You are on page 1of 18

SPE

SPE 17547 Application of Equivalent Drawdown Time in Well Testing


by S.C, Swift, Swift Engineering
SPE Member

Co./OGCl

Copyright198S Society of Petroleum Engineers Th18naper was prepared for presentationat the SPE Rocky Mountain Regional Meeting, held in Casper, WY, May 11-13, 19S8. Th: oar waa selected for preaematlonby an SPE Program Committeefollcwingreview of informationcontained In an abstract submittedby the auth Contentsof the paper, aa presented, have not been reviewed by the Society of Petrolaum Engineers and are subject to correctionby the author,., The material, aa praemted, does not neceeaarllyreflectany posllionof the Societyof PetroleumEngineers,ita officara,or members. Papere presented at SPE meetings are tiubjectto publicationreview by EditorialCommifteeaof the SOcletyof Petroleum Englneere. Permlaslonto copy ia restrictedto an abetract of not more than 300 worda. IIluatratlonemay not be wpled. The abstract shouldcontain Conaplcuoua acknowledgmentof where and by whomthe paper la presented,Write PubllcationaManager, SPE, P.O. sax 833S3(+,Rlchardeon,TX 75083-3S36. Telex, 730989 SPEDAL.

ABSTRACT
The concept of equivalent drawdown time was described by Ram Agarwal in 1980. In reality it has been In common use In well test anaiysis since at Ieasf 1950. However, in 1988, application of the concept Is still not widely understood. The purpose of the present paper Is to encourage its application by clarifying its role in well test analysis.

The most important tool used for step o,,e is type curve analysis. The tool best fitted for step two depends on step one. Obviously type curves (especially type curve analysis) play a critical role in drawdown analysis. Examples of tools used for step two include semilog plots for radial flow behavior, square root plots for iinear flow behavior, cartesian plots for pseudusteady flow behavior, and type curve matching for more complex behaviors such as storage/linear transitions, finite conductivity fractures and dual flow systems. Equivalent drawdown time was Iiteraily invented to allow proper utilization of type curves on tests which involve more than one rate change. Although type curves fostered the creation of equivalent drawdown time, the possibility of much wider application was immediately recognized. When properly applied, the use of equivalent drawdown time reduces the anaiysis of any complex flow to anaiysis of a drawdown test. Therefore any tool used for drawdown analysis, applies directly to complex data which has been transformed to equivalent drawdown data. In short, Ateq can be used for any kind of analysis of complex flow data, not just for type cuwe applications. For exarripie, it could be used to make a radial flow analysis of a Horner rate history or a iinear flow anaiysis of a general multiple rate history. An infinite number of different possible behavioral mechanisms exist. Likewise, an infinite number of different rate histories are possible. The number of different combinations of behaviors and rate histories is bewildering. If equivalent drawdown time is used, we need only understand each mechanism as it pertains to drawdown analysis. The effects of muitiple rate changes are then factored out of the process, removing unnecessary clutter. A NEW PERSPECTIVE When any step two technique is applied to a complex flow test bv first transforming it to equivalent drawdown data, this application might at f;rst appear to be a new method. However,

INTRODUCTION
Drawdown test analysis is easy. Analysis of tests involving more than one rate change is difficult. Equivalent drawdown time makes the difficult into the easy. If that sounds too good to be true, maybe it is, but its almost truel In the real world it comes out more like this. Drawdown test analysis is difficult. Analysis of tests involving more than one rate change is so complicated that in makes drawdown test analysls look easy by comparison. Equivalent drawdown time makes the analysis of more complex tests no more difficult than drawdown test analysis, In the process it gives a simpler perspective of well test anaiysis than is possible without it, By reducing it to two simple steps, equivalent drawdown time make the analysis of compiex flow tests easier to understand. The two steps are: 1. 2. Convert the complex data to drawdown data. Analyze the converted data like a drawdown test,

Obviously, a thorough understanding of drawdown behavior is critical to make this plan of attack work. Drawdown anaiysis itself invoives two steps: 1. Determine the mechanism(s) that is controlling the behavior of the well. 2. Quantify the parameters reiative tG ihat controlling rv.hanism.

References and illustrations at end of paper,

595 .

APPUCATION OF EQUIVALENT D
it is usually simply a new perspective on an older technique. For example when Agarwais equivalent drawdown lima is used for semlioo radial flow analvsls It may have the . ...- . appearance of a new method, But it Is best thought of as a standard Horner plot viewed from a new perspective. Indeed, It will be demonstrated that this most revered Institution of well test anaiysis, the Horner plot is, and aiways has been, an application of equivalent drawdown time. Most people accept the fact that there are several different ways to construct a Horner piot, but regardless of whether you use Horner Ratio (extrapolate to the right) or Horner Time (extrapolate to the ieff) you are still making a Horner plot.
you may choose tO in the future, strictly for convenience, make a semibg plot using Agarwals equivalent drawdown time. This is not a sacrilege- It is, like Horner Ratio and Horner Time, just another way to make a Horner plot. I will leave It to your own conscience as to whether you choose to use such a semilog plot. I wili even tolerate a difference of opinfon as to whether doing so constitutes a new method. But, as for me, and my house, a Horner plot using Agarwals equivalent drawdown time is just a simpier perspective on the same old Horner plot.

kWDOWN TIME IN WELL TESTING

SPE 1754

For cases where there are two ecfual but OPPosite rate changes (Horner rate history) and when both trarisients are in radiai flow this equivalent time is the product of the producing time (tp) and the buildup time (At) divided by the sum of the same two quantities:

AteqA =

tp At/

(tp+At) .. .. .. . .. . ... . .. . ... .. .. .. .. . .. .. .. .. .. . .. .. ....0. (1)

Equivalent drawdown time should be cieariy distinguished from two other time related concepts of weii testin9, (1) reai gas pseudotime and (2) effective flow time. Real gas pseudotime corrects time for noticeable variations in properties of reai gases which are caused by large pressure changes. This concept was also introduced by Agarwal, and is often confused with equivalent drawdown time. Both ooncepts are occasionally referred to as Agarwai time. The two oonoepts are, in fact, totaiiy different. They can be applied separately, or in combination, The present paper wiii totally neglect any corrections for real gas pseudotime in order to focus on the properties of equivalent drawdown time. Effective fiow time is a parameter which is used, quite successfully, to model a more ccmplex fiow history as a less oomplex Horner flow history. GENERAL FORM Equivalent drawdown time can be defined in generai as:
Ateq

Likewise, when equivalent drawdown time iS applied to an arbitrary rate history by using superposition, it really is just a new perspective on multirate anaiysis. Even if the arbitrary rate history reflects the sandface rate caused by wellbore storage, it Is still only a new perspective on older methods. There are few problems in weil test anaiysis that have never been previously addressed, Equivalent drawdown time may occasionally provide a basis for improving existing methods, as was the case for type curve analysis. Even so, the most fundamental value of equivalent drawdown time is that it provides a simpler, more easily understood perspective of existing methods of analysis. PURPOSE The purpose of this paper is to encourage more widespread understanding of the perspective offered by equivalent drawdown tim~. it will attempt to accomplish this objective by presenting a generalized derivation of equivalent drawdown time and examining various ways which the concept can be applied, and the considerations which must be made in so doing. So whats new? Probably nothingl Superposition stili requires addition, radiai flow analysis still requires a semiiog piot, and the sun stiii rises in the east. Even the credit for the new perspective belongs to Agarwai, not to the present author. There is a remote chance that the perspective offered in this paper is siightiy more generai than the one taken by Agarwai (hindsight is 20/201) However, my own hope is that this paper might promote a wider understanding, and an associated wider application, of the concept of equivalent drawdown time within the industry. If this comes to pass, the, I i wiil be satisfied that my effort was not in vain,

tu ~l{X(i=l,N): )-f((tR (Ri*[~((t-ti)/tu

-ti)/tU

)])} .................... ..( 2)

Where: ~(:v~sm arbitrary function of x and ~ 1(y) is its

{X.=1 ,N): (xi)} is the summation of XI over i from 1 to , RI is the vaiue of the i-th rate change divided by a reference rate, qR, t is time measured arbitrary units, from any arbitrary point in any

ti is the time of the i-th rate change, tu is the unit of measurement end for dimensionless time,

tR is a reference time from which Ateq is to be measured. Note that it is assumed that the potentiai, PR, at the reference time, tR, is known. Aithough it is not required for equivalent time, a compiete transformation from (AMP) to (Ateq tAp) requires that the change in potential be known. This change in potentiai is measured from PR. See Appendix A for the derivation and a detailed explanation of terms. KEY POiNTS Application of Equation 2 requires:

THEORY
The concept of equivalent drawdown time was introduced to the weii testing literature in 1980 by Ram Agarwai. It addressed a very specific probiem. Buiidup type curves based oniy on shutin time (At) can be misleading, They often exhibit behavior that has the appearance of radial flow, even when Iittie or no radial fiow effects are present. Agarwai solved this problem eiegantiy by demonstrating that the actual buiidup time (At) shouid be repiaced with an equivalent drawdown time (Ateq).

1. A known rate history (number of rate changes,

N; magnitude of the rate changes, Aqi; and the time of the rate changes, ti).

2. A known function for the weii behavior and its inverse.

596

SPE 17547
3. Three additional parameters: A. B. C Reference Time, tR, Reference Rate, qR, and Characteristic Time, tlr,

SAMUEL C. SWlfT
RATE HISTORY The number of rate changes, N, must be known. If N Is orw then Ateq reduces to test time, AlthoWh this Is a trivial case of Ateq, it does cover five common aIMly!iCd procedures; drawdown, negative drawdown, buildup following extended production, falloff following extended injection, and desuperposition. For desuperposition the change In potential Is the projected change in potential rather than the observed change In potential. If N is two and the rate changes are equal then It is called a Horner rate but opposite (Aql =-Aq2) history, For multirate analysis N Is two or more and the magnitude of the rate changes are arbitrary. The time of each rate change, ti, must be known as well as the magnitude of each rate change, Aq, The coefficients, Ri, are obtained by dividing (normalizing) each rate change by the reference rate. THE FUNCTION Probably the most difficult to understand aspect of Atsq is the role of the function, ~. However, the difficulty is not something unique to the concept of equivalent drawdown time. Determining (or at least assuming) what function describes the behavior of a well is the first step in sny type of analys!s. In general the required function is the characteristic drawdown behavior of the system, so that if an idea I drawdown test were run at the reference rate , qR, the change in potential (AP) as a function of drawdown time (At) would be prec!sely expressed as:
AP = f(At) ................................. ..... ................ ..................(3)

Both the reference time, tR, and the reference rate, qR, are completely arbitrary. Both also fix associated values of potential. The reference level of potential, PR, Is the potential which existed at the time tR. The unit of measurement for dimensionless potential, Pu, is fixed by qR. Note that the parameters tR and qR tTIUSt alWaYS ~ specified. Whether or not the value of tu must be known explicitly, depends on the nature of the function, t, and the type of equivalent drawdown time desired; regular or dimensionless. If tu is required, then the solution becomes trail and error in terms of !U. It is generally not required to know Pu, a priori.

