You are on page 1of 13

SPE 114955

Cementing Horizontal Wells: Complete Zonal Isolation Without


Casing Rotation
M. Carrasco-Teja, I.A. Frigaard, SPE, and B. Seymour, University of British Columbia

Copyright 2008, Society of Petroleum Engineers

This paper was prepared for presentation at the CIPC/SPE Gas Technology Symposium 2008 Joint Conference held in Calgary, Alberta, Canada, 16–19 June 2008.

This paper was selected for presentation by an SPE program committee following review of information contained in an abstract submitted by the author(s). Contents of the paper have not been
reviewed by the Society of Petroleum Engineers and are subject to correction by the author(s). The material does not necessarily reflect any position of the Society of Petroleum Engineers, its
officers, or members. Electronic reproduction, distribution, or storage of any part of this paper without the written consent of the Society of Petroleum Engineers is prohibited. Permission to
reproduce in print is restricted to an abstract of not more than 300 words; illustrations may not be copied. The abstract must contain conspicuous acknowledgment of SPE copyright.

Abstract
There is a common perception that displacing mud in horizontal wellbores is made ineffective by large density differences
due to stratification of the fluid layers along the annulus. Hence the industry commonly employs methods such as casing
rotation to move fluids around the annulus. Whilst such methods may be effective, they are not always necessary. Using a
mathematical model of the process, we show that even with large density differences it is still possible to have a steadily
advancing displacement front that displaces mud all around the annulus without casing rotation. The conditions when this is
possible depend on the fluid rheologies, flow rate and hole geometry, but not the density difference. These conditions are
easily quantified. Although density difference does not enter into the criteria for an effective displacement, the length of the
interface along the wellbore does scale with the density difference. This means that, provided one is able to tolerate additional
fluid volumes, we are able to effectively displace the mud even using fluids with large density differences. This new
understanding of the displacement fluid mechanics opens up new opportunities for cementing difficult horizontal wells, (e.g.
slimhole), where casing rotation is not possible.
Introduction
Over the years there have been a number of studies of mud removal in primary cementing. Prior to the mid-1990’s much
of this related to the cementing of near vertical wells. The initial work focused on identifying specific phenomena with
clearly defined mechanics, e.g., Ref 1, and evolved over the years into systems of design rules based on an understanding of
the displacement mechanics, e.g., Refs 2-6. Thus, by the early 1990’s most service companies and operators had established
design practices for mud removal in primary cementing, but based on their understanding of what happens in vertical wells.
A typical set of design rules includes the following considerations. First, usually there is a preference for displacing in
turbulent regimes. If full turbulence cannot be achieved then a number of conditions should be satisfied: (i) each displacing
fluid should be heavier and more viscous than the fluid it displaces; (ii) the pressure gradient should be adequate to mobilize
fluid on the narrow side of the annulus; (iii) some effort is made to predict if the interface will be stable, i.e., displace as a
steady traveling wave. On top of these conditions there are frequently additional criteria related to chemical-based treatment
of wall layers.
As well as design methodologies, which have typically been based on hydraulic arguments, e.g., Refs 2 & 6, there have
been a number of studies of annular fluid displacements conducted using either CFD or experimental methods, e.g., Refs 7-
11. Although some of these have considered inclined wells, the mechanics of horizontal well displacements remained
unstudied in any detail. When the large growth of horizontal well drilling started in the 1990’s, the industry was not ready
with a tried and tested methodology for cementing, even though some of the potential problems had been identified and
solutions proposed, see e.g., Ref 12-14.
Certain aspects of fluid displacement design in a vertical well evidently carry over to horizontal well cementing, e.g., the
preference for turbulent fluid displacement. However, in extended reach wells the ability to displace in the turbulent regime is
often restricted by pore-frac pressure considerations and hence laminar displacements are more prevalent. In laminar flows
ensuring a frictional pressure hierarchy between fluid stages is sensible, as is ensuring a minimal pressure gradient on the
narrow side of the annulus. However, a density difference is not generally seen as being a good thing. Horizontal wells also
are very eccentric, which presents additional challenges. To compute the coupled effects of density, viscosity and eccentricity
on fluid displacement requires a modeling approach that is at least 2D. Such an approach has been developed in Refs 15-19
and applied principally to near vertical wells. The genesis of this approach can be traced back to Ref 20, although the details
of the analysis and results are quite new. The key results relevant to primary cementing pertain to the identification of steady
traveling wave displacements. These displacements represent the ideal scenario for a fluid displacement as the mud is
2 SPE 114955

