Proceedings of the ASME 2014 33rd International Conference on Ocean, Offshore and Arctic Engineering

June 8-13, 2014, San Francisco, California, USA


Muhammad Zulqarnain
Graduate Student
Louisiana State University
Baton Rouge, LA, USA

Spacers are the intermediate fluids in the primary cementing
process, preceded by the mud and followed by the cement. They
help in mud removal, keep mud and cement separated and
prepare the annulus walls for a good cement bond. The focus of
this study is to analyze the spacer performance in keeping mud
and cement separated in the vertical and horizontal well
sections with eccentricity. Several combinations of eccentricity
variation and displacement rates are simulated using
Computational Fluid Dynamics (CFD) tool to analyze the
temporal and spatial fluid volume fraction distributions in the
annulus and validated against multi-fluid displacement
experiments. A 50 ft vertical section with annular gap of 1.5″
and casing standoff from 5% to 100% and a horizontal section
with variable eccentricity are studied. For initial conditions the
annulus is modeled as filled with mud and subsequently swept
by one annular volume of spacer and cement each respectively.
Mud and cement are treated as Herschel Bulkley fluids and
spacer as a Newtonian fluid.
Qualitative results from simulations are shown in the form
of volume fraction contour plots at different sections of the
annulus and quantitative results in the form of temporal and
spatial volume fraction of each fluid at specific planes in the
entire annulus. For a vertical annular section with constant
eccentricity, some unswept mud is observed on the narrow side
and eventually becomes trapped after the eccentricity is
increased above a threshold value. It is also observed that
increasing displacement rate helps in displacing some of this
trapped mud. After a threshold eccentricity value, the trapped
mud cannot be displaced and the spacer and cement follow the
path of least resistance and flow occurs mainly on the wider

Mayank Tyagi
Associate Professor
Louisiana State University
Baton Rouge, LA, USA

side of the annulus. For a horizontal section with variable
eccentricity (maximum at the mid distance between
centralizers), the flow is observed to move towards wider side
around the center and then returning to the other side (narrow
part) after passing through middle section. Some trapped mud is
also observed in the vicinity of the middle section due to local
maximum eccentricity.
centralization/standoff, flow rate, casing movement, formation
permeability, spacer design and contact time are some of the
most important parameters influencing the primary cementing
[9-11]. This study focuses on the performance of spacer in the
presence of eccentricity in vertical and horizontal well sections
and effect of flow rate.
Problems encountered during horizontal well cementing are
similar to those of vertical well cement job, but are exacerbated
by factors such as wellbore orientation, gravitational forces and
casing eccentricity. Solids settling are not limited to the drilling
mud, but also occur in the cement slurries if proper precautions
are not observed. Deposition of solids in the wellbore is one of
the most severe problems in horizontal wells [2]. Keller et al.
[2] conducted a series of lab experiments in which they
simulated the solid particle settling in a concentric annulus at
various wellbore orientations. They showed that even with
concentric annulus there was a huge difference in the
displacement efficiency, between the top and bottom section of
the annulus at different orientations. The problem was more
severe for certain combination of fluid properties and flow


Copyright © 2014 by ASME


Figure 1: Typical horizontal cemented wellbore cross
section [3]
There is an increased possibility of having narrow annular
channel on lower side and wider channel on other side of the
casing in horizontal well. Insufficient clearance on the narrow
side of the wellbore can result in portions of the well that are
not cemented properly or even not cemented at all. It is due to
the fact that due to excessive viscous forces have to be
overcome in order to mobilize the mud or settled cuttings on
narrow side while there is lower resistance on the wider side of
the annulus. Thus, fluid has the tendency to follow the path of
less resistance and move in the wider side of the annular path.
In some reported field studies [4], tops of the cement on the two
side of the annulus were separated by hundreds of feet.
If casing in the well is not centered, the velocity
distribution across the annulus is distorted, with the flow
favoring the wider side. This may lead to undesirable situations
where the flow regime can be laminar on the narrow side of the
annulus and turbulent on the wider side, because of the
azimuthal variation of local Reynolds number in the annulus
[3]. Pipe centralization can significantly aid in the efficient mud

