Society of Petroleum Engineers

SPE 27343
Sand Production as a Viscoplastic Granular Flow
M.B. Geilikman, M.B. Dusseault, * and F.A. Dullien, * U. of Waterloo
SPE Members
Copyright 1994, Society of Petroleum Engineers Inc.
This peper was prepared for presentation at the International Symposium on Formation Damage COntrol, Lafayette, Loulslena, 9-10 February, 1994.
This paper was selected for presentation by SPE Program COmmittee following review of information contained In an abstract submitted by the authors. Contents
of the paper, as presented, have not been reviewed by Society of Petroleum Engineers and are SUbject to correction by the authors. The material, as presented,
does not necessarily reflect any position of the Society of Petroleum Engineers, lis officers, or members. Papers presented aI SPE meetings are subject to publication
review by Editorial Committee of the Society Of Petroleum Engineers. Permission to copy Is restrlcled to an abstract 01 not more than 300 words. lIIustrallons may
not be copied. The abstract shoUld contain consplclous acknowledgment of where and by whom the paper Is presented. Write Librarlan, SPE, P.O. Box 833836,
Richardson, TX 75083-3836 U.SA. TeleX, 730989 SPEDAL.

ABSTRACT

Mohr-Coulomb material model. This mechanism leads to

massive solid flux (plastic flow), and this plastic flow can be
a major factor in sand production. In this paper, we will try
to consider this mechanism of solid production quantitatively,
analyzing not only instability initiation, but also dynamic and
mass transport properties related to sand production after the
appearance of yielded zone, including the process of radial
propagation of the yielded zone around the wellbore as sand
production continues.

A continuum model for sand production is developed by

considering coupling of fluid flow and rate-sensitive
plastic flow of granular matrix. Initial granular matrix
yielding is reached as a result of wellbore fluid pressure
reduction (drawdown) below some critical level. The
model provides a physical mechanism of yielding front
propagation away from the wellbore as the process of
sand production continues. Mass balance analysis leads
to a relation between cumulative sand production and the
current yielding front location. Different regimes of sand
production, limited and unstable, are considered. Fluid
production enhancement because of simultaneous sand
production turns out to be dependent not only on
instantaneous sand flux, but also on the current radius of
the yielded zone around wellbore, that is, the history of
sand production for the well.

A macroscopic continuum description of fluid-saturated

granular flow should reflect grain-grain and grain-fluid
interactions. Analysis at the grain scale is beyond the scope
of this article, but it is important to understand the major
physical mechanisms which govern the rheology of the
system.
First, fluid-saturated granular media inherit plasticity from
dry ones because plastic behaviour arises directly from

Decreasing of fluid pressure in the wellbore creates an

excessive deviatoric stress that can cause instability,
which was first considered by Bratli and Risnes 1 for a

frictional grain-grain contacts. Dense granular systems have

physical properties that are strongly density-dependenf,3. In
fact, the shear strength of unconsolidated sandstones
decreases dramatically as a function of fabric disturbance4
This means that a yield stress criterion must depend on
porosity, and this will be more specifically addressed later.

References and illustrations at end of paper

Additionally, a viscous (or rate-sensitive) mode of fluid-

INTRODUCTION

41

SPE 27343

SAND PRODUCTION AS A VISCOPLASTIC GRANULAR FLOW

symmetry was used for simplicity, but generalization is not

complicated.

saturated granular medium behaviour appears because of

viscous dissipation in the fluid between grains. This is
the mechanism responsible for high viscous energy
dissipation in dense suspensions5 Unlike Darcy viscous
resistance, this mechanism makes a contribution to
energy dissipation even if an average fluid flux is absent
in the system. This kind of dissipation is necessarily
stipulated by relative motion of grains, and should be
included as an aspect of the granular skeleton rheology
being described in a general macroscopic continuum
model. Interaction between the skeleton and fluxes of
saturated fluid can be described, as usual, by Darcy
forces (hydrodynamic or seepage forces).

