ArticlePDF Available

Interactive Reservoir Simulation

Authors:

Abstract

Shell's new Modular Reservoir Simulator (MoReS) has been equipped with a comprehensive and versatile user interface called FrontEnd. Apart from providing a user-friendly environment for interactive reservoir simulation, FrontEnd serves as a software platform for other dynamic simulation and reservoir-engineering applications. It offers to all supported applications a common user interface, enables the re-use of code and reduces overall maintenance and support costs associated with the embedded applications. Because of its features, FrontEnd facilitates the transfer of research results in the form of operational software to end users. When coupled with MoReS, FrontEnd can be used for pre- and post-processing and interactive simulation. The pre-processing options allow data to be inputted by means of various OSF/Motif widgets containing a spreadsheet, text editors, dialogues and graphical input. The display of the input data as well as the post-processing of all simulation results is made possible by a variety of user-defined plots of tabular (e.g. timestep summary) and array (simulation grid) data. During a simulation user-defined plots can be displayed and edited, allowing a close inspection of the results as they are being calculated. FrontEnd has been equipped with a powerful input command language, which gives the batch user as much flexibility and control over the input as the interactive user. Introduction The introduction of powerful workstations offers opportunities to change the way reservoir simulations are carried out. Most current reservoir-simulation software require the engineer to assemble a data deck with a text editor or dedicated pre-processors, submit the run in batch mode, wait for its execution and then investigate the results using post-processing software. Yet current hardware is powerful enough to run interactive simulations whose results can be monitored as they are being calculated. Such interactive simulation can increase the efficiency of the engineer significantly. Since the data can be immediately displayed and validated, obvious errors can be detected early and alternative scenarios can be quickly evaluated. To make Shell's new Modular Reservoir Simulation package MoReS fully interactive, it has been equipped with a versatile and powerful user interface called FrontEnd. BRIEF DESCRIPTION OF MORES The MoReS reservoir simulator offers a large choice of fluid descriptions, ranging from standard black or volatile oil to user-specified multicomponent mixtures. In addition, tracers, polymers and chemical reactions are handled for various specialist applications (e.g. reservoir souring, environmental applications etc.). The simulator can be operated in either a non-fractured or fractured mode. In the latter case dual porosity, dual permeability and block-to-block flow (oil re-infiltration) can be taken into account. P. 545
lt30019tyofWtrolNm Engktem I
SPE 29146
Interactive Reservoir Simulation
J.M.M. Regtien ***, G.J.A, Por *, M.T. van Stiphout and F.F. van der Vlugt
Shell Research Rijswijk
*SPE Member
This paper W= prepared wpmsentatbn at the Isth SPE Sympo8ium cmReamoir Simulation held in San Antcmim~i U.S.A., 12-15 Febr_ 1ss5.
This paper was adectsd for presentation by an SPE Program C0mmittc8 fdlowlng review of Information contaitmd in an 8b8tmcf submitted by the aufhor(a). C0nt0nf8 of the paper,
MPMWntOd, haw not been mdevmd by the Sodafy of Petrofeum Ewirwm and we subject to correction by the author(a). Tim material, as presented, &ma not neceaawify reflect
MY_ ~t~ *~fy ~p~~um Emi~m. ~off~m, of ~m~. pp -m ~SpE *W We Wb@f to Publi@tbn r- by Edhmial ~mm~ of t~ *kW
of Petroleum Er#irmom. Pennklontocopy ismabbbd @an~dtimtb ~*.lll@@tis m9tib~. ~=**M_nm*ws~@
of whue and by whom the -r is Pm8snted. Write Libmfbn, SPE, P.O. Box SSSSSS,Rbh~dfm ~7~1 U.S.A. TOkX,1SS245 SPEUT.
ABSTRACT
Shell’s new Modular Reservoir Simulator (MoReS) has keen
equipped with acomprehensive and versatile user interface called
FrontEnd. Apart from providing auser-friendly environment for
interactive reservoir simulation, FrontEnd serves as asoftware
platform for other dynamic simulation and reservoir-engineering
applications. It offers to all supported applications acommon user
interface, enables the re-use of code and reduces overall maintenance
and support costs associated with the embedded applications.
Because of its features, FrontEnd facilitates the transfer of research
results in the form of operational software to end users.
When coupled with MoReS, FrontEnd can be used for pre- and post-
processing and interactive simulation. The pre-pmcessing options
allow data to be inputted by means of various 0SFM40tif widgets
containing aspreadsheet, text editors, dialogues and graphical input.
The display of the input data as well as the post-processing of all
simulation results is made possible by avariety of user-defined plots
of tabular (e.g. timestep sumtmwy) and array (simulation grid) data.
During asimulation user-defined plots can be displayed and edited,
allowing aclose inspection of the results as they are being calculated.
FrontEnd has been equipped with apowerfui input command
language, which gives the batch user as much flexibility and control
over the input as the interactive user.
RODUCTION
Theintroduction of powerthl workstations offers opportunities to
change the way reservoir simulations are carried out. Most current
reservoir-simulation software require the engineer to assemble adata
deck with atext edhr or dedkated pmprncessors, submit the run in
batch mode, wait for its execution and then investigate the results
using post-processing software [1]. Yet current hardware is powerful
enough to run interactive simulations whose results can be monitored
as they are being calculated. Such interactive simulation can increase
the efficiency of the engineer significantly. Since the data can be
immediately displayed and validated, obvious errors can be detected
early and alternative scenarios can be quickly evaluated.
To make Shell’s new Modular Reservoir Simulation package MoReS
[2] fully interactive, it has been equipped with aversatile and
powerful user interface called FrontEnd.