removed all around the annulus at the steady pumping speed. Such displacements can be observed routinely in 2D
simulations (Refs 15 & 16), as well as predicted analytically (Refs 17-19). Indeed the approach advanced in Ref 19 gives
explicit conditions on the rheological and other physical parameters in order to ensure a steady displacement. An interesting
interpretation of these conditions is as an extended rule system, but one based on the 2D flow rather than the 1D hydraulics of
the 1990’s. To date the approach in Refs 15-19 has not been extensively analysed for horizontal displacements and that is the
aim here and in our companion paper, Ref 21 that explains much of the mathematical detail. So far our analysis is confined to
annuli that do not rotate or reciprocate during the displacement, although this analysis is also underway.
In the absence of a clear understanding of horizontal fluid displacements, industry practices have still evolved. A
significant proportion of long horizontal wells are simply not cemented. For those that are cemented it is fairly common to
use rotation and/or reciprocation of the casing. The idea of rotation is to move displacing fluids all around the annulus.
Casing reciprocation has the effect of increasing the shear, hence mobilizing gelled mud. Both also have application, post
placement, in the control of gas migration. Some of the studies involving these methods include Refs 22-26. There have also
been a number of interesting studies that look closely at the effects of casing motion on single fluid flows, Ref 27-29. In this
paper we make no comment about the effectiveness or otherwise of casing movement on fluid displacement. Instead we
examine the case of a stationary annulus and attempt to provide some of the missing physical understanding of this simpler
case.
Model Outline
The aim of our paper is to understand the mechanics of a horizontal displacement. Since typical fluid volumes pumped
translate into long annular lengths, we shall simplify matters by considering only two fluids, with fluid 1 displacing fluid 2.
Equally, since the annular geometry changes relatively slowly in the axial direction, relative to a circumferential length-scale,
we consider the displacement flow in an annulus of fixed geometry and inclination. We denote dimensional outer & inner
radii by R̂o and R̂i , respectively. The hat symbol will be used to denote a dimensional quantity. The angle of inclination of
the well from vertical is β , and the eccentricity of the annulus is e , see Figure 1. Physical and rheological parameters for
fluid k = 1,2 are: τˆ k ,Y , κˆ k , n k , ρ̂ k , the yield stress, consistency, power law index and density respectively; ĝ denotes the
gravitational acceleration. We assume that the annulus is initially full of fluid 2, which is displaced by fluid 1 at fixed flow
rate, Q̂ . These 14 dimensional parameters govern the annular displacement flow described, which must be considered as the
simplest realistic model of what occurs in a horizontal cementing displacement.
Via dimensional analysis we may reduce the parametric dependency considerably. In the case where the annulus is
narrow and the fluids are pumped in laminar regimes, inertial effects may be neglected and the flow is dominated by the
principal shear stresses. Model reduction, as described in Ref 15, leads to a Hele-Shaw description of the displacement flow.
This simplified model considers the 2D “gap-averaged” flow in axial and azimuthal directions; see Figure 1. A major benefit
of this model reduction is the fact that, in the simplest situation, the flow is governed by only the following 9 dimensionless
~
groups: τ k ,Y , κ k , n k , k = 1,2 , β , e and a buoyancy number b . For the benefit of the reader in later interpreting our
dimensionless results, we use the following scales:

Length - scale =
π ˆ
2
(
Ro + Rˆi ) Annular halfgap - scale =
1 ˆ
2
(
Ro − Rˆi )
Qˆ Qˆ
Velocity scale =
[
π Rˆo2 − Rˆi2 ] = wˆ * Stream - function scale =
4
(1)

⎧ nk
( ) ⎫
Stress scale = max ⎨κˆk γˆ* + τˆk ,Y ⎬ = τˆ* Shear rate scale =
wˆ *
= γˆ*
k =1, 2 ⎩ ⎭ [
1 ˆ
2
Ro − Rˆi ]
The relation between dimensionless, (no hat symbol), and dimensional variables is given as follows.

Yield stress, τ k ,Y = τˆ k ,Y / τˆ * k = 1,2

Consistency, κ k = κˆ k γˆ * ( ) nk
/ τˆ * k = 1,2 (2)

~
[ρˆ 2 − ρˆ 1 ]gˆ 1 [Rˆ o − Rˆ i ]
Buoyancy number, b = 2
τˆ *
The model that we use to compute our displacement flows is described in detail in Refs 15, 16 & 18, and therefore we do not
repeat the equations here. Essentially it involves solving an elliptic problem for the stream function at each timestep, and
advancing the concentration field using an advection equation.
SPE 114955 3

