Sei sulla pagina 1di 12

ARMA/NARMS 04-494

Sand Production and Instability Analysis in a Wellbore using a Fully


Coupled Reservoir-Geomechanics Model
J. Wang1, R. G. Wan2, A. Settari3, D. Walters4, and Y. N. Liu5
1,4
Taurus Reservoir Solutions Ltd., 2,5 Department of Civil Engineering, University of Calgary, 3 Department of
Chemical and Petroleum Engineering, University of Calgary
Copyright 2004, ARMA, American Rock Mechanics Association

This paper was prepared for presentation at Gulf Rocks 2004, the 6th North America Rock Mechanics Symposium (NARMS): Rock Mechanics Across Borders and Disciplines, held in Houston,
Texas, June 5 – 9, 2004.
This paper was selected for presentation by a NARMS Program Committee following review of information contained in an abstract submitted earlier by the author(s). Contents of the paper, as
presented, have not been reviewed by ARMA/NARMS and are subject to correction by the author(s). The material, as presented, does not necessarily reflect any position of NARMS, ARMA,
CARMA, SMMR, their officers, or members. Electronic reproduction, distribution, or storage of any part of this paper for commercial purposes without the written consent of ARMA 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 acknowledgement of where and by
whom the paper was presented.

ABSTRACT: This paper presents a fully coupled reservoir-geomechanics model with erosion mechanics to address wellbore
instability phenomena associated with sand production within the framework of mixture theory. A Representative Elementary
Volume (REV) is chosen to comprise of five phases, namely solid grains (s), fluidized solids (fs), oil fluid (f), water (w) and gas
(g). The particle transport and balance equations are written to reflect the interactions among phases in terms of mechanical
stresses and hydrodynamics. Constitutive laws (mass generation law, Darcy's law, and stress-strain relationships) are written to
describe the fundamental behaviour of sand erosion, fluid flow, and deformation of the solid skeleton respectively. Subsequently,
the resulting governing equations are solved numerically using Galerkin’s method with a generic nonlinear Newton-Raphson
iteration scheme. Numerical examples in a typical light oil reservoir are presented to illustrate the capabilities of the proposed
model in the absence of the gas phase. It is found that there is an intimate interaction between sand erosion activity and
deformation of the solid matrix. As erosion activity progresses, porosity increases and in turn degrades the material strength.
Strength degradation leads to an increased propensity for plastic shear failure that further magnifies the erosion activity. An
escalation of plastic shear deformations will inevitably lead to instability with the complete erosion of the sand matrix. The self-
adjusted mechanism enables the model to predict both the volumetric sand production and the propagation of wormholes, and
hence instability phenomena in the wellbore.

surface ground, and the need for disposal of sand in


1. INTRODUCTION
an environmentally acceptable manner. Each year,
The production of formation sand has plagued the these issues cost the oil industry hundreds of
oil and gas industry for decades because of its millions of dollars. Furthermore, sand production
adverse effects on wellbore stability and equipment, and control becomes extremely crucial in offshore
while it has also been proven to be a most effective operations where a very low tolerance to sand
way to increase well productivity. When production is allowed. Hence, it is imperative to
hydrocarbon production occurs from shallow and find an efficient computational model that has the
geologically young (or so-called unconsolidated / predictive capability to assist field operators to
weakly consolidated) formations that have little or understand this unique process. The ultimate goal is
no cementation to hold the sand particles together, to design an economical well-production strategy in
the interaction of fluid pressure and stresses within which sand production and operating costs may be
the porous granular material can lead to the reduced to some extent with maximum hydrocarbon
mechanical failure of the formation and unwanted productivity. It is commonly believed that the
mobilization of sand. It has been reported that 10%- mechanism of sand production can be attributed to
40% sand cuts normally stabilize in time to levels geomechanics and multi-phase or foamy oil effects.
less than 5% in heavy oil reservoirs [1], while an However, modelling such a complex problem is a
average of 40% productivity increase was achieved challenging task since it requires multidisciplinary
through sand management in light oil reservoirs [2]. physics to capture the whole range of material
When sand is produced from reservoir formations, it response from sand flow initiation to fluidization.
can cause a number of problems. These include the
In this paper, sand production is treated as an
instability of wellbores, the erosion of pipes, the
erosion process by which a weakly consolidated
plugging of production liners, the subsidence of
sand matrix is disaggregated near perforations of a
wellbore due to a combination of stress changes and as shown in Figure 1. In reality, the individual
multiphase flow. A fully coupled reservoir- distribution varies discontinuously over space.
geomechanics mathematical model is presented to However, an averaging procedure in the spirit of
account for the effects of multiphase flow and mixture theory is used to homogenize each
geomechanics as well as their interaction in a constituent over the REV volume V such that these
consistent manner. Numerical solutions, restricted individuals are substituted with continuous ones that
to a typical light oil reservoir without the influence fill the whole volume. Each phase discontinuity in
of the gas phase, are sought to examine the basic the REV is represented in terms of its own volume
capabilities of this model. As the wellbore pressure fraction, i.e. saturation and porosity.
is lower than reservoir pressure, the erosion process
free & disolved gas gas (dg+fg)
begins as a result of the degradation of the sand fluidized solids (g) M , ρ , dV
g g g
matrix strength and the drag force imposed by fluid fluid
fluid
solids
pressure gradient. The plastic yielding zones wellbore (f) Mf , ρf , dVf dVv
sand, oil,
develop due to the material degradation (erosion) gas cavity or
fluidized solid
(fs) Mfs , ρfs , dVfs
dV
σy wormhole
and stress re-distribution, while the wormholes or σx REV solid
cavities form and propagate in terms of the (s) Ms , ρs , dVs

