SPE 81032
A Comprehensive Approach to Modeling Sanding During Oil Production
Alireza Nouri, Hans Vaziri*, SPE, Hadi Belhaj, Rafiqul Islam, SPE
Dalhousie University, Halifax, Canada
*BP-AMOCO, Houston, Texas, USA
Copyright 2003, Society of Petroleum Engineers Inc.
This paper was prepared for presentation at the SPE Latin American and Caribbean petroleum
Engineering Conference held in Port-of-Spain, Trinidad, West Indies, 27–30 April 2003.
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, as
presented, have not been reviewed by the Society of Petroleum Engineers and are subject to
correction by the author(s). The material, as presented, does not necessarily reflect any
position of the Society of Petroleum Engineers, its officers, or members. Papers presented at
SPE meetings are subject to publication review by Editorial Committees of the Society of
Petroleum Engineers. Electronic reproduction, distribution, or storage of any part of this paper
for commercial purposes 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 where and by whom the paper was presented. Write Librarian, SPE, P.O.
Box 833836, Richardson, TX 75083-3836, U.S.A., fax 01-972-952-9435.
Abstract
Sand production has been a major dilemma facing operating
oil companies over many years, sometimes substantially
increasing production costs. On the other hand, a very
controlled solid production can enhance oil production. A
dependable predictive model is vital for planning production
strategies in order to optimize well production. To date,
despite several research studies, sand production remains the
nightmare of petroleum engineers.
Even though many researchers have tried to predict sand
production in the past, none of them suggested a
comprehensive model that takes care of a variety of
mechanisms at different points and different times. Moreover,
rare models predict sanding rate and volume along with
sanding initiation. This paper presents a comprehensive
numerical modeling of sand production that appreciates the
different behavior of the medium near and far well-bore from
early to late life-time. Sanding criteria were adopted according
to the physics of the problem, by taking the sequential nature
of sand production into consideration.
The numerical model that was used not only assesses sand
production qualitatively but can also give the sanding rate at
different times. This was used to model the observations of
sand production in a large block test, and the sanding rate and
volume generated from numerical modeling agreed with
experimental results. With each stage of increased drawdown
or depletion, a burst of sand took place which enlarged the
cavities initiated from the perforations. The expansion of the
cavity was soon stabilized and this behavior was predicted by
numerical modeling. Moreover, besides considering shear and
tensile failure of the material, the possibility of volumetric
failure has been discussed.
Introduction
Sand Production in the petroleum industry is a phenomenon of
solid particles being produced together with reservoir fluid.
This phenomenon is costing the industry billions of dollars
every year. Corrosion of pipelines and other facilities, sand-oil
separation costs, possible wellbore choke, environmental
effects, reduction of production rate and possible work-overs
for clean-up operations are some examples of these costs. On
the other hand, a controlled sanding or even sand production
invocation has proved to be very effective in increasing
production rate, especially in heavy oil recovery, asphalt wells
and low PI wells 1,2,3.
Sand production takes place if the material around the cavity is
disaggregated and then there is enough fluid flow rate to
produce the grain particles. Disaggregation of the material
initiates from cavity faces and propagates inside the medium.
Material disaggregation can take place if the material fails
under excessive drawdown or depletion or a combination of
them. Depletion and drawdown fail the medium under either
of shear or tensile or volumetric failure mechanisms or a
combination of them. After failure of the rock disaggregates
the material, the resulted grains are produced by the existence
of enough pressure gradients in tension. Friction between the
grains and capillary tension are the resisting forces against
grains movements 4,5.
To date, there has been no comprehensive mathematical model
that considers all the mechanisms associated with sand
production. A model is presented in this paper that takes into
account different failure mechanisms that may play a role in
sanding. Therefore, in the rest of this paper, first, possible
failure mechanisms associated with sanding are explained. For
that, a comprehensive model for simulating the behavior of the
formation against the applied loads in early and late life, near
and far well-bore is presented. The significance of this
numerical model is providing a tool for better understanding
and modeling of sand production by better prediction of the
failure mechanism, sanding rate, and volume.
First Stage: Failure of Rock
Drawdown Induced Failure Drawdown may fail the rock
under shear, tensile or volumetric failure mechanisms.
Shear Failure Induced Sanding Drilling the wellbore or
creating any cavity like perforations, changes the stress pattern
in the medium around the cavity. Effective stresses normal to
cavity face, here called radial stress, drop to zero as soon as
2 SPE 81032
the cavity is created. The radial effective stress profile tends to
the initial effective radial stress away from the cavity. But,
tangential and vertical stresses recover faster than radial
stresses. This induces high shear stresses around the cavity.
Increase of drawdown augments effective stresses in an
interval around the wellbore. This is attributed to the fact that
pore pressure recovers much more slowly compared to total
stresses. Therefore, shear stresses increase around the cavities
once higher drawdowns are used. In this respect, drawdown
acts very similarly to depletion in increasing shear stresses
around the cavities. This effect is sometimes neglected in the
studies of sand production, and a common belief is that
drawdown may induce sanding only because it causes pressure
gradients which may fail the material in tension and worse,
push the sand grains to the wellbore.
The concomitant effect of drawdown on sand production has
appeared in the literature. For instance, Morita and Boyd 5
reported a field case in Alaska with ultra-weak formation
having inter-layers with Unconfined Compressive Strength
(UCS) of 550 kPa (80 psi). After installing a submersible
pump, the well initially produced 1 to 4 % sand which
increased to 20%. It was believed that drawdown induced
shear type failure in the material that was produced by fluid
flow.
Sanfilippo et al. 6 defined the term “total drawdown” as sum of
drawdown and depletion. According to their theory, sand
production takes place only if rock is plasticized around the
perforations by drawdown or a combination of depletion and
drawdown. By increasing drawdown, background sand
production rate increases which was attributed to the increase
of erosion rate from failed material as drawdown increases.
Tensile Failure Induced Sanding Tensile failure sand
production mainly takes place in unconsolidated medium
when flow gradient is high. Any increase in drawdown is
associated with elevation of pressure gradient around the
perforations that in severe cases may lead to tensile failure in
unconsolidated formations. This effect is magnified especially
where permeability of the rock has been harshly reduced as a
consequence of perforation damage.
Tensile failure induced sand production is often sporadic. It
usually produces low magnitudes of sand. Repeated shut in
and bean up increases drag forces on grains and, therefore,
may increase sand production.
Depletion Induced Failure Depletion of the reservoir may
fail the rock under shear or volumetric mechanisms.
Shear Failure Induced Sanding As the reservoir pressure
depletes, effective stresses increase in the medium. Increase of
effective stresses around the well-bore increases the load on
the cavities, perforation cavities or open hole, and augments
shear stresses around the cavities. Additional depletion
induces more increase in shear stresses which adds to the shear
stress induced by drawdown. Shear failure takes place once
shear stresses exceed limit shear strength of the intact rock.
Shear failure mechanism is mainly active around the cavities
where two major criteria are fulfilled. Firstly, shear stresses
are very high and secondly differential deformations are
possible. As sanding continues, effective stresses are
redistributed around the cavity as the failed material is
removed by tensile failure 5. This stress redistribution, in turn,
may fail additional material which provides more grains for
production. Therefore, the progressive cycle of failure-
production continues until a stable geometry of cavity is
formed.
Volumetric Failure Induced Sanding This mechanism is
mainly activated where either lateral displacement is zero or
very small. In this case, shear failure may not take place and
the only disaggregating mechanism would be pore collapse in
the course of volumetric failure. Higher porosity and lower
mechanical strength add to the likelihood of the occurrence of
pore collapse. In the mathematical modeling, once stress path
meets the cap, i.e. volumetric failure surface, volumetric
failure takes place.
In the past, pore collapse phenomenon has been studied to
understand the driving mechanism of land subsidence of oil
fields. Nevertheless, its association with sand production has
entirely been ignored. Under large depletion the overall
disaggregation may take place wherein compressive strains
breakdown the bonding among the sand grains. This depletion
induced volumetric failure starts from near the wellbore and/or
weaker zones and develops to other parts of the reservoir. The
key point is that total stresses around any opening recover very
quickly inwards. On the other hand, pore pressure changes are
much more moderate. Therefore, effective stresses near the
wellbore are usually larger than those of inside the reservoir.
Later, as oil production continues, the pressure of the whole
reservoir may get depleted and pore collapse, which initially
started from near wellbore, may spread to the whole medium.
The overall volumetric failure of the reservoirs consequent to
depletion has been reported for a number of reservoir, e.g.
Ekofisk field 7. Hamilton and Shafer 8 studied pore collapse
characteristics of a high porosity Diatomite and carbonate
rock. The pore collapse pressure, which corresponds to the cap
pressure, highly depends on porosity and UCS (or any another
strength representative) of the rock. As an example, Soares
and Ferreira 9 reported test results of pore collapse
experiments on limestone samples taken from Campos basin
having a porosity in the range of 25 to 35%. Cap pressure of
these samples ranged from almost 40 Mpa for samples with
31% porosity to almost 90 Mpa for those with 20% porosity.
In the other extreme, Hamilton and Shafer 8 experimented fine
grain diatomite with porosity over 50% and reached to a cap
pressure of almost 3 Mpa. They also performed the same test
on carbonate samples with over 20% porosity and the cap
pressure was 24 Mpa. For Ekofisk field that pore collapse of
chalk medium resulted in over four meters of sea floor
subsidence, a cap pressure of almost 22 Mpa was concluded 7.
Similar tests were performed by Ditzhuijzen et al. 10 on samples
from offshore Sarawak. Mouldic limestone and dolomite
samples ranged from 26.8 to 40.7% in porosity, and resulted in
15 to 36 Mpa vertical failure stress for limestone while 33 to
47 Mpa for Dolomitic samples. Smits et al. 11 reported a pore
collapse vertical effective stress of 20 Mpa to 50 Mpa for
Mouldic limestone with porosity in range of 27 to 39%. They
also concluded a failure stress of 18 Mpa to 60 Mpa for
Danian and Maastrichtian chalk samples that were in range of
35 to 50% porosity.
SPE 81032 3

