research article explicit deconvolution of well test data
TRANSCRIPT
Research ArticleExplicit Deconvolution of Well Test Data Dominated byWellbore Storage
K Razminia1 A Hashemi1 A Razminia2 and D Baleanu345
1 Department of Petroleum Engineering Petroleum University of Technology Ahwaz Iran2Dynamical Systems amp Control (DSC) Research Laboratory Department of Electrical Engineering School of EngineeringPersian Gulf University P O Box 75169 Bushehr Iran
3Department of Chemical and Materials Engineering Faculty of Engineering King Abdulaziz University Jeddah Saudi Arabia4Department of Mathematics and Computer Sciences Faculty of Arts and Sciences Cankaya University 06530 Ankara Turkey5 Institute of Space Sciences Magurele-Bucharest Romania
Correspondence should be addressed to A Razminia arazminiagmailcom
Received 27 December 2013 Accepted 28 February 2014 Published 3 April 2014
Academic Editor Abdon Atangana
Copyright copy 2014 K Razminia et al This is an open access article distributed under the Creative Commons Attribution Licensewhich permits unrestricted use distribution and reproduction in any medium provided the original work is properly cited
This paper addresses some methods for interpretation of oil and gas well test data distorted by wellbore storage effects Using thesetechniques we can deconvolve pressure and rate data from drawdown and buildup tests dominated by wellbore storage Some ofthese methods have the advantage of deconvolving the pressure data without rate measurement The two important methods thatare applied in this study are an explicit deconvolution method and a modification of material balance deconvolution method Incases with no rate measurements we use a blind deconvolution method to restore the pressure response free of wellbore storageeffects Our techniques detect the afterflowunloading rate function with explicit deconvolution of the observed pressure data Thepresented techniques can unveil the early time behavior of a reservoir system masked by wellbore storage effects and thus providepowerful tools to improve pressure transient test interpretation Eachmethod has been validated using both synthetic data and fieldcases and each method should be considered valid for practical applications
1 Introduction
In conventional well test analysis the pressure responseto constant rate production is essential information thatpresents the distinct characteristics for a specific type ofreservoir system However in many cases it is difficultto acquire sufficient constant rate pressure response dataThe recorded early time pressure data are often hidden bywellbore storage (variable sandface rates) In some casesouter boundary effects may appear before wellbore storageeffects disappear Therefore it is often imperative to restorethe early time pressure response in the absence of wellborestorage effects to provide a confident well test interpretation[1ndash4]
A considerable amount of work has been done onmultirate (variable) tests during the last 50 years Howeverthese are basically sequential constant rate drawdowns onlytransient pressure is measured and rate is assumed to be
constant during each drawdown test [5] All these worksdeal with the direct problem In other words the constantrate solution (the influence or the unit response function)is convolved (superimposed) with the time dependent innerboundary condition to obtain solutions to the diffusivityequation This process is called convolution The process ofdetermining the influence function is called deconvolution[6ndash8]
Based on [1] deconvolution is a technique used to convertmeasured pressure and sandface rate data into the constantrate pressure response of the reservoir In other words decon-volution provides the pressure response of a wellreservoirsystem free of wellbore storage effects as if the reservoir isproducing at a constant rate Once the deconvolved pressureis obtained conventional interpretationmethods can be usedfor reservoir system identification and parameter estimation
The purpose of the well test interpretation as stated byGringarten et al [9] is to identify the system and determine
Hindawi Publishing CorporationAbstract and Applied AnalysisVolume 2014 Article ID 912395 12 pageshttpdxdoiorg1011552014912395
2 Abstract and Applied Analysis
its governing parameters frommeasured data in the wellboreand at the wellhead This problem is known as the inverseproblem The solution of the inverse problem usually is notunique Gringarten et al [9] pointed out that if the numberand the rage measurements increased the nonuniqueness ofthe inverse problem will be reduced Thus combining sand-face flow rate with pressure measurements will enhance theconventional (including type curve) well test interpretationmethods On the other hand mathematically deconvolutionis also a highly unstable inverse problem because smallerrors in the data can result in large uncertainties in thedeconvolution solution [10 11]
Unfortunately standard deconvolution techniques re-quire accurate measurements of flow rate and pressure atdownhole (or sandface) conditions While accurate pressuremeasurements are commonplace the measurement of sand-face flow rates is rare essentially nonexistent in practiceAs such the deconvolution of wellbore storage distortedpressure test data is problematic In theory this process ispossible but in practice without accurate measurements offlow rates this process cannot be employed [12 13]
In the past 50 years a variety of deconvolution tech-niques have been proposed in petroleum engineering Russell[14] used the approximate direct method to correct thepressure transient data distorted by wellbore storage intothe equivalent pressure function for the constant rate caseGladfelter et al [15] and Fetkovich and Vienot [16] used therate normalization techniques to correct for thewellbore stor-age effectsThese rate normalizationmethods were successfulin some cases So Johnston [17] provided thematerial balancedeconvolution that is a practical approach for the analysis ofpressure transient data distorted by wellbore storage effects
Essentially rate normalization techniques and materialbalance deconvolution are restricted when the lack of ratemeasurement exists van Everdingen [18] and Hurst [19]demonstrated empirically that the sandface flow rate profilecan be modeled approximately using an exponential relationfor the duration of wellbore storage distortion during a pres-sure transient testThe vanEverdingen andHurst exponentialrate model is given in dimensionless form as
119902119863(119905119863) = 1 minus 119890
minus120573119905119863 (1)
Further they showed that the rate-time relationship duringafterflow (for a pressure buildup test) or unloading (in apressure drawdown test) is a function of the pressure dropchange with respect to time and a relatively constant wellborestorage coefficient
Based on a material balance in the wellbore the sandfaceflow rate can be calculated by the following relation given indimensionless form
119902119863(119905119863) = 1 minus 119862
119863
119889119901119908119863
119889119905119863
(2)
where it can be noted that in the development of wellborestorage models solutions (eg type curves) a constant well-bore storage coefficient is always assumed
Equations (1) and (2) laid the groundwork for 120573-deconvolution Joseph and Koederitz [20] and Kuchuk [21]
applied 120573-deconvolution for the analysis of wellbore stor-age distorted pressure transient data The 120573-deconvolutionformula which computes the undistorted pressure dropfunction directly from the wellbore storage affected data isgiven as
119901119904119863
(119905119863) =
1
120573
119889119901119908119863
(119905119863)
119889119905119863
+ 119901119908119863
(119905119863) (3)
It should be noted that (3) is only validwhen the sandface flowrate profile follows an exponential trend as prescribed by (1)[12] In thisworkwe develop an approach that overcomes thisrestriction by treating the 120573 quantity as a variable in generalrather than a constant
The obvious advantage of 120573-deconvolution is that thewellbore storage effects are eliminated and the restoredpressure is obtained using only the given pressure data
This paper reports a modification of Russell method(Appendix A) a modification of 120573-deconvolution and anexplicit deconvolution formula to restore the pressure andrate data and a review on material balance deconvolutionis also presented to deconvolve the pressure and rate dataafter the explicit deconvolution method Then a generalapproach to analyze the deconvolution problem is provided(Appendix B) In verification examples section we show thepower and practical applicability of our approaches withvarious synthetic and field case examples for oil and gas wells
2 Main Results
21 120573-Deconvolution and Its Modification van EverdingenandHurst [22] presented the dimensionless wellbore pressurefor a continuously varying flow rate as
119901119908119863
(119905119863) = 119902119863 (0) 119901119904119863 (119905119863) + int
119905119863
0
1199011015840
119863(120591) 119901119904119863 (119905119863 minus 120591) 119889120591
(4)
An alternative form to (4) can be obtained by integration byparts as
119901119908119863
(119905119863) = 119902119863(119905119863) 119901119904119863 (0) + int
119905119863
0
119902119863 (120591) 119901
1015840
119904119863(119905119863minus 120591) 119889120591
(5)
where 119901119908119863
(119905119863) = (119896ℎ1412119902119861120583)(119901
119894minus 119901119908119891(119905)) 119905
119863=
(000026371198961199051206011205831198881199051199032
119908) 119901119904119863
= 119901119863+ 119878 119901
119908119863(119905119863) is dimen-
sionless wellbore pressure with wellbore storage and skineffects 119901
119863(119905119863) is dimensionless sandface pressure for the
constant rate case without wellbore storage and skin effects 119878is steady state skin factor 1199011015840
119904119863(119905119863) = 119889119901
119904119863(119905119863)119889119905119863 119902119863(119905119863) =
119902119904119891(119905119863)119902119903 1199021015840119863(119905119863) = 119889119902
119863(119905119863)119889119905119863 and 119902
119903is reference
flow rate if the stabilized constant rate is available then 119902119903
should be replaced by 119902119877 119902119904119891(119905) is variable sandface flow rate
(flowmeter reading) and 119902119863(119905119863) is dimensionless sandface
rateIt should be emphasized that (4) and (5) can be applied
for many reservoir engineering problems The linearity ofdiffusivity equation allows us to use (4) and (5) for fractured
Abstract and Applied Analysis 3
layered anisotropic and heterogeneous systems as long as thefluid in the reservoir is single phaseThese equations can alsobe applied to both drawdown and buildup tests if the initialconditions are known For a reservoir with an initial constantand uniform pressure distribution (119901
119863(0) = 0) (4) and (5)
can be expressed as
119901119908119863
(119905119863) = int
119905119863
0
1199021015840
119863(120591) 119901119904119863 (119905119863 minus 120591) 119889120591 (6)
119901119908119863
(119905119863) = 119878119902
119863(119905119863) + int
119905119863
0
119902119863 (120591) 119901
1015840
119904119863(119905119863minus 120591) 119889120591 (7)
Furthermore (6) and (7) also can be expressed as
119901119908119863
(119905119863) = int
119905119863
0
1199021015840
119863(119905119863minus 120591) 119901
119904119863 (120591) 119889120591 (8)
119901119908119863
(119905119863) = 119878119902
119863(119905119863) + int
119905119863
0
119902119863(119905119863minus 120591) 119901
1015840
119904119863(120591) 119889120591 (9)
In (6) and (8) it is assumed that 1199021015840119863(119905119863) exists If 119902
119863(119905119863) is
constant then (7) and (9) must be used Equations (6) and(8) are known as a Volterra integral equation of first kind andthe convolution type
Taking the Laplace transform of (6) yields
119901119908119863
(119904) = 119904119902119863(119904) 119901119904119863
(119904) (10)
Rearranging (10) for the equivalent constant rate pressuredrop function 119901
119904119863(119904) we obtain
119901119904119863
(119904) =119901119908119863
(119904)
119904119902119863(119904)
(11)
By taking the Laplace transform of (1) one can easily obtain
119902119863(119904) =
1
119904minus
1
119904 + 120573 (12)
Substituting (12) into (11) and then taking the inverse Laplacetransform of (11) yield the beta deconvolution formula
119901119904119863
(119905119863) =
1
1205731199011015840
119908119863(119905119863) + 119901119908119863
(119905119863) (13)
where
1199011015840
119908119863(119905119863) =
119889119901119908119863
119889119905119863
(14)
where we note that (13) is specifically valid only for theexponential sandface flow rate profile given by (1) This maypresent a serious limitation in terms of practical applicationof the 120573-deconvolutionmethod To overcome this restrictionwe propose an idea It would be assumed that120573 is not constantin general If 119902
119863(119905119863)data are given the procedure is as follows
119902119863(119905119863) = 1 minus 119890
minus120573(119905119863)119905119863 (15)
Rearranging (15) yields
1 minus 119902119863(119905119863) = 119890minus120573(119905119863)119905119863 (16)
then
minus120573 (119905119863) 119905119863= ln (1 minus 119902
119863(119905119863)) (17)
so
120573 (119905119863) = minus
ln (1 minus 119902119863(119905119863))
119905119863
(18)
Substituting of (18) into (13) instead of 120573 term yields themodified 120573-deconvolution
119901119904119863
(119905119863) =
1
120573 (119905119863)1199011015840
119908119863(119905119863) + 119901119908119863
(119905119863) (19)
Plot of (19) versus (int119905119863
0119902119863(120591)119889120591119902
119863(119905119863)) and (int
119905119863
0(1 minus
119902119863(120591))1198891205911 minus 119902
119863(119905119863)) for drawdown and buildup tests
respectively should be considered more accurate to use as apractical tool for field applications As mentioned before thevariable 120573 quantity (rather than a constant) shows the generaland more accurate behavior of 119902
119863(119905119863) So accuracy of (19) is
also more than (13) It can be emphasized that (13) and (19)are applicable in the wellbore storage duration and for thelater times the given pressure data are correct However theresults of (19) show that this equation is not bad for the latertimes
22 Explicit Deconvolution Formula A few types of approx-imation functions can be used to approximate 119902
119863(119905119863) An
approximate proof is given here to show that the exponentialfunction gives a good representation of 119902
119863(119905119863) By setting
119902119863(119905119863) = 1 minus 119891 (119905
119863) (20)
119891 (0) = 1 (21)
Equation (21) is an initial condition for (20)Taking the time derivative of (20) gives
1199021015840
119863(119905119863) = minus119891
1015840(119905119863) (22)
Taking the time derivative of (2) gives
1199021015840
119863(119905119863) = minus119862
11986311990110158401015840
119908119863 (23)
From (2) and (20) we have
119891 (119905119863) = 1198621198631199011015840
119908119863 (24)
And from (22) and (23) we have
1198911015840(119905119863) = 11986211986311990110158401015840
119908119863 (25)
After dividing right and left hand sides of (25) by (24) theresult is
1198911015840(119905119863)
119891 (119905119863)=11990110158401015840
119908119863(119905119863)
1199011015840119908119863
(119905119863) (26)
Integrating both sides of (26) with respect to 119905119863from 0 to 119905
119863
yields
int
119905119863
0
1198911015840(120591)
119891 (120591)119889120591 = int
119905119863
0
11990110158401015840
119908119863(120591)
1199011015840119908119863
(120591)119889120591 (27)
4 Abstract and Applied Analysis
Using elementary calculus yields
ln (1003816100381610038161003816119891(120591)
1003816100381610038161003816)1003816100381610038161003816119905119863
0= ln (1003816100381610038161003816119891 (119905
119863)1003816100381610038161003816) minus ln (1) = int
119905119863
0
11990110158401015840
119908119863(120591)
1199011015840119908119863
(120591)119889120591
(28)
So
119891 (119905119863) = 119890120572(119905119863) (29)
where
120572 (119905119863) = int
119905119863
0
11990110158401015840
119908119863(120591)
1199011015840119908119863
(120591)119889120591 (30)
Now the explicit deconvolution of data based on (20) can bewritten as follows
119901119904119863
(119905119863) = minus
1
120572 (119905119863) 119905119863
1199011015840
119908119863+ 119901119908119863
(119905119863) (31)
or
119901119904119863
(119905119863) = minus
119905119863
120572 (119905119863)1199011015840
119908119863+ 119901119908119863
(119905119863) (32)
that is similar to (19) The other similarity between (19) and(32) is that the plot of (32) versus (int119905119863
0119902119863(120591)119889120591119902
119863(119905119863)) and
(int119905119863
0(1 minus 119902
119863(120591))119889120591(1 minus 119902
119863(119905119863))) for drawdown and buildup
tests respectively yields the more accurate results and isusable for practical applications The applicability of (32) isalso for duration of wellbore storage effects And for the othertimes the given pressure data are correct However using thismethod for the later times yields not bad results
3 Material Balance Deconvolution
Material balance deconvolution is an extension of the ratenormalization method Johnston [17] defines a new 119909-axisplotting function (material balance time) that provides anapproximate deconvolution of the variable rate pressuretransient problem There are numerous assumptions associ-ated with the material balance deconvolution methods oneof the most widely accepted assumptions is that the rateprofilemust change smoothly andmonotonically In practicalterms this condition should be met for the wellbore storageproblem
The general form of material balance deconvolution isprovided for the pressure drawdown case in terms of thematerial balance time function and the rate-normalizedpressure drop function The material balance time functionis given as
119905119898119887
=119873119901
119902119863 (119905)
(33)
where
119873119901= int
119905
0
119902119863 (120591) 119889120591 (34)
The rate-normalized pressure drop function is given by
Δ119901119908119891 (119905)
119902119863 (119905)
=119901119894minus 119901119908119891 (119905)
119902119863 (119905)
(35)
The material balance time function for the pressure buildupcase is given as
119905119898119887
=119873119901
1 minus 119902119863 (Δ119905)
(36)
where
119873119901= int
Δ119905
0
(1 minus 119902119863 (120591)) 119889120591 (37)
The rate-normalized pressure drop function for the pressurebuildup case is given as
Δ119901119908119904 (Δ119905)
1 minus 119902119863 (Δ119905)
=119901119908119904 (Δ119905) minus 119901
119908119891 (Δ119905 = 0)
1 minus 119902119863 (Δ119905)
(38)
4 Methodology
In this work several methods are provided to restore thepressure datawithout any sandface flow rate informationThevalues of sandface flow rate data for all methods that requirethese data may be obtained from (20) We evaluated a veryold correctionmethod by Russell [14] However we modifiedthis method by least square method but it was found thatthis method is to be unacceptable for all applications Thederivation of this method and its modification is given inAppendix A
The 120573-deconvolution and its modification are anothertechnique that is mentioned in this study The formulationof this method is presented in Section 21 The other explicitdeconvolutionmethod is an approach similar to themodified120573-deconvolution that uses only the pressure data for decon-volution of pressure and rate data The formulation of thismethod was derived in Section 22 The 120573-deconvolution itsmodification and explicit deconvolution formula can be usedfor practical applications The applicability of these methodsis for duration of wellbore storage effects However usingthese approaches for the other times yields not bad results
The other technique is material balance deconvolutionmethod [17] that has sufficiently accurate resultsThismethodcan also be used as a good approach for practical applicationsThe formulas for this method were given in Section 3
In Appendix B deconvolution is expressed as an inverseproblem that is analyzed by a tricky idea Based on thisidea an interesting proof is given for rate normalizationmethod and material balance deconvolution that is a goodapproach to find the errors and limitations of these methodsand overcome them
Abstract and Applied Analysis 5
10minus3 10minus2 10minus1 100 101 102 1030
500
1000
1500
2000
2500
3000
3500
4000
Time (hr)
Dpwf wellbore storaged(Dpwf)d(ln(t)) wellbore storageDps explicit deconvolution methodd(Dps)d(ln(t)) explicit deconvolution methodDps material balance deconvolution methodd(Dps)d(ln(t)) material balance deconvolution methodDps no wellbore storaged(Dps)d(ln(t)) no wellbore storage
Dpwf
and
its lo
garit
hmic
der
ivat
ive (
psi)
Figure 1 Deconvolution for a drawdown test in a homogeneousreservoir
From the above-mentioned methods we use the fol-lowing two techniques to show the verification of theseapproaches
Explicit Deconvolution Formula The final result of explicitdeconvolution method is given as follows
119901119904119863
(119905119863) = minus
119905119863
120572 (119905119863)1199011015840
119908119863+ 119901119908119863
(119905119863) (39)
Plotting of this term versus (int119905119863
0119902119863(120591)119889120591119902
119863(119905119863)) and
(int119905119863
0(1 minus 119902
119863(120591))119889120591(1 minus 119902
119863(119905119863))) for drawdown and buildup
tests respectively yields more accurate resultsThis approachperforms well in field applications
Material Balance Deconvolution The material balance pres-sure drop function for the pressure drawdown test is givenas
Δ119901119908119891 (119905)
119902119863 (119905)
=119901119894minus 119901119908119891 (119905)
119902119863 (119905)
(40)
and the material balance time function for the pressuredrawdown test is given as
119905119898119887
=int119905119863
0119902119863 (120591) 119889120591
119902119863 (119905)
(41)
For the buildup test the material balance pressure dropfunction can be written as
Δ119901119908119904 (Δ119905)
1 minus 119902119863 (Δ119905)
=119901119908119904 (Δ119905) minus 119901
119908119891 (Δ119905 = 0)
1 minus 119902119863 (Δ119905)
(42)
10minus3 10minus2 10minus1 100 101 102 103
Time (hr)
104
103
102
101Dpwf
and
its lo
garit
hmic
der
ivat
ive (
psi)
Dpwf wellbore storaged(Dpwf)d(ln(t)) wellbore storageDps explicit deconvolution methodd(Dps)d(ln(t)) explicit deconvolution methodDps material balance deconvolution methodd(Dps)d(ln(t)) material balance deconvolution methodDps no wellbore storaged(Dps)d(ln(t)) no wellbore storage
Figure 2 Deconvolution for a drawdown test in a homogeneousreservoir
And the material balance time function can be written as
119905119898119887
=int119905119863
0(1 minus 119902
119863 (120591)) 119889120591
1 minus 119902119863 (Δ119905)
(43)
This technique is a good approach for practical applications
5 Verification Examples
We first use few synthetic cases for different reser-voirwellbore systems to validate the applicability of explicitdeconvolution formula and material balance deconvolutionmethod Then the field applications of these techniqueswill be shown by two real cases The unloading rates arecalculated with (20)
The first case is a synthetic test derived from a test designin a vertical oil wellThe reservoir is homogenous and infiniteacting with a constant wellbore storage The wellbore storagecoefficient is 119862
119904= 0015 bblpsi The pressure data are
obtained from the cylindrical source solution (test design)We used the explicit deconvolution formula and material
balance deconvolution method to deconvolve the pressureand unloading rate distorting by wellbore storage effectsTheresults are presented in Figures 1 and 2 for the semilog andlog-log plots in the case of drawdown test and Figures 3and 4 for the semilog and log-log plots in the case of builduptest Besides the deconvolved pressures and their logarithmicderivatives pressure responses with and without wellborestorage are also plotted on the figures for comparisonThrough the deconvolution process we have successfullyremoved the wellbore storage effects so that radial flow canbe identified at early test times The deconvolved pressuresand their logarithmic derivatives are almost identical to
6 Abstract and Applied Analysis
0
500
1000
1500
2000
2500
3000
3500
4000
105104103102101100
Horner time ratio
Dp wellbore storage
Dps explicit deconvolution method(ln
Dps material balance deconvolution
Dps no wellbore storage
ws
and
its lo
garit
hmic
der
ivat
ive (
psi)
Dpws
d(Dp )d(ln )) wellbore storagews (HTR
d(Dps)d )) explicit deconvolution method(HTR
d(Dps)d(ln )) no wellbore storage(HTR)
d(Dps)d(ln )) material balance deconvolution (HTR
Figure 3 Deconvolution for a buildup test in a homogeneousreservoir
10minus3 10minus2 10minus1 100 101 102 103
Time (hr)
104
103
102
101
ws
ws
and
its lo
garit
hmic
der
ivat
ive (
psi)
Dpws
Dp wellbore storage
Dps explicit deconvolution method
Dps material balance deconvolution
Dps no wellbore storage
d(Dp )d(ln )) wellbore storagews (HTR
d(Dps)d )) explicit deconvolution method(HTR
d(Dps)d(ln )) no wellbore storage(HTR)
d(Dps)d(ln )) material balance deconvolution (HTR
(ln
Figure 4 Deconvolution for a buildup test in a homogeneousreservoir
the analytic solutions (without wellbore storage) whichenables us to estimate reservoir parameters accurately
The second case is an oil well drawdown test in adual porosity reservoir The storativity ratio (120596) and inter-porosity flow coefficient (120582) are assumed to be 01 and10minus6 respectively The constant wellbore storage coefficient
is 119862119904= 015 bblpsi The dimensionless wellbore pressure
10minus3 10minus2 10minus1 100 101 102 103
Time (hr)
600
500
400
300
200
100
0
Dpwf wellbore storaged(Dpwf)d(ln(t)) wellbore storageDps explicit deconvolution methodd(Dps)d(ln(t)) explicit deconvolution methodDps material balance deconvolutiond(Dps)d(ln(t)) material balance deconvolutionDps no wellbore storaged(Dps)d(ln(t)) no wellbore storage
Dpwf
and
its lo
garit
hmic
der
ivat
ive (
psi)
Figure 5 Deconvolution for a drawdown test in a dual porosityreservoir
data were obtained from the cylindrical source solution withpseudosteady-state interporosity flow (test design) It is notpossible to identify the dual porosity behavior characteristicof the reservoir from wellbore pressure data because thisbehavior has been masked completely by wellbore storage asindicated in Figures 5 and 6
We deconvolved the pressure and unloading rate andpresent the results in Figures 5 and 6 The deconvolvedresults are reasonably consistent with the analytical solutionAfter deconvolution the dual porosity behavior can beidentified easily from the valley in the logarithmic derivativeof pressure If wellbore storage effects were not removed atearly times it easily could have incorrectly identified thereservoir system as infinite actingThe estimation of reservoirparameters can be obtained now from deconvolved pressure
The third case is an oil well drawdown test in a reservoirwith a sealing fault 250 ft from the well Synthetic pressureresponses were generated by simulation (test design) with aconstant wellbore storage coefficient of 01 bblpsi As we cansee in Figures 7 and 8 the early time infinite acting reservoirbehavior has been masked completely by wellbore storage
To recover the hidden reservoir behavior we performeddeconvolution on the pressure change to remove the wellborestorage effect As in Figures 7 and 8 the deconvolved pressurechange and logarithmic derivative data are almost identicalto the responses without wellbore storageThe deconvolutionresults clearly indicate parallel horizontal lines for the loga-rithmic derivative However without wellbore storage effectselimination this maymislead to identify the reservoir systemas infinite acting So this allows us to correctly identify thereservoir system having a sealing fault
The fourth case is a field example that is taken from theliterature [23] The basic reservoir and fluid properties are
Abstract and Applied Analysis 7
Table 1 Reservoir rock and fluid properties for the first field case
119905119901(hr) ℎ (ft) 119903
119908(ft) 119902
119900(STBD) 119861
119900(RBSTB) 119888
119905(psiminus1) 120583
119900(cp) 120601
1533 107 029 174 106 42 times 10minus6 25 025
Table 2 Reservoir rock and fluid properties for the second field case
119905119901(hr) ℎ (ft) 119903
119908(ft) 119879 (∘F) 119901 (psi) 119902
119900(STBD) 119902
119892(MscfD) 119888
119891(psiminus1) 119879
119889119901(∘F) 119901
119889119901(psi) ∘API 120574
119892
143111 100 0354 2795 7322 104 13925 3 times 10minus6 212 5445 45375 068
10minus3 10minus2 10minus1 100 101 102 103
Time (hr)
103
102
101
100
10minus1
Dpwf wellbore storaged(Dpwf)d(ln(t)) wellbore storageDps explicit deconvolution methodd(Dps)d(ln(t)) explicit deconvolution methodDps material balance deconvolutiond(Dps)d(ln(t)) material balance deconvolutionDps no wellbore storaged(Dps)d(ln(t)) no wellbore storage
Dpwf
and
its lo
garit
hmic
der
ivat
ive (
psi)
Figure 6 Deconvolution for a drawdown test in a dual porosityreservoir
shown in Table 1 In this case we provide the deconvolvedpressure data using the explicit deconvolution formula andthe material balance deconvolution method that is presentedin this work The data are taken from a pressure buildup testand can be reasonably used as field data The deconvolutionresults are shown in Figure 9 in semilog plot As it isillustrated from Figure 9 the corrected pressure provides aclear straight line starting as early as 15 hours in contrast tothe vague line from the uncorrected pressure response
And the last (field) example is a gas condensatewell buildup test following a variable rate productionhistory The duration of last pressure buildup test is361556 hours (from 1889 to 2250556 hours) The basicreservoir and fluid properties for the last drawdown testare shown in Table 2 However the properties of otherproduction tests are not shown here To use our newtechniques effectively for gas wells we use normalizedpseudopressure and pseudotime functions as defined byMeunier et al [24]The final results of explicit deconvolutionformula and material balance deconvolution methodshow two different slopes (Figure 10) where the earlytimes slope is greater than the last one This identifiesthe radial composite behavior of the reservoir With
the corrected pressures the reservoir parameters can also beestimated
6 Conclusions
In this work we have provided some explicit deconvolutionmethods for restoration of constant rate pressure responsesfrom measured pressure data dominated by wellbore storageeffects without sandface rate measurements Based on ourinvestigations we can state the following specific conclusionsrelated to these techniques
(1) The results obtained from the synthetic and fieldexamples show the practical and computational effi-ciency of our deconvolution methods
(2) Modification of 120573-deconvolution is presented wherethe beta quantity is variable rather than a constant
(3) For situations in which there are no downhole ratemeasurements the explicit deconvolution methods(blind deconvolution approaches) can be used toderive the downhole rate function from the pressuremeasurements and restore the reservoir pressureresponse
(4) We can unveil the early time behavior of a reservoirsystemmasked by wellbore storage effects using thesenew approaches Wellbore geometries and reservoirtypes can be identified from this restored pressuredata Furthermore the conventional methods canbe used to analyze this restored formation pressureto estimate formation parameters So the presentedtechniques can be used as the practical tools toimprove pressure transient test interpretation
(5) Deconvolution in well test analysis is often expressedas an inverse problem This approach can be used asan idea to justify the rate normalization method andthe material balance deconvolution method
Appendices
A Russel Method and Its Modification
A method for correcting pressure buildup pressure data freeof wellbore storage effects at early times has been developedby Russell [14] In this method it is not necessary to measurethe rate of influx after closing in at the surface Instead oneuses a theoretical equation which gives the form that thebottom hole pressure should have fluid accumulates in the
8 Abstract and Applied Analysis
10minus3 10minus2 10minus1 100 101 102 103
Time (hr)
250
200
150
100
50
0
Dpwf wellbore storaged(Dpwf)d(ln(t)) wellbore storageDps explicit deconvolution methodd(Dps)d(ln(t)) explicit deconvolution methodDps material balance deconvolutiond(Dps)d(ln(t)) material balance deconvolutionDps no wellbore storaged(Dps)d(ln(t)) no wellbore storage
Dpwf
and
its lo
garit
hmic
der
ivat
ive (
psi)
Figure 7 Deconvolution for a drawdown test in a reservoir with asealing fault
10minus3 10minus2 10minus1 100 101 102 103
Time (hr)
103
102
101
100
Dpwf wellbore storaged(Dpwf)d(ln(t)) wellbore storageDps explicit deconvolution methodd(Dps)d(ln(t)) explicit deconvolution methodDps material balance deconvolutiond(Dps)d(ln(t)) material balance deconvolutionDps no wellbore storaged(Dps)d(ln(t)) no wellbore storage
Dpwf
and
its lo
garit
hmic
der
ivat
ive (
psi)
Figure 8 Deconvolution for a drawdown test in a reservoir with asealing fault
wellbore during buildup This leads to the result that oneshould plot
[119901119908119904 (Δ119905) minus 119901
119908119891 (Δ119905 = 0)]
[1 minus (11198622Δ119905)]
(A1)
versus log (Δ119905) in analyzing pressure buildup data during theearly fill-up period The denominator makes a correction for
104103102101100
Horner time ratio
800
700
600
500
400
300
200
100
0
Dpws
(psi)
Dpws wellbore storageDps explicit deconvolution methodDps material balance deconvolution
Figure 9 Restored pressures for a buildup test in a homogeneousreservoir
0
500
1000
1500
2000
2500
3000
3500
4000
105104103102101100
Horner time ratio
Dpa
(psi)
Dpaw wellbore storage explicit deconvolution method material balance deconvolution
DpasDpas
Figure 10 Restored pressures for a buildup test in a radial compositereservoir
the gradually decreasing flow into the wellbore The quantity1198622is obtained by trial and error as the value which makes the
curve straight at early times After obtaining the straight linesection the rest of the analysis is the same as for any otherpressure buildup This method has the advantage or requiresno additional data over that taken routinely during a shut intest
Due to the time consuming and unconfident of trial anderror to determine 119862
2in Russell method we modify this
method Russellrsquos wellbore storage correction is given as
[119901119908119904 (Δ119905) minus 119901
119908119891 (Δ119905 = 0)]
[1 minus (11198622Δ119905)]
= 119891 (Δ119905 = 1 hr) + 119898119904119897log (Δ119905)
(A2)
Abstract and Applied Analysis 9
Defining the following function and using least squarestechnique obtain the final result
119864 (1198622 119891119898)
=
119873
sum
119894=1
(Δ119901119908119904119894
(1 minus (11198622Δ119905119894))
minus 119891(1198622) minus 119898(119862
2) log(Δ119905
119894))
2
(A3)
Tominimize the function119864 it can differentiate119864with respectto 1198622 119891 and 119898 and setting the results equal to zero After
differentiation it reduces to
120597119864
1205971198622
= minus2
119873
sum
119894=1
(Δ119901119908119904119894
Δ119905119894(1198622minus (1Δ119905
119894))2)
times (Δ119901119908119904119894
(1 minus (11198622Δ119905119894))
minus 119891 (1198622)
minus119898 (1198622) log (Δ119905
119894) ) = 0
(A4)
120597119864
120597119891
= minus2
119873
sum
119894=1
(Δ119901119908119904119894
(1 minus (11198622Δ119905119894))
minus 119891 (1198622) minus 119898 (119862
2) log (Δ119905
119894))
= 0
(A5)120597119864
120597119898
= minus2
119873
sum
119894=1
log (Δ119905119894) (
Δ119901119908119904119894
(1 minus (11198622Δ119905119894))
minus 119891 (1198622)
minus 119898 (1198622) log (Δ119905
119894)) = 0
(A6)
From (A4) we define the following function
119883(1198622)
=
119873
sum
119894=1
(Δ119901119908119904119894
Δ119905119894(1198622minus (1Δ119905
119894))2)
times(Δ119901119908119904119894
(1minus (11198622Δ119905119894))minus119891 (119862
2) minus119898 (119862
2) log (Δ119905
119894))
= 0
(A7)
The roots of (A7) can be found via the following algorithm
1198622(119894+1)
= 1198622(119894)
minus119883 (1198622)
1198831015840 (1198622) (A8)
where
119883(1198622) =
119873
sum
119894=1
(Δ119901119908119904119894
Δ119905119894(1198622minus (1Δ119905
119894))2)
times (Δ119901119908119904119894
(1 minus (11198622Δ119905119894))
minus 119891 (1198622)
minus119898 (1198622) log (Δ119905
119894))
1198831015840(1198622) =
119873
sum
119894=1
(minus2Δ119901119908119904119894
Δ119905119894(1198622minus (1Δ119905
119894))3)
times (Δ119901119908119904119894
Δ119905119894(1198622minus (1Δ119905
119894))2
times [Δ119901119908119904119894
(1 minus (11198622Δ119905))
minus 119891 (1198622)
minus119898 (1198622) log (Δ119905
119894)])
+ (Δ119901119908119904119894
Δ119905119894(1198622minus (1Δ119905
119894))2)
times (minusΔ119901119908119904119894
Δ119905119894(1198622minus (1Δ119905
119894))3minus 1198911015840(1198622)
minus1198981015840(1198622) log (Δ119905
119894))
(A9)
where the unknown functions are given by the followingrelations