APPLICATION
The above discussion defines everything necessary to calculate equivalent drawdown time. However, meaningful application requires an appreciation for the role each item plays. PARAMETERS The characteristic time, !U, is simply a unit of time measurement that is especially appropriate to the test situation under consideration- just as the size of the cup is significant when one requests a ha!f of a cup of coffee, Unlike the tangible cup, tu is difficult to visualize. It can be thought in hours, determined by the of as a time constant, parameters of the problem. Its exact mathematical definition IS in Appandix A. It Is also the time unit that is normally used for dimensionless time. In fact, if Eq 2 is divided by tu one obtains the dimensionless equivalent drawdown time, AtDeq. Conventional Horner Time is a dimensionless equivalent time. The characteristic potential, Pu, is simply a unit of potential measurement that is especially appropriate to the test situation under consideration (similar to !U for time), Characteristic potential does not appear explicitly in the equation for A!eq, but it is linked by definition to the reference rate, qR (see Appendix A). Reference rate is required In the equivalent time equation in that it is the denominator of each of the Ri coefficients. It is totally arbitrary but is often the value of the rate change which occurred at tR (Note, however, that it is not required that tR be selected at a point in time where a rate change occurred), Reference rate can be selected as a conveniently round number, e.g. 100 BOPD, 1 MMCFD, etc. If unity is used then Ateq is tied to the unit response function. The reference time, tR, is totally arbitrary. It would most often be chosen as the time of the most recent rate change, since the effect of that rate change dominates the potential behavior for some period of time thereafter. The value of the potential at tR Is the reference potential, PR, If the change in potential (PR- Pw or AP) is required (e.g. for type curves) then PR must be known. For Agarwal equivalent the, AteqA, tR is selected at the time of shutin, For conventional Horner Time, AtDeqH, tR is selected at the beginning of flow. Horner Time is the same as Horner Ratio except that for Horner Ratio the reference rate is the production rate (+q), Horner Time uses the change in rate (-q) at shutin as the reference rate.

Perhaps you have noticed that our discussion has become somewhat circular. We must know the characteristic response function of the system in order to correct our actual observations, in order to faithfully reflect the characteristic response of the system- thats life! The characteristic response function is only revealed directly, when it Is possible to run an Ideai drawdown test. For any more complex test, including real drawdowns, the observed data is distorted by superposition of multiple images of the characteristic response function. Equivalent drawdown time can remove this distortion, but only if we are skilful enough to select the proper function. Therefore, in a general sense, the selection of j is always trial and error. We must choose a function, make the correction (calculate equivalent drawdown time) and check to see if the corrected function (transformed data) is the same as the assumed function. Thats the bad news. The good news is that we know something about what functions might be appropriate. In fact, petroleum engineers have been picking such functions for many years with at least some success. With any luck at all, ~ should be proportional to some dimensionless solution to the diffusivity equation, so that Eq 3 becomes: ,..........................( 4 ) Ap=f(At)=pu*pD(tD) =pu*pD(At/tlJ) Where PD(tD) iS a dimensionless equation.
SOIUtiOrr.tO

the diffusivity

There are an infinite number of solutions to the diffusivity equation. The following list, while only a finite sampling of an infinite number of possibilities, represent the ones that are most useful in practice.

597

1.
2. 3. 4. 5. 6. 7.

Steady flow solution, Pseudosteady flow solution, Linear flow solution, (Infinite reservoir) Exponential integrai solution , (infinite reservoir) Cyllndricai weiibore soiution, (Infinite reservoir) Fractured welibore solution, (infinite reservoir) Pseudosteady wellbore flow soiution, (Cylindrical wellbore) (infinite reservoir)

PDs PDP PDLI PDEI PDci PDFi PDPCi

For this reason, it is erroneous to assume that equivalent time based on the assumption of any particular behavior can be used to interpret a different kind of behavior. For example, if AtsqA is used and the transformed data had a peculiarity that appeared to be duai flow system behavior, it would be premature to conciude that the test data is from a dual flow system. This couid oniy be substantiated if a function was seiected which faithfully portrays dual system behavior and the resuiting transformation confirmed that the peculiarity is a property of the fundamental drawdown behavior and not Just a distortion caused by muitlple rate changes. A final property of ali functions which are appropriate to equivalent drawdown transformations is that they are zero for aii negative or zero arguments. if a rate change has not yet occurred it has no effect on potential. This saems trlviai, but it must be accounted for in the equations. Mathematlcaiiy, Eq 5 and Eq 6 are incomplete. They must explicitly be forced to zero for negative and for zero arguments. That is the reason for the pound sign (#) in some later equations. EQLfiVALENT TiMES ALREADY iN COMMON USE Although Equation 2 is somewhat abstract, careful study of the factors in its derivation immediately give rise to some important applications. One such application is to demonstrate that several equivalent times in common use are reaiiy speciai cases of ti Ie generai equation. The most obvious is Agarwals equivalent time (AteqA) Much iess weii imown is Tandem linear equivalent time. (AteqT), which differs from Agarwais equivalent time oniy in that iinear flow behavior is assumed, Very weii known, but not often thought of as equivalent times and Horner Time are Horner Ratio ([tp +At]/At) (At/[tp+At]), Both of these familiar quantities are speciai cases of general equivalent time, Eq 2. Both are dimensionless equivalent drawdown times which take their reference time, tR, as the beginning of the flow period and assume radial fiow behavior. Horner Time takes -q as the reference rate where Horner Ratio takes +q, Since Horner Time, Horner Ratio and Agarwais equivalent time share a common rate history and a common assumed functional behavior, and differ oniy in the arbitrary parameters, it is obvious that identical results shouid be expected from radiai flow (semiiog) anaiysls using any one of the three parameters. Corresponding to Horner Ratio (but assuming linear flow behavior is present rather than radial) is tandem root +At)-d(At)]2). The square root of time([d(tp this, parameter is routineiy used for iinear fiow (square root) anaiysis of two equal and opposite rate changes, Tabie 1 summarizes the defining properties cf these equivalent times: Agarwals AteqA is a Speciai case for: (1) two equai and opposite rate changes, (2) both transients in radial flow, and (3) using the time of shutin as the reference time and the change in rate at shu!in as the reference rate. This derivation is in the appendix and the resuit, as expected, is Equation 1. Agarwais equivalent time is properly used oniy when appiied to two equai and opposite rate changes and then only if both transients are in radiai flow. it is weii known that the assumption of two equai and opposite rate changes is an adequate modei for many rate histories that are in fact more complex, and that if the rate changes are iarge (especially if they are recent; that is to say, near the end of the flow period) then probiems can arise. Much iess weil known are the probiems which arise from a faise assumption that the transients are in radiai flow.

Aithough the above list Is iimited, it is stiii too iong to cover in a single paper. Therefore, lets cut it down some. The first two are very simpie solutions, so we wili assume that their effects are well known (a dangerous assumption). PDPCi differs from PDCi oniy in that it has a compiex (selfinferred) flow history. It therefore can, if desired, be reduced to PDCi by applying equivalent drawdown time. The remaining four are all reiated in some way to two simple functions: PDLi=(2/rr)*d(tD) PDLN=(l/2)*in(4 ....................................,.,,,,,.,,,.,,, ......(5) *tDN) ...............................................( 6 )

Where n is a transcendental constant approximated as 3.1416 and d(x) represents the square root of x. Aiso in indicates the naturai logarithm and Y is a transcendental constant approximated as 1.7811. PDLi is the exact solution for iinear fiow. PDLN is the equation on which semiiog anaiysis is based. Since common usage equates radial flow with semliog straight, Eq 6 constitutes the definition of radiai fiow. it should be noted that PDLN is not a solution to the diffusivity equation, but an approximation to PDEi, PDCI, PDFi, etc. which oniy applies for iarge dimensionless times. These two behaviors, linear fiow and radiai fiow, are fundamental. If ii were possible to ?onduct an ideai drawdown test (instantaneous change in sand face rate) on a reai weii, ail such tests wouid exhibit iinear flow initialiy. Also, many real welis exhibit a period of radiai flow behavior which foilows weiibore (inner boundary) effects and precedes outer boundary effects, These two simpie functions are considered extensively in Appendix B. The uitimate advantage of the concept of equivalent drawdown time is that the basic procedure is the same regardless of the complexity of the function ~. Whether using a simpie function iike PDLI, a siightiy more complicated one iike PDCi or a very complex one invoiving inner boundary effects, dual fiow systems and outer boundary effects, the concept and basic procedures are unchanged, They do, however, become more complicated to perform. Any function that successfully transforms complex fiow data to equivalent drawdown data must accurately describe the behavior of the system under consideration during an ideai drawdown, it must do so over the reai time intervai being considered- test time or time of projected performance. Unfortunately this single reai time intervai has associated with it, N different dimensl~vrless time intervals- one for each individual rate change. These intervals are of equai iength but span distinctly different vaiues of dimensionless time. One fundamental fact is ciear. Unless the function selected provides at ieast a reasonable approximation to the ideai drawdown behavior over ali of these N dimensionless Intervals, it will not succeed.

SE 17547

SAMUEL C. SWIFI
and for linear flow are:
AteqL={~(i=f,N):

I .