Second Stage: Push of Grains into the Wellbore by
Seepage Forces
The grains of failed material are produced if the hold forces
between them consisting of friction and capillary cohesion are
smaller than drag forces induced by seepage. According to
Sanfilippo et al. 6, it is impossible to erode away the grains
from a sand compacted in such a way that it has a uniaxial
strength above 1 MPa. Water breakthrough destroys capillary
tension while repeated shut in and bean up may release the
fine intergranular material with fluid flow which in turn
loosens the grain particles. It also exerts significant pressure
gradient on the material which may fail the material in tension.
Moreover, the boosted pressure gradient applies huge drag
forces on the sand grains which may mobilize them.
Whether sanding will occur depends on availability of
capillary cohesion 4, strength properties of the disaggregated
rock mass (such as projected cohesion, friction), pressure
gradient (drawdown & permeability), and fluid velocity.
Modeling of Sanding
The numerical modeling was carried out by using FLAC
12
. Sand production in this model is described in terms of the
coupled mechanisms of fluid flow, stress evolution, and
changing boundary conditions with time.
A non-associated bilinear Mohr-Coulomb strain
hardening/softening model was used in the analysis in which
the failure envelop is a combination of two criteria along with
a tension cut-off. Both bilinear shear and tension cut-off can
harden or soften in compliance with the hardening parameters.
Hardening parameters are a measure of plastic shear and
tensile strains. A general shape of the model is schematically
depicted in Figure 1.
Shear Stress
boundaries are defined. In the model, therefore, the pattern of
cavity development is visible at different stages of the loading.
Another unique feature of the model is the provision of the
tools to simulate all failure mechanisms of shear, tensile, and
volumetric failure. Volumetric failure mode, while
substantially important in the process of sand production, has
not been accounted for in the previous attempts of modeling of
sand production. Showing the possible contribution of
volumetric failure on sand production is the subject of another
paper 13.
Fundamental Mathematical Equations Key formulations,
based on Biot’s 14 consolidation theory, are as follows.
Equilibrium Equation The balance of momentum is:
*
∂σ ij du
+ ρg i = ρ
∂x j dt
where σ ij is the stress tensor and ρ is the density of the
geomaterial.
Compatibility Equation Compatibility equation defines the
relation between the strain rate and velocity gradient:
*
*
* 1 ∂ ui ∂u j
ε ij = [ + ]
2 ∂x j ∂x i
Constitutive Law of Geomaterial The small strain constitutive
law for the geomaterial is defined as:
*
d
(σ ij + P ) = H (σ ij , ε ij , κ )
dt
In this equation, H is the functional form of the
constitutive law, and κ is the hardening parameter.
Continuity Equation Mass balance equation is:

