and optimization
A
Research Thesis
Presented to the Department of Petroleum Engineering,
African University of Science and Technology,
Abuja
By
RECOMMENDED:
________________________________
________________________________
________________________________
________________________________
APPROVED:
________________________________
Supervisor:
________________________________
________________________________
________________________________
Date
ABSTRACT
Well stimulation consists of several methods used for enhancing the natural producing ability of
the r eservoir when p roduction rate declines. A de tailed l iterature r eview of s ome of t he well
published stimulation models are discussed in this research. This d iscussion wa s preceded wi th
an introduction t o f ormation damage concepts and an o verview o f well stimulation m ethods.
Production decline curve analysis is combined with economic discounting concepts to develop a
model that can be used for optimizing stimulation decisions. The model is presented in the form
of a no n-linear programming pr oblem subject t o t he constraints imposed by t he p roduction
facilities, reservoir productivity and the stimulation budget approved by management. Production
data from f our stimulation ca ndidate wells, o ffshore Niger Delta was used to validate the m odel
developed by s etting up a m aximization problem. S olution to the p roblem w as ob tained using
non-linear o ptimization software. The r esult o btained was v erified u sing Wolfram R esearchs
Mathematica 7.0. The results s how that the o ptimization m odel c an be c ombined w ith
stimulation t reatment modules, de veloped f rom i ndustry w ide models, t o q uantify s timulation
benefits. C andidate w ells w ere t hen r anked ba sed on stimulation c ost, p ayout t ime a nd
stimulation b enefit. Hence, th e m odel i s valid f or stimulation ca ndidate s election; and i s
therefore recommended for use in optimizing stimulation decisions.
DEDICATION
This research is dedicated to my Lord Jesus Christ who has been, and will ever be the best role
model anyone could find. And also, to the good people of the Niger Delta.
ACKNOWLEDGEMENT
I wish t o sincerely a ppreciate G od Almighty for H is l ove, c are a nd wonderful works t hat a re
made m anifest i n m y life each da y. Also, m y s incere thanks go to my supervisor, Prof.
(Emeritus) David O. Ogbe for guiding me to success in this work, Dr. Samuel Osisanya and Prof.
Peters Ekwere f or s erving in m y thesis committee, and m y m other, M rs. G race Ugbenyen f or
being there always for me.
The following persons, among others, who contributed in no small measure to the success of this
work deserved to be acknowledged.
My friends: Lymmy B ukie O gbidi, Akpana Paul, R aymond Agav, Habibatu Ahmed, and
Christopher M udi who paid m e several v isits a t A UST t o c heer me u p. T he members o f H ope
Hall Parish, Redeemed C hristian C hurch of G od, Galadimawa, Abuja, who have always been a
warm family to me. Nature will not forgive me if I fail to thank Miss Esther Akinyede who was
kind to provide me with a laptop to continue this work when lightning storm damaged my laptop
on 14 th July 2010 a t Julius N yerere Hall, AUST, Abuja, and I got no help from t he University
even t hough I pl eaded f or assistance. I w ill n ot f ail to m ention Mr. Alfred Emakpose who
assisted me in no small measure to keep things straight when the odds were against me. Finally, I
would like to thank my wonderful new friends, who would be mad at me if I fail to mention their
names; Hatem, Adel, Amar, Fauzan and Andrew, who are here with m e as I write these lines at
The Beaches Hotel, Prestatyn, North Wales, where I neglected some of my schedule to put most
parts of this work together.
TABLE OF CONTENTS
ABSTRACT......iii
DEDICATION.....iv
ACKNOWLEDGEMENT....v
TABLE OF CONTENT...vi
LIST OF FIGURES......x
LIST OF TABLESxi
Matrix Stimulation....2
Fracture Acidizing........4
1.3.3
Hydraulic Fracturing.....6
1.3.4
Recompletion....7
LIST OF FIGURES
3.1 Production Decline Profile for a Stimulated Well. .45
4.1 Effect of oil price on the objective function ...60
4.2 Effect of discount rate on the objective function ....61
4.3 Effect of decline rate on the objective function ..62
4.4 Effect of pre-stimulation production rate on the objective function....64
4.5 Effect of abandonment rate on the objective function ............65
4.6 Effect of stimulation time.....67
4.7 Cost Versus Productivity Ratio Plot for Well BU 3.....71
4.8 Effect of oil price on Well BU 3 model result...76
4.9 Cost Versus Productivity Ratio Plot for Well BU 5 .79
4.10 Effect of oil price on Well BU5 model result.83
LIST OF TABLES
Table 4.1:
Table 4.2:
Table 4.3:
Table 4.4:
Chapter One
Introduction
1.1
1.1.1
The skin effect can be o btained from a w ell te st. It m easures t he extent of da mage in t he n earwellbore z one. The total skin effect obtained from the well test is a composite parameter which
consists of s kin c omponents d ue to mechanical c auses a di sturbance of t he fluid f low
streamline n ormal t o t he w ell, o r formation damage - alteration o f t he natural r eservoir
permeability. It is very important to be able to identify the formation damage component of the
skin s ince t his c an b e r educed by b etter operational practices, or possibly, b e r emoved or
bypassed by stimulation treatments. Formation damage can result from many different operations
such a s dr illing, cementing, perforating, completion/gravel pa cking, production, i njection,
workover, stimulation, etc.
1.2
Well stimulation is a way of increasing well productivity by removing (or by passing) formation
damage in t he n ear-wellbore r egion or by superimposing a highly conductive structure onto the
formation.
Technical Objectives
Remove, reduce or b ypass t he f ormation damage, reduce sand production and cl eaningup the perforations.
Economic Objectives
Increase flow rate and optimize production from the reservoir.
1.3
Several stimulation t echniques e xist bu t t he commonly u sed methods i nclude matrix a cidizing,
fracture a cidizing, fracpack, ex treme o verbalance operations and hy draulic fracturing. These
methods h elp t o optimally increase well or reservoir productive c apacity by providing a net
increase in the productivity index. This increase in productivity index can then be used either to
increase t he p roduction r ate o r t o d ecrease the dr awdown pressure differential. Increase i n
production rate will eventually increase productivity. A decrease in drawdown can help prevent
sand pr oduction and water or gas c oning and/or s hift the p hase e quilibrium in the near-wellbore
region t owards s maller f ractions of condensate. Some of the m ost c ommon s timulation
techniques are discussed in the following sections.
1.3.1
Matrix Stimulation
Matrix stimulation is injecting an acid/solvent into the formation at below the fracturing pressure
of t he formation to d issolve/disperse materials th at im pair well production i n sandstone
reservoirs or to create new, unimpaired flow channels in carbonate r eservoirs. Mineral acids are
most c ommonly us ed in matrix s timulation hence t his t echnique is f requently ca lled ma trix
acidizing. Matrix acidizing is a near-wellbore treatment, with all of the acid reacting within a few
to perhaps as much as 10 ft of the w ellbore in carbonates. Matrix a cidizing l ower permeability
limit is 10mD for oil wells and 1mD for gas wells.
In s andstone, only a small f raction o f the m atrix i s soluble hence r elatively s low r eacting acid
dissolves the permeability-damaging minerals. Carbonate formations are different in that a large
fraction of the matrix is soluble (usually > 50%), hence acid will react rapidly with flow channels
and pores and creates new flow paths by dissolving the formation rock.
As a rule o f thumb, matrix acidizing i s a pplied only in situations where a well has a large skin
effect t hat cannot b e attributed t o mechanical, o peration o r surface p roblems. The r emoval of
damage by matrix a cidizing r equires t hat t he t ype ( or c ause) a nd location of t he damage be
identified before its removal is attempted. The damage identification process involves:
Carrying out specific laboratory testing, such as a reservoir core flushing, to determine if
the identified operations did indeed lead to core damage for the particular combination of
the fluids in question and the reservoir formation
wellbore pressure until it is large enough to overcome compressive earth stresses and the rocks
tensile strength. At this p ressure, t he r ock fails, a llowing a c rack ( fracture) t o be formed.
Continued fluid injection increases the fracture length and width. The injected acid differentially
etches t he formation fracture faces as it r eacts, r esulting i n t he formation of h ighly c onductive
etched channels that remain open after the fracture closes. Two procedures are commonly u sed.
Acid alone i s i njected, or a fluid ( called a pad) that will create a lo ng, wide fracture is injected
and followed by a n a cid. A c onventional fracture a cidizing treatment involves pumping a n a cid
system after fracturing. It may be preceded by a nonacid preflush and usually is overflushed with
a nonacid fluid.
Acid s olubility of th e f ormation is a key f actor i nfluencing w hether f racture acidizing or
proppant treatments should be employed. If the formation is less than 75% acid soluble, proppant
treatments should be used. For acid solubilities between 75 and 85%, special lab work can help
define w hich a pproach should be u sed. Above 85 % acid solubility, fracture acidizing would b e
the most effective approach.
Treatment v olumes for fracture a cidizing a re m uch l arger t han for matrix acidizing t reatments,
being as high as 1,000 to 2,000 gal/ft of perforated interval.
As a ge neral guideline, fracture a cidizing i s us ed on formations with > 80% hyd rochloric a cid
solubility. Low-permeability carbonates ( >20 md) a re t he b est candidates for t hese treatments.
Fluid loss to the matrix and natural fractures can also be better controlled in lower permeability
formations.
The su ccess of t he acid f racturing treatment depends on two ch aracteristics o f t he etched
fracture: effective fracture length (which is a function of the rate of acid consumption, acid fluid
loss ( wormhole formation) a nd acid convection a long t he fracture) a nd e ffective fracture
conductivity (a function of the etched pattern, vo lume of r ock di ssolved, r oughness of etched
surface, rock strength and closure st ress). The acidized fracture length and fracture conductivity
are therefore controlled largely by the treatment design and formation strength.
1.3.3
Hydraulic Fracturing
Hydraulic Fracturing consists of pu mping a viscous fluid at a sufficiently high pressure (greater
than the formation fracture pressure) into the completion interval so that a two winged, hydraulic
fracture is formed. This fracture is then filled with a high conductivity, proppant which holds the
fracture open (maintains a high conductivity path to the wellbore) after the treatment is finished.
Propped hydraulic fracturing is aimed at raising the well productivity by increasing the e ffective
wellbore radius f or w ells c ompleted i n low p ermeability c arbonate or clastic f ormations.
Hydraulic fracturing i s t o improve productivity i n l ow-permeability f ormations, or to pe netrate
near-wellbore damage or for sand control in higher permeability formations.