There are In faot many special cases of Equation 2 which am in oommon use, although they often are not thought of aj equivalent time. obviously, drawdown analysis (radial, linear, or even pseudosteady reservoir limit behavior) are all special oases where N is one. As Agarwai pointed out, the concept of aquivaient time is not limited to type curve analysis. Once the transformation to equivalent drawdown time is made, the data becomes, in every way, drawdown data so that semi-log pbts, square root plots and cartesian plots may be wiled if appropriate. Appropriate in this case means both that the weii is exhibiting a particular behavior and that the transformation was made using that same behavior, Classical Horner Tim% At/(tp+At) , is a dimensionless equivalent drawdown time, exactly the same as Agarwais except that the reference time is at the beginning of the drawdown, the reference rate 1sthe finai rate change, -q , and the dimensionless form is used. if +q is used as the reference rate, then the famiiiar form, (tp+At)/At,is obtained. Note that both of these forms are oommonly used for semilog anaiysls. They generally can not be used for type curve analysis, because the (Initial) potentiai at the reference time Is not known. Indeed, the whole key to Agarwals disoovery of AteqA was the giving Of fuil mathematical resPect to the oommonly used reference potentiai - the (known) potentiai in the weil at shutin. Linear fiow anaiysis is more important than commoniy recognized. This iack of recognition is at least partially due to misusing pressure versus square root of time plots for analysis, when the more compiioated tandem root time plots were required. Much has been said and written about how the Horner rate history must be considered in radial fiow analysis of buiidup data. The simpler single rate change (Miiler, Dyes and Hutchinson history) is not adequate In many cases. This is even more true in linear f!ow than in radial flow. This discussion couid obviously be extendad to include many other possible combinations of dimensionless and dimensional equivalent times, different fiow behaviors, and different reference times. Some of the resuits are interesting, and may even be in use by some experts in weii testing. However, the above discussion covers the most oommoniy seen special cases and further discussion is beyond the scope of this paper. MULTiRATE ANALYSiS FROM THE NEW PERSPECTIVE

Ri*[d#(t-ti)-d# where o 4#(x) = { d(x)

(tR-ti)]}h2

.. . .... .. .... . .. .... .... ... . (11)

Xso ....................................,.,,.. (12)

X>o .......................................... (13)

Hopefully, at this point, the abstract function j(x) has assumed some more concrete meaning. it has in fact been assumed to be equai to common dimensionless potential functions, The function for radial flow is PDLN and the function for ikrear flow is PDLI. The real power of Equation 2 is that ~(x) is not restricted oniy to simple soiutlons to the diffusivity equation, By inserting more com Iicated solutions (models) one can consider any case wh f h may be deemed to be appropriate, e.g. the finite cylindrical wellbore (PDCI), infinite conductivity fracture (PDFi), finite capacity fracture with biiinear flow, dual porosity reservoirs, etc. Although it is beyond the soope of this paper to mnsider any of these cases in detaii, one thing should be obvious. Extension of a special case equivalent drawdown time to a case where it does not apply is at best risky, and in many cases totally wrong. For example A!eqA assumes radial flow for ali transients and a Horner rate history. if it is appiied to a Horner rate history, but iinear fiow behavior, it works fairiy well, but not exactiy, If the flow is linear but tha rate history compiex it can be totaiiy misleading. Agarwais equivalent time can be used generally, as a basis for examining unknown data. However, if the data is seen to be in anything other than radial flow, then Equation 2 should be used to determine the proper equivalent time for the suspacted behavior. The appearance of non-radiai flow behavior subjected to a radiai fiow transformation can be very misleading For exampie, a real resewoir with dual porosity behavior will probabiy not iook like known dual porosity soiutions if AteqA is used. Furthermore, if storage is present aiso, the radiai flow transformation wiil distort the data in a different manner. The bottcm iine is common sense- dont expect radial flow transformations to work on non-radial flow data. Check it out, it might work adequately in some cases, but It certainiy wiil not in others. FLOW HiSTORY PARTiALLY SELF-iNFERRED

The most obvious application of Equation Z beyond those cases already discussed is to consider more complicated rate histories. For exampie, when attempting to analyze an extended drawdown test wifh major disruptions included in the fiow period, it is obvious that the Horner rate history does not pply. For these cases it is convenient to write equivalent drawdown time for an arbitrary fiow history, but stiil assuming that ali transients have a common functional behavior. These cases for radial flow and iinear fiow are derived in the appendix. The iesuits for radiai fiow are: AtsqR= {tU*W4}*{rI(i=l, N): (t-ti) Ri#/(tR -ti)ARi# where: II(i=l ,N):(xi)=xl

} .....................................( 7 )

q X2X3*...*XN

. .. . . .. . . .. .. . . . . .. . . .. . . . . . .. . . . . ..(

8)

(Y*tu/4)Ri xARi# = { xARi

for x s O...............................(9) forx>O ,..,,.,,!..,,..,.. ........... (lo)


599

One soiution to the diffusivity equation is encountered sufficiently often as to deserve speciai consideration, the so tailed storage and skin solution (PDPCI) This solution was introduced by Agarwai and Ramey in 1970 and almost simultaneously by McKinley. After considerable debate, it was generaiiy agreed that in most cases skin couid be adequately accounted for by using the effective weiibore radius, rwa, or rw*eA(-s). This debate took several years and resulted in the presentation of the same basic soiution on numerous log-log l-Agarwal&Ramey; 2-McKinley; 3formats (e.g. 4- Gringarten). All of these formats were Eariougher; essentiai evolutionary steps. Unfortunately many petroieum engineers view these different type curves as fundamentally different methods rather than just different ways of presenting the same information. Each of these presentations ( as weil as others) have particular advantages and disadvantages and may be the most useful for a given test anaiysis.

APPUCATION OF EQUIVALENT DRAWDOWN TfME IN WELL TESTING . ..


This function, PDPCI, can be used in equation 2, however the mathematics, including the parameters, become somewhat cumbersome. A more satisfying approach is to back out the storage effects by treating the afterflow as finite rate changes at the sand face. When the weilbcre storage behavior is simple, it Is easy to make a welibcre modei which calculates the unknown sandface rate from known surface rates. For exampie, if constant weilbcre storage Is assumed the sandface rates are easiiy catcuiated. When these rate changes are taken Into amount the transformed potential data reduces to the weiibcre behavior without storage, in this case PDCi. This is true however, only to the extent that the underlying reservoir behavior conforms to PDCi. Fortunately, it works in many cases. Even a fractured weli (PDFi) conforms closeiy to PDCi. The only significant deviation Is in the transition from PDLI to PDLN; and even there the error is less than 10%. This process Is illustrated in the Waterton #1 example. It can obviously be extended to more complicated cases (continuously changing storage or discretely changing storage for instance) by simpie modifications of the procedure of calculating sandface rate from surface rate, It is interesting to note that Akq R wcrks very PcorlY for this application. Because PDLN behaves peculiarly (iiteraliy unreail) for short drawdcwn times, AteqL Is a much better choice when dealing with afterflow corrections. In fact the auihor believes that AteqL is a preferred tool for first lock usage, but most authorities would regard that as a personal iinear flow bias (to which i proudly confess!). So i will stop short of recommending it for general use, However, it doesnt hurt to check it just in case. It might just be that the weird behavior seen on a given test is a distortion caused by muitiple imaging of PDLI rather than some super sophisticated fundamental behavior. In fact, if this effect as well as minor wellbore perturbations caused by phase redistribution and crossfiow are faithfully eliminated, those seven PD functions previously considered are adequate In the vast majority of cases- I did not say all I BEHAViOR SELF-INFERRED SUMMARY OF APPLICATIONS

SPE 175

Defining the rate history and choosing reference quantities Is straightforward. Choosing the best function for calculating equivalent drawdown time is the major consideration. Because of the special emphasis placed on radiai fiow behavior, AteqR, and its speclai Horner rate form, AtsqAt will be the most common forms used. They wiil probably be the choice for first looks at test data of unknown character. If type curve analysis then confirms the presence of radial flow behavior, quantification can proceed in the conventional manner, using Horner plots or radial superposition plots for radial flow analysis. Because of its Convenience, AteqR wiii gradually become the parameter of choice for radial fiow analysis. If the first look reveals nomadial flow, the type curve should be redone using AteqL. if linear fiow is confirmed analysis can proceed using Tandem root time or ikrear superposition piots. Here again, because it is more convenient, AteqL will gradually become the paramqter of choice for linear flew anaiysis. if the flow behavicr is more complex than simpie ilnear or radial fiow, every effort shcuid be made to find the drawdcwn function which properly transforms the data. The complex function of choice is PDCi, and it wiii give a satisfactory performance in most cases. However, more complicated functions which accurately refiect inner boundary effects, compiex reservoir effects and outer boundary effects wiil occasionally be necessary.

EXAMPLES:
Two greatly different exampies are considered. Both are generated data (not real weiis), however the points made are based on personal experience of the author in dealing with real well tests. The first example, Waterton #1 could be (almost) correctly evaluated uskrg only a data type curve based on Agarwals equivalent time and a semiiog plot based on the standard Horner parameter. Indeed some would point out that use of oniy the Horner analysis arrives at the correct interpretation. Even so, in the opinion of the authcr, use of the general form of equivalent drawdown time for both type curve analysis and radial fiow (semilog) analysis provides a more Iogicaliy consistent approach and a higher degree of confidence in the resuits. The second example, Sombrero #1, Illustrates many of the shortcomings of attempting to use methods which fail to take into account the complex nature of the rate history andor the effects of non-radial flow behavior. In addition, it aiso shows that non-radial flow behavior can be (in the authors experience and cpinion often Is) important in high rate (international oii) production as weil as tight gas production in North America. In fact, non-radial flew effects seem to obey Murphys law - they tend to occur in the most unexpected places ( anywherel ) and at the worst pcssible times (when cash ficw predictions have been based on the assumption of more graduai radial declines). Understanding of non-radial fiow behaviors cannot predict prcduct prices. However, such an understanding can provide a reasonable basis for anticipating many declines in production rate. This knowiedge can be critical in managing the impact of these declines, whether they occur in tight gas sands of North America or high rate oil production from fractured reservoirs anywhere in the world.

The applications discussed above involve a known functional behavior. However Equation 2 is not iimited to any particular function. This gives rise to a particularity important possible application where the observed behavior itself might be used to deduce the actual functional nature of the drawdown behavior of a real well: totaily independent of any assumed (guessed) functional behavior. Indeed this is possible in some cases, as is illustrated in the Sombrero #1 exampie. The authors numerical skiils are ver~ primitive, example illustrates several things to my satisfaction. 1. There is no free iunch - meaning: 1.a Extrapolation critical, and interpolation techniques are but this

1.b

The oniy time this method Is guaranteed to work is when you dont need it, namely if you observe the entire possible range of time values.

