Вы находитесь на странице: 1из 7

SPE 86949

Sand Erosion in Cold Heavy-Oil Production


Yalamas T., Nauroy J.-F., Bemer E., Institut Franais du Ptrole and Dormieux L., Garnier D., Laboratoire des Matriaux
et des Structures du Gnie Civil

Copyright 2004, Society of Petroleum Engineers Inc.


This paper was prepared for presentation at the SPE International Thermal Operations and
Heavy Oil Symposium and Western Regional Meeting held in Bakersfield, California, U.S.A.,
1618 March 2004.
This paper was selected for presentation by an SPE Program Committee following review of
information contained in a proposal 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 a proposal of not more than 300
words; illustrations may not be copied. The proposal 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
Ensuring quantitative prediction of eroded sand volume is a
major challenge for companies, which seek to conduct
effective sand management. This is particularly a key point for
Cold Heavy Oil Production With Sand.
IFP has been studying the geomechanical aspects of heavy
oil cold production for several years from both the
experimental and theoretical points of view [Deruyter & al.,
1998], [Nauroy J.-F., 1999]. While experimental studies of sand
erosion are mostly based on linear flow through cylindrical
cells [Tremblay & al., 1995], IFP has designed a special
oedometric cell to simulate a radial flow around a well drilled
in an unconsolidated sand reservoir.
The current device allowed to observe two types of erosion
figure under CT-scan, defined as "cavity" or "spread erosion".
Two main key parameters governing erosion have
been identified.
Dimensional analysis shows that CHOPS in situ conditions
should lead to a "cavity" type erosion. Various theoretical
models describing sand erosion and mixed sand/fluids flow
have been investigated. An algorithm allowing to predict the
formation of a cavity has notably been defined and already
validated for a 1D numerical modeling. A 2D numerical
modeling is in progress.
Introduction
Cold Heavy Oil Production With Sand (CHOPS) is one of the
most effective production method for heavy oils located in
unconsolidated reservoirs, e. g. in Athabasca. In some cases of
conventional oil production from poorly consolidated
reservoirs, Sand Management appears to be more profitable
than total Sand Control. The challenge is now to predict, for a
given well, the ratio of eroded sand to produced oil as a
function of reservoir and production parameters.

IFP has been studying the geomechanical aspects of heavy


oil cold production for several years from both the
experimental and theoretical point of view [Deruyter & al.,
1998], [Nauroy J.-F., 1999]. IFP has notably designed a special
oedometric cell to simulate a radial flow around a well drilled
in an unconsolidated sand reservoir (Fig. 1).
At the center of the cell, a metallic tube perforated by
several holes and going through the sand bed represents the
well and its perforations. A hydraulically loaded piston applies
an axial stress to the sand bed, while its radial strain is zero.
The fluid is injected at the periphery of the cell to ensure a
radial flow. The drawdown pressure (pressure difference
between the periphery and the center of the cell) is controlled
and can be increased to enhance sand production. While sand
is being produced, the sand bed can be observed
under CT-scan.
We will first present a rapid review of the various
experimental results obtained on sand erosion. We will then
focus on IFP radial flow cell and the two types of erosion
figure ("cavity" and "spread erosion") observed under CT-scan
according to two main key parameters.
As dimensional analysis shows that CHOPS in situ
conditions should lead to a "cavity" type erosion, modeling
efforts have been devoted to the formation of a cavity. We will
point out the main encountered difficulties and describe the
retained solutions.
State of the art in experimental study of
sand erosion
When considering the experimental results that have been
obtained in the past years on sand erosion, several types of
experimental device have to be distinguished: cylindrical cells
with axial flow, plane cells with radial flow and plane cells
with plane strains.
Cylindrical cell with axial flow
The Alberta Research Council has conducted various tests,
which consist in setting an axial flow through a cylindrical
cell, first with one perforation and an imposed flow or
pressure gradient [Tremblay B. & al., 1996, 1998], and then with
two perforations [Tremblay B. & al., 2002].
These experiments have lead to the development in the
sand bed of one or two "wormholes" according to the number
of perforations. But the shape of these "wormholes" appears to
be strongly dependent on the geometry and dimensions of the
cell. Indeed, the diameter of the "wormholes" observed with
the two first cells was around 4 cm. While, in the third cell,

which was larger, one "wormhole" reached a 17-cm diameter.