increasing porosity values. The volumetric oil and Phase diagram

sand productions are also calculated as a function of Fig. 1 Phase components of a REV
time, stresses, and hydrocarbon flow rate.
For solid phase (s), the density of the solid phase
averaged out over a REV of volume dV can be
2. COUPLED MULTIPHASE FLOW AND written as the homogenized solid density (1-φ)ρs ,
GEOMECHANICS FORMULATION where porosity φ = dV
dV , and ρs is the density of the
V

2.1. Mass balance equations solid phase. The mass conservation requires that
The single-phase formulation describing sand ∂[(1 − φ ) ρ s ]
production in a deforming sand matrix was derived + ∇ ⋅ [(1 − φ ) ρ s u& s ] = − m& (2)
∂t
in a series of publications [3, 4]. It has been shown
to be a promising method for modeling sand where u& s is the absolute velocity of the solid phase
production in terms of matching numerical
boundary, and the negative sign of the right hand
calculations with lab test data, both in heavy and
side refers to a solid loss due to erosion since m& is
light oil conditions [5, 6, 7]. In this paper, an
chosen to be the local rate of solid gain per unit
extension to multiphase sand production model is
volume as seen from the fluidized solid phase.
presented within the same framework of mixture
theory, i.e., a coupled black-oil/geomechanics sand Similarly, for the fluidized solid phase (fs), the mass
production model with erosion mechanics is balance equation can be written, i.e.
∂[φS fs ρ fs ]
proposed to further account for the effects of
multiphase flow of three components (gas, water, + ∇ ⋅ [S fsφρ fs u& fs ] = m& (3)
oil) and their interaction with geomechanics. The ∂t
mass balance equation used in formulating the sand
where the fluidized solid saturation at reservoir
production problem is typically written as [dV ]RC
condition (RC) is S fs = [dVVfs ]RC , u& fs is the absolute
∂ρ
+ ∇ ⋅ (ρu& ) = m& (1) velocity of the fluidized solid phase, and ρ fs is the
∂t
density of the fluidized solid phase.
where state variables ρ, u& are the density and the
absolute velocity respectively, and m& is the source The basic assumptions for flow of oil, water and gas
or sink term to account for the local rate of solid phases follow those used in the classical black-oil
loss or gain per unit volume due to erosion. model [8]. The oil phase (o) continuity equation can
be derived at stock tank condition (STC), i.e.
The fluid/gas saturated sand body is idealized as a
∂[S oφρ o / Bo ]
Representative Elementary Volume (REV) which + ∇ ⋅ [S oφρ o u& o / Bo ] = 0 (4)
comprises of five phases, namely solid grains (s), ∂t
fluidized solids (fs), fluid (f), water (w) and gas (g)
where ρ o = fluid density at stock tank condition, ∇ ⋅ (σ eff − ωPm 1) + b = 0 (7)
[Vo ]RC
So = = oil saturation in reservoir condition
[VV ]RC
where b are body forces per unit volume, and ω is
[V +V ]
(RC), Bo = o[Vo ]dgSTCRC = the formation volume factor, a parameter accounting for the compressibility of
the sand grains. The sign convention adopted is that
and u& o = the absolute velocity of the oil phase.
negative stresses are compressive and fluid
Furthermore, the averaged density of gas can be pressures are always positive. The Kronecker delta
divided into two components: free gas S gφρ g / Bg tensor is given by 1 such that 1ij = δ ij . The
[V ] averaged mixture pressure can be defined as
and dissolved gas ρ gφSo , where S g = [VVg ]RCRC ,
[V ] Pm = So Po + S g Pg + S w Pw (8)
Bg = [Vgg ] RC , ρ g = BRos ρ g , ρ g = the gas density at
STC

