You are on page 1of 10
NYes Interpretation of Pressure Buildup Test Using In-Situ Measurement of Afterflow Denis Meunier, SPE, Schlumberger Technical Services In. ‘MJ. Wittmann, SPE, Schlumberger Technical Service Ie. George Stewart, SPE, Herc Wat U. ‘Summary ‘The influence of afterflow on transient pressure response ‘has long been recognized in well test analysis. The type- ‘curve approach based on the simplified mode! of a con- stant wellbore storage coefficient has attempted to assess afterflow effects from pressure data alone. The introduc tion of a new generation of production logging tools that ‘can measure bottomhole flow (BHF) and pressure (BHP) simultaneously allows the afterflow in the well to be monitored atthe surface. Consequently, the well test can i i new form of allow the detection of a straight line on the appropriate semilog plot [e.g., Horner, or Miller-Dyes-Hutchison (MDH) is much reduced. The analysis of the pressure ‘and flow data transmitted from the production logging. device may be carried out atthe wellsite in real time so thatthe test can be continued until the desired objectives are achieved. ‘A field example shows how continuous BHF measure- ‘ment during the wellbore storage period can improve the interpretation of a short-duration buildup test. Introduction ‘The constant rate solution of the flow equation in an in- finite reservoir is the basis of the conventional analysis of well tests. Even ifthe reservoir is bounded, the analysis applies while the reservoir is infinite acting—ie., until the pressure change reaches the outer boundary. For the ‘conventional analysis, itis essential to maintain constant rate, and itis well known that ths is difficult to achieve. ‘The best way to maintain constant rate is to make it zero, hence the wide use of buildup tests. Because of the compressibility of the fluid in the wellbore, the BHF does not adjust instantly to the sur- face flow after a change of rate is imposed atthe wellhead. ‘This phenomenon is called “‘afterflow’" in a buildup test ‘or, more generally, “‘v'ellbore storage effect."” During this time, the sandface rate is not constant, and early data ‘cannot be used in the conventional (i.e., Horner or MDH) analysis, Furthermore, outer boundary effects may develop before the wellbore storage effect disappears, in which case the conventional analysis does not apply. Hence, the necessity of interpreting early data. ‘This interpretation is possible through a more general solution of the flow equation, with a variable sandface Tate as inner boundary condition. The rate can be obtained Coorg 188 Socay of Paotmm Enger JANUARY 1988 from a direct measurement or by other means. The frst attempt known to us of direct measurement in a buildup test was presented by Gladfelter er al.! The afterflow rate was obtained by measuring the rise of the liquid level in. the wellbore. The analysis used a semilog plot ofthe 'cor- rected pressure,” whichis the ratio ofthe pressure drop to the sandface rate variation Another method isto estimate the sandface rte from ‘a mathematical model. Van Everdingen? and Hurst? presented an exponential function of time as an approx- Imation ofthe sandface rate. In the currently accepted type-curve analysis itis estimated from the derivative of pressure with respect o time, onthe assumption that the trelbore storage coefficient and the surface rate are con- sant. A comprehensive review of this technique is presented in Ref. 4 Recently, wireline tools capable of simultaneous measurement of sandface rate and pressure (a8 well as fluid temperature and density) were introduced, The first buildup tests recorded with these tools in ol and gas wells almost invariably revealed abnormal situations sich as very long afterflow or backflow and phase segregation in the wellbore. Such observations justifiably may lead ts to question the validity of sandface rate estimated from pressure data alone, We present an analysis of the wellbore storage concept and show briefly how a direct Sandface measurement could improve the type-curve analysis. ‘Our main purpose is to propose a new analysis based con simultaneously measured sandface rate and pressure ‘The technique makes use of a rate-convolved time func. tion similar to the superposition time function of Odeh ‘and Jones.