Consequently, these tests do not allow to conclude on the
geometry of the flow paths ensuring the rapid communications
observed in situ between the wells.

SPE 86949

This device has been operated since the end of 2001 and
more than 30 tests have been realised.
Axial stress

Plane Cell with radial flow


During his Ph-D, P. Cerasi has realised tests with a HeleShaw cell [Cerasi, 1996]. This cylindrical cell of small
thickness (around 1 cm) has a hole at its center, which allows
to produce the fluid. The confining pressure is not controlled.
While imposing an increasing flow rate, Cerasi has observed
three flow types: laminar filtration of the porous medium,
erosion instability (some canals free of grain appear), and
porous medium liquefaction. The erosion instability features
could be linked with the "wormhole" assumption.

Sand bed

Well

Reservoir
pressure

Fig. 1: Sand production cell

Plane Cell with plane strains


The sand is disposed in a parallelepipedic cell (height: 600
mm, length: 900 mm, thickness: 20 mm) and saturated with
water. A metallic pipe, placed at the center of the cell with two
diametrically opposed holes, represents a casing and its
perforations. This experimental device has been developed at
IFP, as part of a feasibility study of erosion experimental
modeling, to represent a reservoir with a well exploited in cold
production and constitute. A confining pressure was first
applied followed by a hydraulic gradient, which yielded a flow
of water through the sand bed. Some tests have lead to sand
flow whose characteristics give information on the phenomena
taking place inside the sand bed.
Initial tests with Fontainebleau Sand have lead to an
immediate coming of sand after the drawdown. A model sand
with a less regular grain shape ensuring more concavo-convex
contacts has then been chosen. Sand arrivals interrupted by
periods of free of sand water flow have notably been observed.
This could result from the development of stable archs.
The beginning of sand erosion appears to depend on
hydraulic gradient and confining pressure.
Cylindrical cell with radial flow
Taking into account the interesting results obtained with
the feasibility study cell, IFP has designed a new experimental
device (Fig.1). At the center of the cell, a metallic tube
perforated by several holes and going through the sand bed
represents the well and its perforations. A hydraulically loaded
piston applies an axial stress to the sand bed, while its radial
strain is zero. The fluid is injected at the periphery of the cell
to ensure a radial flow. The drawdown pressure (pressure
difference between the periphery and the center of the cell) is
controlled and can be increased to enhance sand production.
While sand is being produced, the sand bed can be observed
under CT-scan. Very significant improvement have
been made:
the fluid flow through the sand bed is radial,
the confining pressure on the sand bed is
well controlled,
the cell is designed to be opered under CT-scan in
order to get pictures of the sand bed during tests,
one of the key points being that the geometry of this
cell does not favour any erosion geometry.

Principal features of the experimental device


The cell geometry is an oedometric one:
internal diameter of the cell: 30 cm
height of the cell: 20 cm
height of the sand bed: 6.5 cm
The metallic tube representing the well has a 1-cm
diameter. In this tube, one or several holes, which are
3 to 5 mm in diameter, represent the perforations.
These holes could be kept closed or opened.
Three independent porous sectors, located at the
periphery of the cell, allow to inject the interstitial
fluid. For the moment, only water has been used.
When both the sectors and the center holes are open, a
radial flow is created in the sand bed.
A piston is placed at the top of the sand bed in order to
applied a uniform axial stress (confining pressure) on
the sand bed.
A GDS pump is used to control the confining pressure
imposed on the piston.
One buffer cell is used to maintain a steady upstream
pressure.
To collect the sand at the exit of the main cell and to
control the downstream pressure, a receptive cell with
a pressure-regulating valve is used.
A LVDT transducer is placed between the piston and
the cap of the cell to record the sand bed axial strain.
Test process
Setting and consolidation
The sand is disposed in the cell in successive layers up to a
height of 6.5 cm and the obtained sand bed is slightly
confined. The sand is then satured with the pore fluid. Only
water has yet been used. The desired confining and pore
pressures are finally applied.
After the sand bed consolidation, CT-scan pictures of the
sand bed every 3-mm are taken and defined as the initial state
of the sand bed.
The perforations, which have been kept closed during the
consolidation phase, are now opened, the downstream and
upstream pressures being equal.