119891 (1198622) =
120573119906 (1198622) minus 120572V (119862
2)
1205732 minus 119873120572(A10)
whose derivative is
1198911015840(1198622) =
1205731199061015840(1198622) minus 120572V1015840 (119862
2)
1205732 minus 119873120572
119898 (1198622) =
119873119906 (1198622) minus 120573V (119862
2)
119873120572 minus 1205732
(A11)
where its derivative is
1198981015840(1198622) =
1198731199061015840(1198622) minus 120573V1015840 (119862
2)
119873120572 minus 1205732
119906 (1198622) =
119873
sum
119894=1
log (Δ119905119894)
Δ119901119908119904119894
(1 minus (11198622Δ119905119894))
(A12)
10 Abstract and Applied Analysis
and after differentiation it yields
1199061015840(1198622) = minus
119873
sum
119894=1
log (Δ119905
119894)
Δ119905119894
Δ119901119908119904119894
(1198622minus (1Δ119905
119894))2
V (1198622) =
119873
sum
119894=1
Δ119901119908119904119894
(1 minus (11198622Δ119905119894))
(A13)
which has the following derivative
V1015840 (1198622) = minus
119873
sum
119894=1
1
Δ119905119894
Δ119901119908119904119894
(1198622minus (1Δ119905
119894))2 (A14)
finally the values of 120572 and 120573 parameters are given by thefollowing formulas
120572 =
119873
sum
119894=1
(log(Δ119905119894))2
120573 =
119873
sum
119894=1
log (Δ119905119894)
(A15)
B General Formula for Inverse Problem
The convolution of (6) yields
119901119904119863
(119905119863) = int
119905119863
0
119870 (120591) 119901119908119863 (119905119863 minus 120591) 119889120591
= int
119905119863
0
119870(119905119863minus 120591) 119901
119908119863 (120591) 119889120591
(B1)
where
119870(119905119863) = 119871minus1(
1
119904119902 (119904)) (B2)
where 119871minus1 is the inverse Lapalce transform operator 119870(119905119863)
can be computed either from the Laplace transforms of119902119863(119905119863) data a curve-fitted equation of 119902
119863(119905119863) data or using
another technique with a relation between119870(119905119863) and 119902
119863(119905119863)
However it would be time consuming to invert all the 119902119863(119905119863)
data in Laplace space and transform it back to real space inaccordance with (B2) Once an approximation is obtained itwill be easy to compute119870(119905
119863) and integrate (B1) to determine
119901119904119863(119905119863) However finding some approximation functions for
119902119863(119905119863) data may be difficult We provide an approach to
overcome this problemThis idea is based on finding anotherrelation between 119870(119905
119863) and 119902
119863(119905119863) that is
ℓ (1) =1
119904=
1
119904119902119863(119904)
119902119863(119904) = ℓ (int
119905119863
0
119870 (120591) 119902119863 (119905119863 minus 120591) 119889120591)
= ℓ(int
119905119863
0
119870(119905119863minus 120591) 119902
119863 (120591) 119889120591)
(B3)or
1 = int
119905119863
0
119870 (120591) 119902119863 (119905119863 minus 120591) 119889120591 = int
119905119863
0
119870(119905119863minus 120591) 119902
119863 (120591) 119889120591
(B4)
The other relations that can be used are
ℓ (119905119863) =
1
1199042=
1
1199042119902119863(119904)
119902119863(119904) = ℓ (int
119905119863
0
119871 (120591) 119902119863 (119905119863 minus 120591) 119889120591)
= ℓ(int
119905119863
0
119871 (119905119863minus 120591) 119902
119863 (120591) 119889120591)
(B5)
or
119905119863= int
119905119863
0
119871 (120591) 119902119863(119905119863minus 120591) 119889120591 = int
119905119863
0
119871 (119905119863minus 120591) 119902
119863(120591) 119889120591
(B6)
ℓ (1) =1
119904=
1
1199042119902119863(119904)
119904119902119863(119904) = ℓ (int
119905119863
0
119871 (120591) 1199021015840
119863(119905119863minus 120591) 119889120591)
= ℓ(int
119905119863
0
119871 (119905119863minus 120591) 119902
1015840
119863(120591) 119889120591)
(B7)
or
1 = int
119905119863
0
119871 (120591) 1199021015840
119863(119905119863minus 120591) 119889120591 = int
119905119863
0
119871 (119905119863minus 120591) 119902
1015840
119863(120591) 119889120591
(B8)
where
119871 (119905119863) = ℓminus1(
1
1199042119902 (119904)) (B9)
Equation (B4) can be used simultaneously with (B1) toanalyze the deconvolution problem
Equation (11) can be written as
119901119904119863
(119904) = 119904119901119908119863
(119904)1
1199042119902119863(119904)
(B10)
or
119901119904119863
(119905119863) = int
119905119863
0
119871 (120591) 1199011015840
119908119863(119905119863minus 120591) 119889120591
= int
119905119863
0
119871 (119905119863minus 120591) 119901
1015840
119908119863(120591) 119889120591
(B11)
The applications of (B6) and (B8) are with (B11)For justifying the rate normalizationmethod andmaterial
balance deconvolution it can use (B1) (B4) ((B11) can alsobe used with (B6) or (B8) for this goal) and generalizedmean value theorem for integrals This theorem is expressedas follows
Theorem 1 Let 119891 and 119892 be continuous on [119886 119887] and supposethat 119892 does not change sign on [119886 119887] Then there is a point 119888 in[119886 119887] such that
int
119887
119886
119891 (119905) 119892 (119905) 119889119905 = 119891 (119888) int
119887
119886
119892 (119905) 119889119905 (B12)
Abstract and Applied Analysis 11
By Theorem 1 there are some values 1198881and 1198882between 0
and 119905119863such that
119901119904119863
(119905119863) = 119901119908119863
(1198881) int
119905119863
0
119870(119905119863minus 120591) 119889120591 (B13)
1 = 119902119863(1198882) int
119905119863
0
119870(119905119863minus 120591) 119889120591 (B14)
By setting 1198881
= 1198882
= 119905119863and dividing the right and left
side of (B13) respectively the rate normalization techniqueis developed However the values 119888
1and 1198882are not equal
to each other and are not equal to 119905119863
in general Thisis the reason of error in rate normalization and materialbalance deconvolution methods 119888
1and 1198882are less than 119905
119863in
these methods for the duration of wellbore storage distortionduring a pressure transient test Since these conditions areapproximately satisfied in material balance deconvolutionthe results of material balance are more accurate and moreconfident than rate normalization
Nomenclature
API API gravity ∘API119861119900 Oil formation volume factor RBSTB
119888119905 Total isothermal compressibility factor
psiminus1
1198622 Arbitrary constant hr
minus1
119862119863 Wellbore storage constant dimensionless
119862119904 Wellbore storage coefficient bblpsi
ℎ Formation thickness ftHTR Horner time ratio dimensionless119896 Formation permeability md119870 Kernel of the convolution integral119871 Kernel of the convolution integral119873119901 Cumulative oil production bbl
119901 Pressure psi119901119886 Normalized pseudopressure psi
119901 Average reservoir pressure psi119901119863 Formation pressure dimensionless
119901119889119901 Dewpoint pressure psi
119901119894 Initial reservoir pressure psi
119901119904119863 Formation pressure including skin
dimensionless119901119904 Formation pressure including skin psi
119901119908119863
Pressure draw down dimensionless119901119908119891 Flowing bottom hole pressure psi
119901119908119904 Shut-in bottom hole pressure psi
119902119863 Sandface flow rate dimensionless
119902119892 Gas flow rate MscfD
119902119900 Oil flow rate STBD
119903119908 Wellbore radius ft
119904 Laplace variable119878 Skin factor119905 Time hr119905119863 Time dimensionless
119905119901 Production time hr
119879 Reservoir temperature ∘F
119879119889119901 Dewpoint temperature ∘F
120572 Explicit deconvolution variabledimensionless
120573 Beta deconvolution variabledimensionless
120574119892 Gas gravity (air = 10)
120582 Interporosity flow coefficient of dualporosity reservoir
120601 Porosity of reservoir rock dimensionless120583119892 Gas viscosity cp
120583119900 Oil viscosity cp
120596 Storativity ratio of dual porosity reservoirminus Laplace transform of a function1015840 Derivative of a function
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
References
[1] Y Cheng W J Lee and D A McVay ldquoFast-Fourier-transform-based deconvolution for interpretation of pressure-transient-test data dominated by Wellbore storagerdquo SPE Reservoir Eval-uation and Engineering vol 8 no 3 pp 224ndash239 2005
[2] M M Levitan and M R Wilson ldquoDeconvolution of pressureand rate data from gas reservoirs with significant pressuredepletionrdquo SPE Journal vol 17 no 3 pp 727ndash741 2012
[3] Y Liu and R N Horne ldquoInterpreting pressure and flow-ratedata from permanent downhole gauges by use of data-miningapproachesrdquo SPE Journal vol 18 no 1 pp 69ndash82 2013
[4] O Ogunrewo and A C Gringarten ldquoDeconvolution of welltest data in lean and rich gas condensate and volatile oil wellsbelow saturation pressurerdquo in Proceedings of the SPE AnnualTechnical Conference and Exhibition Document ID 166340-MS New Orleans La USA September-October 2013
[5] R C Earlougher Jr Advances in Well Test Analysis vol 5 ofMonogragh Series SPE Richardson Tex USA 1977
[6] F Kucuk and L Ayestaran ldquoAnalysis of simultaneously mea-sured pressure and sandface flow rate in transient well testingrdquoJournal of PetroleumTechnology vol 37 no 2 pp 323ndash334 1985
[7] M Onur M Cinar D IIk P P Valko T A Blasingame and PHegeman ldquoAn investigation of recent deconvolution methodsfor well-test data analysisrdquo SPE Journal vol 13 no 2 pp 226ndash247 2008
[8] F J KuchukMOnur andFHollaender ldquoPressure transient testdesign and interpretationrdquo Developments in Petroleum Sciencevol 57 pp 303ndash359 2010
[9] A C Gringarten D P Bourdet P A Landel and V J KniazeffldquoA comparison between different skin and wellbore storagetype-curves for early-time transient analysisrdquo in Proceedings ofthe SPE Annual Technical Conference and Exhibition Paper SPE8205 Las Vegas Nev USA 1979
[10] A C Gringarten ldquoFrom straight lines to deconvolution theevolution of the state of the art in well test analysisrdquo SPEReservoir Evaluation and Engineering vol 11 no 1 pp 41ndash622008
[11] J A Cumming D A Wooff T Whittle and A C GringartenldquoMultiple well deconvolutionrdquo in Proceedings of the SPE Annual
12 Abstract and Applied Analysis
Technical Conference and Exhibition Document ID 166458-MS New Orleans La USA September-October 2013
[12] O Bahabanian D Ilk N Hosseinpour-Zonoozi and T ABlasingame ldquoExplicit deconvolution of well test data distortedby wellbore storagerdquo in Proceedings of the SPE Annual TechnicalConference and Exhibition Paper SPE 103216 pp 4444ndash4453San Antonio Tex USA September 2006
[13] J Jones and A Chen ldquoSuccessful applications of pressure-ratedeconvolution in theCad-Nik tight gas formations of the BritishColumbia foothillsrdquo Journal of Canadian Petroleum Technologyvol 51 no 3 pp 176ndash192 2012
[14] D G Russell ldquoExtension of pressure buildup analysis methodsrdquoJournal of Petroleum Technology vol 18 no 12 pp 1624ndash16361966
[15] R E Gladfelter G W Tracy and L E Wilsey ldquoSelecting wellswhich will respond to production-stimulation treatmentrdquo inDrilling and Production Practice API Dallas Tex USA 1955
[16] M J Fetkovich and M E Vienot ldquoRate normalization ofbuildup pressure by using afterflow datardquo Journal of PetroleumTechnology vol 36 no 13 pp 2211ndash2224 1984
[17] J L Johnston Variable rate analysis of transient well testdata using semi-analytical methods [MS thesis] Texas AampMUniversity College Station Tex USA 1992
[18] A F van Everdingen ldquoThe skin effect and its influence on theproductive capacity of a wellrdquo Journal of Petroleum Technologyvol 5 no 6 pp 171ndash176 1953
[19] W Hurst ldquoEstablishment of the skin effect and its impedimentto fluid flow into a well borerdquo The Petroleum Engineer pp B6ndashB16 1953
[20] J A Joseph and L F Koederitz ldquoA simple nonlinear model forrepresentation of field transient responsesrdquo Paper SPE 11435SPE Richardson Tex USA 1982
[21] F J Kuchuk ldquoGladfelter deconvolutionrdquo SPE Formation Evalu-ation vol 5 no 3 pp 285ndash292 1990
[22] A F van Everdingen and W Hurst ldquoThe application of thelaplace transformation to flow problems in reservoirsrdquo Journalof Petroleum Technology vol 1 no 12 pp 305ndash324 1949
[23] D Bourdet T M Whittle A A Douglas and Y M Pirard ldquoAnew set of type curves simplifies well test analysisrdquo World Oilpp 95ndash106 1983
[24] D F Meunier C S Kabir and M J Wittmann ldquoGas well testanalysis use of normalized pseudovariablesrdquo SPE FormationEvaluation vol 2 no 4 pp 629ndash636 1987
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
2 Abstract and Applied Analysis
its governing parameters frommeasured data in the wellboreand at the wellhead This problem is known as the inverseproblem The solution of the inverse problem usually is notunique Gringarten et al [9] pointed out that if the numberand the rage measurements increased the nonuniqueness ofthe inverse problem will be reduced Thus combining sand-face flow rate with pressure measurements will enhance theconventional (including type curve) well test interpretationmethods On the other hand mathematically deconvolutionis also a highly unstable inverse problem because smallerrors in the data can result in large uncertainties in thedeconvolution solution [10 11]
Unfortunately standard deconvolution techniques re-quire accurate measurements of flow rate and pressure atdownhole (or sandface) conditions While accurate pressuremeasurements are commonplace the measurement of sand-face flow rates is rare essentially nonexistent in practiceAs such the deconvolution of wellbore storage distortedpressure test data is problematic In theory this process ispossible but in practice without accurate measurements offlow rates this process cannot be employed [12 13]
In the past 50 years a variety of deconvolution tech-niques have been proposed in petroleum engineering Russell[14] used the approximate direct method to correct thepressure transient data distorted by wellbore storage intothe equivalent pressure function for the constant rate caseGladfelter et al [15] and Fetkovich and Vienot [16] used therate normalization techniques to correct for thewellbore stor-age effectsThese rate normalizationmethods were successfulin some cases So Johnston [17] provided thematerial balancedeconvolution that is a practical approach for the analysis ofpressure transient data distorted by wellbore storage effects
Essentially rate normalization techniques and materialbalance deconvolution are restricted when the lack of ratemeasurement exists van Everdingen [18] and Hurst [19]demonstrated empirically that the sandface flow rate profilecan be modeled approximately using an exponential relationfor the duration of wellbore storage distortion during a pres-sure transient testThe vanEverdingen andHurst exponentialrate model is given in dimensionless form as
119902119863(119905119863) = 1 minus 119890
minus120573119905119863 (1)
Further they showed that the rate-time relationship duringafterflow (for a pressure buildup test) or unloading (in apressure drawdown test) is a function of the pressure dropchange with respect to time and a relatively constant wellborestorage coefficient
Based on a material balance in the wellbore the sandfaceflow rate can be calculated by the following relation given indimensionless form
119902119863(119905119863) = 1 minus 119862
119863
119889119901119908119863
119889119905119863
(2)
where it can be noted that in the development of wellborestorage models solutions (eg type curves) a constant well-bore storage coefficient is always assumed
Equations (1) and (2) laid the groundwork for 120573-deconvolution Joseph and Koederitz [20] and Kuchuk [21]
applied 120573-deconvolution for the analysis of wellbore stor-age distorted pressure transient data The 120573-deconvolutionformula which computes the undistorted pressure dropfunction directly from the wellbore storage affected data isgiven as
119901119904119863
(119905119863) =
1
120573
119889119901119908119863
(119905119863)
119889119905119863
+ 119901119908119863
(119905119863) (3)
It should be noted that (3) is only validwhen the sandface flowrate profile follows an exponential trend as prescribed by (1)[12] In thisworkwe develop an approach that overcomes thisrestriction by treating the 120573 quantity as a variable in generalrather than a constant
The obvious advantage of 120573-deconvolution is that thewellbore storage effects are eliminated and the restoredpressure is obtained using only the given pressure data
This paper reports a modification of Russell method(Appendix A) a modification of 120573-deconvolution and anexplicit deconvolution formula to restore the pressure andrate data and a review on material balance deconvolutionis also presented to deconvolve the pressure and rate dataafter the explicit deconvolution method Then a generalapproach to analyze the deconvolution problem is provided(Appendix B) In verification examples section we show thepower and practical applicability of our approaches withvarious synthetic and field case examples for oil and gas wells
2 Main Results
21 120573-Deconvolution and Its Modification van EverdingenandHurst [22] presented the dimensionless wellbore pressurefor a continuously varying flow rate as
119901119908119863
(119905119863) = 119902119863 (0) 119901119904119863 (119905119863) + int
119905119863
0
1199011015840
119863(120591) 119901119904119863 (119905119863 minus 120591) 119889120591
(4)
An alternative form to (4) can be obtained by integration byparts as
119901119908119863
(119905119863) = 119902119863(119905119863) 119901119904119863 (0) + int
119905119863
0
119902119863 (120591) 119901
1015840
119904119863(119905119863minus 120591) 119889120591
(5)
where 119901119908119863
(119905119863) = (119896ℎ1412119902119861120583)(119901
119894minus 119901119908119891(119905)) 119905
119863=
(000026371198961199051206011205831198881199051199032
119908) 119901119904119863
= 119901119863+ 119878 119901
119908119863(119905119863) is dimen-
sionless wellbore pressure with wellbore storage and skineffects 119901
119863(119905119863) is dimensionless sandface pressure for the
constant rate case without wellbore storage and skin effects 119878is steady state skin factor 1199011015840
119904119863(119905119863) = 119889119901
119904119863(119905119863)119889119905119863 119902119863(119905119863) =
119902119904119891(119905119863)119902119903 1199021015840119863(119905119863) = 119889119902
119863(119905119863)119889119905119863 and 119902
119903is reference
flow rate if the stabilized constant rate is available then 119902119903
should be replaced by 119902119877 119902119904119891(119905) is variable sandface flow rate
(flowmeter reading) and 119902119863(119905119863) is dimensionless sandface
rateIt should be emphasized that (4) and (5) can be applied
for many reservoir engineering problems The linearity ofdiffusivity equation allows us to use (4) and (5) for fractured
Abstract and Applied Analysis 3
layered anisotropic and heterogeneous systems as long as thefluid in the reservoir is single phaseThese equations can alsobe applied to both drawdown and buildup tests if the initialconditions are known For a reservoir with an initial constantand uniform pressure distribution (119901
119863(0) = 0) (4) and (5)
can be expressed as
119901119908119863
(119905119863) = int
119905119863
0
1199021015840
119863(120591) 119901119904119863 (119905119863 minus 120591) 119889120591 (6)
119901119908119863
(119905119863) = 119878119902
119863(119905119863) + int
119905119863
0
119902119863 (120591) 119901
1015840
119904119863(119905119863minus 120591) 119889120591 (7)
Furthermore (6) and (7) also can be expressed as
119901119908119863
(119905119863) = int
119905119863
0
1199021015840
119863(119905119863minus 120591) 119901
119904119863 (120591) 119889120591 (8)
119901119908119863
(119905119863) = 119878119902
119863(119905119863) + int
119905119863
0
119902119863(119905119863minus 120591) 119901
1015840
119904119863(120591) 119889120591 (9)
In (6) and (8) it is assumed that 1199021015840119863(119905119863) exists If 119902
119863(119905119863) is
constant then (7) and (9) must be used Equations (6) and(8) are known as a Volterra integral equation of first kind andthe convolution type
Taking the Laplace transform of (6) yields
119901119908119863
(119904) = 119904119902119863(119904) 119901119904119863
(119904) (10)
Rearranging (10) for the equivalent constant rate pressuredrop function 119901
119904119863(119904) we obtain
119901119904119863
(119904) =119901119908119863
(119904)
119904119902119863(119904)
(11)
By taking the Laplace transform of (1) one can easily obtain
119902119863(119904) =
1
119904minus
1
119904 + 120573 (12)
Substituting (12) into (11) and then taking the inverse Laplacetransform of (11) yield the beta deconvolution formula
119901119904119863
(119905119863) =
1
1205731199011015840
119908119863(119905119863) + 119901119908119863
(119905119863) (13)
where
1199011015840
119908119863(119905119863) =
119889119901119908119863
119889119905119863
(14)
where we note that (13) is specifically valid only for theexponential sandface flow rate profile given by (1) This maypresent a serious limitation in terms of practical applicationof the 120573-deconvolutionmethod To overcome this restrictionwe propose an idea It would be assumed that120573 is not constantin general If 119902
119863(119905119863)data are given the procedure is as follows
119902119863(119905119863) = 1 minus 119890
minus120573(119905119863)119905119863 (15)
Rearranging (15) yields
1 minus 119902119863(119905119863) = 119890minus120573(119905119863)119905119863 (16)
then
minus120573 (119905119863) 119905119863= ln (1 minus 119902
119863(119905119863)) (17)
so
120573 (119905119863) = minus
ln (1 minus 119902119863(119905119863))
119905119863
(18)
Substituting of (18) into (13) instead of 120573 term yields themodified 120573-deconvolution
119901119904119863
(119905119863) =
1
120573 (119905119863)1199011015840
119908119863(119905119863) + 119901119908119863
(119905119863) (19)
Plot of (19) versus (int119905119863
0119902119863(120591)119889120591119902
119863(119905119863)) and (int
119905119863
0(1 minus
119902119863(120591))1198891205911 minus 119902
119863(119905119863)) for drawdown and buildup tests
respectively should be considered more accurate to use as apractical tool for field applications As mentioned before thevariable 120573 quantity (rather than a constant) shows the generaland more accurate behavior of 119902
119863(119905119863) So accuracy of (19) is
also more than (13) It can be emphasized that (13) and (19)are applicable in the wellbore storage duration and for thelater times the given pressure data are correct However theresults of (19) show that this equation is not bad for the latertimes
22 Explicit Deconvolution Formula A few types of approx-imation functions can be used to approximate 119902
119863(119905119863) An
approximate proof is given here to show that the exponentialfunction gives a good representation of 119902
119863(119905119863) By setting
119902119863(119905119863) = 1 minus 119891 (119905
119863) (20)
119891 (0) = 1 (21)
Equation (21) is an initial condition for (20)Taking the time derivative of (20) gives
1199021015840
119863(119905119863) = minus119891
1015840(119905119863) (22)
Taking the time derivative of (2) gives
1199021015840
119863(119905119863) = minus119862
11986311990110158401015840
119908119863 (23)
From (2) and (20) we have
119891 (119905119863) = 1198621198631199011015840
119908119863 (24)
And from (22) and (23) we have
1198911015840(119905119863) = 11986211986311990110158401015840
119908119863 (25)
After dividing right and left hand sides of (25) by (24) theresult is
1198911015840(119905119863)
119891 (119905119863)=11990110158401015840
119908119863(119905119863)
1199011015840119908119863
(119905119863) (26)
Integrating both sides of (26) with respect to 119905119863from 0 to 119905
119863
yields
int
119905119863
0
1198911015840(120591)
119891 (120591)119889120591 = int
119905119863
0
11990110158401015840
119908119863(120591)
1199011015840119908119863
(120591)119889120591 (27)
4 Abstract and Applied Analysis
Using elementary calculus yields
ln (1003816100381610038161003816119891(120591)
1003816100381610038161003816)1003816100381610038161003816119905119863
0= ln (1003816100381610038161003816119891 (119905
119863)1003816100381610038161003816) minus ln (1) = int
119905119863
0
11990110158401015840
119908119863(120591)
1199011015840119908119863
(120591)119889120591
(28)
So
119891 (119905119863) = 119890120572(119905119863) (29)
where
120572 (119905119863) = int
119905119863
0
11990110158401015840
119908119863(120591)
1199011015840119908119863
(120591)119889120591 (30)
Now the explicit deconvolution of data based on (20) can bewritten as follows
119901119904119863
(119905119863) = minus
1
120572 (119905119863) 119905119863
1199011015840
119908119863+ 119901119908119863
(119905119863) (31)
or
119901119904119863
(119905119863) = minus
119905119863
120572 (119905119863)1199011015840
119908119863+ 119901119908119863
(119905119863) (32)
that is similar to (19) The other similarity between (19) and(32) is that the plot of (32) versus (int119905119863
0119902119863(120591)119889120591119902
119863(119905119863)) and
(int119905119863
0(1 minus 119902
119863(120591))119889120591(1 minus 119902
119863(119905119863))) for drawdown and buildup
tests respectively yields the more accurate results and isusable for practical applications The applicability of (32) isalso for duration of wellbore storage effects And for the othertimes the given pressure data are correct However using thismethod for the later times yields not bad results
3 Material Balance Deconvolution
Material balance deconvolution is an extension of the ratenormalization method Johnston [17] defines a new 119909-axisplotting function (material balance time) that provides anapproximate deconvolution of the variable rate pressuretransient problem There are numerous assumptions associ-ated with the material balance deconvolution methods oneof the most widely accepted assumptions is that the rateprofilemust change smoothly andmonotonically In practicalterms this condition should be met for the wellbore storageproblem
The general form of material balance deconvolution isprovided for the pressure drawdown case in terms of thematerial balance time function and the rate-normalizedpressure drop function The material balance time functionis given as
119905119898119887
=119873119901
119902119863 (119905)
(33)
where
119873119901= int
119905
0
119902119863 (120591) 119889120591 (34)
The rate-normalized pressure drop function is given by
Δ119901119908119891 (119905)
119902119863 (119905)
=119901119894minus 119901119908119891 (119905)
119902119863 (119905)
(35)
The material balance time function for the pressure buildupcase is given as
119905119898119887
=119873119901
1 minus 119902119863 (Δ119905)
(36)
where
119873119901= int
Δ119905
0
(1 minus 119902119863 (120591)) 119889120591 (37)
The rate-normalized pressure drop function for the pressurebuildup case is given as
Δ119901119908119904 (Δ119905)
1 minus 119902119863 (Δ119905)
=119901119908119904 (Δ119905) minus 119901
119908119891 (Δ119905 = 0)
1 minus 119902119863 (Δ119905)
(38)
4 Methodology
In this work several methods are provided to restore thepressure datawithout any sandface flow rate informationThevalues of sandface flow rate data for all methods that requirethese data may be obtained from (20) We evaluated a veryold correctionmethod by Russell [14] However we modifiedthis method by least square method but it was found thatthis method is to be unacceptable for all applications Thederivation of this method and its modification is given inAppendix A
The 120573-deconvolution and its modification are anothertechnique that is mentioned in this study The formulationof this method is presented in Section 21 The other explicitdeconvolutionmethod is an approach similar to themodified120573-deconvolution that uses only the pressure data for decon-volution of pressure and rate data The formulation of thismethod was derived in Section 22 The 120573-deconvolution itsmodification and explicit deconvolution formula can be usedfor practical applications The applicability of these methodsis for duration of wellbore storage effects However usingthese approaches for the other times yields not bad results
The other technique is material balance deconvolutionmethod [17] that has sufficiently accurate resultsThismethodcan also be used as a good approach for practical applicationsThe formulas for this method were given in Section 3
In Appendix B deconvolution is expressed as an inverseproblem that is analyzed by a tricky idea Based on thisidea an interesting proof is given for rate normalizationmethod and material balance deconvolution that is a goodapproach to find the errors and limitations of these methodsand overcome them
Abstract and Applied Analysis 5
10minus3 10minus2 10minus1 100 101 102 1030
500
1000
1500
2000
2500
3000
3500
4000
Time (hr)
Dpwf wellbore storaged(Dpwf)d(ln(t)) wellbore storageDps explicit deconvolution methodd(Dps)d(ln(t)) explicit deconvolution methodDps material balance deconvolution methodd(Dps)d(ln(t)) material balance deconvolution methodDps no wellbore storaged(Dps)d(ln(t)) no wellbore storage
Dpwf
and
its lo
garit
hmic
der
ivat
ive (
psi)
Figure 1 Deconvolution for a drawdown test in a homogeneousreservoir
From the above-mentioned methods we use the fol-lowing two techniques to show the verification of theseapproaches
Explicit Deconvolution Formula The final result of explicitdeconvolution method is given as follows
119901119904119863
(119905119863) = minus
119905119863
120572 (119905119863)1199011015840
119908119863+ 119901119908119863
(119905119863) (39)
Plotting of this term versus (int119905119863
0119902119863(120591)119889120591119902
119863(119905119863)) and
(int119905119863
0(1 minus 119902
119863(120591))119889120591(1 minus 119902
119863(119905119863))) for drawdown and buildup
tests respectively yields more accurate resultsThis approachperforms well in field applications
Material Balance Deconvolution The material balance pres-sure drop function for the pressure drawdown test is givenas
Δ119901119908119891 (119905)
119902119863 (119905)
=119901119894minus 119901119908119891 (119905)
119902119863 (119905)
(40)
and the material balance time function for the pressuredrawdown test is given as
119905119898119887
=int119905119863
0119902119863 (120591) 119889120591
119902119863 (119905)
(41)
For the buildup test the material balance pressure dropfunction can be written as
Δ119901119908119904 (Δ119905)
1 minus 119902119863 (Δ119905)
=119901119908119904 (Δ119905) minus 119901
119908119891 (Δ119905 = 0)
1 minus 119902119863 (Δ119905)
(42)
10minus3 10minus2 10minus1 100 101 102 103
Time (hr)
104
103
102
101Dpwf
and
its lo
garit
hmic
der
ivat
ive (
psi)
Dpwf wellbore storaged(Dpwf)d(ln(t)) wellbore storageDps explicit deconvolution methodd(Dps)d(ln(t)) explicit deconvolution methodDps material balance deconvolution methodd(Dps)d(ln(t)) material balance deconvolution methodDps no wellbore storaged(Dps)d(ln(t)) no wellbore storage
Figure 2 Deconvolution for a drawdown test in a homogeneousreservoir
And the material balance time function can be written as
119905119898119887
=int119905119863
0(1 minus 119902
119863 (120591)) 119889120591
1 minus 119902119863 (Δ119905)
(43)
This technique is a good approach for practical applications
5 Verification Examples
We first use few synthetic cases for different reser-voirwellbore systems to validate the applicability of explicitdeconvolution formula and material balance deconvolutionmethod Then the field applications of these techniqueswill be shown by two real cases The unloading rates arecalculated with (20)
The first case is a synthetic test derived from a test designin a vertical oil wellThe reservoir is homogenous and infiniteacting with a constant wellbore storage The wellbore storagecoefficient is 119862
119904= 0015 bblpsi The pressure data are
obtained from the cylindrical source solution (test design)We used the explicit deconvolution formula and material
balance deconvolution method to deconvolve the pressureand unloading rate distorting by wellbore storage effectsTheresults are presented in Figures 1 and 2 for the semilog andlog-log plots in the case of drawdown test and Figures 3and 4 for the semilog and log-log plots in the case of builduptest Besides the deconvolved pressures and their logarithmicderivatives pressure responses with and without wellborestorage are also plotted on the figures for comparisonThrough the deconvolution process we have successfullyremoved the wellbore storage effects so that radial flow canbe identified at early test times The deconvolved pressuresand their logarithmic derivatives are almost identical to
6 Abstract and Applied Analysis
0
500
1000
1500
2000
2500
3000
3500
4000
105104103102101100
Horner time ratio
Dp wellbore storage
Dps explicit deconvolution method(ln
Dps material balance deconvolution
Dps no wellbore storage
ws
and
its lo
garit
hmic
der
ivat
ive (
psi)
Dpws
d(Dp )d(ln )) wellbore storagews (HTR
d(Dps)d )) explicit deconvolution method(HTR
d(Dps)d(ln )) no wellbore storage(HTR)
d(Dps)d(ln )) material balance deconvolution (HTR
Figure 3 Deconvolution for a buildup test in a homogeneousreservoir
10minus3 10minus2 10minus1 100 101 102 103
Time (hr)
104
103
102
101
ws
ws
and
its lo
garit
hmic
der
ivat
ive (
psi)
Dpws
Dp wellbore storage
Dps explicit deconvolution method
Dps material balance deconvolution
Dps no wellbore storage
d(Dp )d(ln )) wellbore storagews (HTR
d(Dps)d )) explicit deconvolution method(HTR
d(Dps)d(ln )) no wellbore storage(HTR)
d(Dps)d(ln )) material balance deconvolution (HTR
(ln
Figure 4 Deconvolution for a buildup test in a homogeneousreservoir
the analytic solutions (without wellbore storage) whichenables us to estimate reservoir parameters accurately
The second case is an oil well drawdown test in adual porosity reservoir The storativity ratio (120596) and inter-porosity flow coefficient (120582) are assumed to be 01 and10minus6 respectively The constant wellbore storage coefficient
is 119862119904= 015 bblpsi The dimensionless wellbore pressure
10minus3 10minus2 10minus1 100 101 102 103
Time (hr)
600
500
400
300
200
100
0
Dpwf wellbore storaged(Dpwf)d(ln(t)) wellbore storageDps explicit deconvolution methodd(Dps)d(ln(t)) explicit deconvolution methodDps material balance deconvolutiond(Dps)d(ln(t)) material balance deconvolutionDps no wellbore storaged(Dps)d(ln(t)) no wellbore storage
Dpwf
and
its lo
garit
hmic
der
ivat
ive (
psi)
Figure 5 Deconvolution for a drawdown test in a dual porosityreservoir
data were obtained from the cylindrical source solution withpseudosteady-state interporosity flow (test design) It is notpossible to identify the dual porosity behavior characteristicof the reservoir from wellbore pressure data because thisbehavior has been masked completely by wellbore storage asindicated in Figures 5 and 6
We deconvolved the pressure and unloading rate andpresent the results in Figures 5 and 6 The deconvolvedresults are reasonably consistent with the analytical solutionAfter deconvolution the dual porosity behavior can beidentified easily from the valley in the logarithmic derivativeof pressure If wellbore storage effects were not removed atearly times it easily could have incorrectly identified thereservoir system as infinite actingThe estimation of reservoirparameters can be obtained now from deconvolved pressure
The third case is an oil well drawdown test in a reservoirwith a sealing fault 250 ft from the well Synthetic pressureresponses were generated by simulation (test design) with aconstant wellbore storage coefficient of 01 bblpsi As we cansee in Figures 7 and 8 the early time infinite acting reservoirbehavior has been masked completely by wellbore storage
To recover the hidden reservoir behavior we performeddeconvolution on the pressure change to remove the wellborestorage effect As in Figures 7 and 8 the deconvolved pressurechange and logarithmic derivative data are almost identicalto the responses without wellbore storageThe deconvolutionresults clearly indicate parallel horizontal lines for the loga-rithmic derivative However without wellbore storage effectselimination this maymislead to identify the reservoir systemas infinite acting So this allows us to correctly identify thereservoir system having a sealing fault
The fourth case is a field example that is taken from theliterature [23] The basic reservoir and fluid properties are
Abstract and Applied Analysis 7
Table 1 Reservoir rock and fluid properties for the first field case
119905119901(hr) ℎ (ft) 119903
119908(ft) 119902
119900(STBD) 119861
119900(RBSTB) 119888
119905(psiminus1) 120583
119900(cp) 120601
1533 107 029 174 106 42 times 10minus6 25 025
Table 2 Reservoir rock and fluid properties for the second field case
119905119901(hr) ℎ (ft) 119903
119908(ft) 119879 (∘F) 119901 (psi) 119902
119900(STBD) 119902
119892(MscfD) 119888
119891(psiminus1) 119879
119889119901(∘F) 119901
119889119901(psi) ∘API 120574
119892
143111 100 0354 2795 7322 104 13925 3 times 10minus6 212 5445 45375 068
10minus3 10minus2 10minus1 100 101 102 103
Time (hr)
103
102
101
100
10minus1
Dpwf wellbore storaged(Dpwf)d(ln(t)) wellbore storageDps explicit deconvolution methodd(Dps)d(ln(t)) explicit deconvolution methodDps material balance deconvolutiond(Dps)d(ln(t)) material balance deconvolutionDps no wellbore storaged(Dps)d(ln(t)) no wellbore storage
Dpwf
and
its lo