Stress equilibrium equation for solid skeleton:

dO, 0,- 0
Il
_
---a;:-+-r--137Z(vl-v) - 0 (1)
The coefficient 13 is a function of porosity both in Biot's and
in the de la Cruz-Spanos theory?, which is based on volume
averaging procedures. Darcy's law for the case of a mobile
solid matrix can be written as:

(2)

dP = -l:(v -v)
dr
k I s

The equations of continuity for the case of incompressible

fluid and solid phases (we assume this to simplify the
problem, but it is realistic as well) are:

The impetus to develop this model of massive sand

production comes from experience in the heavy oil fields
of Eastern Alberta and Western Saskatchewan. These
fields typically contain heavy oil of 14-18API gravity in
30% porosity cohesionless sandstone reservoirs from 315 m thick with oil saturations typically 88% and gas in
solution in the oil phase. Conventional well tests on
undisturbed reservoirs suggest production rates of less
than 1.5 m 3/day (10 bbl/day). However, if sand is
allowed to enter the wellbore unhindered (l % to 3% by
volume in most cases), production rates eight to ten
times higher than this can be achieved. If sand exclusion
technologies are used (gravel packs, screens), oil
production rates drop to the predicted uneconomic rates
of 1.5 rn 3/day. Thus, sand production is vital for
economic viability, and is also an important production
mechanism in its own right. To date, no model properly
accounting for the physical mechanisms involved and
allowing a prediction of history has existed; we hope the
present model goes some way to rectify this deficiency.

d<\>
dt

dive<\>

0; - d<\> + div[ (l-<\> )v]

VI) =

dt'

(3)

For the granular skeleton, we apply the general HohenemserPrager-Malvern approach expressing rate-sensitive plastic
behaviourS:

11

dV
s
dr

_ <<I> (F

dF

(4)

dO,

-for radial deformation rate, and

V

11-s
r

dF

= -<<I>(F_

(5)

dO o

-for transversal deformation rate. Here, 11 is a viscosity

coefficient which, generally speaking, is a function of
porosity. Function < <I> > is defined as being zero if <I> < 0,
and as being equal to <I> if <I> > O. The explicit form of the
function <I>(F) might vary for different materials. Herein, we
will use a linear expression, <I> = F. In this case,expressions
(4) and (5) define a specific example of linear irreversible
thermodynamic approaches when the fluxes are assumed to
be proportional to the driving forces, with some kinetic
coefficients of proportionality 9.

MODEL OF FLUID-SATURATED MEDIUM WITH

VISCOPLASTIC SKELETON
According to what has been said above, in the analytic
formulation of the problem we consider a two-continuum
medium with the following constituent components:

As an yield function F, the Drucker-Prager type of criterion

for effective stresses including porosity hardening is used:

a. A saturated viscous (incompressible) fluid; and,

b. A viscoplastic skeleton which is described according
to the general approach for rate-sensitive plasticity.

F=

0- 0

'+a(

0+0

--p)-K(<\

(6)

Here, a is a dimensionless parameter describing the

compaction behaviour of the skeleton.
The porosity
hardening function has a negative derivative

The appropriate equations for the specific case of

cylindrical flow are now presented; this high degree of
42

GEILIKMAN, DUSSEAULT, AND DULLIEN

SPE 27343

<0

d1C(<\

state), the equations of continuity (3) can be considered in

linear approximation, Le. by replacing porosity under the
operation of divergence by some constant value <\>y,
corresponding approximately to the porosity of a yielded
zone. Substituting the expression for solids velocity into the
continuity equation for the solid, we obtain an expression for
the evolution of porosity inside the yielded, flowing zone:

(7)

d<\>
if decreasing of porosity leads to hardening of the
skeleton, which is further supposed. An example of the
type of porosity-dependent yield criterion which one
should bear in mind is shown in Figure 1. Rigorously
speaking, near the critical state point, a non-linear
(parabolic) relationship between isotropic and deviatoric
stresses should be taken into account'.
This
complication will be addressed in a further article, and
we will continue to using the Drucker-Prager yield
criterion herein.

<\> (r,t) = <\>

= _ 1 +<X <<I>(F

During massive sand production, the boundary between

yielded and non-yielded zones may be assumed to move
radially away from the borehole. Of the solid mass coming
from the non-yielded zone (sloughing off the non-yielded
zone surface), part goes to increase the radius of the yielded
zone, and part is carried away from the boundary toward the
borehole (Figure 4). A mass balance at the boundary gives
the following equation:

Because flow is uniformly convergent

Recombining these equations leads to:

ovs

or

1-<x

(8)

211

Vs

+ --- =
1 +<X r

( 11)

Location of the boundary between yielded and nonyielded zones

Substituting the explicit expression (6) for the yield

function F into (4) and (5), we obtain:

2<X(1-<\
Set)
y
1 + <X
-2:----::-:2a::"'"

which corresponds to contraction if <X > 0, and to dilation for

negative values of <X. Here, Set) can be considered as the
cumulative solids production. As can be seen from equation
(11), <\>y is the porosity at the initial moment of solid
production, t = 0, when the yielded zone starts to appear and
grow around the borehole; <\>y can be found as a result of the
joint solution together with mass transport inside the
wellbore. The latter can be quantified by comparing the
grain size distribution being produced and that of the entire
material from cores. Because this issue is not addressed
explicitly in this article, <\>y can be taken as one of the
parameters of the theory. Evolution of porosity along the
radial direction is depicted schematically in Figure 3.

Solids velocity and porosity evolution

211

S(t)=. fdt'q(t ' )

As can be deduced from the yield function expression, a

decrease of fluid pressure increases the excessive
deviatoric stress, and can make F non-negative, which
implies instability with respect to plastic flow of the
skeleton'. Solid flux proceeds from the inside of this
zone toward the wellbore (Figure 2). Solids flux
dynamics and the evolution of the yielded zone can be
formulated from the approach which has been described
above, and these dynamic expressions will describe the
sand production mechanism.

1-<x <<I>(F;

r T+ii' rwT+ii'

SAND FLOW INDUCED BY REDUCTION OF

BOREHOLE PRESSURE (DRAWDOWN)

!.;:
=
or

v. < O.
(9)

(1 - <\> ) dR = (1 _ <\> )( dR - v )
n

Solutions of this equation can be represented in the

following form:

dt

dt

(12)

Assuming again small changes of porosity and using

expression (10) for solids velocity, we get the solution of this
equation, which provides an explicit relationship between the
radius of the yielded zone and cumulative solids production:

with an unknown function q(t) which has the sense of

volumetric rate of solid production. Assuming small
porosity changes (reasonable when not far from critical
43

SPE 27343

SAND PRODUCTION AS A VISCOPLASTIC GRANULAR FLOW

the Drucker-Prager type of criterion, equation (6), it means

effectively that not far from the critical point, <X 1. In
any case, according to the original theory of Drucker and
Prager l , the following inequality should be fulfilled: 12cr <
1. Obviously, at the critical point <X = 0, so it is reasonable
to develop a perturbation theory with respect to a small
coefficient of dilatancy <x.

2a

1+ii

Set) =

(1 +<X)( ~y -~n) rw

(12)

2(l-~y)

Equation for stress

First, consider the zero approximation; i.e., let <X = 0. From

the general equation (14), by substituting explicit expressions
for vf and Vs with <X = 0, we obtain in the limit:

To obtain a closed-fonn equation for radial stress, for

example, we can eliminate the tangential stress from
equations (1), (6) and (8), and obtain the following
equation:
da

~ + __

or

l+<xr

[<X(a -p)-K(~)]r

_ PI..l (v -v) +
k

dar _ 2K + PI..l(Q -.!L)2.- 411q =

dr
r
k
y
~y r
r3

411

(1+<X)2 r 2

(14)

O'(r) =p +[2K- P I..l(Q -.!L)]ln(r/r)+

r

~yv

~y

(15)

inside the yielded zone, and

Qn

(18)

r2

The follOWing boundary condition for radial stress at the

wellbore was used to obtain this result: arCrw) = Pw'
Expression (18) contains only one unknown quantity q(t),
taking into account that Qy can be expressed through q(t),
according to fonnula (A6). Furthermore, assuming small
differences in porosity between the yielded and- non-yielded
zones, we can as a first-order approximation neglect
differences between penneabilities ~ and kyo
This
approximation is not necessary for our analysis but it
significantly simplifies the fonnulas. In fact, non-constant
penneability is one of the more interesting aspects of sand
production, which we hope to investigate in further
developments.

VI = - -

rw2

Using a linear approximation with respect to variations

of porosity in the equations of continuity (3), solutions
for fluid velocity read as follows:
1-

1
1 y
+211q(---)

Fluid velocity and fluid pressure

(17)

Integration of this equation provides an expression for radial

stress:

The boundary condition for equation (14) is continuity of

radial stress at the boundary between the yielded and
non-yielded zones. To obtain a solution for radial stress,
it is necessary to substitute in equation (14) an explicit
expression for fluid velocity and fluid pressure.