Hydraulic fracturing is a mechanical process hence it is only necessary to know that formation
damage is present when designing such a treatment. When a well is hydraulically fractured, most
pre-treatment skin e ffects such a s f ormation da mage, perforation skins a nd s kins d ue t o
completion and partial penetrations are bypassed and have no e ffect on the post-treatment w ell
performance. Phase-and r ate-dependent skins effects a re either eliminated or c ontributes i n the
calculation of the fracture skin effects. Generally pre-treatment skin effects are not added to postfracture skin effects.
Hydraulic fracturing differs from fracture acidizing in that hydraulic fracturing fluids usually are
not c hemically r eactive, a nd a pr oppant i s placed i n the f racture t o keep the f racture open and
provide conductivity.
The Inflow P erformance of a Fracture Stimulated w ell i s c ontrolled by a q uantity k nown a s t he
dimensionless fracture conductivity which depends on the fracture permeability conductive
fracture w idth, f ormation permeability and the conductive fracture single wing length. The
fracture c onductivity i s i ncreased by an i ncreased fracture width, a n i ncreased proppant
permeability ( large, more spherical p roppant grains ha ve higher permeability), a nd m inimizing
the permeability damage to the proppant pack from the fracturing fluid.
Propped hy draulic f racture w ell s timulation s hould onl y be c onsidered when the: well i s
connected to adequate produceable r eserves; reservoir pressure is h igh e nough to maintain flow
1.3.4 Recompletion
For wells with certain t ypes of da mage such a s pa rtially or t otally p lugged p erforations,
insufficient perforation de nsity o r low de pth of pe rforation, it m ay b e s ufficient t o r ecommend
recompletion technique. Hence the idea of recompletion is to increase the perforation density or
to increase the depth of perforations. The overall aim of this method is to increase production by
bypassing t he da mage. R ecompletion i s a lso u sed effectively in reducing w ater p roduction. I n
this approach t he w ell i s re-perforated at a new hi gher z one w hile t he pe rforations i n t he wa ter
zone are plugged off.
NPV is the difference between the present value of all receipts and costs, both current and future,
generated a s a r esult of t he stimulation treatment. Future r eceipts a nd costs a re converted i nto
present va lue u sing a discount rate a nd taking i nto a ccount the y ear in which t hey will a ppear.
Another measure of t he economic e ffectiveness i s t he p ayout period (PO); t hat is, t he t ime i t
takes for the cumulative present value of the net well revenue to equal the treatment costs. Other
indicators i nclude i nternal rate of r eturn (I RR), profit-to-investment ratio ( PIR) and gr owth rate
of return (GRR). The NPV (as well as other indicators) is sensitive to the discount rate and to the
predicted future hydrocarbon pr ices. A s with a lmost a ny other e ngineering a ctivities, costs
increase almost li nearly with t he size of t he stimulation tr eatment but (after a certain point) t he
revenues increase onl y marginally or may even decrease. This suggests that there is an optimum
size of t he t reatment t hat will maximize t he N PV. Hence it i s i mportant to select stimulation
candidate wells that have potentials for maximum benefit.
Candidate Selection (Recognition) is the process of identifying a nd selecting wells for treatment
which have the capacity for hi gher production and better economic return. Hence in stimulation
candidate w ell s election, t he w ell s timulation treatment yielding the hi ghest di scounted rate o f
return is the treatment which, in principle, should be carried out first.
1.6
recommending stimulation treatment option and optimizing the stimulation process selected. The
model i s a lso u sed to rank stimulation candidates ba sed o n e conomics. Hence this research will
attempt to answer the question: given the need to stimulate several wells in a field, how do we
rank the wells ba sed on s timulation b enefit and w hat stimulation approach to use in or der to g et
the highest e conomic returns? To answer these questions, a m erit function is developed based
on production decline curve a nalysis and economic discounting concepts. In combination with a
good stimulation treatment module, the model can be used for ranking stimulation candidates.
The research procedure begins i n c hapter one with a n introduction to the c oncept o f skin factor
and w ell s timulation methods. S everal lit eratures o n f ormation da mage a nd s timulation models
are r eviewed in chapter t wo. Chapter t hree c ontains a w ell s creening m odule, design o f s ome
selected stimulation m odules and an o ptimization model w hich c onsists of an objective function
with constraint. The optimization model combines the concept of production decline curves with
economic d iscounting. The m odel de veloped i n chapter three is va lidated in c hapter f our using
actual field data from the Niger Delta.
1.7
This research is intended for stimulation candidate selection in the Niger Delta. Matrix acidizing
technique is the main stimulation technique that has been used up to date in the Niger Delta due
to t he g ood permeability of t he N iger D elta formation. H ence only matrix a cidizing t echnique,
recompletion and gravel packing are considered in the methodology presented in chapter three of
this research. Acid fracturing and hydraulic fracturing are not considered.
Literature Review
In or der t o properly select s timulation candidate w ells, i t i s n ecessary t o first ha ve a n i n-depth
understanding of t he c oncepts of f ormation d amage and w ell s timulation. A lot of researches
conducted o n formation da mage and w ell stimulation methods can be f ound in literatures. We ll
stimulation i s c onsidered a m ajor ke y t o proper r eservoir m anagement, he nce several a uthors
made valid contributions.
2.1
2.1.1 Definition
Civan1 defined formation d amage a s a g eneric t erminology r eferring t o t he i mpairment o f t he
permeability of petroleum bearing formations by various adverse processes. It is an undesirable
operational a nd e conomic problem t hat c an o ccur du ring t he va rious p hases of oi l a nd ga s
recovery f rom s ubsurface r eservoirs including d rilling, production, hydraulic f racturing, and
workover operations. Bennion2 viewed formation da mage as any process that causes a reduction
in t he natural inherent pr oductivity of a n o il a nd ga s pr oducing formation, or a reduction i n the
injectivity o f a water or gas in jection well. Bennion also pointed out that the formation da mage
issue is often overlooked because of ignorance and apathy. In many cases, the operators are not
seriously c oncerned w ith f ormation d amage because of t he b elief t hat i t can be circumvented
later o n, simply b y a cidizing a nd/or h ydraulic fracturing. B ut P orter3 and M ungan4 argued t hat
because formation damage is usually nonreversible, it is better to avoid formation damage rather
than deal with it later on using expensive and complicated procedures.
Invasion of f oreign f luids, s uch as w ater and c hemicals used for i mproved
recovery, drilling mud invasion, and workover fluids;
steady-state pressure difference. A relationship between the skin effect, s, reduced permeability,
and altered zone radius, r d may be expressed as:
R
1 ......2.1
Equation 2.1 is k nown a s Hawkins7 formula. From t he e quation i t can be deduced t hat If <
the well is damaged and > 0; conversely, if > , then < 0 and the well is stimulated. For = 0,
Generally, certain well logs may enable calculation of the damaged radius, r d , whereas pressure
transient analysis may provide the skin effect, s, and reservoir permeability, k. Equation 2.1 may
then be used to calculate the value of the altered permeability .
In the absence of production log data, Frick and Economides8 postulated that, an elliptical cone
is a more plausible shape of da mage distribution along a horizontal well. They developed a skin
effect expression, analogous to the Hawkins formula:
2
4
,
+1
3
+ 1
....2.2
of d amage is n ear t he vertical section of t he well. T hey stated t hat the shape of t he el liptical
cross-section will depend greatly on t he i ndex of a nisotropy. The i ndex of anisotropy is
defined as:
= ...2.3
Piot and Lietard9 expressed the total skin of a well as a sum of the pseudoskin of flow lines from
the f ormation face to t he pi peline and the true s kin du e to f ormation da mage. Economides and
Nolte10 shown t hat t he t otal skin effect i s a c omposite of a n umber of factors, most of which
usually cannot be altered by conventional matrix treatments.
The total skin effect may be written as:
= + + + +
...............2.4
The last term in the right-hand side of Eq. 2.3 represents an array of pseudoskin factors, such as
phase-dependent a nd r ate-dependent e ffects that c ould b e altered b y hy draulic f racturing
treatments. The other three terms are the common skin factors. The th ird term refers to the
damage skin e ffect as defined in equation 2.1. The fi rst term + is the skin effect caused by
partial completion and slant. Cinco-Ley et al.11 documented a detailed approach of estimating the
skin f actor du e t o partial completion a nd slant. T he pa rameters needed for t he estimation a re:
completion t hickness, r eservoir thickness, elevation, a nd penetration r atio. An e xample t o
= + + ..2.5
Karakas and Tariq13 also shown t hat a combination of the damage and p erforation skin e ffects
( ) can be approximated, for a case where the perforations terminate inside the damaged zone,
by:
( ) =
+ = ( ) +
....2.6
is the damaged zone radius, and ( ) is the equivalent openhole skin effect (Eq. 2.1)
= 1.612
0.521
0.271
(0.043) .2.7
(, ) = +
162.6
3.23) .......2.8
1690
1
2
2.9
In equation 2.9, is the time at w hich the two straight lines representing the damage zo ne and
=1
....2.10
where and the undamaged and damaged standard flow rates, respectively.
Using Muskat19 equation for the undamaged flowrate:
2 ( )
...2.11
2( )
....2.12
.....2.13
where and in Equations 2.11 and 2.12 are the fluid viscosity and formation volume factor.
and are t he wellbore and reservoir drainage r adii, and is the r adius of t he d amaged
region.
Combining equation 2.1 a nd 2.13, t he damage r atio can be e xpressed i n t erms o f the effective
skin factor , as:
....2.14
is as defined in equation 2.1. Equation 2.14 gives the production loss by alteration of formation
properties. Leontaritis21 stated t hat r apid flow o f o il a nd water i n t he near-wellbore r egion
..2.15
and are respectively the absolute and relative permeabilities. High viscosity emulsion forms
a stationary block which resists flow. It is usually called emulsion block. If and represent
the v iscosities of oil a nd e mulsion, r espectively, a nd a s teady-state and i ncompressible r adial
and,
2( )
....2.16
2( )
....2.17
Civan22 substituted Equations 2.16 and 2.17 into Eq. 2.10 to obtain the following e xpression for
the damage ratio:
=
...2.18
Equation 2.18 gives a means to calculate the production loss by alteration of fluid properties.
The viscous skin effect is also expressed similar to Zhu et al23 as:
1 .2.19
Flow efficiency ( FE) i s defined a s the r atio o f t he damaged t o u ndamaged formation flow
(production or injection) indices.