garit
hmic
der
ivat
ive (
psi)
Figure 6 Deconvolution for a drawdown test in a dual porosityreservoir
shown in Table 1 In this case we provide the deconvolvedpressure data using the explicit deconvolution formula andthe material balance deconvolution method that is presentedin this work The data are taken from a pressure buildup testand can be reasonably used as field data The deconvolutionresults are shown in Figure 9 in semilog plot As it isillustrated from Figure 9 the corrected pressure provides aclear straight line starting as early as 15 hours in contrast tothe vague line from the uncorrected pressure response
And the last (field) example is a gas condensatewell buildup test following a variable rate productionhistory The duration of last pressure buildup test is361556 hours (from 1889 to 2250556 hours) The basicreservoir and fluid properties for the last drawdown testare shown in Table 2 However the properties of otherproduction tests are not shown here To use our newtechniques effectively for gas wells we use normalizedpseudopressure and pseudotime functions as defined byMeunier et al [24]The final results of explicit deconvolutionformula and material balance deconvolution methodshow two different slopes (Figure 10) where the earlytimes slope is greater than the last one This identifiesthe radial composite behavior of the reservoir With
the corrected pressures the reservoir parameters can also beestimated
6 Conclusions
In this work we have provided some explicit deconvolutionmethods for restoration of constant rate pressure responsesfrom measured pressure data dominated by wellbore storageeffects without sandface rate measurements Based on ourinvestigations we can state the following specific conclusionsrelated to these techniques
(1) The results obtained from the synthetic and fieldexamples show the practical and computational effi-ciency of our deconvolution methods
(2) Modification of 120573-deconvolution is presented wherethe beta quantity is variable rather than a constant
(3) For situations in which there are no downhole ratemeasurements the explicit deconvolution methods(blind deconvolution approaches) can be used toderive the downhole rate function from the pressuremeasurements and restore the reservoir pressureresponse
(4) We can unveil the early time behavior of a reservoirsystemmasked by wellbore storage effects using thesenew approaches Wellbore geometries and reservoirtypes can be identified from this restored pressuredata Furthermore the conventional methods canbe used to analyze this restored formation pressureto estimate formation parameters So the presentedtechniques can be used as the practical tools toimprove pressure transient test interpretation
(5) Deconvolution in well test analysis is often expressedas an inverse problem This approach can be used asan idea to justify the rate normalization method andthe material balance deconvolution method
Appendices
A Russel Method and Its Modification
A method for correcting pressure buildup pressure data freeof wellbore storage effects at early times has been developedby Russell [14] In this method it is not necessary to measurethe rate of influx after closing in at the surface Instead oneuses a theoretical equation which gives the form that thebottom hole pressure should have fluid accumulates in the
8 Abstract and Applied Analysis
10minus3 10minus2 10minus1 100 101 102 103
Time (hr)
250
200
150
100
50
0
Dpwf wellbore storaged(Dpwf)d(ln(t)) wellbore storageDps explicit deconvolution methodd(Dps)d(ln(t)) explicit deconvolution methodDps material balance deconvolutiond(Dps)d(ln(t)) material balance deconvolutionDps no wellbore storaged(Dps)d(ln(t)) no wellbore storage
Dpwf
and
its lo
garit
hmic
der
ivat
ive (
psi)
Figure 7 Deconvolution for a drawdown test in a reservoir with asealing fault
10minus3 10minus2 10minus1 100 101 102 103
Time (hr)
103
102
101
100
Dpwf wellbore storaged(Dpwf)d(ln(t)) wellbore storageDps explicit deconvolution methodd(Dps)d(ln(t)) explicit deconvolution methodDps material balance deconvolutiond(Dps)d(ln(t)) material balance deconvolutionDps no wellbore storaged(Dps)d(ln(t)) no wellbore storage
Dpwf
and
its lo
garit
hmic
der
ivat
ive (
psi)
Figure 8 Deconvolution for a drawdown test in a reservoir with asealing fault
wellbore during buildup This leads to the result that oneshould plot
[119901119908119904 (Δ119905) minus 119901
119908119891 (Δ119905 = 0)]
[1 minus (11198622Δ119905)]
(A1)
versus log (Δ119905) in analyzing pressure buildup data during theearly fill-up period The denominator makes a correction for
104103102101100
Horner time ratio
800
700
600
500
400
300
200
100
0
Dpws
(psi)
Dpws wellbore storageDps explicit deconvolution methodDps material balance deconvolution
Figure 9 Restored pressures for a buildup test in a homogeneousreservoir
0
500
1000
1500
2000
2500
3000
3500
4000
105104103102101100
Horner time ratio
Dpa
(psi)
Dpaw wellbore storage explicit deconvolution method material balance deconvolution
DpasDpas
Figure 10 Restored pressures for a buildup test in a radial compositereservoir
the gradually decreasing flow into the wellbore The quantity1198622is obtained by trial and error as the value which makes the
curve straight at early times After obtaining the straight linesection the rest of the analysis is the same as for any otherpressure buildup This method has the advantage or requiresno additional data over that taken routinely during a shut intest
Due to the time consuming and unconfident of trial anderror to determine 119862
2in Russell method we modify this
method Russellrsquos wellbore storage correction is given as
[119901119908119904 (Δ119905) minus 119901
119908119891 (Δ119905 = 0)]
[1 minus (11198622Δ119905)]
= 119891 (Δ119905 = 1 hr) + 119898119904119897log (Δ119905)
(A2)
Abstract and Applied Analysis 9
Defining the following function and using least squarestechnique obtain the final result
119864 (1198622 119891119898)
=
119873
sum
119894=1
(Δ119901119908119904119894
(1 minus (11198622Δ119905119894))
minus 119891(1198622) minus 119898(119862
2) log(Δ119905
119894))
2
(A3)
Tominimize the function119864 it can differentiate119864with respectto 1198622 119891 and 119898 and setting the results equal to zero After
differentiation it reduces to
120597119864
1205971198622
= minus2
119873
sum
119894=1
(Δ119901119908119904119894
Δ119905119894(1198622minus (1Δ119905
119894))2)
times (Δ119901119908119904119894
(1 minus (11198622Δ119905119894))
minus 119891 (1198622)
minus119898 (1198622) log (Δ119905
119894) ) = 0
(A4)
120597119864
120597119891
= minus2
119873
sum
119894=1
(Δ119901119908119904119894
(1 minus (11198622Δ119905119894))
minus 119891 (1198622) minus 119898 (119862
2) log (Δ119905
119894))
= 0
(A5)120597119864
120597119898
= minus2
119873
sum
119894=1
log (Δ119905119894) (
Δ119901119908119904119894
(1 minus (11198622Δ119905119894))
minus 119891 (1198622)
minus 119898 (1198622) log (Δ119905
119894)) = 0
(A6)
From (A4) we define the following function
119883(1198622)
=
119873
sum
119894=1
(Δ119901119908119904119894
Δ119905119894(1198622minus (1Δ119905
119894))2)
times(Δ119901119908119904119894
(1minus (11198622Δ119905119894))minus119891 (119862
2) minus119898 (119862
2) log (Δ119905
119894))
= 0
(A7)
The roots of (A7) can be found via the following algorithm
1198622(119894+1)
= 1198622(119894)
minus119883 (1198622)
1198831015840 (1198622) (A8)
where
119883(1198622) =
119873
sum
119894=1
(Δ119901119908119904119894
Δ119905119894(1198622minus (1Δ119905
119894))2)
times (Δ119901119908119904119894
(1 minus (11198622Δ119905119894))
minus 119891 (1198622)
minus119898 (1198622) log (Δ119905
119894))
1198831015840(1198622) =
119873
sum
119894=1
(minus2Δ119901119908119904119894
Δ119905119894(1198622minus (1Δ119905
119894))3)
times (Δ119901119908119904119894
Δ119905119894(1198622minus (1Δ119905
119894))2
times [Δ119901119908119904119894
(1 minus (11198622Δ119905))
minus 119891 (1198622)
minus119898 (1198622) log (Δ119905
119894)])
+ (Δ119901119908119904119894
Δ119905119894(1198622minus (1Δ119905
119894))2)
times (minusΔ119901119908119904119894
Δ119905119894(1198622minus (1Δ119905
119894))3minus 1198911015840(1198622)
minus1198981015840(1198622) log (Δ119905
119894))
(A9)
where the unknown functions are given by the followingrelations
119891 (1198622) =
120573119906 (1198622) minus 120572V (119862
2)
1205732 minus 119873120572(A10)
whose derivative is
1198911015840(1198622) =
1205731199061015840(1198622) minus 120572V1015840 (119862
2)
1205732 minus 119873120572
119898 (1198622) =
119873119906 (1198622) minus 120573V (119862
2)
119873120572 minus 1205732
(A11)
where its derivative is
1198981015840(1198622) =
1198731199061015840(1198622) minus 120573V1015840 (119862
2)
119873120572 minus 1205732
119906 (1198622) =
119873
sum
119894=1
log (Δ119905119894)
Δ119901119908119904119894
(1 minus (11198622Δ119905119894))
(A12)
10 Abstract and Applied Analysis
and after differentiation it yields
1199061015840(1198622) = minus
119873
sum
119894=1
log (Δ119905
119894)
Δ119905119894
Δ119901119908119904119894
(1198622minus (1Δ119905
119894))2
V (1198622) =
119873
sum
119894=1
Δ119901119908119904119894
(1 minus (11198622Δ119905119894))
(A13)
which has the following derivative
V1015840 (1198622) = minus
119873
sum
119894=1
1
Δ119905119894
Δ119901119908119904119894
(1198622minus (1Δ119905
119894))2 (A14)
finally the values of 120572 and 120573 parameters are given by thefollowing formulas
120572 =
119873
sum
119894=1
(log(Δ119905119894))2
120573 =
119873
sum
119894=1
log (Δ119905119894)
(A15)
B General Formula for Inverse Problem
The convolution of (6) yields
119901119904119863
(119905119863) = int
119905119863
0
119870 (120591) 119901119908119863 (119905119863 minus 120591) 119889120591
= int
119905119863
0
119870(119905119863minus 120591) 119901
119908119863 (120591) 119889120591
(B1)
where
119870(119905119863) = 119871minus1(
1
119904119902 (119904)) (B2)
where 119871minus1 is the inverse Lapalce transform operator 119870(119905119863)
can be computed either from the Laplace transforms of119902119863(119905119863) data a curve-fitted equation of 119902
119863(119905119863) data or using
another technique with a relation between119870(119905119863) and 119902
119863(119905119863)
However it would be time consuming to invert all the 119902119863(119905119863)
data in Laplace space and transform it back to real space inaccordance with (B2) Once an approximation is obtained itwill be easy to compute119870(119905
119863) and integrate (B1) to determine
119901119904119863(119905119863) However finding some approximation functions for
119902119863(119905119863) data may be difficult We provide an approach to
overcome this problemThis idea is based on finding anotherrelation between 119870(119905
119863) and 119902
119863(119905119863) that is
ℓ (1) =1
119904=
1
119904119902119863(119904)
119902119863(119904) = ℓ (int
119905119863
0
119870 (120591) 119902119863 (119905119863 minus 120591) 119889120591)
= ℓ(int
119905119863
0
119870(119905119863minus 120591) 119902
119863 (120591) 119889120591)
(B3)or
1 = int
119905119863
0
119870 (120591) 119902119863 (119905119863 minus 120591) 119889120591 = int
119905119863
0
119870(119905119863minus 120591) 119902
119863 (120591) 119889120591
(B4)
The other relations that can be used are
ℓ (119905119863) =
1
1199042=
1
1199042119902119863(119904)
119902119863(119904) = ℓ (int
119905119863
0
119871 (120591) 119902119863 (119905119863 minus 120591) 119889120591)
= ℓ(int
119905119863
0
119871 (119905119863minus 120591) 119902
119863 (120591) 119889120591)
(B5)
or
119905119863= int
119905119863
0
119871 (120591) 119902119863(119905119863minus 120591) 119889120591 = int
119905119863
0
119871 (119905119863minus 120591) 119902
119863(120591) 119889120591
(B6)
ℓ (1) =1
119904=
1
1199042119902119863(119904)
119904119902119863(119904) = ℓ (int
119905119863
0
119871 (120591) 1199021015840
119863(119905119863minus 120591) 119889120591)
= ℓ(int
119905119863
0
119871 (119905119863minus 120591) 119902
1015840
119863(120591) 119889120591)
(B7)
or
1 = int
119905119863
0
119871 (120591) 1199021015840
119863(119905119863minus 120591) 119889120591 = int
119905119863
0
119871 (119905119863minus 120591) 119902
1015840
119863(120591) 119889120591
(B8)
where
119871 (119905119863) = ℓminus1(
1
1199042119902 (119904)) (B9)
Equation (B4) can be used simultaneously with (B1) toanalyze the deconvolution problem
Equation (11) can be written as
119901119904119863
(119904) = 119904119901119908119863
(119904)1
1199042119902119863(119904)
(B10)
or
119901119904119863
(119905119863) = int
119905119863
0
119871 (120591) 1199011015840
119908119863(119905119863minus 120591) 119889120591
= int
119905119863
0
119871 (119905119863minus 120591) 119901
1015840
119908119863(120591) 119889120591
(B11)
The applications of (B6) and (B8) are with (B11)For justifying the rate normalizationmethod andmaterial
balance deconvolution it can use (B1) (B4) ((B11) can alsobe used with (B6) or (B8) for this goal) and generalizedmean value theorem for integrals This theorem is expressedas follows
Theorem 1 Let 119891 and 119892 be continuous on [119886 119887] and supposethat 119892 does not change sign on [119886 119887] Then there is a point 119888 in[119886 119887] such that
int
119887
119886
119891 (119905) 119892 (119905) 119889119905 = 119891 (119888) int
119887
119886
119892 (119905) 119889119905 (B12)
Abstract and Applied Analysis 11
By Theorem 1 there are some values 1198881and 1198882between 0
and 119905119863such that
119901119904119863
(119905119863) = 119901119908119863
(1198881) int
119905119863
0
119870(119905119863minus 120591) 119889120591 (B13)
1 = 119902119863(1198882) int
119905119863
0
119870(119905119863minus 120591) 119889120591 (B14)
By setting 1198881
= 1198882
= 119905119863and dividing the right and left
side of (B13) respectively the rate normalization techniqueis developed However the values 119888
1and 1198882are not equal
to each other and are not equal to 119905119863
in general Thisis the reason of error in rate normalization and materialbalance deconvolution methods 119888
1and 1198882are less than 119905
119863in
these methods for the duration of wellbore storage distortionduring a pressure transient test Since these conditions areapproximately satisfied in material balance deconvolutionthe results of material balance are more accurate and moreconfident than rate normalization
Nomenclature
API API gravity ∘API119861119900 Oil formation volume factor RBSTB
119888119905 Total isothermal compressibility factor
psiminus1
1198622 Arbitrary constant hr
minus1
119862119863 Wellbore storage constant dimensionless
119862119904 Wellbore storage coefficient bblpsi
ℎ Formation thickness ftHTR Horner time ratio dimensionless119896 Formation permeability md119870 Kernel of the convolution integral119871 Kernel of the convolution integral119873119901 Cumulative oil production bbl
119901 Pressure psi119901119886 Normalized pseudopressure psi
119901 Average reservoir pressure psi119901119863 Formation pressure dimensionless
119901119889119901 Dewpoint pressure psi
119901119894 Initial reservoir pressure psi
119901119904119863 Formation pressure including skin
dimensionless119901119904 Formation pressure including skin psi
119901119908119863
Pressure draw down dimensionless119901119908119891 Flowing bottom hole pressure psi
119901119908119904 Shut-in bottom hole pressure psi
119902119863 Sandface flow rate dimensionless
119902119892 Gas flow rate MscfD
119902119900 Oil flow rate STBD
119903119908 Wellbore radius ft
119904 Laplace variable119878 Skin factor119905 Time hr119905119863 Time dimensionless
119905119901 Production time hr
119879 Reservoir temperature ∘F
119879119889119901 Dewpoint temperature ∘F
120572 Explicit deconvolution variabledimensionless
120573 Beta deconvolution variabledimensionless
120574119892 Gas gravity (air = 10)
120582 Interporosity flow coefficient of dualporosity reservoir
120601 Porosity of reservoir rock dimensionless120583119892 Gas viscosity cp
120583119900 Oil viscosity cp
120596 Storativity ratio of dual porosity reservoirminus Laplace transform of a function1015840 Derivative of a function
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
References
[1] Y Cheng W J Lee and D A McVay ldquoFast-Fourier-transform-based deconvolution for interpretation of pressure-transient-test data dominated by Wellbore storagerdquo SPE Reservoir Eval-uation and Engineering vol 8 no 3 pp 224ndash239 2005
[2] M M Levitan and M R Wilson ldquoDeconvolution of pressureand rate data from gas reservoirs with significant pressuredepletionrdquo SPE Journal vol 17 no 3 pp 727ndash741 2012
[3] Y Liu and R N Horne ldquoInterpreting pressure and flow-ratedata from permanent downhole gauges by use of data-miningapproachesrdquo SPE Journal vol 18 no 1 pp 69ndash82 2013
[4] O Ogunrewo and A C Gringarten ldquoDeconvolution of welltest data in lean and rich gas condensate and volatile oil wellsbelow saturation pressurerdquo in Proceedings of the SPE AnnualTechnical Conference and Exhibition Document ID 166340-MS New Orleans La USA September-October 2013
[5] R C Earlougher Jr Advances in Well Test Analysis vol 5 ofMonogragh Series SPE Richardson Tex USA 1977
[6] F Kucuk and L Ayestaran ldquoAnalysis of simultaneously mea-sured pressure and sandface flow rate in transient well testingrdquoJournal of PetroleumTechnology vol 37 no 2 pp 323ndash334 1985
[7] M Onur M Cinar D IIk P P Valko T A Blasingame and PHegeman ldquoAn investigation of recent deconvolution methodsfor well-test data analysisrdquo SPE Journal vol 13 no 2 pp 226ndash247 2008
[8] F J KuchukMOnur andFHollaender ldquoPressure transient testdesign and interpretationrdquo Developments in Petroleum Sciencevol 57 pp 303ndash359 2010
[9] A C Gringarten D P Bourdet P A Landel and V J KniazeffldquoA comparison between different skin and wellbore storagetype-curves for early-time transient analysisrdquo in Proceedings ofthe SPE Annual Technical Conference and Exhibition Paper SPE8205 Las Vegas Nev USA 1979
[10] A C Gringarten ldquoFrom straight lines to deconvolution theevolution of the state of the art in well test analysisrdquo SPEReservoir Evaluation and Engineering vol 11 no 1 pp 41ndash622008
[11] J A Cumming D A Wooff T Whittle and A C GringartenldquoMultiple well deconvolutionrdquo in Proceedings of the SPE Annual
12 Abstract and Applied Analysis
Technical Conference and Exhibition Document ID 166458-MS New Orleans La USA September-October 2013
[12] O Bahabanian D Ilk N Hosseinpour-Zonoozi and T ABlasingame ldquoExplicit deconvolution of well test data distortedby wellbore storagerdquo in Proceedings of the SPE Annual TechnicalConference and Exhibition Paper SPE 103216 pp 4444ndash4453San Antonio Tex USA September 2006
[13] J Jones and A Chen ldquoSuccessful applications of pressure-ratedeconvolution in theCad-Nik tight gas formations of the BritishColumbia foothillsrdquo Journal of Canadian Petroleum Technologyvol 51 no 3 pp 176ndash192 2012
[14] D G Russell ldquoExtension of pressure buildup analysis methodsrdquoJournal of Petroleum Technology vol 18 no 12 pp 1624ndash16361966
[15] R E Gladfelter G W Tracy and L E Wilsey ldquoSelecting wellswhich will respond to production-stimulation treatmentrdquo inDrilling and Production Practice API Dallas Tex USA 1955
[16] M J Fetkovich and M E Vienot ldquoRate normalization ofbuildup pressure by using afterflow datardquo Journal of PetroleumTechnology vol 36 no 13 pp 2211ndash2224 1984
[17] J L Johnston Variable rate analysis of transient well testdata using semi-analytical methods [MS thesis] Texas AampMUniversity College Station Tex USA 1992
[18] A F van Everdingen ldquoThe skin effect and its influence on theproductive capacity of a wellrdquo Journal of Petroleum Technologyvol 5 no 6 pp 171ndash176 1953
[19] W Hurst ldquoEstablishment of the skin effect and its impedimentto fluid flow into a well borerdquo The Petroleum Engineer pp B6ndashB16 1953
[20] J A Joseph and L F Koederitz ldquoA simple nonlinear model forrepresentation of field transient responsesrdquo Paper SPE 11435SPE Richardson Tex USA 1982
[21] F J Kuchuk ldquoGladfelter deconvolutionrdquo SPE Formation Evalu-ation vol 5 no 3 pp 285ndash292 1990
[22] A F van Everdingen and W Hurst ldquoThe application of thelaplace transformation to flow problems in reservoirsrdquo Journalof Petroleum Technology vol 1 no 12 pp 305ndash324 1949
[23] D Bourdet T M Whittle A A Douglas and Y M Pirard ldquoAnew set of type curves simplifies well test analysisrdquo World Oilpp 95ndash106 1983
[24] D F Meunier C S Kabir and M J Wittmann ldquoGas well testanalysis use of normalized pseudovariablesrdquo SPE FormationEvaluation vol 2 no 4 pp 629ndash636 1987
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
Abstract and Applied Analysis 3
layered anisotropic and heterogeneous systems as long as thefluid in the reservoir is single phaseThese equations can alsobe applied to both drawdown and buildup tests if the initialconditions are known For a reservoir with an initial constantand uniform pressure distribution (119901
119863(0) = 0) (4) and (5)
can be expressed as
119901119908119863
(119905119863) = int
119905119863
0
1199021015840
119863(120591) 119901119904119863 (119905119863 minus 120591) 119889120591 (6)
119901119908119863
(119905119863) = 119878119902
119863(119905119863) + int
119905119863
0
119902119863 (120591) 119901
1015840
119904119863(119905119863minus 120591) 119889120591 (7)
Furthermore (6) and (7) also can be expressed as
119901119908119863
(119905119863) = int
119905119863
0
1199021015840
119863(119905119863minus 120591) 119901
119904119863 (120591) 119889120591 (8)
119901119908119863
(119905119863) = 119878119902
119863(119905119863) + int
119905119863
0
119902119863(119905119863minus 120591) 119901
1015840
119904119863(120591) 119889120591 (9)
In (6) and (8) it is assumed that 1199021015840119863(119905119863) exists If 119902
119863(119905119863) is
constant then (7) and (9) must be used Equations (6) and(8) are known as a Volterra integral equation of first kind andthe convolution type
Taking the Laplace transform of (6) yields
119901119908119863
(119904) = 119904119902119863(119904) 119901119904119863
(119904) (10)
Rearranging (10) for the equivalent constant rate pressuredrop function 119901
119904119863(119904) we obtain
119901119904119863
(119904) =119901119908119863
(119904)
119904119902119863(119904)
(11)
By taking the Laplace transform of (1) one can easily obtain
119902119863(119904) =
1
119904minus
1
119904 + 120573 (12)
Substituting (12) into (11) and then taking the inverse Laplacetransform of (11) yield the beta deconvolution formula
119901119904119863
(119905119863) =
1
1205731199011015840
119908119863(119905119863) + 119901119908119863
(119905119863) (13)
where
1199011015840
119908119863(119905119863) =
119889119901119908119863
119889119905119863
(14)
where we note that (13) is specifically valid only for theexponential sandface flow rate profile given by (1) This maypresent a serious limitation in terms of practical applicationof the 120573-deconvolutionmethod To overcome this restrictionwe propose an idea It would be assumed that120573 is not constantin general If 119902
119863(119905119863)data are given the procedure is as follows
119902119863(119905119863) = 1 minus 119890
minus120573(119905119863)119905119863 (15)
Rearranging (15) yields
1 minus 119902119863(119905119863) = 119890minus120573(119905119863)119905119863 (16)
then
minus120573 (119905119863) 119905119863= ln (1 minus 119902
119863(119905119863)) (17)
so
120573 (119905119863) = minus
ln (1 minus 119902119863(119905119863))
119905119863
(18)
Substituting of (18) into (13) instead of 120573 term yields themodified 120573-deconvolution
119901119904119863
(119905119863) =
1
120573 (119905119863)1199011015840
119908119863(119905119863) + 119901119908119863
(119905119863) (19)
Plot of (19) versus (int119905119863
0119902119863(120591)119889120591119902
119863(119905119863)) and (int
119905119863
0(1 minus
119902119863(120591))1198891205911 minus 119902
119863(119905119863)) for drawdown and buildup tests
respectively should be considered more accurate to use as apractical tool for field applications As mentioned before thevariable 120573 quantity (rather than a constant) shows the generaland more accurate behavior of 119902
119863(119905119863) So accuracy of (19) is
also more than (13) It can be emphasized that (13) and (19)are applicable in the wellbore storage duration and for thelater times the given pressure data are correct However theresults of (19) show that this equation is not bad for the latertimes
22 Explicit Deconvolution Formula A few types of approx-imation functions can be used to approximate 119902
119863(119905119863) An
approximate proof is given here to show that the exponentialfunction gives a good representation of 119902
119863(119905119863) By setting
119902119863(119905119863) = 1 minus 119891 (119905
119863) (20)
119891 (0) = 1 (21)
Equation (21) is an initial condition for (20)Taking the time derivative of (20) gives
1199021015840
119863(119905119863) = minus119891
1015840(119905119863) (22)
Taking the time derivative of (2) gives
1199021015840
119863(119905119863) = minus119862
11986311990110158401015840
119908119863 (23)
From (2) and (20) we have
119891 (119905119863) = 1198621198631199011015840
119908119863 (24)
And from (22) and (23) we have
1198911015840(119905119863) = 11986211986311990110158401015840
119908119863 (25)
After dividing right and left hand sides of (25) by (24) theresult is
1198911015840(119905119863)
119891 (119905119863)=11990110158401015840
119908119863(119905119863)
1199011015840119908119863
(119905119863) (26)
Integrating both sides of (26) with respect to 119905119863from 0 to 119905
119863
yields
int
119905119863
0
1198911015840(120591)
119891 (120591)119889120591 = int
119905119863
0
11990110158401015840
119908119863(120591)
1199011015840119908119863
(120591)119889120591 (27)
4 Abstract and Applied Analysis
Using elementary calculus yields
ln (1003816100381610038161003816119891(120591)
1003816100381610038161003816)1003816100381610038161003816119905119863
0= ln (1003816100381610038161003816119891 (119905
119863)1003816100381610038161003816) minus ln (1) = int
119905119863
0
11990110158401015840
119908119863(120591)
1199011015840119908119863
(120591)119889120591
(28)
So
119891 (119905119863) = 119890120572(119905119863) (29)
where
120572 (119905119863) = int
119905119863
0
11990110158401015840
119908119863(120591)
1199011015840119908119863
(120591)119889120591 (30)
Now the explicit deconvolution of data based on (20) can bewritten as follows
119901119904119863
(119905119863) = minus
1
120572 (119905119863) 119905119863
1199011015840
119908119863+ 119901119908119863
(119905119863) (31)
or
119901119904119863
(119905119863) = minus
119905119863
120572 (119905119863)1199011015840
119908119863+ 119901119908119863
(119905119863) (32)
that is similar to (19) The other similarity between (19) and(32) is that the plot of (32) versus (int119905119863
0119902119863(120591)119889120591119902
119863(119905119863)) and
(int119905119863
0(1 minus 119902
119863(120591))119889120591(1 minus 119902
119863(119905119863))) for drawdown and buildup
tests respectively yields the more accurate results and isusable for practical applications The applicability of (32) isalso for duration of wellbore storage effects And for the othertimes the given pressure data are correct However using thismethod for the later times yields not bad results
3 Material Balance Deconvolution
Material balance deconvolution is an extension of the ratenormalization method Johnston [17] defines a new 119909-axisplotting function (material balance time) that provides anapproximate deconvolution of the variable rate pressuretransient problem There are numerous assumptions associ-ated with the material balance deconvolution methods oneof the most widely accepted assumptions is that the rateprofilemust change smoothly andmonotonically In practicalterms this condition should be met for the wellbore storageproblem
The general form of material balance deconvolution isprovided for the pressure drawdown case in terms of thematerial balance time function and the rate-normalizedpressure drop function The material balance time functionis given as
119905119898119887
=119873119901
119902119863 (119905)
(33)
where
119873119901= int
119905
0
119902119863 (120591) 119889120591 (34)
The rate-normalized pressure drop function is given by
Δ119901119908119891 (119905)
119902119863 (119905)
=119901119894minus 119901119908119891 (119905)
119902119863 (119905)
(35)
The material balance time function for the pressure buildupcase is given as
119905119898119887
=119873119901
1 minus 119902119863 (Δ119905)
(36)
where
119873119901= int
Δ119905
0
(1 minus 119902119863 (120591)) 119889120591 (37)
The rate-normalized pressure drop function for the pressurebuildup case is given as
Δ119901119908119904 (Δ119905)
1 minus 119902119863 (Δ119905)
=119901119908119904 (Δ119905) minus 119901
119908119891 (Δ119905 = 0)
1 minus 119902119863 (Δ119905)
(38)
4 Methodology
In this work several methods are provided to restore thepressure datawithout any sandface flow rate informationThevalues of sandface flow rate data for all methods that requirethese data may be obtained from (20) We evaluated a veryold correctionmethod by Russell [14] However we modifiedthis method by least square method but it was found thatthis method is to be unacceptable for all applications Thederivation of this method and its modification is given inAppendix A
The 120573-deconvolution and its modification are anothertechnique that is mentioned in this study The formulationof this method is presented in Section 21 The other explicitdeconvolutionmethod is an approach similar to themodified120573-deconvolution that uses only the pressure data for decon-volution of pressure and rate data The formulation of thismethod was derived in Section 22 The 120573-deconvolution itsmodification and explicit deconvolution formula can be usedfor practical applications The applicability of these methodsis for duration of wellbore storage effects However usingthese approaches for the other times yields not bad results
The other technique is material balance deconvolutionmethod [17] that has sufficiently accurate resultsThismethodcan also be used as a good approach for practical applicationsThe formulas for this method were given in Section 3
In Appendix B deconvolution is expressed as an inverseproblem that is analyzed by a tricky idea Based on thisidea an interesting proof is given for rate normalizationmethod and material balance deconvolution that is a goodapproach to find the errors and limitations of these methodsand overcome them
Abstract and Applied Analysis 5
10minus3 10minus2 10minus1 100 101 102 1030
500
1000
1500
2000
2500
3000
3500
4000
Time (hr)
Dpwf wellbore storaged(Dpwf)d(ln(t)) wellbore storageDps explicit deconvolution methodd(Dps)d(ln(t)) explicit deconvolution methodDps material balance deconvolution methodd(Dps)d(ln(t)) material balance deconvolution methodDps no wellbore storaged(Dps)d(ln(t)) no wellbore storage
Dpwf
and
its lo
garit
hmic
der
ivat
ive (
psi)
Figure 1 Deconvolution for a drawdown test in a homogeneousreservoir
From the above-mentioned methods we use the fol-lowing two techniques to show the verification of theseapproaches
Explicit Deconvolution Formula The final result of explicitdeconvolution method is given as follows
119901119904119863
(119905119863) = minus
119905119863
120572 (119905119863)1199011015840
119908119863+ 119901119908119863
(119905119863) (39)
Plotting of this term versus (int119905119863
0119902119863(120591)119889120591119902
119863(119905119863)) and
(int119905119863
0(1 minus 119902
119863(120591))119889120591(1 minus 119902
119863(119905119863))) for drawdown and buildup
tests respectively yields more accurate resultsThis approachperforms well in field applications
Material Balance Deconvolution The material balance pres-sure drop function for the pressure drawdown test is givenas
Δ119901119908119891 (119905)
119902119863 (119905)
=119901119894minus 119901119908119891 (119905)
119902119863 (119905)
(40)
and the material balance time function for the pressuredrawdown test is given as
119905119898119887
=int119905119863
0119902119863 (120591) 119889120591
119902119863 (119905)
(41)
For the buildup test the material balance pressure dropfunction can be written as
Δ119901119908119904 (Δ119905)
1 minus 119902119863 (Δ119905)
=119901119908119904 (Δ119905) minus 119901
119908119891 (Δ119905 = 0)
1 minus 119902119863 (Δ119905)
(42)
10minus3 10minus2 10minus1 100 101 102 103
Time (hr)
104
103
102
101Dpwf
and
its lo
garit
hmic
der
ivat
ive (
psi)
Dpwf wellbore storaged(Dpwf)d(ln(t)) wellbore storageDps explicit deconvolution methodd(Dps)d(ln(t)) explicit deconvolution methodDps material balance deconvolution methodd(Dps)d(ln(t)) material balance deconvolution methodDps no wellbore storaged(Dps)d(ln(t)) no wellbore storage
Figure 2 Deconvolution for a drawdown test in a homogeneousreservoir
And the material balance time function can be written as
119905119898119887
=int119905119863
0(1 minus 119902
119863 (120591)) 119889120591
1 minus 119902119863 (Δ119905)
(43)
This technique is a good approach for practical applications
5 Verification Examples
We first use few synthetic cases for different reser-voirwellbore systems to validate the applicability of explicitdeconvolution formula and material balance deconvolutionmethod Then the field applications of these techniqueswill be shown by two real cases The unloading rates arecalculated with (20)
The first case is a synthetic test derived from a test designin a vertical oil wellThe reservoir is homogenous and infiniteacting with a constant wellbore storage The wellbore storagecoefficient is 119862
119904= 0015 bblpsi The pressure data are
obtained from the cylindrical source solution (test design)We used the explicit deconvolution formula and material
balance deconvolution method to deconvolve the pressureand unloading rate distorting by wellbore storage effectsTheresults are presented in Figures 1 and 2 for the semilog andlog-log plots in the case of drawdown test and Figures 3and 4 for the semilog and log-log plots in the case of builduptest Besides the deconvolved pressures and their logarithmicderivatives pressure responses with and without wellborestorage are also plotted on the figures for comparisonThrough the deconvolution process we have successfullyremoved the wellbore storage effects so that radial flow canbe identified at early test times The deconvolved pressuresand their logarithmic derivatives are almost identical to
6 Abstract and Applied Analysis
0
500
1000
1500
2000
2500
3000
3500
4000
105104103102101100
Horner time ratio
Dp wellbore storage
Dps explicit deconvolution method(ln
Dps material balance deconvolution
Dps no wellbore storage
ws
and
its lo
garit
hmic
der
ivat
ive (
psi)
Dpws
d(Dp )d(ln )) wellbore storagews (HTR
d(Dps)d )) explicit deconvolution method(HTR
d(Dps)d(ln )) no wellbore storage(HTR)
d(Dps)d(ln )) material balance deconvolution (HTR
Figure 3 Deconvolution for a buildup test in a homogeneousreservoir
10minus3 10minus2 10minus1 100 101 102 103
Time (hr)
104
103
102
101
ws
ws
and
its lo
garit
hmic
der
ivat
ive (
psi)
Dpws
Dp wellbore storage
Dps explicit deconvolution method
Dps material balance deconvolution
Dps no wellbore storage
d(Dp )d(ln )) wellbore storagews (HTR
d(Dps)d )) explicit deconvolution method(HTR
d(Dps)d(ln )) no wellbore storage(HTR)
d(Dps)d(ln )) material balance deconvolution (HTR
(ln
Figure 4 Deconvolution for a buildup test in a homogeneousreservoir
the analytic solutions (without wellbore storage) whichenables us to estimate reservoir parameters accurately
The second case is an oil well drawdown test in adual porosity reservoir The storativity ratio (120596) and inter-porosity flow coefficient (120582) are assumed to be 01 and10minus6 respectively The constant wellbore storage coefficient
is 119862119904= 015 bblpsi The dimensionless wellbore pressure
10minus3 10minus2 10minus1 100 101 102 103
Time (hr)
600
500
400
300
200
100
0
Dpwf wellbore storaged(Dpwf)d(ln(t)) wellbore storageDps explicit deconvolution methodd(Dps)d(ln(t)) explicit deconvolution methodDps material balance deconvolutiond(Dps)d(ln(t)) material balance deconvolutionDps no wellbore storaged(Dps)d(ln(t)) no wellbore storage
Dpwf
and
its lo
garit
hmic
der
ivat
ive (
psi)
Figure 5 Deconvolution for a drawdown test in a dual porosityreservoir
data were obtained from the cylindrical source solution withpseudosteady-state interporosity flow (test design) It is notpossible to identify the dual porosity behavior characteristicof the reservoir from wellbore pressure data because thisbehavior has been masked completely by wellbore storage asindicated in Figures 5 and 6
We deconvolved the pressure and unloading rate andpresent the results in Figures 5 and 6 The deconvolvedresults are reasonably consistent with the analytical solutionAfter deconvolution the dual porosity behavior can beidentified easily from the valley in the logarithmic derivativeof pressure If wellbore storage effects were not removed atearly times it easily could have incorrectly identified thereservoir system as infinite actingThe estimation of reservoirparameters can be obtained now from deconvolved pressure