Inner diameter
Outer diameter
Density of spacer
Density of mud
Density of cement
Reynolds Number
Time for one annular sweep
Viscosity of spacer
Viscosity of mud
Viscosity of cement
Plastic Velocity
Yield Stress
Computational Fluid Dynamics
Volume of Fluid
Spacer Volume Fraction

Mud Volume Fraction
Cement Volume Fraction

A commercially available CFD package FluentTM
based on unstructured finite volume formulation of nonNewtonian Navier Stokes equations is used [1]. Volume of
Fluid method (VOF) has been implemented in this study to
track the fluid volume fractions in each cell of
computational domain and reconstruct the interface [6],
between different fluids. VOF method tracks the volume of
each fluid in all cells containing portions of the
interface. The geometric reconstruction scheme is used to
reconstruct the interface in the cells where more than one
fluid is present. This reconstruction scheme represents the
interface between fluids using a piecewise-linear approach. It
assumes that the interface between two fluids has a linear slope
within each cell, and uses this linear shape for calculation of the
advection of fluid through the cell faces. The position of linear
interface relative to the center of each partially-filled cell is
based on information about the volume fraction and its
derivatives in the cell. The volume fraction in each cell is
calculated using the balance of fluxes calculated during the
previous step. For further details on governing equations and
numerical method, reader is referred to [1] and [7].
Experimental results of Tehrani et al. [5] were used for the
validation of numerical study. The experimental setup consisted
of two coaxial cylindrical tubes. The ID and OD of the annulus
were 1.57 inches and 1.97 inches respectively, creating a
concentric gap of 0.20 inches. Total axial length of the tubes
was 9.8 ft. Comparison for both concentric and ecentric cases
were made. The simulated values for the displacement
efficiency and experimental measured value eccentric (e = 0.5)
case with similar and different density fluids are shown Figure2.
A satisfactory agreement is obtained between the experimental
data and the calculated values of displacement efficiency.

Figure 2: Comparison of the simulation results (lines) for
displacement efficiency with the experimental data (symbols)
of Tehrani et al. [5]


Copyright © 2014 by ASME


Table 2: Different density and rheological properties for spacer
fluid in the horizontal setup with variable eccentricity (12
simulation cases) at fixed Reynolds number (displacement rate).

Horizontal Setup with Variable Eccentricity
Casing was assumed to be supported by two centralizers at both
ends and had uniformly distributed load on it. It was assumed
that the horizontal well was completed with 8.5" of open hole
having 5.5" OD P-110 casing. The spacing between the
centralizers was varied to achieve maximum eccentricities of
0.15, 0.3 and 0.6 at centeral ponit of casing. Casing section
length, maximum deflection at center and eccentricity are given
in Table 1.





Table 1: Casing Deflection and Corresponding Eccentricity

Length (ft)














Vertical Setup with Constant Eccentricity

Mud had the density and rheological properties of ρ = 10.2
lb/gal, μp = 34 cP, τy = 13 lb/100ft2 and cement had the density
and rheological properties of ρ = 15.8 lb/gal, μp = 34 cP, τy = 4
lb/100ft2 respectively. Herschel Bulkley rheological model was
used for mud and cement, while the spacer was treated as
Newtonian fluid. Reynolds number for Herschel Bulkley fluid
was calculated by the procedure given by Antonino Merlo et al.
[14]. A pump rate of 10 bbl/min corresponds to Reynolds
number of 3304 was used. Corresponding critical Reynolds
number for turbulence was 1309, resulting in turbulent flow
regime. A two-equation (k-ϵ) turbulent model was used with
standard wall function with default setting for this model. It was
also ensured that the grid clustering along the walls met the wall
distance y+ requirements for the selected turbulence model.
Three variable eccentric cases (Table 2) were considered with
spacer density and viscosity either equal to mud or cement for a
fixed displacement rate corresponding to Reynolds number
3304. All fluids enter the computational domain from the left
and exit from the right.