∂ζ ∂q
= − i + qv
Residual Strength ∂t ∂x i
in which ζ is the variation of fluid content and qv is the
φ2
volumetric fluid source intensity.
Fluid Flow Transport Law Darcy’s law was used to model
the relation of fluid flow velocity-pressure gradient.
Cp ∂
Cr φ1
q i = − k ij (P − ρ w g k xk )
∂x j
Normal Effective Stress In this equation, qi is the specific discharge vector and k ij is
Tension Cut-off
the mobility coefficient.
Figure 1. Material model Fluid Flow Constitutive Law This equation defines the
relation of pressure change of the fluid with the fluid flow rate
The proposed model reflects the general physics of the difference in response to the mechanical deformation and
problem, sanding takes place if the rock is directly failed in variation of fluid content.
tension or the failed rock under shear falls into tension. ∂P ∂ζ ∂ε
Tensile strength and real cohesion of the rock reduces as shear = M( − )
failure develops in the material. This corresponds with the fact ∂t ∂t ∂t
that disintegrated rock under shear failure has no or little In this equation, M is Biot modulus and ε is the volumetric
tensile strength. This concept has the significant advantage Kw
strain. Biot Modulus is defined as where K w is Bulk
that complies with the physics of the problem. As soon as the ϕ
element, i.e. the zone, satisfies the sanding criteria, it is modulus of water and ϕ is the porosity of the geomaterial.
removed from the mesh entirely and consequently, new
4 SPE 81032