* The wireline aspect of the measurement of- fers the possiblity of analysis in real time. We present this analysis for a buildup test in a homogeneous reservoir. ‘Shorly after the original presentation ofthis work, several authors presented more comprehensive applica. tions of direct sandface rate measurements. Fetkovich and Vienot® introduced their rate-normalized type curves, Kucuk and Ayestaran’ used a deconvolution process to identify the reservoir model, and Stewart et al.® applied the rate-convolved time function in nonhomogeneous reservoirs ‘Variable Rate Equation For an initially static well starting production at a time 120, the dimensionless pressure drawdown is modeled Py by the, superposition theorem of van Everdingen and ‘Hurst,? which can be expressed as Peotto)= | an(Pto-ndr, - o where tp is the dimensionless time, p(tp) is the time derivative of pp(tp), the dimensionless constant rate solution of the flow equation, and qp is the dimensionless sandface rate defined below. ‘The skin effect can be included in Eq. 1 by adding the skin pressure drop, sq, to the drawdown to obtain D Pavtta)= | aotPb\t—TAr+saplta). 2) é ‘The dimensionless sandface rate is the rato of the sand: face rate, gay, to some reference BHF rate, qret. Thi reference flow rate is usually the stabilized production rate, 9, converted to bottomhole conditions: eet =B4, -@ but it can be any other fixed value, provided the dimen- sionless pressure drawdown also is defined with respect to this reference rate. Thus, we define the dimensionless pressure drawdown, p,,p, and the dimensionless flow rate, 4p, respectively, as ath Pavtto)=—— .~P of) cee) and aot), o et where gre is any fixed-reference BHF rate. ‘These definitions emphasize thatthe variable rate equa- tion relates BHP and BHF rates exclusively; therefore, its application does not necessitate the knowledge of the surface production rate. Furthermore, it does not require that the production rate be constant, and it applies to any rate schedule. ‘Type-Curve Analysis ‘The type-curve analysis consists of comparing the test data with a set of precomputed solutions of Eq. 2 for various forms of pp functions and values of skin. This com- parison yields the reservoir mode! (pp fuaction) and parameters (kh, skin, etc.). Practically, the precomputed solutions are presented as a set of dimensionless pressure- time graphs (type curves), and the analysis involves find- ing a match between the graph of pressure-time data with ‘one ofthe available type curves on compatible logarithmic Eq. 2 relates three variables—pressure, sandface rate, and time. The sandface rate is computed from pressure data; thus, Eq, 2 is reduced to two variables, and its solu- tion can be presented on a two-dimensional graph. ‘The computation of the sandface rate is based on the constant wellbore storage concept, which states that the variation of the fluid volume in the wellbore is propor- tional to the variation of the BHP, such that (ag-Barse=C age © where q is the production rate and C is the wellbore storage constant. The dimensionless form of this equa- tion is obtained by dividing both sides by Bg, chosen as reference flow rate. (4p—DAtp=—CpdP wo o where gp is defined by Eqs. 3 and 5 and Cp is the dimensionless wellbore storage constant. By substituting 4p from Eq, 7 into Eq. 2, we obtain an equation involv- ing only two variables, pressure and time, provided that Co is constant. The solution of this equation provides the basis for the generation of type curves. ‘According to Ramey" and McKinley," this is ap- plicable to either (1) a partially liquid-filled wellbore with Constant tubing head pressure, in which case C is pro- portional tothe liquid density, or (2) a wellbore complete- Iy filled by a single-phase fluid, in which case c= 8) ‘we where Vy is the wellbore volume and cy the com- Dressibiliy ofthe fuid in the wellbore, Strictly speak- ‘owing oil wells would be eliminated by the above constraint. “The constant wellbore storage concept involves several assumptions, the first of which is that Eq 6 is valid. Tn fact, the determination ofthe sandface rate fom pressure requires measurements of both BHP and tubing head Assuming thatthe tubing head pressure varia- tion is negligible, Eq. 6is valid but two additional con- ditions are necessary for Cp tobe constant in Eq. 7: (1) the