Sei sulla pagina 1di 7

SPE 28041

Results from reservoir depressurisation simulation incorporating phenomena revealed


in a reactive reservoir analogue
B. G. D. Smart. D. H. Tehrani, P.w. H. Olden & M. Jin, Department of Petroleum Engineering, Heriot-Watt
University, Edinburgh, UK

Copyright 1994. Society of Petroleum Engineers

This paper was prepared for presentation at the 1994 Eurock SPEIISRM Rock Mechanics in
Petroleum Engineering Conference held in Delft, The Netherlands, 29-31 August 1994.

ABSTRACT
Field observations of rock mechanical phenomena made in reactive analogues are presented, along with the
output from coupled simulation models of reservoirs incorporating these phenomena. The models
demonstrate that fault activation occurs during depressurisation, altering conductivities and hence
controlling the flow of fluids towards the producing wells.

Als reaktive Analoge gemachte Feldbeobachtungen gesteinsmechanischer Phanomene werden zusammen mit
dem Output aus gekoppelten Simulationsmodellen von Speichern, die diese Phanomene aufweisen,
vorgestellt. Die Modelle demonstrieren das Auftreten von Fehleraktivierung und Scherung der
Ablagerungsebene wahrend einer Druckminderung, wodurch Leitfahigkeiten und damit auch die Kontrolle
des Flusses von FIUssigkeiten zum Produktionsbohrloch geandert werden.

Les phenomenes rocheux mecaniques observes sur Ie terrain dans des systemes analogiques reactifs sont
presentes accompagnes des sorties obtenues de modeles couples de simulation de reservoirs presentant ces
phenomenes. Les modeles demontrent que I' activation de faille et Ie cisaillement de plan de stratification se
produisent a la depressurisation donnat des modifications de conductivite affectant l'ecoulement des fluides
vers Ie puits de production.

1. INTRODUCTION 2. REACTNE RESERVOIR STUDIES


The development and application of petroleum 2.1 The Concept of the Reactive Reservoir
engineering rock mechanics is hampered by the
remoteness of the rock mass, i.e the reservoir, being Reservoir characterisation studies may utilise data
engineered. This may be partially overcome by retrieved from outcrops. Such studies utilise passive
finding suitably stressed analogues of reservoirs and reservoir analogues. In real reservoirs at depth, the
translating observations made of rock deformation reservoir is stressed in a manner which probably
processes in the stressed analogue to the reservoir reflects the modifications imposed by basement
environment. To differentiate between the outcrop structures and reservoir rock types on regional
analogues used in reservoir description studies and tectonic trends. The strata comprising the reservoir
these stressed analogues, the former have been have reacted in geological time to the in situ stress
called passive anlogues, and the latter reactive state by developing continuous ground strains and
analogues. structural features such as faults, joints and bedding
plane parallel shear zones. Despite these
These rock deformation processes may then be equilibrating phenomena, in situ stress fields remain
transposed to the reservoir, using coupled fluid distorted and non-hydrostatic. Through the
flow/rock mechanical models. This is demonstrated expression of these types of reaction, changes in the
for a hypothetical reservoir subjected to in situ stress state during geological time will have
depressurisation. influenced the migration of hydrocarbon through the
formation, and hence the current location of
reserves. It is further argued that production -
induced reduction in pore pressure and the
associated changes in effective in situ stress state