[V ]STC 2.3. Discharge for each phase


stock tank condition, and Rs = [Vdgo ]STC . Hence, the
In anticipation for the description of fluid flow
mass balance for the gas phase is written, i.e. through a porous medium, a volume averaged
∂[S gφρ g / Bg + Rs Soφρ g / Bo ] discharge velocity v j (j= o, w, g) of each fluid
∂t (5) phase relative to the solid matrix (Darcy velocity) is
+ ∇ ⋅ [S g φρ g / Bg u& g + Rs Soφρ g / Bo u& o ] = 0 defined as
v j = S jφ (u& j − u& s ) (9)
Since the water is assumed not to partition in either
the hydrocarbon liquid or the gas phase, the mass Both the detachment and fluidization of solid
balance for the water phase is given as particles are a dynamic process that is complex in
∂[S wφρ w / Bw ] nature. It is a future research task to define the
+ ∇ ⋅ [S wφρ w u& w / Bw ] = 0 (6) interaction between fluidized particle and fluid at a
∂t
micro/macro level. However, the discharge of
[V w ]RC [Vw ]RC fluidized solid phase can be related to the average
where S w = [VV ]RC , Bw = [Vo ]STC can be related to a
velocity of mixture, i.e.
function involving water phase pressures.
v fs = S fsφu& fs = S fs ( v m − u& s ) (10)
In the above, the velocities u& o , u& g and u& w are
defined somewhat differently from what is where the average velocity of mixture is
customary done in the multiphase flow literature. v m = So v o + S g v g + S w v w (11)
They are interstitial velocities, based on an
assumption that the flow area Aj for the any phase j Eqs.(2-6) represent local mass balance equations for
is equal to the total pore (void) area AV times the each individual phase. Successively combining
phase saturation Sj. Therefore, the absolute velocity these equations with Eqs.(9-10), the following five
u& j is related to Darcy velocity v j (see Eq. (9) that governing equations are obtained for each phase,
follow in the next section). i.e.
2.2. Equilibrium equation for the solid matrix ∂φ m&
The interaction between the mechanical behaviour − + ∇ ⋅ [(1 − φ ) u& s ] = − (12)
∂t ρ
of a deforming solid matrix and fluid dynamics
must be incorporated into the governing equations ∂[φ ( S fs − 1)]
in order to describe the coupling effects. The + ∇ ⋅ [S fs v m + (1 − φ + S fs ) u& s ] = 0 (13)
∂t
volume-weighted solid velocity u& s provides the
linkage between the fluid and geomechanical v S φu&  ∂  φS 
aspects of the problem. The latter involves a ∇. o + o s  +  o  = 0 (14)
 Bo Bo  ∂t  Bo 
deforming sand skeleton under an effective stress
field σ eff and the volume-averaged pore mixture v S φu&  ∂  φS 
pressure Pm, which must satisfy momentum ∇. w + w s  +  w  = 0 (15)
balance, i.e.  Bw Bw  ∂t  Bw 
λ
 S φu& R S φu& 
∇ ⋅  v g / Bg + g s + Rs v o / Bo + s o s 
 Bg Bo 
(16) α, β
φS g  λmax
∂  φR S
+  s o + =0
∂t  Bo Bg 

2.4. Constitutive laws


γ p / γ max
p

Eqs.(12-16) must be supplemented with constitutive λ0 λ = λ0 +


α + βγ / γ max
p p

laws describing sand particle erosion, fluid flow,


and deformation of the sand matrix. It is commonly
believed that the driving force causing the solid γ p / γ max
p