surface rate, q, remains constant during the test and (@) the wellbore storage coefficient, C, of Eq. 8 remains constant, itis obvious that q will remain perfectly constant only in shutin wells (where its zer0), However, the wellbore storage coefficient, C, cannot be constant ina shut-in of Wwell Because dissolved gas separates in the tubing, and the welbore Suid compressibility, cy, i largely governed by the gas Compreseibility. For a fowing wel, q isnot strictly constant as long as the tubing head pressure changes. This is more important during the early tran- sient period when the pressures vary rapidly Furthermore, it takes some time to close (or open) a wellhead valve; during this time, the rate, 4, is varying Widely, We have observed that it may take more than 1 ‘minute to close a wellhead valve. This lapse of time can- not be neglected. “These considerations suggest that the estimation ofthe sandface rate from BHP alone is not sufficiently accurate in large oll producers, especialy at early times. This can be verified from simeltaneously measured sandface rate and pressure data. Using Eq, 6 as a definition of the JOURNAL OF PETROLEUM TECHNOLOGY ‘wellbore storage coefficient, we can compute its instan- taneous value from the: sandface rate and the time derivative of pressure. In field units, Eq. 6 becomes A gg-Ba 24 dpypldt ” * where Cis in RBVpsi [res m? /kPa], gin B/D [m/d], and the derivative of pressure, dp/dr, isin psi/hr {kPa/h]. For a shut-in well, where the production rate, q, is zero, 1 ay ane tt 9) “ 24 dp g/t oo Fig. I shows a iypicl graph of the computed wellbore storage cooficient vs shut-in ime during a buildup tet Tecan be seen that the wellbore storage decreases con- tinuously daring the atrflow and that its variation is rather significa. The sharp variations a very early time ae believed to be cased by the varying surface rate ur- ing the closing ofthe wellhead valve. “The constant wellbore storage concept isnot applicable to very early times. The inaccuracy inthe evaluation of the sandface rates reflected at ltr times since the con- volved integra of Ea, 2 includes all daa from 0. Even at Inter times the wellbore storage isnot perfectly con- stan, as illustrated by Fig. LA direct measurement of the sandface rate shoud improve the accuracy ofthe eary- time transient analysis. c= (a) Application of Direct Sandface Measurements to Buildup Tests ‘Type-Curve Normalization. The instantaneous wellbore storage coefficient of Eq, 9 could be used to normalize the time inthe type-curve interpretation. Instead of plot- ting 4c*, one could plot the function a dr ate alec ‘The interpretation is thereafter the same as for a well with ‘a wellbore storage constant, C1, equal to unity. In buildup test, however, itis obvious from Eq. 9b thatthe ‘computed wellbore storage coefficient becomes zero when the aRerflow becomes too small to be measurable and Eq. 10 cannot be used. This could be overcome by assuming that the wellbore storage remains constant and equal 19 the last computed value forthe remainder ofthe buildup test. Alternatively, an extrapolation based on the graph ‘of Fig, 1 could be used. The same difficulty arises in a drawdown test when the sandface rate approaches the stabilized rate, Bg, and it can be overcome in the same way. = +0) Convolution Analysis. Ifthe well has produced at acon- stant rate, q, from #=0 to time f, the pressure drawdown at a later time, 1+ A¢, can be obtained from Eq. 2 by in- tegrating from ¢=0 to f, then from time 1 to ¢+AL. In dimensionless form, using the constant bottomhole rate, ‘Bq, as teference, Eq. 2 becomes Pwoltp + Atp)=Poltp +Atp)—Pp(Atp) 0 +] Pavtnpb(aty—drtsqp (dtp)... JANUARY 1988 Bt tee Fig. 1—Wollbor storage costiciant computed trom afertiow ‘and derwvative of pressure. Eq, 11 is very general. It describes the pressure change for any variable rate schedule following a constant rate of sufficient duration. It applies in particular to buildup ‘ests with afterflow. The shut-in pressure drop may be ‘expressed as the sum of three dimensionless time func- tions, which are (1) the classical buildup time function, given by + (12) Poltp+Atp)~pplAtp), (2) a rate-convolved time function, given by ap Fate (atp—ndr, é (13) and (3) the skin pressure drop, given by sqp(Atp). aay ‘At the beginning of the shut-in