BRIEF DESCRIPTION OF MQBES
llre MoReS
reservoir simulator offers alarge choice of fluid
&scriptions, ranging from standard black or volatile oil to user-
specified multicomponent mixtures. In addition, tracers, pcdymers
and chemical reactions are handled for various specialist applications
(e.g. reservoir souring, environmental applications etc.). The
simulator can be operated in either anon-fractured or fractured mode
[2, 5, 6]. In the latter case dual porosity, dual permeability and block-
to-block flow (oil re-infiltration) can be taken into account. The
program has also been equipped with various options to effectively
** Cumntiy with Brunei Shell Petroleum 545
2Interactive Reservoir Simulation SPE 29146
handle detailed and realistic three-dimensional (statistically derived)
....- J--i --A I. AA.,+.. mf _latiw. mrmenhilitv rmd Canillarv
~WIU~lGU1 lllUU&& AU1lULW “. .-.-” .- y- ..-------- –.- --r-- –.
pressure models is availabl% aspecial module then generates either
static or dynamic pseudo-curves for use on acoarse grid.
The program’s well model accurately simulates vertical, deviated and
horizontal wells, including crossflow. The well-management package
is, by default, coupled to the wells of the subsurface simulation
model, but it can also be used in astand-alone fashion with user-
supplied well potentials. This feature means that acomplicated
.-.....--* ...- k. “.* .. .“A .mnl”~ @L~~Qu~~~~~&~ Qfrioing
llGLWUIF. !,aII UG SW-upal” u..-,
complete simulations.
CRIPTION OF FRONT~
FrontEnd is best described as asoftware platform for dynamic
modelling applications. So far, it has mostly been used for the pre-
and post-processing of both reservoir-simulation input data and
results and the run-time control of (interactive) applications.
FrontEnd offers to all applications it supporkY
acommon graphic user interface for data input and display
access to internal application data via adata dictionary;
apowerful input command language for batch input as well
as for arithmetic manipulation and display of all available
application data (variables, tables, array data);
real-time simulation and animation;
tisll restart captillities;
an on-line documentation system, and
an interface to spreadsheets and third-party 3D
vistralisation programs [3].
The graphic user interface consists of amenu bar and two main
windows. The first window contains the “’Dictionary”(Figure 1),
which manages data transmission to and from the user. The second
window is the “Panorama” window (Figure 2); through it, the user
can SIXthe data displayed. Depending on the user settings, various
methnds to edit or browse through the data are available.
All relevant application data in FrontEnd can be visualized in the
form of icons grouped in ahierarchical structure, like that used in the
well-known Macintosh or Windows desktops. Each data type has its
own unique icon. Aselection of the supported data types is visible in
Figure 1.
Each of the icons allows access to the data in one of several
supported representations in the Panorama window. The number of
representations depends on the object type:
XY cross plot (for tables);
SprUdSk?t (for =JW and tables);
editable dialog with widgeta (all~
colour-fill plots (for arrays);
editable text (all).
Representations combining data from different applications can also
be constructed.
The user can add plot representations that will be stored with the data
for later use. The plot components (axes, legenda, colours, titles, data
points) are filly editable.
The advantages of such an icon-based data dictionary are obvious:
The user always has access to all relevant application data,
which are organised in aclear way.
‘he data can be checked, visualized and edted at any time.
No prior choice of ouQut type is rquired.
Access to the data via “tiedictionary is possible not only by way of
interactive methods but also through the text input (e.g. for batch
running); for the latter the FrontEnd input command language is
used.
Interactive simulation allows the engineer to visualise any of the
results while the simulator is time-stepping. At any moment,
however, he or she can interrupt the simulation, change the input
(e.g. well controls) and then continue. Except for the grid dimensions
and the maximum number of phases and components, any data can
be changed before or even after reservoir initialisation.
Such interactive simulation increases the efficiency of the engineer
significantly, because:
results are available immediately;
incorrect results can be detected on the spot, allowing the
user to abort the rms and thereby spare computer resources;
“what if’ scenarios can otlen be executed much more
quickly.
r-s,=t.~~for ins~.~, the cause of convergence problems (e.g., an
..-..--,
erroneous transmissibility or very small pm volume in alarge or
complex simulation) can be quickly spotted and corrected.
Four exawspiks:
1.
2.
3.
4.
Weff planning is made easier with FrontEnd, since it enables the
user to define awell’s trajectory and its perforations. The
simulator can then be requested to calculate the position and
properties of perforations and to determine the well’s potential.
By varying the trajectory, one can determine an optimum
location for the well.
Alarge simulation model can be run on ahigh-end workstation,
whilst the FrontEnd windows are visible on aPC using X-
windows and TCP/IP network software. Windows to other PC-
baaed computer applications, such as word-processing and
spreadsheet pmgrasrss,can be placed next to the FrontEnd
windows. The progress of the simulation can be monitored by
activating any of the default or user-defined output and plot
options. The table or array plots will then be updated at every
timestep.
Wells that are shut-in smexpectdy ‘becauseof iifi die-out can
be examined in detail by setting up plots of the intake pressure
curves and the inflow performance of the well. The parameters
necessary to interpolate between the data in the multi-
dimensional vertical-flow or lift tables are taken fmm the well,
and the well inflow performance curve reflects the actual
gridblock data.
Aseparate well-potential cdxdation dialog has been added that
can be used at any time to study the effects of various
546
.
SPE 29146 J.M.M. Regtien, G.J.A. Per, M.T. van Stiphout and F.F. van der Vlugt 3
constraints imposed on awell. The resulting (potential) flows
..d “.p=.,l~. *s,,lt;IK7 fmm tire mn.ctIimitimrconstraintCanbe
-,” p,””.,..-” .“ “.....= . .. ... -- - .----- ......—..=----- --—---
displayed on the computer screen or printed.