VI = - Q y -

(16)

in the non-yielded zone. Clearly, inside the non-yielded

zone V s = 0. Volumetric rates Qit) and (l(t) can be
determined from conditions of continuity of fluid flux
and pressure at the boundary between the yielded and
non-yielded zones. One must integrate the equation for
fluid pressure (2) separately inside the yielded and nonyielded zones, and this is done in Appendix A.

In order to find q(t), continuity of radial stress at the

boundary between yielded and non-yielded zones is assumed,
which leads to:
dS [Pl..l lnp (1- lnp ) + 211 (1_p-2)] =

dt

lnp e

= a (R) - P - [2Kr

rw2

(19)

P(Pe -Pw) ]lnp

lnpe

where p = R/rw ' Pc = rJrw , and the relation between Set) and
R(t) is given by equation (13).

SOLUTION FOR SMALL DILATANCY

Near the critical point (Figure 1), the slope of the yield
criterion is small. If we approximate the yield curve by

Now consider an evolving case, beginning from a stable non-

44

SPE 27343

GEILIKMAN, DUSSEAULT, AND DULLIEN

yielded state around the wellbore, and then gradually

decreasing fluid pressure at the wellbore. At some
critical fluid pressure Pc' the yield criterion starts being
fulfilled at r = rw ' From this condition and from
expression (B4) for the radial stress in the non-yielded
zone (Appendix B), one may find:
Pc = cr,(R)

IRz'.

lC
(l-p-2 ) == cr -lC
Pc = cr~ - _"_
1+<X
~
~"

(20)

Despite 11

~rw

because of the obvious and strong inequality: rw2 k. For

this case, an approximation can be deduced from equation
(19):
k

(22)

==

lnp~

dt

Now let us consider how a non-zero dilatancy can change the

general situation presented to this point. The most important
qualitative contribution appears from a porosity-dependent
strength lC(~). Assuming small variations of porosity, in the
yielded zone one may write:
==

lC(~y) + 1lC'1 2<X(l-~;:S(t;

(l + <X)

As can be seen, the maximum radius of disturbance

increases exponentially as a function of drawndown fluid
pressure in the wellbore, .1p == Pc - Pw(00), where Pw(00) is
a final (asymptotic) value of borehole pressure.
An exact solution of the non-linear equation (19) is not
possible at this time in analytical form.
Before
presenting specific results, we will compare by order of
magnitude the two terms on the left-hand side of
equation (19). It is well-known that calculation of
viscosity coefficients for fluid-saturated dense granular
media is a difficult problem, and these coefficients can
only be estimated parametrically12. In our notation:

(24)

where d is a grain diameter, and h is a characteristic

thickness of the fluid layer between grains where viscous
dissipation goes on. Certainly this thickness must be
some function of porosity, h(q, but an explicit
connection is not yet clear for dense granular systems.

45

(26)

Pc -Pw(t) -2lC+J3 p~-Pw(t)

In(R(t)/rw)
lnp~

Inp~

d
11 - ~h

In(R(t)/rw)} ==

The initial stage of growth of the yielded zone depends on

the temporal increment of Pw(t).

lC(~)

(23)

(25)

(21)

PC-Pw(oo)}

2 lC _ 13 P~ - Pw (00)

being far more probable than the reverse:

13 ~ dS(t) {I

As wellbore fluid pressure is continuously decreased, the

yielded zone starts spreading away from wellbore, and
this process is accompanied by sand production. The
yielded zone stops growing and massive sand production
ceases when the right-hand side of equation (19)
becomes zero. From this condition the maximum radius
of the yielded zone can be deduced:
RM = rwexp {

11 k < 1

for Pc 1. These results allow rewriting of expression

(B4) for radial stress at the boundary as follows:
cr,(R) == cr~ -lC" == Pc

(27)

r:; r

i+<i'

After substitution of equation (27) into (14), an explicit

integration may be performed for non-zero <X values. Letting
<X ~ 0, only one term is retained, permitting a new qualitative
insight to demonstrate the role of hardening of fabric in
massive sand production processes. The following equation
is arrived at:

J3~ dS{I_lnp(t)} == Pc-Pw(t) -2lC+

k

dt

lnp~

Inp(t)

(28)

where the last term on the right-hand side arises because of

hardening. In the early stages of the sand production process
this term is small at <X 1. However, this term may
become significantly contributive at later stages, reducing the
magnitude of the maximum radius of the yielded zone, and
therefore reducing the duration of sand production. In any
case, if <X > 0 (porosity-contractile hardening), sand
production terminates at some finite radius of the yielded
zone, thus the yielded zone stops growing.

