Problem Sets
Introduction
This chapter deals mostly with problem sets that the
author has created for use in industry short courses.
‘They have been used many times, and potential bugs
should be resolved. The data sets have been prepared for
both Computer Modelling Group and Geoquest software.
Originally, there was going to be a copy of all the data
sets included in the back of the book. However, the book
‘was getting somewhat long and it would have taken a
considerable period of time for readers to reproduce the
data sets manually For these reasons, the data sets will,
instead, be located on a website at hup//vwwwappliesd:
reservoinengingering com. The files can be downloaded
from this website free of charge. This will also allow for,
future revisions, additional exercises, and perhaps differ
ent data set formats,
‘There are also some final thoughts on reservoir simu:
Tation at the end of the chapter
Grid Problem Set
This probiem set is designed to illustrate the effects of
grid density on the accuracy of solutions obtained from
‘numerical simulation, This is a simple example that uti-
{izes a fivespot waterllood pattern, perhaps one of the
most common situations tat one may encounter. Pay
particular attention to the water cut performance as the
grid becomes progressively finerPractical
Reservoir Simulation
This is a relatively simple problem set, The second
objective is t allow the reader to become familiar with
2 simulation data set, since no simulation experience
is assumed
Element of symmetry
Only a portion of a vomplete fivesput has been od
eled. This saves on the amount of computing time and
memory required. The basic grid for he frst run is shown
in Figure 201
The element of symmetry has alsu been extended to
the edge of the model. The sielator assumes that the
Wells are located in the center of the grid block. This
degree of accuracy allows simulator results to be com
pared against analytical solutions, although this is not
covered in this chapter.”
‘Two changes are required 10 effect comparison with
analytical solutions. First, the permeability of the edge
blocks must be cut in half, since the outside portion does
not exist. In addition, the porosity of the edge blocks
must be halved also, to give the correct total pore vot
lume, Note also that, by applying the 50% factor to per
se oo
wiwsedel $- ayot?
" #
\
meabilities and porosities on all four edges, the comer
blocks will get this treatment twice. Hence, the porosity
and permeability in the comer blocks will be 2:
ir original values.
Changes required
Required changes have been highlighted in the fo
towing section. Note that only afew qumbers need 10 be
changed in the data set
Analysis
Although this problem set may seem simplified, it
contains some important implications. A number of peo-
ple use a minimum of two grid blocks between wells as a
rule of thumb for grid definition. The author suggests that
this is not a good criterion, as explained in chapter 10.
The simulator provides comparable results on recov
ery factors; however, itis apparent wat very different
results are obtained for water breakthrough times. Often,
this is an important factor in determining pore volumes
during history matching
Fig, 20-1 Idealized Blement of SymmeoyConing Problem Set
This prubiom set hos a number of interesting twists. ft
highlights the differeuces in behavior between gas and
water coning, The final solution is semewhat counterine
jitive. It is a dramatic example of why datum pressure
should be used to analyze flow in the reservoir I is easy
to muve abead btindly using straight pressure rather than
actual tlow potential (datum pressure). Recall that pote
tial includes the appropriate hydrostatic head—ie., the
gravity term —with the respective ptiase pressures.
Constraints
‘The objective of this problem set is to
maximize oil recovery 00 an offshore plat
form, There is a water cut restriction due
Process
The author recommends using a graphical approacl
to understanding the probleyn, This meats exaluating the
aduetion and then plot both
pressure and datuin pressure, What is dominating the
reservoir performance?
flow in the reservoir Plot j
Grid
Note that a radial grid has beer used, Most simulators
hhave this capability Radial models can be very elficient
forsinglewell probleins
con PROMEM SET
to limitations with the topside facilites.
The water cut must be kept below 20%, '
Further, the topside facilities are limited to
a GOR of 750 mim’, The only variables
that can be changed are the location of
the perforations and the bottombole flow:
ing pressure. These items have been high
lighted in the data set.