SPE 86949

700
13

600

21

23

15
20

28

22

500
Pceff (kPa)

Production phase
The downstream pressure is then progressively decreased
to a given value and a radial flow appears between the
injection sectors and the perforations. Depending on the
hydraulic gradient, confining pressure and relative density, we
do or do not observe sand comings.
When sand comings are observed, we stop the hydraulic
flow by closing the perforations and take new CT-scan
pictures of the sand bed.
We repeat this production phase several times for each test.

Cavity

32

30

31

17
27

400
12

300

11

Spread erosion

200
100
0

Preliminary tests
Sand nature influence
In Canadian reservoirs (e.g. Elk Point), the sand is both
very fine (d50<300 m) and tangled. Dusseault and Van
Domselaar (1982) give a porosity of 30 to 32% for Cold Lake
field and 28 to 34% for Athabasca field. They point out the
high number of concavo-convex contacts, which result of sand
dissolution and recrystallization phenomena [Dusseault &

60

65

70

75

80

85

90

95

100

Initial Relative Density (%)

Fig. 2: Prediction of erosion figures

The exact geometry of the concerned area is difficult to


define accurately and is not exactly the same for all the tests.
But it always starts at the perforation and spreads toward
the periphery.

al., 1978].

From a granulometric point of view, this sand is very close


to Fontainebleau sand. We thus first used this sand, which led
to immediate and massive sand coming. We then chose a
rougher model sand showing more concavo-convex contacts
and thus an apparent cohesion.
Perforation diameter
Three different perforation diameters have successively
been tested: 3-mm, 4-mm, and 5-mm, leading to different size
ratio between the sand grains and perforations. With the
smaller perforations, no coming of sand could be observed
denoting a scale effect. The 5-mm diameter perforation is
about 8 times larger than the biggest grains. This value has to
be compared to the in situ size ratio of 1 to 100 between the
sand grains and the perforations, which could not be
reproduced with the cell.
Tests results
All the tests presented here have been realized with our
model sand and a 5-mm diameter perforation.
Erosion schemes
Two erosion schemes have been observed (Fig. 3 & 4):
development of a cavity,
spread erosion (sand coming massively with a drop of
density in one area of the sand bed).
The main governing parameters appear to be the effective
consolidation pressure and the initial relative density of the
sand bed before consolidation (fig 2).
We can notice that the initial relative density varies from
65 to 85 % depending essentially of the setting of the sand bed
(which is done by hand).
Spread erosion
During tests 8, 11, 12, 20, 27 and 28, massive sand coming
has been observed since the first production phase. When the
cell was opened, at the end of the test, we noticed a weak area
of lesser density in the sand bed, which also appears clearly on
CT-scan pictures (Fig. 3).

Test n 27
Weak density area

Cavity close to the periphery

After the first production phase. After the second production phase.
Fig. 3: Erosion localization (test n 27)

Knowing the final mass of produced sand, we tried to


estimate the density of the weak area.
In the case of test 27 (Fig. 3), we obtained a relative
density in the weak area less by 40% than the relative density
after consolidation.
Development of a cavity
Cavity close to the well
For a higher density or a more important confining
pressure, the erosion process is very different. Prior to induce
sand coming under an increasing drawdown, we generally had
to reduce the effective confining pressure.
The sand is no more coming massively. We observe
alternatively sand bursts (around 300 mm3 of weak sand) and
"grain to grain" comings, while CT-Scan pictures show the
formation of a cavity in front of the perforation. The cavity is
first very localized in front of the perforation, but its height
and its diameter increase as water and sand are produced (Fig.
4). The final mass of produced sand is small compared to the
massive sand production associated to spread erosion. As long
as we product water, we keep on producing sand and so
enlarging the cavity. But, if the water flow is stopped, the
cavity is mechanically stable.