SAND PRODUCTION AS A VISCOPLASTIC GRANULAR FLOW

The situation changes qualitatively if (X < 0, which

corresponds to porosity-dilatant weakening during
granular flow (as expected for chalk). The volumetric
rate of sand production as a function of yielded zone
radius is plotted in Figure 5 (and schematically, as a
function of time, in Figure 6) according to equation (28).
At some critical magnitude of the dilation coefficient (x,
or more precisely, the control dimensionless parameter

l; =

(X

IK' I (C\>y -C\>n)

(29)

(Pc-Pw)(I-C\>y)

ENHANCEMENT OF FLUID PRODUCTION DUE

TO SIMULTANEOUS SAND PRODUCTION
Enhancement of fluid production under a fixed gradient
of fluid pressure appears because the motion of grains
reduces viscous flow resistance from the granular matrix.
From equations (15) and (A6) we obtain (for (X ~ 0) the
volumetric rate of fluid production:
In (R(t)/rw )
In(r Ir )
~

(30)

where

Q == k (p~ -P w )
o

CONCLUSION
A new, physics-based model of massive sand production has
been proposed. The model accounts for a viscoplastic
rheology of the fluid-saturated granular fabric, arising from
viscous dissipation in the intergranular fluid. This dissipation
prevents appearance of high gradients of granular velocity,
and hence, after surpassing a yield stress, the granular fabric
may flow as a dense suspension (or a "heavy fluid").
If the fluid pressure in the wellbore is lowered below some

growth of the yielded zone, and hence cumulative sand

production, become unstable. Solids production rate
behaves in a non-monotonic manner, increasing along
with the increase of the yielded zone radius. Because of
this, it is possible to consider that particular scenario of
sand production as a catastrophic one.

Q[ = Qo + q(t).

SPE 27343

(31)

11 1n (r/rw )

would have been fluid production if there were no

simultaneous solids production. As can be seen from
equation (30), enhancement of fluid production depends
not only on the current magnitude of solids production,
q(t), but also on the previous history of the process, that
is, on the size of the yielded zone, R(t). Thus, solids
production decreases in time, but the radius increases in
time. As a result of these two competitive tendencies,
fluid production rate enhancement will likely have a
maximum at some intermediate time during the
continuous process of sand production. In order to
quantify possible maximum rates of fluid production, it
is necessary to know the specific dynamics of fluid
pressure in the wellbore, Pw(t).

46

critical magnitude, the shear yield condition becomes

fulfilled, and a yielded and flowing zone starts growing
around the wellbore. Solids production proceeds by a
mechanism which can be called "unravelling of fabric"
(similar to unravelling of knitting) because the motion of the
boundary between intact and flowing zones is in a direction
opposite (away from the wellbore) to the direction of mass
transport (toward the wellbore). Analysis of mass balance at
the boundary shows that this mechanism is possible only if
the porosity of the flowing, yielded zone is higher than
porosity of the non-yielded zone. Thus, as a qualitative
precursor to sand production instabilitity, an expansion of
granular fabric (dilation) must be considered in the regions
approaching instability. It should be noted that expansion of
powder fabric has been observed as a precursor to
fluidization by gas flow 13
The initial material state, expressed by the position on a
porosity-dependent stress diagram (Figure 1), determines if
dilatant weakening or contractile hardening will be the path
followed upon a significant spatiotemporal variation in
stresses. Porosity-contractile hardening along with granular
flow reduces the final radius of the disturbed zone as well as
the duration of solids production. However, if granular flow
is accompanied by dilation, it can lead to a number of
scenarios ranging from finite growth of the yielded zone
(and, hence, of cumulative solid production) to the case of
unstable sand production. In the latter, volumetric rate of
sand production behaves nonmonotonously; it decreases at
early stages of the process, then it exibits a minimum, and
afterwards it begins increasing slowly.
This kind of
nonmonotonous behaviour is reminicsent of the appearance
of density waves which have been observed for dry granular
flow in hoppers14
Flow of granular materials mitigates resistance to fluid flow,
which in turn enhances fluid production rate. Interestingly
enough, enhancement of fluid production rate depends not

SPE 27343

GEILIKMAN, DUSSEAULT, AND DULLIEN

only on the instantaneous solids production rate, but also

on the current size of the disturbed, yielded zone around
wellbore. This makes a great deal of sense, and as a
result of this, the maximum fluid production rate for a
well that is experiencing sand production will not in
general coincide with the maximum solids flux. This
effect may be noticed in the history of some wells in

Engineering Research Council of Canada through a Strategic

Grant to support sand production research, and, in part, by
the Sand Production Mechanisms and Management Project of
the Center for Frontier Engineering Research, Edmonton,
Alberta. We would like to thank the participants of this
project, especially RChalatumyk for interesting discussions.
We are also thankful to RBratli for useful comments.