Description
Hid
‘The physical reservoir properties are
shown in Figure 20-2. The reservoir has
+
been divided into layers. Note that the
formation permeability and porosity vary
vertically with depth. There is also a gas-
ve
oil and a wateroil contact, which are 5
located at a depth of 1,500 and 1,359 Lt
meters, respectively
As with the earlier data sets, the fluid 4
properties and rock properties, such as
capillary pressure and relative perm:
ability remain the same. Note also that
depths and pressures are the same, or very
similar, throughout all of the problem sets, |
Fly, 20-2 Coning Model Grid
> 508Practical Reservoir Simulation
Results
The resutts of your various runs should be tabulated as
shown in Table 20-1
In addition, make a description of the reservoir
physics. What are the relative coning susceptibilities of
gas versus water?
Analysis of coning problem
A production plot shows that the peak GOR produc
tion does nor occur at the end of the run. Therefore, this
{raint myst be plotted against time.
The reservoir flow developed from raw pressure is
very misleading when compared to the potential or pres:
sure datum approach. Cleatly, the high-permeability lay
ers act a5 a conduit, Gas is also more prone su coning
thao water, Despite the larger density difference, the gas
has a significantly Jower viscosity: The latter dominates
the propensity t0 cone gas ahead of water In fact, in
is possible to inctease recovery by com
pleting wells in the water leg and aot the vil section (as
indicated in chapter 16 in the section dealing with water
ancl gas cuning, this is a technique known as “reverse
Reservoir simulation is usually associated with study.
ing areal movements of fluid and secondary recovery
processes—"big picture” issues. This proble
strates that it also can be useful for completion design
set demon:
Table 20-1 Blank Table for Recording Completion Sensitivity Results
M3/M3
Base
col afor| al e}olre|—Problem Sets
Cross-Sectional Study Problem Set
This is the most difficult exercise, since pseudo rela-
tive permeability is the most dificult concept in the text.
Io the authors opinion, pseudos are widely misunder
stood. In fact, aithough this text covers relative perme-
ity and pseudo relative permeability concepts in
chapters 7 and 8, respectively, there's nothing like putting
these concepts into practice to really get a good feel for
them, The author has found that itis easy to look at the
forrmulae and gain a general idea from a textbook: it is
another matter entirely to actually implement pseudo ret
ative permeability correctly in a model, Aside from the
basic experience, it provides another valuable insight. It
ispossible to duplicate the results ofa detaited simulation
with accuracy using pseudo relative permeability. This
exercise will give the reader an instinctive feel for what
can be done.
Process
‘A detailed sixtayer run is provided to start. This is a
reservoir engineering problem.
|. Fist, understand the reservoir from 2 qualitative
perspective. This means understanding the
structure, the layering, and the type of fluids
in place.
2. Next,run the program provided.
3. Transfer the production results to graphical for
mat. While a graphical postprocessor can be
used, readers are advised to plot the results by
hand.lt gives one more time to think about
What is happening in the reservois Using a
postprocessor may see to be more exciting
or clever; however it can lead to shortcuts
in analysis.
4, Observe how the saturations change in the
reservoir What is actually happening?
A graphical postprocessor greatly speeds
up this process. Note that these sections
also can be plotted by hand, and this is
appropriate at times,
5. Write a short description of the relevant
reservoir mechanisms,
6, Develop wateroil ck pseudo relative permeabil
{ies.This is done on a grid column by grid cot
tumn basis. specific example of how to do this is
outlined in the following section. Inthe author's
experience. translating the theory to practice in
this area is the most uifficult
Simplify the model from a sixlayer model toa
single-layer model, Fist, an average permeability
and average porosity must be generated. Second,
an average capillary pressure curve is developed
In this simplified example,each layer has the
same S, and P. curve This would rarely be the
ccase in actual practice.A set of rock pseudo rela
tive permeability curves needs to be developed,
ie.k,, and &,.,. Were simplifications required in
this process? How certain is the reader of the
general applicability?