=
.................2.20
where and denote t he a verage reservoir fluid and flowing well bottom hole pressures,
Mukherjee a nd Economides24 presented the f low ef ficiency o f v ertical w ells f or radial and
..2.21
= 1
2.22
where and denote the formation permeabilities before and after damage, respectively.
.2.23
Li e t al26 and a lso L ee a nd Kasap27 stated t hat b ecause t he d egree o f damage variation in t he
near-wellbore region, i t is more appropriate to express t he total skin, used in any of the
=1
=1
= =
..2.24
2.2
The optimal volume of a cid for a particular acidizing job may b e selected ba sed on a laboratory
acid response curve or an acidizing model28. These models consider both the modification of the
pore structure as it dissolves and the change in acid concentration as a function of both time and
position within the pore system.29
Dullien30 presented a c omprehensive literature r eview of t he models a nd the methods us ed t o
determine p ore-size d istributions i n a po rous medium. Scheidegger31 reviewed capillary models
and concluded that to predict quantities that relate to the geometric structure of a porous medium,
such as permeability and capillary pressure, an empirical correlation factor called tortuosity must
be introduced. Scheschter and Gidley32 proposed a capillary model to describe matrix acidizing.
The t wo-mineral m odel l umps all m inerals i nto on e of t wo c ategories: f ast reacting and s low
reacting species; a nd it i s t he m ost c ommon model i n u se today.
36, 41 -42
Schechter43 categorizes
fieldspars, a uthogenic clays, a nd a morphous silica a s fast-reacting, w hile d etrital c lay p articles
and qu artz gr ains are the pr imary s low-reacting mi nerals. This model a s presented by
Economides a nd N olte44 consists o f material b alances ap plied t o t he H F a cid a nd r eactive
minerals, which for linear flow, such as in core-flood, can be written as:
( )
= , + , (1 ) ..2.25
[(1 ) ] =
[(1 ) ] =
.....2.26
.2.27
where is the concentration of hydrofluoric acid (HF) in solution and is its molecular
weight, is t he a cid flux, is th e d istance, and are the s pecific s urface a reas p er unit
are t he dissolving pow ers of 100% H F, a nd and are t he densities of t he fast- and s lowreacting minerals, respectively, denoted by the subscripts F and S.
When t he e quations above are m ade d imensionless f or a c ore-flood of l ength with c onstant
porosity, two dimensionless groups were observed for each mineral: the Damkohler number
and the acid capacity number . These two groups d escribe the kinetics and t he stoichiometry of the
HF-mineral reactions. The shape of the acid reaction front depends on t he Damkhler number . The
acid ca pacity n umber regulates h ow m uch l ive acid reaches t he f ront, in ot her w ords, it
affects the frontal propagation rate directly.
The Damkhler number is the ratio of the rate of acid consumption to the rate of acid convection,
which for the fast-reacting mineral is:
()
(10 )0 ()
...2.28
The acid capacity number is the ratio of the amount of mineral dissolved by the acid occupying a
unit vol ume o f rock por e s pace to the amount o f m ineral present in the u nit vol ume o f rock,
which for the fast-reacting mineral is:
()
....2.29
(10 )0
The dimensionless form of equations 2.25 through 2.27 can onl y b e solved nu merically in t heir
general f orm, th ough a nalytical s olutions a re p ossible for certain simplified situations.
Schechter 43 presented an a pproximate solution to these equations that is valid for relatively high
Damkhler number ( () > 10). Numerical m odels providing solutions t o t hese equations,
such as that presented by Taha et al.36are frequently used for sandstone acidizing design.
Bryant45, and also, da Motta et al.46 shown that at elevated temperatures the sandstone acidizing
process i s n ot well d escribed by t he t wo-mineral m odel. These studies suggest that the r eaction
of fluosilicic acid with aluminosilicate (fast-reacting) minerals may be quite significant. Thus, an
additional acid and mineral must be considered to accommodate the following reaction, which is
added to the two-mineral model:
H 2 SiF 6 + fast-reacting mineral
Si(OH) 4 + Al fluorides
...2.30
Precipitation Models
Permeability Models
Predicting permeability change a s acid dissolves some of the formation minerals and precipitate
is f ormed i s a necessary s tep n eeded to predict the f ormation response to acidizing. The
permeability increases a s t he pores a nd pore t hroats a re enlarged b y mineral dissolution. At th e
same t ime, small particles ar e r eleased a s c ementing m aterial i s dissolved, a nd some of t hese
particles lodge (perhaps temporarily) in pore throats, reducing the permeability. Any precipitates
formed a lso t end t o d ecrease the permeability. T he formation of carbon d ioxide ( CO 2 ) a s
carbonate mi nerals a re dissolved m ay a lso cause a t emporary r eduction i n t he r elative
permeability t o li quids.48The complex n ature o f the p ermeability response h as m ade its
theoretical pr ediction f or r eal sandstones impractical. For t his r eason empirical correlations
relating the permeability increase to the porosity change during a cidizing are u sed. Guin et al.49
however a chieved s ome s uccess when a more i deal systems su ch a s si ntered disks was
considered. Labrid50 presented the following useful relationship:
= ..................2.31
= [45.7( )] ..2.32
2.33
In Eq. 2.31 through 2.33, and are the initial permeability and porosity and and are the
permeability and porosity after acidizing. and are empirical constants. In Eq. 2.33, and
best fit data f or pha coides s andstone. The b est a pproach i n u sing t hese correlations i s t o select
Schechter and Gidley32 used a m odel o f pore growth and collision to study the natural tendency
for wormholes to form when r eaction i s mass transfer l imited. I n t his model, t he c hange i n the
cross-sectional area of a pore is expressed as:
= 1 2.34
where is the pore cross-sectional area, is the time, and is a pore growth function that does
depend on t ime. If > 0, s maller pores gr ow faster than l arger p ores a nd wormhole cannot
form; when < 0, larger pores grow faster than smaller pores and wormhole will develop. They
diffusion controls the overall reaction rate. This m odel does not give a c omplete picture of the
wormholing process because it does not include the effect of fluid loss from the pores.
Mechanistic Models
Hung et al.56 considered fluid loss in their cylindrical model of the wormhole gr owth, a nd also
took i nto a ccount a number o f factors, i ncluding t he c ontributions of both a cid diffusion a nd
convection resulting from fluid l oss t o t he walls of t he wormhole where t he a cid reacts. They
found t hat the w ormhole velocity i ncreases linearly w ith the i njection rate i nto the w ormhole,
implying that t he v olume of a cid needed to pr opagate a
independent of injection rate. The model also predicts that wormhole velocity will be constantly
decreasing because t he a cid flux t o t he end of t he wormhole i s de creasing a s t he wormhole
length increases ( grows). The w ormhole ve locity is e xpressed in t erms o f the acid ca pacity
number (which had been defined for a fast-reacting mineral in Eq. 2.29) as:
= ...2.35
where and are the flux a nd a cid c oncentration ( mass fraction), t he subscript o refers to th e
initial condition, the subscript e refers to conditions evaluated at the e nd or tip o f the w ormhole,
and L is the length of the wormhole.
Network Models
Hofefner and Fogler55 presented n etwork m odels in which the porous medium is approximated
as a
concentration i n each capillary is calculated a nd the radii of the capillaries are i ncreased as
dissolution oc curs. These models a ppear t o give t he b est r epresentation o f w ormhole b ehavior
over a wide range of conditions, but they are difficult to generalize for treatment design.
Stochastic Models
2 3
..2.36
influence of acid diffusion but does not take into account fluid loss; therefore, this equation does
not indicate a plateau value as the wormhole lengthens. Thus, the equation is only applicable to
short w ormholes where fluid loss i s n ot a factor, and it should not be u sed for t he pr ediction of
wormhole penetration l ength. For a c onstant i njection r ate, t he skin e ffect pr edicted b y t he
Daccord et al.s model is:
If there is a damaged zone,
=
13
23
...2.37
If there is no damaged zone or if the wormholes penetrated beyond the damaged region,
= 1 +
13
23
...2.38
dimension equal to about 1.6 and is the cumulative volume of acid injected. Eq. 2.37 and 2.38
do not apply if the injection rate is changing during the treatment because of the dependence of
the wormhole velocity on injection rate in the Daccord et al.s model.
Volumetric Model
expressed as:
= 2 +
.....2.39
where , the w ormholing e fficiency, is de fined as the fraction of r ock d issolved in the r egion
= .2.40
where is the number of pore volumes of acid injected at the time of wormhole breakthrough
at the end of the core. The skin effect during injection is expressed as:
If there is a damaged zone,
....2.41
If there is no damaged zone or if the wormholes penetrated beyond the damaged region,
1
= 1 +
2
..2.42
..2.43
dissolution rate constant which depends on the sum of resistances in series, i.e.
1+
1
1
1
+
+
1 3
....2.44
Kr is the effective surface reaction constant. K1 and K3 are the mass transfer coefficients for the
reactants a nd products, r espectively. Eq. 2 .43 and 2. 44 can be u sed t o determine t he r ate of
carbonate dissolution in any flow geometry, provided that a n appropriate expression for the rate
of mass transfer is available.
2.3
The f ollowing e quations d escribed linear flow of a cid dow n a fracture, with fluid l eakoff a nd
acid diffusion to the fracture walls.
( )
= 0 2.45
(, , = 0) = 0 2.46
( = 0, , ) = () .2.47
= (1 ) ....2.48
where is the acid concentration, is the flux along the fracture, is the transverse flux due
viscous fingering o f l ow-viscosity a cid through a vi scous pad, the e ffect of the a cid o n leak-off
behavior, a nd va rious fracture geometries. Neerode and Williams64 also pr esented a solution t o
the a bove e quations by a ssuming a steady state, laminar flow of a N ewtonian fluid between
parallel plates with constant fluid loss flux along the fracture. They presented the solution for the
concentration p rofile as a f unction of t he leakoff P eclet n umber. At l ow Peclet n umbers,
diffusion controls a cid propagation, while a t hi gh P eclet numbers, fluid l oss i s t he c ontrolling
factor.