Typical simulation results in horizontal wells


We now present some results for a 2 fluid displacement flow at constant flow rate, fixing the inclination: β = π / 2 , i.e., a
fully horizontal well. The flow is now governed by 8 dimensionless parameters: 6 rheological parameters, annular
eccentricity and buoyancy number. Dimensionless spatial coordinates are (φ , ξ ) , where 0 ≤ φ ≤ 1 ; φ = 0 denotes the wide
side of the annulus, φ = 1 denotes the narrow side. The ξ -coordinate measures axial depth upwards from bottom-hole
towards surface. The annular gap half-width is denoted H (φ ) , where to leading order:
H (φ ) = [1 + e cos πφ ] , (3)
see Figure 1c.
We consider our displacements to be of 2 primary types: either heavy-light (HL) or light-heavy (LH). In the HL
~ ~
displacement the displacing fluid is heavier, b < 0 , and in the LH displacement the displacing fluid is lighter, b > 0 .
Intuitively we expect that the heavier fluid will slump towards the lower side of the annulus during displacement, and
therefore we expect the type of behaviours illustrated schematically in Figure 2.
Although having 8 dimensionless parameters makes an exhaustive parametric study of simulations near impossible, after
running a large number of simulations, we realize that a number of qualitatively similar flow behaviours are observed. Figure
3 shows the results of HL displacements for 2 Newtonian fluids in an annulus with fairly modest eccentricity, e = 0.1 . The
viscosities are chosen in order to give a steady displacement in the absence of any density difference, κ 1 = 1, κ 2 = 0.25 . In
each figure the direction of flow is from left to right and the top (wide) side of the annulus is towards the top of the page.
Figure 3a illustrates the interface position at successive timesteps throughout the displacement. We observe that after a short
transient the interface shape propagates left to right without changing profile. This is an ideal steady traveling wave
displacement.
Figure 3b shows the results of a computation using the same parameters as Figure 3a, but conducted in a moving frame of
reference, ξ − t , moving at the mean flow speed. In this frame of reference the interface becomes steady after a short time.
We show the results after t=40 time units. The thick black line denotes the interface position and the thinner lines give the
moving frame streamlines, spaced at 1% of the base flow rate. Thus, we see that there is a secondary flow about the steady
state, of magnitude 5-6% of the mean flow. The direction of flow behind the interface is from wide side to narrow side, and
that in front of the interface is from narrow side to wide side.
In the absence of a density difference we would expect that, due the annular eccentricity, the interface advances along the
~
wider side of the annulus. We observe however, that the density difference, here b = −2 , is sufficient to reverse this trend.
The interface slumps slightly towards the base of the annulus. For comparison, Figures 3c & d show the same results as
~ ~
Figure 3b, but for two larger buoyancy parameters, b = −10 & b = −50 , respectively. We observe that an increased
buoyancy number results in an increased slumping along the bottom of the annulus. As the interface elongates it takes a
progressively longer time to attain the steady profile, but the model still converges to a traveling wave solution. In order to
relate the results in Figure 3 back to a typical set of dimensional variables, we may note that these dimensionless parameters
would be found when displacing a 12.5cP fluid of density 1.26SG with a 50cP fluid of density 1.4SG, by pumping through an
annular gap between cylinders of radii 2in. & 3in., at flow rates between 35.6gpm and 1.8gpm.
As further examples of HL displacements, we show in Figure 4 the effects of adding a yield stress to the displaced fluid
~
and of increasing the eccentricity, e = 0.25 . Figure 4a shows the results at b = −2 when buoyancy effects are not large
enough to overcome the increased eccentricity and the displaced fluid yield stress. There is a tendency for the fluid to
~
elongate along the wide side of the annulus. Figures 4b & c show the effects of increasing the buoyancy to b = −10 . The
tendency of the interface to advance unsteadily along the top is suppressed and we observe a slumping behaviour toward the
bottom, and convergence towards a steady traveling wave profile. The shape of the final steady state is shown in Figure 4c in
the moving frame of reference. The magnitude of the secondary flow has increased up to around 15% of the base flow rate.
Figure 5 shows the same computations, but with an increased eccentricity, e = 0.4 . The unsteady nature of the interface at
~ ~
lower b = −2 is very evident, but again increasing the size of the buoyancy parameter to b = −10 results in a stable steady
traveling wave, (Figures 5b & c).
For HL displacements, the main features we have observed are: (i) elongation of a steady, stable interface profile as the
size of the buoyancy parameter is increased; (ii) competition between eccentricity and buoyancy, each favouring unsteady
fingering in different parts of the annulus; (ii) the existence of critical parameter ranges at which the behaviour changes from
steady to unsteady. The range of qualitative behaviours does not appear affected unduly by whether fluids are Newtonian or
non-Newtonian, although fluid rheology has an important role in determining the specific behaviour.
Lastly, we should note that the unsteady interfaces in Figures 4a and 5a are mechanically unstable, i.e., the heavier fluid is
fingering ahead on top of a lighter fluid. This configuration is likely to become unstable over time due to small-scale
gravitational fingering at the interface. Such effects are not modeled in our simplified approach but are considered in Refs 30-
31.
4 SPE 114955