detachment from the sand matrix is due to


Fig. 2 Relation between erosion and plastic shear strain
hydrodynamics and geomechanics. Based on
phenomenology, a possible functional form of mass
generation can be obtained from the inverse of φ3  φ − φ0 
k = k0 1 or k = k0 exp  A 1 (20)
filtration theory as proposed in refs. [9,10], i.e. (1 − φ ) 2
 1 − φ0 
m& where k 0 and φ0 are constants and A is a fitting
= λ (1 − φ ) S fs v m if v m ≥ v crm
ρs (17) parameter. Turning to solid skeleton deformations, a
=0 if v m < v crm more adequate constitutive law based on plasticity
and incorporating stress dilatancy aspects must be
where v crm is the critical average velocity of mixture used, considering that sand behaviour is mainly
dissipative and dominated by grain slippage,
below which no sand production occurs. The
rearrangement, dilation and destructuration. A yield
erosion coefficient λ provides a length scale that
function F ( σ) based on Mohr-Coulomb is
can be linked to the accumulated plastic strains γ p
considered adequate, while a plastic potential
through the following relationship, i.e. function G must be introduced to calculate plastic
γ p / γ max strains. The flow rule basically defines the plastic
λ = λ (γ p ) = λ0 +
p
(18) strain increment vector as the normal to the plastic
α + βγ p / γ max
p
potential function G and its magnitude determined
where α and β are constants to be determined, from the plastic multiplier Λ , i.e.
while λ0 is a constant, and γ max
p
corresponds to the ∂G
dε p = dΛ
maximum plastic shear strain calculated for the ∂σ eff
entire domain. Eq.(18) gives a hyperbolic variation dΛ ≥ 0 if F ( σ) = 0 and dF = 0 (21)
of λ with respect to normalized plastic shear strain dΛ = 0 if F ( σ) = 0 and dF < 0
γ p / γ max
p
, i.e. with increasing plastic strains, λ
becomes larger which in turn increases erosion In order to describe the internal damage due to the
activity as implied in Eq.(17), see Figure 2. degradation of the porous medium as the erosion
proceeds, it is assumed that the material properties
As for describing fluid flow, semi-empirical Darcy's such as cohesion C and friction angle ϕ drop
law is used to establish the relation between linearly with porosity φ, i.e.
pressure gradient ∇Pj and volumetric fluid mixture
1−φ 1−φ
flux per unit area v j . Thus, C = C0 and ϕ = ϕ 0 (22)
1 − φ0 1 − φ0
k
vj = − ⋅ ∇Pj (19) where C 0 and ϕ 0 are constants. Such a simple
µ
damage law enables this model to account for the
where j= o, w, g, m, k is the effective permeability effect of the degradation of the porous medium due
tensor that can be related to porosity via the to the erosion. More precisely, the plastic
Carman-Kozeny equation or its variant, i.e. deformation of sand matrix increases the erosion
potential according to Eq.(18). In return, the erosion weighting functions equal to interpolation
process also weakens the sand matrix through functions) over the entire domain Ω to above
degradation of its strength properties, see Eq.(22). governing equations in turn together with
In order to complete the derivation of governing discretizing time derivatives by standard finite
equations, we have to define the capillary pressure difference formula and also linearizing time
Pc relationship. The most practical method is to use variables, a system of five non-linear equations is
an empirical correlation relating the capillary obtained with its generic form, i.e.
pressure and phase saturations [8], i.e. Wn +1 ( Vn +1 ) = H n +1 ( Vn ) (25)
Pcow = P0 − Pw = f ( S o , S w )
(23) in which W and H are functionals which originate
Pcog = P0 − Pw = f ( So , S g ) from Eqs.(12-16) and subscripts n and n+1 refer to
time stations t n and tn +1 respectively. Eq.(25)
In conclusion, we have eight equations for solving
eight field unknowns, namely, represents the standard non-linear matricial
equations that can be solved via iterative schemes
φ , S fs , Pj ( j = o, g , w) and u i (i = 1, 2, 3) in the
such as the Newton-Raphson method. If superscript
three-dimensional case. k denotes the iteration number during successive
attempts to final solution, then expanding Eq.(25)
3. STABILIZED FINITE ELEMENT using the Taylor’s series leads to
SOLUTIONS
k
Although the writing of the governing equations is ∂W
Wnk+1 ( Vnk+1 ) + ∆Vnk+1 = H n +1 ( Vnk ) (26)
rather straightforward, both their finite element ∂V n +1
discretization and solution are challenging due to
the nature of the equations and field variables. Hence, the increment of vector V at the end of
Numerical instability arises in terms of node-to- iteration k is
node oscillations. Over the past several years, the
authors developed a generic numerical stabilization [ ] [H
∆Vnk+1 = J kn +1
−1
n +1 ( Vnk ) − Wnk+1 ( Vnk+1 ) ] (27)
scheme - an optimized local mean technique. By
k
enriching main field variables with high gradient in which J n1 is the Jacobian of the linearized
terms, sharp non-local changes can be captured in system, i.e.
the computations to ensure stable solutions. Then, k
the enriched field variables enter into the governing ∂W
J k
n +1 = (28)
equations of physics by way of averaging of the ∂V n +1
field values in the neighbourhood of a continuum
point, see details in [11]. Thereafter, the finite Successive iterations are performed until the
element discretization of the modified governing convergence criteria are satisfied, i.e.
equations is ready to be expressed in terms of
Vnk++11 − Vnk+1 < ε (29)
variables V, i.e. the nodal displacement
ui (i = 1, 2, 3) , phase pressure P j ( j = o, g , w) ,
where Vnk++11 = Vnk+1 + ∆Vnk+1 and ε is a small value.
porosity φ, and fluidized sand saturation S fs .
Hence, the incremental form of the equations to be
V ( x, t ) = N k ( x ) V (t ) (24) solved at the element level emerges as

 [A1] [A2] [A3] [A4] k ∆S kfsn+1 


where V stands for S fsp , φ p , p jp , uip , and N p are  [B1]

[B2] [B3] [B 4]  ∆φnk+1 
respectively fluidized solid saturation, porosity,
fluid pressure, displacement, and interpolation  [C1] [C 2] [C 3] [C 4]  ∆p kn+1  (30)
function at node p, for p=1 to nh , the total number

[D1] [D 2] [D3] [D 4] n+1  ∆u kn+1 
of nodes. It is again recalled that Einstein index [
= X ( S kfsn+1 , φnk+1 , p kn +1 , u kn +1 ) ]
notation is used with repeated indices implying
summation and the index p is dummy. Applying where [X] a vector containing known variables at
Galerkin’s method of weighted residual (with previous iteration, and [A1], [A2],... [D2], [D3] are
sub-matrices pertinent to fluid, solid, fluidized are chosen to be 0.001 and 0.25 respectively. The
solid, and stress-deformation properties [12]. The simulation is conducted as follows. First, the initial
procedures of Newton-Raphson algorithm are listed state of the reservoir is computed based on an oil
in Table 1. From a practical point view, we have to saturation pressure of 27.6 MPa and an external
address properly the various coupling strategies, i.e. stress of 42 MPa is imposed on both wellbore and
decoupling, explicit, and implicit coupling outer boundaries. Then, the stress around wellbore
techniques before proceeding with the fully coupled is changed to a reservoir pressure of 27.6 MPa to
reservoir/geomechanics simulation [13]. simulate the open-hole completion. Finally, a 3
Table 1. Procedures for Newton-Raphson scheme MPa drawdown is applied at three perforations (P1,
P2, and P3) as shown in Figure 3.
1. Set the initial value k=0 and initial values for each variable
k
2. Calculate the Jacobian matrix J n +1 according to Eq.(28) The length of each perforation is 0.25 m with a
0.012m diameter for P2, and a 0.006m diameter for
3. Calculate the right hand side X in Eq.(30)
4. Solve Eq.(30) both P1 and P3. These, in fact, refer to eight
5.Check for convergence perforations for the full well configuration. The
IF: Eq.(29) is satisfied THEN initial porosity and erosion coefficient in the
Go to next time step perforations are set to 0.6 and 3 m-1 respectively to
ELSE account for the disturbance caused by the
Go to : 2 with new trial value for each variable and k=k+1 perforation process, while they are set to 0.25 and 2
ENDIF
m-1 in the remainder part of the reservoir formation.
Finally, the entire finite element grid is comprised
4. NUMERICAL EXAMPLES
of 3840 nodes and 3705 4-nodes elements and the
In the following simulation, a numerical example of time step size used in the analysis is 0.005 day for a
a light oil reservoir in North Sea is examined under total time span of 5 days investigated. Table 2
hydrodynamics and geomechanics, while examples shows the material properties (fluid and
in heavy oil reservoirs can be found in a series of geomechanics) used in the simulation.
publications [5, 6, 7]. In this paper, no gas phase Table 2. Model parameters
effect is presented, given the space restriction. λ0 = 2 or 3 m-1 ρs = 2.7 g/cm3 ρo = 0.8 g/cm3
0.5 K0x = 0.5 Darcy K0y = 0.1Darcy µ = 5 cp
C0 = 6 MPa E = 2 GPa υ = 0.25
ο
ϕ0 = 30 σext = 42MPa P0= 27.6 MPa
0.4 α=0.008 β=0.1