The conductivity ( ) of an acid fracture depends on a stochastic process. Nierode and Kruk65
presented the following correlation for the acid fracture conductivity based on the ideal fracture
width
,
where
and for
= 1 2 .2.49
1 = 1.47 107 2.47 2.50
In Eq. 2. 49 t hrough 2. 52, is the f racture closure s tress and is the r ock e mbedment
2(1)
2.53
where is the volumetric dissolving power of the acid, is the total volume of acid injected,
0 ..2.54
For lower values of Peclet number (< 3), this average overestimate the well productivity, hence
Ben-Naceur and Economides67 presented a harmonic a verage which better a pproximates the
behavior of the fractured well as:
0 /
..2.55
Ben-Naceur and Economides67 also presented a series of performance type curves for a cidfractured wells producing at a constant bottomhole flowing pressure of 500 psi.
2.4
..2.56
.. ...2.57
....2.58
Prats68 showed t hat for a s teady-state f low, a fracture affects productivity t hrough t he
dimensionless equivalent (effective) wellbore r adius
which i s related t o the fracture h alf-
...2.59
where is expressed in terms of the equivalent skin effect and the wellbore radius as:
= ...................................................................................2.60
= 0.5 .2.61
ln 0.472 +0.5 +0.5 + +
....................................2.62
= 1.6, is t he optimum value of the dimensionless fracture conductivity for which the
= + + + +
...2.63
the skin resulting from limited perforation height, is the skin due to turbulent (non-Darcy)
flow, is t he skin du e t o wellbore deviation, is the skin due to gravel packing, and
is the skin resulting from a small perforation. Nitters et al then suggested the ranking of
7, Jones defined the ratio of rates before and after stimulation (the stimulation ratio, ) as:
2
1
7+1
7+2
.2.64
where is flow ra te, is the s kin factor, and t he s ubscripts 1 an d 2 r efer t o before an d a fter
stimulation.
( + + ) 2.65
where + is the skin factor due to partial penetration and deviation, is the total skin
8+
............................................. 2.66
They concluded that if R 0.6, then the well is a good stimulation candidate in the Niger Delta.
Afolabi et al.81 also presented candidate selection criterion that is based on minimum economic
reserve, productivity Index (PI) of less than 10bpd/psi, flow efficiency of less than 0.5 and the PI
decline rate that is greater than 30%.
Jennings82 presented a
concluded that well stimulation tr eatments in high-productivity wells a llow better r eservoir
management through sustained productivity and more uniform reservoir depletion throughout the
life of the well, and that good wells make better candidates for matrix stimulation.
Kartoatmodjo et al.83 presented a risk-based c andidate selection a pproach by c onsidering the
range of probability of all the possible outcomes in a stimulation campaign u sing Monte Carlo
simulation technique. They concluded that decision risk analysis is a valuable tool for candidate
selection. Stimulation c andidate selection c ampaign ba sed on highest expected ga in a nd/or
lowest expected risk has also been reported.84
The published literatures reviewed did not consider a detailed and efficient optimization process
for s timulation candidate selection, especially i n t he N iger D elta, a nd hence t he n eed f or this
study.
Methodology
This methodology is a modification of the modular approach to stimulation decisions presented
by S inson et al.85 The m odels pr esented are de rived f rom i ndustry-wide a ccepted well
stimulation procedures and techniques.
3.1
It i s a ssumed that from well t est da ta, t he well pr oblem could b e diagnosed a nd then m atched
with either of acidizing, gravel-packing or re-completion. It is also assumed that all wells can be
acidized, recompleted or gravel-packed successfully if necessary.
Diagnose each well pr oblem. For w ells w ith s kin va lues s howing formation da mage problems,
acidizing i s t he r ecommended t reatment. Wells with m echanical pr oblems s uch a s pa rtially or
totally plugged perforations, i nsufficient perforation density, l ow de pth of perforation o r w ater
production, r e-completion i s r ecommended. I f t he pr oblem i s sand production, t hen gravel
packing i s r ecommended. A s imple screening module flow chart f or t his s ection i s s hown i n
Appendix A.
3.2
The treatment m odels p resented in t his s ection are to b e used f or the s timulation t reatment
design. The choice of which model to use is dependent on the nature of well problem diagnosed
and the result of the screening module.
Determine the present fracture gradient for the well. If the instantaneous shut-in
pressure value is not available, use the following equation to calculate the fracture
gradient:
= + ( )
where:
..3.1
= overburden gradient (1.0 psi/ft for formation depth less than 10,000ft or 1.2 psi/ft
for depth greater than 10,000ft)
= depth of formation, ft
Predict the maximum possible injection rate that does not fracture the formation
using:
where:
, =
4.917106
.3.2
= safety margin for the pressure, psi (usually 200 to 500 psi)
= viscosity of the injected fluid, cp
= drainage radius, ft
= wellbore radius, ft
= formation volume factor, bbl/STB (it has a value of 1 for incompressible fluids)
Using Equation 3.2 with zero value for the skin effect gives the maximum pump rate during the
treatment.
If the injection fluid is N ewtonian, and at pumping rates that a re le ss than 9 bbl/min, t he coil
tubing friction pressure can be calculated using:
where:
...3.3
If p ipe or c oil tubing f riction pressure is co nsidered, the maximum s urface p ressure f or w hich
fluids can be injected without fracturing the formation is calculated using:
, = + ...3.4
where:
, = ..3.5
Determine the volume of mud acid to use
It i s a ssumed that t he a cid volume r equired is equal to the pore volume of t he damaged zone.
Also, i t i s a ssumed that a cid flows t hrough the porous media with a front t hat i s u niform a nd
stable, then the acid injection is piston-like and the first acid in is the last acid out. The mud acid
volume is estimated using:
where:
= 7.48[(2 2 )] ...3.6
where:
= 7.48
(1) [2 2 ]
..3.7
where:
= ..3.8
= cost of acid used per unit volume, $/gal
In sandstone it is difficult to increase the permeability above the natural state because of reaction
kinetics li mitations, r eaction stoichiometry a nd economics. In th is th esis, the maximum
formation productivity r atio for sandstone a cidizing, given s ome s et of reservoir parameters, is
defined b y the reciprocal of the flow efficiency, and is approximated from Equation 2.21, using
the semi-steady state definition:
where:
0.472
+
..3.9
0.472
3.2.1.1 Summary
1.
2.
Predict the maximum possible injection rate that does not fracture the formation.
3.
4.
5.
6.
7.
8.
where:
2 .3.10
= wellbore radius, ft
where:
...3.11
= + 0.5 ..3.12
and:
0.03
..3.13
= reservoir permeability, md
= + 3.14
Calculate cost of recompletion
where:
= ...3.15
= number of perforations
The productivity index for a semi-steady state condition is u sed to define the productivity ratio,
and hence, defining the productivity ratio as the reciprocal of the flow efficiency, it is expressed
as:
where:
0.472
+
0.472
...3.16
100% or less of t his volume m ay be c onsidered as the e xcess g ravel. T his volume e nsures
complete screen/formation coverage by the gravel.
where:
1 =
4144
2
2
3.17
This volume must be filled up completely with gravel in order to have an efficient pack.
2 =
where:
4144
2
2
.3.18
3 = ..3.19
where:
= 0.5 1.5 ft3/ft (for the zones that have produced sands).
= 1 () + 2 + 3 3.20
where:
= total volume of gravel needed, ft3
= fraction of the blank/casing annulus needed to be filled (in this thesis, taken to
be 60% - 90%)
= 7.48 3.21
where:
= bulk density of gravel, ppg. (It is the density of the bulk that includes the air
between the grains).
where:
42
....3.22
where:
= ..3.23
= 1 +
...3.24
..3.25
where:
= slurry density, ppg
where:
96
..3.26
= + .3.27
where:
= cost of gravel packing, $
= cost of gravel, $/ft3
3.3
0.472
+
0.472
.3.28
At s ome point du ring the pr oducing l ife o f a w ell, t he pr oduction rate m ay become s o low and
well diagnosis may result in th e need f or well stimulation. Figure 3. 1 s hows the production
profile (production rate vs. t ime) of a well t hat a t some point during its producing l ife was
profitably stimulated. This figure shall serve a s the theoretical basis for the model developed in
the following sections.
Consider F igure 3.1. T he curve ABC r epresents t he well pre-stimulation decline curve profile.
The w ell initial p roduction r ate is . At point B, t he w ell is c onsidered f or s timulation. T he
curve DEF i s t he r esulting pos t-stimulation production pr ofile. The p roduction r ate is the
corresponding to the production rate . At time the stimulated w ell i s open for production.
Thus, t he difference between t he t imes and is the d uration of t he stimulation job. The
BCHI. The i nitial production r ate a fter stimulation i s r epresented b y which c orresponds t o
point D in Figure 3.1. The well is now produced along the curve DEF until the abandonment rate
The model formulated in the following section u ses the production profile described above and
an e xponential decline curve analysis with e conomic concept of c ontinuous discounting. The
derivation of the model for several cases of hyperbolic decline is presented in Appendix F.
A
D
E
C
Time, days
7. The abandonment rate of the well is the same for both the p re-stimulation and poststimulation profile.
...3.29
= ..3.30
The u ltimate g oal o f well stimulation i s to exploit t he r eservoir p rofitably. I n optimizing well
stimulation processes, the measure of effectiveness is the net i ncremental post-stimulation
production subject to the limitations imposed by the system. Therefore an objective function will
be defined to maximize the net post-stimulation production. The objective function is defined as:
= ............................................................................3.31
where is the discounted production from stimulation, is the discounted production loss
from stimulation, and is the discounted production equivalent to total stimulation cost.
The e xponential d ecline curve analysis s hall b e used to derive the mathematical expressions for
each of the components of Equation 3.31. Before proceeding with the derivation, it is necessary
to define some of the variables in Figure 3.1. First, let us shift the time axis such that the time at
the start of the stimulation job is set to zero. Then let be the duration of the stimulation job
and be the abandonment time of the post-stimulation production profile.
A. Discounted Incremental Post-Stimulation Production,
The discounted incremental production resulting from the stimulation process is derived from the
area enclosed by DEFH in Figure 3.1 by:
= ( ) () ..3.32
where is the exponential decline rate per day. Substituting for and from Equations
3.29 and 3.30, Equation 3.32 is expressed as:
= ( ) ..3.33
Evaluating the integral on the right hand side of equation 3.33 yields:
()
()
()
() () ...3.34
The abandonment production rate for the post-stimulation production forecast is given by:
= ( ) .3.35
Substituting for from E quation 3. 29 and r earranging Equation 3 .35, the economic l ife
=
1
= + .3.36
()
( )
+()
() ..3.37
=
=
( )
()
( )
()
( )
()
.3.38
The concept of production loss is similar to the idea of opportunity cost. The production loss is
an essential component of the ob jective f unction that t akes care of the z ero-production time
during stimulation.