Two examples of LH displacements are presented in Figure 6. For the LH displacement both buoyancy and eccentricity
act to promote unsteady fingering along the top of the annulus. Only a favourable rheology difference can hope to stabilize
displacements. However, as before, we are able to observe that the interface elongates initially and approaches a steady
traveling wave profile. For other simulations with less favourable rheology differences we were unable to achieve steady
state solutions.
Semi-analytical model for elongating interfaces
As we have seen in many of the examples presented, it is quite possible to achieve a steady traveling wave displacement front
for both HL and LH displacements. For many of our examples the axial length of the steady state is of O(1); in dimensional
terms this means of the order of the circumference. For cementing these cases are ideal as not only do the annuli displace
fully and at constant rate, but also the small slump length means that fluid volume calculations (nearly always based on plug-
like displacements) will be reasonable.
It is perhaps true to say that our chief interest is in identifying when the interface elongates over lengths that are large.
~
This occurs either when the interface is simply unsteady, i.e. no steady traveling wave is found, or when steady and if b
~
becomes too large. To analyse the case of a long interface, we rescale axial lengths with b >> 1 and reduce the model further
using classical arguments from lubrication theory. This procedure is detailed in Refs 19 and 21. The volumetric interface
position is denoted Φi (ξ , t ) and is defined by:
φi (ξ ,t )
Φi (ξ , t ) = ∫ H (φ )dφ , (4)
0

where φi (ξ , t ) is the interface position. After following the procedure in Ref 21, for a HL displacement we find the following
partial differential equation governing the evolution of the interface:

∂ ∂ ⎛ ∂φi ⎞
~ Φ i + q⎜ Φ i , α − sin πφi ⎟ = 0 , (5)
∂t ∂z ⎝ ∂z ⎠

where t and z are the rescaled slow timescale and long length-scale, respectively, and α = b cos β ~ O(1) when we consider
~ ~

small inclinations from horizontal. The flux function q represents the stream function at the interface, which may be
computed rapidly under the lubrication approximation using numerical integration, or for some fluid combinations can be
evaluated analytically.
Steady state displacements
~
Writing, x = z − t , shifting to a moving frame of reference, we see that a steady traveling wave solution must satisfy:

⎛ ∂φ ⎞
q⎜ Φi , α − i sin πφi ⎟ = Φi , (6)
⎝ ∂x ⎠
which is a differential equation for the shape of the steady state. In Ref 21 it is shown that Eqn (6) only can be solved, to give
a steady state shape if the following condition holds
q(Φi , αi ) ≥ Φi ∀Φi ∈ [0,1] (7)
This effectively is our criterion for a steady traveling wave displacement. A very similar treatment of the LH displacement is
given in Ref 21.
Physically, what Eqn (6) says is that the slope of the interface should be such as to make the stream function at the
interface equal to the volumetric interfacial position, which is straightforward mass conservation. It can be shown that
increasing the slope of the interface reduces the stream function at the interface. Therefore, the condition given by Eqn (7)
simply says that if there is zero interface slope the flux function q would need to exceed the volumetric interface position.
Examples results for HL displacements
To illustrate the above results, we show first in Figure 7 results for 2 Newtonian fluids. For different wellbore inclinations,
close to horizontal, we plot the critical viscosity ratio κ 1 / κ 2 that needs to be exceeded in order for there to be a steady
traveling wave displacement. This critical ratio is plotted against the eccentricity of the annulus. In Figure 7a we observe the
competition between density difference and eccentricity, in that the minimal ratio is often found at intermediate eccentricity,
i.e., in concentric annuli the viscosity ratio must compensate for the entire effect of buoyancy, and at large eccentricities,
where eccentricity dominates buoyancy again, only a large viscosity ratio results in steady displacement. At intermediate
eccentricities the two effects counter each other and a reduced viscosity ratio is needed. As the wellbore inclination
approaches horizontal and then turns slightly uphill a smaller viscosity ratio is needed: the inclination compensates partly for
the buoyancy effect.
SPE 114955 5

