Sie sind auf Seite 1von 236
Contributions in Petroleum Geology & Engineering (ff) 8 Contributions in Petroleum Geology and Engineering Volume 8 Well Test Analysis Copyright © 1001 by Gulf Publishing Company, Houston, "Tenas. All rights reserved. Printed in the United States of “America. This book, or parts thereof, may not be "reproduced in any form without permission ‘of the publisher. Printed on Acid-Pree Paper (2) Library of Congress Cataloging-in-Publication Data | ‘Sabet, Mohamed A. Wel test analysis M.A. Sabet. p. em, (Contributions in petroleur geology & cngincering: 8) Includes bibliographical references and index. ISBN O-872O1 584 (hardcover, printed on acid-free apes) | 1. Ol! welling. T. Tie. TE Seles, "TNSTLSISS| 1901 22/3982" 0287—8e20 -3968 cP ISBN 0-87201-584X ISBN 0-87201-066-X (Series) 1 3, Contents Preface . cece Vii Fundamentals of Drawdown Testing ... Heuristic Derivation of the Drawdown Equation, 1; Park Jones ‘Method, 16; The Pressure Derivative Method, 20; The Skin Effect, 23; ‘The Radial Flow Equation, 27; Detection of a Linear Boundary, 32; Computer Programs to Evaluate the Exponential Integral, 38; Radius cof Investigation, 40; The Steady State Case, 47; Conclusions, 48, Fundamentals of Pressure Buildup Testing Principle of Superposition, 53; Derivation of Horner's Equation, 54; Derivation of MDH Equation, 60; Determination of P, 70; Detec- tion of a Linear Boundary; 98; Conclusions, 103, Fundamentals of Multirate Flow Testing ..... = 105 ‘Two-Rate Flow Testing, 105; Multirate Flow Testing, 112; Modified ‘Two-Rate Flow Test Analysis, 124; Approximation of Flow by Straight Lines, 129; Superposition, Convolution, and Deconvolu- tion, 140. Wellbore Effects... . - 146 Wellbore Storage Eifects, 146; Type-Curve Analysis, 153; Real ‘Time Detection of End of Wellbore Storage, 157; Gringarten’s ‘Type-Curves, 160; Bourdet’s Type-Curves (Pressure Derivative Method), 168, Hyydraulieally Froctured Wells, 175; Partial Penetra- tion and Partial Perforation, 182; Other Wellbore Effects, 183; Con- clusions, 184. 5. Gas Well Testing . 186 ‘The Real Gas Potential, 186; Adaptation of Liguid Well sting Equations to Gas Well Testing, 190; Wellbore Storage Effects, 199; Deliverability, 195; Real Gas Pseudo Time, 207; Wells Producing from Low Permeability Reservoirs, 215. 6. Testing of Naturally Fractured Reservoirs - 221 Origin of Fractures, 222, A Conceptual Overview of the Models, 296; Model by Warren and Root, 298; Model by Streltsova, 233; Model by Kazemi, 237; Model by deSwaan, 238; Model by Naju- rieta, 242; Model by Adams et al., 253; Type-Curve Analysis, 260; ‘The Pressure Derivative Method, 264; Conclusions, 272. 7. Testing of Layered Reservoirs .....+4.+sessseseeeeeee+ 274 Commingled Reservoirs, 277; Crossflow Reservoirs, 280; Composite. Reservoirs, 281; Conclusions, 289, 8. Interference and Pulse ‘Testing .... aol Interpretation of Interference Test Data, 292; Interpretation of Pulse ‘Test Data, 903; Effects of Skin and Wellbore Storage on Interference Data, 317; Effects of Wellbore Storage on Pulse Test Data, 319; Pulse Test Design, 821; Interpretation of Pulse ‘Tests by PC-BOAST, 323 9, Drill Stem, Closed Chamber, and Slug Testing «+. ..+.. 330 Drill Stem “esting (DST), 331; Closed Chamber Testing (CCT), 348; Slug Testing (ST), 350; Closed Chamber Slug Testing, 961 10. Injection Well Testing . Basic Concepts and Definitions, ls, 377; The Empirical Approach to the Falloff Test Interpretation Problem, 384; Conclusions, 388. LL. A Bridge Between Theory and Practice . . = 390 ‘The Dimensionless Form of the Radial Flow Equation, 391; Essene tials of Laplace ‘Ikansform, 392; The Inverse Laplace Transform, 398; Constant Rate Solution of the Radial Flow Equation in a Bounded, Cireular Reservoir, 400; Wellbore Storage and Skin Ef- fect, 410; Multilayered, Commingled Reservoirs, 419: Naturally Fractured Reservolis, 424; Slug Testing, 431; Testing of Hydrauli- cally Fractured Wells: A Case of Finite Conductivity Vertical Frac- ‘ure, 437; Testing of Hydraulically Fractured Wells: A Case of Inf- nite’ Conductivity Vertical Fracture, 445; Deconvolution, 447; Conclusion, 451 = 454 Preface ‘This book is an outgrowth of a five-day course which I have presented repeatedly in the United States, Europe, the Middle East, Southeast Asia, and Australia, The book is aimed at practicing engineers and