For the purpose of clarity of illustration, the figures


are plotted in the vicinity of the wellbore, within the
0.3
first 1 m, 2 m and 5m as indicated in XY axes
respectively.
0.2 4.1. Deformations and yielding after open-hole
completion and perforations
In order to examine the wellbore instability and
0.1
P3
sand production, it is essential to understand the
P2 open-hole completion and perforation process. The
perforations P1 process is simulated by lowering the initial stresses
0 42 MPa at inner holes to the initial reservoir
0 0.1 0.2 0.3 0.4 0.5
x(m) pressure and the outer ones are kept to initial stress
extends to 5 m conditions after reservoir initialization.
Fig. 3. Mesh layout near wellbore showing perforations.
It is noted that a plastic zone is developed as shown
Figure 3 shows a close-up of the finite element in Figure 4. This is due to the stress re-distribution
mesh representing one quarter of a section of a around wellbore and the existence of a weakened
vertical well of inner radius r0 = 0.1 m with the zone in the perforations (φ0=0.6) during the drilling
outer boundary of the well extending to 5 m. The process. It is critical to capture the developed plastic
initial fluidized sand saturation Sfso and porosity φ0 zones due to drilling and perforation, since the
erosion coefficient λ is linked to plastic shear strain 1

as defined in Eq.(18) - the larger the plastic shear


strains are, the more intensive the erosion activity time=0.3days 0.14
is. This enables the simulator to automatically 0.75 0.13
0.12
capture the disturbance caused by open-hole 0.11
0.11
completion and perforation in terms of the initial 0.10
0.09
values of erosion coefficient and porosity around 0.08

y(m)
0.5 0.07
wellbore and perforations at the beginning of the 0.06
0.05
drawdown. 0.04
0.03
0.02
1 0.01
0.25

P1
0.75 P2
P3
0
0 0.25 0.5 0.75 1
x(m)
Fig. 5 Fluidized sand saturation profile at time t=0.3 day.
y(m)

0.5
2

Plastic yielded zones

0.25 time=0.6days 0.14


1.5 0.13
0.12
0.11
0.11
P1 0.10
P2
0.09
P3
0 0.08
y(m)

0 0.25 0.5 0.75 1 1 0.07


x(m) 0.06
0.05
Fig. 4 Plastic yielded zones developed after open-hole 0.04
completion and perforations (before drawdown). 0.03
0.02
0.01
0.5
4.2. Evolution of fluidized sand saturation
From this section on, we look at the field variable
profiles due to drawdown. Figures 5-7 illustrate the
spatial distribution of the fluidized sand saturation 0
0 0.5 1 1.5 2
Sfs at four different times t=0.3 day, 0.6 day, 2 days x(m)
and 5 days after drawdown. It is noticed that a sharp Fig. 6 Fluidized sand saturation profile at time t=0.6 day.
rise in fluidized sand saturation develops in the 5
region near the perforations P1 and P2 with the
remaining part of the well being at near initial
values of Sfso. The amplification factor for fluidized 4 time=2days 0.14

sand saturation near the perforation, defined as the 0.13


0.12

current saturation value over the initial one, is about 0.11


0.11
0.10
70 times at location P1 for time t=0.3 day, 110 3
0.09
0.08
times at location P2 for time t=0.6 day, and 140
y(m)

0.07
0.06
times at location P3 for time t=5 days respectively. 0.05
2 0.04
These numbers indicate that there is a dramatic 0.03
0.02
increase in the creation of fluidized sand 0.01

corresponding to sand production. In general, an 1


increase in fluidized sand saturation is governed by
the relative rates at which volume of fluidized sand
Vfs and void volume VV are changing, since Sfs = 0
0 1 2 3 4 5
Vfs/VV. This sharp change is due to the physics of x(m)
Fig. 7 Fluidized sand saturation profile at time t=2 days.
the problem described as follows. Initially, erosion
preferentially occurs in the x-direction near
perforation P1 since the horizontal permeability is of wellbore and perforations, very high fluid fluxes
five times greater than the vertical one. As most of prevail, which in turn give way to high fluidized
the sand particles are mobilized to produce a very 2
loose matrix, further erosion takes place in regions
where more sand particles are available. 12.00
11.29
Figure 8 shows a decreased fluidized sand 1.5 time=0.3days
10.57
9.86
saturation profile, which indicates a decline in 9.14
8.43
erosion activity because there is no material left for 7.71
7.00

y(m)
the erosion around wellbore. 1 6.29
5.57
4.86
5 4.14
3.43
2.71
2.00
0.5
4 time=5days 0.14
0.13
0.12
0.11
0.11 0
3 0.10 0 0.5 1 1.5 2
0.09 x(m)
0.08
Fig. 9 Erosion coefficient λ distribution at time t=0.3 days.
y(m)