AlbertalS~

As a final general statement, the model we have

proposed herein has limitations arising because of our
incomplete understanding of the processes in situ and
because of the simplifications made to arrive at an
analytical expression. Nevertheless, the model seems to
qualitatively capture much of the "typical" behaviour
observed in heavy oil wells in Alberta.

NOMENCLATURE
F
k

=
=
=
=
=
=

Yield function for the granular matrix

Permeability
n,y
Indices of intact and yielded zone
Fluid pressure (general)
p
Pc
Reservoir pressure
Pw
WeUbore pressure
q
= Volumetric rate of sand production
Q
= Volumetric rate of fluid production
= Radial distance from the weUbore center
r
rw ' rc = Radius of wellbore, and radius of reservoir
= Radius of yielded zone
R
RM
= Final radius of yielded zone at ex = 0
S
= Cumulative sand production
t
= Time
VI' VS
= Fluid and solid velocity
ex
= Coefficient of dilation
P = Biot's constant
11
= Viscosity coefficient of the skeleton
l(
= Strength function
Jl
= Viscosity of fluid
crr, cro, crc = Radial, tangential, and far-field stress
cj)
= Porosity
<l>(F) = Rheological function

ACKNOWLEDGEMENTS

REFERENCES
1. Bratli, RK. and Risnes, R, "Stability and Failure of Sand
Arches", Soc. Petrol. Eng. J., April 1981, p.236-248.
2. Dusseault, M.B., and Morgenstern, N.R, "Shear Strength
of Athabasca Oil Sands", Can. Geotech. J., v.15, 1978,
p.216-238.
3. Agar, J.G., Morgenstern, N.R, and Scott, J.D., "Shear
Strength and Stress-Strain Behaviour of Athabasca Oil
Sand at Elevated Temperature and Pressure", Can.
Geotech. I, v.24, 1987, p.l-IO.
4. Dusseault, M.B., and Van Domselaar, H.R,
"Unconsolidated Sand Sampling in Canadian and
Venezuelan Oil Sands", Second UNITAR
Conference on the Future of Heavy Crudes and Tar
Sands, Caracas, Venezuela, 1982.
5. Frankel, N.A., Acrivos, A., "On the Viscosity of a
Concentrated Suspension of Solid Spheres", Chern. Eng.,
v.22 1967, p.847-853.
6. Jaeger, lC., and Cook, N.G.W., Fundamentals of
Rock Mechanics, Methuen & Co., 1969.
7. Cruz de la, V., and Spanos, T.J.T., "Thermomechanic
Coupling During Seismic Wave Propagation in a
Porous Media", J. Geophys. Res., v.94, 1989, p.637-642.
8. Perzyna, P., "Fundamental Problems in
Viscoplasticity", in: Advances in Applied Mechanics, v.
9, pp. 243-377, Acad.Press, N.-Y., London, 1966.
9. Landau, L.D., and Lifshits, E.M., Statistical Physics,
part I, Pergamon, 1976.

This work was supported by the Natural Sciences and

47

SPE 27343

SAND PRODUCTION AS A VISCOPLASTIC GRANULAR FLOW

10. Schofield,A., and Wroth, P., Critical State Soil

Mechanics. Mc-Graw-hill, 1968.

py(R) = Pn(R)

This allows us to express fluid rates

solid rate q(t):

11. Drucker,D.C., and Prager, W., "Soil Mechanics and

Plastic Analysis or Limit Design", Q. Appl. Math.,
v.lO, 1952, p.157-162.

On and Qy

(A4)
through the

12. Jenkins, 1.T., and McTigue, D.F., "Transport

Processes in Concentrated Suspensions", in: Two
Phase Flows and Waves, D.D.Joseph and
D.G.Schaeffer eds., Springer, 1990.
13. Rietema, K., The Dynamics of Fine Powders,
Elsevier, 1991.