Experimental Data Kooijman et al. 15 performed a large scale
laboratory sand production on a perforated block sample. The
effect of drawdown as well as depletion was experimented on
the sample. It was observed that with the increase of both
drawdown and depletion sand bursts took place. This paper
attempts to numerically model the process of sanding by
simulating its progressive failure and seepage associated sand
release. Sanding volume and rate as well as fluid production
rate are all estimated.
A sandstone block of 0.70m*0.70m*0.81m was tested by
application of varying horizontal and vertical stresses and
pressures. A hole of 0.105 m diameter was created in the
center and the fluid was flowed from external circumference
to internal hole in the course of the experiment.
Material Parameters In order to capture the physics of the
failure and consider the possible contribution of all of the
aforementioned failure mechanisms, a Mohr-Coulomb mixed
hardening/softening model was used in the simulations.
According to the model, tensile failure takes place when stress
tensor meets the tension cut-off. Shear failures occurs when
the stress tensor touches the shear envelop. Theoretically,
shear failure may only take place adjacent to the cavity faces
where there is a possibility of differential deformation to take
place. Pore collapse is defined when stress content meets a
volumetric yield surface, i.e. a cap. Table 1 lists a summary of
the physical and mechanical parameters of the outcrop
material experimented. According to the mineralogy tests, the
rock sample composed of 70% quartz and 30% mixed
feldspars and rock fragments. The fluid passed through the
sample throughout the test was an oil mixture with a viscosity
of almost 10 CP 15.
basis, one cannot get the same response as the base of the
analysis was not on the grain size. Therefore, a slightly larger
hole size must be used to make this work.
Boundary Conditions The test program contains three main
stages which consisted of perforation job, increase of
drawdown and effective stress increase (which modeled
depletion). Boundary conditions were defined to reflect the
real testing conditions and are shown in Figure 2. The
magnitude of the applied stresses and pressures in different
stages are also summarized in Table 2. The values showed in
this table are nominal. The real values corrected for the
geometrical and loading efficiency effects are shown in
brackets.
A zero pore pressure was defined in the perforation edge that
modeled the atmospheric pressure in the wellbore. In the outer
boundary, a varying fluid pressure was defined that simulated
the applied changing pressure in that edge as shown in Figure
3.
For the mechanical part, displacement was fixed at the casing
touching points. At the top boundary, the vertical stresses were
defined as listed in Table 2. A variable stress was applied at
the outer right boundary as shown in Figure 3. Lower edge
boundary is fixed in vertical direction.
It must be notified that the time scale in Figure 3 accounts
for the effective real time. In other words, the extended shut in
times were eliminated in the numerical modeling process.
variable top pressure
normal velocity=0