Air inside and outside casing
deflection (in)


A 50 ft. vertical section with annular gap of 1.5″ and casing
standoff from 5% to 100% were also simulated in this study.
Only eccentricity and Reynolds number were varied while
keeping all of the other parameters same (Table 3). The hole
and casing geometry and fluid properties were same as were for
the concentric vertical cases [8].
Table 3: Vertical section setup and flow conditions at different
eccentricity and Reynolds number (15 simulation cases)

Case e


Figure 2: Computational domain to show the casing (green
surface) inside the wellbore (grey surface) having emax = 0.60 at
the center of the annulus.


ρs = ρm μs = μw Sub
(lbm/gal) (cp) case

4 0.05 13.11


5 0.25 13.11





7 0.75 13.11


8 0.95 13.11





Copyright © 2014 by ASME

From the vertical simulations of concentric annulus, it was
found that the spacer performs best when it has the same density
as of mud and viscosity equal to that of water [7]. Therefore,
the density and viscosity of spacer were fixed to be that of mud
and water respectively. With increasing eccentricity, the
yielding fluid behavior becomes very important, and therefore
the fluid theologies of mud and cement were modeled with
Herschel Bulkley rheological model.

sweep. Earlier breakthrough times of spacer imply that it would
have penetrated (or mixed) into the mud volume ahead of it due
to interfacial instabilities [12, 13]. Similar arguments are also
true for cement volume sweep except that ideally it should
breakthrough at t = 1.16 τ and should attain the maximum
volume fraction of 1.0 at about t =1.5 τ. The plateau region
within one annular volume being pumped would imply that no
further improvement in placed volume fraction could be
obtained by pumping more fluid volume and would hold true
for both spacer and cement.

Horizontal configuration with variable eccentricity
Interpreting of the volume fraction plots
Instantaneous average volume fractions of three fluids
involved (mud-spacer-cement) are reported in the last one third
annular sections for the horizontal configurations and in the last
10 ft section of vertical configurations. These “exit” sections
are referred as observation section in the following discussion.
Presumably all fluids could have significant volume fractions in
the annulus for this particular section. In an ideal situation, if
the spacer were to act as a fluid plug by not only physically
separating the mud and cement but also helping in easy removal
of mud from the annulus, the highest displacement efficiencies
could be achieved with φc =1.
In a representative temporal sequence plot (Fig. 3), volume
fractions of three fluids are plotted against the fluid annular
sweeps: first half annular volume sweep of spacer followed by
one annular volume sweep of the cement. Three important
instances on these plots to describe the effectiveness of the
spacer in separating mud and cement along with the efficiency
of the displacement process are: i) spacer and cement
breakthrough times, ii) the slopes of spacer and cement fraction
curves, and iii) the plateau region attained by the spacer curve.

Results of only cases 1 and 3 are discussed here and for
detailed discussion with other cases, the reader is referred to
Case1 emax=0.15: Two representative subcases a & d
were selected (Table 2) as it was found that altering the spacer
viscosity does not influence the flow behavior while the density
changes had significant effects on the displacement process and
the value of the final cement volume fraction in the annulus.
Temporal volume fraction variations in the observation section
are shown in Figure 4.


Spacer Plateau


Figure 4: Temporal fluid volume fractions variation in the last
one third sections of the annulus for different spacer densities
(Table 2: 1a – top, and 1d - bottom).
Figure 3: Temporal fluid volume fraction variation in
observation section of the annulus.
If the spacer were to act like a plug between cement and
mud, then it should breakthrough at the observation section no
earlier than t = 0.66 τ and its volume fraction should approach
1.0 at about t = τ ideally, where τ is the time for one annular

For subcase (a) when both the spacer and mud have the
same densities and viscosities, the spacer breaks through in the
observation section at about the ideal breakthrough time of 0.66
τ, showing that the interface between spacer and mud is stable.
The annulus in this simulation is only 34 ft long, for very long
annular sections, the result could be different. The spacer curve