0.07
0.06
0.05
2 0.04 2
0.03
0.02
0.01
12.00
1 11.29
time=0.6days
1.5 10.57
9.86
9.14
8.43
7.71
0 7.00
0 1 2 3 4 5
y(m)

x(m) 1 6.29
5.57
Fig. 8 Fluidized sand saturation profile at time t= 5 days. 4.86
4.14
3.43

4.3. Evolution of erosion coefficient and cavity 2.71


2.00
0.5
propagation
As defined in Eq.(17), the erosion coefficient is a
function of plastic shear strain. This indicates that
0
most erosion activity is confined and intensified in 0 0.5 1 1.5 2
x(m)
only plastic shearing regions. The larger the plastic Fig. 10 Erosion coefficient λ distribution at time t=0.6 day.
shear is, the more intensive the erosion is. In other
words, the erosion activity aligns itself with the 2

plastic yielded zones where plastic shearing of the


material is most prevalent. Figures 9-11 show the 12.00
11.29
distribution of erosion coefficient λ with time 1.5 time=5days
10.57
9.86
around the wellbore. The erosion activity is most 9.14
8.43
intense around the wellbore and perforations at the 7.71
7.00
very beginning, and then propagates further inside
y(m)

1 6.29
5.57
the perforations where the sand matrix has a weak 4.86
4.14
material strength (initial porosity 0.6), and in the x- 3.43
2.71
direction where the pore pressure depletion is the 0.5
2.00

fastest due to high permeability in x-direction


initially. This is due to increasing erosion activity
taking place as porosity increases and ultimately 0
0 0.5 1 1.5 2
degrades the material strength. These will be x(m)
discussed in later sections. Fig. 11 Erosion coefficient λ distribution at time t=5 days.
Figure 12 shows the initiation of erosion at the sand mass fluxes as dictated by the erosion law, see
perforations at time t=0.3 day. In fact, at the edges Eq.(16). However, the maximum erosion activity
does not start simultaneously at all perforations as 5

shown in Figure 12. In fact, the most intensive time=5days


Porosity

erosion activity follows geomechanically yielded 4


0.77
0.73
0.70
zones and a preferential direction of high flux, i.e. 0.66
0.63
x-direction. Figure 13 shows the coalescence of 0.59
0.56
eroded zones around perforations P1 and P2 into a 3
0.53
0.49

y(m)
ring of loose sand of about 0.5 m in radius. The 0.46
0.42
porosity values approach 0.77 and physically 2 0.39
0.35
correspond to the formation of a cavity and 0.32
0.28
mechanical failure of the wellbore. Figure 14 shows
a snapshot of the fully developed zone of high 1

porosity that is initiated at the perforations, and


which localizes along the plastic yielded zones and 0
0 1 2 3 4 5
high flux regions. x(m)
1 Fig. 14 Porosity profile at time t=5 days.
Porosity
time=0.3days 0.5
0.77
0.73
0.70
0.75 0.66
0.63
0.59 0.4
0.56
0.53
0.49
y(m)

0.5 0.46 time=0.3days


0.42
0.3
0.39
0.35
y(m)

0.32
0.28
0.25
0.2

0 0.1
0 0.25 0.5 0.75 1 P1
x(m) P2

Fig. 12 Porosity profile at time t=0.3 day.


P3
0
0 0.1 0.2 0.3 0.4 0.5
2 x(m)
Porosity
Fig. 15 Fluid flux profile at time t=0.3 days.
time=0.6days
0.77
0.73 0.5
0.70
1.5 0.66
0.63
0.59
0.56
0.53 0.4
0.49
y(m)

1 0.46
0.42
0.39 time=0.6days
0.35
0.3
0.32
0.28
y(m)

0.5

0.2

0
0 0.5 1 1.5 2
x(m) 0.1
P1
Fig. 13 Porosity profile at time t=0.6 day. P2

P3
4.4. Fluid flux and pressure distribution 0
0 0.1 0.2 0.3 0.4 0.5
As the cavity enlarges, the permeability of the x(m)
reservoir increases since it is a function of porosity Fig. 16 Fluid flux profile at time t=0.6 days.
in Eq.(19). The gradually increased permeability
the high fluid flux dominates in three perforations in
enhances the well productivity. It is expected that
Figure 15 at the beginning of drawdown. Then, the
direction of large fluid fluxes shows a bias towards 4.5. Displacements and stresses
high porosity regions as shown in Figure 16, i.e. In this section, we look at the plastic shear strain
mostly x-direction in anisotropic permeability case. and stresses distribution in the well. The pressure
It is also worth to mention that the erosion process induced drag forces develop excessive plastic shear
increases the fluid flux by degrading the sand strains around perforations in both x- and y-
matrix where more regions progressively yield direction (maximum value is about 9% after 5 days
plastically due to the high fluid flux and stress in Figure 19). It is also noted that the material
redistribution. Figure 17 shows an increased flux strength parameters, i.e. cohesion C and friction
region around the wellbore at time t=5 days. angle ϕ follow the same distribution as that of
0.5 porosity with time since they are defined as a linear
function of porosity in Eq.(22).
2
0.4
0.090
0.086
0.081
time=5days time=5days 0.077
0.3 0.073
1.5
0.069
0.064
y(m)