geologists who ‘want to understand the basics, limitations, and applications of well testing _methods to be able to practice wel test analysis with proficiency. Therefore, this book is not a handbook of well tet analysis, nor is ita theoretical over- ‘view; it isa book that explains theories, emphasizes simplified mathematics, and presents numerous examples to demonstrate how the theories may be applied to real-world problems. ‘The material presented in this book is applicable in several disciplines: po- ‘troleum engineering and geology; groundwater hydrology; and toxie waste Aisposal in deep wells, which is a branch of environmental engineering and geology. Each of these disciplines uses a different system of units, and the system of units used by one diseipline in the United States is often different from that used by the same discipline in Europe and in other parts of the ‘world. Thus, any attempt to satisfy the needs ofall these disciplines would hhave resulted in utter confusion and chaos. For this reason I had to choose ‘only one system of units for the entire book and I chose the U.S. oilfield sys- tem, not because I think it is a better system, but because I am more familiar with that system. Readers who use other systems of units should begin by converting their measurements of rate and pressure, respectively, to barrels, per day and pounds per square inch, After gaining sufficient confidence in the subject, they will then be able to adapt the formulae in this book to their particular system of units In writing the book, Thave addressed myself to those readers who in their undergraduate curricula have had two years of mathematics, and have since forgotten their mathematics and are now rusty. Ihave also assumed that, except for the availability of a very limited field library, the readers have no ready access to extensive libraries, either because time is limited, or because well-stocked libraries are not within reach, With these two constraints in mind, Thave attempted to make the mathematical presentation as clear and simple as possible, and I have tried to treat the different topics as completely as possible. ‘The first ten chapters should prepare the reader to practice well test anal- ysis with a good understanding of the basic theory. Chapter 11 prepares eaders to expand their knowledge by reading research articles from the lit- ‘erature. But even in Chapter 11 I have attempted to maintain the presenta- tion of the theory in a simplified form. “The book contains a few computer programs. The purpose of these pro- grams is threefold: to help in demonstrating the theory; to show that seem ‘nely difficul tes results ean easily be programmed on the computer; and to ‘encourage the readers to write their own programs. Thus, these programs fare by no means efficient, because they are not intended for everyday use. M.A. Sabet Littleton, Colorado ertrtutons Perot Gesoay &Engering ft 1 Fundamentals of Drawdown Testing ‘The fundamental objectives of drawdown testing are to obtain the aver- age permeability, k, of the reservoir rock within the drainage area of the tvell, and to assess the degree of damage or stimulation induced inthe vicin- ity of the wellbore through drilling and completion practices. Other objec- tives are to determine the pore volume, Vp, and to detect reservoir inhomo- sgeneities within the drainage area of the well. This chapter is aimed at introducing the reader to all of these objectives. ‘There isa difference, however, between objectives and realizations. Al- though the objective is to determine the permeability, k, the test actually yields the transmissibility, T = k hig. Tb obtain k, we must assume that we mow the pay thickness, h, and the fluid viscosity, z. Likewise, our objective is to obtain Vp, but the test yields Vp C,, and we have to assume that we know the total compressibility, Cy, in order to determine V,. HEURISTIC DERIVATION OF THE DRAWDOWN EQUATION Let us assume that we are dealing with a reservoir that can be divided into large, horizontal segments such that each segment is uniform with re- spect to porosity, permeability, thickness, water saturation, and rock com- pressibility. Furthermore, each segment is saturated with water and black ‘oll of constant viecosity and compressibility, and only the oil is mobile. Each segment is drained by one or more wells, and each well penetrates and is ‘open through the entire pay section of the reservoir. If all these conditions are met, then the flow to each well will be radial, ic, the oil moves toward the well equally from all dizeetions. 