test, the influence of afterflow is preponderant. When the afterflow becomes sufficiently small, the skin pressure drop (Eq. 14) ap- proaches 2er0, and the shut-in pressure is governed by the rate-convoived buildup time function, Mpltp.Atp)=poltp +Atp)—Po(Atp) +f D ‘qo(ripb(tp—7d7, 05) in which the time function of Eq. 13 acts as a correcting term to the classical buildup time function. If afterflow data are available from the beginning of the test, the func- tion Mp can be computed for anytime. Then, if one plots the buildup pressure against ths time function instead of just the classical buildup time function, one should ob- tain a straight line right from the end of the afterflow. ‘Compared with the conventional analysis (e.g., the Homer analysis), early data that were discarded by the “one-and-one-half log cycle" rule! can be interpreted. In practice, itis not necessary to wait for about 30 times the duration of afterflow before reaching the semilog straight line, and the time required for buildup tests theoretically could be shortened by a factor of 30. as Practical Computation. Noting that pp(0)=0, and qp(0)=1, an integration by parts of Eq. 13 gives J? averpp tarp -2idr=pp(Atp) sp +f" abtipptarp—ner - (this is also known as Duhamel’s principle’), and the ‘complete expression forthe dimensionless shut-in pressure drop from Eqs. 11, 15, and 16 is a6) Pos =Ppoltp+Atp)+Ep(Atp)+sqp(4tp), --- (17) with sp Epttp)=| abpp(Atp—rr, .. <8) é where Ep(Afp) is the “sandface rate-convolved time function."* ‘The working equations are developed in the Appendix for the classical case of radial flow in infinite reservoir, ‘assuming the logarithmic approximation is valid. The field units version of Eq. 17 is Pi~Ps =mllog(e+ 01) +E(A1) +3gp(A0)}, ...-19) 20) Slog: ~3.23+0.87s, a near All quantities in Eqs. 19 through 21 are expressed in ‘customary oilfield units (psi, STB/D, md, hours, RB/STB, cp, ft, and psi). If all quantities are expressed in SPE- preferred SI units (kPa, m?/d, um?, hours, res m?/stock- tank m?, Pas, m, and kPa~'), the constants 162.6 and 3.23 should be replaced by 2121 and 5.09, respectively. ‘E(s) can be computed from the numerical expression (Appendix) Ey TC an= + cs with Fg te, 1008 eto)“ Ho Ey “HI + BE -1lont tte -enloaH2)) + [eq —ti-)loatty unt ~ tty tog, 101 + +n 9n— NORE ~tn1) 0.434190 -4n)s a +. @2) where fo isthe shut-in time and qo, 91, 42 --- 4 qn ate the sandface rates measured at times f9, £1, 12 «++ Wass ye Note that At=t, fo and qo = Bg. ‘The expression of Z, in'Eq. 22 was obtained by ap- roximating the afterflow by a succession of straight line segments between the measurement points.'“ Had we ap- proximated the afterflow by steps (i.e., constant rates), it would have been identical with the superposition time function of Odeh and Jones. '5 Modified Horner Graph. Eq. 19 can be written in the condensed form Pi-Pus=m(M +590). (23a) where Me=log(t+A1)+E(A) 224) is the rate-convolved buildup time function defined in dimensionless form by Eq. 15 and gp is short for qp(A0). ‘The three unknown parameters p;, m, and 3 can be ‘computed through a three-dimensional regression on the variables Py, M, and qp, which are obtainable from direct measurement. A more convenient graphical method is presented here. ‘When the afterflow becomes sufficiently small, Eq. 23a reduces to =m M, -(23b) Pi-Prtg a ‘A graph of the shutin pressure, Pz, VS. the rate- convolved buildup time function, M, should become a straight line after the end of the afterflow. The slope of the line, m, is the same as the classical semilog slope, and its extrapolation to an infinite shut-in time gives pj (or p* in a bounded reservoir) However, the afterflow never actually reaches 2210 because flow continues as long as the buildup pressure is different from the reservoir pressure. One should graph (-+5qp) rather than only Mf so that all the data points fall on the straight line. The unknown coefficient, 5, can be obtained from the flow equation before shut-in, 225) Pi-Pg=milog 143), which, combined with Eq, 19, gives t =P es —Pug=milog E80 p= Pas Peg =ilon ECA) +50-ap)). (262) For A that is small compared with 1, we can approx- imate 1/(¢+A1) as 1, and Eq. 26 reduces to dp=m{—B(a)+51—gp)). (266) Choosing Ar such that D(4/)=0, and rearranging this equation, we get Prso—Pog m4 5.5 an JOURNAL