155
can cause compaction, reactivating faults and These observations suggest that subsiding strata
bedding plane parallel shear zones. During these accomodate changes in stress and the asscoiated
phases of activation, shear on the discontinuities strains by allowing shear to occur on certain
sometimes creates permeability barriers, although in "dominant " bedding or parting planes. The
some other situations high permeability conduits distribution of these parting planes within the
may be produced. The development of the sequence of strata that comprise the reactive
understanding of these geomechanical phenomena analogue, and hence the reservoir, characterise the
requires extensive access to a reactive reservoir manner in which strata deform. In particular the
analogue. A reactive analogue is defined here as an aggregated flexural stiffness of the strata is
appropriate sedimentary analogue of the reservoir determined by the distance between dominant
type under study, with the additional qualifications partings, and subsidence-induced fractures originate
that it lies at depth and is therefore stressed. The in relative tension at the partings and migrate
analogue is likely also to contain significant downwards.
structural features. Furthermore sufficient access
should be provided to that analogue to enable the In addition to these observations, anisotropic
stress state and the location and properties of horizontal stress fields are regularly detected in
discontinuities to be evaluated. coalmines, and the activation of faults has been
associated with the redistribution of stresses around
Access to the reactive analogue may be provided mine excavations.( Hasegawa, Wetmiller and
through a large number of boreholes, or more Gendzwill, 1989).
comprehensively by mining. In particular coal mines
are reactive analogues for shaly reservoirs. Awareness of these phenomena has been utilised in
configuring the stress sensitive model described
below, which adresses the simulation of fault
2.2 Rock Deformation Processes Observed in activation by production.
Reactive Analogues
Given the relative ease of access, quantitative and 3. STRESS SENSITIVE SIMULATION
qualitataive observations have been made in the
reactive analogue, revealing quantitative data about 3.1 The Simulation Process and Model
the activation of joint sets, faults, bedding plane Configurations
parallel shear and in situ stress state. The data
presented here focusses on one phenomenon A Brent-like reservoir with a fault was chosen to
substantiallly ignored but fundamental to the manner examine the likely impact of stress sensitive
in which strains associated with depressurisation- simualtion. This was constructed with the
induced compaction is absorbed by the layered background knowledge that faults may change
structure of the typical shaly reservoir, i.e. bedding transmissibility during production (Bayat and
plane parallel shear. Coal mine strata deformation Tehrani 1985)
processes are represented in Figure I, which is a
compilation of measurements and observations Stress sensitivity was introduced into the reservoir
made at several sites as the strata reacted to the simualtion by coupling a finite element stress
passage of a longwall coal face at the right-hand side analysis programme (ANSYS) with a reservoir
of the figure. simulator. (ECLIPSE). ANSYS predicts the
mechanical effects, including changes in
Measurements of change in strata dip in permeability, caused by production-induced
region A at sites in the South Wales and increases in effective stress. The modified
Scottish Coalfields. ( Isaac, Smart and permeabilites are used as input data for ECLIPSE,
Roberts 1982, Smart and Haley 1987) which relates production to permeability and ouputs
pore pressures. Thus effective stresses are
Observations made on the influence of determined for input to ANSYS. The coupling of
parting planes and strata thickness on the ANSYS and ECLIPSE is demonstrated in Figure 2.
caving process at sites in the UK and
Australia. ( Smart and Aziz, 1986 ) For the purpose of this discussion the descriptions of
model configurations can be separated into the
Observation of shear on parting planes details of the ANSYS stress analysis model and the
along the line CD at sites in the UK and ECLIPSE reservoir simulation model. The ANSYS
Australia (RowlandsI992, Smart 1992) model was also utilised on its own to examine the
sensitivity of fault re-activation to various
Observation of tensile fractures, initiating parameters and together with ECLIPSE in fully
on parting planes in the region E, at sites in coupled production scenarios.
the UK ( Holland and Smart 1990)