1 2 Well Tist Analysis ‘We will focus our attention on one well which drains a circular, 80-2ere avea of the reservoir This means that all the oil thatthe well produces comes ‘nly from the 80-acre drainage area. In other words, the outer boundary of the drainage isa closed, no-flow boundary. This does not mean that the boundary is formed by any physical barriers such as sealing faults or perme- ability pinchouts. "To understand how such a boundary could form without any physical barriers, consider the uniform segment of the reservoir of Figure 1-1 in ‘which a number of wells are drilled on a grid pattern, If all the wells prodhice atthe same rate then to determine the drainage area of any well in the pattern, one would simply divide the distances between the wells in half Of couse, if we shut-in one well then the neighboring wells would ‘extend their drainage areas to include the drainage area ofthe shut-in well, Tes therefore evident that the size and shape ofthe drainage boundary of a swell producing from a homogeneous reservoir depends on the relative post tons and production rates of the neighboring wells. However, as long as no Changes in the production rates ovcur, the drainage boundary ofeach well in Figure 1-1. Well pattem in a homogeneous reservoir segment, showing a drain- ‘age area with a no-fiow boundary. Fundamentals of Drawdown Testing 3 the pattern is a no-flow boundary, i, no flow takes place across the boundary. Later we will present methods for estimating the size of the drainage area of a well, but without additional data its exact shape cannot bbe determined by well testing alone. “To perform a drawdown test on our well, we fist close the well until the pressure stabilizes throughout its drainage area, We wil eall the stl Then Tower a pressure recorder to a level slightly ‘above the perforations, and open the well to produce at a strate. The presure recorder records the pressure P, until the instant the wells opened for production. Thereafter, it records the flowing bottom-hole pressure, Puy versis Hime. From the recorded Pai itis posible to plot AP = P, ~ Pyrver~ sus time, as shown io Figure 1.2 20: eesas 80: f 90. APHPI-Pat, psi 110. 120 130. 140: 150. ‘ 30 v0 the Figure 1-2. Rectangular plot ofthe frst 100 hrs of dravidown data from Table 1-1 \ “Let us suppose that with a flow rate of 100 B/D we obtain the data shown {in Table 1-1, The first 100 hours of these data are plotted on Figure 1-2. The Figure is characterized by throe segments: an almost vertical straight line be- ‘ween 0 and 0.5 hour; a curve between 0.5 and 20 or 30 hrs; and a sloping, straight line thereafter. Infact, a quick check on the data of Table 1-1 shows that beyond 40 lus AP is a linear function of tise. For example, between 40 4 Wall Tist Analysis Table 1-1 Pressure Drawdown Data (= 100 BID, ry = 1,059 fy = 0.89 M1) t =P t aP t ‘a ow o_o) 0) os 001 sr2t 10 9263 ~—~—«650~~«BST oo =o 18s m0 | 13022, oe te 004 e778 © 4010280160005 bos = 714985 tO5B1 «2800781 04 7477100 100804000207 94 016 7035 18011347 650028867 025, 8175 25011688 800.0——805.10 0.80 sose 400 1205014000 58.68 and 65 hrs, the slope is equal to 0.2429 psi/hr, which isthe same as the slope Detween 40 and 1,000 hrs. Figure 1-3 isa plot of the early data of Table 1-1 with an expanded time scale, Itis evident that these data do not fall on a straight line; they form a ‘curve, Therefore, based on our observations of Figures 1-2 and 1-3, we must conchide that Figure 1-2 is characterized by only two segments: a curve be- ‘tween O and 40 hrs; and a straight line thereafter. 