= 0 ()
= 0 .3.39
= ()
() 0
= ()
() 1 ..3.40
The t otal stimulation c ost, which includes site preparation cost, equipment mobilization &
demobilization cost a nd the stimulation tr eatment cost, can be c onverted to i ts e quivalent
discounted production as:
= .......................................................................................3.41
where is the total cost of the stimulation treatment in dollars, and is the price (in dollars) per
barrel of oil.
( )
()
( )
()
()
() 1 ...3.42
Let
.3.43
1 = ()
.3.44
2 = 3.45
3 = () ...3.46
4 =
...3.47
(1)
5 =
......3.48
Therefore the objective function as defined in Equation 3.42 can be expressed in the form:
3.4
= 1 2 3 5 1 2 3 4 1 (3 1) .3.49
+ ..3.50
Using the definitions of Equations 3.33, 3.39 and 3.41; Equation 3.50 can be expressed as:
( ) + 0 ..3.51
By examining Equations 3.42 through 3.49, this constraint can be expressed as:
1 2 3 5 1 2 3 4 + 1 (3 1) .....3.52
In a practical sense, this constraint is satisfied if and only if the value o f the o bjective function
is positive, that is:
0 ..3.53
3.4.2 Constraint 2: Remaining Reserve Limitation
The recovery from the stimulation should not e xceed the r emaining produceable oil i n pl ace
(reserve). Mathematically, this constraint can be expressed as:
( ) .3.54
where is t he remaining oil reserve in place during stimulation. Solving Equation 3.54 we
get:
[ ] 3.55
....3.56
Simplifying,
.....3.57
....3.58
....3.59
Let
1 =
2 =
Substituting Equations 3.58 and 3.59 into 3.57, this constraint can be written as:
1 2 ...3.60
3.4.3 Constraint 3: Flow String Capacity
The pr oduction r ate a fter s timulation should n ot e xceed t he maximum d esign capacity o f t he
flow string. In the case of gas wells, this constraint is imposed by the gas pipeline capacity.
= ( )
= ( ) ..3.61
The m aximum production rate is obtained w hen the well is opened f or production just after
stimulation, i.e. at time = (see Fig. 3.1). Using this substitution in Equation 3.61, constraint
3 can then be formulated as:
therefore:
....3.62
where is the maximum design capacity (flow rate) for the well tubing string.
..3.63
3.64
where is the m aximum productivity ratio t hat c an be o btained given the reservoir a nd
3.5
1 .3.65
From t he design module pr esented i n section 3 .21 t hrough 3.23, i t c ould be observed t hat t he
input design parameters determine the stimulation cost (C) a nd t he maximum pr oductivity ratio
(F). For e xample, f rom the acidizing d esign m odule in s ection 3.21, it c ould b e seen that t he
stimulation c ost depends o n t he vo lume of acid pumped, and also the volume of a cid pumped
will d etermine t he extent o f damage r emoval ( productivity r atio). T his discussion shows t hat a
relationship can be formulated between the stimulation cost and the productivity ratio based on
the de sign m odule. Hence, in order to use the model presented in section 3 .3 as a n optimization
model, it i s necessary t o develop a stimulation cost v ersus productivity ratio r elationship ba sed
on the design module presented.
The combined effects of the treatment and reservoir variables are lumped into a stimulation cost
versus productivity equation of the form:
= ...3.66
stimulation cost v ersus p roductivity r atio. It is this equation that in corporates the stimulation
option into the optimization model. Hence we must substitute Equation 3.66 into Equation 3.49
in order to use the model.
3.6
Combining the objective function a nd the constraints, the optimization model formulated can be
summarized as:
Maximize:
= 1 2 3 5 1 2 3 4 1 (3 1) .3.49
subject to:
1. Break-even Requirement:
1 2 3 5 1 2 3 4 + 1 (3 1) ....3.52
1 2 ...3.60
4. Budget Allocation:
...3.62
...3.63
.3.64
1 ...3.65
The constants , 1 , 2 , 3, 4 and 5 are as defined in Equation 3.43 through 3 .48; 1 and 2
are defined in Equation 3.58 and 3.59; and are from Equation 3.66.
It is important to note that the optimization model is a non-linear programming (NLP) problem.
The o bjective function consists o f t wo variables, na mely productivity r atio and total
stimulation cost . The two variables are related based on the discussion presented in section 3.5.
This research investigated the matrix stimulation cost and performance data presented by Vogt et
al.88 An a ttempt was ma de to obtain a relationship between total stimulation cost and
productivity ratio . The data is presented in Table 1 of Appendix B. A regression analysis on the
data using M icrosoft Excel s hows a trend between pr oductivity r atio a nd t otal stimulation c ost
similar to the form presented in Equation 3.66. Therefore, to use this model, the stimulation cost
versus productivity ratio constants and must be obtained from the design module. The design
and op timization m odel included i n t he a ccompany compact di sk ( CD) of t his t hesis only
requires the input, stimulation design parameters, to generate the constants.
3.7
In this research, the model was solved u sing the Solver in Microsoft Excel and also Whats Best
10.0 LINDO S ystems o ptimization so ftware. The r esults obtained w ere v erified b y c omparing
the solution with t hat ob tained by us ing Mathematica 7.0 software developed by Wolfram
Research. The Solver i mplemented i n E xcel ( developed by Frontline Systems) u ses numerical
iterative methods (generalized reduced gr adient m ethod) to s olve e quations a nd t o o ptimize
linear and n onlinear functions w ith e ither c ontinuous or i nteger variables. But Solver has some
limitations hence t he ne ed t o verify t he r esults. Wolframs Mathematica 7.0 on t he ot her ha nd
uses several numerical algorithms for constrained no nlinear optimization. T he a lgorithms ar e
categorized into gradient-based methods and direct search methods. Gradient-based methods use
first d erivatives ( gradients) or second d erivatives ( Hessians). Examples a re t he sequential
Chapter Four
4.1
Sensitivity Analysis
The parameters i n t he following section were va ried a nd the values of t he o ptimal objective
function obtained are plotted against the productivity ratio f or e ach parameter va lue. The input
data used for the sensitivity analysis are presented in Table 4.1.
4.1.1
The p rice o f o il d etermines the amount of revenue derived from the s timulation. T herefore an
increase in the price of oil is a ccompanied with a n increase in the optimal point of the objective
function a s shown in Fig. 4.1. The price of oil is purely an e conomic input to the optimization
model. T he decision t o perform well s timulation depends o n t he current price o f oil. H ence t he
higher t he p rice of oil, t he greater t he benefit derived from s timulation. It is important to n ote
that b elow a pr oductivity ratio o f a bout 3 .2, the di scounted pr oduction will not c hange w ith the
price of oil, but the overall monetary benefit will reduce when the price of oil falls.
4.1.2 Effect of Discount Rate
Fig. 4.2 illustrates the effect of the interest rate on t he objective f unction. The value of t he
discount rate was varied from 5% to 20%. The discounted production decreases with an increase
in t he discount r ate. The di scount rate can be viewed a s a n a dditional cost of stimulation. T he
higher the discount rate, the higher the cost of money and well stimulation, and consequently, the
lower the benefit to be derived from the stimulation job.
4.1.3 Effect of Decline Rate
The effect of the exponential decline rate on the objective function is shown in Fig. 4.3. For this
analysis, the value of the decline constant was varied between 0.032/yr and 0.32/yr. It is noticed
that the smaller the exponential decline rate, the higher the stimulation benefit. The exponential
2200 psi
Drainage Radius, r e
1053 ft
Wellbore Radius, r w
0.3 ft
20 ft
Depth of Formation
12000 ft
7 ft
200 md
20 md
Porosity
25%
1 bbl/stb
0.45 psi/ft
1.04
0.57 cp
Pump Rate
2 bbl/min
200 psi
1.75 inches
$ 38 per gal
0.4 psi/ft
1000 stb/d
Abandonment Rate, q a
200 stb/d
Duration of Stimulation, t s
2 days
3 MM stb
80 $/stb
10%
10000 stb/d
1.2 MM $
6000
5000
4000
3000
2000
1000
0
0
8000
7000
6000
5000
I = 5%
4000
I = 10%
3000
I = 15%
I = 20%
2000
1000
0
0
30000
25000
20000
15000
D = 0.032/yr
D = 0.16/yr
10000
D = 0.32/yr
5000
0
0
decline rate is the parameter that controls the concavity of the objective function. The smaller the
value of the exponential decline constant for a w ell pr oduction pr ofile, the more the benefit we
could get if such well i s considered f or s timulation. In pr actice, w e ha ve no c ontrol ov er the
value of the decline rate constant. H owever, it gives u s a direct i nsight i nto candidate selection
for stimulation decisions. Smaller decline rate is desirable for profitable stimulation decisions.
4.1.4 Effect of Pre-Stimulation Production Rate
The effect of pre-stimulation production rate on t he objective function is illustrated i n F ig. 4 .4.
The value of the pre-stimulation production rate was varied from 500stb/d to 1500stb/d. A higher
pre-stimulation pr oduction r ate indicates a higher r eservoir energy dr ive. T he main goal of
stimulation i s t o i ncrease pr oduction u sing t he r eservoir e nergy a s t he driving force in moving
the o il from th e r eservoir i nto th e wellbore. If t he r eservoir h as litt le or n o e nergy, stimulation
benefit w ill b e small. Th is is clearly illustrated in the figure. Since a higher pre-stimulation
production will give a higher optimal point in the objective function, therefore, from the figure, a
higher p re-stimulation production rate will give a higher optimal stimulation benefit. This
suggests that as production declines during production, there should be an optimal time in which
it is best to initial stimulation jobs. Because of the huge impact of this pre-stimulation production
rate on t he ob jective function, this pa rameter must be given a major attention in the selection of
stimulation candidates
4.1.5 Effect of Abandonment Rate
The e ffect of t he a bandonment r ate o n t he stimulation d ecision i s shown i n F ig. 4. 5. The
abandonment r ate i s v aried b etween 100stb/d a nd 50 0stb/d. I t i s observed t hat i ncreasing t he
abandonment r ate r esults i n decrease i n t he overall s timulation b enefit. The a bandonment r ate
can be i nterpreted in terms of t he r emaining r ecoverable o il i n t he r eservoir. A hi gher
abandonment r ate m eans a h igher a mount of r ecoverable oil r emaining i n t he r eservoir. B ut a
reduced incremental production is expected because when the abandonment rate is set high, the
incremental p roduction w ill be reduced since w e h ave a l imit t o w hich w e ca n produce.