Table 1. Physical and mechanical properties of the

Castlegate sandstone (Kooijman et al. 1992)
atmospheric pressure 0.80 mvariable reservoir pressure
variable horizontal pressure
TWC UCS ϕ κ eff C Φ
(MPa) (MPa) (MPa)
(mD) (degrees) Fixed in vertical direction
Castlegate normal velocity=0
(dry)
40 10 0.26 5.0 37
0.40 m
Castlegate
(wet)
30 9 0.26 370 3.9 35
Figure 2. Boundary conditions

The Finite Difference Mesh Considered problem consists of Table 2. Test scheme
a section of the sample with inner diameter of 10.2cm, and Description of σv σh Pp Pw
outer radius of 40cm. The finite difference mesh includes 2880 Phases
zones. Axial symmetry was assumed in the analyses, with
Phase (MPa) (MPa) (MPa) (MPa)
velocities and excess pore pressure as degrees of freedoms.
Finite difference mesh was set with finely spaced zones near 0 Perforation 10.33 (9.30) 4.13(3.72) 0 0
the wellbore. These zones were designed coarser away from 1 Perforation 13.77(12.40) 7.58(6.82) 3.44 .83
the wellbore so as to reduce the total number of them, while at
2 Intermediate 10.33 (9.30) 4.13(3.72) 0 0
the same time not compromising the accuracy of the
computations. 11.02-13.09 4.82-6.89
Furthermore, there was a need to make the opening for the 3A-D Increase DD .69-2.76 0
(9.92-11.78) (4.33-6.20)
ring at a level to enable a continuum representation of the
particulate system to work. While small diameter perforations 4 Depletion 19.98(17.98) 9.64(8.68) 2.41 0
are large enough to allow very fine sand particles to get 5 Depletion 26.86(24.18) 12.40(11.16) 2.41 0
through, in the numerical modeling with continuum mechanics
SPE 81032 5

Numerical Modeling Results

25
Effective stresses, exterior boundary

σ'v Calibration Calibration consists of adjusting the input data

20 until the computed sanding rate and volume match the
pore pressure (MPa)

experimental values. It is not unreasonable to adjust the input

15 data because these data are imperfectly known. In the
numerical simulations, elastic parameters, bulk and shear
10 modulus, were varied in the calibration process.
σ'h