SPE 86949

Initial state

State 1
Water produced: 5 L

State 2
Water produced: 11 L

Final state
Water produced: 17,5 L

Fig. 4: Cavity close to the well (test n17)

Cavity within the sand bed


Tests 30 & 31 have finally led to the formation of a cavity,
but have shown a massive sand production at the start. For
these tests, the cavities have not been initiated close to the well
but within the sand bed. In the case of test 30, several
production phases have been conducted and led to a gradual
increase of the cavity (Fig. 5).
These results show the effect of local heterogeneities in the
sand bed density.

Numerical modeling
The mechanisms involved in the erosion of weakly
consolidated sandstone are very complex. They notably
depend on the geomechanical properties of the sand, on the
flow properties of the fluids (oil, water and gas) and on the
nature of the sand/fluid and fluid/fluid interactions. The
modeling aim is to predict sand production rate and its effect
on the oil production rate and the oil recovery ratio.
Three different approaches can be found in the literature:
the development of a cylindrical zone of enhanced
permeability around the well [Dusseault & al., 1994], the
development of wormholes (enhanced permeability channels)
from the well [Tremblay & al., 1999], [Dusseault & al., 1997], or a
homogenized approach considering the development of an
heterogeneous area damaged by sand erosion [Shao & al.,
2002], [Wang & al., 2000].
We chose to distinguish two zones: a zone of intact
reservoir, which has a poroelastic behavior, and a zone of
slurry (mixture of oil and sand), which follows the behavior of
a Poiseuille fluid. No assumption on the geometry of the
boundary between these two zones is made. We then defined
an algorithm, which allows to predict the evolution of the
moving boundary between the two zones. This algorithm has
been successfully programmed with SCILAB for a
1D modeling.
Modeling assumption
The object is first to develop a tool able to predict the
flows of sand and oil during a test in IFP cell. In a second
time, this tool will be transposed to real cases.
This paper deals with a 1D modeling. If we consider an
experimental production phase where sand is coming, the sand
bed could be represented by an external zone of intact sand,
which has a poroelastic behavior, and an internal zone of
slurry (mixture of oil and sand), which follows the behavior of
a Poiseuille fluid (Fig. 6).
"Slurry"

"Intact" sand
P0
First prodution phase

Last prodution phase

Fig. 5: Cavity within the sand bed (test n30)

Conclusion
The conducted tests have clarified the link between erosion
schemes, confining pressure and initial relative density.
According to our results, reservoir conditions would lead to
the development of a "cavity", whose geometry will depend on
the reservoir heterogeneity.
All the tests have been made with water. It would then be
interesting to study the influence of the fluid viscosity on the
erosion schemes.
Another key parameter could be the granulometry of
the sand.

P2
x

V(t)
L(t)

P1

Fig. 6: Representation of the sand bed

We then have to solve an "academic" hydromechanics


calculation on the porous medium (intact part of the reservoir)
[Dormieux L., Bourgeois E., 2002], with two main difficulties:
the boundary between the slurry and the porous
medium evolves as more sand is produced, so the
problem to solve has a geometry varying with time,
the boundary conditions (pressure and displacement)
at the interface are a priori unknown.

SPE 86949

Relation between V, vs and Q

Resolution algorithm

V : slurry velocity
V=ks(P0-P1)/L(t)
[m/s]
v s : skeleton velocity in porous medium
[m/s]
f
Q : filtrationvector in porous medium Q = (v v s )
[m/s]
Mass conservation at the slurry/porous
interface gives:
(v (fL ) L& ) = c(V L& )
for oil

s
(1 )(v ( L ) L& ) = (1 c)(V L& ) for sand
which yields :

We consider an initial state defined by a cavity of length


L(t0)=L with no slurry flow V(t0)=0 and a uniform pressure in
the porous medium P1(t0)=P2(t0)=P0(t0)=1.

medium
L(t)