2. Even using a iimited range of data and very limited numericai skills this technique makes it possible to Icok though the clutter caused by complex rate histories and see the vague outiine of the underlying drawdown behavior. This view can in some cases be a mere reaiistic view than that obtained by presupposing a certain functional behavior. (e.g. radlai, iinear, biiinear, dual porosity, storage, steady state, pseudosteady state, etc.)

JE 17547

SAMUEL C. WI/WI
PU = -0.86857*m ..................,,..,,..,.,.....!. ....................... (14) Note that technicality PU is negative since -q was selected as the reference rate. in order to insure a proper (positive) value for permeability, the qfl must be signed in the definition of Pu. The reference rate for this example is negative (qR=q). When the reference rate is positive then the slope, m, is negative so that Eq 14 still applies. 2. Completion efficiency is derived from the point in time where the semilog straight line crosses the potential existing in the well just prior to tR, (AtR), The characteristic time, tu , fOr this teSt is 4*AtR/V Or 2.245 ~AtR. tu = 2.2458*AtR .............................................................. ( 15 ) 3. if the test is a shutln test, as is this example, the extrapolated potentiai at infinite shutin time assumlp. radlai fiow describes aii transients, P*R, Is often of Interest. It can be calculated from the equation of the semilog straight line evaiuated at the limit of AteqR as the shutin time approaches infinity .This limit is aiways the value of AteqR evaiuated at tR,
[AteqR(tR)]. AteqR(tR)=(t-tl )hR1(t.t2)h R2=...*

These two examples illustrate expanded applications that are possible using the generalized form ot equivalent drawdown time. Each case first examines the type curve to determine what flow mechanism(s) are controlling the well behavior in each time region. After identifying the controlling mechanism, the appropriate analytical tool for that behavior is then examined; semilog plot for radial flow; square root plot for linear flow. This analysis was done on a personal standard application programs. computer using

WATERTON #1: PREFRAC TEST ON TIGHT GAS WELL Table #2 shows the data for Waterton #1. Note that effects of the variations of gas properties are ignored for simplicity. Both pseudopressure and pseudotime may be required in many real cases, The use of those concepts would only obscure the basic purpose of this example, to illustrate the application of generalized equivalent drawdown time. The obvious approach to this probiem wouid be to use AbqR and correct for each rate change as it occurs at the sand face. However, as is pointed out irr Appendix B - Special Case for Stepwise ShutIn, AteqR does some undesirable mathematical tricks when each stepwlse change in rate is inciuded. However, the desired effect can be obtained by using two transformations which do work, AteqLA and A~qR . The first transformation iS AteqL applied to each Stepwh R3t0 change caused by afterflow, hence the additional A subscript indicating that afterfiow is inciuded. This transformation is defined for aii times but is vaiid only if the behavior is linear flow. Table #3 shows the procedure used to obtain At@A. The second transformation is for radial flow behavior, but is oniy valid for zero afterflow. The approximate type cuwe for this data can be obtained by using the first transformation to define the eariy time behavior and the second to define the latter (radiai flow) part . Both are shown on Figure #1. Since this is generated data, the actuai type curve for a singie rate change (AteqX) is shown for reference. ( Too bad we cant do that for reai data ). The type curve defined as the upper envelope of these approximations clearly represents a reasonable facsimile of the underlying resewoir behavior. WATERTON #1 : TYPE CURVE ANALYSIS The behavior of the data using AkqR Cieariy shows Storage behavior from 0.001 hr to 0.02 hr, foilowed by a storageradial transition extending to 1 hr, foiiowed by radial fiow. As such one wouid expect that a semiiog anaiysis would yield vaiid radiai flow parameters (k and S) when applied to the iatter portion of the data. Also, type curve matching of normai type curves or of derivative group type curves might be appiied to the data beyond 0.02 hrs using any of the commoniy availabie theoretical type cuwes which exhibit storage-radiai transitions. However, be warned that the shape of the transition using AteqA wiii in generai not be the same as the drawdown transition since pure radiai fiow does not inciude pseudosteady weiibore flow effectsi WATERTON #1 : RADIAL FLOW ANALYSiS Figure two is a semilog piot USin9 AteqR. Quantification radial fiow parameters can proceed in typical fashion. of

(t-tN-l )~RN. I ..................

................................ (16)

For a Horner rate history this is tp; for this example it is 512 hours. The equation for the semiiog straight iine is: pw = pR.(l/2)*pU*in( 4*AteqR/tU) .............................( 1 7)

Therefore the extrapolated potential is: P*R = pR-(l/2)*pu *ifl{4*[AteqR(tR) ]/tU} .................( 18)

Bear in mind the sign of qR and therefore PU f. (his e)(afTIpie is negative, : EXTENDED SOMBRERO #l iNTERNATIONAL OiL DISCOVERY EVALUATION OF

Tabie 4 shows the derived test data for this example. It was generated using an (assumed 1) piecewise linear dimensionless potentiai function. S-ch a function is definitely not a solution to the diffusivity equation, but represents pure iinear fiow, followed by pure biiinear fiow, foliowed by a second pure iinear fiow period. As such it has a basic similarity to a f(acture system of limited extent, draining a larger reserve c.>ntained in tight rocks where some pressure drop occurs in the fracture system. It is fundamentally simiiar to the finite capacity fracture soiutions, commonly used in analysls of tight gas production. SOMBRERO #1 : TYPE CURVE ANALYSIS Figure 3 shows the Type Curve for this weii. Included are severai standard equivalent times. This figure couid easily be interpreted as an example of why equivalent drawdown time doesnt work- its too confusing. However the things that make the plot complicated are integrai parts of well behavior, and as such can be present in most any real test data. A much more reasonable attitude toward Fig 3 is that iife is complicated but if you reaily understand equivalent drawdown time, it can be a vaiuabie asset in explaining complex flow data. The first three parameters plotted ciearly illustrate that neither At, nor Ateq A, nor Ateq T Mkibiy refiect the underlying drawdown behavior. They aii appear to indicated 601

10 The slope (m) of the semiiog straight reveals the quality of the reservoir. The characteristic potential (Pu) for this test (based on qR = -q) is -2*m/in(l O) or -0.87 *m, Conventional reservoir quaiity parameters (transmissibility, permeability thickness and permeability) are then derived from the definition of Pu.

APPLICATION -. . .. _. ... . . . . OF EQUIVALE~


some radial flow followed by something, maybe the beginnings of depletion effects. These implications are totally erroneous. They fail because none faithfully reflects the actual rate history of the weii. The first time parameter, At, assumes no effect on potential from prior rate changes- that is prior to the main one under czrnskferation. The other two (AfaqA and A!eqT) fail primarily because they use the Horner rate history, and for this case that is an inappropriate model. A minor effect is that neither AfeqA nor AteqT is the apprOfXiate behavior function. This iast reason is the primary explanation of why AfeqL, and especially AteqR, fail. Aithou9h they eccurateiY refiect the rate history, they faii to use the appropriate functional behavior. it is worth notin9 that AfeqL WO~S much better than AteqR. ( i am biased for linear flow, beoause in mY experience it works better in the many of the welis that i testeven high rate oii wells. In fact, one could justifiably argue that the additional improvement gained by the Seif inferred function, Ataqs is not worth the extra troubie.) However, one of the main points of this example is that sometimes the observed behavior, AP versus At can be used to obtain a better view of the underlying characteristic behavior that can be obtained from any a priori assumed behavior. Sure enough, Ateqs is the best of aii the data sets shown. Except, of course, Ateqx, which is eXact, and would not be available f?r a reai weill Either AteqL or Ateqs give a reasonably good view of the actual ideai drawdown behavior of this well. SOMBRERO #1 : LINEAR FLOW ANALYSiS Figure 4 is a square root plot using Ateqs. Quantification linear flow parameters can proceed in typicai fashion.

DRAWDOWN TIME IN WELL TESTING

SPE 17547

For a Horner rate history this Is tp; for this example it is 63.83 hours. The equation for the square root straight line is: pW=p R-[2*pLJ/d(@/d( tlJ)]*d(AteqL) ,,,..,...,..,,..,.,.,, (2 1 )

Therefore the extrapolated potentiai is:


p* L=p R-[2*pu/d(~)/d( tu)]*d(AteqL[tR ]).,,., ....... ..(22)

Bear in mind the sign of qR, and therefore Pu, for this example is negative. Note that for this exarnpie two linear flow behaviors are seen, or nearty seen. The most important in terms of iong term behavior is the one which is being approached in the latter portion of the data. This wouid represent the behavior of the fracture face for a massive fracture or the behavior of the matrix for a naturaiiy fractured reservoir. if the square root straight iine is seiected at the end of this data then the characteristic iinear fiow parameter and the ioss in potentiai relative to this region is obtained. If the eariy line is seiected then these same parameters are obtained for the near weiibore region.

CONCLUSIONS
The foiiowing conclusions are drawn. 1. Equivalent drawdown time is a powerful tool for reducing complex fiow histories to equivalent drawdown histories. A In theory, any rate history can be so transformed. in practice, it is essential that the functional nature of the characteristic response function of the system be defined. This definition can be from a priori knowiedge, an assumed behavior or from extensive observation.

Of

B.

The slope (mL) of the square root straight iine reveals 1. the iinear flow parameter, Pu/d(tLr). This parameter shows the combined effect of cross-sectional area to fiow and permeability. The characteristic linear fiow parameter [P@d(tu )] fOr this tC%t (based on qR = -q) iS d(7t)*ITIL/2 or -0m86623*mL. Conventional linear fiow parameters (cross-sectional area to flow or fracture iength) are then derived from the definition of Plj/d(tu) if the other required parameters are known. PIJ/4(tu) = -0.88623*mL ................................... ..........( 1 9)

2.

Equivalent drawdown time depends strongiy on the nature of this functional behavior of the system. Some correction for the effects of muitipie rate changes is essential unless the final rate change is totaliy dominates the potentiai changes. To be certain that this is so, the finai rate change must be separated from the next most recent rate change by more than 100 times the duration of the test. Agarwais equivalent time (Horner rate history; radiai fiow assumed): 4.a Applies to a specific rate history and then only when ail transients are in radiai fiow, 4,b Works perfectiy within these restrictions 4,c Has widespread and important application because of the importance of radial fiow behavior and of the flexibility of the Horner rate history in satisfactorily approximating other rate histories. Compiex rate histories andlor other fiow behaviors demand other forms of equivalent time. Generalized equivalent drawdown time is easiiy simplified to speciai cases if the functional form is assumed. Under favorable conditions generalized equivalent drawdown time can be used to obtain functional form if rate history is known.