156
3.1.1 The ANSYS Rock Mechanical Model 3.2 Reservoir Simulation Model
The mesh geometry of the model was set up so that The basic reservoir simulation data for the ECLIPSE
basic model parameters could be easily changed. model was taken from published literature ( Struijk
These included:- and Green, 1991) and is shown in Table 2. For the
purpose of the modelling programme all data was
1. Overall extent of the model converted into metric units so that it was possible to
2. Dip angle of beds read across pressure values from ECLIPSE (bar) to
3. Dip angle of the fault ANSYS (MPa) with only a factor of 10 conversion
4. Thickness and number of beds through between them. Expanded reservoir fluid property
reservoir data was produced to construct the simulation
5. Offset thickness between faulted halves of model, and in the absence of actual laboratory data,
reservoir. some hypothetical saturation functions were used.
The initial model was taken to extend from -1000 m
to + 1000 m in the X -direction, -750 m to +750 m in 3.3 Fault Reactivating Sensitivity Study
the Z-direction and 0 m to 100m in the Y -direction,
measured at the centre of the fault (ECLIPSE axes In order to gain an insight into the likely behaviour
convention). Subsequently the X-direction range of the ANSYS rock mechanical model to
was increased to -1400 m to +1400 m. In this depressurisation for various parameters, without the
configuration the ANSYS model comprised 1214 more time-consuming inclusion of an ECLIPSE
elements of which 660 were two-dimensional solid simulation, a sensitivity study was carried out on the
and 554 two-dimensional interface elements. The ANSYS model alone.
model was taken to be formulated in plane strain The parameters that the sensitivity study varied
which was thought to be applicable to the Brent were:-
field structure in the north-south direction. An - Fault angle: 45, 60 and 75.
example of the model mesh is shown in Figure 3. - In situ stress state,with the vertical stress
being held constant at 65MPa and
The rock mechanical model was assumed to be the horizontal stress varied at 45,50 and
composed of two different linear elastic materials, 55MPa
their mechanical properties as listed in Table 1. In - Coefficient of friction on the fault - 0.1 to
addition, the potential for bedding plane parallel 0.5
shear was accommodated by incorporating nterfaces In this study both sides of the reservoir were
between strata. The coefficient of friction on the depressurised by equal amounts, pressure being
fault and bedding layer interfaces was varied in the dropped until activation of the fault was detected for
range 0.1 to 0.5. a given stress state and coefficient of friction.
Typical results for fault slip are shown in Figure 6
while the results for critical combinations of all
3.1.2 Coupled Modelling Features three parameters are shown in Figure 7. Note that
the magnitude of fault slip mirrors the throw in the
A number of modelling features were also built into fault, originating at the fault centre and moving
the rock mechanical model so that it could represent outwards.
the effect of pore pressure change on the rock
matrix. These included a stress I permeability
relationship for the rock matrix as shown in Figure 3.4 Fully Coupled Production Case
4. Although this is not determined for the Brent
reservoir it is based on published data and laboratory The ECLIPSE model was run in a fully coupled
measurements (Smart, Somerville and Hamilton, production scenario with the ANSYS model for the
1989). An important feature of the curve is that the following case:-.
permeability ratio klko for the rock changes from
1.0 to 0.5 as the mean effective stress increases from The configuration of the ECLIPSE model was:-
Oto 100 MPa.
-wac and GOC located far below and above
The sealing characteristics of the fault were based respectively, the reservoir so that primarily only
on a hypothetical relative shear displacement - oil was produced, in a depletion type of
transmissibility functions as shown in Figure 5. Two recovery.
different characteristics were modelled - one in
which the fault switched from non-seal to seal and The configuration of the ANSYS model was:-
the other in which it switched from seal to non-seal
as relative shear displacement across the fault was -Permeability of each reservoir block a function
activated with reservoir depressurisaton. of effective stress on the block.