The FrontEnd input command language first of all provides keyword
input for the applications to be run in batch mode. For production
runs this mode of operation is often preferred, since it avoids
repeating the same sequence of interactive actions. Secondly, it
enables the user to customise FrontEnds firnctionahty. TM essential
feature is best illustrated by example.
The implementation of well management in areservoir simulator is
-~-.. .------- ,.~. . . . . -k...nm .mn,,-.fc . . wall ql.”.in-ment nnlh+=c
UILG1l aauulw UL Illally -,,au~w JUq”w. k., (s.7““.. J.. . . . ..~-. .. y-------
tend to be field-specific. Hard coding these in the simulator itself
would result in amassive and costly programming effort. With
FrontEnd, however, several optional programming constmcts can he
called upon to modify or extend the simulator’s functionality:
access to all relevant data via the dictionary;
creation of user-defined data objects;
variable substitution;
table manipulation;
arithmetic;
“if” statements and” while” loops;
apre-defined function library;
user-defined functions;
flexible unit conversions;
user-defined file I/O;
ethe C:eatiml &id dkirlg c! p!ots;
access to the operating system.
The input command language also allows the specification of what
are called monitors: functions that are evaluated by the application
whenever aset of user-defined conditions are met. The first one and
ah.lf .J.m.~ fif 17rnmtl%d”. tmm-ti,mq have alrem-tw chmun that itc inmnt
a..-1 ,bG?J. “1 ,,Vllw.1” .“p.....”. ..”7 “..—, ...”.... ...”. ..” ...~.-.
command language enhances the applications it supports.
Five examples:
1.
2.
3.
4,
Screening economics-present vrdue, cash flow, net revenue
and cumulative cash flow-can be automatically calculated
from the output of asimulation mn on the basis of user-defined
discount factors, oil prices, and capital and operating expense
mrmfilac N.-wI. mf the dmw.ithqc hd tn he had rndd 9S thev
F.”...-”. ..“..- “. “.- .“-- ... . .. ..— .- “- .. --- —-, --. -,
can all be included in the user’s input data.
Correlations often describe PVT and saturation data or porosity-
permeability relationships. Such correlations can now be coded
and customised by the user. Although alibrary of various
indu$m t.~rrt=l~t;nmiQ nmvidd the IIcer can m-mvand modify
..J -.. ”.”..-... .. ~.- ..—-, --- ---- --.--.=, —-- ---
them to suit local conditions.
Well-management policies can be implemented by comparing
well production data with targets. Wells and intervals can be
shut-in or opened up on any of the user-defined criteria. Water
breakthrough in one well may thus automatically result in
certain actions at other wells.
Alarge simulation model in history-match mode can be made to
stop by itself when the computed results deviate too much from
the historically observed data. This option saves troth the
computer’s and the engineer’s time.
5. The simulator can he used to study the feasibility of atime-lapse
seismic survey, by calculating relevant seismic parameters from
the simulated reservoir properties. The property data are
gathered automatically during the simulation and, directly after
its completion, the changes in seismic parameters showing the
movement of fluid contacts are made available.
The most important function of FrontEnd as apre-processor is to
check data. All basic data types, such as variables, tables and arrays,
~h~r. ~Svstem of ch~ks on rnInjrnUrn md maximum allowed values.
,-------
Tables contain checks on the required monotonicity of the data in the
columns. The built-in capabilities for the handling of dimensional
units automatically ensure that two numbers of different quantities
(e.g. pressure and mass) are not equated, added, subtracted or used in
an assignment. Adefault unit system can be selected, or alocal set of
units can be defined as the default. In addition, data can be entered in
terms of any unit merely by adding the unit expression after the data
themselves.
All tabular input (PVT, relative permeabilities, capillary pressures,
vertical flow, historical production data) can be plotted and
inspected. Tables can be re-interpolated or filled using user-defined
correlations. When two two-phase relative permeability tables are
used in combination with athree-phase model, the resulting three-
nhace relative nermeahi Iities can be @oti in aternary dl~=~.
r---- ------ .- ~- . ...-_-. ..._-_ _
Most of the well data are generally stored in corporate databases;
they commonly need reformatting before they can be used as
simulator input. FrontEnd has therefore been equipped with an
interface to aproject database that automatically converts well
tldinitinn (tmicctnrv and completions) and performance &tS
---------- .--J -- - ,-
(historical rates and pressures) directly into FrontEnd objects that the
simulator can use.
An extensive on-line help system with search and browse facilities is
coupled to the FrontEnds data dictionary. All icons contain heip
information, and the collected information pertaining to all
application objects forms areference manual that can be printed as a
PostScript file. The user can use the same functionality to document
~h~~~rn.~i~tiQnrn~ei:
P-t pr=w@MMms .
.
The reservoir simulator MoReS as well as other FrontEnd-supported
applications can create and fill tables and arrays that can later be used
by FrontEnd. For example, the simulator generates, for all wells and
surface network nodes, default tables of presaums, rates, ratios
(GOR, BSW, etc.), cumulative volumes of fluids producedinjected.
These tables can be plotted and compared with observed data. In
.A; .ti .“ nuni. “sri m,Clunril,t.rit-mcrmmio~ O&r ~- ccm.p~red ~rr
1----””.. ...--9 -- .“-- r------- —.—-
one plot. Similarly, the properties of the subsurface grid can be
displayed in the colour-fill plots. Any of the existing array data can
be used to calculate new parameters. For example, awater%od-
movable oil array can be calculated from auser-defined algorithm
incorporating existing data.