P0(t0)=P2
V=0
Slurry

Q( L ) + v (sL ) = V

Hydromechanical study of the porous medium


Initial conditions
P = P1
At t = 0 0
L = 0
x=L(t)

x=H

Hydraulics

Q( L ) + v(sL ) = V

p=P2

Mechanics

xx = P0

=0

Concerned
equation
Diffusion
Mechanical
balance

Constitutive equations

( xx ) + f x = 0
Mechanical balance:
x
The volumetric force, f x = k f (x) has to be introduced
due to the 1D modeling to represent the friction stresses at the
porous medium contour ([kf]=[Pa.m-2]). It will not appear in
the 2D modeling.
2
& =k P
Diffusion equation:
p
x 2
where is the volumetric flow of fluid mass and kp is the
hydraulic conductivity of the porous medium.

Sand behavior:
The intact sand is represented by a brittle poroelastic
medium.
d xx = (0 + 2 )d xx bdP

dP
d = M + b xx

with xx C
where:
0 and are Lame coefficients
b is Biot coefficient (b=1)
M is Biot modulus

Porous
medium
, Q, vs

=0
P2

Fig. 7: Initial boundary conditions on the porous medium

The stress states of pressure and stress in the porous


medium are then the following:
xx ( x) = 1

P( x) = 1
( x) = ( x) + P ( x) = 0
xx
eff
We then drop P1 to zero, while P0 and P2 are kept
unchanged in a first time. The pressure gradient between P0
and P1 induces a displacement velocity in the slurry:
P (t ) P1 (t )
V (t ) = k s 0
with V = V e x .
L(t )
We then run a calculation in the porous medium with these
boundary conditions. It is important to note that the fluid
pressure in the porous medium at the interface slurry/porous
medium is not imposed. This pressure is a result of the
poroelastic calculation in the porous medium and is not a
priori equal to P0. However, at equilibrium, these two
pressures have to be identical. This is ensured through an
iterative calculation, which yields the correct P0(t) value.
The calculation conducted with new boundary conditions
gives new pressure and stress distributions in the porous
medium. Several cases are possible:
The effective stress is in every point lower than the
sand strength. The sand bed is then mechanically
steady and we can skip to the next time step.
There is an area in the porous medium where the
effective stress is higher than the sand strength. The
sand bed is then eroded up to the first encountered
point of this area called xrup. Staying in the same time
step, we run a new calculation with updated boundary
conditions:
interface position: Lnew(t)=L(t)+xrup
slurry pressure at the interface: P0(t)=-(xrup)
P (t ) P1
slurry velocity: V (t ) = k s 0
Lnew (t )

SPE 86949

Main results
The numerical results presented here do not pretend to be
quantitative. The object is only to present a
qualitative behavior.
Steady state
The 1D modeling allows to analytically determine an
asymptotic steady state where pressure and stress no more
depend on time. This steady state is physically admissible if
the maximum effective stress in the sand bed is lower than the
sand strength, that is if:
1
tanh[ AH (1 AL )] 1
Ac AH [1 + ( Ak 1) AL ]
where :
Ak =

kp
ks

; AL =

kf
L
C
; Ac =
; AH = H
H
P
0 + 2

Using the above equation to determine material


characteristics and pressure boundary conditions theoretically
leading to a steady state after an erosion of half the length of
the sand bed (AL=1/2), we verified that, fed with those
parameters, our numerical model actually tended towards a
steady state (Fig. 8).
Stress (MPa)
Theoretical curve obtained for
steady state

Compression
Traction
Curve obtained with numerical
program (after 4000 time steps)

Fig. 8: Effective stress in the sand bed

Production curves

Cumulated Oil Volume

Cumulated Eroded Sand


Volume at the interface (*10)
Time
Cumulated Produced Sand
Volume at the well (*10)

Conclusions
The first results obtained with our sand erosion algorithm are
very encouraging. Indeed, when this algorithm will be
introduced in a 3D finite elements program, we will be able to
predict the flows of sand and oil from only a few classical
fluid and sand characteristics. Note particularly that no
assumption on the geometry of the cavity is required.
The calculations could be improved through more accurate
constitutive laws for the sand, the slurry and the oil, especially
in the case of heavy oil.
A 2D modeling is in progress and its results will be
compared with the tests conducted in IFP cell.
Even if the study has initially been initiated for CHOPS,
the results can be adapted to conduct sand management for
conventional oils.
Acknowledgments
We would like to thank Total, and especially Philippe
Marchina, for their support on the experimental work and
INRIA, ENPC and the SCILAB Consortium for the
development of this free scientific software.
References