157
-Transmissibility across the fault a function of Region 2 it can be seen that this starts to flow at 150
the shear displacement on the fault which can days as the sealing characteristcs of the fault begin
either change from non-sealing to seal or vice to change from seal to non-seal. This has the effect
versa. of causing an increase in well bottom hole pressure
thereby maintaining the oil production rate at 2000
An ECLIPSE only simulation (ie. ignoring the rock- m 3 /d for an extra 73 days compared to the non-seal
mechanical effects apart from compressibility) was to seal case before the producing well control
carried out with fixed rock matrix permeabilities for changes to BHP. The Eclipse only simulation here
each reservoir block and fixed fault has the shortest duration because the well is only
transmissibilities through every time step against producing from Region 1 of the reservoir since the
which the ANSYS coupled simulation could be rest of the reservoir oil is trapped in Region 2 by the
compared. sealing fault. These hypothetical simulations point
to the importance of rock mechanical aspects of
Figure 6 shows the typical development of shear reservoir performance prediction that could have a
displacement along the fault in the rock mechanical significant impact on decisions regarding the
model with time. development of a field.
For the case with fault transmissibility changing
from non-seal to sealing, the coupled model 4.CONCLUSIONS
simulation predicts economic oil production lasting
340 days before the producer is shut. The A stress-sensitive fluid flow simulator which
comparative Eclipse only (with no rock mechanics requires minimal manual intervention has been
input) simulation lasts 500 days with a constructed by coupling ECLIPSE with ANSYS
corresponding increase in field total oil production. via programs which manipulate input/output data.
These results are presented together in Figure 8
which show plots of field oil production rate, total The stress-sensitive simulation of a fault in a Brent-
oil production, oil in place for Region 2 (see Figure like reservoir suggests that the fault can be re-
3) and oil producer well bottom hole pressure. activated by depressurisation. The amount of
Comparison of the results indicate that if the rock depressurisation required to cause the re-activation
mechanical effects are ignored in this example, is a function of the in situ stress state, fault angle
Eclipse could produce misleadingly optimistic and the coefficient of friction on the fault.
results.
For the simplified Brent-like situation modelled,
The differences in the simulations may be accounted accommodation of a decrease in fault
for by the fact that as the transmissiblity of the fault transmissibility which is related to shear
is reduced as a pressure differential develops across displacement in the reservoir leads to a significant
it. This leads to a further increase in shear difference in simulated reservoir performance
displacement on the fault and further decrease in compared to that predicted by conventional non
transmisibility, with associated reduction in oil stress sensitive simulation for the last stages of
production rate which takes place sooner together reservoir life.
with reduced overall total oil production because oil
becomes "trapped" in Region 2. The results from the parametric study suggest that
sensitivity of fault re-activation, (and hence
For the case in which fault transmissibility changes production), to stress is reservoir specific, requiring
from seal to non-sealing the coupled model dedicated studies.
simulation predicts economic oil production lasting
440 days before the producer is shut. In contrast to
the situation above, the Eclipse only simulation 5. ACKNOWLEDGE~NTS
lasts less time at 234 days ie. this is the reverse of
the previous simulation and Eclipse is giving The authors gratefully acknowledge the sponsorship
pessimistic results. Again for the longer lasting case provided by the DTI for the simulation work
there is a corresponding increase in field total oil presented.
production. The results of these simulations are
presented in Figure 9. From the graph of well
bottom hole pressure (WBHP) vs time it can be seen 6. REFERENCES
that where the well is producing from Region 1
only and Region 2 is isolated by the fault, the 1. Isaac A.K., Smart B.G.D., Roberts D." The Re-
pressure difference across the fault causes the fault use of Gateroads serving Retreat Faces at Betws
plane to open up after 150 days of production. The Colliery" The Mining Engineer June 1982727-738.
resulting flow of oil from Region 2 to Region 1
causes the pressure in Region 1 to build up. This 2. Smart B.G.D, Haley S.M., " Further Development
gives the producing well longer life and more of the Roof Strata Tilt Concept for Pack Design and
recovery. From the plot of reservoir oil in place for

158
the Estimation of Stress Development in a Caved Table 2: Basic reservoir simulation data
Waste" Mining Science and Technology, 5 (1987)
121-130. Trap
Type Unconformity: tilted fault block

3.Smart B.G.D., Aziz N.!., "The Influence of Depth to crest


Lowest closing contour
25'2 m (8240 ft) TVSS
2835 m (9300 ft) TVSS
Caving in the Hirst and Bulli Seams on Powered Gas/oil contact 2609 m (8560 ft) TVSS
Oil/water contact 2755 m (9040 ft) TVSS
Supports" Ground Movement and Control related to Gas column 98 m (320 ft)
Coal Mining Symposium AusIMM Dlawara Branch Oil column '46 m (480 ft)

1986 Pay zone


Formation Brent Group
Age Middle Jurassic
4. Rowlands D., "Surface Subsidence Associated Gross thickness (average; range) 247 m (8'0 It): 238259 m (780-850 ft)

with Partiel Extraction: An Australian Case Study" Porosity (average: range)