The third case is an oil well drawdown test in a reservoirwith a sealing fault 250 ft from the well Synthetic pressureresponses were generated by simulation (test design) with aconstant wellbore storage coefficient of 01 bblpsi As we cansee in Figures 7 and 8 the early time infinite acting reservoirbehavior has been masked completely by wellbore storage
To recover the hidden reservoir behavior we performeddeconvolution on the pressure change to remove the wellborestorage effect As in Figures 7 and 8 the deconvolved pressurechange and logarithmic derivative data are almost identicalto the responses without wellbore storageThe deconvolutionresults clearly indicate parallel horizontal lines for the loga-rithmic derivative However without wellbore storage effectselimination this maymislead to identify the reservoir systemas infinite acting So this allows us to correctly identify thereservoir system having a sealing fault
The fourth case is a field example that is taken from theliterature [23] The basic reservoir and fluid properties are
Abstract and Applied Analysis 7
Table 1 Reservoir rock and fluid properties for the first field case
119905119901(hr) ℎ (ft) 119903
119908(ft) 119902
119900(STBD) 119861
119900(RBSTB) 119888
119905(psiminus1) 120583
119900(cp) 120601
1533 107 029 174 106 42 times 10minus6 25 025
Table 2 Reservoir rock and fluid properties for the second field case
119905119901(hr) ℎ (ft) 119903
119908(ft) 119879 (∘F) 119901 (psi) 119902
119900(STBD) 119902
119892(MscfD) 119888
119891(psiminus1) 119879
119889119901(∘F) 119901
119889119901(psi) ∘API 120574
119892
143111 100 0354 2795 7322 104 13925 3 times 10minus6 212 5445 45375 068
10minus3 10minus2 10minus1 100 101 102 103
Time (hr)
103
102
101
100
10minus1
Dpwf wellbore storaged(Dpwf)d(ln(t)) wellbore storageDps explicit deconvolution methodd(Dps)d(ln(t)) explicit deconvolution methodDps material balance deconvolutiond(Dps)d(ln(t)) material balance deconvolutionDps no wellbore storaged(Dps)d(ln(t)) no wellbore storage
Dpwf
and
its lo
garit
hmic
der
ivat
ive (
psi)
Figure 6 Deconvolution for a drawdown test in a dual porosityreservoir
shown in Table 1 In this case we provide the deconvolvedpressure data using the explicit deconvolution formula andthe material balance deconvolution method that is presentedin this work The data are taken from a pressure buildup testand can be reasonably used as field data The deconvolutionresults are shown in Figure 9 in semilog plot As it isillustrated from Figure 9 the corrected pressure provides aclear straight line starting as early as 15 hours in contrast tothe vague line from the uncorrected pressure response
And the last (field) example is a gas condensatewell buildup test following a variable rate productionhistory The duration of last pressure buildup test is361556 hours (from 1889 to 2250556 hours) The basicreservoir and fluid properties for the last drawdown testare shown in Table 2 However the properties of otherproduction tests are not shown here To use our newtechniques effectively for gas wells we use normalizedpseudopressure and pseudotime functions as defined byMeunier et al [24]The final results of explicit deconvolutionformula and material balance deconvolution methodshow two different slopes (Figure 10) where the earlytimes slope is greater than the last one This identifiesthe radial composite behavior of the reservoir With
the corrected pressures the reservoir parameters can also beestimated
6 Conclusions
In this work we have provided some explicit deconvolutionmethods for restoration of constant rate pressure responsesfrom measured pressure data dominated by wellbore storageeffects without sandface rate measurements Based on ourinvestigations we can state the following specific conclusionsrelated to these techniques
(1) The results obtained from the synthetic and fieldexamples show the practical and computational effi-ciency of our deconvolution methods
(2) Modification of 120573-deconvolution is presented wherethe beta quantity is variable rather than a constant
(3) For situations in which there are no downhole ratemeasurements the explicit deconvolution methods(blind deconvolution approaches) can be used toderive the downhole rate function from the pressuremeasurements and restore the reservoir pressureresponse
(4) We can unveil the early time behavior of a reservoirsystemmasked by wellbore storage effects using thesenew approaches Wellbore geometries and reservoirtypes can be identified from this restored pressuredata Furthermore the conventional methods canbe used to analyze this restored formation pressureto estimate formation parameters So the presentedtechniques can be used as the practical tools toimprove pressure transient test interpretation
(5) Deconvolution in well test analysis is often expressedas an inverse problem This approach can be used asan idea to justify the rate normalization method andthe material balance deconvolution method
Appendices
A Russel Method and Its Modification
A method for correcting pressure buildup pressure data freeof wellbore storage effects at early times has been developedby Russell [14] In this method it is not necessary to measurethe rate of influx after closing in at the surface Instead oneuses a theoretical equation which gives the form that thebottom hole pressure should have fluid accumulates in the
8 Abstract and Applied Analysis
10minus3 10minus2 10minus1 100 101 102 103
Time (hr)
250
200
150
100
50
0
Dpwf wellbore storaged(Dpwf)d(ln(t)) wellbore storageDps explicit deconvolution methodd(Dps)d(ln(t)) explicit deconvolution methodDps material balance deconvolutiond(Dps)d(ln(t)) material balance deconvolutionDps no wellbore storaged(Dps)d(ln(t)) no wellbore storage
Dpwf
and
its lo
garit
hmic
der
ivat
ive (
psi)
Figure 7 Deconvolution for a drawdown test in a reservoir with asealing fault
10minus3 10minus2 10minus1 100 101 102 103
Time (hr)
103
102
101
100
Dpwf wellbore storaged(Dpwf)d(ln(t)) wellbore storageDps explicit deconvolution methodd(Dps)d(ln(t)) explicit deconvolution methodDps material balance deconvolutiond(Dps)d(ln(t)) material balance deconvolutionDps no wellbore storaged(Dps)d(ln(t)) no wellbore storage
Dpwf
and
its lo
garit
hmic
der
ivat
ive (
psi)
Figure 8 Deconvolution for a drawdown test in a reservoir with asealing fault
wellbore during buildup This leads to the result that oneshould plot
[119901119908119904 (Δ119905) minus 119901
119908119891 (Δ119905 = 0)]
[1 minus (11198622Δ119905)]
(A1)
versus log (Δ119905) in analyzing pressure buildup data during theearly fill-up period The denominator makes a correction for
104103102101100
Horner time ratio
800
700
600
500
400
300
200
100
0
Dpws
(psi)
Dpws wellbore storageDps explicit deconvolution methodDps material balance deconvolution
Figure 9 Restored pressures for a buildup test in a homogeneousreservoir
0
500
1000
1500
2000
2500
3000
3500
4000
105104103102101100
Horner time ratio
Dpa
(psi)
Dpaw wellbore storage explicit deconvolution method material balance deconvolution
DpasDpas
Figure 10 Restored pressures for a buildup test in a radial compositereservoir
the gradually decreasing flow into the wellbore The quantity1198622is obtained by trial and error as the value which makes the
curve straight at early times After obtaining the straight linesection the rest of the analysis is the same as for any otherpressure buildup This method has the advantage or requiresno additional data over that taken routinely during a shut intest
Due to the time consuming and unconfident of trial anderror to determine 119862
2in Russell method we modify this
method Russellrsquos wellbore storage correction is given as
[119901119908119904 (Δ119905) minus 119901
119908119891 (Δ119905 = 0)]
[1 minus (11198622Δ119905)]
= 119891 (Δ119905 = 1 hr) + 119898119904119897log (Δ119905)
(A2)
Abstract and Applied Analysis 9
Defining the following function and using least squarestechnique obtain the final result
119864 (1198622 119891119898)
=
119873
sum
119894=1
(Δ119901119908119904119894
(1 minus (11198622Δ119905119894))
minus 119891(1198622) minus 119898(119862
2) log(Δ119905
119894))
2
(A3)
Tominimize the function119864 it can differentiate119864with respectto 1198622 119891 and 119898 and setting the results equal to zero After
differentiation it reduces to
120597119864
1205971198622
= minus2
119873
sum
119894=1
(Δ119901119908119904119894
Δ119905119894(1198622minus (1Δ119905
119894))2)
times (Δ119901119908119904119894
(1 minus (11198622Δ119905119894))
minus 119891 (1198622)
minus119898 (1198622) log (Δ119905
119894) ) = 0
(A4)
120597119864
120597119891
= minus2
119873
sum
119894=1
(Δ119901119908119904119894
(1 minus (11198622Δ119905119894))
minus 119891 (1198622) minus 119898 (119862
2) log (Δ119905
119894))
= 0
(A5)120597119864
120597119898
= minus2
119873
sum
119894=1
log (Δ119905119894) (
Δ119901119908119904119894
(1 minus (11198622Δ119905119894))
minus 119891 (1198622)
minus 119898 (1198622) log (Δ119905
119894)) = 0
(A6)
From (A4) we define the following function
119883(1198622)
=
119873
sum
119894=1
(Δ119901119908119904119894
Δ119905119894(1198622minus (1Δ119905
119894))2)
times(Δ119901119908119904119894
(1minus (11198622Δ119905119894))minus119891 (119862
2) minus119898 (119862
2) log (Δ119905
119894))
= 0
(A7)
The roots of (A7) can be found via the following algorithm
1198622(119894+1)
= 1198622(119894)
minus119883 (1198622)
1198831015840 (1198622) (A8)
where
119883(1198622) =
119873
sum
119894=1
(Δ119901119908119904119894
Δ119905119894(1198622minus (1Δ119905
119894))2)
times (Δ119901119908119904119894
(1 minus (11198622Δ119905119894))
minus 119891 (1198622)
minus119898 (1198622) log (Δ119905
119894))
1198831015840(1198622) =
119873
sum
119894=1
(minus2Δ119901119908119904119894
Δ119905119894(1198622minus (1Δ119905
119894))3)
times (Δ119901119908119904119894
Δ119905119894(1198622minus (1Δ119905
119894))2
times [Δ119901119908119904119894
(1 minus (11198622Δ119905))
minus 119891 (1198622)
minus119898 (1198622) log (Δ119905
119894)])
+ (Δ119901119908119904119894
Δ119905119894(1198622minus (1Δ119905
119894))2)
times (minusΔ119901119908119904119894
Δ119905119894(1198622minus (1Δ119905
119894))3minus 1198911015840(1198622)
minus1198981015840(1198622) log (Δ119905
119894))
(A9)
where the unknown functions are given by the followingrelations
119891 (1198622) =
120573119906 (1198622) minus 120572V (119862
2)
1205732 minus 119873120572(A10)
whose derivative is
1198911015840(1198622) =
1205731199061015840(1198622) minus 120572V1015840 (119862
2)
1205732 minus 119873120572
119898 (1198622) =
119873119906 (1198622) minus 120573V (119862
2)
119873120572 minus 1205732
(A11)
where its derivative is
1198981015840(1198622) =
1198731199061015840(1198622) minus 120573V1015840 (119862
2)
119873120572 minus 1205732
119906 (1198622) =
119873
sum
119894=1
log (Δ119905119894)
Δ119901119908119904119894
(1 minus (11198622Δ119905119894))
(A12)
10 Abstract and Applied Analysis
and after differentiation it yields
1199061015840(1198622) = minus
119873
sum
119894=1
log (Δ119905
119894)
Δ119905119894
Δ119901119908119904119894
(1198622minus (1Δ119905
119894))2
V (1198622) =
119873
sum
119894=1
Δ119901119908119904119894
(1 minus (11198622Δ119905119894))
(A13)
which has the following derivative
V1015840 (1198622) = minus
119873
sum
119894=1
1
Δ119905119894
Δ119901119908119904119894
(1198622minus (1Δ119905
119894))2 (A14)
finally the values of 120572 and 120573 parameters are given by thefollowing formulas
120572 =
119873
sum
119894=1
(log(Δ119905119894))2
120573 =
119873
sum
119894=1
log (Δ119905119894)
(A15)
B General Formula for Inverse Problem
The convolution of (6) yields
119901119904119863
(119905119863) = int
119905119863
0
119870 (120591) 119901119908119863 (119905119863 minus 120591) 119889120591
= int
119905119863
0
119870(119905119863minus 120591) 119901
119908119863 (120591) 119889120591
(B1)
where
119870(119905119863) = 119871minus1(
1
119904119902 (119904)) (B2)
where 119871minus1 is the inverse Lapalce transform operator 119870(119905119863)
can be computed either from the Laplace transforms of119902119863(119905119863) data a curve-fitted equation of 119902
119863(119905119863) data or using
another technique with a relation between119870(119905119863) and 119902
119863(119905119863)
However it would be time consuming to invert all the 119902119863(119905119863)
data in Laplace space and transform it back to real space inaccordance with (B2) Once an approximation is obtained itwill be easy to compute119870(119905
119863) and integrate (B1) to determine
119901119904119863(119905119863) However finding some approximation functions for
119902119863(119905119863) data may be difficult We provide an approach to
overcome this problemThis idea is based on finding anotherrelation between 119870(119905
119863) and 119902
119863(119905119863) that is
ℓ (1) =1
119904=
1
119904119902119863(119904)
119902119863(119904) = ℓ (int
119905119863
0
119870 (120591) 119902119863 (119905119863 minus 120591) 119889120591)
= ℓ(int
119905119863
0
119870(119905119863minus 120591) 119902
119863 (120591) 119889120591)
(B3)or
1 = int
119905119863
0
119870 (120591) 119902119863 (119905119863 minus 120591) 119889120591 = int
119905119863
0
119870(119905119863minus 120591) 119902
119863 (120591) 119889120591
(B4)
The other relations that can be used are
ℓ (119905119863) =
1
1199042=
1
1199042119902119863(119904)
119902119863(119904) = ℓ (int
119905119863
0
119871 (120591) 119902119863 (119905119863 minus 120591) 119889120591)
= ℓ(int
119905119863
0
119871 (119905119863minus 120591) 119902
119863 (120591) 119889120591)
(B5)
or
119905119863= int
119905119863
0
119871 (120591) 119902119863(119905119863minus 120591) 119889120591 = int
119905119863
0
119871 (119905119863minus 120591) 119902
119863(120591) 119889120591
(B6)
ℓ (1) =1
119904=
1
1199042119902119863(119904)
119904119902119863(119904) = ℓ (int
119905119863
0
119871 (120591) 1199021015840
119863(119905119863minus 120591) 119889120591)
= ℓ(int
119905119863
0
119871 (119905119863minus 120591) 119902
1015840
119863(120591) 119889120591)
(B7)
or
1 = int
119905119863
0
119871 (120591) 1199021015840
119863(119905119863minus 120591) 119889120591 = int
119905119863
0
119871 (119905119863minus 120591) 119902
1015840
119863(120591) 119889120591
(B8)
where
119871 (119905119863) = ℓminus1(
1
1199042119902 (119904)) (B9)
Equation (B4) can be used simultaneously with (B1) toanalyze the deconvolution problem
Equation (11) can be written as
119901119904119863
(119904) = 119904119901119908119863
(119904)1
1199042119902119863(119904)
(B10)
or
119901119904119863
(119905119863) = int
119905119863
0
119871 (120591) 1199011015840
119908119863(119905119863minus 120591) 119889120591
= int
119905119863
0
119871 (119905119863minus 120591) 119901
1015840
119908119863(120591) 119889120591
(B11)
The applications of (B6) and (B8) are with (B11)For justifying the rate normalizationmethod andmaterial
balance deconvolution it can use (B1) (B4) ((B11) can alsobe used with (B6) or (B8) for this goal) and generalizedmean value theorem for integrals This theorem is expressedas follows
Theorem 1 Let 119891 and 119892 be continuous on [119886 119887] and supposethat 119892 does not change sign on [119886 119887] Then there is a point 119888 in[119886 119887] such that
int
119887
119886
119891 (119905) 119892 (119905) 119889119905 = 119891 (119888) int
119887
119886
119892 (119905) 119889119905 (B12)
Abstract and Applied Analysis 11
By Theorem 1 there are some values 1198881and 1198882between 0
and 119905119863such that
119901119904119863
(119905119863) = 119901119908119863
(1198881) int
119905119863
0
119870(119905119863minus 120591) 119889120591 (B13)
1 = 119902119863(1198882) int
119905119863
0
119870(119905119863minus 120591) 119889120591 (B14)
By setting 1198881
= 1198882
= 119905119863and dividing the right and left
side of (B13) respectively the rate normalization techniqueis developed However the values 119888
1and 1198882are not equal
to each other and are not equal to 119905119863
in general Thisis the reason of error in rate normalization and materialbalance deconvolution methods 119888
1and 1198882are less than 119905
119863in
these methods for the duration of wellbore storage distortionduring a pressure transient test Since these conditions areapproximately satisfied in material balance deconvolutionthe results of material balance are more accurate and moreconfident than rate normalization
Nomenclature
API API gravity ∘API119861119900 Oil formation volume factor RBSTB
119888119905 Total isothermal compressibility factor
psiminus1
1198622 Arbitrary constant hr
minus1
119862119863 Wellbore storage constant dimensionless
119862119904 Wellbore storage coefficient bblpsi
ℎ Formation thickness ftHTR Horner time ratio dimensionless119896 Formation permeability md119870 Kernel of the convolution integral119871 Kernel of the convolution integral119873119901 Cumulative oil production bbl
119901 Pressure psi119901119886 Normalized pseudopressure psi
119901 Average reservoir pressure psi119901119863 Formation pressure dimensionless
119901119889119901 Dewpoint pressure psi
119901119894 Initial reservoir pressure psi
119901119904119863 Formation pressure including skin
dimensionless119901119904 Formation pressure including skin psi
119901119908119863
Pressure draw down dimensionless119901119908119891 Flowing bottom hole pressure psi
119901119908119904 Shut-in bottom hole pressure psi
119902119863 Sandface flow rate dimensionless
119902119892 Gas flow rate MscfD
119902119900 Oil flow rate STBD
119903119908 Wellbore radius ft
119904 Laplace variable119878 Skin factor119905 Time hr119905119863 Time dimensionless
119905119901 Production time hr
119879 Reservoir temperature ∘F
119879119889119901 Dewpoint temperature ∘F
120572 Explicit deconvolution variabledimensionless
120573 Beta deconvolution variabledimensionless
120574119892 Gas gravity (air = 10)
120582 Interporosity flow coefficient of dualporosity reservoir
120601 Porosity of reservoir rock dimensionless120583119892 Gas viscosity cp
120583119900 Oil viscosity cp
120596 Storativity ratio of dual porosity reservoirminus Laplace transform of a function1015840 Derivative of a function
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
References
[1] Y Cheng W J Lee and D A McVay ldquoFast-Fourier-transform-based deconvolution for interpretation of pressure-transient-test data dominated by Wellbore storagerdquo SPE Reservoir Eval-uation and Engineering vol 8 no 3 pp 224ndash239 2005
[2] M M Levitan and M R Wilson ldquoDeconvolution of pressureand rate data from gas reservoirs with significant pressuredepletionrdquo SPE Journal vol 17 no 3 pp 727ndash741 2012
[3] Y Liu and R N Horne ldquoInterpreting pressure and flow-ratedata from permanent downhole gauges by use of data-miningapproachesrdquo SPE Journal vol 18 no 1 pp 69ndash82 2013
[4] O Ogunrewo and A C Gringarten ldquoDeconvolution of welltest data in lean and rich gas condensate and volatile oil wellsbelow saturation pressurerdquo in Proceedings of the SPE AnnualTechnical Conference and Exhibition Document ID 166340-MS New Orleans La USA September-October 2013
[5] R C Earlougher Jr Advances in Well Test Analysis vol 5 ofMonogragh Series SPE Richardson Tex USA 1977
[6] F Kucuk and L Ayestaran ldquoAnalysis of simultaneously mea-sured pressure and sandface flow rate in transient well testingrdquoJournal of PetroleumTechnology vol 37 no 2 pp 323ndash334 1985
[7] M Onur M Cinar D IIk P P Valko T A Blasingame and PHegeman ldquoAn investigation of recent deconvolution methodsfor well-test data analysisrdquo SPE Journal vol 13 no 2 pp 226ndash247 2008
[8] F J KuchukMOnur andFHollaender ldquoPressure transient testdesign and interpretationrdquo Developments in Petroleum Sciencevol 57 pp 303ndash359 2010
[9] A C Gringarten D P Bourdet P A Landel and V J KniazeffldquoA comparison between different skin and wellbore storagetype-curves for early-time transient analysisrdquo in Proceedings ofthe SPE Annual Technical Conference and Exhibition Paper SPE8205 Las Vegas Nev USA 1979
[10] A C Gringarten ldquoFrom straight lines to deconvolution theevolution of the state of the art in well test analysisrdquo SPEReservoir Evaluation and Engineering vol 11 no 1 pp 41ndash622008
[11] J A Cumming D A Wooff T Whittle and A C GringartenldquoMultiple well deconvolutionrdquo in Proceedings of the SPE Annual
12 Abstract and Applied Analysis
Technical Conference and Exhibition Document ID 166458-MS New Orleans La USA September-October 2013
[12] O Bahabanian D Ilk N Hosseinpour-Zonoozi and T ABlasingame ldquoExplicit deconvolution of well test data distortedby wellbore storagerdquo in Proceedings of the SPE Annual TechnicalConference and Exhibition Paper SPE 103216 pp 4444ndash4453San Antonio Tex USA September 2006
[13] J Jones and A Chen ldquoSuccessful applications of pressure-ratedeconvolution in theCad-Nik tight gas formations of the BritishColumbia foothillsrdquo Journal of Canadian Petroleum Technologyvol 51 no 3 pp 176ndash192 2012
[14] D G Russell ldquoExtension of pressure buildup analysis methodsrdquoJournal of Petroleum Technology vol 18 no 12 pp 1624ndash16361966
[15] R E Gladfelter G W Tracy and L E Wilsey ldquoSelecting wellswhich will respond to production-stimulation treatmentrdquo inDrilling and Production Practice API Dallas Tex USA 1955
[16] M J Fetkovich and M E Vienot ldquoRate normalization ofbuildup pressure by using afterflow datardquo Journal of PetroleumTechnology vol 36 no 13 pp 2211ndash2224 1984
[17] J L Johnston Variable rate analysis of transient well testdata using semi-analytical methods [MS thesis] Texas AampMUniversity College Station Tex USA 1992
[18] A F van Everdingen ldquoThe skin effect and its influence on theproductive capacity of a wellrdquo Journal of Petroleum Technologyvol 5 no 6 pp 171ndash176 1953
[19] W Hurst ldquoEstablishment of the skin effect and its impedimentto fluid flow into a well borerdquo The Petroleum Engineer pp B6ndashB16 1953
[20] J A Joseph and L F Koederitz ldquoA simple nonlinear model forrepresentation of field transient responsesrdquo Paper SPE 11435SPE Richardson Tex USA 1982
[21] F J Kuchuk ldquoGladfelter deconvolutionrdquo SPE Formation Evalu-ation vol 5 no 3 pp 285ndash292 1990
[22] A F van Everdingen and W Hurst ldquoThe application of thelaplace transformation to flow problems in reservoirsrdquo Journalof Petroleum Technology vol 1 no 12 pp 305ndash324 1949
[23] D Bourdet T M Whittle A A Douglas and Y M Pirard ldquoAnew set of type curves simplifies well test analysisrdquo World Oilpp 95ndash106 1983
[24] D F Meunier C S Kabir and M J Wittmann ldquoGas well testanalysis use of normalized pseudovariablesrdquo SPE FormationEvaluation vol 2 no 4 pp 629ndash636 1987
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
4 Abstract and Applied Analysis
Using elementary calculus yields
ln (1003816100381610038161003816119891(120591)
1003816100381610038161003816)1003816100381610038161003816119905119863
0= ln (1003816100381610038161003816119891 (119905
119863)1003816100381610038161003816) minus ln (1) = int
119905119863
0
11990110158401015840
119908119863(120591)
1199011015840119908119863
(120591)119889120591
(28)
So
119891 (119905119863) = 119890120572(119905119863) (29)
where
120572 (119905119863) = int
119905119863
0
11990110158401015840
119908119863(120591)
1199011015840119908119863
(120591)119889120591 (30)
Now the explicit deconvolution of data based on (20) can bewritten as follows
119901119904119863
(119905119863) = minus
1
120572 (119905119863) 119905119863
1199011015840
119908119863+ 119901119908119863
(119905119863) (31)
or
119901119904119863
(119905119863) = minus
119905119863
120572 (119905119863)1199011015840
119908119863+ 119901119908119863
(119905119863) (32)
that is similar to (19) The other similarity between (19) and(32) is that the plot of (32) versus (int119905119863
0119902119863(120591)119889120591119902
119863(119905119863)) and
(int119905119863
0(1 minus 119902
119863(120591))119889120591(1 minus 119902
119863(119905119863))) for drawdown and buildup
tests respectively yields the more accurate results and isusable for practical applications The applicability of (32) isalso for duration of wellbore storage effects And for the othertimes the given pressure data are correct However using thismethod for the later times yields not bad results
3 Material Balance Deconvolution
Material balance deconvolution is an extension of the ratenormalization method Johnston [17] defines a new 119909-axisplotting function (material balance time) that provides anapproximate deconvolution of the variable rate pressuretransient problem There are numerous assumptions associ-ated with the material balance deconvolution methods oneof the most widely accepted assumptions is that the rateprofilemust change smoothly andmonotonically In practicalterms this condition should be met for the wellbore storageproblem
The general form of material balance deconvolution isprovided for the pressure drawdown case in terms of thematerial balance time function and the rate-normalizedpressure drop function The material balance time functionis given as
119905119898119887
=119873119901
119902119863 (119905)
(33)
where
119873119901= int
119905
0
119902119863 (120591) 119889120591 (34)
The rate-normalized pressure drop function is given by
Δ119901119908119891 (119905)
119902119863 (119905)
=119901119894minus 119901119908119891 (119905)
119902119863 (119905)
(35)
The material balance time function for the pressure buildupcase is given as
119905119898119887
=119873119901
1 minus 119902119863 (Δ119905)
(36)
where
119873119901= int
Δ119905
0
(1 minus 119902119863 (120591)) 119889120591 (37)
The rate-normalized pressure drop function for the pressurebuildup case is given as
Δ119901119908119904 (Δ119905)
1 minus 119902119863 (Δ119905)
=119901119908119904 (Δ119905) minus 119901
119908119891 (Δ119905 = 0)
1 minus 119902119863 (Δ119905)
(38)
4 Methodology
In this work several methods are provided to restore thepressure datawithout any sandface flow rate informationThevalues of sandface flow rate data for all methods that requirethese data may be obtained from (20) We evaluated a veryold correctionmethod by Russell [14] However we modifiedthis method by least square method but it was found thatthis method is to be unacceptable for all applications Thederivation of this method and its modification is given inAppendix A
The 120573-deconvolution and its modification are anothertechnique that is mentioned in this study The formulationof this method is presented in Section 21 The other explicitdeconvolutionmethod is an approach similar to themodified120573-deconvolution that uses only the pressure data for decon-volution of pressure and rate data The formulation of thismethod was derived in Section 22 The 120573-deconvolution itsmodification and explicit deconvolution formula can be usedfor practical applications The applicability of these methodsis for duration of wellbore storage effects However usingthese approaches for the other times yields not bad results
The other technique is material balance deconvolutionmethod [17] that has sufficiently accurate resultsThismethodcan also be used as a good approach for practical applicationsThe formulas for this method were given in Section 3
In Appendix B deconvolution is expressed as an inverseproblem that is analyzed by a tricky idea Based on thisidea an interesting proof is given for rate normalizationmethod and material balance deconvolution that is a goodapproach to find the errors and limitations of these methodsand overcome them
Abstract and Applied Analysis 5
10minus3 10minus2 10minus1 100 101 102 1030
500
1000
1500
2000
2500
3000
3500
4000
Time (hr)
Dpwf wellbore storaged(Dpwf)d(ln(t)) wellbore storageDps explicit deconvolution methodd(Dps)d(ln(t)) explicit deconvolution methodDps material balance deconvolution methodd(Dps)d(ln(t)) material balance deconvolution methodDps no wellbore storaged(Dps)d(ln(t)) no wellbore storage
Dpwf
and
its lo
garit
hmic
der
ivat
ive (
psi)
Figure 1 Deconvolution for a drawdown test in a homogeneousreservoir
From the above-mentioned methods we use the fol-lowing two techniques to show the verification of theseapproaches
Explicit Deconvolution Formula The final result of explicitdeconvolution method is given as follows
119901119904119863
(119905119863) = minus
119905119863
120572 (119905119863)1199011015840
119908119863+ 119901119908119863
(119905119863) (39)
Plotting of this term versus (int119905119863
0119902119863(120591)119889120591119902
119863(119905119863)) and
(int119905119863
0(1 minus 119902
119863(120591))119889120591(1 minus 119902
119863(119905119863))) for drawdown and buildup
tests respectively yields more accurate resultsThis approachperforms well in field applications
Material Balance Deconvolution The material balance pres-sure drop function for the pressure drawdown test is givenas
Δ119901119908119891 (119905)
119902119863 (119905)
=119901119894minus 119901119908119891 (119905)
119902119863 (119905)
(40)
and the material balance time function for the pressuredrawdown test is given as
119905119898119887
=int119905119863
0119902119863 (120591) 119889120591
119902119863 (119905)
(41)
For the buildup test the material balance pressure dropfunction can be written as
Δ119901119908119904 (Δ119905)
1 minus 119902119863 (Δ119905)
=119901119908119904 (Δ119905) minus 119901
119908119891 (Δ119905 = 0)
1 minus 119902119863 (Δ119905)
(42)
10minus3 10minus2 10minus1 100 101 102 103
Time (hr)
104
103
102
101Dpwf
and
its lo
garit
hmic
der
ivat
ive (
psi)
Dpwf wellbore storaged(Dpwf)d(ln(t)) wellbore storageDps explicit deconvolution methodd(Dps)d(ln(t)) explicit deconvolution methodDps material balance deconvolution methodd(Dps)d(ln(t)) material balance deconvolution methodDps no wellbore storaged(Dps)d(ln(t)) no wellbore storage
Figure 2 Deconvolution for a drawdown test in a homogeneousreservoir
And the material balance time function can be written as
119905119898119887
=int119905119863
0(1 minus 119902
119863 (120591)) 119889120591
1 minus 119902119863 (Δ119905)
(43)
This technique is a good approach for practical applications
5 Verification Examples
We first use few synthetic cases for different reser-voirwellbore systems to validate the applicability of explicitdeconvolution formula and material balance deconvolutionmethod Then the field applications of these techniqueswill be shown by two real cases The unloading rates arecalculated with (20)
The first case is a synthetic test derived from a test designin a vertical oil wellThe reservoir is homogenous and infiniteacting with a constant wellbore storage The wellbore storagecoefficient is 119862
119904= 0015 bblpsi The pressure data are
obtained from the cylindrical source solution (test design)We used the explicit deconvolution formula and material
balance deconvolution method to deconvolve the pressureand unloading rate distorting by wellbore storage effectsTheresults are presented in Figures 1 and 2 for the semilog andlog-log plots in the case of drawdown test and Figures 3and 4 for the semilog and log-log plots in the case of builduptest Besides the deconvolved pressures and their logarithmicderivatives pressure responses with and without wellborestorage are also plotted on the figures for comparisonThrough the deconvolution process we have successfullyremoved the wellbore storage effects so that radial flow canbe identified at early test times The deconvolved pressuresand their logarithmic derivatives are almost identical to
6 Abstract and Applied Analysis
0
500
1000
1500
2000
2500
3000
3500
4000
105104103102101100
Horner time ratio
Dp wellbore storage
Dps explicit deconvolution method(ln
Dps material balance deconvolution
Dps no wellbore storage
ws
and
its lo
garit
hmic
der
ivat
ive (
psi)
Dpws
d(Dp )d(ln )) wellbore storagews (HTR
d(Dps)d )) explicit deconvolution method(HTR
d(Dps)d(ln )) no wellbore storage(HTR)
d(Dps)d(ln )) material balance deconvolution (HTR
Figure 3 Deconvolution for a buildup test in a homogeneousreservoir
10minus3 10minus2 10minus1 100 101 102 103
Time (hr)
104
103
102
101
ws
ws
and
its lo
garit
hmic
der
ivat
ive (
psi)
Dpws
Dp wellbore storage
Dps explicit deconvolution method
Dps material balance deconvolution
Dps no wellbore storage
d(Dp )d(ln )) wellbore storagews (HTR
d(Dps)d )) explicit deconvolution method(HTR
d(Dps)d(ln )) no wellbore storage(HTR)
d(Dps)d(ln )) material balance deconvolution (HTR
(ln
Figure 4 Deconvolution for a buildup test in a homogeneousreservoir
the analytic solutions (without wellbore storage) whichenables us to estimate reservoir parameters accurately
The second case is an oil well drawdown test in adual porosity reservoir The storativity ratio (120596) and inter-porosity flow coefficient (120582) are assumed to be 01 and10minus6 respectively The constant wellbore storage coefficient
is 119862119904= 015 bblpsi The dimensionless wellbore pressure
10minus3 10minus2 10minus1 100 101 102 103
Time (hr)
600
500
400
300
200
100
0
Dpwf wellbore storaged(Dpwf)d(ln(t)) wellbore storageDps explicit deconvolution methodd(Dps)d(ln(t)) explicit deconvolution methodDps material balance deconvolutiond(Dps)d(ln(t)) material balance deconvolutionDps no wellbore storaged(Dps)d(ln(t)) no wellbore storage
Dpwf
and
its lo
garit
hmic
der
ivat
ive (
psi)
Figure 5 Deconvolution for a drawdown test in a dual porosityreservoir
data were obtained from the cylindrical source solution withpseudosteady-state interporosity flow (test design) It is notpossible to identify the dual porosity behavior characteristicof the reservoir from wellbore pressure data because thisbehavior has been masked completely by wellbore storage asindicated in Figures 5 and 6
We deconvolved the pressure and unloading rate andpresent the results in Figures 5 and 6 The deconvolvedresults are reasonably consistent with the analytical solutionAfter deconvolution the dual porosity behavior can beidentified easily from the valley in the logarithmic derivativeof pressure If wellbore storage effects were not removed atearly times it easily could have incorrectly identified thereservoir system as infinite actingThe estimation of reservoirparameters can be obtained now from deconvolved pressure
The third case is an oil well drawdown test in a reservoirwith a sealing fault 250 ft from the well Synthetic pressureresponses were generated by simulation (test design) with aconstant wellbore storage coefficient of 01 bblpsi As we cansee in Figures 7 and 8 the early time infinite acting reservoirbehavior has been masked completely by wellbore storage
To recover the hidden reservoir behavior we performeddeconvolution on the pressure change to remove the wellborestorage effect As in Figures 7 and 8 the deconvolved pressurechange and logarithmic derivative data are almost identicalto the responses without wellbore storageThe deconvolutionresults clearly indicate parallel horizontal lines for the loga-rithmic derivative However without wellbore storage effectselimination this maymislead to identify the reservoir systemas infinite acting So this allows us to correctly identify thereservoir system having a sealing fault
The fourth case is a field example that is taken from theliterature [23] The basic reservoir and fluid properties are
Abstract and Applied Analysis 7
Table 1 Reservoir rock and fluid properties for the first field case
119905119901(hr) ℎ (ft) 119903
119908(ft) 119902
119900(STBD) 119861
119900(RBSTB) 119888
119905(psiminus1) 120583
119900(cp) 120601
1533 107 029 174 106 42 times 10minus6 25 025
Table 2 Reservoir rock and fluid properties for the second field case
119905119901(hr) ℎ (ft) 119903
119908(ft) 119879 (∘F) 119901 (psi) 119902
119900(STBD) 119902
119892(MscfD) 119888
119891(psiminus1) 119879
119889119901(∘F) 119901
119889119901(psi) ∘API 120574
119892
143111 100 0354 2795 7322 104 13925 3 times 10minus6 212 5445 45375 068
10minus3 10minus2 10minus1 100 101 102 103
Time (hr)
103
102
101
100
10minus1
Dpwf wellbore storaged(Dpwf)d(ln(t)) wellbore storageDps explicit deconvolution methodd(Dps)d(ln(t)) explicit deconvolution methodDps material balance deconvolutiond(Dps)d(ln(t)) material balance deconvolutionDps no wellbore storaged(Dps)d(ln(t)) no wellbore storage
Dpwf
and
its lo
garit
hmic
der
ivat
ive (
psi)
Figure 6 Deconvolution for a drawdown test in a dual porosityreservoir
shown in Table 1 In this case we provide the deconvolvedpressure data using the explicit deconvolution formula andthe material balance deconvolution method that is presentedin this work The data are taken from a pressure buildup testand can be reasonably used as field data The deconvolutionresults are shown in Figure 9 in semilog plot As it isillustrated from Figure 9 the corrected pressure provides aclear straight line starting as early as 15 hours in contrast tothe vague line from the uncorrected pressure response
And the last (field) example is a gas condensatewell buildup test following a variable rate productionhistory The duration of last pressure buildup test is361556 hours (from 1889 to 2250556 hours) The basicreservoir and fluid properties for the last drawdown testare shown in Table 2 However the properties of otherproduction tests are not shown here To use our newtechniques effectively for gas wells we use normalizedpseudopressure and pseudotime functions as defined byMeunier et al [24]The final results of explicit deconvolutionformula and material balance deconvolution methodshow two different slopes (Figure 10) where the earlytimes slope is greater than the last one This identifiesthe radial composite behavior of the reservoir With
the corrected pressures the reservoir parameters can also beestimated