5 Comparison of Experimental and Numerical Results

0 by initial productivity index attained form the numerical
0 2 4 6
Time (hours)
Figure 3. Loading history (boundary conditions)
Analysis Process In the analysis process, first the insitu
stresses were defined in the medium. Next, the perforation
stage was simulated. For this, effective stresses in the
perforating edge were reduced to zero. The zones
corresponding to the perforation tunnels were nullified and
boundary conditions were defined in the newly created
boundary. In stage 2, pore pressure was decreased to zero
while other total stresses were also decreased accordingly,
keeping effective stresses constant. Drawdown and later
depletion stages were followed as indicated in Table 3.
The loading process comprises of four drawdown steps which
was followed by two depletion stages. In each drawdown step,
an increase of 620 kPa (90 psi) was induced to the exterior
fluid pressure, while interior pressure was kept constant at
atmospheric. At the same time, the same amount was added to
the applied top and right total stresses. In the last drawdown
step, pressure difference was decreased as much as 344 kPa
(50 psi) in order to reduce the flow rate which was exceeding
the maximum range of the flow-meter.
Depletion effect was also studied by two subsequent increase
of applied stress while keeping the drawdown constant at 2400
kPa (350 psi). In each depletion stage, an increase of 6200 kPa
(900 psi) in vertical stress and 2480 kPa (360 psi) in horizontal
stress were applied. As indicated in Table 3 and also reflected
in Figure 3, a shut-in and bean-up stage was considered
between the two depletion stages in the numerical analysis.
This was to simulate the corresponding shut-in of the night of
the end of first testing day and following bean-up of the next
day.
pp
Figure 4 compares the ratio of productivity index divided
8 10 12 14
simulations and experiments. Productivity index is defined as
the ratio of flow rate to drawdown (PI=Q/DD).
It is shown in Fg. 4 that productivity index changes in the
course of the experiment. Change of effective stresses alters
the porosity and rearranges the grains and, as such, affects
permeability. In general, an increase in mean effective stresses
decrease the permeability as the material compacts. The two
drops in the ratio of productivity index in the graph of Fig. 4
are associated with the increase of hydrostatic stress in the two
depletion phases. Increased shear stresses, on the other hand
may augment the permeability if the material dilates and shear
bands form. In the experiments, significant shear stresses
develop around the perforations which progress inward as the
cavity grows inward. This can increase the permeability of the
zones affected by shear failure. Material compaction mostly
takes place where hydrostatic effective stress increases and is
intensified in looser material. Dilation and the formation of
shear bands, on the other hand, occur where shear stresses are
high and the material is dense. In the numerical modeling,
permeability change with effective stress variation was not
modeled.
Furthermore, flow rate gradually increases, as sand grains
continue to be released. This is because the sample dimension
is comparable to the dimension of the cavity which progresses
towards the outer boundary as sand grain are produced. In fact,
it is inferred from Fig. 4 that this cause has played an
important role in the variation of the ratio of productivity
index.
1.6
1.4
1.2
1
Numerical
PI/PI0

Table 3. Numerical analysis steps 0.8

Experimental
σv (MPa) σh (MPa) Pp (MPa) t (hr) 0.6
Initial conditions 12.40 6.82 3.44 0 0.4
Intermediate 9.30 3.72 0 0.28
0.2
DDA 9.92 4.34 0.689 1.36
DDB 10.54 4.96 1.38 2.08 0
0 1000 2000 3000 4000 5000
DDC 11.16 5.58 2.07 3.00
DDD1 11.78 6.20 2.76 4.26 Cumulative flow (l)
DDD2 11.78 6.20 2.41 5.96
DEPA 17.98 8.68 2.41 7.54 Figure 4. Comparison of ratio of PI/(initial PI) resulted from
Shut-in 15.81 6.51 0 7.68 experiments and numerical modeling
Bean-up 17.98 8.68 2.41 8.54
DEPB 24.18 11.16 2.41 9.34
6 SPE 81032

One of the most important features of our numerical

modeling is its capability in forecasting the volume and rate of 0.2

sanding. The volume of sand being produced at any time 0.18

corresponds with the volume of the removed element which 0.16

0.14
satisfies the removal criteria.