OF PETROLEUM TECHNOLOGY TABLE 1—WELL, ROCK AND FLUID PROPERTIES Production rate 9, STB/D 9,200 OWFVE 6, ABsTE 28 Homer preduction time tp, hours 158 Flowing pressure Py, Psa 1946.85 i viscosity », ep 3.26 Total compresatbiltyc,, psi” 0.000011 Poroaty 027 Not pay thickness Af 100 This analysis is comparable to the Homer analysis, ~E(A0) replacing logAr and pro replacing pig. Tt is shown in the Appendix that af sufiiently late shut‘in times = @8) Ate T(AN) arom = log At 40.2172, ar where 41, is the duration of afterflow obtained by ap- proximating the afterflow by a straight line on a graph vs, time, Eq. 28 shows that when Ay is more than about 20 times the duration of afterflow, Ar,, the M function is practically equal to the Horner time function and that E reaches zero shortly after 1 hour (ifthe afterflow is suf- ficiently shorter than 1 hour). We call this graph the “‘modified Horner graph.” Real-Time Analysis. With pressure and flow data being. ‘measured simultaneously by a wireline tool, an analysis, is possible while the data are being recorded. The analysis is based on Eq. 26a after dividing both sides by (1—4): -@9) 9 2, 1-40 1-40 Eq, 29 relates a pressure function and a time function, ‘A graph of these functions should be a straight line. Since they can be calculated from pressure and rate ‘measurements, the graph can be constructed and analy2- ed at any time during the test. The test can be terminated when the analysis is conclusive. We call this graph the “sandface rate-convolution (SFRC) graph.”” ‘hours Fig. 2—Buildup test data—pressure and dimensionlass sand- face rate. ‘The pressure function on the left side of Eq. 29 is the “corrected pressure” of Gladifelter eta. It is interesting to note that ifthe sandface rate i linear with time (“ramp schedule”), the time function in Eq. 29 is, according to Eq. A-II of the Appendix, ~uar) =a ogAr—0.434, 0) ‘This shows that Gladfelter’s approach is essentially cor- rect ifthe aterflow can be approximated by a straight line and hs semilog graph ofthe corrected pressure yields not only the semilog slope but also the skin (more exactly, 5-05). Field Example ‘The following example is an 8-hour buildup test in an oil ‘well. Well, fluid, and rock properties are reported in Table 1. Pressure and sandface rate were recorded simultaneous- ly every 3 seconds with a wireline pressure/Mowmeter combination tool. The data used for the convolution analysis are listed in Table 2. For clarity, only 44 out of the 9,600 recorded points have been retained, which is Fig. 3—Type-curve graph JANUARY 198 Fig. 4—Type-curve graph corrected for variable wellbore ‘storage ur ‘TABLE 2—BUILDUP DATA AND COMPUTED DATA Flowmeter cua) ‘enough to approximate the afterflow adequately by straight segments (Wwe have verified that using all the recorded points or as few as 20 of them does not change the inter- pretation significantly). Fig. 2 shows the pressure and dimensionless sandface rate behavior atthe beginning ofthe test. Approximating the afterflow by a straight line on this graph gives about 0.3 hour as duration of aterflow. According to Eq. 28, the semilog straight line should start after 0.3x20=6 hours’ buildup. 'No satisfactory type-curve match could be found with the usual pressur-time graph of Fig. 3 (note that we have used Agarwal's effective shut-in time for matching with drawdown type curves). The 1/1 slope line is much bet- ter defined on the wellbore storage corrected graph of Fig. 4, and a good match can be obtained with the Cpe* =10° type curve. The match point is Po=2.1 for Ap=1.00 and a dy tnlea=i0s tor (Se ay tte 0h a0) =E.Mt=a9) (a) 2066 30214 Toes 327.02 i237 aaast T2271 33601 T22e 335.80 T2078 348.20 Tts78 355.65, Tas 359.48. Ties 362.96 ii7e2 367.75 rises -373.08 iis g7490 iiss? ara97 Tis 9 g7aa7 Tyee 378.98 rise 3654 i127 38875 Tai01 388.47 Tine geet T1038 300.08, Toss 30089 Toss 302.08, Lo7e1 3042 Togas aoe.t2 Tosa 408.89 Tomos | ain.65, Toles 41560, T0209 © az0.8s Tons | aar77 Toss 431.08 oo = a3a.68 ose = aaast 0147 44248 Oz aas.ae 0264 480.48 0365 45455 ower 458.65 0503 e285 0579 46705 0657 471.26 7s = 47851 0g = 47978 0890 = 483.30 which yields k=419 md, C, =0.95 (dimensionless), and 25. For the corrected graph we used the wellbore storage coefficient shown in Fig. 5 and listed in Table 2. The coef- ficient is