6 Conclusions
In this work we have provided some explicit deconvolutionmethods for restoration of constant rate pressure responsesfrom measured pressure data dominated by wellbore storageeffects without sandface rate measurements Based on ourinvestigations we can state the following specific conclusionsrelated to these techniques
(1) The results obtained from the synthetic and fieldexamples show the practical and computational effi-ciency of our deconvolution methods
(2) Modification of 120573-deconvolution is presented wherethe beta quantity is variable rather than a constant
(3) For situations in which there are no downhole ratemeasurements the explicit deconvolution methods(blind deconvolution approaches) can be used toderive the downhole rate function from the pressuremeasurements and restore the reservoir pressureresponse
(4) We can unveil the early time behavior of a reservoirsystemmasked by wellbore storage effects using thesenew approaches Wellbore geometries and reservoirtypes can be identified from this restored pressuredata Furthermore the conventional methods canbe used to analyze this restored formation pressureto estimate formation parameters So the presentedtechniques can be used as the practical tools toimprove pressure transient test interpretation
(5) Deconvolution in well test analysis is often expressedas an inverse problem This approach can be used asan idea to justify the rate normalization method andthe material balance deconvolution method
Appendices
A Russel Method and Its Modification
A method for correcting pressure buildup pressure data freeof wellbore storage effects at early times has been developedby Russell [14] In this method it is not necessary to measurethe rate of influx after closing in at the surface Instead oneuses a theoretical equation which gives the form that thebottom hole pressure should have fluid accumulates in the
8 Abstract and Applied Analysis
10minus3 10minus2 10minus1 100 101 102 103
Time (hr)
250
200
150
100
50
0
Dpwf wellbore storaged(Dpwf)d(ln(t)) wellbore storageDps explicit deconvolution methodd(Dps)d(ln(t)) explicit deconvolution methodDps material balance deconvolutiond(Dps)d(ln(t)) material balance deconvolutionDps no wellbore storaged(Dps)d(ln(t)) no wellbore storage
Dpwf
and
its lo
garit
hmic
der
ivat
ive (
psi)
Figure 7 Deconvolution for a drawdown test in a reservoir with asealing fault
10minus3 10minus2 10minus1 100 101 102 103
Time (hr)
103
102
101
100
Dpwf wellbore storaged(Dpwf)d(ln(t)) wellbore storageDps explicit deconvolution methodd(Dps)d(ln(t)) explicit deconvolution methodDps material balance deconvolutiond(Dps)d(ln(t)) material balance deconvolutionDps no wellbore storaged(Dps)d(ln(t)) no wellbore storage
Dpwf
and
its lo
garit
hmic
der
ivat
ive (
psi)
Figure 8 Deconvolution for a drawdown test in a reservoir with asealing fault
wellbore during buildup This leads to the result that oneshould plot
[119901119908119904 (Δ119905) minus 119901
119908119891 (Δ119905 = 0)]
[1 minus (11198622Δ119905)]
(A1)
versus log (Δ119905) in analyzing pressure buildup data during theearly fill-up period The denominator makes a correction for
104103102101100
Horner time ratio
800
700
600
500
400
300
200
100
0
Dpws
(psi)
Dpws wellbore storageDps explicit deconvolution methodDps material balance deconvolution
Figure 9 Restored pressures for a buildup test in a homogeneousreservoir
0
500
1000
1500
2000
2500
3000
3500
4000
105104103102101100
Horner time ratio
Dpa
(psi)
Dpaw wellbore storage explicit deconvolution method material balance deconvolution
DpasDpas
Figure 10 Restored pressures for a buildup test in a radial compositereservoir
the gradually decreasing flow into the wellbore The quantity1198622is obtained by trial and error as the value which makes the
curve straight at early times After obtaining the straight linesection the rest of the analysis is the same as for any otherpressure buildup This method has the advantage or requiresno additional data over that taken routinely during a shut intest
Due to the time consuming and unconfident of trial anderror to determine 119862
2in Russell method we modify this
method Russellrsquos wellbore storage correction is given as
[119901119908119904 (Δ119905) minus 119901
119908119891 (Δ119905 = 0)]
[1 minus (11198622Δ119905)]
= 119891 (Δ119905 = 1 hr) + 119898119904119897log (Δ119905)
(A2)
Abstract and Applied Analysis 9
Defining the following function and using least squarestechnique obtain the final result
119864 (1198622 119891119898)
=
119873
sum
119894=1
(Δ119901119908119904119894
(1 minus (11198622Δ119905119894))
minus 119891(1198622) minus 119898(119862
2) log(Δ119905
119894))
2
(A3)
Tominimize the function119864 it can differentiate119864with respectto 1198622 119891 and 119898 and setting the results equal to zero After
differentiation it reduces to
120597119864
1205971198622
= minus2
119873
sum
119894=1
(Δ119901119908119904119894
Δ119905119894(1198622minus (1Δ119905
119894))2)
times (Δ119901119908119904119894
(1 minus (11198622Δ119905119894))
minus 119891 (1198622)
minus119898 (1198622) log (Δ119905
119894) ) = 0
(A4)
120597119864
120597119891
= minus2
119873
sum
119894=1
(Δ119901119908119904119894
(1 minus (11198622Δ119905119894))
minus 119891 (1198622) minus 119898 (119862
2) log (Δ119905
119894))
= 0
(A5)120597119864
120597119898
= minus2
119873
sum
119894=1
log (Δ119905119894) (
Δ119901119908119904119894
(1 minus (11198622Δ119905119894))
minus 119891 (1198622)
minus 119898 (1198622) log (Δ119905
119894)) = 0
(A6)
From (A4) we define the following function
119883(1198622)
=
119873
sum
119894=1
(Δ119901119908119904119894
Δ119905119894(1198622minus (1Δ119905
119894))2)
times(Δ119901119908119904119894
(1minus (11198622Δ119905119894))minus119891 (119862
2) minus119898 (119862
2) log (Δ119905
119894))
= 0
(A7)
The roots of (A7) can be found via the following algorithm
1198622(119894+1)
= 1198622(119894)
minus119883 (1198622)
1198831015840 (1198622) (A8)
where
119883(1198622) =
119873
sum
119894=1
(Δ119901119908119904119894
Δ119905119894(1198622minus (1Δ119905
119894))2)
times (Δ119901119908119904119894
(1 minus (11198622Δ119905119894))
minus 119891 (1198622)
minus119898 (1198622) log (Δ119905
119894))
1198831015840(1198622) =
119873
sum
119894=1
(minus2Δ119901119908119904119894
Δ119905119894(1198622minus (1Δ119905
119894))3)
times (Δ119901119908119904119894
Δ119905119894(1198622minus (1Δ119905
119894))2
times [Δ119901119908119904119894
(1 minus (11198622Δ119905))
minus 119891 (1198622)
minus119898 (1198622) log (Δ119905
119894)])
+ (Δ119901119908119904119894
Δ119905119894(1198622minus (1Δ119905
119894))2)
times (minusΔ119901119908119904119894
Δ119905119894(1198622minus (1Δ119905
119894))3minus 1198911015840(1198622)
minus1198981015840(1198622) log (Δ119905
119894))
(A9)
where the unknown functions are given by the followingrelations
119891 (1198622) =
120573119906 (1198622) minus 120572V (119862
2)
1205732 minus 119873120572(A10)
whose derivative is
1198911015840(1198622) =
1205731199061015840(1198622) minus 120572V1015840 (119862
2)
1205732 minus 119873120572
119898 (1198622) =
119873119906 (1198622) minus 120573V (119862
2)
119873120572 minus 1205732
(A11)
where its derivative is
1198981015840(1198622) =
1198731199061015840(1198622) minus 120573V1015840 (119862
2)
119873120572 minus 1205732
119906 (1198622) =
119873
sum
119894=1
log (Δ119905119894)
Δ119901119908119904119894
(1 minus (11198622Δ119905119894))
(A12)
10 Abstract and Applied Analysis
and after differentiation it yields
1199061015840(1198622) = minus
119873
sum
119894=1
log (Δ119905
119894)
Δ119905119894
Δ119901119908119904119894
(1198622minus (1Δ119905
119894))2
V (1198622) =
119873
sum
119894=1
Δ119901119908119904119894
(1 minus (11198622Δ119905119894))
(A13)
which has the following derivative
V1015840 (1198622) = minus
119873
sum
119894=1
1
Δ119905119894
Δ119901119908119904119894
(1198622minus (1Δ119905
119894))2 (A14)
finally the values of 120572 and 120573 parameters are given by thefollowing formulas
120572 =
119873
sum
119894=1
(log(Δ119905119894))2
120573 =
119873
sum
119894=1
log (Δ119905119894)
(A15)
B General Formula for Inverse Problem
The convolution of (6) yields
119901119904119863
(119905119863) = int
119905119863
0
119870 (120591) 119901119908119863 (119905119863 minus 120591) 119889120591
= int
119905119863
0
119870(119905119863minus 120591) 119901
119908119863 (120591) 119889120591
(B1)
where
119870(119905119863) = 119871minus1(
1
119904119902 (119904)) (B2)
where 119871minus1 is the inverse Lapalce transform operator 119870(119905119863)
can be computed either from the Laplace transforms of119902119863(119905119863) data a curve-fitted equation of 119902
119863(119905119863) data or using
another technique with a relation between119870(119905119863) and 119902
119863(119905119863)
However it would be time consuming to invert all the 119902119863(119905119863)
data in Laplace space and transform it back to real space inaccordance with (B2) Once an approximation is obtained itwill be easy to compute119870(119905
119863) and integrate (B1) to determine
119901119904119863(119905119863) However finding some approximation functions for
119902119863(119905119863) data may be difficult We provide an approach to
overcome this problemThis idea is based on finding anotherrelation between 119870(119905
119863) and 119902
119863(119905119863) that is
ℓ (1) =1
119904=
1
119904119902119863(119904)
119902119863(119904) = ℓ (int
119905119863
0
119870 (120591) 119902119863 (119905119863 minus 120591) 119889120591)
= ℓ(int
119905119863
0
119870(119905119863minus 120591) 119902
119863 (120591) 119889120591)
(B3)or
1 = int
119905119863
0
119870 (120591) 119902119863 (119905119863 minus 120591) 119889120591 = int
119905119863
0
119870(119905119863minus 120591) 119902
119863 (120591) 119889120591
(B4)
The other relations that can be used are
ℓ (119905119863) =
1
1199042=
1
1199042119902119863(119904)
119902119863(119904) = ℓ (int
119905119863
0
119871 (120591) 119902119863 (119905119863 minus 120591) 119889120591)
= ℓ(int
119905119863
0
119871 (119905119863minus 120591) 119902
119863 (120591) 119889120591)
(B5)
or
119905119863= int
119905119863
0
119871 (120591) 119902119863(119905119863minus 120591) 119889120591 = int
119905119863
0
119871 (119905119863minus 120591) 119902
119863(120591) 119889120591
(B6)
ℓ (1) =1
119904=
1
1199042119902119863(119904)
119904119902119863(119904) = ℓ (int
119905119863
0
119871 (120591) 1199021015840
119863(119905119863minus 120591) 119889120591)
= ℓ(int
119905119863
0
119871 (119905119863minus 120591) 119902
1015840
119863(120591) 119889120591)
(B7)
or
1 = int
119905119863
0
119871 (120591) 1199021015840
119863(119905119863minus 120591) 119889120591 = int
119905119863
0
119871 (119905119863minus 120591) 119902
1015840
119863(120591) 119889120591
(B8)
where
119871 (119905119863) = ℓminus1(
1
1199042119902 (119904)) (B9)
Equation (B4) can be used simultaneously with (B1) toanalyze the deconvolution problem
Equation (11) can be written as
119901119904119863
(119904) = 119904119901119908119863
(119904)1
1199042119902119863(119904)
(B10)
or
119901119904119863
(119905119863) = int
119905119863
0
119871 (120591) 1199011015840
119908119863(119905119863minus 120591) 119889120591
= int
119905119863
0
119871 (119905119863minus 120591) 119901
1015840
119908119863(120591) 119889120591
(B11)
The applications of (B6) and (B8) are with (B11)For justifying the rate normalizationmethod andmaterial
balance deconvolution it can use (B1) (B4) ((B11) can alsobe used with (B6) or (B8) for this goal) and generalizedmean value theorem for integrals This theorem is expressedas follows
Theorem 1 Let 119891 and 119892 be continuous on [119886 119887] and supposethat 119892 does not change sign on [119886 119887] Then there is a point 119888 in[119886 119887] such that
int
119887
119886
119891 (119905) 119892 (119905) 119889119905 = 119891 (119888) int
119887
119886
119892 (119905) 119889119905 (B12)
Abstract and Applied Analysis 11
By Theorem 1 there are some values 1198881and 1198882between 0
and 119905119863such that
119901119904119863
(119905119863) = 119901119908119863
(1198881) int
119905119863
0
119870(119905119863minus 120591) 119889120591 (B13)
1 = 119902119863(1198882) int
119905119863
0
119870(119905119863minus 120591) 119889120591 (B14)
By setting 1198881
= 1198882
= 119905119863and dividing the right and left
side of (B13) respectively the rate normalization techniqueis developed However the values 119888
1and 1198882are not equal
to each other and are not equal to 119905119863
in general Thisis the reason of error in rate normalization and materialbalance deconvolution methods 119888
1and 1198882are less than 119905
119863in
these methods for the duration of wellbore storage distortionduring a pressure transient test Since these conditions areapproximately satisfied in material balance deconvolutionthe results of material balance are more accurate and moreconfident than rate normalization
Nomenclature
API API gravity ∘API119861119900 Oil formation volume factor RBSTB
119888119905 Total isothermal compressibility factor
psiminus1
1198622 Arbitrary constant hr
minus1
119862119863 Wellbore storage constant dimensionless
119862119904 Wellbore storage coefficient bblpsi
ℎ Formation thickness ftHTR Horner time ratio dimensionless119896 Formation permeability md119870 Kernel of the convolution integral119871 Kernel of the convolution integral119873119901 Cumulative oil production bbl
119901 Pressure psi119901119886 Normalized pseudopressure psi
119901 Average reservoir pressure psi119901119863 Formation pressure dimensionless
119901119889119901 Dewpoint pressure psi
119901119894 Initial reservoir pressure psi
119901119904119863 Formation pressure including skin
dimensionless119901119904 Formation pressure including skin psi
119901119908119863
Pressure draw down dimensionless119901119908119891 Flowing bottom hole pressure psi
119901119908119904 Shut-in bottom hole pressure psi
119902119863 Sandface flow rate dimensionless
119902119892 Gas flow rate MscfD
119902119900 Oil flow rate STBD
119903119908 Wellbore radius ft
119904 Laplace variable119878 Skin factor119905 Time hr119905119863 Time dimensionless
119905119901 Production time hr
119879 Reservoir temperature ∘F
119879119889119901 Dewpoint temperature ∘F
120572 Explicit deconvolution variabledimensionless
120573 Beta deconvolution variabledimensionless
120574119892 Gas gravity (air = 10)
120582 Interporosity flow coefficient of dualporosity reservoir
120601 Porosity of reservoir rock dimensionless120583119892 Gas viscosity cp
120583119900 Oil viscosity cp
120596 Storativity ratio of dual porosity reservoirminus Laplace transform of a function1015840 Derivative of a function
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
References
[1] Y Cheng W J Lee and D A McVay ldquoFast-Fourier-transform-based deconvolution for interpretation of pressure-transient-test data dominated by Wellbore storagerdquo SPE Reservoir Eval-uation and Engineering vol 8 no 3 pp 224ndash239 2005
[2] M M Levitan and M R Wilson ldquoDeconvolution of pressureand rate data from gas reservoirs with significant pressuredepletionrdquo SPE Journal vol 17 no 3 pp 727ndash741 2012
[3] Y Liu and R N Horne ldquoInterpreting pressure and flow-ratedata from permanent downhole gauges by use of data-miningapproachesrdquo SPE Journal vol 18 no 1 pp 69ndash82 2013
[4] O Ogunrewo and A C Gringarten ldquoDeconvolution of welltest data in lean and rich gas condensate and volatile oil wellsbelow saturation pressurerdquo in Proceedings of the SPE AnnualTechnical Conference and Exhibition Document ID 166340-MS New Orleans La USA September-October 2013
[5] R C Earlougher Jr Advances in Well Test Analysis vol 5 ofMonogragh Series SPE Richardson Tex USA 1977
[6] F Kucuk and L Ayestaran ldquoAnalysis of simultaneously mea-sured pressure and sandface flow rate in transient well testingrdquoJournal of PetroleumTechnology vol 37 no 2 pp 323ndash334 1985
[7] M Onur M Cinar D IIk P P Valko T A Blasingame and PHegeman ldquoAn investigation of recent deconvolution methodsfor well-test data analysisrdquo SPE Journal vol 13 no 2 pp 226ndash247 2008
[8] F J KuchukMOnur andFHollaender ldquoPressure transient testdesign and interpretationrdquo Developments in Petroleum Sciencevol 57 pp 303ndash359 2010
[9] A C Gringarten D P Bourdet P A Landel and V J KniazeffldquoA comparison between different skin and wellbore storagetype-curves for early-time transient analysisrdquo in Proceedings ofthe SPE Annual Technical Conference and Exhibition Paper SPE8205 Las Vegas Nev USA 1979
[10] A C Gringarten ldquoFrom straight lines to deconvolution theevolution of the state of the art in well test analysisrdquo SPEReservoir Evaluation and Engineering vol 11 no 1 pp 41ndash622008
[11] J A Cumming D A Wooff T Whittle and A C GringartenldquoMultiple well deconvolutionrdquo in Proceedings of the SPE Annual
12 Abstract and Applied Analysis
Technical Conference and Exhibition Document ID 166458-MS New Orleans La USA September-October 2013
[12] O Bahabanian D Ilk N Hosseinpour-Zonoozi and T ABlasingame ldquoExplicit deconvolution of well test data distortedby wellbore storagerdquo in Proceedings of the SPE Annual TechnicalConference and Exhibition Paper SPE 103216 pp 4444ndash4453San Antonio Tex USA September 2006
[13] J Jones and A Chen ldquoSuccessful applications of pressure-ratedeconvolution in theCad-Nik tight gas formations of the BritishColumbia foothillsrdquo Journal of Canadian Petroleum Technologyvol 51 no 3 pp 176ndash192 2012
[14] D G Russell ldquoExtension of pressure buildup analysis methodsrdquoJournal of Petroleum Technology vol 18 no 12 pp 1624ndash16361966
[15] R E Gladfelter G W Tracy and L E Wilsey ldquoSelecting wellswhich will respond to production-stimulation treatmentrdquo inDrilling and Production Practice API Dallas Tex USA 1955
[16] M J Fetkovich and M E Vienot ldquoRate normalization ofbuildup pressure by using afterflow datardquo Journal of PetroleumTechnology vol 36 no 13 pp 2211ndash2224 1984
[17] J L Johnston Variable rate analysis of transient well testdata using semi-analytical methods [MS thesis] Texas AampMUniversity College Station Tex USA 1992
[18] A F van Everdingen ldquoThe skin effect and its influence on theproductive capacity of a wellrdquo Journal of Petroleum Technologyvol 5 no 6 pp 171ndash176 1953
[19] W Hurst ldquoEstablishment of the skin effect and its impedimentto fluid flow into a well borerdquo The Petroleum Engineer pp B6ndashB16 1953
[20] J A Joseph and L F Koederitz ldquoA simple nonlinear model forrepresentation of field transient responsesrdquo Paper SPE 11435SPE Richardson Tex USA 1982
[21] F J Kuchuk ldquoGladfelter deconvolutionrdquo SPE Formation Evalu-ation vol 5 no 3 pp 285ndash292 1990
[22] A F van Everdingen and W Hurst ldquoThe application of thelaplace transformation to flow problems in reservoirsrdquo Journalof Petroleum Technology vol 1 no 12 pp 305ndash324 1949
[23] D Bourdet T M Whittle A A Douglas and Y M Pirard ldquoAnew set of type curves simplifies well test analysisrdquo World Oilpp 95ndash106 1983
[24] D F Meunier C S Kabir and M J Wittmann ldquoGas well testanalysis use of normalized pseudovariablesrdquo SPE FormationEvaluation vol 2 no 4 pp 629ndash636 1987
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
Abstract and Applied Analysis 5
10minus3 10minus2 10minus1 100 101 102 1030
500
1000
1500
2000
2500
3000
3500
4000
Time (hr)
Dpwf wellbore storaged(Dpwf)d(ln(t)) wellbore storageDps explicit deconvolution methodd(Dps)d(ln(t)) explicit deconvolution methodDps material balance deconvolution methodd(Dps)d(ln(t)) material balance deconvolution methodDps no wellbore storaged(Dps)d(ln(t)) no wellbore storage
Dpwf
and
its lo
garit
hmic
der
ivat
ive (
psi)
Figure 1 Deconvolution for a drawdown test in a homogeneousreservoir
From the above-mentioned methods we use the fol-lowing two techniques to show the verification of theseapproaches
Explicit Deconvolution Formula The final result of explicitdeconvolution method is given as follows
119901119904119863
(119905119863) = minus
119905119863
120572 (119905119863)1199011015840
119908119863+ 119901119908119863
(119905119863) (39)
Plotting of this term versus (int119905119863
0119902119863(120591)119889120591119902
119863(119905119863)) and
(int119905119863
0(1 minus 119902
119863(120591))119889120591(1 minus 119902
119863(119905119863))) for drawdown and buildup
tests respectively yields more accurate resultsThis approachperforms well in field applications
Material Balance Deconvolution The material balance pres-sure drop function for the pressure drawdown test is givenas
Δ119901119908119891 (119905)
119902119863 (119905)
=119901119894minus 119901119908119891 (119905)
119902119863 (119905)
(40)
and the material balance time function for the pressuredrawdown test is given as
119905119898119887
=int119905119863
0119902119863 (120591) 119889120591
119902119863 (119905)
(41)
For the buildup test the material balance pressure dropfunction can be written as
Δ119901119908119904 (Δ119905)
1 minus 119902119863 (Δ119905)
=119901119908119904 (Δ119905) minus 119901
119908119891 (Δ119905 = 0)
1 minus 119902119863 (Δ119905)
(42)
10minus3 10minus2 10minus1 100 101 102 103
Time (hr)
104
103
102
101Dpwf
and
its lo
garit
hmic
der
ivat
ive (
psi)
Dpwf wellbore storaged(Dpwf)d(ln(t)) wellbore storageDps explicit deconvolution methodd(Dps)d(ln(t)) explicit deconvolution methodDps material balance deconvolution methodd(Dps)d(ln(t)) material balance deconvolution methodDps no wellbore storaged(Dps)d(ln(t)) no wellbore storage
Figure 2 Deconvolution for a drawdown test in a homogeneousreservoir
And the material balance time function can be written as
119905119898119887
=int119905119863
0(1 minus 119902
119863 (120591)) 119889120591
1 minus 119902119863 (Δ119905)
(43)
This technique is a good approach for practical applications
5 Verification Examples
We first use few synthetic cases for different reser-voirwellbore systems to validate the applicability of explicitdeconvolution formula and material balance deconvolutionmethod Then the field applications of these techniqueswill be shown by two real cases The unloading rates arecalculated with (20)
The first case is a synthetic test derived from a test designin a vertical oil wellThe reservoir is homogenous and infiniteacting with a constant wellbore storage The wellbore storagecoefficient is 119862
119904= 0015 bblpsi The pressure data are
obtained from the cylindrical source solution (test design)We used the explicit deconvolution formula and material
balance deconvolution method to deconvolve the pressureand unloading rate distorting by wellbore storage effectsTheresults are presented in Figures 1 and 2 for the semilog andlog-log plots in the case of drawdown test and Figures 3and 4 for the semilog and log-log plots in the case of builduptest Besides the deconvolved pressures and their logarithmicderivatives pressure responses with and without wellborestorage are also plotted on the figures for comparisonThrough the deconvolution process we have successfullyremoved the wellbore storage effects so that radial flow canbe identified at early test times The deconvolved pressuresand their logarithmic derivatives are almost identical to
6 Abstract and Applied Analysis
0
500
1000
1500
2000
2500
3000
3500
4000
105104103102101100
Horner time ratio
Dp wellbore storage
Dps explicit deconvolution method(ln
Dps material balance deconvolution
Dps no wellbore storage
ws
and
its lo
garit
hmic
der
ivat
ive (
psi)
Dpws
d(Dp )d(ln )) wellbore storagews (HTR
d(Dps)d )) explicit deconvolution method(HTR
d(Dps)d(ln )) no wellbore storage(HTR)
d(Dps)d(ln )) material balance deconvolution (HTR
Figure 3 Deconvolution for a buildup test in a homogeneousreservoir
10minus3 10minus2 10minus1 100 101 102 103
Time (hr)
104
103
102
101
ws
ws
and
its lo
garit
hmic
der
ivat
ive (
psi)
Dpws
Dp wellbore storage
Dps explicit deconvolution method
Dps material balance deconvolution
Dps no wellbore storage
d(Dp )d(ln )) wellbore storagews (HTR
d(Dps)d )) explicit deconvolution method(HTR
d(Dps)d(ln )) no wellbore storage(HTR)
d(Dps)d(ln )) material balance deconvolution (HTR
(ln
Figure 4 Deconvolution for a buildup test in a homogeneousreservoir
the analytic solutions (without wellbore storage) whichenables us to estimate reservoir parameters accurately
The second case is an oil well drawdown test in adual porosity reservoir The storativity ratio (120596) and inter-porosity flow coefficient (120582) are assumed to be 01 and10minus6 respectively The constant wellbore storage coefficient
is 119862119904= 015 bblpsi The dimensionless wellbore pressure
10minus3 10minus2 10minus1 100 101 102 103
Time (hr)
600
500
400
300
200
100
0
Dpwf wellbore storaged(Dpwf)d(ln(t)) wellbore storageDps explicit deconvolution methodd(Dps)d(ln(t)) explicit deconvolution methodDps material balance deconvolutiond(Dps)d(ln(t)) material balance deconvolutionDps no wellbore storaged(Dps)d(ln(t)) no wellbore storage
Dpwf
and
its lo
garit
hmic
der
ivat
ive (
psi)
Figure 5 Deconvolution for a drawdown test in a dual porosityreservoir
data were obtained from the cylindrical source solution withpseudosteady-state interporosity flow (test design) It is notpossible to identify the dual porosity behavior characteristicof the reservoir from wellbore pressure data because thisbehavior has been masked completely by wellbore storage asindicated in Figures 5 and 6
We deconvolved the pressure and unloading rate andpresent the results in Figures 5 and 6 The deconvolvedresults are reasonably consistent with the analytical solutionAfter deconvolution the dual porosity behavior can beidentified easily from the valley in the logarithmic derivativeof pressure If wellbore storage effects were not removed atearly times it easily could have incorrectly identified thereservoir system as infinite actingThe estimation of reservoirparameters can be obtained now from deconvolved pressure
The third case is an oil well drawdown test in a reservoirwith a sealing fault 250 ft from the well Synthetic pressureresponses were generated by simulation (test design) with aconstant wellbore storage coefficient of 01 bblpsi As we cansee in Figures 7 and 8 the early time infinite acting reservoirbehavior has been masked completely by wellbore storage
To recover the hidden reservoir behavior we performeddeconvolution on the pressure change to remove the wellborestorage effect As in Figures 7 and 8 the deconvolved pressurechange and logarithmic derivative data are almost identicalto the responses without wellbore storageThe deconvolutionresults clearly indicate parallel horizontal lines for the loga-rithmic derivative However without wellbore storage effectselimination this maymislead to identify the reservoir systemas infinite acting So this allows us to correctly identify thereservoir system having a sealing fault
The fourth case is a field example that is taken from theliterature [23] The basic reservoir and fluid properties are
Abstract and Applied Analysis 7
Table 1 Reservoir rock and fluid properties for the first field case
119905119901(hr) ℎ (ft) 119903
119908(ft) 119902
119900(STBD) 119861
119900(RBSTB) 119888
119905(psiminus1) 120583
119900(cp) 120601
1533 107 029 174 106 42 times 10minus6 25 025
Table 2 Reservoir rock and fluid properties for the second field case
119905119901(hr) ℎ (ft) 119903
119908(ft) 119879 (∘F) 119901 (psi) 119902
119900(STBD) 119902
119892(MscfD) 119888
119891(psiminus1) 119879
119889119901(∘F) 119901
119889119901(psi) ∘API 120574
119892
143111 100 0354 2795 7322 104 13925 3 times 10minus6 212 5445 45375 068
10minus3 10minus2 10minus1 100 101 102 103
Time (hr)
103
102
101
100
10minus1
Dpwf wellbore storaged(Dpwf)d(ln(t)) wellbore storageDps explicit deconvolution methodd(Dps)d(ln(t)) explicit deconvolution methodDps material balance deconvolutiond(Dps)d(ln(t)) material balance deconvolutionDps no wellbore storaged(Dps)d(ln(t)) no wellbore storage
Dpwf
and
its lo
garit
hmic
der
ivat
ive (
psi)
Figure 6 Deconvolution for a drawdown test in a dual porosityreservoir
shown in Table 1 In this case we provide the deconvolvedpressure data using the explicit deconvolution formula andthe material balance deconvolution method that is presentedin this work The data are taken from a pressure buildup testand can be reasonably used as field data The deconvolutionresults are shown in Figure 9 in semilog plot As it isillustrated from Figure 9 the corrected pressure provides aclear straight line starting as early as 15 hours in contrast tothe vague line from the uncorrected pressure response
And the last (field) example is a gas condensatewell buildup test following a variable rate productionhistory The duration of last pressure buildup test is361556 hours (from 1889 to 2250556 hours) The basicreservoir and fluid properties for the last drawdown testare shown in Table 2 However the properties of otherproduction tests are not shown here To use our newtechniques effectively for gas wells we use normalizedpseudopressure and pseudotime functions as defined byMeunier et al [24]The final results of explicit deconvolutionformula and material balance deconvolution methodshow two different slopes (Figure 10) where the earlytimes slope is greater than the last one This identifiesthe radial composite behavior of the reservoir With
the corrected pressures the reservoir parameters can also beestimated
6 Conclusions
In this work we have provided some explicit deconvolutionmethods for restoration of constant rate pressure responsesfrom measured pressure data dominated by wellbore storageeffects without sandface rate measurements Based on ourinvestigations we can state the following specific conclusionsrelated to these techniques
(1) The results obtained from the synthetic and fieldexamples show the practical and computational effi-ciency of our deconvolution methods
(2) Modification of 120573-deconvolution is presented wherethe beta quantity is variable rather than a constant
(3) For situations in which there are no downhole ratemeasurements the explicit deconvolution methods(blind deconvolution approaches) can be used toderive the downhole rate function from the pressuremeasurements and restore the reservoir pressureresponse
(4) We can unveil the early time behavior of a reservoirsystemmasked by wellbore storage effects using thesenew approaches Wellbore geometries and reservoirtypes can be identified from this restored pressuredata Furthermore the conventional methods canbe used to analyze this restored formation pressureto estimate formation parameters So the presentedtechniques can be used as the practical tools toimprove pressure transient test interpretation
(5) Deconvolution in well test analysis is often expressedas an inverse problem This approach can be used asan idea to justify the rate normalization method andthe material balance deconvolution method
Appendices
A Russel Method and Its Modification
A method for correcting pressure buildup pressure data freeof wellbore storage effects at early times has been developedby Russell [14] In this method it is not necessary to measurethe rate of influx after closing in at the surface Instead oneuses a theoretical equation which gives the form that thebottom hole pressure should have fluid accumulates in the
8 Abstract and Applied Analysis
10minus3 10minus2 10minus1 100 101 102 103
Time (hr)
250
200
150
100
50
0
Dpwf wellbore storaged(Dpwf)d(ln(t)) wellbore storageDps explicit deconvolution methodd(Dps)d(ln(t)) explicit deconvolution methodDps material balance deconvolutiond(Dps)d(ln(t)) material balance deconvolutionDps no wellbore storaged(Dps)d(ln(t)) no wellbore storage
Dpwf
and
its lo
garit
hmic
der
ivat
ive (
psi)
Figure 7 Deconvolution for a drawdown test in a reservoir with asealing fault
10minus3 10minus2 10minus1 100 101 102 103
Time (hr)
103
102
101
100
Dpwf wellbore storaged(Dpwf)d(ln(t)) wellbore storageDps explicit deconvolution methodd(Dps)d(ln(t)) explicit deconvolution methodDps material balance deconvolutiond(Dps)d(ln(t)) material balance deconvolutionDps no wellbore storaged(Dps)d(ln(t)) no wellbore storage
Dpwf
and
its lo
garit
hmic
der
ivat
ive (
psi)
Figure 8 Deconvolution for a drawdown test in a reservoir with asealing fault
wellbore during buildup This leads to the result that oneshould plot
[119901119908119904 (Δ119905) minus 119901
119908119891 (Δ119905 = 0)]
[1 minus (11198622Δ119905)]
(A1)
versus log (Δ119905) in analyzing pressure buildup data during theearly fill-up period The denominator makes a correction for
104103102101100
Horner time ratio
800
700
600
500
400
300
200
100
0
Dpws
(psi)
Dpws wellbore storageDps explicit deconvolution methodDps material balance deconvolution
Figure 9 Restored pressures for a buildup test in a homogeneousreservoir
0
500
1000
1500
2000
2500
3000
3500
4000
105104103102101100
Horner time ratio
Dpa
(psi)
Dpaw wellbore storage explicit deconvolution method material balance deconvolution
DpasDpas
Figure 10 Restored pressures for a buildup test in a radial compositereservoir
the gradually decreasing flow into the wellbore The quantity1198622is obtained by trial and error as the value which makes the
curve straight at early times After obtaining the straight linesection the rest of the analysis is the same as for any otherpressure buildup This method has the advantage or requiresno additional data over that taken routinely during a shut intest
Due to the time consuming and unconfident of trial anderror to determine 119862
2in Russell method we modify this
method Russellrsquos wellbore storage correction is given as
[119901119908119904 (Δ119905) minus 119901
119908119891 (Δ119905 = 0)]
[1 minus (11198622Δ119905)]
= 119891 (Δ119905 = 1 hr) + 119898119904119897log (Δ119905)
(A2)
Abstract and Applied Analysis 9
Defining the following function and using least squarestechnique obtain the final result
119864 (1198622 119891119898)
=
119873
sum
119894=1
(Δ119901119908119904119894
(1 minus (11198622Δ119905119894))
minus 119891(1198622) minus 119898(119862
2) log(Δ119905
119894))
2
(A3)
Tominimize the function119864 it can differentiate119864with respectto 1198622 119891 and 119898 and setting the results equal to zero After
differentiation it reduces to
120597119864
1205971198622
= minus2
119873
sum
119894=1
(Δ119901119908119904119894
Δ119905119894(1198622minus (1Δ119905
119894))2)
times (Δ119901119908119904119894
(1 minus (11198622Δ119905119894))
minus 119891 (1198622)
minus119898 (1198622) log (Δ119905
119894) ) = 0
(A4)
120597119864
120597119891
= minus2
119873
sum
119894=1
(Δ119901119908119904119894
(1 minus (11198622Δ119905119894))
minus 119891 (1198622) minus 119898 (119862
2) log (Δ119905
119894))
= 0
(A5)120597119864
120597119898
= minus2
119873
sum
119894=1
log (Δ119905119894) (
Δ119901119908119904119894
(1 minus (11198622Δ119905119894))
minus 119891 (1198622)
minus 119898 (1198622) log (Δ119905
119894)) = 0
(A6)
From (A4) we define the following function
119883(1198622)
=
119873
sum
119894=1
(Δ119901119908119904119894
Δ119905119894(1198622minus (1Δ119905
119894))2)
times(Δ119901119908119904119894
(1minus (11198622Δ119905119894))minus119891 (119862
2) minus119898 (119862
2) log (Δ119905
119894))
= 0
(A7)
The roots of (A7) can be found via the following algorithm
1198622(119894+1)
= 1198622(119894)
minus119883 (1198622)
1198831015840 (1198622) (A8)
where
119883(1198622) =
119873
sum
119894=1
(Δ119901119908119904119894
Δ119905119894(1198622minus (1Δ119905
119894))2)
times (Δ119901119908119904119894
(1 minus (11198622Δ119905119894))
minus 119891 (1198622)
minus119898 (1198622) log (Δ119905
119894))
1198831015840(1198622) =
119873
sum
119894=1
(minus2Δ119901119908119904119894
Δ119905119894(1198622minus (1Δ119905
119894))3)
times (Δ119901119908119904119894
Δ119905119894(1198622minus (1Δ119905
119894))2
times [Δ119901119908119904119894
(1 minus (11198622Δ119905))
minus 119891 (1198622)
minus119898 (1198622) log (Δ119905
119894)])
+ (Δ119901119908119904119894
Δ119905119894(1198622minus (1Δ119905
119894))2)
times (minusΔ119901119908119904119894
Δ119905119894(1198622minus (1Δ119905
119894))3minus 1198911015840(1198622)
minus1198981015840(1198622) log (Δ119905
119894))
(A9)
where the unknown functions are given by the followingrelations
119891 (1198622) =
120573119906 (1198622) minus 120572V (119862
2)
1205732 minus 119873120572(A10)
whose derivative is
1198911015840(1198622) =
1205731199061015840(1198622) minus 120572V1015840 (119862
2)
1205732 minus 119873120572
119898 (1198622) =
119873119906 (1198622) minus 120573V (119862
2)
119873120572 minus 1205732
(A11)
where its derivative is
1198981015840(1198622) =
1198731199061015840(1198622) minus 120573V1015840 (119862
2)
119873120572 minus 1205732
119906 (1198622) =
119873
sum
119894=1
log (Δ119905119894)
Δ119901119908119904119894
(1 minus (11198622Δ119905119894))
(A12)
10 Abstract and Applied Analysis
and after differentiation it yields
1199061015840(1198622) = minus
119873
sum
119894=1
log (Δ119905
119894)
Δ119905119894
Δ119901119908119904119894
(1198622minus (1Δ119905
119894))2
V (1198622) =
119873
sum
119894=1
Δ119901119908119904119894
(1 minus (11198622Δ119905119894))
(A13)
which has the following derivative
V1015840 (1198622) = minus
119873
sum
119894=1
1
Δ119905119894
Δ119901119908119904119894
(1198622minus (1Δ119905
119894))2 (A14)
finally the values of 120572 and 120573 parameters are given by thefollowing formulas
120572 =
119873
sum
119894=1
(log(Δ119905119894))2
120573 =
119873
sum