3.

Note that technicality PU is negative since -q was seiected as the reference rate. In order to insure a proper (positive) value for permeability, the qR must be signed in the definition of Pu. The reference rate for this example is negative (qR=q). When the reference rate is positive then the slope, m, is negative so that Eq 19 stiil appiies. 2. Loss in potential between the observation point and the region causing the linear fiow is simpiy the difference between the intercept of the iinear fiow straight iine and PR.

4.

5. if the test is a shutin test, as is this exampie, the 3. extrapolated potential at infinite shutin time, assuming linear flow describes ali transients, P*L, is often of interest. it can be calculated from the equation of the square root straight iine evaluated at the iimit of AteqL as the shutin time approaches infinity,This iimit is aiways the value of AkqL evaluated at tR, [AteqL(tR)]. Ateq L(tFr)=[R I d(t-tl )* R2*~(!-t2)*...
FtN.l*~(t-t&J.l

6.

7. *
......... (20)

)]A2 ,.... .............., .!!.,.,..,.,,,,,

PE 17547

SAMUEL C. SWIFT f%
Cross-sectional area to flow at the well The value of potential in the well at the particular time, tR. The ValUe of tR (and therefore the value of PR ) is totally arbitrary. The value of potential in the well at any time, t. The value of potential in the well at the particular time, tl. For a drawdown test, this is the initial (static; quiescent) potential. The unit of change for dimensionless potential. This notation is used when the referenoe rate is fixed at some arbitrary value. The unit of change for dimensionless potential based on the first rate change, Aql. Ratio of the I-th change in rate (Aqi) to the reference change in rate (qU). Ratio of the first change in rate (Aql ) to the reference change In rate ( qu ), Production rate during drawdown, Production rate foiiowing the i-th rate change. The magnitude change in rate. Arbitrary of the arbitrary reference

Ct f(x)

Total compressibility of the system An arbitray function of x

Pw Pw@tl

fl(x)

The arbitrary function of x associated with the i-th rate change PIJ The inverse function of t(x) , so that x = f-l ( f(x)). Formation thickness An index value referring to a particular rate change. If the maximum value of i is 1 then it may be thought of as indicating the initial condition. Formation permeability ~ line. qi w PIJl

f-(Y) h

-.

Ri RI

k m ML N

Slope of semiiog (radial flow) straight iine. Slope of square root (Linear flow)straight

Total number of rate changes in a known rate history. Dimensionless Potentiai unique units (Pu). change measured in

PD

m rw

reference

rate. a right

PDCI

PD for a right circular Cyiinder located in an Infinite resewolr and produced at constant rate. PD for the Exponential Integral solution. PD for Pseudosteady flow,

Welibore radius for a well having circular cyiinder for a welibore. Arealiy equivalent weilbore

rw A t

area, Aw/(2rcrw)

PDEI PDP

Real time in any unit measured from any fixed point of time. The arbitrary time which at which the i.th rate change occurs. Dimensionless Time change measured in unique units (tu). Dimensionless Time measured from the beginning of flow (for a single rate change) and in unique units (tit). Dimensionless Time from the i-th rate change to the reference time, tR, and measured in unique units (tU). Dimensionless Time from the i-th rate change to the current time, t , and measured in unique units (tu), Producing time for buildup; faiioff, injection time for

ti PD for a right circular Cylinder located in an infinite reservoir and produced at constant rate with initiai P seudosteady fiow from the wellbore (Storage). PDLI PDLN PD for iinear fiow. The semilog tD /V)) straight iine function, (1/2) (ln(4 !DRi FDs Pm PD for Steady flow. Dimensionless Potentlai change measured from Pw@tl , in unique units (Pu). If there is only one change in rate then FDU is Identicai to PD. However if move than one rate change Is invoived, then PDU is referenced to an arbitrary rate qu. Note that this process requires the introduction of the coefficients, Ri. Dimensionless Potential change measured from PR , in unique units (Pu). Totai sum of the Dimensionless Potential changes caused by each of the severai rate changes, Aqi

tD

tDu

tDui

tp

Pm

PDUT

603

L
tlJ tl
x

tR

Arbitrary reference time; fixes PR. Characteristic time; the unit of dimensionless time; 0* f.LCtrwAA2fk Time at which the first rate change occurred Arbitrary (normaliy y = f(x) independent) variable, e.g

10
Y

APPLICATION OF EQUIVALENT DRAWDOWN TIME IN WELL TESTfNG


Arbitrary (normally dependant) variable, y = ~(x) but also seen as x = tl( y ) e.g l_l(l=l,N):(xI) Repeated from 1 to N. Fluid viscosity Transcendental mathematical approximately 1.7811 multiplication

SPE 1754 . . -.
of XI over 1,

B Ap At

Formation volume factor, reservoir volume per surface volume. Change in potential; PR-Pw Test time, simple flow time for a drawdown; shutin time for buildup; etc. Special dimensionless equivalent time for Horner rate history; radial flow assumed; qR=q; tR=O; same as Horner time, Special dimensionless equivalent time for Horner rate history; radial flow aSSUfTIed; qR=q; tR=O; same as Horner ratio. General equivalent drawdown time with no restrictions as to rate history or as to the well behavior. Special equivalent drawdown time with two equal and opposite rate changes and with radial flow assumed for all transients, This is the most widely known form of A~q. Special equivalent drawdown time with an arbitrary rate history but with linear flow assumed for all transients. Same as AteqL except afterflow rates are calculated and rate changes due to afterflow are included. Special equivalent drawdown time with an arbitrary rate history but with radial flow assumed for all transients. Special equivalent drawdown time with an arbitrary rate history but with the functional behavior Sslf defined. While this sounds Ideal for all cases, in reality the entire range of the function Is never known from actual data, so that some extrapolation is always required, This sometimes gives rise to Intolerable errors, It the old story, the more data the better the interpretation, assuming the data to be reliable. Special equivalent drawdown time with two equal and opposite rate changes and with linear flow assumed for all transients (Tandem linear rate changes). Equivalent time at which the semilog straight line crosses the value of PR. The magnitude of the change in rate at the i.th rate change The magnitude of the change in rate at the first rate change. For a drawdown test, this is q minus zero, or simply, q. ,N):(xi) Summation of xi over i, from 1 to N. Porosity Transcendental mathematical approximately 3.1416. constant;
604

k Y

constant;

ACKNOWLEDGEMENTS
This paper would never have been written were it not for those students who participated in my couree, Pressure Transient Analysls in Tight Rock held In Calgary December 7-11, 1987. Their Interest, enthusiasm and comprehension were an inspiration to me, To each of them go my sincere thanks.

AtDcqH

AtDeqHR

REFERENCES 1. Miller, C, C., Dyes, A, B., and Hutchinson, C,A.,Jr.: The


Estimation of Permeability and Reservoir Pressure From Bottom Hole Pressure Build-Up Characterlstlcs, Trarr.,AlME(1950 )189, 91-104 2, Horner, D.R.:Pressure Build-Up in Wells, Proc,, Third World Pet. Cong., The Hauge. (1951 ) Sec.11, 503-523 Scott, J.O.:The Effect of Vertical Fractures on Transient Pressure Behavior of Wells, J. Pet. Tech (Decl 963) 1365-1369; Trans., AlME,228. Matthews,C.S, and Russell, D.G,:Pressure Bu//dup and How Tests in We//s,Monograph Series, SPE, Dallas(l 967) 1. Millheim, Keith K. and Cichowitz, Leo : Testing and Analyzing Low-Permeability Fractured Gas Wells, (Febl 968) 193-198; SPE1 768 J,Pet. Tech. Trarrs,,AlME,243 Agarwal, Ram G., A1-Hussalny, Rafi, and Ramey, H.J., Jr.:An Investigation of Wellbore Storage and Skin Effect in Unsteady Liquid Flow: 1. Analytical Treatment: Sot. Pet, Eng. J.(Septl 970) 279-290; Trarrs.,AIME, 949 Theory and Practice of Testing of Gas Wells, third edition, Energy Resources Consewation Board, Calgary, Alta. (1975) Earlougher, R.C, Jr.: Advances in Well Test Analysis, Monograph Series, SPE, Dallas(l 977) 5. Agarwal, Ram G.:Real Gas Pseudo.Time A New Function for Pressure Buildup Analysis of MHF Gas Wells, paper 8279 presented at the 1979 SPE Annual Technical Conference and Exposition, Las Vegas, Sept23-26.

Atsq

AteqA

3.

AteqL

4.

AteqLA

5.

AtsqR

6.

Ateqs

7.

8,

AteqT

9.

AtR

Aqi

10. Agarwal, Ram G.:A New Method to Account for Producing Time Effects When Drawdown Type Curves are Used to Analyze Pressure Buildup and Other Test Data, paper 9289 presented at the 1980 SPE Annual Technical Conference and Exposition, Dallas, Sept21 -24. 11, Swift, Tulsa Samuel (1984) C.: Practlca/ We// Testing, O G C 1,

Aql

X(i=l
0

SPE 17547 APPENDIX A: GENERAL CASE

SAMUEL C. SWIH DERIVATION


This oan be written as:

11

SfMPLE FLOW TEST For a simple flow test (drawdown test):

PDIJ =
Where FtI =

..................................... (A-e) Rifh)u) ,,..,,..,..,,,,.,,

f%=

i(tD)

....................!.,................,,. ,,,,,..,,,.., .......... (A-1 )

Aq~lqu ,............................................................ (A-7)

Where: PD is the Dimensionless Potential. tD is the Dlmenslonless Time. Dimensionless Potential is the change in actual potentiai, measured in characteristic units for the particular test conditions under consideration; or: pD = (P@tl-Pw)/PLi 1....................................... (A-2)

Where the additional subscript U is added to indicate that all changes are measured from the initlai potentiai but that these changes are measured in dimensionless units which are referenced to an arbitrary rate change, qu. PDu = (fw@tl-pw)/pu ................................... (A-8)

Where: pu = qu *B/( 2*z*k*h/p) subscript U ................................(A.9) on tD u is SOleiy fOr

Where: Pw@tl is a constant equal to the pdential in the weii at t = tl, before the initial change in rate occurred. Pw is the potentiai in the weli at any time, a variable. PU 1 is the characteristic potential p[ ticuiar test conditions; or pul =Aql*13/(2*n*k* change for these

Use of the consistency.