Permeability (average: range)
21%; 16-28%
650 mD: , 0-6000 mD
Effects of Geomechanics on Mine Design Balkema
Hydrocarbons
1992. Oil gravity 876 kg/m3 (54.7 Ibmlf'3)
Oil type Low sulphur light crude
Gas gravity 0.74
5. Smart B.G.D., "Controls Exerted by Dominant Saturation pressure 298-397 barsa (4326-5750 psia)
Parting Planes over the Deformation of Tabular Gasioil raho (average)
Condensate yield
0.281 sm3/sm3 (1.58 Kscf/stb)
0.775 rm3/sm3 (138 bbllMscf)
Deposits" Effects of Geomechanics on Mine Design Formation volume factor (average) 1.80 rm3/sm3 (1.80 bbl/stb)

Balkema 1992. Formation water


Salinity 25.000 ppm NaCI eqv
Resistivity 0.236 ohm m at 2SC (77'F)
6. Holland C.I., Smart B.G.D., " Beyond PI 1982/6"
The Mining Engineer luly 1990 29-32 . Reservoir conditions
95.6"C 1204"F)
Temoerature
Pressure 399 barsa (5785 psia)
7. Hasegawa H.S., Wetrniller R.I., Gendzwill D.I., " Pressure- gradient in reservoir 0.062 bar/m (0.2i4 psilfH oil

Induced "Seismicity in Mines in Canada - An


Overview" Pure Appl. Geophys. 129 423-453.
8. Knott, S.D. 1993. Fault Seal Analysis in the
North Sea. The American Association of Petroleum
Geologists Bulletin V.77, No.5, pp 778-792.
9. Bayat M.G., Tehrani D.H., " The Thistle Field-
Analysis of its Past Performance and Optimisation B
of its Future Development" SPE 13989/1 Offshore
Europe, Aberdeen 1985.
10. Smart B.G.D., Somerville, I. McL. & Hamilton,
S.A., 1989. Absolute Permeability at Reservoir Fig 1: Observed strata deformation processes
Conditions. University of Strathclyde Dept. Mineral
Resources Engng. report to Britoil pIc.
11. Struijk, A.P. Green, R.T. 1991. The Brent
Field, Block 211129, UK North Sea. From Abbotts,
I.L. (ed.), United Kingdom Oil and Gas Fields 25
Commemorative Volume, Geological Society
Memoir No. 14, pp. 63-72.

Table 1: Rock mechanical properties

Material Elastic Poissons' Compress


Description Ratio -ibility
No. Modulus
(GPa) (llbar)
I Reservoir rock 12.0 0.2 5.8 xlOs
2 Over/underburden rock 1.2 0.4 NA

159
Hypothetical Relative Shear Dlsplacament -
Transmissibility Characteristics
-------- -I 1.0
r---------~--------_.
1 l--~
0.8
/
'/ I I I

0.6
.II III Non-seal to seal

~I Seal to non-sea
~ I 1
:a
.
0;
0.4

IA
....
E 0.2
c I~
Pre-prgcaujog Ph_ lII.,r--
~ 0.0
o 100 200 300 400 500 600
Relative Shear Displacement (mm)

Fig 5: Fault displacement / transmissibility

FAU. T ANClE 60 P~PRE~ 400 be,. - COL I


COEFF. FRICTION 0.2 300 I..,. - cot. 2
1 N'P. STRESSES 65( V) ,50( H) HP. 200 b .. r - COl. 3
1 100 ,. - COl ..
L Solytion
____ pba._ (x IOU-I)
4 . - ...

Post-procaaaing Program 3.5 - - - - ... ---- --1--1---1--1--1---1---1


save data file & pJot diagram:
stresses. displacement. pressure
production rate. etc. 3 - --1--- --- - - --c---.- ._f--

2.5 - - - - - - - - - - --I~ ::""'\-_. - ..- --


Fig 2: ANSYS / ECLIPSE coupling
".
t..