Copyright © 2014 by ASME

does not have a plateau region, which emphasizes the need for
more spacer volume should have been pumped and only half
annular volume was not sufficient to achieve the plateau regime.
The cement breakthrough time is slightly delayed primarily due
to the presence of more volume of other fluids (mud and spacer)
ahead of it.
Volume fraction contours in the last one half part of the
annulus are shown in Figure 5. Most of the unswept mud and
spacer is present at the top side of the annulus near the exit.
Gravity influences fluid segregation as well as casing sagging
leading variable eccentricity of the annulus. Due to gravity, the
mud and the spacer being lighter than cement will override in
the top part of the annulus while the cement will be displacing
these fluids. At the center of the casing section (max. deflection
point), the top side has an expansion and the lower side has
reduction in the flow area, but overall the flow cross-sectional
area remains constant. After the middle region, the top side has
reduction in flow area and the lower side has expansion in the
flow area. Due to this geometry of the annulus with variable
eccentricity, a decrease in pressure occurs at the top side in the
middle and increase in pressure on the lower side at middle.
After the middle section, the pressure on the top side increases
and on lower side decreases. The consequences of such a
pressure distribution on the flow field are to move fluids
upward and towards the exit to move further downstream.
Therefore, unswept mud and spacer are present towards the top
side near the exit.

Case 3 emax = 0.6: With increase of the casing deflection due
to sagging at the center, the flow field is dramatically altered.
The upward push for the moving fluid is reflected in the upward
curvature of streamline plots both for case (3a) and (3d) shown
in Figure 6.

Figure 6: Streamline plot colored by mixture volume fraction in
the entire annular section for cases (3a) and (3d) respectively.
Delay in the cement breakthrough time occurs due to the
differential density effects between the spacer and the cement.
The spacer overrides on the cement and mostly fills the upper
portion of the annulus and due to restriction in the flow path in
the lower part of the annulus, the cement lags behind the
expected breakthrough time. Same phenomenon contributes in
the early breakthrough of spacer in conjunction with the annular




Figure 5: Planer and cross-sectional view of fluid volume
fraction contours in the last one half section of the annulus for
cases (1a) and (1d) respectively.
When the spacer density is increased to reach cement
density, the spacer breakthrough time is little bit delayed and
the cement breaks in earlier as compared to the case (1a). Due
to differential density effects, the mud displacement process is
improved and the spacer volume fraction attains higher value of
0.88 as compared to 0.78 for case (1a). Unswept spacer fraction
after one cement annular flow increase and is evident in the
volume fraction contour plot. Due to its increased density, the
spacer is also left on the lower side of the annulus alongside
both annular walls.


Figure 7: Temporal fluid volume fraction variation in last one
third sections of the annulus for subcases (a) and (d)
A more severe case occurs when the spacer has same
density and viscosity as cement. A large quantity of unswept
spacer is present on the lower side of the annulus (Fig. 8). In


Copyright © 2014 by ASME

this case, the differential density effects are not present and the
cement on the lower side has not reached a cross sectional plane
at the 75% of the annular length while the cement on the upper
side has already reached the exit.

Figure 8: Planer and cross-sectional view of fluid volume
fraction contours in the last one half section of the annulus
Vertical configuration with constant eccentricity
Results of only two cases 5 and 8 are presented and
discussed here and the reader is referred to [7] for details on
comparative results and analysis. In this set of simulations the
50 ft. section is further divided into subsection (I, II, III, IV &
V) each of 10 ft length and the fluid volume fractions are
averaged over these sections. In this way, different sections near
exit can be compared with each other in terms of number of
fluid sweeps in each section and the volume fractions of
unswept fluids. For example, when the spacer front based on
average velocity reaches to the end of annular section, the
section (I) is swept by 4 times while section (V) is swept only
once by the spacer.
Case 5 e=0.25: Variation of various fluid volume
fractions for different Reynolds number (100 for case 5a and
400 for case 5c) are shown in Figure 9. The increase in the
spacer volume fraction is clearly visible with increasing
Reynolds number. One significant observation is that the spacer
can be seen in the narrow side of the annulus for the Reynolds
number of 400, while for the lower Reynolds number (100) only
the mud is present in the section (IV), Figure 10. It implies that
for some eccentric cases in which the eccentricity is not very
severe, the mud displacement efficiency can be enhanced by
increasing displacement rate. The spacer left over after one
annular sweep increases with displacement rate but increase is
not as significant as was in the case of low eccentric case of
0.05. The spacer maximum fraction reached in this case is
around 0.8 which is less then as compared 0.85 with
eccentricity of 0.05, this is due to the reason that more mud is
present in the observation section at all times during the sweep.
Comparison of fluid volume fractions in different sections
(I-V) with the increasing Reynolds number showed that the mud
fraction left in section (IV) increased (Figure 10).