computed from sandface rate and pressure up 10 0.4 hour (when the afterflow becomes too small to be measured accurately), then extrapolated as inversely pro- portional to pressure (in an attempt to account forthe com- pressibility ofthe free gas in the tubing). This coefficient also is used for extrapolation of the sandface rate after (0.4 hour. The sandface rate used in the computations, 4 is presented in Fig. 7 and listed in Table 2. Fig. 6 shows the modified Horner graph compared with the conventional one. The semilog slope, m, was first determined from the points after 1 hour, when the afterflow becomes negligible. The pressure py.ao was ob- tained by interpolation of the data in Table 2, and 5 com- puted from Eq, 27. Then, ¥ was adjusted $0 that all the points fall on a straight line. The values reported on Fig. 6 (p*=2,405, m=$7.05, and py-o=2,279) were points, The results are k JOURNAL OF PETROLEUM TECHNOLOGY Bt owes Fig. 5—Wellore storage (WBS) costficint. Fig. 7 shows the SFRC graph compared with the MDH graph. Note that we have used Agarwal's effective shut- in time for the MDH graph, and, likewise, we have not approximated 1/+dr as unity in Eq, 26. The effective time function of the graph is thar £,=E(A1) +g: ep After 0.5 hours (computed from the E axis, Ar is not quite equivalent during the afterflow as shown by Eq. 30), a well defined straight line is obtained. The values reported on the graph were obtained through linear regres- sion, and the results are k=403.1 md and s=1.55. Discussion In the example, the results from the convolution analysis and the type-curve analysis are somewhat different. In fact, the type-curve analysis is not applicable because the reservoir is not homogeneous. This is revealed by the SFRC graph of Fig. 7, which distincly exhibits wo straight lines. This heterogeneity is unnoticeable on the other graphs. Pure radial flow starts only after 0.5 hour, ‘and earlier points cannot match with the same type curve. Figs, 6 and 7 clearly show thatthe convolution analysis, yields the same straight line as the conventional Horner ‘or MDH analysis, and thatthe semilog straight line starts ‘more than one cycle ahead compared with the conven- tional graphs. This allows a much better definition ofthe straight line, especially in this case, where only the last hours of the test are availabe for semilog analysis. This ‘g00d definition also brings more confidence into the cor- rectness of the analysis; thus, the match of Fig. 4 is er- ‘The SFRC analysis is preferable because the modified Homer graph is quite sensitive to the choice of 5. It also reveals early-time heterogeneities that are likely to be overlooked by the other methods, leading to erroneous, results although a seemingly good match can be obtained. ‘The convolution analysis is not limited to buildup tests in homogeneous reservoirs. The minimum requirements for this type of analysis are as follows. 1, Stabilized rate may be followed by any variable rate schedule. This includes drawdown tests (stabilized rate ‘equals zero). JANUARY 1985 ER satan nana, Fig. 6-Moditied Horner graph compared with Homer graph. Fig. 7—Sandtace rate convolution (SFRO) graph compared wath Miler-Dyes- Hutchinson (MOH) grapn, 2. Simultaneous measurements must be taken of sand {face pressure and rate, Only relative accuracy is needed in the rate measurement since its the dimensionless sand- face rate that is involved in the analysis. Currently available spinner-type flowmeters are well adapted for this since they have a lineageresponse over a large range. 3. Knowledge of the reservoir model (Le., pp func~ tion) is needed. Ifthe logarithmic approximation is valid, the analysis is as explained previously. Recent developments of the method allow the determination of the reservoir model as well as parameters.” ‘A wireline measurement provides several advantages. ‘The surface monitoring allows immediate detection of a ‘measurement failure. The test can be interrupted and restarted after correcting the problem. Unwanted com- plications such as phase segregation can be recognized atan early stage and the test redesigned accordingly. The simple observation of the afterflow allows prediction of the optimal duration of the test With a surface recording, the data storage capacity is, practically unlimited; thus, the