Figures 7b & c show examples of the steady state shapes that can be computed when the viscosity ratio exceeds the values in
Figure 7a. Figure 7b fixes the viscosity ratio at κ 1 / κ 2 = 1 , considers a perfectly horizontal well, α = 0 , and plots the steady
state shapes for various eccentricities. It is interesting to note that increasing the eccentricity (over the range shown) has the
counter-intuitive effect of decreasing the length of the interface along the axis. Figure 7c fixes the eccentricity and plots the
effects of varying the viscosity ratio. Here the effects are more intuitive: a larger ratio of κ 1 / κ 2 produces a shorter steady
state profile.
Power law fluid displacements are explored in Figures 8 & 9. In general the effects of the shear-thinning power law index
combines with the fluid consistency to give fluid pairs that are either more or less viscous. The qualitative picture is then
similar to that for Newtonian fluids insofar as viscosity ratio is concerned, i.e., increasing the power law index of fluid 1
results in a more viscous displacing fluid, all other parameters remaining equal, and thus a reduced ratio κ 1 / κ 2 necessary for
a steady state. This effect is exemplified by comparing for example Figure 7c with Figure 8d and Figure 9d. Qualitatively the
other effects presented in Figures 8 & 9 (inclination, eccentricity) are very similar to those observed for Newtonian fluids.
For yield stress fluids we are also able to compute critical conditions for a steady state displacement, but now the ratios of
consistencies does not appear as an easily isolated parameter. Instead we may plot critical values of any of the parameters. As
an example we have chosen to plot critical values of yield stresses in Figure 10. In general, we can observe that weaker ratios
κ 1 / κ 2 require a larger yield stress in fluid 1 in order to displace in the steady state.
LH displacements
The analysis may be repeated for LH displacements with only minor modifications (see Ref 21). Recall that in LH
displacements both eccentricity and buoyancy act together in pushing the lighter fluid along the wider top of the annulus.
Thus, by comparison to the HL displacements, larger viscosity ratios are generally needed in order to find steady traveling
wave displacements. For brevity, we present just one example of results, in Figure 11, which may be compared with Figure 7.
We observe that the critical ratio κ 1 / κ 2 increases monotonically with eccentricity, tending asymptotically to infinity as the
eccentricity approaches 1, i.e., zero standoff. Similarly, as there is no competition between buoyancy and eccentricity,
increases in eccentricity lead to longer axial lengths of steady states.
Conclusions
Probably the most significant result from the perspective of cementing horizontal wells is the demonstration that it is possible
to have steady, traveling wave, displacement solutions even in situations where there are significant density differences
between fluids. For vertical wells this is of course well known, and effective mud removal strategies in vertical wells often
include a positive density hierarchy between successive fluids pumped.
At the outset it appears that a density difference must be detrimental to a horizontal displacement. Whilst we have seen
that certain combinations of fluid properties do result in unsteady horizontally stratifying interfaces, we have also seen that
for large parameter regimes steady state displacements can be found. This is of great importance for long extended reach
wells where it may not always be feasible to rotate or reciprocate the casing during displacement due to risks of either
shearing the pipe or of compromising well security with swab/surge frictional pressures.
The implication is that many such wells can be effectively cemented without rotation provided that: (i) we are prepared to
accept that there may be a long stratified region in which the two fluids advance steadily along the annulus; (ii) we design
fluid volumes accordingly to effectively discard the length of the steady state. Evidently, this only works where we are able
to circulate to surface any excess fluids or may leave the fluids downhole in an overlapping cased section. In any case, the
modeling techniques outlined here enable us to quickly calculate conditions for steady displacement and the length of the
steady displacement profile.
Some of the physical features of our model solutions are as follows:
• For strictly horizontal wells, the size of the density difference does not govern conditions for there to be a steady
~
traveling wave, displacement front. However, the shape of the front is scaled linearly with the density difference via b .
Essentially it appears that if the fluid rheology and geometry are such as to allow a steady state, in the absence of a
density difference, then the interface simply spreads to such an extent that the density difference becomes insignificant in
determining stability, only shape!
• For HL displacements we can find conditions where increasing the density of the displacing fluid 1 can in fact shorten
the length of the steady state front.
• For HL displacements we can find conditions where increasing the eccentricity of the annulus can shorten the length of
the steady state, or may even changean unstable displacement to a stable displacement!
The last two effects are quite counter-intuitive physically. Since casings in horizontal wells are typically eccentric, it is
suggested that a small positive density difference (HL) is used in horizontal well cementing. When LH displacements are
considered, eccentricity and the density difference act together, making steady displacements harder to achieve. In general
this regime should be avoided. Precise conditions for the existence of a steady state displacement can be computed very
easily. Most of the results presented in this paper took less than a minute to compute on a laptop. Thus, incorporating this
new physical understanding into sets of design rules is certainly feasible.
Our ongoing research looks at the effects of casing rotation and reciprocation on this type of displacement model. The
results in this paper do not discount the effectiveness of such techniques, but simply question their necessity.
6 SPE 114955

Acknowledgements
This research has been carried out at the University of British Columbia, supported financially by Schlumberger and by
NSERC Canada, through Collaborative Research and Development project 245434, and also from NSERC Canada through
strategic project grant 306682. This latter is supported by Schlumberger and by Trican Ltd. This financial support is
gratefully acknowledged and we appreciate the interaction with our industrial partners.