119894=1
log (Δ119905119894)
(A15)
B General Formula for Inverse Problem
The convolution of (6) yields
119901119904119863
(119905119863) = int
119905119863
0
119870 (120591) 119901119908119863 (119905119863 minus 120591) 119889120591
= int
119905119863
0
119870(119905119863minus 120591) 119901
119908119863 (120591) 119889120591
(B1)
where
119870(119905119863) = 119871minus1(
1
119904119902 (119904)) (B2)
where 119871minus1 is the inverse Lapalce transform operator 119870(119905119863)
can be computed either from the Laplace transforms of119902119863(119905119863) data a curve-fitted equation of 119902
119863(119905119863) data or using
another technique with a relation between119870(119905119863) and 119902
119863(119905119863)
However it would be time consuming to invert all the 119902119863(119905119863)
data in Laplace space and transform it back to real space inaccordance with (B2) Once an approximation is obtained itwill be easy to compute119870(119905
119863) and integrate (B1) to determine
119901119904119863(119905119863) However finding some approximation functions for
119902119863(119905119863) data may be difficult We provide an approach to
overcome this problemThis idea is based on finding anotherrelation between 119870(119905
119863) and 119902
119863(119905119863) that is
ℓ (1) =1
119904=
1
119904119902119863(119904)
119902119863(119904) = ℓ (int
119905119863
0
119870 (120591) 119902119863 (119905119863 minus 120591) 119889120591)
= ℓ(int
119905119863
0
119870(119905119863minus 120591) 119902
119863 (120591) 119889120591)
(B3)or
1 = int
119905119863
0
119870 (120591) 119902119863 (119905119863 minus 120591) 119889120591 = int
119905119863
0
119870(119905119863minus 120591) 119902
119863 (120591) 119889120591
(B4)
The other relations that can be used are
ℓ (119905119863) =
1
1199042=
1
1199042119902119863(119904)
119902119863(119904) = ℓ (int
119905119863
0
119871 (120591) 119902119863 (119905119863 minus 120591) 119889120591)
= ℓ(int
119905119863
0
119871 (119905119863minus 120591) 119902
119863 (120591) 119889120591)
(B5)
or
119905119863= int
119905119863
0
119871 (120591) 119902119863(119905119863minus 120591) 119889120591 = int
119905119863
0
119871 (119905119863minus 120591) 119902
119863(120591) 119889120591
(B6)
ℓ (1) =1
119904=
1
1199042119902119863(119904)
119904119902119863(119904) = ℓ (int
119905119863
0
119871 (120591) 1199021015840
119863(119905119863minus 120591) 119889120591)
= ℓ(int
119905119863
0
119871 (119905119863minus 120591) 119902
1015840
119863(120591) 119889120591)
(B7)
or
1 = int
119905119863
0
119871 (120591) 1199021015840
119863(119905119863minus 120591) 119889120591 = int
119905119863
0
119871 (119905119863minus 120591) 119902
1015840
119863(120591) 119889120591
(B8)
where
119871 (119905119863) = ℓminus1(
1
1199042119902 (119904)) (B9)
Equation (B4) can be used simultaneously with (B1) toanalyze the deconvolution problem
Equation (11) can be written as
119901119904119863
(119904) = 119904119901119908119863
(119904)1
1199042119902119863(119904)
(B10)
or
119901119904119863
(119905119863) = int
119905119863
0
119871 (120591) 1199011015840
119908119863(119905119863minus 120591) 119889120591
= int
119905119863
0
119871 (119905119863minus 120591) 119901
1015840
119908119863(120591) 119889120591
(B11)
The applications of (B6) and (B8) are with (B11)For justifying the rate normalizationmethod andmaterial
balance deconvolution it can use (B1) (B4) ((B11) can alsobe used with (B6) or (B8) for this goal) and generalizedmean value theorem for integrals This theorem is expressedas follows
Theorem 1 Let 119891 and 119892 be continuous on [119886 119887] and supposethat 119892 does not change sign on [119886 119887] Then there is a point 119888 in[119886 119887] such that
int
119887
119886
119891 (119905) 119892 (119905) 119889119905 = 119891 (119888) int
119887
119886
119892 (119905) 119889119905 (B12)
Abstract and Applied Analysis 11
By Theorem 1 there are some values 1198881and 1198882between 0
and 119905119863such that
119901119904119863
(119905119863) = 119901119908119863
(1198881) int
119905119863
0
119870(119905119863minus 120591) 119889120591 (B13)
1 = 119902119863(1198882) int
119905119863
0
119870(119905119863minus 120591) 119889120591 (B14)
By setting 1198881
= 1198882
= 119905119863and dividing the right and left
side of (B13) respectively the rate normalization techniqueis developed However the values 119888
1and 1198882are not equal
to each other and are not equal to 119905119863
in general Thisis the reason of error in rate normalization and materialbalance deconvolution methods 119888
1and 1198882are less than 119905
119863in
these methods for the duration of wellbore storage distortionduring a pressure transient test Since these conditions areapproximately satisfied in material balance deconvolutionthe results of material balance are more accurate and moreconfident than rate normalization
Nomenclature
API API gravity ∘API119861119900 Oil formation volume factor RBSTB
119888119905 Total isothermal compressibility factor
psiminus1
1198622 Arbitrary constant hr
minus1
119862119863 Wellbore storage constant dimensionless
119862119904 Wellbore storage coefficient bblpsi
ℎ Formation thickness ftHTR Horner time ratio dimensionless119896 Formation permeability md119870 Kernel of the convolution integral119871 Kernel of the convolution integral119873119901 Cumulative oil production bbl
119901 Pressure psi119901119886 Normalized pseudopressure psi
119901 Average reservoir pressure psi119901119863 Formation pressure dimensionless
119901119889119901 Dewpoint pressure psi
119901119894 Initial reservoir pressure psi
119901119904119863 Formation pressure including skin
dimensionless119901119904 Formation pressure including skin psi
119901119908119863
Pressure draw down dimensionless119901119908119891 Flowing bottom hole pressure psi
119901119908119904 Shut-in bottom hole pressure psi
119902119863 Sandface flow rate dimensionless
119902119892 Gas flow rate MscfD
119902119900 Oil flow rate STBD
119903119908 Wellbore radius ft
119904 Laplace variable119878 Skin factor119905 Time hr119905119863 Time dimensionless
119905119901 Production time hr
119879 Reservoir temperature ∘F
119879119889119901 Dewpoint temperature ∘F
120572 Explicit deconvolution variabledimensionless
120573 Beta deconvolution variabledimensionless
120574119892 Gas gravity (air = 10)
120582 Interporosity flow coefficient of dualporosity reservoir
120601 Porosity of reservoir rock dimensionless120583119892 Gas viscosity cp
120583119900 Oil viscosity cp
120596 Storativity ratio of dual porosity reservoirminus Laplace transform of a function1015840 Derivative of a function
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
References
[1] Y Cheng W J Lee and D A McVay ldquoFast-Fourier-transform-based deconvolution for interpretation of pressure-transient-test data dominated by Wellbore storagerdquo SPE Reservoir Eval-uation and Engineering vol 8 no 3 pp 224ndash239 2005
[2] M M Levitan and M R Wilson ldquoDeconvolution of pressureand rate data from gas reservoirs with significant pressuredepletionrdquo SPE Journal vol 17 no 3 pp 727ndash741 2012
[3] Y Liu and R N Horne ldquoInterpreting pressure and flow-ratedata from permanent downhole gauges by use of data-miningapproachesrdquo SPE Journal vol 18 no 1 pp 69ndash82 2013
[4] O Ogunrewo and A C Gringarten ldquoDeconvolution of welltest data in lean and rich gas condensate and volatile oil wellsbelow saturation pressurerdquo in Proceedings of the SPE AnnualTechnical Conference and Exhibition Document ID 166340-MS New Orleans La USA September-October 2013
[5] R C Earlougher Jr Advances in Well Test Analysis vol 5 ofMonogragh Series SPE Richardson Tex USA 1977
[6] F Kucuk and L Ayestaran ldquoAnalysis of simultaneously mea-sured pressure and sandface flow rate in transient well testingrdquoJournal of PetroleumTechnology vol 37 no 2 pp 323ndash334 1985
[7] M Onur M Cinar D IIk P P Valko T A Blasingame and PHegeman ldquoAn investigation of recent deconvolution methodsfor well-test data analysisrdquo SPE Journal vol 13 no 2 pp 226ndash247 2008
[8] F J KuchukMOnur andFHollaender ldquoPressure transient testdesign and interpretationrdquo Developments in Petroleum Sciencevol 57 pp 303ndash359 2010
[9] A C Gringarten D P Bourdet P A Landel and V J KniazeffldquoA comparison between different skin and wellbore storagetype-curves for early-time transient analysisrdquo in Proceedings ofthe SPE Annual Technical Conference and Exhibition Paper SPE8205 Las Vegas Nev USA 1979
[10] A C Gringarten ldquoFrom straight lines to deconvolution theevolution of the state of the art in well test analysisrdquo SPEReservoir Evaluation and Engineering vol 11 no 1 pp 41ndash622008
[11] J A Cumming D A Wooff T Whittle and A C GringartenldquoMultiple well deconvolutionrdquo in Proceedings of the SPE Annual
12 Abstract and Applied Analysis
Technical Conference and Exhibition Document ID 166458-MS New Orleans La USA September-October 2013
[12] O Bahabanian D Ilk N Hosseinpour-Zonoozi and T ABlasingame ldquoExplicit deconvolution of well test data distortedby wellbore storagerdquo in Proceedings of the SPE Annual TechnicalConference and Exhibition Paper SPE 103216 pp 4444ndash4453San Antonio Tex USA September 2006
[13] J Jones and A Chen ldquoSuccessful applications of pressure-ratedeconvolution in theCad-Nik tight gas formations of the BritishColumbia foothillsrdquo Journal of Canadian Petroleum Technologyvol 51 no 3 pp 176ndash192 2012
[14] D G Russell ldquoExtension of pressure buildup analysis methodsrdquoJournal of Petroleum Technology vol 18 no 12 pp 1624ndash16361966
[15] R E Gladfelter G W Tracy and L E Wilsey ldquoSelecting wellswhich will respond to production-stimulation treatmentrdquo inDrilling and Production Practice API Dallas Tex USA 1955
[16] M J Fetkovich and M E Vienot ldquoRate normalization ofbuildup pressure by using afterflow datardquo Journal of PetroleumTechnology vol 36 no 13 pp 2211ndash2224 1984
[17] J L Johnston Variable rate analysis of transient well testdata using semi-analytical methods [MS thesis] Texas AampMUniversity College Station Tex USA 1992
[18] A F van Everdingen ldquoThe skin effect and its influence on theproductive capacity of a wellrdquo Journal of Petroleum Technologyvol 5 no 6 pp 171ndash176 1953
[19] W Hurst ldquoEstablishment of the skin effect and its impedimentto fluid flow into a well borerdquo The Petroleum Engineer pp B6ndashB16 1953
[20] J A Joseph and L F Koederitz ldquoA simple nonlinear model forrepresentation of field transient responsesrdquo Paper SPE 11435SPE Richardson Tex USA 1982
[21] F J Kuchuk ldquoGladfelter deconvolutionrdquo SPE Formation Evalu-ation vol 5 no 3 pp 285ndash292 1990
[22] A F van Everdingen and W Hurst ldquoThe application of thelaplace transformation to flow problems in reservoirsrdquo Journalof Petroleum Technology vol 1 no 12 pp 305ndash324 1949
[23] D Bourdet T M Whittle A A Douglas and Y M Pirard ldquoAnew set of type curves simplifies well test analysisrdquo World Oilpp 95ndash106 1983
[24] D F Meunier C S Kabir and M J Wittmann ldquoGas well testanalysis use of normalized pseudovariablesrdquo SPE FormationEvaluation vol 2 no 4 pp 629ndash636 1987
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
6 Abstract and Applied Analysis
0
500
1000
1500
2000
2500
3000
3500
4000
105104103102101100
Horner time ratio
Dp wellbore storage
Dps explicit deconvolution method(ln
Dps material balance deconvolution
Dps no wellbore storage
ws
and
its lo
garit
hmic
der
ivat
ive (
psi)
Dpws
d(Dp )d(ln )) wellbore storagews (HTR
d(Dps)d )) explicit deconvolution method(HTR
d(Dps)d(ln )) no wellbore storage(HTR)
d(Dps)d(ln )) material balance deconvolution (HTR
Figure 3 Deconvolution for a buildup test in a homogeneousreservoir
10minus3 10minus2 10minus1 100 101 102 103
Time (hr)
104
103
102
101
ws
ws
and
its lo
garit
hmic
der
ivat
ive (
psi)
Dpws
Dp wellbore storage
Dps explicit deconvolution method
Dps material balance deconvolution
Dps no wellbore storage
d(Dp )d(ln )) wellbore storagews (HTR
d(Dps)d )) explicit deconvolution method(HTR
d(Dps)d(ln )) no wellbore storage(HTR)
d(Dps)d(ln )) material balance deconvolution (HTR
(ln
Figure 4 Deconvolution for a buildup test in a homogeneousreservoir
the analytic solutions (without wellbore storage) whichenables us to estimate reservoir parameters accurately
The second case is an oil well drawdown test in adual porosity reservoir The storativity ratio (120596) and inter-porosity flow coefficient (120582) are assumed to be 01 and10minus6 respectively The constant wellbore storage coefficient
is 119862119904= 015 bblpsi The dimensionless wellbore pressure
10minus3 10minus2 10minus1 100 101 102 103
Time (hr)
600
500
400
300
200
100
0
Dpwf wellbore storaged(Dpwf)d(ln(t)) wellbore storageDps explicit deconvolution methodd(Dps)d(ln(t)) explicit deconvolution methodDps material balance deconvolutiond(Dps)d(ln(t)) material balance deconvolutionDps no wellbore storaged(Dps)d(ln(t)) no wellbore storage
Dpwf
and
its lo
garit
hmic
der
ivat
ive (
psi)
Figure 5 Deconvolution for a drawdown test in a dual porosityreservoir
data were obtained from the cylindrical source solution withpseudosteady-state interporosity flow (test design) It is notpossible to identify the dual porosity behavior characteristicof the reservoir from wellbore pressure data because thisbehavior has been masked completely by wellbore storage asindicated in Figures 5 and 6
We deconvolved the pressure and unloading rate andpresent the results in Figures 5 and 6 The deconvolvedresults are reasonably consistent with the analytical solutionAfter deconvolution the dual porosity behavior can beidentified easily from the valley in the logarithmic derivativeof pressure If wellbore storage effects were not removed atearly times it easily could have incorrectly identified thereservoir system as infinite actingThe estimation of reservoirparameters can be obtained now from deconvolved pressure
The third case is an oil well drawdown test in a reservoirwith a sealing fault 250 ft from the well Synthetic pressureresponses were generated by simulation (test design) with aconstant wellbore storage coefficient of 01 bblpsi As we cansee in Figures 7 and 8 the early time infinite acting reservoirbehavior has been masked completely by wellbore storage
To recover the hidden reservoir behavior we performeddeconvolution on the pressure change to remove the wellborestorage effect As in Figures 7 and 8 the deconvolved pressurechange and logarithmic derivative data are almost identicalto the responses without wellbore storageThe deconvolutionresults clearly indicate parallel horizontal lines for the loga-rithmic derivative However without wellbore storage effectselimination this maymislead to identify the reservoir systemas infinite acting So this allows us to correctly identify thereservoir system having a sealing fault
The fourth case is a field example that is taken from theliterature [23] The basic reservoir and fluid properties are
Abstract and Applied Analysis 7
Table 1 Reservoir rock and fluid properties for the first field case
119905119901(hr) ℎ (ft) 119903
119908(ft) 119902
119900(STBD) 119861
119900(RBSTB) 119888
119905(psiminus1) 120583
119900(cp) 120601
1533 107 029 174 106 42 times 10minus6 25 025
Table 2 Reservoir rock and fluid properties for the second field case
119905119901(hr) ℎ (ft) 119903
119908(ft) 119879 (∘F) 119901 (psi) 119902
119900(STBD) 119902
119892(MscfD) 119888
119891(psiminus1) 119879
119889119901(∘F) 119901
119889119901(psi) ∘API 120574
119892
143111 100 0354 2795 7322 104 13925 3 times 10minus6 212 5445 45375 068
10minus3 10minus2 10minus1 100 101 102 103
Time (hr)
103
102
101
100
10minus1
Dpwf wellbore storaged(Dpwf)d(ln(t)) wellbore storageDps explicit deconvolution methodd(Dps)d(ln(t)) explicit deconvolution methodDps material balance deconvolutiond(Dps)d(ln(t)) material balance deconvolutionDps no wellbore storaged(Dps)d(ln(t)) no wellbore storage
Dpwf
and
its lo
garit
hmic
der
ivat
ive (
psi)
Figure 6 Deconvolution for a drawdown test in a dual porosityreservoir
shown in Table 1 In this case we provide the deconvolvedpressure data using the explicit deconvolution formula andthe material balance deconvolution method that is presentedin this work The data are taken from a pressure buildup testand can be reasonably used as field data The deconvolutionresults are shown in Figure 9 in semilog plot As it isillustrated from Figure 9 the corrected pressure provides aclear straight line starting as early as 15 hours in contrast tothe vague line from the uncorrected pressure response
And the last (field) example is a gas condensatewell buildup test following a variable rate productionhistory The duration of last pressure buildup test is361556 hours (from 1889 to 2250556 hours) The basicreservoir and fluid properties for the last drawdown testare shown in Table 2 However the properties of otherproduction tests are not shown here To use our newtechniques effectively for gas wells we use normalizedpseudopressure and pseudotime functions as defined byMeunier et al [24]The final results of explicit deconvolutionformula and material balance deconvolution methodshow two different slopes (Figure 10) where the earlytimes slope is greater than the last one This identifiesthe radial composite behavior of the reservoir With
the corrected pressures the reservoir parameters can also beestimated
6 Conclusions
In this work we have provided some explicit deconvolutionmethods for restoration of constant rate pressure responsesfrom measured pressure data dominated by wellbore storageeffects without sandface rate measurements Based on ourinvestigations we can state the following specific conclusionsrelated to these techniques
(1) The results obtained from the synthetic and fieldexamples show the practical and computational effi-ciency of our deconvolution methods
(2) Modification of 120573-deconvolution is presented wherethe beta quantity is variable rather than a constant
(3) For situations in which there are no downhole ratemeasurements the explicit deconvolution methods(blind deconvolution approaches) can be used toderive the downhole rate function from the pressuremeasurements and restore the reservoir pressureresponse
(4) We can unveil the early time behavior of a reservoirsystemmasked by wellbore storage effects using thesenew approaches Wellbore geometries and reservoirtypes can be identified from this restored pressuredata Furthermore the conventional methods canbe used to analyze this restored formation pressureto estimate formation parameters So the presentedtechniques can be used as the practical tools toimprove pressure transient test interpretation
(5) Deconvolution in well test analysis is often expressedas an inverse problem This approach can be used asan idea to justify the rate normalization method andthe material balance deconvolution method
Appendices
A Russel Method and Its Modification
A method for correcting pressure buildup pressure data freeof wellbore storage effects at early times has been developedby Russell [14] In this method it is not necessary to measurethe rate of influx after closing in at the surface Instead oneuses a theoretical equation which gives the form that thebottom hole pressure should have fluid accumulates in the
8 Abstract and Applied Analysis
10minus3 10minus2 10minus1 100 101 102 103
Time (hr)
250
200
150
100
50
0
Dpwf wellbore storaged(Dpwf)d(ln(t)) wellbore storageDps explicit deconvolution methodd(Dps)d(ln(t)) explicit deconvolution methodDps material balance deconvolutiond(Dps)d(ln(t)) material balance deconvolutionDps no wellbore storaged(Dps)d(ln(t)) no wellbore storage
Dpwf
and
its lo
garit
hmic
der
ivat
ive (
psi)
Figure 7 Deconvolution for a drawdown test in a reservoir with asealing fault
10minus3 10minus2 10minus1 100 101 102 103
Time (hr)
103
102
101
100
Dpwf wellbore storaged(Dpwf)d(ln(t)) wellbore storageDps explicit deconvolution methodd(Dps)d(ln(t)) explicit deconvolution methodDps material balance deconvolutiond(Dps)d(ln(t)) material balance deconvolutionDps no wellbore storaged(Dps)d(ln(t)) no wellbore storage
Dpwf
and
its lo
garit
hmic
der
ivat
ive (
psi)
Figure 8 Deconvolution for a drawdown test in a reservoir with asealing fault
wellbore during buildup This leads to the result that oneshould plot
[119901119908119904 (Δ119905) minus 119901
119908119891 (Δ119905 = 0)]
[1 minus (11198622Δ119905)]
(A1)
versus log (Δ119905) in analyzing pressure buildup data during theearly fill-up period The denominator makes a correction for
104103102101100
Horner time ratio
800
700
600
500
400
300
200
100
0
Dpws
(psi)
Dpws wellbore storageDps explicit deconvolution methodDps material balance deconvolution
Figure 9 Restored pressures for a buildup test in a homogeneousreservoir
0
500
1000
1500
2000
2500
3000
3500
4000
105104103102101100
Horner time ratio
Dpa
(psi)
Dpaw wellbore storage explicit deconvolution method material balance deconvolution
DpasDpas
Figure 10 Restored pressures for a buildup test in a radial compositereservoir
the gradually decreasing flow into the wellbore The quantity1198622is obtained by trial and error as the value which makes the
curve straight at early times After obtaining the straight linesection the rest of the analysis is the same as for any otherpressure buildup This method has the advantage or requiresno additional data over that taken routinely during a shut intest
Due to the time consuming and unconfident of trial anderror to determine 119862
2in Russell method we modify this
method Russellrsquos wellbore storage correction is given as
[119901119908119904 (Δ119905) minus 119901
119908119891 (Δ119905 = 0)]
[1 minus (11198622Δ119905)]
= 119891 (Δ119905 = 1 hr) + 119898119904119897log (Δ119905)
(A2)
Abstract and Applied Analysis 9
Defining the following function and using least squarestechnique obtain the final result
119864 (1198622 119891119898)
=
119873
sum
119894=1
(Δ119901119908119904119894
(1 minus (11198622Δ119905119894))
minus 119891(1198622) minus 119898(119862
2) log(Δ119905
119894))
2
(A3)
Tominimize the function119864 it can differentiate119864with respectto 1198622 119891 and 119898 and setting the results equal to zero After
differentiation it reduces to
120597119864
1205971198622
= minus2
119873
sum
119894=1
(Δ119901119908119904119894
Δ119905119894(1198622minus (1Δ119905
119894))2)
times (Δ119901119908119904119894
(1 minus (11198622Δ119905119894))
minus 119891 (1198622)
minus119898 (1198622) log (Δ119905
119894) ) = 0
(A4)
120597119864
120597119891
= minus2
119873
sum
119894=1
(Δ119901119908119904119894
(1 minus (11198622Δ119905119894))
minus 119891 (1198622) minus 119898 (119862
2) log (Δ119905
119894))
= 0
(A5)120597119864
120597119898
= minus2
119873
sum
119894=1
log (Δ119905119894) (
Δ119901119908119904119894
(1 minus (11198622Δ119905119894))
minus 119891 (1198622)
minus 119898 (1198622) log (Δ119905
119894)) = 0
(A6)
From (A4) we define the following function
119883(1198622)
=
119873
sum
119894=1
(Δ119901119908119904119894
Δ119905119894(1198622minus (1Δ119905
119894))2)
times(Δ119901119908119904119894
(1minus (11198622Δ119905119894))minus119891 (119862
2) minus119898 (119862
2) log (Δ119905
119894))
= 0
(A7)
The roots of (A7) can be found via the following algorithm
1198622(119894+1)
= 1198622(119894)
minus119883 (1198622)
1198831015840 (1198622) (A8)
where
119883(1198622) =
119873
sum
119894=1
(Δ119901119908119904119894
Δ119905119894(1198622minus (1Δ119905
119894))2)
times (Δ119901119908119904119894
(1 minus (11198622Δ119905119894))
minus 119891 (1198622)
minus119898 (1198622) log (Δ119905
119894))
1198831015840(1198622) =
119873
sum
119894=1
(minus2Δ119901119908119904119894
Δ119905119894(1198622minus (1Δ119905
119894))3)
times (Δ119901119908119904119894
Δ119905119894(1198622minus (1Δ119905
119894))2
times [Δ119901119908119904119894
(1 minus (11198622Δ119905))
minus 119891 (1198622)
minus119898 (1198622) log (Δ119905
119894)])
+ (Δ119901119908119904119894
Δ119905119894(1198622minus (1Δ119905
119894))2)
times (minusΔ119901119908119904119894
Δ119905119894(1198622minus (1Δ119905
119894))3minus 1198911015840(1198622)
minus1198981015840(1198622) log (Δ119905
119894))
(A9)
where the unknown functions are given by the followingrelations
119891 (1198622) =
120573119906 (1198622) minus 120572V (119862
2)
1205732 minus 119873120572(A10)
whose derivative is
1198911015840(1198622) =
1205731199061015840(1198622) minus 120572V1015840 (119862
2)
1205732 minus 119873120572
119898 (1198622) =
119873119906 (1198622) minus 120573V (119862
2)
119873120572 minus 1205732
(A11)
where its derivative is
1198981015840(1198622) =
1198731199061015840(1198622) minus 120573V1015840 (119862
2)
119873120572 minus 1205732
119906 (1198622) =
119873
sum
119894=1
log (Δ119905119894)
Δ119901119908119904119894
(1 minus (11198622Δ119905119894))
(A12)
10 Abstract and Applied Analysis
and after differentiation it yields
1199061015840(1198622) = minus
119873
sum
119894=1
log (Δ119905
119894)
Δ119905119894
Δ119901119908119904119894
(1198622minus (1Δ119905
119894))2
V (1198622) =
119873
sum
119894=1
Δ119901119908119904119894
(1 minus (11198622Δ119905119894))
(A13)
which has the following derivative
V1015840 (1198622) = minus
119873
sum
119894=1
1
Δ119905119894
Δ119901119908119904119894
(1198622minus (1Δ119905
119894))2 (A14)
finally the values of 120572 and 120573 parameters are given by thefollowing formulas
120572 =
119873
sum
119894=1
(log(Δ119905119894))2
120573 =
119873
sum
119894=1
log (Δ119905119894)
(A15)
B General Formula for Inverse Problem
The convolution of (6) yields
119901119904119863
(119905119863) = int
119905119863
0
119870 (120591) 119901119908119863 (119905119863 minus 120591) 119889120591
= int
119905119863
0
119870(119905119863minus 120591) 119901
119908119863 (120591) 119889120591
(B1)
where
119870(119905119863) = 119871minus1(
1
119904119902 (119904)) (B2)
where 119871minus1 is the inverse Lapalce transform operator 119870(119905119863)
can be computed either from the Laplace transforms of119902119863(119905119863) data a curve-fitted equation of 119902
119863(119905119863) data or using
another technique with a relation between119870(119905119863) and 119902
119863(119905119863)
However it would be time consuming to invert all the 119902119863(119905119863)
data in Laplace space and transform it back to real space inaccordance with (B2) Once an approximation is obtained itwill be easy to compute119870(119905
119863) and integrate (B1) to determine
119901119904119863(119905119863) However finding some approximation functions for
119902119863(119905119863) data may be difficult We provide an approach to
overcome this problemThis idea is based on finding anotherrelation between 119870(119905
119863) and 119902
119863(119905119863) that is
ℓ (1) =1
119904=
1
119904119902119863(119904)
119902119863(119904) = ℓ (int
119905119863
0
119870 (120591) 119902119863 (119905119863 minus 120591) 119889120591)
= ℓ(int
119905119863
0
119870(119905119863minus 120591) 119902
119863 (120591) 119889120591)
(B3)or
1 = int
119905119863
0
119870 (120591) 119902119863 (119905119863 minus 120591) 119889120591 = int
119905119863
0
119870(119905119863minus 120591) 119902
119863 (120591) 119889120591
(B4)
The other relations that can be used are
ℓ (119905119863) =
1
1199042=
1
1199042119902119863(119904)
119902119863(119904) = ℓ (int
119905119863
0
119871 (120591) 119902119863 (119905119863 minus 120591) 119889120591)
= ℓ(int
119905119863
0
119871 (119905119863minus 120591) 119902
119863 (120591) 119889120591)
(B5)
or
119905119863= int
119905119863
0
119871 (120591) 119902119863(119905119863minus 120591) 119889120591 = int
119905119863
0
119871 (119905119863minus 120591) 119902
119863(120591) 119889120591
(B6)
ℓ (1) =1
119904=
1
1199042119902119863(119904)
119904119902119863(119904) = ℓ (int
119905119863
0
119871 (120591) 1199021015840
119863(119905119863minus 120591) 119889120591)
= ℓ(int
119905119863
0
119871 (119905119863minus 120591) 119902
1015840
119863(120591) 119889120591)
(B7)
or
1 = int
119905119863
0
119871 (120591) 1199021015840
119863(119905119863minus 120591) 119889120591 = int
119905119863
0
119871 (119905119863minus 120591) 119902
1015840
119863(120591) 119889120591
(B8)
where
119871 (119905119863) = ℓminus1(
1
1199042119902 (119904)) (B9)
Equation (B4) can be used simultaneously with (B1) toanalyze the deconvolution problem
Equation (11) can be written as
119901119904119863
(119904) = 119904119901119908119863
(119904)1
1199042119902119863(119904)
(B10)
or
119901119904119863
(119905119863) = int
119905119863
0
119871 (120591) 1199011015840
119908119863(119905119863minus 120591) 119889120591
= int
119905119863
0
119871 (119905119863minus 120591) 119901
1015840
119908119863(120591) 119889120591
(B11)
The applications of (B6) and (B8) are with (B11)For justifying the rate normalizationmethod andmaterial
balance deconvolution it can use (B1) (B4) ((B11) can alsobe used with (B6) or (B8) for this goal) and generalizedmean value theorem for integrals This theorem is expressedas follows
Theorem 1 Let 119891 and 119892 be continuous on [119886 119887] and supposethat 119892 does not change sign on [119886 119887] Then there is a point 119888 in[119886 119887] such that
int
119887
119886
119891 (119905) 119892 (119905) 119889119905 = 119891 (119888) int
119887
119886
119892 (119905) 119889119905 (B12)
Abstract and Applied Analysis 11
By Theorem 1 there are some values 1198881and 1198882between 0
and 119905119863such that
119901119904119863
(119905119863) = 119901119908119863
(1198881) int
119905119863
0
119870(119905119863minus 120591) 119889120591 (B13)
1 = 119902119863(1198882) int
119905119863
0
119870(119905119863minus 120591) 119889120591 (B14)
By setting 1198881
= 1198882
= 119905119863and dividing the right and left
side of (B13) respectively the rate normalization techniqueis developed However the values 119888
1and 1198882are not equal
to each other and are not equal to 119905119863
in general Thisis the reason of error in rate normalization and materialbalance deconvolution methods 119888
1and 1198882are less than 119905
119863in
these methods for the duration of wellbore storage distortionduring a pressure transient test Since these conditions areapproximately satisfied in material balance deconvolutionthe results of material balance are more accurate and moreconfident than rate normalization
Nomenclature
API API gravity ∘API119861119900 Oil formation volume factor RBSTB
119888119905 Total isothermal compressibility factor
psiminus1
1198622 Arbitrary constant hr
minus1
119862119863 Wellbore storage constant dimensionless
119862119904 Wellbore storage coefficient bblpsi
ℎ Formation thickness ftHTR Horner time ratio dimensionless119896 Formation permeability md119870 Kernel of the convolution integral119871 Kernel of the convolution integral119873119901 Cumulative oil production bbl
119901 Pressure psi119901119886 Normalized pseudopressure psi
119901 Average reservoir pressure psi119901119863 Formation pressure dimensionless
119901119889119901 Dewpoint pressure psi
119901119894 Initial reservoir pressure psi
119901119904119863 Formation pressure including skin
dimensionless119901119904 Formation pressure including skin psi
119901119908119863
Pressure draw down dimensionless119901119908119891 Flowing bottom hole pressure psi
119901119908119904 Shut-in bottom hole pressure psi
119902119863 Sandface flow rate dimensionless
119902119892 Gas flow rate MscfD
119902119900 Oil flow rate STBD
119903119908 Wellbore radius ft
119904 Laplace variable119878 Skin factor119905 Time hr119905119863 Time dimensionless
119905119901 Production time hr
119879 Reservoir temperature ∘F
119879119889119901 Dewpoint temperature ∘F
120572 Explicit deconvolution variabledimensionless
120573 Beta deconvolution variabledimensionless
120574119892 Gas gravity (air = 10)
120582 Interporosity flow coefficient of dualporosity reservoir
120601 Porosity of reservoir rock dimensionless120583119892 Gas viscosity cp
120583119900 Oil viscosity cp
120596 Storativity ratio of dual porosity reservoirminus Laplace transform of a function1015840 Derivative of a function
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
References
[1] Y Cheng W J Lee and D A McVay ldquoFast-Fourier-transform-based deconvolution for interpretation of pressure-transient-test data dominated by Wellbore storagerdquo SPE Reservoir Eval-uation and Engineering vol 8 no 3 pp 224ndash239 2005
[2] M M Levitan and M R Wilson ldquoDeconvolution of pressureand rate data from gas reservoirs with significant pressuredepletionrdquo SPE Journal vol 17 no 3 pp 727ndash741 2012
[3] Y Liu and R N Horne ldquoInterpreting pressure and flow-ratedata from permanent downhole gauges by use of data-miningapproachesrdquo SPE Journal vol 18 no 1 pp 69ndash82 2013
[4] O Ogunrewo and A C Gringarten ldquoDeconvolution of welltest data in lean and rich gas condensate and volatile oil wellsbelow saturation pressurerdquo in Proceedings of the SPE AnnualTechnical Conference and Exhibition Document ID 166340-MS New Orleans La USA September-October 2013
[5] R C Earlougher Jr Advances in Well Test Analysis vol 5 ofMonogragh Series SPE Richardson Tex USA 1977
[6] F Kucuk and L Ayestaran ldquoAnalysis of simultaneously mea-sured pressure and sandface flow rate in transient well testingrdquoJournal of PetroleumTechnology vol 37 no 2 pp 323ndash334 1985
[7] M Onur M Cinar D IIk P P Valko T A Blasingame and PHegeman ldquoAn investigation of recent deconvolution methodsfor well-test data analysisrdquo SPE Journal vol 13 no 2 pp 226ndash247 2008
[8] F J KuchukMOnur andFHollaender ldquoPressure transient testdesign and interpretationrdquo Developments in Petroleum Sciencevol 57 pp 303ndash359 2010
[9] A C Gringarten D P Bourdet P A Landel and V J KniazeffldquoA comparison between different skin and wellbore storagetype-curves for early-time transient analysisrdquo in Proceedings ofthe SPE Annual Technical Conference and Exhibition Paper SPE8205 Las Vegas Nev USA 1979
[10] A C Gringarten ldquoFrom straight lines to deconvolution theevolution of the state of the art in well test analysisrdquo SPEReservoir Evaluation and Engineering vol 11 no 1 pp 41ndash622008
[11] J A Cumming D A Wooff T Whittle and A C GringartenldquoMultiple well deconvolutionrdquo in Proceedings of the SPE Annual
12 Abstract and Applied Analysis
Technical Conference and Exhibition Document ID 166458-MS New Orleans La USA September-October 2013
[12] O Bahabanian D Ilk N Hosseinpour-Zonoozi and T ABlasingame ldquoExplicit deconvolution of well test data distortedby wellbore storagerdquo in Proceedings of the SPE Annual TechnicalConference and Exhibition Paper SPE 103216 pp 4444ndash4453San Antonio Tex USA September 2006
[13] J Jones and A Chen ldquoSuccessful applications of pressure-ratedeconvolution in theCad-Nik tight gas formations of the BritishColumbia foothillsrdquo Journal of Canadian Petroleum Technologyvol 51 no 3 pp 176ndash192 2012
[14] D G Russell ldquoExtension of pressure buildup analysis methodsrdquoJournal of Petroleum Technology vol 18 no 12 pp 1624ndash16361966
[15] R E Gladfelter G W Tracy and L E Wilsey ldquoSelecting wellswhich will respond to production-stimulation treatmentrdquo inDrilling and Production Practice API Dallas Tex USA 1955
[16] M J Fetkovich and M E Vienot ldquoRate normalization ofbuildup pressure by using afterflow datardquo Journal of PetroleumTechnology vol 36 no 13 pp 2211ndash2224 1984
[17] J L Johnston Variable rate analysis of transient well testdata using semi-analytical methods [MS thesis] Texas AampMUniversity College Station Tex USA 1992
[18] A F van Everdingen ldquoThe skin effect and its influence on theproductive capacity of a wellrdquo Journal of Petroleum Technologyvol 5 no 6 pp 171ndash176 1953
[19] W Hurst ldquoEstablishment of the skin effect and its impedimentto fluid flow into a well borerdquo The Petroleum Engineer pp B6ndashB16 1953
[20] J A Joseph and L F Koederitz ldquoA simple nonlinear model forrepresentation of field transient responsesrdquo Paper SPE 11435SPE Richardson Tex USA 1982
[21] F J Kuchuk ldquoGladfelter deconvolutionrdquo SPE Formation Evalu-ation vol 5 no 3 pp 285ndash292 1990
[22] A F van Everdingen and W Hurst ldquoThe application of thelaplace transformation to flow problems in reservoirsrdquo Journalof Petroleum Technology vol 1 no 12 pp 305ndash324 1949
[23] D Bourdet T M Whittle A A Douglas and Y M Pirard ldquoAnew set of type curves simplifies well test analysisrdquo World Oilpp 95ndash106 1983
[24] D F Meunier C S Kabir and M J Wittmann ldquoGas well testanalysis use of normalized pseudovariablesrdquo SPE FormationEvaluation vol 2 no 4 pp 629ndash636 1987
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
Abstract and Applied Analysis 7
Table 1 Reservoir rock and fluid properties for the first field case
119905119901(hr) ℎ (ft) 119903
119908(ft) 119902
119900(STBD) 119861
119900(RBSTB) 119888
119905(psiminus1) 120583
119900(cp) 120601
1533 107 029 174 106 42 times 10minus6 25 025
Table 2 Reservoir rock and fluid properties for the second field case
119905119901(hr) ℎ (ft) 119903
119908(ft) 119879 (∘F) 119901 (psi) 119902
119900(STBD) 119902
119892(MscfD) 119888
119891(psiminus1) 119879
119889119901(∘F) 119901
119889119901(psi) ∘API 120574
119892
143111 100 0354 2795 7322 104 13925 3 times 10minus6 212 5445 45375 068
10minus3 10minus2 10minus1 100 101 102 103
Time (hr)
103
102
101
100
10minus1