Consequently, a reduced incremental production will eventually decrease the optimal point of the
objective function. Therefore, the abandonment rate is a major factor that influences the choice
of stimulation candidate selection.
12000
10000
8000
6000
qo = 500stb/d
qo = 1000stb/d
qo = 1500stb/d
4000
2000
0
0
6000
5000
4000
3000
qa = 100stb/d
qa = 300stb/day
qa = 500stb/d
2000
1000
0
0
4.2
In this section, the optimization model is applied with the acidizing treatment model to quantify
stimulation benefit derivable from four typical acidizing jobs, and also, to rank the w ells f or the
stimulation process. Production data from four wells: Well BU 1, Well BU 2, Well BU 3 and Well
BU 4 were u sed t o va lidate t he model. The four w ells completed i n M ay 200 4 are l ocated in
Bestfields, offshore Niger Delta. This high permeability field is located in a water depth of 200m.
The a verage pe ak p roduction recorded i n J anuary 2006 from e ach of t he four wells i s
7000stb/day. Production d ecline s tarts a fter a 3 -year peak pr oduction period. The a vailable
production data f or each o f t he four wells shows t hat t he d ecline pr ofile for e ach w ell i s
exponential. The wells are being considered as potential candidates for acidizing after a well test
confirms the presence of acid removable damage. The field data is presented in Table 4.2. These
data s erved a s input da ta for t he a cidizing d esign a nd optimization model. A dditional da ta used
were t aken from published l iteratures by Ofoh a nd H eikal89, Nnanna et al.73, Nnanna an d
Ajienka76, and Onyekonwu80.
The data in Table 4.2 are used to formulate the Bestfield Model, which gives an insight into how
the model ca n be used to optimize acidizing candidate w ell s election process in the Niger Delta.
The design and optimization model is available in the included CD.
4.2.1 Formulation of the Bestfield Model
In combination with the data provided in Table 4.2, lets assume that the remaining recoverable
reserve is 500 MM bbls, and the tubing maximum design flow rate for each well is 12500stb/d.
8000
7000
6000
5000
4000
Stim. Time = 1 day
3000
2000
1000
0
0
Well BU 2
Well BU 3
Well BU 4
3200
3200
3200
3200
1000
1000
1000
1000
0.3
0.3
0.3
0.3
120
85.6
148.29
68.2
9000
9060
8950
9000
6.8
6.4
6.5
3500
3300
4600
3500
900
450
510
580
Porosity (%)
25
25
25
25
1.159
1.159
1.159
1.159
0.45
0.45
0.45
0.45
1.04
1.04
1.04
1.04
0.57
0.57
0.57
0.57
3.5
3.5
3.5
3.5
200
200
200
200
1.75
1.75
1.75
1.75
30
30
30
30
(psi/ft)
0.4
0.4
0.4
0.4
4000
4100
3900
5200
250
250
250
250
0.000519
0.000568
0.000547
0.000533
Also lets assume an average oil price of $80/bbl, effective discount rate of 10% and a maximum
acidizing budget of $1200000 per well.
Using these data, the model is formulated as follows:
Step A: Enter each well data given into the Acidizing Design and Optimization Model and
generate the Cost versus Productivity-Ratio relationship for each well.
The Cost versus Productivity-Ratio plots for each of the four wells was generated by the design
model. T he a nalysis i n t he following section is for Well BU 3. The a nalysis for t he Well BU 1,
Well BU 2 and Well BU 4 is similar, h ence o nly t he results w ere discussed. T he C ost ve rsus
Productivity-Ratio plot for the Well BU 3 data is shown in Fig. 4.7.
Step B: Obtain the relationship between the stimulation cost (C) and productivity ratio (F) in
form of power equation of a trendline through a log-log regression of the data.
For the Well BU 3 input data, the equation is obtained as:
= 87.156.499 .................4.1
The above relationship, a s presented in Equation 4.1 was obtained from a regression analysis o f
the simulated data generated by the design model using Microsoft Excel. The design model used
the Well BU 3 input data to account f or c ost as s hown in Equation 4.1 ba sed on t he damaged
radius of the well, which is one of the parameters with greatest influence on the acidizing design.
The acidizing design and optimization model will generate this equation once the data input step
is completed.
Step C: Use Equations 3.49, 3.52, 3.60, 3.62, 3.63, 3.64 & 3.65 to formulate the objective
function and its constraints.
The constants needed to define the objective function can be calculated easily using the equation
listed a bove. It i s i mportant t o point ou t that w hile u sing the a cidizing a nd d esign model
included in the CD, one does not need to calculate the objective function as presented below. The
program is designed to calculate the objective function, set up the constraints and then awaits the
user t o call a s olver pr ogram f or t he o ptimization step. Hence, S tep C i s only i ncluded for t he
purpose of proper understanding of how the model and its constraints were formulated.
The non linear programming model formulated a s a maximization problem using the Well BU 3
data is presented below.
:
1. Break-even Constraint:
3.21 ...4.5
1 4.7
4.20 4.8
From t he non-linear programming optimization problem presented a bove i t c ould b e seen t hat,
simply, we seek an optimum value for the productivity ratio which ha s a lower and upper bound
of 1 and 3.21 respectively. This is true because the limit sets by the facility constraints (Equation
4.5) is more binding than the maximum productivity r atio a ttainable given t he r eservoir a nd
treatment parameters (Equation 4.8).
Figure 4.7 (a) and (b): Cost versus Productivity Ratio plots for Well BU 3
Cost-Productivity Ratio Plot
(Cartesian Plot)
3000000.00
2500000.00
Cost, $
2000000.00
1500000.00
1000000.00
500000.00
0.00
1.00
1.50
2.00
2.50
3.00
3.50
4.00
4.50
5.00
Productivity Ratio, F
(a)
Cost-Productivity Ratio Plot
(Log-Log Plot)
10000000.00
Cost, $
1000000.00
100000.00
10000.00
C = 87.15F 6.499
1000.00
1.00
10.00
Productivity Ratio, F
(b)
Step D: Find a solution to the non-linear programming model formulated in Step C above.
In this research, two different solvers, which use different a lgorithms, were used to get an
optimum solution to the model. The solvers are: Frontline Systems Microsoft Excel Solver and
Whats Best 10.0 LINDO S ystems op timization s oftware. T he solution t o t he model was t hen
verified u sing t he W olfram R esearchs Mathematica 7.0. The results a re discussed i n t he
following sections.
4.2.2 Solution of the Well BU 3 Model
The Well BU 3 Model was solved u sing t he Microsoft Excel Solver and Whats Best 10.0, the
results a re presented i n Appendix C and D . The s olutions gave the same r esult for t he optimal
point.
The value of the objective function at optimal point is:
= 106868.12
The results obtained using the Microsoft Excel Solver and Whats Best 10.0, is taken as the actual
value of the objective function at the optimal point. Hence, the following discussions are based
on this result.
4.2.3 Discussion of the Well BU 3 Model Result
This optimal v alue of the objective f unction is 106868.12bbls, meaning that i f this well i s
considered for s timulation, given that t he a ssumptions c onsidered i n section 4 .2.1 a re binding,
the benefit derivable is 106868.12bbls of oi l. The life of the well is estimated to be 19.6 years.
The payout time on the acidizing cost is also estimated to be 0.94 day.
The A nswer R eport for Well BU 3 in Appendix C, Section 1, shows t hat five o ut of t he six
constraints are not binding. The only binding constraint is the tubing string capacity. H ence, the
optimal solution was f ound within th e lim its o f a ll t he constraints. That is t o s ay th at a ll
constraints are satisfied. No constraint is violated.
The tubing string capacity constraint is binding, meaning that if the tubing f low capacity is
increased, there will be more benefit from this project, but on the other hand, this extremely high
rate w ill kill o ur w ell s ooner than la ter. In g eneral, f or a c onstraint to b e bi nding m eans any
movement to the right would still give a better result to the objective function.
From the sensitivity report it could be seen that the value of the Lagrange Multiplier associated
with th e flow string c apacity c onstraint is 31221.9668. T his gi ves a n i dea of t he fractional
change of the ob jective f unction i f t he flow s tring c apacity c onstraint c hanges b y 1stb/day.
Hence, if the flow string capacity is increased by 1stb/day, the benefit derivable from stimulation
will increase by 31221.9668 bbls. Hence, the v alue o f t he Lagrange Multiplier will help th e
stimulation design engineer to know i f it is necessary to increase the stimulation benefit by
adjusting the constraints. It also gives the estimate of the derivable benefit.
Considering Well BU 1, t he u ltimate s olution obt ained is s hown in A ppendix C, S ection 2 . All
constraints are n ot b inding, m eaning that t he optimum po int of the objective f unction w as
attained before any of the constraint bound was reached. Hence any shift to the right or left of the
optimum point will only decrease the stimulation benefit.
Discounted
Production
Forcast
Oil Price
Payout
(days)
Stimulation
Benefit ($)
Ranking
4000
18.6
64936.28
80
1.62
5194902.4
4th
736662.37
4000
18.9
106275.38
80
0.74
8502030.4
3rd
1200000
1058634.07
4000
19.6
106868.12
80
1.06
8549449.6
1st
1200000
626820.32
4000
20.1
106652.90
80
0.63
8532232.0
2nd
Well
Stimulation
Budget ($)
Stimulation
Cost ($)
($/month)
BU1
1200000
1096162.22
BU2
1200000
BU3
BU4
LOE
(years)
(bbls)
($)
160000
140000
120000
100000
Oil Price = $40/bbl
80000
60000
40000
20000
0
0
Productivity Ratio, F
4.3
The data used for this case study was taken from published literatures by Nnanna et al.73, Nnanna
and A jienka76, a nd Onyekonwu80. T hese data ar e u sed to formulate t he Well BU 5 Model. The
data for Well BU 5 is presented in Table 4.4.
4.3.1 Formulation of the Well BU 5 Model
Lets assume that the remaining recoverable reserve in the drainage area of this well is 2.5 MM
bbls, and the tubing maximum design flow rate is 10000 stb/d. Also lets assume that the price of
oil is $80/bbl, the effective discount rate is 10% and the maximum budget for the acidizing job is
set at $1000000.
The model is formulated as follows:
Step A: Enter t he da ta gi ven i nto t he A cidizing Design and Optimization Model and ge nerate
the Cost versus Productivity-Ratio data .