8. Run the new single-layer model. Plot the results
and compare them to the first model run with six
layers. How do the results compare?
9. What happens ifthe model is run without the
pseudo relative permeability curves? Recall that
each of the layers actually utilized only one
unique (lab) relative permeability relationship. it
this run doesn’t match the sixlayer model, how
much is it off” How severe isthe layering? Can a
Dykstra Parsons ratio be calculated, and how
severe is this compared to most reservoirs? What
does this tell us about the relative permeability
‘curves that we actually use or history match in
reservoir simulation?
10. Finally, develop pseudo relative permeabilisy
curves for the wells. Tis takes into account local
wellbore effects. Note that different curves may be
required for all three wells.
11. Ifone were completing a study.one more step
‘would be required, which is to transfer al this
information into a single-layer reservoir model
‘What other issues would have to be resolved?”Pract
I Re
rvgir Simutation
Cross-sectional analysis
In this problem set, an apparently simple problem set
actualy requites a great deat of detailed analysis for cor
rect interpretation, The author recommends that. (0 truly
understand this material, more is required than simply
reading or teviewing the solution that follows.
Saturations show that the water fingers up the high
permeability streak in the middie of the zone, as shown
in Figure 20-3
Gas breakout occurs atte top of the structure, This
gives a high GOR ia the top well, The production per
formance for the wells s shown in Figure 20-4
‘The wells show increasing water production, which
related to flank water influx
The pressure distribution in the reservoir is shown ia
jure 20-5, The averagé pressure drops in the first five
years, After this, the aquifer maintains the pressure at
approximately 7400 kPa,
The production performance for the field is shown,
in Figure 20-6. Oil rates, water cuts, and GOR trends
are indicated
lk may be concluded from this that a formation or
rock relative permeability characterization will not be
sufficient. The effects of partial completion must atso
be included.
Less use of pseudo relative permeability is made today
than in the past. However, on large simulations, this may
still be a practical necessity The generation of these
curves is time intensive. Ulilities are available to generate
these curves. Note that shifting the end-points on the
pseudo relative permeability is often sufficient to achieve
the required result
This is by no means the end of the problem. After
this, it would be necessary to include these curves in an
areal simulation. If there is cross-flow into the cross sec
tion, the history match generated may aot be represent
tative, This has occurred in practice on studies per.
formed by the author.
‘This problem set also demonstrates another benefit
that can be derived from using a crosssectional study,
‘The problem has been reduced in complexity, which
allows one to concentrate on understanding the basic
mechanisms in the reservoir.
ig 20-3 Cross-Sectional Model—Viscous FingeringFig. 20-4 Cross Sectional Model—Gas Breakout
Fig 20-5 Cross-Sectional Model—Reservoir PressurePractical Reservoir Simulation
Fig, 20-6 Original Reservoir Model with Si Layers
UE the reader were actually doing this study for an
employer ora client, what would be written up regard-
ing the risks and assumptions of the results? The pseu-
do curves are also rate dependent. How sensitive are
the curves to areal variations in rate? One of the major
reasons for implementing simulation is waterllooding
Government conservation regulations historically
allowed one to produce at higher rates when the proj
ect was recognized (ie., accredited). Typically, the per-
mitted rates were doubled. Do the relative permeability
curves used apply to the new rates? This would mean
changing the relative permeability curves on the water
injection sensitivities, In some cases, the reservoir sim-
ulator will not allow one to change the relative perme-
ability curves during the run section and is usually set
inthe initial conditions.
sree
In a more modern context, 34D simulations are the
norm today However, many people formed their percep-
tion of simulation results when the use of these danger
‘ous shortcuts were necessary. Note that upscaling can
make use of pseudo relative permeability so these issues
are lar from dead. Most upscaling is developed assuming
single-phase flow in the authors opinion, many of the
‘current “discoveries” in upscaling were previously found.