References
1. R.H. McLean, C.W. Manry and W.W. Whitaker, Displacement Mechanics in Primary Cementing. Society of Petroleum Engineers
paper number SPE 1488, (1966).
2. A. Jamot, Deplacement de la boue par le latier de ciment dans l’espace annulaire tubage-paroi d’un puits. Revue Assoc. Franc.
Techn. Petr., 224, pp. 27-37, (1974).
3. C.F. Lockyear and A.P. Hibbert, Integrated Primary Cementing Study Defines Key Factors for Field Success. Journal of Petroleum
Technology, December 1989, pp. 1320-1325.
4. C.F. Lockyear, D.F. Ryan and M.M. Gunningham, Cement Channeling: How to Predict and Prevent. Society of Petroleum
Engineers paper number SPE 19865, (1989).
5. D. Guillot, H. Hendriks, F. Callet and B. Vidick, Mud Removal. Chapter 5 in Well Cementing. editor E.B. Nelson, Schlumberger
Educational Services, Houston, (1990).
6. M. Couturier, D. Guillot, H. Hendriks and F. Callet, Design Rules and Associated Spacer Properties for Optimal Mud Removal in
Eccentric Annuli. Society of Petroleum Engineers, paper number SPE 21594, (1990).
7. J. Jakobsen, N. Sterri, A. Saasen, B. Aas, I. Kjoenes and A. Vigen, Displacements in Eccentric Annuli During
Primary Cementing in Deviated Wells. Society of Petroleum Engineers, paper number SPE 21686, (1991).
8. A. Tehrani, J. Ferguson and S.H. Bittleston, Laminar Displacement in Annuli: A Combined Experimental and Theoretical Study.
Society of Petroleum Engineers paper number SPE 24569, (1992).
9. A. Tehrani, S.H. Bittleston and P.J.G. Long, Flow instabilities during annular displacement of one non-Newtonian fluid by
another. Experiments in Fluids 14, pp. 246-256, (1993).
10. E. Biezen, N. van der Werff and K. Ravi, Experimental and numerical study of drilling fluid removal from a horizontal wellbore.
Society of Petroleum Engineers, paper number SPE62887, (2000).
11. N. Barton, G.L. Archer & D.A. Seymour, Computational fluid dynamics improves liner cementing operation. Oil & Gas Journal,
September 26, pp. 92-98, (1994)
12. S.R. Keller, R.J. Crook, R.C. Haut and D.S. Kulakofsky, Deviated-Wellbore Cementing: Part 1 - Problems. Journal of Petroleum
Technology, August 1987, pp. 955-960.
13. R.J. Crook, S.R. Keller, and M.A. Wilson, Deviated Wellbore Cementing: Part 2 - Solutions. Journal of Petroleum Technology,
August 1987, pp. 961-966.
14. F.L. Sabins, Problems in Cementing Horizontal Wells. Journal of Petroleum Technology, April 1990, pp. 398-400.
15. S.H. Bittleston, J. Ferguson & I.A. Frigaard, Mud removal and cement placement during primary cementing of an oil well. J.
Engng. Math., 43, pp. 229-253, (2002).
16. S. Pelipenko & I.A. Frigaard, Two-dimensional computational simulation of eccentric annular cementing displacements. IMA J
Appl. Math, 64(6). pp. 557-583, (2004).
17. S. Pelipenko & I.A. Frigaard, Effective and Ineffective Strategies for Mud Removal and Cement Slurry Design, Society of
Petroleum Engineers paper number SPE 80999 (2003)
18. S. Pelipenko & I.A. Frigaard, On steady state displacements in primary cementing of an oil well. J. Engng. Math., 48(1), pp. 1-26,
(2004).
19. S. Pelipenko & I.A. Frigaard, Visco-plastic fluid displacements in near-vertical narrow eccentric annuli: prediction of travelling
wave solutions and interfacial instability. J Fluid Mech., 520 pp. 343-377, (2004)
20. M. Martin, M. Latil and P. Vetter, Mud Displacement by Slurry during Primary Cementing Jobs - Predicting Optimum Conditions.
Society of Petroleum Engineers paper number SPE 7590, (1978).
21. M. Carrasco-Teja, I.A. Frigaard, B. Seymour & S. Storey, Viscoplastic fluid displacements in horizontal narrow eccentric annuli:
stratification and traveling wave solutions, J. Fluid Mech, accepted for publication (2008), to appear.
22. D.L. Sutton & K.M. Ravi, Low rate pipe movement during cement gellation to control gas migration and improve cement bond
Society of Petroleum Engineers paper number SPE 22776, (1991).
23. D. Gust & R.R. MacDonald, Rotation of a long liner in a shallow long reach well. J. Petr. Tech. April 1989, pp 401-404.
24. H. Kinzel & J. Martens, The application of new centralizer types to improve zonal isolation in horizontal wells. Society of
Petroleum Engineers paper number SPE 50438, (1998).
25. S.A. McPherson, Cementation of horizontal wellbores. Society of Petroleum Engineers paper number SPE 62893, (2000).
26. M. Savery, R. Darbe & W. Chin, Modelling fluid interfaces during cementing using a 3D mud displacement simulator. Offshore
Technology Conference paper number OTC 18513, (2007).
27. M.P. Escudier & I.W. Gouldson, Concentric annular flow with centerbody rotation of a Newtonian and a shear-thinning liquid.
Int. J. Heat Fluid Flow, 16(3), pp. 156-162, (1995).
28. M. P. Escudier, I. W. Gouldson, P. J. Oliveira & F. T. Pinho, Effects of inner cylinder rotation on laminar flow of a Newtonian
fluid through an eccentric annulus. Int. J. Heat Fluid Flow, 21(1), pp. 92-103, (2000).
29. M. P. Escudier, P. J. Oliveira & F. T. Pinho, Fully developed laminar flow of purely viscous non-Newtonian liquids through annuli,
including the effects of eccentricity and inner-cylinder rotation. Int. J. Heat Fluid Flow, 23(1), pp. 52-73, (2002).
30. M. Moyers-Gonzalez & I.A. Frigaard, Kinematic instabilities in two-layer eccentric annular flows, part 1: Newtonian
fluids, J. Engng. Math., DOI: 10.1007/s10665-007-9178-y (2007).
SPE 114955 7