Dpwf wellbore storaged(Dpwf)d(ln(t)) wellbore storageDps explicit deconvolution methodd(Dps)d(ln(t)) explicit deconvolution methodDps material balance deconvolutiond(Dps)d(ln(t)) material balance deconvolutionDps no wellbore storaged(Dps)d(ln(t)) no wellbore storage
Dpwf
and
its lo
garit
hmic
der
ivat
ive (
psi)
Figure 6 Deconvolution for a drawdown test in a dual porosityreservoir
shown in Table 1 In this case we provide the deconvolvedpressure data using the explicit deconvolution formula andthe material balance deconvolution method that is presentedin this work The data are taken from a pressure buildup testand can be reasonably used as field data The deconvolutionresults are shown in Figure 9 in semilog plot As it isillustrated from Figure 9 the corrected pressure provides aclear straight line starting as early as 15 hours in contrast tothe vague line from the uncorrected pressure response
And the last (field) example is a gas condensatewell buildup test following a variable rate productionhistory The duration of last pressure buildup test is361556 hours (from 1889 to 2250556 hours) The basicreservoir and fluid properties for the last drawdown testare shown in Table 2 However the properties of otherproduction tests are not shown here To use our newtechniques effectively for gas wells we use normalizedpseudopressure and pseudotime functions as defined byMeunier et al [24]The final results of explicit deconvolutionformula and material balance deconvolution methodshow two different slopes (Figure 10) where the earlytimes slope is greater than the last one This identifiesthe radial composite behavior of the reservoir With
the corrected pressures the reservoir parameters can also beestimated
6 Conclusions
In this work we have provided some explicit deconvolutionmethods for restoration of constant rate pressure responsesfrom measured pressure data dominated by wellbore storageeffects without sandface rate measurements Based on ourinvestigations we can state the following specific conclusionsrelated to these techniques
(1) The results obtained from the synthetic and fieldexamples show the practical and computational effi-ciency of our deconvolution methods
(2) Modification of 120573-deconvolution is presented wherethe beta quantity is variable rather than a constant
(3) For situations in which there are no downhole ratemeasurements the explicit deconvolution methods(blind deconvolution approaches) can be used toderive the downhole rate function from the pressuremeasurements and restore the reservoir pressureresponse
(4) We can unveil the early time behavior of a reservoirsystemmasked by wellbore storage effects using thesenew approaches Wellbore geometries and reservoirtypes can be identified from this restored pressuredata Furthermore the conventional methods canbe used to analyze this restored formation pressureto estimate formation parameters So the presentedtechniques can be used as the practical tools toimprove pressure transient test interpretation
(5) Deconvolution in well test analysis is often expressedas an inverse problem This approach can be used asan idea to justify the rate normalization method andthe material balance deconvolution method
Appendices
A Russel Method and Its Modification
A method for correcting pressure buildup pressure data freeof wellbore storage effects at early times has been developedby Russell [14] In this method it is not necessary to measurethe rate of influx after closing in at the surface Instead oneuses a theoretical equation which gives the form that thebottom hole pressure should have fluid accumulates in the
8 Abstract and Applied Analysis
10minus3 10minus2 10minus1 100 101 102 103
Time (hr)
250
200
150
100
50
0
Dpwf wellbore storaged(Dpwf)d(ln(t)) wellbore storageDps explicit deconvolution methodd(Dps)d(ln(t)) explicit deconvolution methodDps material balance deconvolutiond(Dps)d(ln(t)) material balance deconvolutionDps no wellbore storaged(Dps)d(ln(t)) no wellbore storage
Dpwf
and
its lo
garit
hmic
der
ivat
ive (
psi)
Figure 7 Deconvolution for a drawdown test in a reservoir with asealing fault
10minus3 10minus2 10minus1 100 101 102 103
Time (hr)
103
102
101
100
Dpwf wellbore storaged(Dpwf)d(ln(t)) wellbore storageDps explicit deconvolution methodd(Dps)d(ln(t)) explicit deconvolution methodDps material balance deconvolutiond(Dps)d(ln(t)) material balance deconvolutionDps no wellbore storaged(Dps)d(ln(t)) no wellbore storage
Dpwf
and
its lo
garit
hmic
der
ivat
ive (
psi)
Figure 8 Deconvolution for a drawdown test in a reservoir with asealing fault
wellbore during buildup This leads to the result that oneshould plot
[119901119908119904 (Δ119905) minus 119901
119908119891 (Δ119905 = 0)]
[1 minus (11198622Δ119905)]
(A1)
versus log (Δ119905) in analyzing pressure buildup data during theearly fill-up period The denominator makes a correction for
104103102101100
Horner time ratio
800
700
600
500
400
300
200
100
0
Dpws
(psi)
Dpws wellbore storageDps explicit deconvolution methodDps material balance deconvolution
Figure 9 Restored pressures for a buildup test in a homogeneousreservoir
0
500
1000
1500
2000
2500
3000
3500
4000
105104103102101100
Horner time ratio
Dpa
(psi)
Dpaw wellbore storage explicit deconvolution method material balance deconvolution
DpasDpas
Figure 10 Restored pressures for a buildup test in a radial compositereservoir
the gradually decreasing flow into the wellbore The quantity1198622is obtained by trial and error as the value which makes the
curve straight at early times After obtaining the straight linesection the rest of the analysis is the same as for any otherpressure buildup This method has the advantage or requiresno additional data over that taken routinely during a shut intest
Due to the time consuming and unconfident of trial anderror to determine 119862
2in Russell method we modify this
method Russellrsquos wellbore storage correction is given as
[119901119908119904 (Δ119905) minus 119901
119908119891 (Δ119905 = 0)]
[1 minus (11198622Δ119905)]
= 119891 (Δ119905 = 1 hr) + 119898119904119897log (Δ119905)
(A2)
Abstract and Applied Analysis 9
Defining the following function and using least squarestechnique obtain the final result
119864 (1198622 119891119898)
=
119873
sum
119894=1
(Δ119901119908119904119894
(1 minus (11198622Δ119905119894))
minus 119891(1198622) minus 119898(119862
2) log(Δ119905
119894))
2
(A3)
Tominimize the function119864 it can differentiate119864with respectto 1198622 119891 and 119898 and setting the results equal to zero After
differentiation it reduces to
120597119864
1205971198622
= minus2
119873
sum
119894=1
(Δ119901119908119904119894
Δ119905119894(1198622minus (1Δ119905
119894))2)
times (Δ119901119908119904119894
(1 minus (11198622Δ119905119894))
minus 119891 (1198622)
minus119898 (1198622) log (Δ119905
119894) ) = 0
(A4)
120597119864
120597119891
= minus2
119873
sum
119894=1
(Δ119901119908119904119894
(1 minus (11198622Δ119905119894))
minus 119891 (1198622) minus 119898 (119862
2) log (Δ119905
119894))
= 0
(A5)120597119864
120597119898
= minus2
119873
sum
119894=1
log (Δ119905119894) (
Δ119901119908119904119894
(1 minus (11198622Δ119905119894))
minus 119891 (1198622)
minus 119898 (1198622) log (Δ119905
119894)) = 0
(A6)
From (A4) we define the following function
119883(1198622)
=
119873
sum
119894=1
(Δ119901119908119904119894
Δ119905119894(1198622minus (1Δ119905
119894))2)
times(Δ119901119908119904119894
(1minus (11198622Δ119905119894))minus119891 (119862
2) minus119898 (119862
2) log (Δ119905
119894))
= 0
(A7)
The roots of (A7) can be found via the following algorithm
1198622(119894+1)
= 1198622(119894)
minus119883 (1198622)
1198831015840 (1198622) (A8)
where
119883(1198622) =
119873
sum
119894=1
(Δ119901119908119904119894
Δ119905119894(1198622minus (1Δ119905
119894))2)
times (Δ119901119908119904119894
(1 minus (11198622Δ119905119894))
minus 119891 (1198622)
minus119898 (1198622) log (Δ119905
119894))
1198831015840(1198622) =
119873
sum
119894=1
(minus2Δ119901119908119904119894
Δ119905119894(1198622minus (1Δ119905
119894))3)
times (Δ119901119908119904119894
Δ119905119894(1198622minus (1Δ119905
119894))2
times [Δ119901119908119904119894
(1 minus (11198622Δ119905))
minus 119891 (1198622)
minus119898 (1198622) log (Δ119905
119894)])
+ (Δ119901119908119904119894
Δ119905119894(1198622minus (1Δ119905
119894))2)
times (minusΔ119901119908119904119894
Δ119905119894(1198622minus (1Δ119905
119894))3minus 1198911015840(1198622)
minus1198981015840(1198622) log (Δ119905
119894))
(A9)
where the unknown functions are given by the followingrelations
119891 (1198622) =
120573119906 (1198622) minus 120572V (119862
2)
1205732 minus 119873120572(A10)
whose derivative is
1198911015840(1198622) =
1205731199061015840(1198622) minus 120572V1015840 (119862
2)
1205732 minus 119873120572
119898 (1198622) =
119873119906 (1198622) minus 120573V (119862
2)
119873120572 minus 1205732
(A11)
where its derivative is
1198981015840(1198622) =
1198731199061015840(1198622) minus 120573V1015840 (119862
2)
119873120572 minus 1205732
119906 (1198622) =
119873
sum
119894=1
log (Δ119905119894)
Δ119901119908119904119894
(1 minus (11198622Δ119905119894))
(A12)
10 Abstract and Applied Analysis
and after differentiation it yields
1199061015840(1198622) = minus
119873
sum
119894=1
log (Δ119905
119894)
Δ119905119894
Δ119901119908119904119894
(1198622minus (1Δ119905
119894))2
V (1198622) =
119873
sum
119894=1
Δ119901119908119904119894
(1 minus (11198622Δ119905119894))
(A13)
which has the following derivative
V1015840 (1198622) = minus
119873
sum
119894=1
1
Δ119905119894
Δ119901119908119904119894
(1198622minus (1Δ119905
119894))2 (A14)
finally the values of 120572 and 120573 parameters are given by thefollowing formulas
120572 =
119873
sum
119894=1
(log(Δ119905119894))2
120573 =
119873
sum
119894=1
log (Δ119905119894)
(A15)
B General Formula for Inverse Problem
The convolution of (6) yields
119901119904119863
(119905119863) = int
119905119863
0
119870 (120591) 119901119908119863 (119905119863 minus 120591) 119889120591
= int
119905119863
0
119870(119905119863minus 120591) 119901
119908119863 (120591) 119889120591
(B1)
where
119870(119905119863) = 119871minus1(
1
119904119902 (119904)) (B2)
where 119871minus1 is the inverse Lapalce transform operator 119870(119905119863)
can be computed either from the Laplace transforms of119902119863(119905119863) data a curve-fitted equation of 119902
119863(119905119863) data or using
another technique with a relation between119870(119905119863) and 119902
119863(119905119863)
However it would be time consuming to invert all the 119902119863(119905119863)
data in Laplace space and transform it back to real space inaccordance with (B2) Once an approximation is obtained itwill be easy to compute119870(119905
119863) and integrate (B1) to determine
119901119904119863(119905119863) However finding some approximation functions for
119902119863(119905119863) data may be difficult We provide an approach to
overcome this problemThis idea is based on finding anotherrelation between 119870(119905
119863) and 119902
119863(119905119863) that is
ℓ (1) =1
119904=
1
119904119902119863(119904)
119902119863(119904) = ℓ (int
119905119863
0
119870 (120591) 119902119863 (119905119863 minus 120591) 119889120591)
= ℓ(int
119905119863
0
119870(119905119863minus 120591) 119902
119863 (120591) 119889120591)
(B3)or
1 = int
119905119863
0
119870 (120591) 119902119863 (119905119863 minus 120591) 119889120591 = int
119905119863
0
119870(119905119863minus 120591) 119902
119863 (120591) 119889120591
(B4)
The other relations that can be used are
ℓ (119905119863) =
1
1199042=
1
1199042119902119863(119904)
119902119863(119904) = ℓ (int
119905119863
0
119871 (120591) 119902119863 (119905119863 minus 120591) 119889120591)
= ℓ(int
119905119863
0
119871 (119905119863minus 120591) 119902
119863 (120591) 119889120591)
(B5)
or
119905119863= int
119905119863
0
119871 (120591) 119902119863(119905119863minus 120591) 119889120591 = int
119905119863
0
119871 (119905119863minus 120591) 119902
119863(120591) 119889120591
(B6)
ℓ (1) =1
119904=
1
1199042119902119863(119904)
119904119902119863(119904) = ℓ (int
119905119863
0
119871 (120591) 1199021015840
119863(119905119863minus 120591) 119889120591)
= ℓ(int
119905119863
0
119871 (119905119863minus 120591) 119902
1015840
119863(120591) 119889120591)
(B7)
or
1 = int
119905119863
0
119871 (120591) 1199021015840
119863(119905119863minus 120591) 119889120591 = int
119905119863
0
119871 (119905119863minus 120591) 119902
1015840
119863(120591) 119889120591
(B8)
where
119871 (119905119863) = ℓminus1(
1
1199042119902 (119904)) (B9)
Equation (B4) can be used simultaneously with (B1) toanalyze the deconvolution problem
Equation (11) can be written as
119901119904119863
(119904) = 119904119901119908119863
(119904)1
1199042119902119863(119904)
(B10)
or
119901119904119863
(119905119863) = int
119905119863
0
119871 (120591) 1199011015840
119908119863(119905119863minus 120591) 119889120591
= int
119905119863
0
119871 (119905119863minus 120591) 119901
1015840
119908119863(120591) 119889120591
(B11)
The applications of (B6) and (B8) are with (B11)For justifying the rate normalizationmethod andmaterial
balance deconvolution it can use (B1) (B4) ((B11) can alsobe used with (B6) or (B8) for this goal) and generalizedmean value theorem for integrals This theorem is expressedas follows
Theorem 1 Let 119891 and 119892 be continuous on [119886 119887] and supposethat 119892 does not change sign on [119886 119887] Then there is a point 119888 in[119886 119887] such that
int
119887
119886
119891 (119905) 119892 (119905) 119889119905 = 119891 (119888) int
119887
119886
119892 (119905) 119889119905 (B12)
Abstract and Applied Analysis 11
By Theorem 1 there are some values 1198881and 1198882between 0
and 119905119863such that
119901119904119863
(119905119863) = 119901119908119863
(1198881) int
119905119863
0
119870(119905119863minus 120591) 119889120591 (B13)
1 = 119902119863(1198882) int
119905119863
0
119870(119905119863minus 120591) 119889120591 (B14)
By setting 1198881
= 1198882
= 119905119863and dividing the right and left
side of (B13) respectively the rate normalization techniqueis developed However the values 119888
1and 1198882are not equal
to each other and are not equal to 119905119863
in general Thisis the reason of error in rate normalization and materialbalance deconvolution methods 119888
1and 1198882are less than 119905
119863in
these methods for the duration of wellbore storage distortionduring a pressure transient test Since these conditions areapproximately satisfied in material balance deconvolutionthe results of material balance are more accurate and moreconfident than rate normalization
Nomenclature
API API gravity ∘API119861119900 Oil formation volume factor RBSTB
119888119905 Total isothermal compressibility factor
psiminus1
1198622 Arbitrary constant hr
minus1
119862119863 Wellbore storage constant dimensionless
119862119904 Wellbore storage coefficient bblpsi
ℎ Formation thickness ftHTR Horner time ratio dimensionless119896 Formation permeability md119870 Kernel of the convolution integral119871 Kernel of the convolution integral119873119901 Cumulative oil production bbl
119901 Pressure psi119901119886 Normalized pseudopressure psi
119901 Average reservoir pressure psi119901119863 Formation pressure dimensionless
119901119889119901 Dewpoint pressure psi
119901119894 Initial reservoir pressure psi
119901119904119863 Formation pressure including skin
dimensionless119901119904 Formation pressure including skin psi
119901119908119863
Pressure draw down dimensionless119901119908119891 Flowing bottom hole pressure psi
119901119908119904 Shut-in bottom hole pressure psi
119902119863 Sandface flow rate dimensionless
119902119892 Gas flow rate MscfD
119902119900 Oil flow rate STBD
119903119908 Wellbore radius ft
119904 Laplace variable119878 Skin factor119905 Time hr119905119863 Time dimensionless
119905119901 Production time hr
119879 Reservoir temperature ∘F
119879119889119901 Dewpoint temperature ∘F
120572 Explicit deconvolution variabledimensionless
120573 Beta deconvolution variabledimensionless
120574119892 Gas gravity (air = 10)
120582 Interporosity flow coefficient of dualporosity reservoir
120601 Porosity of reservoir rock dimensionless120583119892 Gas viscosity cp
120583119900 Oil viscosity cp
120596 Storativity ratio of dual porosity reservoirminus Laplace transform of a function1015840 Derivative of a function
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
References
[1] Y Cheng W J Lee and D A McVay ldquoFast-Fourier-transform-based deconvolution for interpretation of pressure-transient-test data dominated by Wellbore storagerdquo SPE Reservoir Eval-uation and Engineering vol 8 no 3 pp 224ndash239 2005
[2] M M Levitan and M R Wilson ldquoDeconvolution of pressureand rate data from gas reservoirs with significant pressuredepletionrdquo SPE Journal vol 17 no 3 pp 727ndash741 2012
[3] Y Liu and R N Horne ldquoInterpreting pressure and flow-ratedata from permanent downhole gauges by use of data-miningapproachesrdquo SPE Journal vol 18 no 1 pp 69ndash82 2013
[4] O Ogunrewo and A C Gringarten ldquoDeconvolution of welltest data in lean and rich gas condensate and volatile oil wellsbelow saturation pressurerdquo in Proceedings of the SPE AnnualTechnical Conference and Exhibition Document ID 166340-MS New Orleans La USA September-October 2013
[5] R C Earlougher Jr Advances in Well Test Analysis vol 5 ofMonogragh Series SPE Richardson Tex USA 1977
[6] F Kucuk and L Ayestaran ldquoAnalysis of simultaneously mea-sured pressure and sandface flow rate in transient well testingrdquoJournal of PetroleumTechnology vol 37 no 2 pp 323ndash334 1985
[7] M Onur M Cinar D IIk P P Valko T A Blasingame and PHegeman ldquoAn investigation of recent deconvolution methodsfor well-test data analysisrdquo SPE Journal vol 13 no 2 pp 226ndash247 2008
[8] F J KuchukMOnur andFHollaender ldquoPressure transient testdesign and interpretationrdquo Developments in Petroleum Sciencevol 57 pp 303ndash359 2010
[9] A C Gringarten D P Bourdet P A Landel and V J KniazeffldquoA comparison between different skin and wellbore storagetype-curves for early-time transient analysisrdquo in Proceedings ofthe SPE Annual Technical Conference and Exhibition Paper SPE8205 Las Vegas Nev USA 1979
[10] A C Gringarten ldquoFrom straight lines to deconvolution theevolution of the state of the art in well test analysisrdquo SPEReservoir Evaluation and Engineering vol 11 no 1 pp 41ndash622008
[11] J A Cumming D A Wooff T Whittle and A C GringartenldquoMultiple well deconvolutionrdquo in Proceedings of the SPE Annual
12 Abstract and Applied Analysis
Technical Conference and Exhibition Document ID 166458-MS New Orleans La USA September-October 2013
[12] O Bahabanian D Ilk N Hosseinpour-Zonoozi and T ABlasingame ldquoExplicit deconvolution of well test data distortedby wellbore storagerdquo in Proceedings of the SPE Annual TechnicalConference and Exhibition Paper SPE 103216 pp 4444ndash4453San Antonio Tex USA September 2006
[13] J Jones and A Chen ldquoSuccessful applications of pressure-ratedeconvolution in theCad-Nik tight gas formations of the BritishColumbia foothillsrdquo Journal of Canadian Petroleum Technologyvol 51 no 3 pp 176ndash192 2012
[14] D G Russell ldquoExtension of pressure buildup analysis methodsrdquoJournal of Petroleum Technology vol 18 no 12 pp 1624ndash16361966
[15] R E Gladfelter G W Tracy and L E Wilsey ldquoSelecting wellswhich will respond to production-stimulation treatmentrdquo inDrilling and Production Practice API Dallas Tex USA 1955
[16] M J Fetkovich and M E Vienot ldquoRate normalization ofbuildup pressure by using afterflow datardquo Journal of PetroleumTechnology vol 36 no 13 pp 2211ndash2224 1984
[17] J L Johnston Variable rate analysis of transient well testdata using semi-analytical methods [MS thesis] Texas AampMUniversity College Station Tex USA 1992
[18] A F van Everdingen ldquoThe skin effect and its influence on theproductive capacity of a wellrdquo Journal of Petroleum Technologyvol 5 no 6 pp 171ndash176 1953
[19] W Hurst ldquoEstablishment of the skin effect and its impedimentto fluid flow into a well borerdquo The Petroleum Engineer pp B6ndashB16 1953
[20] J A Joseph and L F Koederitz ldquoA simple nonlinear model forrepresentation of field transient responsesrdquo Paper SPE 11435SPE Richardson Tex USA 1982
[21] F J Kuchuk ldquoGladfelter deconvolutionrdquo SPE Formation Evalu-ation vol 5 no 3 pp 285ndash292 1990
[22] A F van Everdingen and W Hurst ldquoThe application of thelaplace transformation to flow problems in reservoirsrdquo Journalof Petroleum Technology vol 1 no 12 pp 305ndash324 1949
[23] D Bourdet T M Whittle A A Douglas and Y M Pirard ldquoAnew set of type curves simplifies well test analysisrdquo World Oilpp 95ndash106 1983
[24] D F Meunier C S Kabir and M J Wittmann ldquoGas well testanalysis use of normalized pseudovariablesrdquo SPE FormationEvaluation vol 2 no 4 pp 629ndash636 1987
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
8 Abstract and Applied Analysis
10minus3 10minus2 10minus1 100 101 102 103
Time (hr)
250
200
150
100
50
0
Dpwf wellbore storaged(Dpwf)d(ln(t)) wellbore storageDps explicit deconvolution methodd(Dps)d(ln(t)) explicit deconvolution methodDps material balance deconvolutiond(Dps)d(ln(t)) material balance deconvolutionDps no wellbore storaged(Dps)d(ln(t)) no wellbore storage
Dpwf
and
its lo
garit
hmic
der
ivat
ive (
psi)
Figure 7 Deconvolution for a drawdown test in a reservoir with asealing fault
10minus3 10minus2 10minus1 100 101 102 103
Time (hr)
103
102
101
100
Dpwf wellbore storaged(Dpwf)d(ln(t)) wellbore storageDps explicit deconvolution methodd(Dps)d(ln(t)) explicit deconvolution methodDps material balance deconvolutiond(Dps)d(ln(t)) material balance deconvolutionDps no wellbore storaged(Dps)d(ln(t)) no wellbore storage
Dpwf
and
its lo
garit
hmic
der
ivat
ive (
psi)
Figure 8 Deconvolution for a drawdown test in a reservoir with asealing fault
wellbore during buildup This leads to the result that oneshould plot
[119901119908119904 (Δ119905) minus 119901
119908119891 (Δ119905 = 0)]
[1 minus (11198622Δ119905)]
(A1)
versus log (Δ119905) in analyzing pressure buildup data during theearly fill-up period The denominator makes a correction for
104103102101100
Horner time ratio
800
700
600
500
400
300
200
100
0
Dpws
(psi)
Dpws wellbore storageDps explicit deconvolution methodDps material balance deconvolution
Figure 9 Restored pressures for a buildup test in a homogeneousreservoir
0
500
1000
1500
2000
2500
3000
3500
4000
105104103102101100
Horner time ratio
Dpa
(psi)
Dpaw wellbore storage explicit deconvolution method material balance deconvolution
DpasDpas
Figure 10 Restored pressures for a buildup test in a radial compositereservoir
the gradually decreasing flow into the wellbore The quantity1198622is obtained by trial and error as the value which makes the
curve straight at early times After obtaining the straight linesection the rest of the analysis is the same as for any otherpressure buildup This method has the advantage or requiresno additional data over that taken routinely during a shut intest
Due to the time consuming and unconfident of trial anderror to determine 119862
2in Russell method we modify this
method Russellrsquos wellbore storage correction is given as
[119901119908119904 (Δ119905) minus 119901
119908119891 (Δ119905 = 0)]
[1 minus (11198622Δ119905)]
= 119891 (Δ119905 = 1 hr) + 119898119904119897log (Δ119905)
(A2)
Abstract and Applied Analysis 9
Defining the following function and using least squarestechnique obtain the final result
119864 (1198622 119891119898)
=
119873
sum
119894=1
(Δ119901119908119904119894
(1 minus (11198622Δ119905119894))
minus 119891(1198622) minus 119898(119862
2) log(Δ119905
119894))
2
(A3)
Tominimize the function119864 it can differentiate119864with respectto 1198622 119891 and 119898 and setting the results equal to zero After
differentiation it reduces to
120597119864
1205971198622
= minus2
119873
sum
119894=1
(Δ119901119908119904119894
Δ119905119894(1198622minus (1Δ119905
119894))2)
times (Δ119901119908119904119894
(1 minus (11198622Δ119905119894))
minus 119891 (1198622)
minus119898 (1198622) log (Δ119905
119894) ) = 0
(A4)
120597119864
120597119891
= minus2
119873
sum
119894=1
(Δ119901119908119904119894
(1 minus (11198622Δ119905119894))
minus 119891 (1198622) minus 119898 (119862
2) log (Δ119905
119894))
= 0
(A5)120597119864
120597119898
= minus2
119873
sum
119894=1
log (Δ119905119894) (
Δ119901119908119904119894
(1 minus (11198622Δ119905119894))
minus 119891 (1198622)
minus 119898 (1198622) log (Δ119905
119894)) = 0
(A6)
From (A4) we define the following function
119883(1198622)
=
119873
sum
119894=1
(Δ119901119908119904119894
Δ119905119894(1198622minus (1Δ119905
119894))2)
times(Δ119901119908119904119894
(1minus (11198622Δ119905119894))minus119891 (119862
2) minus119898 (119862
2) log (Δ119905
119894))
= 0
(A7)
The roots of (A7) can be found via the following algorithm
1198622(119894+1)
= 1198622(119894)
minus119883 (1198622)
1198831015840 (1198622) (A8)
where
119883(1198622) =
119873
sum
119894=1
(Δ119901119908119904119894
Δ119905119894(1198622minus (1Δ119905
119894))2)
times (Δ119901119908119904119894
(1 minus (11198622Δ119905119894))
minus 119891 (1198622)
minus119898 (1198622) log (Δ119905
119894))
1198831015840(1198622) =
119873
sum
119894=1
(minus2Δ119901119908119904119894
Δ119905119894(1198622minus (1Δ119905
119894))3)
times (Δ119901119908119904119894
Δ119905119894(1198622minus (1Δ119905
119894))2
times [Δ119901119908119904119894
(1 minus (11198622Δ119905))
minus 119891 (1198622)
minus119898 (1198622) log (Δ119905
119894)])
+ (Δ119901119908119904119894
Δ119905119894(1198622minus (1Δ119905
119894))2)
times (minusΔ119901119908119904119894
Δ119905119894(1198622minus (1Δ119905
119894))3minus 1198911015840(1198622)
minus1198981015840(1198622) log (Δ119905
119894))
(A9)
where the unknown functions are given by the followingrelations
119891 (1198622) =
120573119906 (1198622) minus 120572V (119862
2)
1205732 minus 119873120572(A10)
whose derivative is
1198911015840(1198622) =
1205731199061015840(1198622) minus 120572V1015840 (119862
2)
1205732 minus 119873120572
119898 (1198622) =
119873119906 (1198622) minus 120573V (119862
2)
119873120572 minus 1205732
(A11)
where its derivative is
1198981015840(1198622) =
1198731199061015840(1198622) minus 120573V1015840 (119862
2)
119873120572 minus 1205732
119906 (1198622) =
119873
sum
119894=1
log (Δ119905119894)
Δ119901119908119904119894
(1 minus (11198622Δ119905119894))
(A12)
10 Abstract and Applied Analysis
and after differentiation it yields
1199061015840(1198622) = minus
119873
sum
119894=1
log (Δ119905
119894)
Δ119905119894
Δ119901119908119904119894
(1198622minus (1Δ119905
119894))2
V (1198622) =
119873
sum
119894=1
Δ119901119908119904119894
(1 minus (11198622Δ119905119894))
(A13)
which has the following derivative
V1015840 (1198622) = minus
119873
sum
119894=1
1
Δ119905119894
Δ119901119908119904119894
(1198622minus (1Δ119905
119894))2 (A14)
finally the values of 120572 and 120573 parameters are given by thefollowing formulas
120572 =
119873
sum
119894=1
(log(Δ119905119894))2
120573 =
119873
sum
119894=1
log (Δ119905119894)
(A15)
B General Formula for Inverse Problem
The convolution of (6) yields
119901119904119863
(119905119863) = int
119905119863
0
119870 (120591) 119901119908119863 (119905119863 minus 120591) 119889120591
= int
119905119863
0
119870(119905119863minus 120591) 119901
119908119863 (120591) 119889120591
(B1)
where
119870(119905119863) = 119871minus1(
1
119904119902 (119904)) (B2)
where 119871minus1 is the inverse Lapalce transform operator 119870(119905119863)
can be computed either from the Laplace transforms of119902119863(119905119863) data a curve-fitted equation of 119902
119863(119905119863) data or using
another technique with a relation between119870(119905119863) and 119902
119863(119905119863)
However it would be time consuming to invert all the 119902119863(119905119863)
data in Laplace space and transform it back to real space inaccordance with (B2) Once an approximation is obtained itwill be easy to compute119870(119905
119863) and integrate (B1) to determine
119901119904119863(119905119863) However finding some approximation functions for
119902119863(119905119863) data may be difficult We provide an approach to
overcome this problemThis idea is based on finding anotherrelation between 119870(119905
119863) and 119902
119863(119905119863) that is
ℓ (1) =1
119904=
1
119904119902119863(119904)
119902119863(119904) = ℓ (int
119905119863
0
119870 (120591) 119902119863 (119905119863 minus 120591) 119889120591)
= ℓ(int
119905119863
0
119870(119905119863minus 120591) 119902
119863 (120591) 119889120591)
(B3)or
1 = int
119905119863
0
119870 (120591) 119902119863 (119905119863 minus 120591) 119889120591 = int
119905119863
0
119870(119905119863minus 120591) 119902
119863 (120591) 119889120591
(B4)
The other relations that can be used are
ℓ (119905119863) =
1
1199042=
1
1199042119902119863(119904)
119902119863(119904) = ℓ (int
119905119863
0
119871 (120591) 119902119863 (119905119863 minus 120591) 119889120591)
= ℓ(int
119905119863
0
119871 (119905119863minus 120591) 119902
119863 (120591) 119889120591)
(B5)
or
119905119863= int
119905119863
0
119871 (120591) 119902119863(119905119863minus 120591) 119889120591 = int
119905119863
0
119871 (119905119863minus 120591) 119902
119863(120591) 119889120591
(B6)
ℓ (1) =1
119904=
1
1199042119902119863(119904)
119904119902119863(119904) = ℓ (int
119905119863
0
119871 (120591) 1199021015840
119863(119905119863minus 120591) 119889120591)
= ℓ(int
119905119863
0
119871 (119905119863minus 120591) 119902
1015840
119863(120591) 119889120591)
(B7)
or
1 = int
119905119863
0
119871 (120591) 1199021015840
119863(119905119863minus 120591) 119889120591 = int
119905119863
0
119871 (119905119863minus 120591) 119902
1015840
119863(120591) 119889120591
(B8)
where
119871 (119905119863) = ℓminus1(
1
1199042119902 (119904)) (B9)
Equation (B4) can be used simultaneously with (B1) toanalyze the deconvolution problem
Equation (11) can be written as
119901119904119863
(119904) = 119904119901119908119863
(119904)1
1199042119902119863(119904)
(B10)
or
119901119904119863
(119905119863) = int
119905119863
0
119871 (120591) 1199011015840
119908119863(119905119863minus 120591) 119889120591
= int
119905119863
0
119871 (119905119863minus 120591) 119901
1015840
119908119863(120591) 119889120591
(B11)
The applications of (B6) and (B8) are with (B11)For justifying the rate normalizationmethod andmaterial
balance deconvolution it can use (B1) (B4) ((B11) can alsobe used with (B6) or (B8) for this goal) and generalizedmean value theorem for integrals This theorem is expressedas follows
Theorem 1 Let 119891 and 119892 be continuous on [119886 119887] and supposethat 119892 does not change sign on [119886 119887] Then there is a point 119888 in[119886 119887] such that
int
119887
119886
119891 (119905) 119892 (119905) 119889119905 = 119891 (119888) int
119887
119886
119892 (119905) 119889119905 (B12)
Abstract and Applied Analysis 11
By Theorem 1 there are some values 1198881and 1198882between 0
and 119905119863such that
119901119904119863
(119905119863) = 119901119908119863
(1198881) int
119905119863
0
119870(119905119863minus 120591) 119889120591 (B13)
1 = 119902119863(1198882) int
119905119863
0
119870(119905119863minus 120591) 119889120591 (B14)
By setting 1198881
= 1198882
= 119905119863and dividing the right and left
side of (B13) respectively the rate normalization techniqueis developed However the values 119888
1and 1198882are not equal
to each other and are not equal to 119905119863
in general Thisis the reason of error in rate normalization and materialbalance deconvolution methods 119888
1and 1198882are less than 119905
119863in
these methods for the duration of wellbore storage distortionduring a pressure transient test Since these conditions areapproximately satisfied in material balance deconvolutionthe results of material balance are more accurate and moreconfident than rate normalization
Nomenclature
API API gravity ∘API119861119900 Oil formation volume factor RBSTB
119888119905 Total isothermal compressibility factor
psiminus1
1198622 Arbitrary constant hr
minus1
119862119863 Wellbore storage constant dimensionless
119862119904 Wellbore storage coefficient bblpsi
ℎ Formation thickness ftHTR Horner time ratio dimensionless119896 Formation permeability md119870 Kernel of the convolution integral119871 Kernel of the convolution integral119873119901 Cumulative oil production bbl
119901 Pressure psi119901119886 Normalized pseudopressure psi
119901 Average reservoir pressure psi119901119863 Formation pressure dimensionless
119901119889119901 Dewpoint pressure psi
119901119894 Initial reservoir pressure psi
119901119904119863 Formation pressure including skin
dimensionless119901119904 Formation pressure including skin psi
119901119908119863
Pressure draw down dimensionless119901119908119891 Flowing bottom hole pressure psi
119901119908119904 Shut-in bottom hole pressure psi
119902119863 Sandface flow rate dimensionless
119902119892 Gas flow rate MscfD
119902119900 Oil flow rate STBD
119903119908 Wellbore radius ft
119904 Laplace variable119878 Skin factor119905 Time hr119905119863 Time dimensionless
119905119901 Production time hr
119879 Reservoir temperature ∘F
119879119889119901 Dewpoint temperature ∘F
120572 Explicit deconvolution variabledimensionless
120573 Beta deconvolution variabledimensionless
120574119892 Gas gravity (air = 10)
120582 Interporosity flow coefficient of dualporosity reservoir
120601 Porosity of reservoir rock dimensionless120583119892 Gas viscosity cp
120583119900 Oil viscosity cp
120596 Storativity ratio of dual porosity reservoirminus Laplace transform of a function1015840 Derivative of a function
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
References
[1] Y Cheng W J Lee and D A McVay ldquoFast-Fourier-transform-based deconvolution for interpretation of pressure-transient-test data dominated by Wellbore storagerdquo SPE Reservoir Eval-uation and Engineering vol 8 no 3 pp 224ndash239 2005
[2] M M Levitan and M R Wilson ldquoDeconvolution of pressureand rate data from gas reservoirs with significant pressuredepletionrdquo SPE Journal vol 17 no 3 pp 727ndash741 2012
[3] Y Liu and R N Horne ldquoInterpreting pressure and flow-ratedata from permanent downhole gauges by use of data-miningapproachesrdquo SPE Journal vol 18 no 1 pp 69ndash82 2013
[4] O Ogunrewo and A C Gringarten ldquoDeconvolution of welltest data in lean and rich gas condensate and volatile oil wellsbelow saturation pressurerdquo in Proceedings of the SPE AnnualTechnical Conference and Exhibition Document ID 166340-MS New Orleans La USA September-October 2013
[5] R C Earlougher Jr Advances in Well Test Analysis vol 5 ofMonogragh Series SPE Richardson Tex USA 1977
[6] F Kucuk and L Ayestaran ldquoAnalysis of simultaneously mea-sured pressure and sandface flow rate in transient well testingrdquoJournal of PetroleumTechnology vol 37 no 2 pp 323ndash334 1985
[7] M Onur M Cinar D IIk P P Valko T A Blasingame and PHegeman ldquoAn investigation of recent deconvolution methodsfor well-test data analysisrdquo SPE Journal vol 13 no 2 pp 226ndash247 2008
[8] F J KuchukMOnur andFHollaender ldquoPressure transient testdesign and interpretationrdquo Developments in Petroleum Sciencevol 57 pp 303ndash359 2010
[9] A C Gringarten D P Bourdet P A Landel and V J KniazeffldquoA comparison between different skin and wellbore storagetype-curves for early-time transient analysisrdquo in Proceedings ofthe SPE Annual Technical Conference and Exhibition Paper SPE8205 Las Vegas Nev USA 1979
[10] A C Gringarten ldquoFrom straight lines to deconvolution theevolution of the state of the art in well test analysisrdquo SPEReservoir Evaluation and Engineering vol 11 no 1 pp 41ndash622008
[11] J A Cumming D A Wooff T Whittle and A C GringartenldquoMultiple well deconvolutionrdquo in Proceedings of the SPE Annual
12 Abstract and Applied Analysis
Technical Conference and Exhibition Document ID 166458-MS New Orleans La USA September-October 2013
[12] O Bahabanian D Ilk N Hosseinpour-Zonoozi and T ABlasingame ldquoExplicit deconvolution of well test data distortedby wellbore storagerdquo in Proceedings of the SPE Annual TechnicalConference and Exhibition Paper SPE 103216 pp 4444ndash4453San Antonio Tex USA September 2006
[13] J Jones and A Chen ldquoSuccessful applications of pressure-ratedeconvolution in theCad-Nik tight gas formations of the BritishColumbia foothillsrdquo Journal of Canadian Petroleum Technologyvol 51 no 3 pp 176ndash192 2012
[14] D G Russell ldquoExtension of pressure buildup analysis methodsrdquoJournal of Petroleum Technology vol 18 no 12 pp 1624ndash16361966
[15] R E Gladfelter G W Tracy and L E Wilsey ldquoSelecting wellswhich will respond to production-stimulation treatmentrdquo inDrilling and Production Practice API Dallas Tex USA 1955
[16] M J Fetkovich and M E Vienot ldquoRate normalization ofbuildup pressure by using afterflow datardquo Journal of PetroleumTechnology vol 36 no 13 pp 2211ndash2224 1984
[17] J L Johnston Variable rate analysis of transient well testdata using semi-analytical methods [MS thesis] Texas AampMUniversity College Station Tex USA 1992
[18] A F van Everdingen ldquoThe skin effect and its influence on theproductive capacity of a wellrdquo Journal of Petroleum Technologyvol 5 no 6 pp 171ndash176 1953
[19] W Hurst ldquoEstablishment of the skin effect and its impedimentto fluid flow into a well borerdquo The Petroleum Engineer pp B6ndashB16 1953