9 oc = t+—] | it | Se Boo 062 00s 006, ojo 012 04 O16 Producing Time tp (hrs) Figure 1-3. Rectangular plot of eatly drawdown data from Table 1-1 with an ex panded scale. Fundamentals of Drawdown Testing 5 Figure 1-4 is the same as Figure 1.2 except that the time seale has been changed. Figure 1-4 suggests that the curve ends at about 15 hrs, not at 40 hrs as Figure 1-2 shows. However, if we check the data in Table 1-1 between 10 and 16 hrs we get a linear slope of 0.597 pai/hr, and between 16 and 25 his the linear slope is 0.978 psi/hr, Both slopes are quite different from the Linear slope of 0.2429 psi/hr which was determined beyond 40 hrs. We must therefore conclude that in spite of appearances, the curvature begins at 0 and continues to 40 hrs. 2 AP=(Pi-Pul) psi 0 130 eo 0 es ed coe cata) Producing Time tp (hrs) Figure 1-4. Same plot as Figure 1-2, but with a different scale. ‘We summarize our observations as follows: 1. A plot on rectangular graph paper of the pressure drop AP versus time ‘as observed in a well producing at a constant rate from a bounded rainage area consists of a curve followed by a straight line. The curve begins at t = 0. This curve represents what is known as the unsteady state, the early transient state, or the infinite acting state, All three ex- pressions are used interchangeably. The straight line represents what is known as the semi-steady state, the pseudosteady state, or the quasi- steady state 2. The beginning of the semi-steady state cannot be visually discerned from the plot because the scale of the plot may influence our judg- ‘ment. Since real field data are usually noisy (je., contain spurious 6 Well ast Analysis ‘pressures caused by temperature variations, solar lunar gravitational fides, closing and opening of wells in the field, and presence of 8). ‘sual determination ofthe beginning ofthe straight line on the rectan- iqular plot (Figure 1-2) could be wrong. Figure 1-5 isa plot of the same data given in Table 1-1 versus the log of time. Comparison between Figures 1-2 and 1-5 shows that the early tran- sient state is represented on Figure 1-5 by a straight line, and the semi-steady state is represented by a curve that Is concave downward, It is clear from Figure 5 that the early transient state ends and the semt-steady state begins at about 40 hrs. Thus, the beginning ofthe straight line on the rectangular ‘plot could be better determined on the semi-log plot, but this is not always the case when one is dealing with actual field data. Later on we will learn the Park Jones method, which could be helpful in determining the beginning of the semi-steady state, 8 tt 2 2 r fe a 8 ee ie TOP Producing Time tp (hrs) Figure 1-5. Semiog plot of drawdown data from Table (Our observations can be stated in a different way. During the early tran sient state, AP is a linear function of log t. Thus, AP = mlog t + Constant where m = the absolute value ofthe slope ofthe straight line on the semi-log plot (Figure 1-5). Fundamentals of Drawdown Tasting 7 For brevity, we will refer to the absolute value ofthe slope as the lope. Tt should be clear that the slope, m, describes the rate of decline ofthe flowit om-hole pressure, Pyts du eri Vale OF the constant, we choose t = 1 hr, then log Ie follows that the constant is equal to APg. (or AP). Then the equa tion could be writen as AP = mlogt + APiie gay ‘During the semi-steady state, AP is a linear funetion of time. Thus, AP = mnpst + Cue (2) where ms, = slope ofthe straight line on the rectangular plot (Figure 1-2), “This lope describes the rate of decline ofthe flowing bottom-hole pressure, Pais during the psendosteady state (ps) ya = poeudosteady state intercept Te is evident that we cannot write Cpa = APgr 1» since at t= 0, AP = 0. Therefore, to determine Cye we must choose t > 40 hrs and substitute in Equation 1-2, or extrapolate the straight line backwards and read the inter- cept on the ordinate, TReturning to Equation 1-1, the unsteady state equation, we wish to in- «quire about the nature ofthe slope m. For this purpose, we reason as fol- lows: autihLyBy cong ey ewe wold ee i Dots ole pressure Py to decline at a rate which is higher than the rate ob- _ ayn WH served at smaller values of g, ie. the slope, m, of the straight line Fi -5 must be proportional to q. However, since q is measured in igure Ny stock tank barrels (STB), we must multiply q by the formation volume factor B, t0 obtain the flow rate in reservoir barrels (RB). be greater in the case of a low permeability reservoir than in the ease 2. The slope, m, must be related to the permeability, k, of the reservoir rock. Its evident that for the same q, the rate of pressure decline must the net reservoir thickness h. For Constant q, the larger the value of h, the lower the value of m. ‘The slope, m, must also be proportional to the oil viscosity, . The sigher the viscosity, the higher the rate of pressure drop, m. Seabee