C
.:~=-rt \. --+--1-1

'::=-::~;;;:1
(f)

.
ex: 0""
o ESt ~
COL 4
-0.5f--- \ - - - ---I---I---+--I--.~--+--.I-----j

-I L..._._ I--.'---f---
-500 -300 -100 100 300 500
-400 -200 0 200 400
Oi st. <11 on II di sconti nui ty (m)

Fig 6: Typical fault slip plots

0.6
Fig 3: Finite element model of Brent-like reservoir 0.5
0.4
Coefficient of 0.3
Friction 0.2
Hypothetical Stress - Permeability Curve
0.1
1.0 o o
1.8600~-2X
1 I I

'" I Y = 0.99836 - + 3.1 445e-4x"2


- 2.6973e-6x"3 + 9.1485e-9 x"4 Fault Angle
0.8
..... R"2 1.000 =
~
(deg)
0.6 .I I Ratio of Horizontal

-...... I
0

0.4
I to Vertical Stress

0.2
I Fig 7: Critical combinations of parameters

0.0
o 20
I
40 60 80 100
Stress (MPa)

Fig 4: Stress / penneability sensitivity

160
rop,. VS TIME Me 1 3-& ROIPL AEON 2 "S TIME MC13-0
"OPA "s
TIM MC12-5
FOPR VS T .... E MC1t-o
ROIPL RECI. 2 VS TIME .... C12-S
ROIPL REG,. 2 VS TIME MC11-0
....
to.:s/o
__ 3 ...v 11M"
.fO__ O
2.0 ,...........- - - - - - -..... ----...,.:------------- ~O -----------------------------,

t
\'.
\:
~
\
1
1
1
1
2
-
-
-
..,...~,.,.,..,.,..,.~~~~_:_:_:_
--.....:.........,. I
1

.. .. .... .... 1
\ 1
1.2 \ 1 1
\ 1 2.2 - I
\ 1 I
0." ~
~
I1 .... I
I
I
I
0.4 "\\ I1 1.4
I
I
~, .' 1 I
'.:<".:.: 1 I
o.oo~.o~~~7t~0~OC..O~~~2~O~0~.0~~~~of~~4~o~o~.ro~~~a~o~o~.~o~~~.~o~o.o t.OO,~.O~~~~1rO~O~.0~~~2~O~O~.o~-L~~~0~'b
O~AYW~L-~4~0~O.o~-L~a~o~o~.o~-L~a~o~o.o

WBHP OP- 1 va TINE ....e 13-0 ,.OPT vS TIME ......::: 1 3 - .


wBHP OP- 1 vS T .... E .... C12-S FOPT VS TINS: "'C12-5
Y/BHP OP-1 VS T ..... E MC1t-O FOPT "S TIME Me 11-0
.AA .... ."'3
-10--3

700.0
----------------~:~~~~---~--~
400.0

380.0
//
800.0
.20.0 ..... ..........
500.0
""
L"":------
280.0
~', .........'-
2"'0.0
..... 400.0
".
200.0

1180.0
"'-,
........---..:~.~------.
300.0

120.0 .z00.0

ao.o
AO.O

'00.0 200.0 400.0 800.0 .00.0

Fig 8: ECLIPSE output for cases in which fault transmissibility changes from non-seal to seal

roPft \IS T .... E ....CZ-3 AOIPL fltEON 2 va TIME .....C2-3


"OPA VS T .... E MC2-1 ROIPL RECN 2 VS TI .... E ....C 2 - '

".0
!t'cl':'!AY
~==--- .....------...._--------------I
...."
_to __ 8

...
\ 1 3.0 -
\ 1
\ 1 2
\
\
'.2 \
\ 2.2

\
D
\
\
\
....
\
D." \ 1."
\
\
.....

WBH'" 0 .. -1 va TINE MC2-3


P'O"T va TI .... E ....(;2-3
WBHP OP-1 VS TINE MC2-1 F"OPT VS T .... E ....C2-'

.......... .... 3
-~=~-------------- -::.:-=-=--~~-------I ..
400.0 .00.0
I
1
0.0 I
.20.0 &00.0 1
I
2.0.0 I
~
....00.0 1
aAD.O
\ I
200.0 \ ,, ~OO.O
----------------------1
I
1.0.0
'----\ I
1ao.o
200.0 I
\ I
ao.o \ 100.0
I
\ I
40.0
\ I

Fig 9: ECLIPSE output for cases in which fault transmissibility changes from seal to non-seal

161

Potrebbero piacerti anche