[20] J A Joseph and L F Koederitz ldquoA simple nonlinear model forrepresentation of field transient responsesrdquo Paper SPE 11435SPE Richardson Tex USA 1982
[21] F J Kuchuk ldquoGladfelter deconvolutionrdquo SPE Formation Evalu-ation vol 5 no 3 pp 285ndash292 1990
[22] A F van Everdingen and W Hurst ldquoThe application of thelaplace transformation to flow problems in reservoirsrdquo Journalof Petroleum Technology vol 1 no 12 pp 305ndash324 1949
[23] D Bourdet T M Whittle A A Douglas and Y M Pirard ldquoAnew set of type curves simplifies well test analysisrdquo World Oilpp 95ndash106 1983
[24] D F Meunier C S Kabir and M J Wittmann ldquoGas well testanalysis use of normalized pseudovariablesrdquo SPE FormationEvaluation vol 2 no 4 pp 629ndash636 1987
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
Abstract and Applied Analysis 9
Defining the following function and using least squarestechnique obtain the final result
119864 (1198622 119891119898)
=
119873
sum
119894=1
(Δ119901119908119904119894
(1 minus (11198622Δ119905119894))
minus 119891(1198622) minus 119898(119862
2) log(Δ119905
119894))
2
(A3)
Tominimize the function119864 it can differentiate119864with respectto 1198622 119891 and 119898 and setting the results equal to zero After
differentiation it reduces to
120597119864
1205971198622
= minus2
119873
sum
119894=1
(Δ119901119908119904119894
Δ119905119894(1198622minus (1Δ119905
119894))2)
times (Δ119901119908119904119894
(1 minus (11198622Δ119905119894))
minus 119891 (1198622)
minus119898 (1198622) log (Δ119905
119894) ) = 0
(A4)
120597119864
120597119891
= minus2
119873
sum
119894=1
(Δ119901119908119904119894
(1 minus (11198622Δ119905119894))
minus 119891 (1198622) minus 119898 (119862
2) log (Δ119905
119894))
= 0
(A5)120597119864
120597119898
= minus2
119873
sum
119894=1
log (Δ119905119894) (
Δ119901119908119904119894
(1 minus (11198622Δ119905119894))
minus 119891 (1198622)
minus 119898 (1198622) log (Δ119905
119894)) = 0
(A6)
From (A4) we define the following function
119883(1198622)
=
119873
sum
119894=1
(Δ119901119908119904119894
Δ119905119894(1198622minus (1Δ119905
119894))2)
times(Δ119901119908119904119894
(1minus (11198622Δ119905119894))minus119891 (119862
2) minus119898 (119862
2) log (Δ119905
119894))
= 0
(A7)
The roots of (A7) can be found via the following algorithm
1198622(119894+1)
= 1198622(119894)
minus119883 (1198622)
1198831015840 (1198622) (A8)
where
119883(1198622) =
119873
sum
119894=1
(Δ119901119908119904119894
Δ119905119894(1198622minus (1Δ119905
119894))2)
times (Δ119901119908119904119894
(1 minus (11198622Δ119905119894))
minus 119891 (1198622)
minus119898 (1198622) log (Δ119905
119894))
1198831015840(1198622) =
119873
sum
119894=1
(minus2Δ119901119908119904119894
Δ119905119894(1198622minus (1Δ119905
119894))3)
times (Δ119901119908119904119894
Δ119905119894(1198622minus (1Δ119905
119894))2
times [Δ119901119908119904119894
(1 minus (11198622Δ119905))
minus 119891 (1198622)
minus119898 (1198622) log (Δ119905
119894)])
+ (Δ119901119908119904119894
Δ119905119894(1198622minus (1Δ119905
119894))2)
times (minusΔ119901119908119904119894
Δ119905119894(1198622minus (1Δ119905
119894))3minus 1198911015840(1198622)
minus1198981015840(1198622) log (Δ119905
119894))
(A9)
where the unknown functions are given by the followingrelations
119891 (1198622) =
120573119906 (1198622) minus 120572V (119862
2)
1205732 minus 119873120572(A10)
whose derivative is
1198911015840(1198622) =
1205731199061015840(1198622) minus 120572V1015840 (119862
2)
1205732 minus 119873120572
119898 (1198622) =
119873119906 (1198622) minus 120573V (119862
2)
119873120572 minus 1205732
(A11)
where its derivative is
1198981015840(1198622) =
1198731199061015840(1198622) minus 120573V1015840 (119862
2)
119873120572 minus 1205732
119906 (1198622) =
119873
sum
119894=1
log (Δ119905119894)
Δ119901119908119904119894
(1 minus (11198622Δ119905119894))
(A12)
10 Abstract and Applied Analysis
and after differentiation it yields
1199061015840(1198622) = minus
119873
sum
119894=1
log (Δ119905
119894)
Δ119905119894
Δ119901119908119904119894
(1198622minus (1Δ119905
119894))2
V (1198622) =
119873
sum
119894=1
Δ119901119908119904119894
(1 minus (11198622Δ119905119894))
(A13)
which has the following derivative
V1015840 (1198622) = minus
119873
sum
119894=1
1
Δ119905119894
Δ119901119908119904119894
(1198622minus (1Δ119905
119894))2 (A14)
finally the values of 120572 and 120573 parameters are given by thefollowing formulas
120572 =
119873
sum
119894=1
(log(Δ119905119894))2
120573 =
119873
sum
119894=1
log (Δ119905119894)
(A15)
B General Formula for Inverse Problem
The convolution of (6) yields
119901119904119863
(119905119863) = int
119905119863
0
119870 (120591) 119901119908119863 (119905119863 minus 120591) 119889120591
= int
119905119863
0
119870(119905119863minus 120591) 119901
119908119863 (120591) 119889120591
(B1)
where
119870(119905119863) = 119871minus1(
1
119904119902 (119904)) (B2)
where 119871minus1 is the inverse Lapalce transform operator 119870(119905119863)
can be computed either from the Laplace transforms of119902119863(119905119863) data a curve-fitted equation of 119902
119863(119905119863) data or using
another technique with a relation between119870(119905119863) and 119902
119863(119905119863)
However it would be time consuming to invert all the 119902119863(119905119863)
data in Laplace space and transform it back to real space inaccordance with (B2) Once an approximation is obtained itwill be easy to compute119870(119905
119863) and integrate (B1) to determine
119901119904119863(119905119863) However finding some approximation functions for
119902119863(119905119863) data may be difficult We provide an approach to
overcome this problemThis idea is based on finding anotherrelation between 119870(119905
119863) and 119902
119863(119905119863) that is
ℓ (1) =1
119904=
1
119904119902119863(119904)
119902119863(119904) = ℓ (int
119905119863
0
119870 (120591) 119902119863 (119905119863 minus 120591) 119889120591)
= ℓ(int
119905119863
0
119870(119905119863minus 120591) 119902
119863 (120591) 119889120591)
(B3)or
1 = int
119905119863
0
119870 (120591) 119902119863 (119905119863 minus 120591) 119889120591 = int
119905119863
0
119870(119905119863minus 120591) 119902
119863 (120591) 119889120591
(B4)
The other relations that can be used are
ℓ (119905119863) =
1
1199042=
1
1199042119902119863(119904)
119902119863(119904) = ℓ (int
119905119863
0
119871 (120591) 119902119863 (119905119863 minus 120591) 119889120591)
= ℓ(int
119905119863
0
119871 (119905119863minus 120591) 119902
119863 (120591) 119889120591)
(B5)
or
119905119863= int
119905119863
0
119871 (120591) 119902119863(119905119863minus 120591) 119889120591 = int
119905119863
0
119871 (119905119863minus 120591) 119902
119863(120591) 119889120591
(B6)
ℓ (1) =1
119904=
1
1199042119902119863(119904)
119904119902119863(119904) = ℓ (int
119905119863
0
119871 (120591) 1199021015840
119863(119905119863minus 120591) 119889120591)
= ℓ(int
119905119863
0
119871 (119905119863minus 120591) 119902
1015840
119863(120591) 119889120591)
(B7)
or
1 = int
119905119863
0
119871 (120591) 1199021015840
119863(119905119863minus 120591) 119889120591 = int
119905119863
0
119871 (119905119863minus 120591) 119902
1015840
119863(120591) 119889120591
(B8)
where
119871 (119905119863) = ℓminus1(
1
1199042119902 (119904)) (B9)
Equation (B4) can be used simultaneously with (B1) toanalyze the deconvolution problem
Equation (11) can be written as
119901119904119863
(119904) = 119904119901119908119863
(119904)1
1199042119902119863(119904)
(B10)
or
119901119904119863
(119905119863) = int
119905119863
0
119871 (120591) 1199011015840
119908119863(119905119863minus 120591) 119889120591
= int
119905119863
0
119871 (119905119863minus 120591) 119901
1015840
119908119863(120591) 119889120591
(B11)
The applications of (B6) and (B8) are with (B11)For justifying the rate normalizationmethod andmaterial
balance deconvolution it can use (B1) (B4) ((B11) can alsobe used with (B6) or (B8) for this goal) and generalizedmean value theorem for integrals This theorem is expressedas follows
Theorem 1 Let 119891 and 119892 be continuous on [119886 119887] and supposethat 119892 does not change sign on [119886 119887] Then there is a point 119888 in[119886 119887] such that
int
119887
119886
119891 (119905) 119892 (119905) 119889119905 = 119891 (119888) int
119887
119886
119892 (119905) 119889119905 (B12)
Abstract and Applied Analysis 11
By Theorem 1 there are some values 1198881and 1198882between 0
and 119905119863such that
119901119904119863
(119905119863) = 119901119908119863
(1198881) int
119905119863
0
119870(119905119863minus 120591) 119889120591 (B13)
1 = 119902119863(1198882) int
119905119863
0
119870(119905119863minus 120591) 119889120591 (B14)
By setting 1198881
= 1198882
= 119905119863and dividing the right and left
side of (B13) respectively the rate normalization techniqueis developed However the values 119888
1and 1198882are not equal
to each other and are not equal to 119905119863
in general Thisis the reason of error in rate normalization and materialbalance deconvolution methods 119888
1and 1198882are less than 119905
119863in
these methods for the duration of wellbore storage distortionduring a pressure transient test Since these conditions areapproximately satisfied in material balance deconvolutionthe results of material balance are more accurate and moreconfident than rate normalization
Nomenclature
API API gravity ∘API119861119900 Oil formation volume factor RBSTB
119888119905 Total isothermal compressibility factor
psiminus1
1198622 Arbitrary constant hr
minus1
119862119863 Wellbore storage constant dimensionless
119862119904 Wellbore storage coefficient bblpsi
ℎ Formation thickness ftHTR Horner time ratio dimensionless119896 Formation permeability md119870 Kernel of the convolution integral119871 Kernel of the convolution integral119873119901 Cumulative oil production bbl
119901 Pressure psi119901119886 Normalized pseudopressure psi
119901 Average reservoir pressure psi119901119863 Formation pressure dimensionless
119901119889119901 Dewpoint pressure psi
119901119894 Initial reservoir pressure psi
119901119904119863 Formation pressure including skin
dimensionless119901119904 Formation pressure including skin psi
119901119908119863
Pressure draw down dimensionless119901119908119891 Flowing bottom hole pressure psi
119901119908119904 Shut-in bottom hole pressure psi
119902119863 Sandface flow rate dimensionless
119902119892 Gas flow rate MscfD
119902119900 Oil flow rate STBD
119903119908 Wellbore radius ft
119904 Laplace variable119878 Skin factor119905 Time hr119905119863 Time dimensionless
119905119901 Production time hr
119879 Reservoir temperature ∘F
119879119889119901 Dewpoint temperature ∘F
120572 Explicit deconvolution variabledimensionless
120573 Beta deconvolution variabledimensionless
120574119892 Gas gravity (air = 10)
120582 Interporosity flow coefficient of dualporosity reservoir
120601 Porosity of reservoir rock dimensionless120583119892 Gas viscosity cp
120583119900 Oil viscosity cp
120596 Storativity ratio of dual porosity reservoirminus Laplace transform of a function1015840 Derivative of a function
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
References
[1] Y Cheng W J Lee and D A McVay ldquoFast-Fourier-transform-based deconvolution for interpretation of pressure-transient-test data dominated by Wellbore storagerdquo SPE Reservoir Eval-uation and Engineering vol 8 no 3 pp 224ndash239 2005
[2] M M Levitan and M R Wilson ldquoDeconvolution of pressureand rate data from gas reservoirs with significant pressuredepletionrdquo SPE Journal vol 17 no 3 pp 727ndash741 2012
[3] Y Liu and R N Horne ldquoInterpreting pressure and flow-ratedata from permanent downhole gauges by use of data-miningapproachesrdquo SPE Journal vol 18 no 1 pp 69ndash82 2013
[4] O Ogunrewo and A C Gringarten ldquoDeconvolution of welltest data in lean and rich gas condensate and volatile oil wellsbelow saturation pressurerdquo in Proceedings of the SPE AnnualTechnical Conference and Exhibition Document ID 166340-MS New Orleans La USA September-October 2013
[5] R C Earlougher Jr Advances in Well Test Analysis vol 5 ofMonogragh Series SPE Richardson Tex USA 1977
[6] F Kucuk and L Ayestaran ldquoAnalysis of simultaneously mea-sured pressure and sandface flow rate in transient well testingrdquoJournal of PetroleumTechnology vol 37 no 2 pp 323ndash334 1985
[7] M Onur M Cinar D IIk P P Valko T A Blasingame and PHegeman ldquoAn investigation of recent deconvolution methodsfor well-test data analysisrdquo SPE Journal vol 13 no 2 pp 226ndash247 2008
[8] F J KuchukMOnur andFHollaender ldquoPressure transient testdesign and interpretationrdquo Developments in Petroleum Sciencevol 57 pp 303ndash359 2010
[9] A C Gringarten D P Bourdet P A Landel and V J KniazeffldquoA comparison between different skin and wellbore storagetype-curves for early-time transient analysisrdquo in Proceedings ofthe SPE Annual Technical Conference and Exhibition Paper SPE8205 Las Vegas Nev USA 1979
[10] A C Gringarten ldquoFrom straight lines to deconvolution theevolution of the state of the art in well test analysisrdquo SPEReservoir Evaluation and Engineering vol 11 no 1 pp 41ndash622008
[11] J A Cumming D A Wooff T Whittle and A C GringartenldquoMultiple well deconvolutionrdquo in Proceedings of the SPE Annual
12 Abstract and Applied Analysis
Technical Conference and Exhibition Document ID 166458-MS New Orleans La USA September-October 2013
[12] O Bahabanian D Ilk N Hosseinpour-Zonoozi and T ABlasingame ldquoExplicit deconvolution of well test data distortedby wellbore storagerdquo in Proceedings of the SPE Annual TechnicalConference and Exhibition Paper SPE 103216 pp 4444ndash4453San Antonio Tex USA September 2006
[13] J Jones and A Chen ldquoSuccessful applications of pressure-ratedeconvolution in theCad-Nik tight gas formations of the BritishColumbia foothillsrdquo Journal of Canadian Petroleum Technologyvol 51 no 3 pp 176ndash192 2012
[14] D G Russell ldquoExtension of pressure buildup analysis methodsrdquoJournal of Petroleum Technology vol 18 no 12 pp 1624ndash16361966
[15] R E Gladfelter G W Tracy and L E Wilsey ldquoSelecting wellswhich will respond to production-stimulation treatmentrdquo inDrilling and Production Practice API Dallas Tex USA 1955
[16] M J Fetkovich and M E Vienot ldquoRate normalization ofbuildup pressure by using afterflow datardquo Journal of PetroleumTechnology vol 36 no 13 pp 2211ndash2224 1984
[17] J L Johnston Variable rate analysis of transient well testdata using semi-analytical methods [MS thesis] Texas AampMUniversity College Station Tex USA 1992
[18] A F van Everdingen ldquoThe skin effect and its influence on theproductive capacity of a wellrdquo Journal of Petroleum Technologyvol 5 no 6 pp 171ndash176 1953
[19] W Hurst ldquoEstablishment of the skin effect and its impedimentto fluid flow into a well borerdquo The Petroleum Engineer pp B6ndashB16 1953
[20] J A Joseph and L F Koederitz ldquoA simple nonlinear model forrepresentation of field transient responsesrdquo Paper SPE 11435SPE Richardson Tex USA 1982
[21] F J Kuchuk ldquoGladfelter deconvolutionrdquo SPE Formation Evalu-ation vol 5 no 3 pp 285ndash292 1990
[22] A F van Everdingen and W Hurst ldquoThe application of thelaplace transformation to flow problems in reservoirsrdquo Journalof Petroleum Technology vol 1 no 12 pp 305ndash324 1949
[23] D Bourdet T M Whittle A A Douglas and Y M Pirard ldquoAnew set of type curves simplifies well test analysisrdquo World Oilpp 95ndash106 1983
[24] D F Meunier C S Kabir and M J Wittmann ldquoGas well testanalysis use of normalized pseudovariablesrdquo SPE FormationEvaluation vol 2 no 4 pp 629ndash636 1987
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
10 Abstract and Applied Analysis
and after differentiation it yields
1199061015840(1198622) = minus
119873
sum
119894=1
log (Δ119905
119894)
Δ119905119894
Δ119901119908119904119894
(1198622minus (1Δ119905
119894))2
V (1198622) =
119873
sum
119894=1
Δ119901119908119904119894
(1 minus (11198622Δ119905119894))
(A13)
which has the following derivative
V1015840 (1198622) = minus
119873
sum
119894=1
1
Δ119905119894
Δ119901119908119904119894
(1198622minus (1Δ119905
119894))2 (A14)
finally the values of 120572 and 120573 parameters are given by thefollowing formulas
120572 =
119873
sum
119894=1
(log(Δ119905119894))2
120573 =
119873
sum
119894=1
log (Δ119905119894)
(A15)
B General Formula for Inverse Problem
The convolution of (6) yields
119901119904119863
(119905119863) = int
119905119863
0
119870 (120591) 119901119908119863 (119905119863 minus 120591) 119889120591
= int
119905119863
0
119870(119905119863minus 120591) 119901
119908119863 (120591) 119889120591
(B1)
where
119870(119905119863) = 119871minus1(
1
119904119902 (119904)) (B2)
where 119871minus1 is the inverse Lapalce transform operator 119870(119905119863)
can be computed either from the Laplace transforms of119902119863(119905119863) data a curve-fitted equation of 119902
119863(119905119863) data or using
another technique with a relation between119870(119905119863) and 119902
119863(119905119863)
However it would be time consuming to invert all the 119902119863(119905119863)
data in Laplace space and transform it back to real space inaccordance with (B2) Once an approximation is obtained itwill be easy to compute119870(119905
119863) and integrate (B1) to determine
119901119904119863(119905119863) However finding some approximation functions for
119902119863(119905119863) data may be difficult We provide an approach to
overcome this problemThis idea is based on finding anotherrelation between 119870(119905
119863) and 119902
119863(119905119863) that is
ℓ (1) =1
119904=
1
119904119902119863(119904)
119902119863(119904) = ℓ (int
119905119863
0
119870 (120591) 119902119863 (119905119863 minus 120591) 119889120591)
= ℓ(int
119905119863
0
119870(119905119863minus 120591) 119902
119863 (120591) 119889120591)
(B3)or
1 = int
119905119863
0
119870 (120591) 119902119863 (119905119863 minus 120591) 119889120591 = int
119905119863
0
119870(119905119863minus 120591) 119902
119863 (120591) 119889120591
(B4)
The other relations that can be used are
ℓ (119905119863) =
1
1199042=
1
1199042119902119863(119904)
119902119863(119904) = ℓ (int
119905119863
0
119871 (120591) 119902119863 (119905119863 minus 120591) 119889120591)
= ℓ(int
119905119863
0
119871 (119905119863minus 120591) 119902
119863 (120591) 119889120591)
(B5)
or
119905119863= int
119905119863
0
119871 (120591) 119902119863(119905119863minus 120591) 119889120591 = int
119905119863
0
119871 (119905119863minus 120591) 119902
119863(120591) 119889120591
(B6)
ℓ (1) =1
119904=
1
1199042119902119863(119904)
119904119902119863(119904) = ℓ (int
119905119863
0
119871 (120591) 1199021015840
119863(119905119863minus 120591) 119889120591)
= ℓ(int
119905119863
0
119871 (119905119863minus 120591) 119902
1015840
119863(120591) 119889120591)
(B7)
or
1 = int
119905119863
0
119871 (120591) 1199021015840
119863(119905119863minus 120591) 119889120591 = int
119905119863
0
119871 (119905119863minus 120591) 119902
1015840
119863(120591) 119889120591
(B8)
where
119871 (119905119863) = ℓminus1(
1
1199042119902 (119904)) (B9)
Equation (B4) can be used simultaneously with (B1) toanalyze the deconvolution problem
Equation (11) can be written as
119901119904119863
(119904) = 119904119901119908119863
(119904)1
1199042119902119863(119904)
(B10)
or
119901119904119863
(119905119863) = int
119905119863
0
119871 (120591) 1199011015840
119908119863(119905119863minus 120591) 119889120591
= int
119905119863
0
119871 (119905119863minus 120591) 119901
1015840
119908119863(120591) 119889120591
(B11)
The applications of (B6) and (B8) are with (B11)For justifying the rate normalizationmethod andmaterial
balance deconvolution it can use (B1) (B4) ((B11) can alsobe used with (B6) or (B8) for this goal) and generalizedmean value theorem for integrals This theorem is expressedas follows
Theorem 1 Let 119891 and 119892 be continuous on [119886 119887] and supposethat 119892 does not change sign on [119886 119887] Then there is a point 119888 in[119886 119887] such that
int
119887
119886
119891 (119905) 119892 (119905) 119889119905 = 119891 (119888) int
119887
119886
119892 (119905) 119889119905 (B12)
Abstract and Applied Analysis 11
By Theorem 1 there are some values 1198881and 1198882between 0
and 119905119863such that
119901119904119863
(119905119863) = 119901119908119863
(1198881) int
119905119863
0
119870(119905119863minus 120591) 119889120591 (B13)
1 = 119902119863(1198882) int
119905119863
0
119870(119905119863minus 120591) 119889120591 (B14)
By setting 1198881
= 1198882
= 119905119863and dividing the right and left
side of (B13) respectively the rate normalization techniqueis developed However the values 119888
1and 1198882are not equal
to each other and are not equal to 119905119863
in general Thisis the reason of error in rate normalization and materialbalance deconvolution methods 119888
1and 1198882are less than 119905
119863in
these methods for the duration of wellbore storage distortionduring a pressure transient test Since these conditions areapproximately satisfied in material balance deconvolutionthe results of material balance are more accurate and moreconfident than rate normalization
Nomenclature
API API gravity ∘API119861119900 Oil formation volume factor RBSTB
119888119905 Total isothermal compressibility factor
psiminus1
1198622 Arbitrary constant hr
minus1
119862119863 Wellbore storage constant dimensionless
119862119904 Wellbore storage coefficient bblpsi
ℎ Formation thickness ftHTR Horner time ratio dimensionless119896 Formation permeability md119870 Kernel of the convolution integral119871 Kernel of the convolution integral119873119901 Cumulative oil production bbl
119901 Pressure psi119901119886 Normalized pseudopressure psi
119901 Average reservoir pressure psi119901119863 Formation pressure dimensionless
119901119889119901 Dewpoint pressure psi
119901119894 Initial reservoir pressure psi
119901119904119863 Formation pressure including skin
dimensionless119901119904 Formation pressure including skin psi
119901119908119863
Pressure draw down dimensionless119901119908119891 Flowing bottom hole pressure psi
119901119908119904 Shut-in bottom hole pressure psi
119902119863 Sandface flow rate dimensionless
119902119892 Gas flow rate MscfD
119902119900 Oil flow rate STBD
119903119908 Wellbore radius ft
119904 Laplace variable119878 Skin factor119905 Time hr119905119863 Time dimensionless
119905119901 Production time hr
119879 Reservoir temperature ∘F
119879119889119901 Dewpoint temperature ∘F
120572 Explicit deconvolution variabledimensionless
120573 Beta deconvolution variabledimensionless
120574119892 Gas gravity (air = 10)
120582 Interporosity flow coefficient of dualporosity reservoir
120601 Porosity of reservoir rock dimensionless120583119892 Gas viscosity cp
120583119900 Oil viscosity cp
120596 Storativity ratio of dual porosity reservoirminus Laplace transform of a function1015840 Derivative of a function
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
References
[1] Y Cheng W J Lee and D A McVay ldquoFast-Fourier-transform-based deconvolution for interpretation of pressure-transient-test data dominated by Wellbore storagerdquo SPE Reservoir Eval-uation and Engineering vol 8 no 3 pp 224ndash239 2005
[2] M M Levitan and M R Wilson ldquoDeconvolution of pressureand rate data from gas reservoirs with significant pressuredepletionrdquo SPE Journal vol 17 no 3 pp 727ndash741 2012
[3] Y Liu and R N Horne ldquoInterpreting pressure and flow-ratedata from permanent downhole gauges by use of data-miningapproachesrdquo SPE Journal vol 18 no 1 pp 69ndash82 2013
[4] O Ogunrewo and A C Gringarten ldquoDeconvolution of welltest data in lean and rich gas condensate and volatile oil wellsbelow saturation pressurerdquo in Proceedings of the SPE AnnualTechnical Conference and Exhibition Document ID 166340-MS New Orleans La USA September-October 2013
[5] R C Earlougher Jr Advances in Well Test Analysis vol 5 ofMonogragh Series SPE Richardson Tex USA 1977
[6] F Kucuk and L Ayestaran ldquoAnalysis of simultaneously mea-sured pressure and sandface flow rate in transient well testingrdquoJournal of PetroleumTechnology vol 37 no 2 pp 323ndash334 1985
[7] M Onur M Cinar D IIk P P Valko T A Blasingame and PHegeman ldquoAn investigation of recent deconvolution methodsfor well-test data analysisrdquo SPE Journal vol 13 no 2 pp 226ndash247 2008
[8] F J KuchukMOnur andFHollaender ldquoPressure transient testdesign and interpretationrdquo Developments in Petroleum Sciencevol 57 pp 303ndash359 2010
[9] A C Gringarten D P Bourdet P A Landel and V J KniazeffldquoA comparison between different skin and wellbore storagetype-curves for early-time transient analysisrdquo in Proceedings ofthe SPE Annual Technical Conference and Exhibition Paper SPE8205 Las Vegas Nev USA 1979
[10] A C Gringarten ldquoFrom straight lines to deconvolution theevolution of the state of the art in well test analysisrdquo SPEReservoir Evaluation and Engineering vol 11 no 1 pp 41ndash622008
[11] J A Cumming D A Wooff T Whittle and A C GringartenldquoMultiple well deconvolutionrdquo in Proceedings of the SPE Annual
12 Abstract and Applied Analysis
Technical Conference and Exhibition Document ID 166458-MS New Orleans La USA September-October 2013
[12] O Bahabanian D Ilk N Hosseinpour-Zonoozi and T ABlasingame ldquoExplicit deconvolution of well test data distortedby wellbore storagerdquo in Proceedings of the SPE Annual TechnicalConference and Exhibition Paper SPE 103216 pp 4444ndash4453San Antonio Tex USA September 2006
[13] J Jones and A Chen ldquoSuccessful applications of pressure-ratedeconvolution in theCad-Nik tight gas formations of the BritishColumbia foothillsrdquo Journal of Canadian Petroleum Technologyvol 51 no 3 pp 176ndash192 2012
[14] D G Russell ldquoExtension of pressure buildup analysis methodsrdquoJournal of Petroleum Technology vol 18 no 12 pp 1624ndash16361966
[15] R E Gladfelter G W Tracy and L E Wilsey ldquoSelecting wellswhich will respond to production-stimulation treatmentrdquo inDrilling and Production Practice API Dallas Tex USA 1955
[16] M J Fetkovich and M E Vienot ldquoRate normalization ofbuildup pressure by using afterflow datardquo Journal of PetroleumTechnology vol 36 no 13 pp 2211ndash2224 1984
[17] J L Johnston Variable rate analysis of transient well testdata using semi-analytical methods [MS thesis] Texas AampMUniversity College Station Tex USA 1992
[18] A F van Everdingen ldquoThe skin effect and its influence on theproductive capacity of a wellrdquo Journal of Petroleum Technologyvol 5 no 6 pp 171ndash176 1953
[19] W Hurst ldquoEstablishment of the skin effect and its impedimentto fluid flow into a well borerdquo The Petroleum Engineer pp B6ndashB16 1953
[20] J A Joseph and L F Koederitz ldquoA simple nonlinear model forrepresentation of field transient responsesrdquo Paper SPE 11435SPE Richardson Tex USA 1982
[21] F J Kuchuk ldquoGladfelter deconvolutionrdquo SPE Formation Evalu-ation vol 5 no 3 pp 285ndash292 1990
[22] A F van Everdingen and W Hurst ldquoThe application of thelaplace transformation to flow problems in reservoirsrdquo Journalof Petroleum Technology vol 1 no 12 pp 305ndash324 1949
[23] D Bourdet T M Whittle A A Douglas and Y M Pirard ldquoAnew set of type curves simplifies well test analysisrdquo World Oilpp 95ndash106 1983
[24] D F Meunier C S Kabir and M J Wittmann ldquoGas well testanalysis use of normalized pseudovariablesrdquo SPE FormationEvaluation vol 2 no 4 pp 629ndash636 1987
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
Abstract and Applied Analysis 11
By Theorem 1 there are some values 1198881and 1198882between 0
and 119905119863such that
119901119904119863
(119905119863) = 119901119908119863
(1198881) int
119905119863
0
119870(119905119863minus 120591) 119889120591 (B13)
1 = 119902119863(1198882) int
119905119863
0
119870(119905119863minus 120591) 119889120591 (B14)
By setting 1198881
= 1198882
= 119905119863and dividing the right and left
side of (B13) respectively the rate normalization techniqueis developed However the values 119888
1and 1198882are not equal
to each other and are not equal to 119905119863
in general Thisis the reason of error in rate normalization and materialbalance deconvolution methods 119888
1and 1198882are less than 119905
119863in
these methods for the duration of wellbore storage distortionduring a pressure transient test Since these conditions areapproximately satisfied in material balance deconvolutionthe results of material balance are more accurate and moreconfident than rate normalization
Nomenclature
API API gravity ∘API119861119900 Oil formation volume factor RBSTB
119888119905 Total isothermal compressibility factor
psiminus1
1198622 Arbitrary constant hr
minus1
119862119863 Wellbore storage constant dimensionless
119862119904 Wellbore storage coefficient bblpsi
ℎ Formation thickness ftHTR Horner time ratio dimensionless119896 Formation permeability md119870 Kernel of the convolution integral119871 Kernel of the convolution integral119873119901 Cumulative oil production bbl
119901 Pressure psi119901119886 Normalized pseudopressure psi
119901 Average reservoir pressure psi119901119863 Formation pressure dimensionless
119901119889119901 Dewpoint pressure psi
119901119894 Initial reservoir pressure psi
119901119904119863 Formation pressure including skin
dimensionless119901119904 Formation pressure including skin psi
119901119908119863
Pressure draw down dimensionless119901119908119891 Flowing bottom hole pressure psi
119901119908119904 Shut-in bottom hole pressure psi
119902119863 Sandface flow rate dimensionless
119902119892 Gas flow rate MscfD
119902119900 Oil flow rate STBD
119903119908 Wellbore radius ft
119904 Laplace variable119878 Skin factor119905 Time hr119905119863 Time dimensionless
119905119901 Production time hr
119879 Reservoir temperature ∘F
119879119889119901 Dewpoint temperature ∘F
120572 Explicit deconvolution variabledimensionless
120573 Beta deconvolution variabledimensionless
120574119892 Gas gravity (air = 10)
120582 Interporosity flow coefficient of dualporosity reservoir
120601 Porosity of reservoir rock dimensionless120583119892 Gas viscosity cp
120583119900 Oil viscosity cp
120596 Storativity ratio of dual porosity reservoirminus Laplace transform of a function1015840 Derivative of a function
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
References
[1] Y Cheng W J Lee and D A McVay ldquoFast-Fourier-transform-based deconvolution for interpretation of pressure-transient-test data dominated by Wellbore storagerdquo SPE Reservoir Eval-uation and Engineering vol 8 no 3 pp 224ndash239 2005
[2] M M Levitan and M R Wilson ldquoDeconvolution of pressureand rate data from gas reservoirs with significant pressuredepletionrdquo SPE Journal vol 17 no 3 pp 727ndash741 2012
[3] Y Liu and R N Horne ldquoInterpreting pressure and flow-ratedata from permanent downhole gauges by use of data-miningapproachesrdquo SPE Journal vol 18 no 1 pp 69ndash82 2013
[4] O Ogunrewo and A C Gringarten ldquoDeconvolution of welltest data in lean and rich gas condensate and volatile oil wellsbelow saturation pressurerdquo in Proceedings of the SPE AnnualTechnical Conference and Exhibition Document ID 166340-MS New Orleans La USA September-October 2013
[5] R C Earlougher Jr Advances in Well Test Analysis vol 5 ofMonogragh Series SPE Richardson Tex USA 1977
[6] F Kucuk and L Ayestaran ldquoAnalysis of simultaneously mea-sured pressure and sandface flow rate in transient well testingrdquoJournal of PetroleumTechnology vol 37 no 2 pp 323ndash334 1985
[7] M Onur M Cinar D IIk P P Valko T A Blasingame and PHegeman ldquoAn investigation of recent deconvolution methodsfor well-test data analysisrdquo SPE Journal vol 13 no 2 pp 226ndash247 2008
[8] F J KuchukMOnur andFHollaender ldquoPressure transient testdesign and interpretationrdquo Developments in Petroleum Sciencevol 57 pp 303ndash359 2010
[9] A C Gringarten D P Bourdet P A Landel and V J KniazeffldquoA comparison between different skin and wellbore storagetype-curves for early-time transient analysisrdquo in Proceedings ofthe SPE Annual Technical Conference and Exhibition Paper SPE8205 Las Vegas Nev USA 1979
[10] A C Gringarten ldquoFrom straight lines to deconvolution theevolution of the state of the art in well test analysisrdquo SPEReservoir Evaluation and Engineering vol 11 no 1 pp 41ndash622008
[11] J A Cumming D A Wooff T Whittle and A C GringartenldquoMultiple well deconvolutionrdquo in Proceedings of the SPE Annual
12 Abstract and Applied Analysis
Technical Conference and Exhibition Document ID 166458-MS New Orleans La USA September-October 2013
[12] O Bahabanian D Ilk N Hosseinpour-Zonoozi and T ABlasingame ldquoExplicit deconvolution of well test data distortedby wellbore storagerdquo in Proceedings of the SPE Annual TechnicalConference and Exhibition Paper SPE 103216 pp 4444ndash4453San Antonio Tex USA September 2006
[13] J Jones and A Chen ldquoSuccessful applications of pressure-ratedeconvolution in theCad-Nik tight gas formations of the BritishColumbia foothillsrdquo Journal of Canadian Petroleum Technologyvol 51 no 3 pp 176ndash192 2012
[14] D G Russell ldquoExtension of pressure buildup analysis methodsrdquoJournal of Petroleum Technology vol 18 no 12 pp 1624ndash16361966
[15] R E Gladfelter G W Tracy and L E Wilsey ldquoSelecting wellswhich will respond to production-stimulation treatmentrdquo inDrilling and Production Practice API Dallas Tex USA 1955
[16] M J Fetkovich and M E Vienot ldquoRate normalization ofbuildup pressure by using afterflow datardquo Journal of PetroleumTechnology vol 36 no 13 pp 2211ndash2224 1984
[17] J L Johnston Variable rate analysis of transient well testdata using semi-analytical methods [MS thesis] Texas AampMUniversity College Station Tex USA 1992
[18] A F van Everdingen ldquoThe skin effect and its influence on theproductive capacity of a wellrdquo Journal of Petroleum Technologyvol 5 no 6 pp 171ndash176 1953
[19] W Hurst ldquoEstablishment of the skin effect and its impedimentto fluid flow into a well borerdquo The Petroleum Engineer pp B6ndashB16 1953
[20] J A Joseph and L F Koederitz ldquoA simple nonlinear model forrepresentation of field transient responsesrdquo Paper SPE 11435SPE Richardson Tex USA 1982
[21] F J Kuchuk ldquoGladfelter deconvolutionrdquo SPE Formation Evalu-ation vol 5 no 3 pp 285ndash292 1990
[22] A F van Everdingen and W Hurst ldquoThe application of thelaplace transformation to flow problems in reservoirsrdquo Journalof Petroleum Technology vol 1 no 12 pp 305ndash324 1949
[23] D Bourdet T M Whittle A A Douglas and Y M Pirard ldquoAnew set of type curves simplifies well test analysisrdquo World Oilpp 95ndash106 1983
[24] D F Meunier C S Kabir and M J Wittmann ldquoGas well testanalysis use of normalized pseudovariablesrdquo SPE FormationEvaluation vol 2 no 4 pp 629ndash636 1987
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
12 Abstract and Applied Analysis
Technical Conference and Exhibition Document ID 166458-MS New Orleans La USA September-October 2013
[12] O Bahabanian D Ilk N Hosseinpour-Zonoozi and T ABlasingame ldquoExplicit deconvolution of well test data distortedby wellbore storagerdquo in Proceedings of the SPE Annual TechnicalConference and Exhibition Paper SPE 103216 pp 4444ndash4453San Antonio Tex USA September 2006
[13] J Jones and A Chen ldquoSuccessful applications of pressure-ratedeconvolution in theCad-Nik tight gas formations of the BritishColumbia foothillsrdquo Journal of Canadian Petroleum Technologyvol 51 no 3 pp 176ndash192 2012
[14] D G Russell ldquoExtension of pressure buildup analysis methodsrdquoJournal of Petroleum Technology vol 18 no 12 pp 1624ndash16361966
[15] R E Gladfelter G W Tracy and L E Wilsey ldquoSelecting wellswhich will respond to production-stimulation treatmentrdquo inDrilling and Production Practice API Dallas Tex USA 1955
[16] M J Fetkovich and M E Vienot ldquoRate normalization ofbuildup pressure by using afterflow datardquo Journal of PetroleumTechnology vol 36 no 13 pp 2211ndash2224 1984
[17] J L Johnston Variable rate analysis of transient well testdata using semi-analytical methods [MS thesis] Texas AampMUniversity College Station Tex USA 1992
[18] A F van Everdingen ldquoThe skin effect and its influence on theproductive capacity of a wellrdquo Journal of Petroleum Technologyvol 5 no 6 pp 171ndash176 1953
[19] W Hurst ldquoEstablishment of the skin effect and its impedimentto fluid flow into a well borerdquo The Petroleum Engineer pp B6ndashB16 1953
[20] J A Joseph and L F Koederitz ldquoA simple nonlinear model forrepresentation of field transient responsesrdquo Paper SPE 11435SPE Richardson Tex USA 1982
[21] F J Kuchuk ldquoGladfelter deconvolutionrdquo SPE Formation Evalu-ation vol 5 no 3 pp 285ndash292 1990
[22] A F van Everdingen and W Hurst ldquoThe application of thelaplace transformation to flow problems in reservoirsrdquo Journalof Petroleum Technology vol 1 no 12 pp 305ndash324 1949
[23] D Bourdet T M Whittle A A Douglas and Y M Pirard ldquoAnew set of type curves simplifies well test analysisrdquo World Oilpp 95ndash106 1983
[24] D F Meunier C S Kabir and M J Wittmann ldquoGas well testanalysis use of normalized pseudovariablesrdquo SPE FormationEvaluation vol 2 no 4 pp 629ndash636 1987
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of