547
4Interactive Reservoir Simulation SPE 29146
Plots can be combined in multiple views (Figure 3). Any number of
these views can be defined for interactive or batch plotting. Plots of
@e surface network, for instance, can easily be generated (Figure 4).
All flow rates and active well-management policies cart be accessed
by selecting the various objects in the network plot.
Three-dimensional visuaiisation has riot beetr iinpielmfiritedkr
FrontEnd, given the 3D reservoir-simulation visualization software
available in the OFI market [3]. Instead, interfarm have been written
so that simulation results and gridded data can be visualized with two
leading commercial packages.
Quality assurance of the simulation rnr@ by peer m+GIVia.i
.- .- —-.,:.-.
review is made quite easy. The reviewer merely opens arestart file
and systematictdly checks the basic input and the resultw there is no
need to do additional simulations.
Tables and arrays can be written in afile format for expott to
spreadsheet programs or in any user-defined file format. Plot files
c~~ ~. gen~m~ in postscript fofmat for printing on their own or for
inclusion in other documents and presentation material.
lmmmmMn
Both MoRes and FrontEnd have been developed using the latest
industry standards and technology. The implementation languages
.-. e%. *h. M., ;.t@f.eP mm-lFn~-77 for&e CidCUkltiOn core Of
us, b.“. “.” ..s.,. .. . ..-+ -.-. . .
the applications. The design is fully object oriental the Clanguage
has been used to implement aclass structure with standatd software
~=,wnen~, ~i~ Ohierts. that mav ifierit properties ~d data from
., —-, -— —.-,
other objects. OSF/Motif and Xl 1are the packages used to
implement the graphic user interface and the screen graphics. The
POSC user-interface guidelines [4] have been followed as much as
Possibl% plot file output follows postscript standards. Adherence to
programming standards is strict, resulting in an excellent portability.
There is only one version of the current package, which is
operational on various hardware platforms.
RONGND PRO-
...--.: -.I:~, ..~w.n-tumAat+it. s},t.re~~fi~! IJ~C. W~~h-
TiW auaptabk fulkuoll=a~, U. ..V8.U.U A., .. .-
its first major application, the MoReS reservoir simulator, has
prompted Shell software developem to rely on FrontEnd as user
interface for other reservoir-modelling applications. Tbe following
applications are now available in the FrontEnd environment
REDUCE: areservoir simulation pre-pmcessing package that
compresses &tm”led(geostatktical) geological models of multi-
rnillionceh!rinto auser-dejined simulation grid (Figure 5).
The compression includes not only the scaling up of properties like
porosity and permeability but also the calculation of comerpoints
and transmissibilities for the simulator grid blocks. Various
averaging techniques are available to coarsen the geological model
c., .-:.- . . .. :..... +...+ frm.”A-*-rmi”i”e fm~me~,
Wniie PreSWvliig IQ IILUaLu@k UU.. ..V” =.- . .. . .. ... .—
PROJECTDAT~SE: on inte~ace that connects thereservoir
simulator to petroleum-engineering databases and applicatwns,
coqrorate &tabases (historical production&to ~generic well
&to like well tmjectories andptvjoratwns) orpre-processing
applications (grid generation; PVT table;, pseudo relutive
permersbilities and capillary pressures; il@tabks).
PROJBCT DATABASE stores all these data allowing
engineers
working on the same project to share them. FrontEnd can be use
to browse through the database (with the aid of the dictionary)
an
load datq into the reservoir simulator.
FRACNET: awell-test package for horizontal wells in sparsely
f?actured reservoirs.
FRACNEI’ generates arbitrary fracture geometries and
calculates
the response of awell connected to the fmcture network as a
fii=titi, u. UI..W. x.W---- .. . . .
,.s M-. 17m.md k,Id to sp:fy t~e fractuR
geomet
and other test parameters as well as to visualise and analyze the
well-test data.
SUBCAL arnodellingpackage that con be used to predict
reservo
compaction and swfhce subsidence as aresult of theproduction
of
hydrocarbons.
AvarietyGfmde:iirig CPU..- -. ..—.— . .. . .. .-...-=.
.&,.... .- mat+ .vailahlp thmll~h
FrontEnd, including an optional link with the reservoir
simulator
OtbCrzippiicztimisthat aii bekig rieve!qree wrrtodifk! huse
within the FrontEnd environment are
1. apackage for forecasting acompany’s medium- and long-ter
production on the basis of individual teservoir forecasts
originating from various sources, from decline curves to
full-
field reservoir simulatiotu
2. agas-field planning td [7] consisting of athird-party
surfac
network model, asubsurface simulator (reservoir simulator
and/or material-balance tank models) and adriver that handl
the contracts and steers the subsurface and surface simulator
amaster-slave mode.
Auser-friendly pre- and post-processing environment for
reservoir simulation has been developed. It adds anew,
interactive dimension to reservoir .$imuiSdOmi—india”e
WX.W
to Mrelevant simulation data at any titnq data validation
and
visualization of results in various forms; and full user control
the course of asimulation as it is being run.
Application of the fully integrated FrontEnd/MoReS
package
has Ied to pre- and post-processing that are more efficient
and
higher quality than that done with conventional simulation
sotlware, which consists of asuite of separate pre- and post-
processing programs.
One of the key elements of FrontEnd has been its flexible
inp
mmmand language. The language’s programming features
enable the user to enhance data analysis and customise the
simulator’s functionality.
FrontEnd has also been successfully developed into a
platform
for dynamic rwervoir-modelling applications other than
reservoir simulation. It offers the user acommon user
interfac
for all embedded applications. The re-use of code and
reduce
overall maintenance and suppott coats are additional benefits
Because of itt versatiiiey and s“a!iiitjy, ~i%on’~-dlets new
.
