.
0
0 1 2 3 4 5 6 7 8
7000 10 10 10 10 10 10 10 10 10
TAE
No Rotation
100 RPM
6000 200 RPM Fig. 9 RFAC as a function of the effective Taylor number, TAE
400 RPM
600 RPM
5000
DP/DL [Pa/m]

4000

3000

2000

1000

0
0 20 40 60 80 100 120 140 160 180 200
Q [l/min]
Fig. 7 DP/DL VS. Q FOR FLUID NN04A FOR VARIOUS ROTATION
RATES IN A CONCENTRIC ANNULUS, δ=0.0.
SPE 57579 A SLIM HOLE HYDRAULICS MODEL 13

-1
1.5 Ω (κ,0.0) ↓ 10
↑ Ω (κ,0.1)
Exp Data
1.4 Model
↑ Ω (κ,0.2)
Experimental Error Bounds
1.3 ↑ Ω (κ,0.3)

1.2 ↑ Ω (κ,0.4)

1.1
Ω (κ, δ)

Ω (κ,0.5) ↑ -2
10

f
1
Ω (κ,0.6) ↑
0.9

Ω (κ,0.7) ↑
0.8

↓ Ω (κ,0.9) Ω (κ,0.8) ↑
0.7

Ω (κ,1.0) ↑
0.6
-3
10
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 2 3 4
10 10 10
κ
ReGEN
Fig. 13 f vs. REGEN comparison between experimental data and
Fig. 10 Values for the Ω( κ,δ
δ) function.
model predictions for fluid NN01.

-1
10
Exp Data, δ=0.0
Exp Data, δ=0.7
Model, δ=0.0
Model δ=0.7 10
-1

-2
10
f

-2
10
f

Data 0 RPM
-3
10 Model 0 RPM
2 3 4 5
10 10 10 10 Data 200 RPM
Re Model 200 RPM
Fig. 11 The friction factor f vs. Re for water, model predictions vs. Data 500 RPM
Model 500 RPM
Experimental data. -3
10
2 3 4
10 10 10
0 ReGEN
10
Fig. 14 f vs. REGEN for various rotation rates for fluid NN01,
Model comparison between model prediction and experimental data.
Exp Data, δ=0.0
Exp. Data δ=0.7
Exp error 0.14
Exp. data
Model: δ=0.0
Frictional Pressure Loss [bar/m]

0.12
10
-1 Model: δ=1.0

0.1

0.08
f

-2 0.06
10
0.04

0.02

-3
0
10
2 3 4 5
10 10 10 10 -0.02
REGEN 75.7 75.8 75.9 76 76.1 76.2 76.3 76.4 76.5
Fig. 12 The friction factor f vs. REGEN for water, model predictions Time
vs. Experimental data. Fig. 15 Comparison between model and full scale experimental data
from the upper part of the research well.
14 S. HANSEN, A MERLOAND R ROMMETVEIT SPE 57579

0.07 1.05

Exp. data Ω PL(κ, δ,0.7)

Model: δ=0.0 -δ=0.1
0.06 Model: δ=1.0
δ=0.2
1
Frictional Pressure Loss [bar/m]

δ=0.3
0.05
δ=0.4
0.04
0.95 δ=0.5

δ=0.6
0.03
δ=0.7

0.02
0.9 δ=0.8
-δ=0.9
δ=0.0
0.01 δ=1.0
0.85
0 0.2 0.4 0.6 0.8 1
0 κ
75.8 75.9 76 76.1 76.2 76.3 76.4 76.5
Time Fig. 18 Evaluation of ΩPL(κ
κ,δ
δ,0.7) as a function of κ for various
Fig. 16 Comparison between model and full scale experimental data values of δ.
from lower part of research well.
1.04

Ω BH(κ, δ,0.15) -δ=0.1

1.02 δ=0.2

δ=0.3
1
δ=0.4

0.98 δ=0.5

δ=0.6
0.96
δ=0.7

δ=0.8
0.94
δ=0.0 -δ=0.9
δ=1.0
Ro 0.92
0 0.2 0.4 0.6 0.8 1
ε κ
Fig. 19 Evaluation ΩBH(κ
κ,δ
δ,0.15) as a function of κ for various values
of δ.
Ri
1.04
Ω HB(κ, δ,0.05,0.8)
1.02 -δ=0.1
δ=0.2
1
δ=0.3

0.98 δ=0.4

Fig. 17 The eccentric annulus. δ=0.5

0.96

δ=0.6
0.94
δ=0.7
0.92
δ=0.8

0.9 -δ=0.9
δ=0.0
δ=1.0
0.88
0 0.2 0.4 0.6 0.8 1
κ
Fig. 20 Evaluation ΩHB(κ
κ,δ
δ,0.05,0.8) as a function of κ for various
values of δ.