«with pseudo relative permeabilityAreal Problem Set
This problem set isthe author’ favorite, Its not a tite
exercise, and is based on some direct experiences in
Halésay Comiation name) reservoirs in northwestern
Alberta and cortheastert British Columbia, The abjec:
tives are outlined in the following
1 Demonstrate the extent to which outside inthy
jeces affect the development outcome of the
resersvitThis makes the exercise very interesting
in that it appeosimates the actual development or
exploitation process.
2. Provide input on a crossdiscipline basis
Successful simulations are strongly influenced
by geological input and impact development
decisions The ontcome can have a strong intlue
ence on future locations or may provide val
able insight into what is fundamentally a geo-
logical focus.An example of this is the exter
sion of a pool. Simulation input and results
‘often influence production engineering isues.
such as well completions
3. Keep a business focus. Historically, many people
involved in reservoir simulation have beer the
‘more technically oriented engineers. Very often,
management really wants economic iemptica-
tions zo be considered at the technical level,
This may be as simple as including the eco-
‘nomics in a format that can be used readily in
an economics model
4. Get the geological model right.This is the most
dificult par ofthe simulation. This example is
very stylized in thatthe input was created in an
orderly fashion. Having said this, the model has
sone efements of actual reservoir complexity.
Although this is substantially less than reality the
problems of geological description are apparent.
To reiterate, this exercise affords an ideal opportunity
to make mistakes, The results are not reported to anyone,
and there is no real money on the line.
Outline
The genera idea is this. Assume that you are operat
ing a small oil company and that you ane allocated a cer
tain amount of money from which to obtain land and
drill. A reservoir code! has beet built, ane well results
are obtained when money ts spent. fn acklition, you can
derive revenue fiom pavduction with which to drill fur
ther wells or purchase fad. You will be provided with a
generalized play and some seismic data, The final objec
live isto devise a geological map aud develop a reservoir
simulation model.
Financial
You company is awarded! 5.000 KS to start. Oil prices
are $24.90 per barrel, and the royalty rate is a lat 2
‘The gas price is $2.50 per scf with a 15% royalty rate
Operating costs are $2,500/well/month, regardless of
whether isan oll or gas well
Geological model
‘The basic geological concept is shown in Figure 20-7
(heavy. grey lines are lag shales). The sands are a combi
nation of shoreface and channel sands. In the problem
set area, the sea was located towards the bottom of the
maps andlor gris when this reservoir was being deposi
ed. Geophysics were run with the results shown in Figure
20-8, This isan amplitude variation map and is somewhat
representative of porosity and sand thickness develope
ment
Land sales
‘When this exercise is used as part of a course, land is,
sold via auction in the strle adopted by the western
provincial governments. This tansforms this problem set
into a competitive game. Its therefore possible to have
one person assume the role of game master and others
can competitively model/develop the reservoir
“The land sale is very bureaucratic. Par of this reflects
the realities of dealing with government, and part of itis
pure administration. Bids must be provided on quarter
section by section data using the forms provided and
blue ink! The frst time the author attempted to do this
exercise, several hours were required to determine the
results from the fst lan sale, The form in Table 20-2 wasPractical Reservoir Simulation
Beach Swale
Wash- inlet ridge Barrier
7 island
Shoreline
‘\\ of ocean
\
Outer shelf
Fig, 20-8 Geophysical Amplitude Survey Results, Areal Problem SetTable 20-2 Official Government Form for Submitting Land Bids (Read instructions carefully and literally)
Offer to Purchase Land From Government
(use red ink)
Section Number
LSDs (circle)
Nef
SE sw nw
Amount, g
Signed
Company
adopted so that each parcel of land could be rapidly sort
ed during lunch. This greatly facilitated the progress of
the game
Note that land! is sold by the quarter section. The avail-
able leases are shown in Figure 20-9. Pay careful atten
tion to the following section on permitted drilling and
production spacing locations.