Figure 9: Temporal fluid volume fraction variation in last 10 ft
sections of the annulus for cases (5a) and (5c) respectively.

Figure 10: Comparison of different fluid volume fractions along
the 50 ft. vertical section.
The lower three sections (I-III) in both cases indicate perfect
cement job. The top section (V) has a large fraction of bypassed
mud along with some fraction of spacer as well. The decrease in
cement fraction in the section (V) with increasing displacement
rate can be explained by the fact that depending on the
displacement rate the interfaces of cement-spacer and spacermud behave differently. Both Rayleigh-Taylor [12] and
Saffman-Taylor [13] instabilities grows and mixing of fluids
occurs and as a result, some of the spacer is also left behind,
while below a certain critical displacement rate the instabilities
does not grow rapidly and the sweep is good. The ratio of mud
to cement is used to define the green (φm /φc ≤ 0.5), yellow
(0.5< φm /φc < 0.1) and red (φm /φc ≥ 0.1) coloring for the
sections shown in Figure 10.


Copyright © 2014 by ASME

Case 8 e=0.95: As the eccentricity increased from 0.25 to
0.95, it can be observed (Figure 11) that in the given three
sections, the volume fraction of unswept mud decreases. The
reason for this decrease is the immobile fluid in a very narrow
gap. Since the volume fraction is calculated as the average in
the cross-sectional area and section length, the flow
contributions of wide section compared to narrow section are
very large. Most of the flow takes the path of wider side and the
mud is confined or static in the narrow gap. A major difference
from the previous cases with lower eccentricity is that the small
amount of unswept mud is observed even in the lowest two
sections as well (Fig. 11).



Figure 12: Temporal fluid volume fraction variation in last
section for cases (8a) and (8d) respectively




Figure 11: Fluid volume fraction in different sections (left –
first 10 ft., middle – 20 to 30 ft. section, right – last 10 ft.) for
the case 8c.
The severity of unswept mud on the narrow side of the
annulus can be seen in Figure 11. The continuous volume
fraction distribution (i.e. layer or channel of mud) starting from
entrance section to exit section of the annulus is present. If the
cement job in this high eccentricity annulus were evaluated in
only terms of the displacement efficiency, it might provide
completely misleading interpretation in terms of goodness of
the cement job.

The effect of increasing the displacement rate on various
fluid volume fractions in the observation section is shown in
Figure 12. With increasing the Reynolds number from 100 to
400 the flat plateau region attained by the spacer is raised from
0.66 volume fraction of spacer to 0.7 and correspondingly mud
fraction flat region is decreased from 0.32 to 0.29, showing
slight improvement in the mud sweep efficiency at higher rates.

Mud displacement efficiency is not a good indicator of
the severity of mud left in the annulus, for eccentric
geometries even a small fraction of left over mud may
be occupying a large region along the narrow side of
the annulus.
Gravitational segregation of different density fluids
along with variable eccentricity flow path in horizontal
well section makes the cement displacement
challenging. Due to variable eccentricity the flow on
the top side experiences an expansion in the flow area
and then contraction, while on the lower side the fluid
encounters reduction in flow area at the center. These
geometric variations combined with differential
density, viscosity and displacement rates result in a
complex flow fields.
Maximum cement volume fraction after one and half
sweep was obtained with spacers lighter than cement.
Even for smaller length horizontal sections, gravity
segregation of fluids was observed and its effects were
significant, resulting in the unswept mud along the top
portion of the annulus towards the exit.
For the case with maximum casing deflection resulting
in eccentricity of 0.6, the spacer with same density as
cemented performed very poorly, resulting in a
continuous spacer layer on the lower side of annulus.