SPE 29146 J.M.M. Regtien, G.J.A. For, M.T. van Stiphout and F.F. van der Vlugt 5
research developments quickly become operational software.
products,
1.
2.
3.
4.
5.
6.
7.
Rogers, W.L., Ingalls, L.J. and Praa@ S.J.: “Prc- and
Postprocessing for Reservoir Simulation”, SPE 20360.
Per, G.J.A., Boerrigter, P.M., Maas, J.G., and de Vries, A.S.: “A
fractured reservoir simulator capable of modelling block-block
interaction”, SPE 19807.
Wells, M. and Watkins, H.: “3-D visualization of multivariate
reservoir simulation data on complex grids”, presented at the
1992 IBM Europe Institute conference on Computational
Methods and Tools in Reservoir modelting, 17-22 August 1992,
Obcrlech, Austria.
POSC User Interface Style Guide, version 1.0
.,.. n:a...... r=m“..AW.11,.. T.W2_,,”4..-A -“—...:-
val, “,, MI,,,, G.ti. al,” ,, m-, , , . . ,., a&Lul Fa4 ,Cz.w v“,,
simulation and field developmen~ Natih field (Oman)”, SPE
22917.
Bccrrigter, P.M., van der Lcemput, L.E.C., Pieters, J., Wit, K.,
and Ypma, J.G.J.: “Fractured Rmervoir Simulation: Case
Histories”, SPE 25615.
Hooi, H.R.., Goobie, L., Cook, R. and Choi, J., The integrated
Team Approach to the Optimization of aMature Gas Field,
paper SPE 26144, presented at the SPE Gas Technology
Symposium, Calgary, 28-30 June 1993 @mceedings pp. 73-80)
The authors wish to thank the management of Shell Research BV and
Shell International Petroleum Maatschappij for permission to
publish this work.
549
m
Cn
o
A
.R)
Fig,,1Dictionary dispiay of (soieotsd) reservoir
simulation data.
Fig. 2Panorama dispiay of adata oioject (saturation array).
. . .
SPE 29146
0
[*MS] !W31VM PoJdl10
iiis
[Slsmd tlo9
dDr.no.113073 551 FIG. 3
SPE 29146
N
0
dDr.rto,l13074 552 FIGS. 4,5
Article
Turbidite reservoirs with lobe stratigraphic architecture are typically encountered in deep-water environments and are present in a large number of basins. In the past, a comprehensive study to identify the key parameters governing the dynamic connectivity of turbidite channel reservoirs formed the basis for the development of the 3D reservoir connectivity factor concept allowing to account for un-modeled fine-scale stratigraphic details by computing geologically consistent field recovery discount factors. Following-up on the success of this approach, it was decided to proceed similarly with turbidite reservoirs containing lobe stratigraphic architecture, by first starting to investigate the parameters driving the performance of these reservoirs. Fine-scale reservoir models are constructed, and a large number of sensitivity-study type flow simulations are performed testing the recovery factor sensitivity to a large number of geological parameters. Analysis of the resulting data indicates that the distributary channel shale-drape coverage and its permeability, together with the lobe axis permeability, are important parameters affecting the connectivity of turbidite lobe reservoirs. Other parameters such as the lobe margin permeability, the lobe infill and lobe complex shale-drape coverages are secondary parameters. The impact of the channel incision depth, the geological differences between proximal (more channelized) and distal (more lobate) environments and the impact of the chosen well spacing are only noticeable when shale-drape coverages are high, particularly for the distributary channel drapes. The effect of well spacing appears to be stronger in the distal environment. Through extensive visualizations of the simulated oil saturations, we further conclude that the location of by-passed oil depends on the facies permeabilities and on the shale-drape coverages. Our base case scenario shows that lobe axes are better swept than distributary channels and lobe margins. At 100% channel shale-drape coverage, the oil located within the channels can be totally by-passed for some channels depending on well locations. When all shale-drape coverages are high (at or close to 100%), by-passed oil is observed in every facies. Contrarily to lower drape coverages, this complicated situation does not allow for a per facies localization of by-passed oil. At high shale-drape coverage, it is further noticed that the sweep efficiency of deeply incising channel models is lower than the one simulated with shallower incising channels. Two main reasons contribute to this effect: first, the significant volume of by-passed oil that is located in the channels (deeply incising channels being bigger, less oil is swept), and second the fact that deep channels compartmentalize the reservoir more than shallow ones at high shale-drape coverages. The ensuing results are compared with those for turbidite channel reservoirs. It is concluded that shale-drape coverage has considerably less impact in lobate than in channelized systems. At equivalent shale-drape coverages, turbidite lobe reservoirs have a greater connectivity than turbidite channel reservoirs.
Article
Full-text available
Fault modeling has become an integral element of reservoir simulation for structurally complex reservoirs. Modeling of faults in general has major implications for the simulation grid. In turn, the grid quality control is very important in order to attain accurate simulation results. We investigate the dynamic effects of using stair-step grid (SSG) and corner-point grid (CPG) approaches for fault modeling from the perspective of dynamic reservoir performance forecasting. We have performed a number of grid convergence and grid-type sensitivity studies for a variety of simple, yet intuitive faulted flow simulation problems with gradually increasing complexity. We have also explored the added value of the multipoint flux approximation (MPFA) method over the conventional two-point flux approximation (TPFA) to increase the accuracy of reservoir simulation results obtained on CPGs. Effects of fault seal modeling on grid-resolution convergence and grid-type sensitivity have also been briefly examined. For simple geometries, both SSG and CPG can be used for fault modeling with similar accuracy in conjunction with the pillar-grid approach. This is evidenced by the fact that simulation results from SSG and CPG converge to identical solutions. SSG and CPG yield different results for more complex geometries. Simulation results approach to a converged solution for relatively fine SSGs. However, a SSG only provides an approximation to the fault geometry and reservoir volumes when the grid is coarse. On the other hand, non-orthogonality errors are increasingly evident in relatively more complex faulted models on CPGs and such errors cannot be addressed by grid refinement. It has been observed that MPFA partially addresses the discretization errors on non-orthogonal grids but only from the total flux accuracy perspective. However, transport related errors are still evident. Grid convergence behaviors and grid effects are quite similar with or without fault seal modeling (i.e., dedicated fault-zone modeling by use of scaled-up seal factors) for simple geometries. However, in more complex test cases, we have observed that it is more difficult to achieve converged results in conjunction with fault seal modeling due to increased heterogeneity of the underlying problem.