measurement can be made at very short time intervals to ensure that enough data are available for a close approximation of the afterflow by straight segments (1-second intervals or les are possible). In addition, a real-time analysis is possible during the test. The analysis involves the computation of the sand- 9 face rate-convolved time function. This computation can be made with a programmable pocket calculator. However, logging equipment provides this computation ‘and displays all the graphs used in the analysis without ‘interrupting the measurement, ‘Conclusions wi ae nen fe el i Saas crm te mnt Le att as ap tegin n oB sicker ero a cece epi ntti “hs ome een ctw a oy eel wet ent) PE ct ee tee re aaa ee em ie SSeS SR a pen pin a necessary computation and graphing capabilities. A pro- sree cme eg ri the analysis. hig wen le sey wate eee ate Nomenclature B = formation volume factor, RB/STB {res m?/stock-tank m3} ey = wellbore fluid compressibility, psi~ (kPa) cy = total system compressibility, psi! tkpa-") C = wellbore storage constant, RB/psi [m*/kPal ‘unity wellbore storage constant, ‘dimensionless C(2) = variable wellbore storage coefficient, RB/psi {m?/kPa] = pay thickness, f (m} effective formation permeability, md conventional semilog slope, psi/log cycle [kPa/log cycle) (M = rate-convolved buildup time function i = initial pressure, psi (KPa) Pwo = wellbore pressure drop, dimensionless Pug = bottomhole flowing pressure, psi [kPa] Pu ™ dottomhole shut-in pressure, psi [kPa] Po(p) = constant rate solution of the flow equation Po, = shut-in pressure drop, dimensionless p* = pressure extrapolated to infinite shut-in time, psi (kPa) Pine = pressure on straight-line portion of Homer semilog plot 1 hour after beginning of transient test, psi (KPa) e-o = pressure at time when E=0, psi [kPa] = pressure change, psi {kPa} C h k 4 = production rate, STB/D [stock-tank m3/d) ret = reference bottomhole flow rate, RB/D [res m?/4] qq sandface rate, RB/D {res m?/d] 4p = sandface rate, dimensionless GDeag sandface rate when E=0, dimensionless Ta = wellbore radius, ft {m] ‘5 = skin factor, dimensionless F = skin coefficient defined by Eq. 21 1 = time, hour tp = time, dimensionless time from end of stabilized rate, hour duration of afterflow, hour At, = Agarwal’s effective shut-in time, hour Ate = time from end of stabilized rate, ‘compensated for wellbore storage variations, hr-psi/RB [h-kPa/res m°] V,, = wellbore volume, bbl {m?] i = bottomhole fluid viscosity, cp [Pas] = sandface rate-convolved time function, dimensionless LZ, = effective sandface rate-convolved time function, corrected for production time, dimensionless 5, = numerical approximation of the sandface rate-convolved time function, RB [m?] 1 = integration variable ¢ = formation porosity, fraction Subscripts and Superscripts D = dimensionless i= ith data point rn = last data point (corresponds to shut-in time 40) 0 = first data point, at start of test (corresponds to Armd) derivative with respect to time Acknowledgment ‘We thank S.P. Noik of Schlumberger for providing the first recorded field data and suggesting investigation of the afterflow process. References 1, Glade, RE, Tracy, G.W., and Wise, LE. “Sloting Wels ‘Which Wil Respond to Preducion-Smulation Treatment," Dil ‘and Prod, Proc, API, Dallas (1985) 117-19. 2. van Everdingen, A..!*“The Skin Effect and Is Inueace on he Productive Cepacty of « Wel” Tra, AIME (1953) 198, 171-76. 3, Hurt, We: “Eaablshment ofthe Skin Effect and Re Impediment to Fuld Flow Into a Well Bore,” Pet Eng. (Oct. 1953) BE-BIG. 4. Geingarea, A.C. etal: “A Comparison Between Different Skin and: Wellore Storage Type-Curves for Early-Time Transient ‘Analysis paper SPE 8208 presented atthe 1979 SPE Annual ‘Technlcal Conference and Exhibition, Las Vegas, Sept. 23-26. 5, Od A,S. and Jones, LG. “Pressure Drawdown Analysis, Varisle Rate Case," Pet, Tech (Aug. 1968) 960-58; Trans, AIME, 234. 6, Pethvich, MJ. and Vient, M.E.: "Rte Normalization of Buty Pressure By Using Afteriow Data," J. Pe. Tech (De. 1984) mis JOURNAL OF PETROLEUM TECHNOLOGY 7. Kucuk, F. and Ayestaran, L.: “Analysis of Simltanously ‘Measured Pressure and Saniface Flow Rate in Transient Well ‘Tesing,” paper SPE 1217 presented atthe 1983 SPE Annual ‘echnical Conference and Exibiton, San Francisco, Ot schedule for pubicaion, J. Pet. Tach. (Feb. 1985) 8 Stwar, G., Witmann, MJ, and. Meunier, D.: “ARerfiow ‘Measurement ad Deconvohiion in Wel Tet Anais,” paper SPE 12174 presented at te 1983 SPE Annual Technical Conference and Edubidon, Sun Francis, Oc. 58. 