In this study, the spacer volume pumped for horizontal
section is only half of the total annular volume and for
vertical section, it is exactly one annular volume followed


Copyright © 2014 by ASME

by one annular volume of cement. For the future extension
of such simulation study, it is suggested that the simulation
should be carried out to the point such that the spacer
fraction in the annulus reaches an asymptotic value with
minimum residual mud to better understand and quantify
the phenomenon. Similarly, the total pumped cement
volume fraction must be increased with simulation time to
achieve its asymptotic value.

[10] Frigaard, I.A. and Pelipenko,S. 2003. Effective and
Ineffective Strategies for Mud Removal and Cement Slurry
Design. Paper SPE 80999 presented at SPE Latin American
and Caribbean Petroleum Engineering Conference, Port-ofSpain, Trinidad, West Indies. 27–30 April. doi:
[11] Haut, Richard C. and Crook, Ronald J. 1979. Primary
Cementing: The Mud Displacement Process. Paper SPE
8253-MS presented at the SPE Annual Technical
Conference and Exhibition, Las Vegas, Nevada, USA, 2326 September, doi: 10.2118/8253-MS.

Authors are thankful to Prof. A. K. Wojtanowicz and Prof.
J. R. Smith at the Craft & Hawkins Department of Petroleum
Engineering, LSU, for the valuable discussions. Authors would
also like to express their gratitude for Chevron Emerging
Faculty Fund for providing the financial support for (M.Z.) to
carry out this research work.

[12] Rayleigh 1882. Investigation of the character of an
incompressible heavy fluid of variable density. Proc.,
doi: 10.1112/plms/s1-14.1.170.

[1] Fluent User’s Guide, 2012. Volume of Fluid method.

[13] Saffman P. G. & Taylor G. I., 1958, The penetration of
a fluid into a porous medium or Hele-Shaw cell containing
a more visous fluid. Proc. Royal Society, 245, 312-329.

[2] Keller, S.R., Crook R.J., Haut R.C. and Kulakofaky
D.S. 1987. Deviated-Wellbore Cementing: Part 1 –
Problems. J Pet Technology 39 (8): 955-960; SPE-11979PA.doi: 10.2118/11979-PA.

[14] Merlo, Antonino, Maglione, Roberto, Piatti, Cesare,
1995, An Innovative Model for Drilling Fluid Hydraulics.
Paper SPE- 29259-MS presented at SPE Asia Pacific Oil
and Gas Conference, Kuala Lumpur, Malaysia, 20-22
March, SPE-29259-MS. doi: 10.2118/29259-MS.

[3] Nelson, E.B., and Guillot, D. 2006, Well Cementing,
Schlumberger Publications, Second Edition.
[4] Sabins, Fred L., 1990. Problems in Cementing
Horizontal Wells. J Pet Technology 42 (4): 398-400; SPE20005-PA.doi: 10.2118/20005-PA.
[5] Tehrani, M.A. Bittleston, S.H. and Long, P.J.G. 1993.
Flow instabilities during annular displacement of one
non-Newtonian fluid by another. Experiments in Fluids
14: 246-256.
[6] Youngs D.L., 1982. Time-Dependent Multi-Material
Flow with Large Fluid Distortion. Numerical Methods for
Fluid Dynamics (ed. K.W. Morton and M.J. Baines), pp.
273-285, New York Academic Press.
[7] Zulqarnain, M., 2012. Simulations of the Primary
Cement Placement in Annular Geometries during Well
Completion Using Computational Fluid Dynamics (CFD),
MS Thesis, Louisiana State University, Baton Rouge.
[8] Wilson, M.A. and Sabins, F.L. 1988. A Laboratory
Investigation of Cementing Horizontal Wells. SPE Drill
Eng 3 (3): 275-280. SPE- 16928-PA. doi: 10.2118/16928PA.
[9] Clark, J.B., Howard, G.C., 1948. Factors to be
Considered in Obtaining Proper Cementing of Casing. API,
Drilling and Production Practice, 257-72: API.


Copyright © 2014 by ASME