A commoniy seen example of the function , f ,isthe function for the semilog straight line which several solutions to the diffusivity equation approach for sufficiently iarge vaiues of time. pD = = PDLN(tDU) (1 /2)*in(4*tD V is

h/~) ................................ (A-3)

U/V) ..........................................(A-l o) Eulers constant, 1.7811 and the

Where Aqt is change in volumetric rate, B is formation volume factor, k is permeability, h is thickness, and ~ is viscosity. Likewise, Dimensionless Time is the change in actuai time, measured in units which are characteristic of the parilcular test conditions under consideration; or:

Where,

iipp?OXhWitiOfl !S vaiid fOr valUeS of tDu greater than 20.

COMPLEX FLOW TEST For a complex flow test:


PDUT =

Z(i=l .N):[Ri*~i(tDUi) ]................................(A.l 1)

Where: RI
Aqi

=
=

Aq~qu .................. ................ ..... ..o..... .... ......... (A-1 2)

qi.q(i.l) .........................................................(A. ~3)

tD =

(t-tl )/tu.......................... .........................


...!.

(A-4)

qi is production rate foiiowing the i-th rate change, and qu is the arbitrary rate used in Pu. N is the total number of rate changes i is an index designating a particular rate change, and

Where tl is the time at which the initiai change in rate occurred - a constant. t is time measured variabie. from an arbitrary reference, a

tu is the characteristic time particular test conditions; or: tlj = O*p*Ct*rw

change

for

these

X(i=l ,N): designates a summation from i=l to i=N. Note that for rate changes within a singie weilbore, tu is normaliy constant, so that: tDLJi u (t-ti)/tlj .......................................................(A.l4)

.........oo.(A ......oo.. A-5) ( A/k ................m

Where a is fractional porosity, p is Viwrsity, ct is totai compressibility, rwA is areaiiy equivalent weiibore radius, and k is permeability.

if it is assumed that there is no difference in functional form for aii rate changes: fi(x) So that:
pDUT = Z(i=l.N):[Ri'i( tDUi)] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ..(A.le)

f(x) .................................................

...!.

. . . . . . . . .

(A-15)

Note that for a simple flow test, Aql was (qO) or simply q, the productbn rate. For a comptex test each rate change causes a potential change which Is proportional to the magnitude of that particular rate change. Analogy to the simple case is presewed by using the arbitrary rate, qu, both to normalize the coefficients of the time functions (Ri) and to fix the value of PIJ. EQUIVALENT DRAWDOWN TIME

In general the dimensionless pressure terms in the preceding equation invoive multiple rate changes, and therefore are obtained from the standard superposition equation for the complex case derived eariier:

I I
I

pDIJT(tDIJi)=~(i=t ,N):Ri*f(tDui)....... ........... ...


and,

.,0,..,

,..,.

(A-22)

pDuT(tDRi)=~(i=i ,hJ):Fti*f(k)Ri)...........(A-23) ...... .......... ......


Therefore: pDR*~(i=l,N):Ri*[~( tDUi)-~(tDRi)] ...........................(A-24)

The concept of equivalent drawdown time, as used in well testing, involves two separate ideas. First is the classical mathematical concept of equivalency - attempting to reduce a complex problem to an equivalent simple one. Second, is recognizing that the reference level for potential, from which the change in potential is measured, is arbitrary. Consider the second idea fi,-1. In order to measure potential (or anything eise for that matter) it is necessanr to define the starting point (reference level or value) for the quantity and the units in which changes from that ievel wiil be measured. For ordinary dimensionless potentiai the reference vaiue is Pw@tl (the quiescent condition) and the unit is PU. However, both the reference potential level and the units of its measurement are totaliy and separately, arbitrary. The unit of measurement for potential is fixed by seiecting an arbitrary reference rate. The reference level of potentiai from which changes in potential are to be measured is fixed by seiecting a reference time, tR. in order to emphasize that the reference conditions are arbitrary, some redefinition of the quantities used to measure dimensionless potentiai is required. The unit of measurement which has been previously defined is Pu. in the simpie case, Pu, was determined by the rate, q. But in the more generai case, PU, was determined by the arbitraty rate, qU. This was possible only because of the introduction of the coefficients, R i, which normalize each actual rate change, Aqi, to the arbitrary rate, qu. Therefore no change is required in PU as iong as it is fully understood that qu, and therefore PU, is totaiiy arbitrary, and that the rmrmaiizing coefficients, Ri, are required in the summation. The effect of the use of an arbitrary reference level of potentiai is more subtle. instead of measuring changes in potential from the initiai condition they wiil now be measured from the value of potentiai that occurred at the arbitrary time, tR. This potential is defined as the potential in the weli at tR, or, pR. The finai effect of both the arbitrary reference rate and the (independently) arbitrary reference Ievei of potential is manifested in the reference dimensioniess potentiai: PDR= (PR.Pw)/PU .......................................................(A.l 7) PDR= [(Pw@tl-Pw)-(Pw@tl PDR= pDUT(tDUi).pDUT(tDRi) -PR)]/PIJ ........................(A- 18) .....................................(A.l9) time evaiuated at any

I I
I
I

If we now return to the first idea invoived in equivalent drawdown time, we must ask if the compiex case shown above can be modeled as an equivalent Simple case by USin9 AtDeq. Obviousiy if this is possible then: pDR=~(AtDeq) ................................................................. (A-25) So that ecfuivaient drawdown time, if it exists, is the soiutir-m to: i(AtDeq)=~(i=l, N): Ri*[~(tDUi)-t(tDRi) ] ...................(A-26)

I I

Proceeding with the soiution,


AtDeq=f-l(z(i=l, fd): Ri*[f(tDui).f(tDRi)

]} ..... .... ......................... .......(A.27)

Where ~-1 is the inverse function off Or by using the definition of AtDeq: Ateq = tU*~l{Z(i=l
[~((t-ti)/tlJ

, or

x = fll { ~[ x ] }.

,N):Ri*
)-f((t13 -ti)/tlJ )]} . . .. . . . .. . . .. . .. .. . .. ...(A-26)

This is the general form of equivalent drawdown time for an arbitrary rate history and an arbitrary weil behavior,

I
I

APPENDIX
AGARWAL

B: SPECIAL

CASES

EQUIVALENT TIME

The development done by Agarwai was in fact a speciai case of Ateq for a Horner rate history and assuming radiai flow behavior. in terms of the generai Ateq development, these assumptions are: ........................................................!.00..> .,..!(B-l ) =
=
=

IN=2

Aql
Aq2
@

+q ....................................................................(B.2)
-q .. .... ..... ....... ......... .......!. ............ ........ ............ (B-3)
.q .. . .. . .. . . . . .. . .. . . . . .. . .. . . . .. . . .. . .. . . . . .." .. . . ..`". "..." ".. ..""(B"4)

Where tDIJi is ordinary dimensionless time, t: tDui=(t.ti)/tU

I
I

R1 R2

= =

+q (-q

)/(-q) )/(-q)

= =

-1 ................................(B-5) ................(B-6) +1 ,..........(B.6)

ltR=
tl = =

tz-tl

tp...........o..........(B.7) ..........

.. ... . .... .... ... .. ... . ... .... . ... .... . ... .... . ... . ... .... .. ... ..(A.2O) o.......................................................`............(B.8) tp ....................................................................(B.9)
At ..... ..... ............. ..... ........... ..... .... .... .... ..... .....(B.lo)

and tDRi iS dimensionless tiITIe evaluated at the referenCe time, tR: ...........................o...................................(A.2l) tDRi=(tR-ti)/tlJ
I

tz

t-tz =

606

SPE17547 t-tl =
o

SAMUELC.SWIFI
tp+At ............................................................(B.l 1)

13

Deflnlng A~q, subject to the above assumption as AbqR:


AteqA ={tu}*{Y/4}*{i oA[2*y/tn(i

Xso ......................4 B.12) O(4*XIV)] x>O..,.....................(B-13)

0)]} ....................(B.28)

f(x)={ [In(l 0)/2] *[logl V ~-l(y) = y = X(i=l ,N):Ri*[logl O#([t-ti]ltU) -lOgl O#([tR-ti]/tU)] ..................o....(B-29) X < 0.......................(B-3O) ={ V/4}* {10 A[2y/ln(10)l} ..............................(B-15) loglo#(x)=l [ln(l O)/2] [loglo(4*x/v)] v= or: Y s X(i=l, N):logl O{(t-tijRi#/ (tR-ti)ARi#} .........................(B-33) .....(B.31) X>o ,,,,.,, ........31)

1.7811 ...........................................................(B-14)

Note that either of these approximations may do very strange ihii~~s for ti rms greater ihaii O but less than 20 q tu (20 time constants). Defining A@, subject to the abOVe assumption as AteqA:

1,7811 ................................(B-32)

AteqA={tlJ}*{~/4 }*{ioA[2*y/[fl(
Y Y= = ~(1-l,2):Ri*[~(

lo)]}

(B-16) (B-17)

tDUi)-~(tDRi)]

where: -f([t-o]/tu) +f([tp-o]/tlJ +f([t-tPl/tu -t([tP-tPl/tu) O)/2]*{ + + ) ) (B-16) xARi#={ xfiR\ for x>O...........................(B-35) (V
q

.,.........(B .............(B.34) tu / 4)Ri for XSO

Y = [ln(l

lloglo(4*[t-o]/ttJ/v)] [loglo(4*[tp-o] /tlJ/v)] [loglo(4*[t-tp] /tlf/v)] [0]} .............................................(B.l9)

Note that both time difference terms in equation 6-31 are kept separate in order to avoid oonfusion in ihe application of the Ri# function. Either, or both, arguments may be less than or equal to zero. So that:
AteqR =

Clearly: [2*y/ln(l O)]=


tp+At)/tLt/V)] .,......,............... (B-2o)

{tu*y/4}*{~(i=l, N):(t-ti)ARi#/ (tR -ti) AR i# } ..................................................(B.36)