0.060
0.056
0.051
0.2 0.047

y(m)
0.043
1
0.039
0.034
0.030
0.026
0.1 0.022
P1
0.017
P2 0.013
0.5
0.009
0.004
P3
0 0.003
0 0.1 0.2 0.3 0.4 0.5 0.001
x(m) 0.000
0.000
Fig. 17 Fluid flux profile at time t=5 days.
0
0 0.5 1 1.5 2
x(m)
Due to the initial anisotropic permeability Fig. 19 Plastic shear strain distribution at time t=5 days.
conditions, the dissipation of fluid pressures around
the well also occurs in regions of high 5
(Pa)
permeabilities, i.e. x-direction. As sand is being -7.00E+06
-7.53E+06
produced, the fluid pressure slowly depletes more -8.05E+06
-8.58E+06
4
from initial values of 27.6 MPa on the outside -9.11E+06
-9.63E+06
boundary to 24.5 MPa than at perforations P1, P2, time= 5 days -1.02E+07
-1.07E+07
and P3 around the wellbore, as shown in Figure 18. 3 -1.12E+07
-1.17E+07
-1.23E+07
y(m)

-1.28E+07
-1.33E+07
-1.38E+07
5 2 -1.44E+07
-1.49E+07
-1.54E+07
Time=5days (Pa) -1.59E+07
2.74E+07 -1.65E+07
4 2.72E+07 1 -1.70E+07
2.70E+07
2.69E+07
2.67E+07
2.65E+07
3 2.63E+07 0
2.61E+07 0 1 2 3 4 5
2.59E+07 x(m)
y(m)

Fig. 20 Effective stress σxx at time t=5 days.


2.57E+07
2.55E+07
2.54E+07
2 2.52E+07
2.50E+07
2.48E+07
Considering the wellbore stability, it is very
important to look at the stress distribution after sand
1
production. Figures 20-22 show the distribution of
effective stresses σxx, σyy, τxy at 5 days after
0
0 1 2 3 4 5
drawdown. Due to fluid pressure reduction through
x(m) three perforations, drag forces are imposed upon
Fig. 18 Pore pressure distribution at time t=5 days.
three perforations, causing a reduced stress σxx in P3
whereas an increased stress σxx around P1 in Figure qoil = ∫∫ v f dS ; qsand = ∫∫ S fs v f dS (31)
20. Also, the stress σyy is reduced in P1 and S S

increased around P3, as shown in Figure 21. Figure Figure 23 gives both the oil and sand rates over the
22 shows the tangential stress profile distribution. time of fluid drawdown. We observe that the sand
The high stress values indicate a highly sheared production rate rapidly increases in an initial phase
zone. Depending on the re-distribution of pore to reach a peak value in approximately 0.5 day.
pressure and stress during erosion, the high shear During this time period, the oil rate gradually
stress zone shifts and grows, which in turn causes increases as well. Then, this phase is followed by a
the evolution of plastic shear yielded zones. decline in sand production rate corresponding to the
5
(Pa)
decrease in availability of sand grains. However, the
-7.00E+06
-7.53E+06
oil rate continues to increase given the enhancement
4
-8.05E+06
-8.58E+06
in permeability of the reservoir induced by sand
-9.11E+06
-9.63E+06
production. This trend is also observed in oilwells
time=5 days -1.02E+07
-1.07E+07 under sand production.
-1.12E+07
3
-1.17E+07 12000 1200
-1.23E+07
y(m)

-1.28E+07

sand rate (kg/day/m)


-1.33E+07 10000 1000

oil rate (kg/day/m)


-1.38E+07
2 -1.44E+07 8000 800
-1.49E+07
-1.54E+07
-1.59E+07 6000 600
-1.65E+07
1 -1.70E+07
4000 oil rate 400
sand rate
2000 200
0
0 1 2 3 4 5 0 0
x(m)
0 1 2 3 4 5 6
Fig. 21 Effective stress σyy at time t=5 days.
time (days)
5 Fig. 23 Oil and sand rate history at anisotropic permeability
(Pa)
3.00E+06
2.84E+06
conditions.
2.69E+06
4 2.53E+06
2.38E+06
25000 3000
2.22E+06
2500

sand rate (kg/day/m)


time=5 days 2.07E+06
20000
oil rate (kg/day/m)

1.91E+06
1.76E+06
3 2000
1.60E+06
1.45E+06 15000 oil rate
y(m)

1.29E+06 1500
1.14E+06 sand rate
9.82E+05 10000
2 8.26E+05 1000
6.71E+05
5.16E+05 5000 500
3.61E+05
2.05E+05
1 5.00E+04 0 0
0 1 2 3 4 5 6