Article
Full-text available
Many properties of complex porous media such as reservoir rocks are strongly affected by heterogeneity at different scales. Complex depositional and diagenetic processes have a strong control on the pore structures, leading to systems with a wide range of pore sizes covering many orders of magnitude in length scales. This poses a significant challenge for digital rock analysis since a single resolution image and associated simulation model cannot capture all the relevant length scales in sufficient detail due to limitations in computer memory and speed. The scale-transgressive effects of heterogeneity must therefore be accounted for through a multiscale digital rock workflow. We introduce a generalized multiscale imaging and pore-scale modelling workflow to derive transport properties of complex rocks having broad pore size distributions. A dry/wet micro-CT imaging sequence is used to spatially map the porosity and the connectivity of resolved and unresolved porous regions. The unresolved porosity regions are classified into different porosity classes or rock types. The resulting 3D rock-type map and the porosity map are combined and transformed into a multiscale pore network model. Resolved pores are treated in a conventional pore network manner while unresolved network elements are treated as a continuum Darcy-type porous medium. Similar to conventional continuum models, each Darcy pore is populated with single and multiphase flow properties. These properties are derived from high-resolution rock-type models constructed from backscatter SEM images and/or high-resolution micro-CT images of sub-samples. The multiscale digital rock workflow is applied to two heterogeneous rock samples: a mixed wet thinly laminated reservoir sandstone and an oil wet reservoir carbonate. Experimentally measured mercury–air primary drainage and oil–water imbibition capillary pressure curves (after ageing to restore wettability) are used to anchor the multiscale pore network model. Waterflood relative permeability is calculated in a blind test and compared with high-quality experimental data. A very encouraging agreement between computed and measured properties is found.
Article
Local upscaling techniques fall short in capturing the dynamic effects of laterally continuous non-local geologic heterogeneities, for example, high-permeability channels, vertically thin, yet, horizontally extensive shale layers. Global upscaling techniques have been introduced to remedy the shortcomings of local scale-up approaches. In order to address the upscaling challenges associated with non-local heterogeneities, we have developed a practical and effective flow-based transmissibility scale-up method that computes coarse-scale transmissibilities using flux derived from single-phase flow simulations with generic global boundary conditions, referred to as the global transmissibility upscaling (GTU) method. GTU targets more accurate upscaling of laterally continuous non-local heterogeneities including the ones modeled on distorted cornerpoint grids. A variant of GTU is developed in this context that uses fluxes computed by use of the multi-point flux approximation (MPFA) discretization technique at the fine scale. While various global upscaling methods were described in the literature, they are rarely used in industrial applications due to challenges related to implementation. We describe in detail an effective practical implementation of the GTU method in a comprehensive industrial-grade simulation package and share our learnings with GTU on aggressively upscaled models with non-local heterogeneities. We demonstrate the application of GTU to several challenging reservoir models. Comparisons of GTU-based model responses are made with ones stemming from local upscaling techniques popularly used in the industry using fine-scale model predictions as reference. Investigated numerical examples demonstrate that the GTU method leads to notable improvements in the accuracy delivered by coarse-scale models. Especially, the application of GTU is most impactful for high-upscaling-ratio models with complex fine-scale connectivity because GTU preserves the fine-scale behavior more authentically compared to local techniques. We demonstrate that GTU naturally complements an innovative regional multiphase upscaling (RMU) method on aggressively upscaled models with strong multiphase flow associated with displacement-type recovery mechanisms. It has also been observed that using MPFA on cornerpoint grids to obtain fine-scale fluxes for GTU gives rise to more accurate coarse-scale pressure behavior. We have also observed that GTU is especially effective when combined with a regional multiphase upscaling method for ultimate accuracy in simulations of displacement-based recovery mechanisms.
Article
Field-development optimization and optimization at the pattern scale are crucial to maximize the value of thermal enhanced-oil-recovery (EOR) projects. Application of a field net-present-value (NPV)-based pattern optimization algorithm honoring field-scale surface and subsurface constraints for in-situ-upgrading (IUP) projects has been described in the recent past. In this paper, we describe the development and application of a novel field-development-optimization capability, including the optimization of the ramp-up phase to accelerate the production to achieve a faster cash flow and high surface-facility utilization. We integrate this new capability into a robust field NPV optimization platform. A two-stagefield-development optimization algorithm is developed in this work. First, the steady-state pattern is optimized using the field-scale pattern optimization algorithm while honoring field-scale constraints and using a combined surface and subsurface performance-indicator-driven objective function. Ramp-up pattern designs are optimized separately using a solely pattern-scaleperformance-driven objective function in this stage. A preliminary pattern-delay time optimization follows next to precondition the problem for the subsequent field-scale optimization stage. The ramp-up pattern and pattern-delay times are optimized using a constant steady-state pattern in the second step of the algorithm. An appropriately penalized field-NPV-based objective function is used in this step to enforce field-scale surface and subsurface constraints. Optimization results on a realistic example application indicate that the time to oil-rate plateau could be significantly reduced on the order of multiple years while honoring the surface production constraints. This requires the use of an optimized ramp-up pattern in conjunction with the optimal steady-state pattern. The ramp-up pattern is approximately two patterns wide and features an increased heater density to deliver production acceleration. It is also notably more robust against the effects of subsurface uncertainties.