[tog10(4*tp*At/(

where l_l(i=l,N):(xi) and

XI*;.2*X3*...*XN

. . .. . . .. . . .. .. .. .. .. . . .. . . . ...!.. (B-37)

Or: (10 A[2*y/ln(10)])= {4*tp*At/(tp+At) And:


At~qA=(tLJ}*{~/4)* {4*tp*At/(tp+At)lt@4)

(Y xARi#={ /tU/Y } ,..,,.,.,0,..,.... .,.,,,.,,,...!!. (B-21 ) EQUIVALENT

.,........................(B-34) tu / 4)ARi for XSO for x>O,..........................(B-35) LINEAR

xARi TIME-

. . . . . . . . .. . . . . . .. . . . . . . .. . . . . . . . . .. ...(B-22)

If the flow history is arbitrary but linear flow behavior is assumed: X<o .......o................... .......(l3-38) f(x) ={ 0 2

So that, finaliy:
AteqA = tp
q

At /(tp+At) . ... . .... .....(B.... . ..... .... .. .. ....(B-23)

q SQRT(x/rr)

X <0

.................................(B-39)

EQUIVALENT

TIME-

RADIAL f-l(Y) = 7c*(y/2)A2 .,,,,.,.,,..,,., ..............................(B-4o)

If the flow history is arbitrary but radial flow behavior is assumed: .......(B-24) X50,..,............(B f(x) = {0 [ln(l

ARBITRARY RATE HISTORY


Defining Ateq, subject {tlJ}*(n}*{[y/2 to the &x3Vt3 aSSUIT@On ]A2} %5

AteqL:

O)/2]*
x> O .. ................. ......(B-25)
AteqL =

[ioglo(4*x/Y)]

....................4B ..............4B-41 )

Y= I.7811 ~-l (y) s {W4}*{10A[2*y/in(

..,,.,...,,,,...,.., ................(B-26)

q {Z(i=l ,N):Ri* {2/SO RT(fi*tu)} [DSQT#(t-ti)-DSQT# (tR-ti)]} ..........................(B-42)


o

lO)l} . .... .. . ....!.. .. .... .. ...(B-27) X<o .... ........ .... .................(B-43) X>o ..................................(B.44) DSQT#(x)=( sQRT(4x/7c) or:

Note that either of these approximations may do very strange greater than O but less than things for tImes 20 q tu ( 20 time constants ).

OuI

APPLICATION OF EQUIVALENT DRAWDOWN TIME IN WELL TESTING


AtaqL={z(\=l ,N):RI* [SQRT#(t-ti)-SQRT# where: o SQRT#(x)={ SQRT(X) HORNER RATE HISTORY For this special case: N=2 Aql Aq2 %t R1 R2 tR tl tz t-t2 t-tl = = = = = = = = = = ..................................................o ...............(B.46) ..,..,,.., ............(B-49) +q .....................................,,....0 .q ..................................................................(B.5O) .q ..................................................................(B.5l) (+q)!(-q) (-q)/( tz-tl -q) = = = -1 .......................................(B-52) +1 .......................................(B-53) tp .......................................(B-54) X>o.,,,.,..............................(B-47) Xso .....................,......0......(B-46) (tR-ii)]}A2 ..,...,...,...,., ......(B-45)

SPE 17547

The behavior of the parameter, tu , is tied to the mathematical properties of the function, ~ , For the PDLI function which is weil behaved near t=tR, the parameter, tu , falls out of the equation. So that the above equation reduces to Eq (B-43). Note that splitting the summation has no effect. Unfortunately, the wideiy used and more familiar logarithmic approximation, PDLN is not so kind. In the case of the logarithmic approximation, the summation must be split yet a second time, to see what Is OCCurring.Let tL be the time of the iast rate change which occurs before the time, t, that the current measurement measurement of potential is made. Then the sum from L+l to N is zero so that the summation In Eq(B-64) is truncated at L.
Ateq ~
+2(i=M+l

tU*~-l(X(i=l

,M):Ri
/tU)]} .,, .,.,!,....,.,.. ,. .,,,.., (B-64)

q [i((t-ti)/tU

)-~((tR-ti)/tU)] ,L):Ri*[j((t-ti)

When ~ is assumed to be PDLN: AteqR= {ll(i=l,M): [(tti)/(tRti) ]ARi} q {~(i=M+l ,L):(t-ti)ARi} q {(tU*Y/4)fi[l -Z(i=M+l ,L) :Ri]} .....(B ..................(B-65) This last term is the probleml For the special case where the exponent, [1 -X(i=M+l ,L):Ri] , Is zero (after the enfY of afterflow) this term setties down and behaves. However prior to that time the value of tu affects tke value of equivalent time. Furthermore this effect changes with time (as L Increases), ieaving no effect only when the true sandface rate is zero. As if this were not bad enough, when the sandface rate changes continuously some of the dimensionless times are such that the function PDLN may be a poor or intolerable approximation. PCLN is a vaiid approximation oniy for tD > 20, it is poor for W4 < tD e 20 and it becomes physically intolerable fOr tD c W4 , giving pressure change contribution with the opposite sign of the real changes. Therefore, AteqR, is a poor candidate for analyzin9 rate changes which are influenced by afterflow. Authors note: This incidentally explains why Horner plots on buiidup data in tight weiis (infiuencecf by oniy iinear flow and storage) appear to break over to a fotaily meaningless semilog straight iine. This break is not caused by a storage/radial transition but rather by misiocation of the storage affected data. if one contemplates the known behavior of storage/linear transitions, no such break should be seen. Therefore such a break is frequently interpreted as finaiiy getting to radial flow. The type curve for these cases is self justifying when the speciai case of At~qR , narnelY, A~qAt is used, since the same errors occur on the iog-iog plot. A very abrupt transition is seen which implies radiai fiow with severe damage. if AteqL is used then Only linear fiow is seen. This is a ciassic case of two distinctly different weii behaviors causing neariy identicai observed data. Either couid be correct, but the physically correct one is normaliy indicated by the circumstances, not by the data. That is why krmwiedge and experience wiii always be a critical element of well test interpretation. However, AteqR often does become weil behaved again for iong shutins. in these circumstances afterflow approaches zero and tD ( hopefully but not necessarily ) is much greater than 20. if the iimit of Eq B-62 is considered as t approaches 00 then t-ti = t-tL ; and the exponent approaches zero, so that:
linlit(t=_)Ateq R=l/i(H(i=l,L): [(tR-ti)lARi) . .. .. . . .. . . .. . . .. . . . ..(B-66)

O..................................................................(B.55) m tp ...............a.,,,.....,..,, .,,..,,.,,,..,,, ...................(B-56) At.................................................................,(B-57) tp+At ..............................................................(B.56) {[-l] *[ SQRT(tp+At)-SQRT( tp)] +[l][SQRT(At)]}A2 ....................................(B-59)

AteqL =

Or: AteqL=(SQRT(tp) +SQRT(At)-SQRT(tP+At))A2................. . ..(B-60) STEPWISE SHUTIN

An Interesting special case arises when a shutin test occurs as a stepwise process. This might be a surface shutin with wellbore storage (constant or changing ; or an intentionally stepwise process). In this Jase an interesting complication arises due to the fact that the coefficients of the time terms (usually 4/tuPf) do not quite disappear. For this case the rate history is broken into two parts. It is assumed that there are N total rate changes, but only M rate changes occur prior to the beginning of the test. Note that the rate change marking the beginning of the test falls in the second group ( i = M+l ). Also: tR = t(M+l) ...................................................................(B.6l)

@ =-X(i=l, M):(Aqi) (Negative of rate prior to stwtin)


So

. . . . .. .. .. .. . . .. .. .. . . . .. . . ..(B-62)

that:

Ateq=tu*~-l {Z(i=l,M):Ri *[ f((tti)/tu )-t((tR-ti)/tu)l +X(i=M+l ,N):Ri*[~((t-ti) /tU)]} .... .... ................(B-63)

608

SPE17547
Which fora Horner rate Mstory, L=l, reduces to:

SAMUEL C. SWIFT

15 .-

lirnit(t*co)Ate~R = tp ... ,,..,0.,,,.,.,,,, ..,. ,,, ,.,, .,....,,. ,,, ,.,,.,,,.,.. .,(B-67) Whlchclearly shows that If the flow time was insufficient to see radial flow effects, no amount of buildup time will give radial flow parameters.

APPENDIX

C: NUMERICAL

METHODS

The procedure used Is primitive numerical analysis (since the author lacks extensive expertise in the sublect). It is Included hereto clari~the actual procedure used, but also in the hope that It will shortly be brought somewhere near the state of the art In application of numerical procedures. My first attempt Involved the following steps: 1. Assume that the obsewed potential change versus the actual time change was a first approximation to the unknown function. (Note obvious problems are the methods of extrapolation to times before and after the observed data and to a lesser extent interpolation, I used (in log-log space) linear !merpolaticn and linear extrapolation from the nearest two points.)
2, Calculate equivalent time using known rate history and the previous approximation to ~(x).

3, Use Ap versus the transformed approximation to ~(x).

Ateq as the ungraded

4, Repeat steps 2 and 3 until it converged, Although this worked after a fashion, it tended to clump the data about vertical lines. After about 10 iterations the change from iteration to iteration increased, and so did the absolute error ( for this example I knew the answer ), My next attempt was to modify step three to: 3. Use Ap versus a weighed average of the old and new values (M*Ateq:old + Ateq:new)l(M+l ) as of Ateq , Afeq:app= the ungraded approximation to ~(x). M equal to 1 and 2 were tried. The results were only slightly obvious. better, with clumping still

My final attempt was to use step 3, as modified with M = 1 (a simple average), but to smooth the new average with a least squares fit straight line thru 5 points nearest to Ap. The new approximation to Ateq was then taken to be the value on this line at Ap, After 13 iterations (and much soul searching about deadlines and doing less than the best possible job) this looked good enough to me,

609

TABLE 1 EQUIVALENT DRAWDOWN

itMES IN COMMON SE U

N-m.

... .. .. ... .. .. ... .. . .... ..Symbol . . .. . . . ..~ . . . ... . ... . ...qR

tR,,, .. . .. .. R.sun . ...,...