Well drilling and completion costs
Wells cost an even 1,000 KS and completions 250 KS.
Well dling locations must be specified exactly on the
grid and shown diagrammatically. It is also possible to
obtain core data at the cost of 50 K$ per well.
Drilling locations and
production requirements
Wells must be drilled in the
specified target areas shown in
Figure 20-10. Although wells may
be drilled in any of the target
areas, oil wells can only be pro-
duced from the oll targets and gas
Caveat emptor
Bad things happen in wea fife, and this exercise is no
exception, Fist, dry holes are quite possible and AEUB
regulations are generally in etfect. You may get oil ga, or
water You may drill wells with no reservoir development
or that have poor permeability
Reservoir engineering data
All reservoir engineering data is provided ia ehis exer
cise, ast was in the other examples, This has been done
to concentrate on the geological model, The Core poros-
ity versus permeability is shown in Figure 20-11 for both
the channel and beach sand facies, A base map is also
provided for mapping as shown in Figure 20-12, A num
ber of maps will be required and photocopying the vari
‘ous maps is suggested by the author
swells can only fe produced from
the gas targets. There is an area of
overlap between these targets.
‘These four areas within a section
are ihe safest place to drill,
‘The government (game-
master) may allow you to pro-
duce out of target upon special
application, provided no other
operators have a bona fide
objection. The government is
conscious of maximizing royal
ties. Such applications normally
invatve costs.
Fig 20-9 Land Parcels sealable for SaiePractical Reservoir Simulation
518 <
Fig, 20-10 Allowable Driling Locations
PRS Areal Problem Set
no
ce
Fh
S| ok tHeeers
ia a
=| Gas TRREETS
—* Bar Sand
3 10+
3
£ 1 —® Channel |
5 Sand
a
0 0.1 0.2
Porosity (fraction)
Fig. 20-11 Core Data for Areal Data Set
03
Well results
(confidentiality)
When a well is drilled, the
logs and production results are
held confidential from your
competitors for a period of one
year. During a fivelay course,
this equates to one day. Alter
that, a copy of your logs will be
distributed to your competitors
This means that, at any given
time, every group in the class
will be dealing with a different
set of data
Analysis
This is a nasty problem, To
date, the author has not heard
of anyone who has solved it and
correctly determined the geols-
ay. Note in particular that there
are three overlapping sands.
This problem is discussed in
chapter 181i the section dealing
with structural style effect, The
extent of the channels haz not
been completely identified
Moreover, this problem set is
highly ideatized
The contacts are the same
across all three sands. They are,
in efect, in partial communica-
tion by virtue of the intersecting
channels; tis is known as juxta-
position, The geological model
was based on actual experience
and is representative of an area
in Northern Alberta. tn the
authors opinion, itis impossible
to solve this problem without a
geological model
The problem set slike real lte
in thata partial set of information
is available. Unlike a real reser
soir, a correct solution is known.
This is an interesting control pos
sibility Can we actually deter3
[.
| * cf:
[EE
Fig. 20-12 Base Sap of Areal Problem Set Study Area
mine the right answer with simulation? The author doubts
that itis possible to arrive at a correct, or more to the
point, a unique answer
At least one group teking the course went broke
drilling in the gas cap. They were shut in by the oil oper
ators, which is consistent with Alberta oil and gas con
servation laws. One of the participants worked for the
AEUB, He was none too pleased! Other groups went
broke drilling in the water leg or in low-permeability
areas, This is an interesting aspect as well. Winning the
game is not just a matter of technical skill. One must
assess risk and take appropriate hedging action. The sit
uation is competitive,
One other issue worth investigation is What is the
optinal recovery scheme? Some areas are low perme:
ability that do not make for economical wells. The prop:
er location in channels would indicate intuitively a bet:
ter development plan.