31. M. Moyers-Gonzalez & I.A. Frigaard, Kinematic instabilities in two-layer eccentric annular flows, part 2: shear-
thinning and yield stress effects, submitted to J. Engng Math., 2007.

Figure 1. Schematic of the model geometry: (a) the near-horizontal annulus; (b) annular cross-section showing coordinates and
eccentricity; (c) unwrapped annular gap – the computational domain.

Figure 2. Schematic of the two different types of mechanically stable displacement: (a) heavy fluid displacing lighter fluid (HL); (b)
light fluid displacing heavier fluid (LH).
8 SPE 114955

Figure 3. Example HL displacements in a horizontal annulus for two Newtonian fluids with dimensionless parameters:
κ 1 = 1, κ 2 = 0.25, e = 0.1 . ~
(a) b = −2 : figure shows successive interface profiles advancing along the annulus, separated by a
fixed time interval. Figures (b)-(d) show the streamlines and interface (heavy line) in a frame of reference moving at the mean
displacement speed. Computations have reached steady state and are shown at t = 40 . Each streamline contour spacing
~ ~ ~
represents 1% of the total flow rate. Parameters: (b) b = −2 ; (c) b = −10 ; (d) b = −50 .
SPE 114955 9

Figure 4. HL displacements in a horizontal annulus for non-Newtonian fluids with parameters: τ Y ,1 = 0, τ Y , 2 = 1, n k = 1,


~ ~
κ 1 = 1, κ 2 = 0.25, e = 0.25 . Figures (a)&(b) show interface profiles separated by fixed time intervals: (a) b = −2 , (b) b = −10 .
~
b = −10 , shows streamlines and interface (heavy line) in a frame of reference moving at the mean displacement
Figure (c), also for
speed. Computations have reached steady state and are shown at t = 40 . Each streamline contour spacing represents 3% of the
total flow rate.

Figure 5. As figure 4 but with increased eccentricity, e = 0.4 .


10 SPE 114955

Figure 6. Example LH displacements in a horizontal annulus. (a) & (b) 2 Newtonian fluids with dimensionless parameters:
~
κ 1 = 1, κ 2 = 0.25, e = 0.2, b = −2 . Figure (a) successive interface profiles advancing along the annulus, separated by a fixed time
interval. Figure (b) the interface (heavy line) and moving frame streamlines in a frame of reference moving at the mean displacement
speed, t = 40 . Each streamline contour spacing represents 3% of the total flow rate. Figures (c) & (d) as for (a) & (b) but for non-
Newtonian fluids: τ1,Y = 1, κ 2 = 0.05 .