Agarwal(TandemRadial) EquivalentTlma AganvalTandam Linear Equlvalant Time CammonTandemUnear EqulvalantTime Homer Tlma Hornar Ratio

A&A AteqT AteqT At~H A*R

Rarflal(PDLN) Llnaar(PDLl) Llnear(PDLl) Radlal(PDLN) Radlal(PDLN)

-q

tp

tpAt/(tp+At) [d(tp)+~(At)-~( Ip+At)]A2

-q +q -q +q

Ip O O O

[i(tp+At)-d(Al)]~2

At/[lP+At] [tP+At]/Al

Nota that both of tha commonly used Hornar parameter are actually equivalent drawdowntlmaa retarancadto the lime that flow was Initlalad.

TABLE 2 WELL TEST DATA FOR WA7ERTON #l

PARAM~ERS 0.045 0.00012 0.316 v. 0.67 CM 0.00313

1: 0-

0.::43 0.693

[$)) RbbUMCF

~: rw -

( pall) (ft)

(bbl/pal)

i 1 2 3 4 5 Observed ET(hr) 0.00 6.00 12.00 1S.00 24.00 Potentlela BHP(psla) 4255.00 2302,69 2029.00 2356.05 2314.56

RATE HISTORY tl(hr) 0 12 24 36 48

Aql ( MCFD ) 1000 -250 -250 -250 -250

*** CALCULATE

FROM

POTENTIAL

BEHAVIOR *****

teff n 46+36+24+12

120hr

30!00 36.00
42.00 46.00000 4a.00020

2756,31 2783.2S
3274.51 3338.70 3339.16

At(hr) 0.00000 0.00020 0.00050 0.00100 0.00200 0.00500 0.01000 0.02000 0.05000 0.10000 0.20000 0.50000 1.00000 2.00000
5.00000 10.00000 20.00000 4S.00000

Ap(psl) 0.00
0.46

AteqA(hr) AteqT(hr) AtaqR(hr) AtaqL(hr) AteqLA(hr)


0.00000

0.00000
0.00020

0.00000
0.00020

0.00000
0.00020

0.00000
0.00000

0.00020

E s

48.00050 48.00100 46.00200 48.00500


46.01000 48.02000 46.05000 46.10000 48.20000 48,50000 49.00000 50.00000 53.00000 58.00000 66.00000 96.00000

3339.80 3340.83 3342.81 3348.33


3356.56 3370,93 3404.69 3444.77 3497,58 3562.76 3654.17 3730!07 3S36.66 3923.62 4013.29 4116.69

1.10 2,13 4.11 9.63


17.68 32.23 65.99 106,07 158.88 244.06 315.47 391.37 49S.18 585.12 674.59 777.99

0.00050 0,00100 0.00200 0.00500


0,01000 0.02000 0.0499S 0.09992 0.19967 0.49793 0.99174 1.96721 4,60000 9.23077 17.14266 34,26571

0.00050 0.00100 0.00199 0.00497


0.00991 0.01974 0.04696 0.09713 0.19192 0.46626 0. S1096 1.75113 4.04062 7.37109 12.91891 24.21794

0.00050 0.00100 0.00200 0.00500


0.01001 0.02003 0,05017 0.10069 0.20277 0,51718 1.06S00 2.26623 8.55794 15.58745 37,78200 105,0000

0.000s0 0.00100 0.00201 0.00503


0.01006 0.02023 0,05092 0.10259 0.20730 0,52833 1.07754 2,20453 5.64S96 11.24003 21,25617 42.02279

0.00000 0.00002 0.00C06 0.00035


0.00125 0.00463 0.02097 0.06133 0.17058 0.57744 1.41175 3.47102 11.72269 30.36207 80.19631 272.0291

SPE 17543
610

TABLE 3 WA7ERTON OATA@RRECTED FOR AF7ERFLCIW #l

Note

qFf Out: -.250

. qsurface If thesloraga capacity oftlw wllbore changas, (0.01M52) anh changed anyetep C c at LhdlofAteqtAas ET goesto - is 512 hr

rncfd fU-lE+40: (Th18reduces AleqPDCiAto 4teqLA) Storage prior 10 flmf Shutfn is ignored Storage aflerflnal shutin Ia constan! al 0.00452 rwflpd AIIer shulin qsf h glvon by((Ap[l+l) .Apl)/(AI[i+l]-All) 0.00452.24}

I 1

ET (hr] 13HP(psla) Al (hr) 0,0000


12.0000 24.0000 36,0000 40.0000 48.0002 48.0006 48.0010 48.0020 48.0050 48.0100 48.0200 48.0500 48.1000 48.2ooo 48.5000 49.0000 50.0000 53.0000 58.0000 68.0000 96.0000

AP (Psi) @ (mCfd)Aql (mcfd) !000,0


750,0 500.0 250.0 246.0 231 0 224,3 215.5 190.6 179,1 155,6 122.1 67.0 57.3 30,8 15.5 8.2 3.9 1.9 1.0 0.4 1000,00 -250.00 -250.00 -250.00 .2.01 .16.98 .6.72 -8.77 -16,92 .20.51 .23,29 .33.68 .35.11 -29.70 -28.5o .15.32 .7.28 .4.37 -1,96 .0.92 -0.57 .0.40

RI
-4.0000 1.0000 1.0000 1.0000 0.0081 0.0679 0.0269 0.0351 0.0637 0,0821 0.0932 0.1348 0.1405 0.1168 0.1080 0.0613 0.0291 0.0175 0.0079 0,0037 0.0023 0.0016

AteqLA

(hr)

2 3 4 5 6 : 9 10 11
12 13 14 15 16 17 18 19 20 21 22

lU_l E+40 0.000000 0.000001 0.000004 0.000016 0.000065 0.000354 0.0013 0.0046 0.0210 0.0613 0,1708 0.5774 1,4117 3.4710 11.7229 30,3021 80.1963 272,0291

3336,7o 3339,16 3339,60 3340,63 3342.81 334s.33 3356.58 3370,93 3404.69 3444.77 3497.58 3562.76 3654.17 3730.07 3836.68 3923.82 4013.29 4118.69

0.0000
0.0002 0.0005 0.0010 0.0020 0.0050 0.0100 0.0200 0.0500 0.1000 0.2000 0.5000 t.000o 2.0000 5.0000 10.0000 20.0000 48.0000

0.00
0.46 f.lo 2.13 4.11 9.63 17.88 32.23 65.99 106.07 158.66 244.06 315.47 391.37 496.18 585.12 674.59 777.99

TABLE 4
SOMBRERO #l uUILDUPT EST DATA PARAMETERS h= P= 0. 36o 0,69 1.3 (II) (Cp) 0.19 0. ct - 0.000125 0.316 W. qRm tR = PR 25oOO 1440 4255 (BPD (hr) (pSt )

( psi.1) ( It )

RATE HISTORY I 1 2 3 Ii(hr) o 1439.75 1440 Aql ( SPD ) 5000 20000 -25000

POTENTIAL

BEHAVIOR

.. . ... .. .. . .. .. ..
Ip (hr) ET(hr) 1440,00 1440,01 1440.02 1440.05 1440.10 1440.20 1440.50 1441,00 1442,00 1445,00 1450,00 1464.00 1476.00 1408.00 1512,00 1524.00 1536.00 1560.00 1584.00 1008.00 i3HP(psia) 3839.24 3854,41 32!00.47 3872,14 3077,01 3801,90 3067,97 3692,16 3696.10 3900.15 3909.28 3925,1t 3934,61 3942.33 3954,78 3959.98 3984.74 3973.19 3960.55 3967.10 A!(hr) 0,00 0.01 0.02 0,05 0.10 0.20 0.50 1.00 2,00 5,00 10,00 24.oo 36.00 46.oo 72.oo 64.oO 98.00 120.00 144,00 t88.00 Ap(psi) 0.00 15.17 21.23 32.90 37,77 42,86 48.73 52.92 56.07 SO.91 70.02 S5,88 95.37 103.09 115,52 120.75 125.51 133.95 141.32 147.S6 155,66 220.42 348.52 492.S8 AlaqX(hr) 0.0000 0.0095 0.0166 0.0446 0.0690 0.1123 0.1912 0.2856 0.3544 0.4664 0.0140 1.8432 2.8037 3.6283 5.4932 8.0015 6.4840 7.3082 8.2204 6,9998 i: 50 100 . 208.2 AteqT(hr) 0.0000 0.0100 0.0199 0.0497 0.0991 0.1974 0,4917 0.970s 1.9178 4,7405 9.0941 21,2241 28,1871 34.7994 50.7796 51.303S 55.9792 69.3870 76.9587 03.6953 AleqR(hr) 0.0000 0.0097 0,0160 0,0432 0.0784 0.1250 0.2076 0.2759 0,3448 0.4374 0,5119 0,s157 0.80S4 0.7079 0,7683 0!7093 0.S096 0.8443 0.8733 0.S961 AteqL(hr) 0.0000 0.0085 0.0160 0.0343 0.0589 0.09s1 0.1705 0.2529 0.3708 0.6239 0.9501 1.0806 2.2174 2.7oOO 3.6019 4.0156 4.4121 5.1614 5.S024 6.5233 AteqA(hr) 0.0000 0,0100 0,0200 0.0500 0.1000 0.1999 0.4991 0,9985 1.9S62 4,9147 9.6647 22,1550 32,0025 41.1469 57,60S0 85,0425 72.0125 84.7232 98.0222 106.1324

Number of 13 AleqS(fIr) 0.0000 0.0004 0.0182 0.0427 0.0753 C.1191 0.1905 0.2817 0.3521 0.5640 0.9290 1.9300 2.8S44 3.3S36 4.5790 5.0804 5.5498 6.3549 7.0622 7.7616

611

SPE 17547

4400
1000

AteqX
4000

100
3300 ~ s 3600

10

3400

32U0 .1

10

100

1000

M-R

(l@

1 .001

.01

.1

10

100

Fig.

2- Samilog plot for Vifaterton#1

Ateq (hr]

Fig. 1- Typa curve for WatertorI #1

4000 1 B~ Q
u

u El
El

AteqR 100 , J , m
q

3900 # I

4 p A

P*L :4255 Ateqs* = [0.2*d(1440)+0.8*~(0.25)]2 =63.63

.01

.1

10

100

3800 0

1 dAte@(~hr)

Fig. 3-Type curve for Sombrero #1

fig.

4- Square

root

plot

for Sombrero #1

You might also like