9. van Everiagen, ALP. and Hurst, W.: “The Applicaton of the [Laplace Transformation o Flow Problems in Reservoirs,” Trans, ‘AIME (1949) 186, 305-24, 10, Ramey, HJ. te | Pessre Bull-Up and Drawdown of Gas Wel (Feb, 1968) 23-38, Trans, AIME, 234. 1, Mekinley, RM: "Wellore Transmissblty From ARerfiow: Dominated Presure Buleup Data," J. Pe. Tech Guly 1971) 863-72: Trane, AIME, 281. 12, Matiows, CS. and Russel, DG: Presure Bland Flow Tests: a Wel, Monograph Seis. SPE, Richarson (1967) 1 16nd 11S. 13, Ramey, HJ. In: "racial Use of Moder Well Test Analyse," paper SPE 5878 presented atthe 1976 SPE California Regional Meeing, Long Beach, Api 7-9, 14, Mebdwards,D.G.:""Maltvel Variable Rate Test Analysis." Soe. Pat Eng. J. (Aug. 1981) $88.46 1S. Oaeh, A'S. and Jones, LG. “Two-Rate Pow Test, Variable Rate CCase~Applleation to Gat-Lif and Pumping Wells.” J Pet. Tech. (Gan, 1974) 93-99; Trans. AIME, 257. APPENDIX Numerical Computation of the Sandface Rate-Convolved Time Function Ifthe pp function can be approximated by the logarithm of time, Eq, 17 of the-text becomes (in field units) Pies =millog(+A1+ D(A) +sqn(42)}, (A-1) with 2(aH= | gb(rlogtar—n)r 2 (A2) im and ¥ in Eq. A-1 have the classical expressions: 162.689 m -(A3) oe c and k Flog (a4) ed 4p is the dimensionless sandface rate defined by Eq. $ of the text: (an ap(an~ 2 AS) Ba [At time 10 (shut-in time), ao0=1. « as JANUARY 1985 Ramp Schedule. Assuming thatthe sandface rate is linear with time, its derivative is constant and equal to 2) where gp is short for 4 p(Ai). Combined with Eq. A-7, Eq. A-2 becomes at ran=4 ogtar-ne as é ‘The integral of Eq. A-8 can be obtained from the integral Of the natural logarithm: J In(ar—2)dr=(ar—2)[1 ~In(ae—7)]. (A-9) Expressed in base 10 logarithm, Eg. A9 is J logtar—ayr=(ar— 710.434 =log(ar—7) (A10) and Eq, A-8 takes the simple form (an) =(ap — 1Mloga-0.434), ay ‘Variable Rate Schedule ‘The sandface rate is measured at times (a-12) fos the fa ee thoes tas and the corresponding measured data are Qn ees (Ae13) Go. 1 G2 ++ where fo is the time of shut-in; thus, g9=Bq is the reference flow rate. For the numerical computation, we define the sum Ea (ata) Peloton with Zn Ban= 4% ‘The integration of Eq. A-14 can be made as follows. fiefs Pala Pi sell af or, written in a different form, Zn EAE, +. HAE A. HAE, (AIS) 181 If the rate can be approximated by a succession of straight Line segments between the measurement points, the derivative is constant in each of those intervals; thus, ‘each elementary AE, is given by AE ;=4/' “log(ar—nar tt with ain 1, = SN, 1 ogl ty =F) Cg tog ty =1))-0.434( gi Gi-1), --(A18) ‘and the numerical approximation of Eq. A-1S is tent Eye DAE, +( aq Gn-iO8 th +0.434(q0~4n). -(A-19) [Note that we have made explicit the last term of the sum, AE, because of its particular form. Late Shut-In Times. Assuming a ramp schedule, the afterflow will cease at a shut-in time Ar,. Eq. A-1] ap- plies only during the afterflow period. For later times, Eq. A-19 should be used. The straight line flow schedule can be represented (as in Eqs. A-12 and A-13) by Ar=0, tg, At and gp =1, 0, 0, and application of Eq. A-19 gives (Ai) = —[Ar logar—(ar—Ar,)log(Ar—Ar4)) -(A20) 12. which can be rearranged as 2(a0)= ~Iog ae (“-1)iog ( ) a, 40434, (2D Loving Ye, we get ae a 1) mg-oo.4(-)(-« woas(-ir bs. i Ife is small (ie., Ar is large compared with Ar), Eq, ‘A-21 can be approximated as staying avs cay ‘isc ha te ator coca ia lea he Air attic soared ene ores opsbe ae iw intake sonrbagu aap src Sritcnci thane moceenr beginee ae Soc ais hag toe tou tac Tt Sona vi er fo sapien beara hao i a he eae esperar aekiore Spin ene ne ghee ny do Re ieererd otktpeere measurement, the start of the semilog straight line can Spanair Searcy SI Metric Conversion Factors bbl x 1.589873 E-Ol = m? ep x Loe E-03 = Pas Rx 308" -E-Ol =m psi x 6.894757 E+00 = kPa psi"! x 6.896757 E400 = kPa“! ‘ogra manera a he acy of Param Eager tee Oa. 28 {5B how estntorpouon Ag 1180s mance ei A Bt awt Pager SPE thse tn pete sth te Ex OX Saw name, Baa, aca 1 JOURNAL OF PETROLEUM TECHNOLOGY

You might also like