Sand cut (gr/l)

0.12
Figure 5 shows a comparison of the sand-cut extracted Expe rim e nta l
0.1
from experimental and numerical results. Sand-cut is defined Nim e ric a l

as the ratio of sanding rate to flow rate. Figure 5 shows that 0.08
0.06
large amount of sanding took place in the first drawdown. This
0.04
amount is partly supplied from the perforating induced
0.02
shattered material plus some aluminum charge powder.
0
Numerical modeling also predicts comparatively large amount
0 1000 2000 3000 4000 5000
of sanding induced by shear failure and release of some
material as marked by the fist peak in Fig 5. Cumulative flow (l)

4.5 Figure 6. Comparison of Sand-cut resulted from the experiments

and numerical analysis excluding the first drawdown related data
4
3.5
3
Sand cut (gr/l)

2.5 Expe rim e nta l

2 Nim e ric a l
1.5
1
0.5
0 40
0 1000 2000 3000 4000
Cumulative flow (l)
Figure 5. Comparison of Sand-cut resulted from the experiments
and numerical analysis
As shown in Fig. 6, sand-cut dropped in the next
drawdowns, when still, with each drawdown a sand burst took
place. Sanding predictions with numerical modeling show a
good agreement with the experimental results. It must be
mentioned that in the sand production modeling, it is the
prediction of the order of magnitude of sanding rate or volume
which is most important. Moreover, the prediction must
capture the right mechanism of sanding. It is its general
agreement with the real data that is essential. The models
exact prediction of sanding is neither important nor possible
by any means. In that sense, the model well forecasted the
sanding as occurred in the experiments.
Increase of effective stress in the next stage of the experiment
simulated depletion of the pore pressure in the real reservoir.
This part is reflected in the last two peaks of Figure 6. With
each increase in effective stresses, a new burst of sanding
occurred. This corresponds with the increased effective
stresses in the outer boundary which translates to more
intensified shear stresses on and near the cavity face.
Figure 7 shows the cumulative sanding in the course of
experiment as predicted by numerical modeling. It is noticed
that the portion associated with the initial burst of sanding in
the first drawdown in addition to that corresponding to the two
depletion stages constituted the majority of sand production.
100
90
80
Cumulative Sand (gr)
70
60
50
5000 30
20
10
0
0.00 2.00 4.00 6.00 8.00 10.00
Time (hours)
Figure 7. Cumulative sanding weight
Summary and Conclusions
Both drawdown and depletion may induce sand production
in a similar way. In fact, depletion increases the effective
stress in the reservoir which in effect, translates to shear
stresses on the cavity face. Increase of drawdown, on the other
hand, increases the effective stresses only in the zone near the
cavity face. This is because pore pressures do not recover as
quickly as total stresses from cavity face toward the reservoir.
Therefore, in a confined zone near the cavity, effective
stresses increase. Similar to depletion effect, shear stresses
around the cavity increase which may lead to more shear
failure. These two stages have been simulated in the
experiments by changing the boundary stresses and pressure
according to the effects that they induce on and around the
cavity face.
Experimental and numerical results showed a burst of sand
production associated with each drawdown and depletion. In
the case of the experiments, cavity growth soon slowed down.
The growth may, in turn, not get stabilized and lead to
continuous sanding and cavity propagation.
Numerical modeling well captured each sand burst and
consequent stabilization. The rate and volume of sanding
reasonably well agreed with the experimental results. This
shows that the criteria used in the modeling are capturing the
SPE 81032 7