Conference Paper
The effectiveness of a low complexity chemical flooding formulation that was developed for application in offshore environments was evaluated in a single well test offshore South China Sea. The SP formulation uses seawater with no additional water treatment beyond that which is normally performed for water flooding (filtration, de-oxygenation, etc.). The ability to use a formulation based on only seawater avoids water treatment and reduces complexity for commercial implementation. The Single Well Test was conducted in an existing producer with its wellhead on an existing production platform offshore. The injection facilities were placed on work barge that also had accommodation to house the people that executed the injection test. The facilities on the barge consisted of seawater lift and filtering equipment, surfactant and polymer mixing capability, and high-pressure pumps to deliver the mixed fluids through a flexible high-pressure hose to the platform. A laboratory to support the QA/QC of the injected fluids was also installed on the barge. Standard Single Well Chemical Tracer test procedures were applied to determine the remaining oil saturation after waterflood and after injection of the SP formulation. The Single Well Test was completed in accordance with the objective without any recordable HSSE cases ahead of schedule. The composition of the injected SP formulation was well within the specifications resulting in good injectivity of the formulation. The single well tracer tests conducted after the SP and polymer injection indicate that the residual oil saturation was reduced to values between 7-9% (Sorc). The main contributions described in the paper are: demonstration that a seawater-based SP formulation can be effectively mixed and injected from a barge offshore. the interpretation of the single well tracer tests conducted after the SP and polymer injection indicate that the residual oil saturation was reduced to values similar to those observed in core floods with the SP formulation in the laboratory.
Article
The subsurface hybridization of the in-situ-upgrading process (IUP) and steam injection (i.e., the hybrid-IUP/steam recovery scheme) integrates the advantages of process robustness and bitumen upgrading of the IUP with the relatively low capital-expenditure (CapEx) requirement and faster subsurface heat delivery of steam-injection-driven heating. The hybrid-IUP/steam scheme significantly reduces CapEx compared with the IUP recovery scheme, and increases robustness and reliability compared with the solely steam-injection-driven recovery processes because it is less sensitive to variability in geology and reservoir properties. We first demonstrate the advantages of the hybrid-IUP/steam recovery scheme using a mechanistic model for a real-life heavy-oil reservoir. An integrated surface/subsurface-economic evaluation using simplified economic indicators shows that the hybrid scheme has attractive attributes. Then, we further mature the hybrid scheme in terms of increasing its robustness with respect to subsurface uncertainties through a multirealization robust optimization workflow. By use of this workflow, pattern designs that are less prone to subsurface uncertainties have been developed, thereby improving the robustness of main subsurface-economic-performance indicators. Two promising design families and example designs stemming from these families have been identified that exceed standalone-IUP performance in terms of subsurface-recovery-performance indicators in a statistical sense. In this context, cumulative-probability distribution functions (CDFs) of main pattern-performance indicators have been computed using realistic dual-permeability/dual-porosity (DP/DP) mechanistic subsurface models by rigorously taking into account the effects of major subsurface uncertainties with a design-of-experiments (DOE) and Monte Carlo sampling-driven uncertainty-quantification workflow applied after robust optimization. The key conclusions of this study are the following: Hydrocarbon recovery from the upper section of the reservoir (formation above the heaters) is critical for driving the benefits of the hybrid scheme beyond standalone-IUP performance. The presence of a reasonable level of vertical fracture connectivity between steam-injection and electrically heated formations plays a significant role in the improved performance over the standalone-IUP recovery scheme. Shifting the steam injection to the same formation with the heaters results in suboptimal pattern performance for the investigated design families. Effective lateral heating with steam is important for the economic performance of the hybrid scheme. Fracture-architecture-related heterogeneities [namely, bed-bounded and bed-crossing (mesofractures)] play a significant role in lateral heating with steam (supported by electrical heaters). The optimal operation strategy features relatively early and high-rate steam injection as well as using 12 to 14 heaters, two producers, and one steam injector per a 92- to 110-m-wide pattern. Ceasing steam injection upon a significant amount of live-steam breakthrough at the producer is required during the main steam-heating period for optimal recovery performance. Injectivity plays an important role for the hybrid-pattern performance. A realistic enhanced-injectivity scenario noticeably improves pattern performance.
Article
This paper advances the understanding of foam transport in heterogeneous porous media for enhanced oil recovery (EOR). Specifically, we investigate the dependence of methane foam rheology on the rock permeability at the laboratory scale and then extend the observations to the field scale with foam modeling techniques and reservoir simulation tools. The oil recovery efficiency of conventional gasflooding, waterflooding, and water-alternating-gas (WAG) processes can be limited by constraints such as bypassing effects (including both viscous fingering and channeling mechanisms) and gravity override. The problem can be more severe if the reservoir is highly fractured or heterogeneously layered in the direction of flow. Foam offers the promise to address the three issues simultaneously by better controlling the mobility of injected fluids. However, limited literature data of foam-flooding experiments were reported using actual reservoir cores at harsh conditions. In this paper, a series of methane (CH4) foam-flooding experiments were conducted in three different actual cores from a proprietary reservoir at an elevated temperature. It is found that foam rheology is significantly correlated with the rock permeability. To quantify the mobility control offered by foam, we calculated the apparent viscosity on the basis of the measured pressure drop at steady state. Interestingly, the apparent viscosity was found to be selectively higher in the high-permeability cores compared with that in the low-permeability zones. We parameterized our system using a texture-implicit-local-equilibrium model (STARS™ simulator, Computer Modelling Group, Calgary, Alberta, Canada) to illustrate the dependence of foam parameters on rock permeability. In addition, we created a two-layered model reservoir using an in-house simulator called modular reservoir simulator (MoReS; Shell Research, Rijswijk, The Netherlands) to elucidate the role of different driving forces for fluid diversion at the field level. We took into consideration the combined effect of gravitational, viscous force, and capillary forces in our simulation. We show that the gravitational forces prevent the gas from sweeping the lower part of the reservoir. However, the poor sweep can be ameliorated by intermittent surfactant injection to generate foam. In addition, the capillary force which hinders the gas (nonwetting phase) from entering the low-permeability region can be effectively leveraged to redistribute the fluids in the porous media, resulting in better sweep efficiency. We conclude that foam if properly designed can effectively improve the conformance of the WAG EOR in the presence of reservoir heterogeneity.