The Cost versus Productivity-Ratio plot for Well BU 5 is shown in Fig. 4.9.
Step B: Obtain t he relationship be tween the s timulation c ost (C) a nd productivity ratio (F) in
form of power equation of a trendline through a log-log regression of the data.
For the input data, this equation is obtained as:
= 27.826.187 .................4.9
The above relationship, a s presented in Equation 4.1 was obtained from a regression analysis of
the model generated cost data using Microsoft Excel.
3850 psi
Drainage Radius, r e
1000 ft
Wellbore Radius, r w
0.3 ft
68.2 ft
Depth of Formation
12100 ft
6 ft
1050 md
100 md
Porosity
25%
1.159 bbl/stb
0.45 psi/ft
1.04
0.57 cp
Pump Rate
2 bbl/min
200 psi
1.75 inches
$ 30 per gal
0.4 psi/ft
500 stb/d
Abandonment Rate, q a
150 stb/d
Duration of Stimulation, t s
1 day
80 $/stb
1400000.00
1200000.00
1000000.00
Cost, $
800000.00
600000.00
400000.00
200000.00
0.00
1.00
1.50
2.00
2.50
3.00
3.50
4.00
Productivity Ratio, F
4.50
5.00
5.50
6.00
Step C: Using Equations 3.49, 3.52, 3.60, 3.62, 3.63, 3.64 & 3.65, the non linear programming
model can be formulated as a maximization problem as presented below.
:
1. Break-even Constraint:
20 .4.13
1 ..4.15
4.79 ..4.16
In t his pr oblem, we s eek a n o ptimum value for t he productivity r atio with a l ower a nd upp er
bound of 1 and 4.79 respectively. Equation 4.16 is more binding than Equation 4.13.
4.3.2 Solution of the Well BU 5 Model
The Well BU 5 Model was solved u sing t he Microsoft Excel Solver and Whats Best 10.0, t he
results are presented in Appendix C and D, Case Study 2. The solutions gave the same result for
the optimal point. The value of the objective function at optimal productivity ratio is:
= 10846.95
The o ptimal p roductivity r atio i s e qual to 4 .18. The o bjective function be havior i s vi ewed
Mathematica 7.0. The plot generate i s shown i n Appendix E . Also, using Mathematica 7.0 to
solve t he n on-linear pr ogramming optimization pr oblem, the r esult for t he ob jective function i s
10276.7 bbls, while the value of the productivity ratio at the optimal point is 3.97. The solution
using Mathematica 7.0 is also shown in Appendix E.
The slight variation in the results obtained is due to same reasons as discussed in section 4.2.2.
4.3.3 Discussion of the Well BU 5 Model Result
This op timal v alue of t he o bjective f unction i s 10846.95 bbls, meaning th at i f th is well i s
considered f or s timulation, gi ven t hat the assumptions considered in s ection 4.3.1 a re binding,
the benefit derivable from the stimulation equals ($80/bbl 10846.95 bbls ), i.e. $867,756.
The Answer Report for Case Study 2 Section 1, in Solvers Result section of Appendix C shows
that t he six c onstraints a re n ot bi nding. T his m eans t hat t he o ptimal solution was f ound within
the limits of all the constraints. Hence no constraint is violated.
If we a ssume t hat t he stimulation budget a pproved by management i s $185, 000 for t his well,
then t he ou tput o f t his model will be a s shown i n Appendix C , Case Study 2 - Section 2 . T he
Answer Report Section shown that the budget allocation constraint is now binding. This means
that if management is willing to allocate more money to this project, there will be more benefit.
From the sensitivity report it could be seen that the value of the Lagrange Multiplier associated
with the budget allocation constraint is 0.0004047. H ence i f the stimulation budget is increased
by $1, the benefit derived from stimulation will increase by 0.0405%. This can be interpreted in
a much better sense a s $ 1 increase i n stimulation budget will r esult i n a n additional production
benefit of 0.0004047 bbls.
The a mount in d ollars, X, needed to be added to the present budget in order to get an optimum
result can be roughly approximated with the following relationship (valid only for this case):
100
1100
..4.17
budget allocation constraint. is obtained from the optimization model sensitivity result.
For the above result the amount needed to be a dded i n order to get an op timum benefit i s
estimated as:
1000.000405185000
(11000.000405)
$7809
12000
10000
8000
Oil Price = $40/bbl
6000
4000
2000
0
0
Productivity Ratio, F
Chapter Five
5.1 Conclusion
This r esearch seeks a method to quantify stimulation benefits derivable from different candidate
wells, a nd us e the result t o rank economically profitable candidates. To achieve t his, a d esign
module was developed for a cidizing based o n t he works o f Schechter a nd G idley32 , a nd
Economides and Nolte86 . O ther design m odules were a lso developed f or gravel packing a nd
recompletion stimulation techniques.
The o ptimization model derived i n t his r esearch c ombines the o utput from t he stimulation
treatment de sign m odule w ith production de cline c urve analysis and economic c ontinuous
discounting c oncepts. The o bjective f unction is f ormulated in the f orm of
a no n-linear
this mo del, stimulation decisions should be based on the cost of the project, payout time and the
stimulation benefit.
A sensitivity a nalysis on t he r esults of t he m odel w as a lso performed. B ased on t he s ensitivity
analysis and the results of the Bestfield Model, it can be concluded that:
1. The proposed methodology and models can be quantitatively used to estimate the benefits
derivable from stimulation options like: acidizing, gravel-packing and recompletion.
2. The mo del and non -linear o ptimization model can be u sed t o r ank c andidate w ells for
selective stimulation. Hence it can be used for stimulation candidate well selection.
3. Below a p roductivity r atio of 3, t he di scounted pr oduction from a cidizing does n ot
depend on the price of oi l. H owever, the ove rall m onetary be nefit derived fro m
stimulation depends on price of oil.
4. The o ptimization m odel can a lso b e u sed to study t he effect o f t he treatment parameters
on the objective function.
5.2 Recommendation
The following recommendations are presented to highlight areas of additional research to
improve the methodology and models developed in this research.
It is recommended that the model be used t o quantify stimulation be nefit derivable from a
stimulation decision o nce a well has been matched t o either of acidizing, gravel packing or
recompletion. F or e ffective use o f the model, i t i s r ecommended t hat t he l ease o perating cost
(LOE) and also, federal and state taxes be considered before ranking the wells for stimulation.
The effect of t he pr e-stimulation p roduction rate on t he o ptimal point of t he objective function
(Fig. 4 .4) needs further investigations. S uch study will help to k now optimal t ime to initiate
stimulation jobs during the producing life of a well.
It i s al so r ecommended, for f urther s tudy, that a nother a pproach, other t hat t he pr oduction
decline curve analysis, that can be used to quantify the stimulation benefit be investigated. This
References
1. Civan, F .: Reservoir Formation Damage Fundamentals, Modeling, Assesment
and Mitigation, Gulf Publishing Company, Houston, Texas (2000) p.1
2. Bennion, B ., Formation D amageThe I mpairment o f th e I nvisible, b y the
Inevitable and Uncontrollable, R esulting i n an Indeterminate Reduction of t he
Unquantifiable! Journal of Canadian Petroleum Technology, V ol. 3 8, No. 2,
February 1999, pp. 11-17.
3. Porter, K. E ., An O verview of F ormation D amage, Journal of Petroleum
Technology, Vol. 41, No. 8, 1989, pp. 780-786.
4. Mungan, N ., Discussion o f an Overview o f F ormation D amage, Journal of
Petroleum Technology, Vol. 41, No. 11, Nov. 1989, p. 1224
5. Amaefule, J. O., Ajufo, A., Peterson, E., & Durst, K., "Understanding Formation
Damage Processes: An E ssential I ngredient for I mproved Measurement a nd
Interpretation o f R elative P ermeability Data," SPE 16232 p aper, SPE Production
Operations Symposium, 1987, Oklahoma City, Oklahoma
6. Van Everdingen, A.F. a nd H urst, W .: The Application o f the L aplace
Transformation to Flow Problems in Reservoirs, Trans., AIME (1949) 186, 305
324.
7. Hawkins, M.F.: A Note on the Skin Effect, Journal of Petroleum Technology
(December 1956) 8, 356357.
8. Frick, T .P. a nd Economides, M .J.: Horizontal We ll Damage Characterization
and Removal, SPE Production & Facilities (February 1993) 8, No. 1, pp.1522.
9. Piot, B . M., & L ietard, O . M ., Nature of Formation Damage in Reservoir
Stimulation, in Economides, M. J. & Nolte, K. S., (eds.), Reservoir Stimulation,
Schlumberger Education Services, Houston, Texas, 1987.
10. Economides, M.J., and N olte, K.G., Reservoir Simulation, Third Edition. Wiley,
N.Y. (hardbound) 2000, Chapter One, p. 1-12
11. Cinco-Ley, H., Ramey, H.J. Jr. and Miller, F.G.: Pseudoskin Factors for Partially
Penetrating D irectionally D rilled Wells, paper SPE 5589, presented a t t he S PE
Annual Technical C onference and Exhibition, Dallas, Texas, USA (September
28October 1, 1975a).
12. Harris, M.H.: The Effect o f P erforating o n Well Productivity, paper SPE 1236,
Journal of Petroleum Technology (April 1966) 18, pp. 518528.
13. Karakas, M. a nd T ariq, S .: Semi-Analytical Productivity M odels for P erforated
Completions, paper S PE 18247, p resented a t t he S PE Annual Technical
Conference and Exhibition, Houston, Texas, USA (October 25, 1988).
14. Economides, M.J., a nd N olte, K.G., Reservoir Simulation, T hird E dition. W iley,
N.Y. (hardbound) 2000, Chapter One, p. 1-13
15. Yan, J ., J iang, G ., & Wu, X., Evaluating o f F ormation D amage C aused by
Drilling a nd Completion Fluids i n H orizontal W ells, Journal of Canadian
Petroleum Technology, Vol. 36, No. 5, 1997, pp. 36-42.