essence of sanding, and therefore, the use of this numerical 4. Vaziri, H., B.Barree, Y.Xiao, I.Palmer and M.Kutas, “What Is the
tool in sanding prediction looks very promising. Magic of Water in Producing Sand?”, SPE 77683, SPE Annual
Technical Conference and Exhibition, San Antonio, Texas, 2002.
Nomenclature
cp= plastic projected cohesion 5. Morita, N., P.A. Boyd, “Typical Sand Production Problems: Case
Studies and Strategies for Sand Control”, SPE 22739, 66th
cr= plastic real cohesion
Annual Conference, Dallas, TX, 1991.
DD= drawdown
g= gravitational force 6. Sanfilippo F., Brignoli M., Giacca D. and Santarelli F.J., “Sand
H= functional form of the constitutive law Production: From Prediction to Management”, SPE 38185, SPE
k ij = mobility coefficient. European Formation Damage Conference, The Netherland,
1997.
K w = Bulk modulus of water
M= Biot modulus 7. Lawrence W.T., R W. Douglas, H.E. Farrell, “Effect of Reservoir
P= pore pressure Depletion and Pore Pressure Drawdown on In Situ Stress and
PI= productivity index Deformation in the Ekofisk Field, North Sea”, Rock Mechanics
Pp =exterior boundary pressure and Multidisiplinary Science, Balkema, Rotterdom, 1991.
Pw= wellbore pressure
qi = specific discharge vector 8. Hamilton, J.M. and J.L. Shafer, “Measurement of Pore
Compressibility Charachteristics in Rock Exhibiting Pore
qv = volumetric fluid source intensity. Collapse and Volumetric Creep”, SCA conference paper
Q= flow rate Number 9124, 1991.
t= time
TWC = thick wall cylinder test strength 9. Soares A.C. and F.H. Ferreira, “An Experimental Study for
* Mechanical Formation Damage”, SPE 73734, SPE international
u i = velocity in i direction symposium and exhibition on formation damage control,
UCS= unconfined compressive strength Lafayette, Louisiana, 2002.
ζ = fluid content
10. Ditzhuijzen P.J.D. and J.A. Waal, “Reservoir Compaction and
ε = volumetric strain
Surface Subsidence in the Central Luconia Gas bearing
ϕ = porosity of the geomaterial Carbonates, Offshore Sarwak, East Malaysia”, SPE 12400, SPE
φ = friction angle 5th offshore south east Asia conference, Singapore, 1984.
κ = the hardening parameter
ρ = density of the geomaterial. 11. Smits R.M.M., A. Waal and J.F.C. Kooten, “Prediction of Abrupt
Reservoir Compaction and Surface Subsidence Due to Pore
σ ij = stress tensor Collapse in Carbonates”, SPE 15642, 61th annual SPE
conference, New Orleans, LA, 1986.
σh= horizontal stress applied on exterior boundary
σv = vertical stress applied on top boundary 12. Itasca Consulting Group: FLAC- Fast Lagrangian Analysis of
Continua- User’s Guide 3.4, Minneapolis, Minnesota, 1998.
Acknowledgements
The authors would like to thank NSERC, Killam trustees 13. Nouri A., H. Vaziri, H. Belhaj, R. Islam, Effect of Volumetric
and ACPI for their financial support. Failure on Sand Production in Oil-Wellbores, SPE 80448, SPE
Asia Pacific Oil and Gas Conference and Exhibition, Jakarta,
References: Indonesia, April 2003.
1. Dusseault, M.B. and S.El-Sayed, “Heavy-Oil Production
Enhancement by Encouraging Sand Production”, SPE 59276, 14. Biot, M.A., “General Theory of Three Dimensional
SPE International Symposium on Formation Damage, Lafayette, Consolidation”, Journal of Applied Physics 12, 155-164, 1941.
Louisiana, 2000.
15. Kooijman A.P., P.M. Halleck and C.A.M. Veeken, “Large-Scale
2. Wang, Y. and C.C. Chen, “Improved Production and Sand (Cold) Laboratory Sand Production Test”, SPE 24798, 67th Annual
Production in Conventional and Heavy Oil Reservoirs- A Field Technical Conference and Exhibition, Washington DC, 1992.
Case and Simulation”, SPE 57290, SPE Asia Pacific improved
oil recovery Conference, Kuala Lumpur, Malaysia, 1999.

3. Smith, G.E., “Fluid Flow and Sand Production in Heavy Oil

Reservoirs under Solution Gas Drive”, SPE 15094, SPE
California regional meeting, Oakland, California, 1986.