Article
Full-text available
SCAL parameters (i.e., Relative Permeability and Capillary Pressure curves) are key inputs to understand and predict reservoir behavior in all phases of development. Techniques to measure relative permeability and capillary pressure have been well established and applied to a wide variety of core samples both from sandstone and carbonate reservoirs. On the other hand, we frequently encounter quality compromised data due to challenges in experimental procedures, lack of understanding of measurement techniques, and poor quality of raw data. As a result, relative permeability is often viewed as a parameter with large uncertainties and a fitting parameter in history matching. A special core analysis program was recently carried out on selected core samples from a deep-water sandstone reservoir in the Gulf of Mexico. In this frontier, relative permeability has been ranked among the top subsurface uncertainties. It greatly impacts the production forecast and field development plan. However, due to the high temperature, high salinity and fluid compatibility issues, the core measurements faced very specific challenges and a good relative permeability dataset has not been obtained in the past for this area. In this work, we demonstrate that a quality set of relative permeability data can be obtained through close collaboration across disciplines, a properly designed protocol, adequate engagement with the laboratory, timely QA/QC of experimental raw data, and appropriate interpretation incorporating numerical simulations. Well-defined and constrained relative permeability curve shave been derived with the combination of steady state and centrifuge techniques. The average trend can be described by a residual oil saturation of 22%, end-point relative permeabilities of 0.6 and 0.2 to oil and water, respectively and Corey exponents between 2 and 3.
Article
Full-text available
Enhancing recovery of the remaining gas in ageing fields presents a commercial opportunity. As such,NAMtook the decision to apply on-site generated N2 injection for the first time. The De Wijk gas field was selected based on size and geological setting, allowing the demonstration of N2 flooding of depleted gas in diverse reservoir types, as well as testing the efficiency of residual gas sweeping in the watered-out reservoir. De Wijk is one of the oldest onshore natural gas fields in The Netherlands, having produced c. 15 Bcm (billion cubic metres) of high calorific gas with a natural N2 content of 5-11% for more than 60 years, bringing it close to the end of field life. Enhanced gas recovery in this field would extend the field life and that of the Ten Arlo gas-processing system, thereby also lengthening tail-end production of other connected fields. In De Wijk, the existing 35 wells were used as dedicated injectors and producers. The production rates from the latter are used to help better understand reservoir flow behaviour. Integration of geological analysis and available production data helped to implement the N2 enhanced gas recovery technology in this mature field. Two processes of enhanced gas recovery were tested: a gas-gas displacement in the depleted gas leg (Nitrogen Assisted Depletion Drive: NADD); and gas displacing residual gas in the water leg (Nitrogen Enhanced Residual Gas: NERG). Observations after 1 year of N2 injection showed that planned v. actual performance of the NADD technology is comparable with an increase in production of 100 000-160 000 Nm³/day (normal cubic metre per day). NERG application restored a watered-out well; however, further investigation of this application is required. It is demonstrated that a detailed geological understanding of, for example, permeable Rogenstein oolite intervals and the lateral connectivity of internal high-permeability streaks in Volpriehausen sandstones is crucial to the successful application of enhanced gas recovery techniques in such ageing fields.
A fractured reservoir simulator capable of modelling block-block interaction
  • G J A Per
  • P M Boerrigter
  • J G Maas
  • A S De Vries
Per, G.J.A., Boerrigter, P.M., Maas, J.G., and de Vries, A.S.: "A fractured reservoir simulator capable of modelling block-block interaction", SPE 19807.
3-D visualization of multivariate reservoir simulation data on complex grids
  • M Wells
  • H Watkins
Wells, M. and Watkins, H.: "3-D visualization of multivariate reservoir simulation data on complex grids", presented at the 1992 IBM Europe Institute conference on Computational Methods and Tools in Reservoir modelting, 17-22 August 1992, Obcrlech, Austria.
Fractured Rmervoir Simulation: Case Histories
  • P M Bccrrigter
  • L E C Van Der Lcemput
  • J Pieters
  • K Wit
  • J G J Ypma
Bccrrigter, P.M., van der Lcemput, L.E.C., Pieters, J., Wit, K., and Ypma, J.G.J.: "Fractured Rmervoir Simulation: Case Histories", SPE 25615.
The integrated Team Approach to the Optimization of a Mature Gas Field, paper SPE 26144, presented at the SPE Gas Technology Symposium, Calgary, 28-30
  • H R Hooi
  • L Goobie
  • R Cook
  • J Choi
Hooi, H.R.., Goobie, L., Cook, R. and Choi, J., The integrated Team Approach to the Optimization of a Mature Gas Field, paper SPE 26144, presented at the SPE Gas Technology Symposium, Calgary, 28-30 June 1993 @mceedings pp. 73-80)