Figure 7. Newtonian fluids in HL displacement: (a) Conditions on the viscosity ratio κ1 / κ 2 at different eccentricities in order to
have a steady traveling wave for different near-horizontal inclinations:α / κ 2 = -0.6 ∇, -0.3 ◊, 0 , 0.3 Ο, 0.6 Δ. Steady conditions are
found above the curves. (b) Shapes of steady interface for κ 1 / κ 2 = 1 and various eccentricities: e = 0.2 ∇, 0.15 ◊, 0.1 , 0.05 Ο.
Perfectly horizontal well, α = 0 . (c) Shapes of steady interface for e = 0.1 and various viscosity ratios: κ 1 / κ 2 = 1.6 ∇, 1.4 ◊, 1.2
, 1 Ο. Perfectly horizontal well, α = 0 .
SPE 114955 11

Figure 8. Power-law fluids in HL displacement: (a) n1 = 0.5 , n 2 = 1 . Conditions on the consistency ratio κ 1 / κ 2 at different
eccentricities in order to have a steady traveling wave for different near-horizontal inclinations: α / κ 2 = -0.6 ∇, -0.3 ◊, 0 , 0.3 Ο, 0.6
Δ. Steady conditions are found above the curves. (b) n1 = 0.5 , α = 0 . Conditions on the consistency ratio κ 1 / κ 2 at different
power law indices in order to have a steady traveling wave for different power law indices: n 2 = 1 ∇, 0.8 ◊, 0.6 , 0.4 Ο, 0.2 Δ. (c)
Shapes of steady interfaces for κ 1 / κ 2 = 1.5 , n1 = 0.5 , n 2 = 1 , and various eccentricities: e = 0.2 ∇, 0.15 ◊, 0.1 , 0.05 Ο.
Perfectly horizontal well, α = 0 . (d) Shapes of steady interfaces for e = 0.1 , n1 = 0.5 , n 2 = 1 , and various viscosity ratios:
κ 1 / κ 2 = 1.6 ∇, 1.4 ◊, 1.2 , 1 Ο. Perfectly horizontal well, α = 0 .
12 SPE 114955

Figure 9. Power-law fluids in HL displacement: (a) n1 = 1 , n 2 = 0.5 . Conditions on the consistency ratio κ1 / κ 2 at different
eccentricities in order to have a steady traveling wave for different near-horizontal inclinations: α / κ 2 = -0.6 ∇, -0.3 ◊, 0 , 0.3 Ο, 0.6
Δ. Steady conditions are found above the curves. (b) n 2 = 0.5 , α = 0 . Conditions on the consistency ratio κ 1 / κ 2 at different
power law indices in order to have a steady traveling wave for different power law indices: n1 = 1 ∇, 0.8 ◊, 0.6 , 0.4 Ο, 0.2 Δ. (c)
Shapes of steady interfaces for κ 1 / κ 2 = 1 , n1 = 1 , n 2 = 0.5 , and various eccentricities: e = 0.2 ∇, 0.15 ◊, 0.1 , 0.05 Ο. Perfectly
horizontal well, α = 0 . (d) Shapes of steady interfaces for e = 0.1 , n1 = 1 , n 2 = 0.5 , and various viscosity ratios: κ 1 / κ 2 = 1.6 ∇,
1.4 ◊, 1.2 , 1 Ο. Perfectly horizontal well, α = 0 .

Figure 10. Yield stress fluids in HL displacement: (a) n1 = 0.5 , n 2 = 1 ; (b) n1 = 1 , n 2 = 0.5 . Conditions on the yield stresses at
different consistency ratios κ 1 / κ 2 in order to have a steady traveling wave: κ 1 / κ 2 = 2 ∇, 1 .5 ◊, 1 , 0.5 Ο. Perfectly horizontal
well, α = 0 , with eccentricity e = 0.2. Steady conditions are found above the curves.
SPE 114955 13

Figure 11. Newtonian fluids in LH displacement: (a) Conditions on the viscosity ratio κ1 / κ 2 at different eccentricities in order to
have a steady traveling wave for different near-horizontal inclinations: α / κ 2 = -0.6 ∇, -0.3 ◊, 0 , 0.3 Ο, 0.6 Δ. Steady conditions are
found above the curves. (b) Shapes of steady interface for κ 1 / κ 2 = 1.8 and various eccentricities: e = 0.2 ∇, 0.15 ◊, 0.1 , 0.05 Ο.
Perfectly horizontal well, α =0.

You might also like