16. McLeod, Jr., H. O a nd C oulter, Jr., A. W.: The Stimulation T reatment P ressure
Record An Overlooked Formation Evaluation Tool Journal of Petroleum
Technology (August, 1969) p. 952
17. Raymond, L . R . a nd H udson, J . L .: Short-Term Well Te sting to D etermine
Wellbore Damage, Journal of Petroleum Technology (Oct. 1966) 1363-1370
18. Amaefule, J. O., Kersey, D. G., Norman, D. L., & Shannon, P. M., Advances in
Formation Da mage Assessment a nd C ontrol Strategies, CIM P aper No. 88 -3965, Proceedings of t he 39th Annual T echnical Meeting o f P etroleum S ociety of
CIM an d C anadian G as P rocessors Association, Ju ne 1 2-16, 1988, C algary,
Alberta, 16 p. 65-2
19. Muskat, M., The Flow of Homogeneous Fluids Through Porous Media, McGrawHill Book Co., New York, New York, 1937
20. Civan, F .: Reservoir Formation Damage Fundamentals, Modeling, Assesment
and Mitigation, Gulf Publishing Company, Houston, Texas (2000) p.688
21. Leontaritis, K. J ., Asphaltene Ne ar-Wellbore F ormation D amage Modeling,
SPE 39446 p aper, P roceedings of t he 1998 S PE Formation Damage C ontrol
Conference, February 18-19, 1998, Lafayette, Louisiana, pp. 277-288.
22. Civan, F .: Reservoir Formation Damage Fundamentals, Modeling, Assesment
and Mitigation, Gulf Publishing Company, Houston, Texas (2000) p.688
23. Zhu, D ., H ill, A. D ., & Morgenthaler, L . N ., Assessment o f M atrix Ac idizing
Treatment R esponses i n G ulf of M exico W ells, Proceedings of t he 199 9 S PE
Mid-Continent O perations Symposium held in Oklahoma City, Oklahoma, USA,
March 28-31, 1999.
48. Economides, M.J., Hill, A.D. and Ehlig-Economides, C.A.: Petroleum Production
Systems, Englewood Cliffs, New Jersey, USA, Prentice-Hall (1994) p.359
49. Guin, J.A., Schechter, R.S. and Silberberg, I.H.: Chemically Induced Changes in
Porous Media, Ind. & Eng. Chem. Fund. (February 1971) 10, No.1, pp. 5054.
50. Labrid, J .C.: Thermodynamic and Kinetic Aspects of Argillaceous S andstone
Acidizing, paper S PE 5165, Journal of Petroleum Technology (April 1975), pp.
117128.
51. Lambert, M.E.: A Statistical Study of Reservoir Heterogeneity, MS thesis, The
University of Texas at Austin, Austin, Texas, USA (1981).
52. Lund, K. a nd Fogler, H .S.: Acidization V. T he Prediction of t he M ovement o f
Acid a nd P ermeability F ronts i n S andstone, Chemical Engineering Science
(1976) 31, No. 5, pp. 381392.
53. McLeod, H .O. J r.: Matrix Acidizing, paper SPE 13752, Journal of Petroleum
Technology (December 1984) 36, pp. 20552069.
54. Shaughnessy, C.M. and Kunze, K.R.: Understanding Sandstone Acidizing Leads
to Improved Field Practices, paper SPE 9388, Journal of Petroleum Technology
(July 1981), pp. 11961202.
55. Hoefner, M.L. and Fogler, H .S.: Pore Evolution a nd Channel Formation During
Flow and Reaction in Porous Media, AIChEJ. (Jan., 1988) 34, No.1, pp. 4554.
56. Hung, K.M., Hill, A.D. and Sepehrnoori, K.: A Mechanistic Model of Wormhole
Growth i n Carbonate M atrix A cidizing a nd Acid Fracturing, paper SPE 16886,
Journal of Petroleum Technology, (January 1989) 41, No. 1, 5966.
57. Daccord, G ., T ouboul, E . a nd L enormand, R .: Carbonate A cidizing: T oward a
Quantitative M odel of t he Wormholing Phenomenon, p aper SPE 16887, SPE
Production Engineering (February 1989), pp. 6368.
58. Pichler, T., Frick, T.P., Economides, M.J. and Nittmann, J.: Stochastic Modeling
of W ormhole Growth i n C arbonate A cidizing with B iased R andomness, pa per
SPE 25004, p resented a t t he S PE European P etroleum C onference, Cannes,
France (November 1618, 1992).
59. Economides, M.J., Hill, A.D. and Ehlig-Economides, C.A.: Petroleum Production
Systems, Englewood Cliffs, New Jersey, USA, Prentice-Hall (1994) p.400
60. Fredd, C.N. and Fogler, H.S.: Influence of Transport and Reaction on Wormhole
Formation in Porous Media, AIChE J. (September 1998b), pp. 19331949.
Nomenclature
C max
cost of perforation, $
q max
index of anisotropy
vertical permeability, md
horizontal permeability, md
reservoir permeability, md
reservoir permeability, md
Pr
number of perforations
wellbore radius, ft
ROIP
porosity, fraction
porosity, fraction
specific gravity of the acid (or density of acid in g/cc)
Appendix A
A Simple Well Screening Flow Chart
No
No
No
No
Re-evaluate
well problem
Positive
skin effect?
Mechanical
problems?
(e.g plugged perf.)
Yes
high water
production?
Yes
high sand
production?
Yes
Yes
gravel
packing
recompletion
recompletion
matrix
acidizing
Appendix B
Appendix C
Solver Results
CASE STUDY 1
(a)
(b)
Appendix D
Whats Best 10.0 Results
CASE STUDY 1
CASE STUDY 2
Appendix E
Mathematica 7.0 Results
CASE STUDY 1
Mathematica 7.0 inpretation of input data and results for Well BU 3 Model
CASE STUDY 2
Appendix F
Derivation of the Objective Function for Other Decline Cases
The derivation of the optimization model presented in this section is modified from the published
work of Sinson et al.85 Let u s start by deriving t he o ptimization model for the ge neral de cline
curve analysis.
F.1
Where:
(F.1)
[1+ ]
= hyperbolic constant
The general form for the discounted production from stimulation, , is expressed as:
[1+ ( )]
(F.3)
(F.4)
( 1 + )
(F.2)
(F.5)
1
1+ ]
1
(F.6)
(F.7)
= 0
[1+ ( )]
(F.8)
(F.9)
Note that the general form of the solution for and using integral transformations is of
the form:
= 2
= 1
(F.11)
(F.12)
(F.13)
= ()
2
(F.10)
1 1
Note that equation (F.13) is the most common form of hyperbolic decline curve.
Let:
1
(F.14)
= ,
and =
(F.15)
F.2
(F.16)
=
Where:
1+ ( )
(F.17)
Such that:
and
(F.18)
= 1 +
(F.19)
( 1 + )
(F.20)
Substituting (F.19) and (F.20) into the original equation, (F. 17):
Integrating:
( )
( )
(F.21)
(1 + )
Simplifying:
and:
=
( )
( )
(F.23)
(F.24)
(F.22)
(F.25)
Where:
= abandonment production rate
( )
(F.26)
= 0
1+
(F.27)
Such that:
(F.18)
= 1 +
=
and
(F.19)
( 1 + )
We get:
and
Therefore:
(F.28)
(1 + )
(F.20)
(F.29)
(F.30)
(F.31)
= 1 (2 3 ) 1 (4 ) 6 5
(F.32)
(F.33)
Where:
1 =
( )
2 =
3 =
4 =
5 =
F.2.2 Constraints
(F.34)
(F.35)
(F.36)
6 =
(F.37)
(F.38)
(F.39)
The same constraints formulated in the exponential case also applied here.
Constraint 1. Break-even point
The incremental revenue from any stimulation decision should be greater than or at least equal to
the cost of the project.
(1 + ) + 0 1+
(F.40)
Performing the integration and using the definition of given in equation (F.25) and using the
constants above, we get:
1 (2 3 ) 1 (4 ) 6 5
(F.41)
The recovery from the stimulation cannot exceed the remaining oil in place.
(1 +
)
[ ]
(F.42)
(F.43)
and:
where:
(F.44)
7 (8 9 ) 10
(F.45)
ln
7 =
8 =
9 =
10 =
(F.46)
(F.47)
(F.48)
(F.49)
(F.50)
(F.51)
The maximum attainable pr oductivity ratio from stimulation de pends on t he reservoir pr operties
and treatment parameters.
(F.52)
(F.53)
Where 0 and 1 are the intercept and slope of a regression line through the data.
F.2.3 Form of the NLP
Equations (F.33) and (F.53) define the NLP model for the harmonic case.
Maximize:
= 1 (2 3 ) 1 (4 ) 6 5
(F.33)
Subject to:
1 (2 3 ) 1 (4 ) 6 5
7 (8 9) 10
= 100 1
(F.41)
(F.45)
(F.50)
(F.51)
(F.52)
(F.53)
Where:
1+ 2 ( )
(F.54)
= 1
Such that:
and
(F.55)
(F.56)
(F.57)
( 1 +
=
)
2
Substituting equations (F.55), (F.56) and (F.57) into (F. 54), then simplifying:
=
(F.58)
0.5
(F.59)
0.5
2
2 +
where:
2
1 =
2 = +
3 =
(F.60)
+ 1 3 2 3
0.5
2 0.5
(F.61)
(F.62)
(F.63)
= 0
1+ 2
(F.64)
1+
2
1+ 2
1 +
Simplifying further:
1+ 2
(F.65)
(F.66)
We can define:
4 =
1+ 2
(F.67)
We can define:
5 =
Therefore:
(F.69)
(F.70)
= 5
6 =
3 8 6
0.5
9 + 7 0.5
+ 1 3 2 3 4 5
0.5
7 =
0.5
8 =
F.3.2 Constraints
9 =
(F.32)
(F.71)
(F.72)
(F.73)
(F.74)
(F.75)
1+ 2 ( )
+ 0
1+ 2
(F.76)
Evaluating:
0.5
2
0.5
+
2
+ 1 3 2 3 4 + 5
(F.77)
1+ 2 ( )
(F.78)
1+ 2 2
Where:
7 0.5 1 1
7 0.5 9
10 =
(F.79)
(F.80)
(F.81)
(F.82)
The ma ximum attainable pr oductivity ratio from stimulation de pends on t he reservoir pr operties
and treatment parameters.
(F.52)
(F.53)
Where 0 and 1 are the intercept and slope of a regression line through the data.
A.3.3 Form of the NLP
The equation (F.71), together with all the constraints considered, can be summarized as:
Maximize:
Subject to:
3 8 6
0.5
9 + 7 0.5
+ 1 3 2 3 4 5
(F.71)
0.5
2
0.5
+
2
10
7 0.5 1 1
7 0.5 9
+ 1 3 2 3 4 + 5
F.4
Summary
= 100 1
(F.77)
(F.80)
(F.82)
(F.51)
(F.52)
(F.53)