Intact sand
bed

Slurry

sand volume eroded at the slurry/porous medium interface and


the concentration of oil (resp. sand) in the slurry (Fig. 9).
Taking into account the velocity of the slurry and its
concentration in sand, we can also deduce the volume of sand
effectively produced at the well.

Oil Concentration
in the slurry

Fig. 9: Produced volume in function of time

The model gives at each time step the length of the cavity
and the velocity of the slurry, which allows to determine: the
cumulated oil volume produced at the well, the cumulated

1. Deruyter C., Moulu J.-C., Nauroy J.-F., Renard G., Sarda J.-P.
Bibliographie sur la "Production froide des huiles visqueuses",
ARTEP, 1998.
2. Nauroy, J.F., 1999, "Mcanismes de production massive de sable
dans les huiles lourdes", Rapport IFP n45577.
3. Tremblay, B., Sedgwick, G., and Forshner, K., 1995, "Imaging of
sand production in a horizontal pack by X-ray computed
tomography", SPE 30248.
4. Tremblay B., Sedgwick G. and Forshner K., "Simulation of Cold
Production in Heavy Oil Reservoirs: Wormhole Dynamics",
SPE 35387, 1996.
5. Tremblay B., Sedgwick G. and Don Vu, "CT Imaging of
Wormhole Growth under Solution-gas Drive", SPE
39638, 1998.
6. Tremblay B., Oldakowski K., "Wormhole Growth and interaction
in a large sand pack", SPE 39638, 2002.
7. Cerasi P., "Etude de la croissance dune instabilit drosion dans
un milieu poreux non consolid. Application langiognse",
Thse de luniversit Paris 7-Denis Diderot, 1996.
8. Dusseault M. and Van Domselaar H. "Unconsolidated sand
sampling in canadian and venezuelan oil sands", Heavy Crude
and Tar Sands, 1982.
9. Dusseault M. and Morgerstern N., "Shear strength of Athabasca
oil sands", Canadian Geotechnical Journal 15, 2, 1978.
10. Dusseault M., Dulien F. and Geilikman M., "Sand production as a
viscoplastic granular flow", SPE 27343, 1994.
11. Tremblay B., Yuan J. Y. and Babchin A., "A wormhole network
model of cold production in heavy oil", SPE 54097, 1999.
12. Dusseault M. and Geilikman M., "Dynamics of wormholes and
enhancement of fluid production", 48th An. Tech. Meeting of the
Petroleum Society, 1997.
13. Shao J.F. and Marchina P., "A damage mechanics approach for
the modelling of sand production in heavy oil reservoirs",
SPE/ISRM 78167, 2002.

SPE 86949

14. Wan R.G. and Wang J., "Modelling sand production within a
continuum
mechanics
framework",
CIPC
2000,
Calgary, Alberta.
15. Dormieux L., Bourgeois E., "Introduction la mcanique des
milieux poreux", Presses des Ponts et Chausses, 2002.

Notations
c : oil concentration in the slurry
: porous medium porosity
Q : filtration vector in porous medium (Q= (vf- vs))
[m/s]
f
v : fluid velocity in porous medium
[m/s]
v s : skeleton velocity in porous medium
[m/s]
V : slurry velocity
V=ks(P0-P1)/L(t) [m/s]
ks : hydraulic conductivity in the slurry
[m2Pa-1s-1]
kp : hydraulic conductivity in the porous medium [m2Pa-1s-1]
P1 : well pressure
[Pa]
P2 : reservoir pressure
[Pa]
P0 : slurry pressure at slurry/sand interface
[Pa]
: axial strain
xx : total axial stress
[Pa]
eff : effective axial stress
[Pa]
C : sand strength
[Pa]

Вам также может понравиться