0
time (days)
0 1 2 3 4 5
x(m)
Fig. 22 Effective tangential stress τxy at time t=5 days. Fig. 24 Oil and sand rate history at isotropic permeability
conditions.
4.6. Volumetric sand production and oil rates
In the previous sections, detailed spatial As a comparison, an initial isotropic permeability
distributions of governing field variables with time case is also computed with kx0=ky0=0.5 Darcies. As
were discussed and the analysis revealed local expected, more sand and higher oil rates are
phenomena during sand production. From an obtained as larger initial reservoir permeability
engineering point of view, we would be interested prevails in y-direction, see Figure 24. The same
in examining the total oil and volumetric sand peak value of fluidized sand saturation is calculated,
production rates as integrated over the total but a smoother decline curve of sand rate is
perforation area S (P1, P2, and P3) of the wellbore. obtained in isotropic case, since there is no erosion
Hence, lag due to anisotropic permeability conditions.
5. CONCLUSIONS Journal of Canadian Petroleum Technology. 41:4, 46–
52.
A fully coupled reservoir/geomechanics numerical
4. Wan, R.G. and J. Wang 2004. Analysis of sand
model is presented based on an extension of a production in unconsolidated oil sand using a coupled
theoretical and numerical model that the authors erosional-stress-deformation model. Journal of
have developed in the past to address sand Canadian Petroleum Technology. 43:2, 47–53.
production as an erosion problem coupled with 5. Wan, R.G. and J. Wang: 2002. A Coupled Stress-
hydro- and geo-mechanical effects. This is done Deformation Model for Sand Production using
within the framework of mixture theory in which Streamline Upwind Finite Elements. In Proceedings of
mechanics and transport equations are written for the Eighth International Symposium on Numerical
Models in Geomechanics – NUMOG VIII, Rome, Italy,
each of the concerned phases, i.e. solid, fluid (oil,
10-12 April, 2002, eds. Pande & Pietruszczak, 301–
water), gas, and fluidized solid. 309. A. A. Balkema, Rotterdam. ISBN 90 5809 359 X
Leaving aside gas-related issues, it is found that 6. Wan, R.G. and J. Wang. 2004. Modelling of sand
sand production is a function of stress, time, and production and wormhole propagation in an oil
fluid rate. Sand erosion activity is strongly linked to saturated sand pack using stabilized finite element
methods. Journal of Canadian Petroleum Technology.
geomechanics and there is an intimate interaction
43:4, 46–53.
between sand erosion activity and deformation of
the solid matrix. As the erosion activity progresses, 7. Wan, R. G. and J. Wang. 2003. Modeling Sand
Production and Erosion Growth under Combined Axial
porosity increases and in turn degrades the material and Radial Flow. SPE International Thermal
strength. Strength degradation leads to an increased Operations and Heavy Oil Symposium and
propensity for plastic shear failure that further International Horizontal Well Technology Conference
magnifies the erosion activity. An escalation of SPE 80139. Calgary, Canada, 4–7 November 2002.
plastic shear deformations will inevitably lead to 8. Aziz, K, and A. Settari. 1979. Petroleum reservoir
wellbore instability with the complete erosion of the simulation. London. Elservier Applied Sci.
sand matrix. The self-adjusted mechanism enables 9. Vardoulakis, M. Stavropoulou and P. Papanastasiou.
the model to predict both the volumetric sand 1996. Hydromechanical aspects of the sand production
production and the propagation of wormholes. problem. Transport in Porous Media. 22, 225-244.

The multiphase results including gas phase will be 10. M. Stavropoulou, P. Papanastasiou and I. Vardoulakis.
1998. Coupled wellbore erosion and stability analysis.
presented in a forthcoming paper. The proposed Int. J. Numer. Anal. Methods Geomech. 22, 749-769
model can be used for wellbore stability analysis
and design in open-hole completions, perforation 11. Wang, J. and R.G. Wan. 2004. Computation of Sand
Fluidization Phenomena using Stabilized Finite
pattern design, as well as volumetric sand prediction Elements, Finite Elements in Analysis and Design (in
at different pumping strategies in terms of press).
optimization of the hydrocarbon production.
12. Wang, J. 2003. Mathematical and numerical modeling
of sand production as a coupled geomechanics-
6. ACKNOWLEDGEMENTS hydrodynamics problem. Calgary. (PH. D. dissertation)
The authors wish to express their sincere gratitude 13. Settari, A. and D. A. Walters. 2001. Advances in
for funding provided by Alberta Ingenuity Fund coupled geomechanical and reservoir modeling with
(AIF) and the National Science and Engineering applications to reservoir compaction. SPE Journal. 9:
334–342.
Research Council of Canada (NSERC).

REFERENCES
1. Tremblay, B., G. Sedgwick, and D. Vu. 1999. CT
imaging of wormhole growth under solution – gas
drive. SPE Reservoir Journal. 2: 1, 37–45.
2. Papamichos, E. and E. M. Malmanger. 2001. A sand
erosion model for volumetric sand predictions in a
north sea reservoir. SPE Reservoir Evaluation and
Engineering. 44–50.
3. Wan, R.G. and J. Wang. 2002. Modelling sand
production within a continuum mechanics framework.

Potrebbero piacerti anche