Each time the author has observed others attempt to
solve this problem, there have been different develop-
ments, This is an artifact of history and chance. People
interpret the seismic differently and use different profiles
for drilling. [t would be interesting to investigate how
much variance would occur in recovery oF net present
value (NPV) based on different development scenarios.
One branch of statistical analysis allows us o evaluate rel
ative probabilities by combining a grouping ol experts
analysis. This would be an interesting investigation also,
During one course offered by the author, there was
sufficient time for the groups to prepare a report. The pro-
duction forecasts for the three different groups were
entirely different! This was used for course evaluation,
The scope of the report was to recommend a develop-
ment plan based on mapping and reservoir simulation.
‘This course was attended by several complete exploita-
tion teams
Procedure
‘The author has worked to simplify the procedure so
that it is possibfe to enter the geological input more quick-
ly without having to resort to a contour package. This, like
real life, isa major part of simulation input.Practical Reservair Simulation
Beyond This Book
In this book, the author hopes to have changed the
reader's attitude towards simulation. Getting good results
is a matter of putting together many dierent elements
and disciplines. Simulation technique is not enough. In
the authors opinion, the biggest hang-up usually comes
from not having a good visualization or sufficient geolo-
gy. Ideally the techniques and details of running the sim
ulator are second nature, and the engineer can concen
trate on getting the correct input. This is not realistic, ia
particular for the first few simulations. Cheat (fairly) by
reading geological papers on reservoirs similar to the one
being modeled, by reviewing technical papers on offset
reservoirs, by reading all government applications, by
using a systematic approach, by careful data organiza:
tion, and by setting aside time to specifically review glob-
al issues. Its not easy. and it takes time to do it correctly,
Technique
Many techniques have been developed in the field of
reservoirengineering. itis not possible to remember all of
the information that is available. Even Albert Einstein is
said to have looked up his own number in the telephone
book. The person who asked for his number was 50 sur
prised that they asked him why he had not committed it
to memory His reply was to the effect that he had enough
important things (theoretical physics) to remember. He
‘was happy simply to know where to find the rest.
Develop a set of reference materials. Ata minimum, a
practicing reservoir engineer should have one or two of
the classic reservoir engineering textbooks and several of
the SPE and Petroleum Society of CIM monographs.
Continuous development
Some areas of reservoir engineering are still progress.
ing rapidly, Such areas include pressure transient analysis,
compositionat behavior, naturally fractured reservoirs,
and reservoir characterization with geostatistics. These
‘areas require that one read some of the professional jour-
nals such as the Journal of Canadian Petroleum
Technology (JCPT), Journal of Peiroleum Technology
UPN), Society of Petroleum Engineering Journal (SPE),
and SPE Reservoir Evaluation & Engineering (SPEREE), At
‘one time, operating companies supported membership
in a reasonable number of professional societies. This is
1no longer the case. The author recommends that at least
an engineer belong to APEGGA, the Petroleum Society of
CIM, and the Society of Petroleum Engineers,
Contacts
Finally, friends may be able to help. This will not
always be possible since they may be working for com
petitors. The company may have a few gurus in some
areas of technical reservoir engineering, geology, or sim
ulation. These people are amazing Very often, they
remember details of wells drilled more than 20 years ear-
lier and will recall why certain decisions were made
‘There are not as many of them as there used to be, but
few are known to the author.
An engineer can hire a consultant to critique the
approach. Some consultants are reluctant to do this ince
they want to do the work. On the other hand, one very
good log analyst prefers to advise consultants. A
Canadian heavy oil expert often comes on site for a day
and reviews projects or data. Apparently. his suggestions.
can keep a number of people busy for a considerable
period, even when the project is at a mature stage, Many
consultants, including the author, do not abject to pro-
viding advice as long as they are paid for the time
expended,
Conclusion
Itis hoped the reader found the material in this book
to be useful. Please contact the author if you find any
errors (there are some undoubtedly). Good luck with
simulation studies that you do, have to review, or that you
commission others to do.