APPENDIX B. Stress in non-yielded incompressible zone.

14. Baxter, G.W., Behringer, RP., Fagert, T., and
Johnson, G.A., "Pattern Formation in Flowing Sand".
Phys. Rev. Lett., v.62, 1987, p.2825-2828.

We will consider the non-yielded zone as an incompressible

one.
This is a typical and well-known simplifying
assumption7 which allows reduction of the number of
parameters in the formulation. From a physical point of
view, elasticity is not important (at least, qualitatively) for a
transition to a viscoplastic state. The limit of
incompressibility means that bulk modulus tends to infinity
and Poisson's ratio v ~ 1/2 . A stress solution for a fluid
saturated porous cylinder has been found earlier by Risnes,
Bratli, and Horsrud16 Taking the limit of incompressibility in
their formulas, we obtain in our notations:

15. C-FER Report Mechanisms and Management of

Solids Production. Edmonton, 1992.
16. Risnes, R, Bratli, RK., and Horsrud, P., "Sand
Stress Around a Wellbore", Soc. Petrol. Eng. 1.,
December 1982, p.883-898.
APPENDIX A. Fluid flow through two-zone medium:
flowing yielded zone and non-yielded zone.

O',(r) = 0't + [0't -O',(R)]

From Darcy's law (2) and the expressions for solid and
fluid velocity (10), (15) and (16) we can find the fluid
pressure, taking into account the presence of solid flow
in the yielded zone (for a sake of simplicity we take a
limit a ~ 0 ):
p/r) = Pw +

~ [Qy ky

q(t)] In(rlrw)

(AI)

<\>y

O'e(r) = 0't + [0'e -O',(R)]

R2
2

rt -R

[1-(rt Ir)2]

(Bl)

[1 +(rt Ir)2]

(B2)

R2
2

r t -R

where:
(B3)

for rw < r < R, and in non-yielded zone:

p (r)
n

= pt

~Q

- _n In(r Ir)

(A2)

At the boundary with the yielded zone (r = R), the stresses

must obey the yield condition (6): F = O. As a result we find:

for R < r < re At the boundary between yielded and

non-yielded zones, the following conditions for fluid flux
and pressure should be fulfilled:
<\>n vn = <\>y v y I,=R

O',(R) = O'e -

(A3)

48

l-R 2 Ir;
2

1 +aR 2 lrt

{len - a[ O't -p(R)]}

(B4)

GEILIKMAN, DUSSEAULT, A.1'l'D DULLIEN

SPE 27343

Intact zone

yielded zone

.--.--- I

> 10

contraction

dilation
....
.... " .....

< 10

"

R(t)

Fig. 2 Yielded, flowing zone

with moving boundary.

Fig. 1 Porosity-dependent yield curve.

I
I
I
I
I

~'-----

___---+I.....~~.=~2.~~.~t
yielded zone

yielded zone

Intact zone

Intact zone

v..
Tw

R(t)

(l-I)d~
,

R+J1.R

Fig. 4 Solids mass balance at the moving boundary

between yielded and intact zone.

Fig. 3 - Schematic evolution of porosity

for contractile granular flow ( a > 0 ).

49

SPE 27343

SAND PRODUCTION AS A VISCOPLASTIC GRANULAR FLOW

10

0.2

~ =

uIx'i (4), -4>,,)

dilatant weakening
~
5 . 10. 5

=.

(PC -Pll) (1-4>,)

dilatant weakening
1.95 . 10 - 5

~c

=.

'ill

contractile hardening
1;=5-10'5

\10

porosity = const
~=O

R.. I r. = 50).

Fig. 5 Volumetric rate of solid production vs radius of yielded zone (for

q(t)

q(O)

",
2 (Pc-pw)k
1
q =-------

p~

In(t/1:)

'\

~>~c
t /1:

'i

q -

krw(pc-pw)
--"--""::"2- - -

P~RJlln

exp [ -

(RJl/rw)

Fig. 6 Volumetric rate of solid production vs time (schematIcaIIYJ;

q(O)

= { Ip;(O) I k r.2 (IIy.<<II,J I ~ J.1 (1-11 y) }112 ;

'[ ={~ J.1 r.

(IIy<<IIn) I k (PcP.) (1-11 y) }.

50

trw
2

1: Rj{ln (RJl/rw)