ohne aire po Biome se nati 8 Well Test Analysis We conclude, therefore, that: m oe But kh Ca Bu : kb oe where ©, = constant ‘To determine the value of the constant, Cy, in Equation 1-3, we need to ‘measure m from Figure 1-5. We also need to know some rock and fluid properties, Let us assume that on the basis of our knowledge from well logs, ‘core analyses, and PVT data, we obtained the following information: 50 md h=s0ft 0.8 ep 1.85 RBISTB ‘To determine m from Figure 1-5, we choose data points at the beginning and at the end of any logarithmic cycle. For example, from Figure 1-5, we could ‘choose the data points at t = 0.1 and 1.0 hr. The slope, m, is given by: ime TAT ~ 92.33 ~ fog 0.1 log LO = 17.56 psifeycle Had we chosen another cycle, we would have obtained the same result. For example, choosing the data at t = 1.0 and 10 hrs, 92.93 ~ 109.89 Tog I= log 10 = 17.56 paileycle We could, of course, choose other points that do not fall on the ends of a cycle, The computation of m isthe same as before. As an example, fort = 0.2 and 0.7 br, we get m= 80-90 Tog0.2—Tog 0.7 = 18.38 paileyele Fundamentals of Drawdown Testing 9 ‘which is essentially the same value obtained earlier, given the fact that the ‘presures were read off the plot of Figure 1-5. It is important to develop the habit of reading the slope off the straight line on the semi-log plot, not from the pressure-time data table. ‘Now we are ready to determine the value of the constant, Cy, in Equation 1a: mkh = Bee aBoe 17.56 x 50 x 20 . i % 1.35 x 0.8 \ Ol = 162.6 » cordingly, m= 1028.48 , : Accordingly, ce 4) and Equation 1-1 could be writen as: 62.6 4 Bo = Y TES AB Hog t+ Ay (5) ‘Tofind the dimensions of m, we substitute the dimensions of q, (L°/T); 1, (MILT); k, (L4}; and h, (L), The dimensions of m are determined to be that of pressure (M/L T°), which means that log tis dimensionless. This is true, since In t = fdtit, and In t = 2.303 log t. Having found the expression for m in Equation 1-1, we now wish to find the expression for mg, in Equation 1-2, Before we do that, however, we must develop a clear understanding of the meaning of the early transient (un- steady or infinite acting) state and the semi-steady (pseudosteady or quasi steady) state, For this purpose it is necessary to refer to Figures 1-6a—1-6e. ‘When the well is first put on produetion at a constant rate q, the pressure in the wellbore drops and a pressure disturbance spreads through the drain- age area of the well. The rate at which the pressure disturbance spreads does not depend on the distance to the drainage boundary. That is why the tran: sient state is also called the infinite acting state. During the infinite acting state the rate of decline of P.; and the manner by which the pressure distur: bance spreads through the reservoir are determined by the reservoir ane £luid characteristics such as the porosity, @, viscosity, 4, permeability k, ané total compressibility, Cz, Note that these characteristics do not include the flow rate, q. In other words, the rate of spread of the pressure disturbance i independent of q. This is very much the same as the spread of sound wave through a medium. Sound travels in air at the same speed whether one whis pers or shouts, However, our ability to hear depends on the amplitude of the 10 Wall Tat Analysts Sy ST Pate Figure 1-68, A well in a Figure t-6b, Well produced ata bounded, circtiar drainage area, constant rate, q. Pressure tran- ‘The well ie closed. Piisanimagt —_sients at t= {, and at t= te aro nary surface. fot parallel Infinite acting state. mn owt Figure 1-60, Beginning of som Figure 16d. The semi-steady Steady stato, Traneionts have f- slate. Transients at ty, t, and ts nally reached drainage bound. are paral. ary. Figure 4-00. The walla produo Ing in a semisteady state. Note that AP = P, — Pye = Pe ~ Pw. Figure 1-6. Schematic diagrams showing unsteady and semi-steady states, Fundamentals of Drawdown Testing 11 sound wave. Likewise, our ability to detect the pressure drop, Ap, depends ‘on q. The larger q is, the larger the pressure drop and the easier itis to de- tect. "There is a strong analogy between the manner by which sound travels in sir and the way a pressure disturbance travels in the reservoir, If one stands ‘at some distance away from a vertical cliff and holler, then until the sound ‘wave reaches the cliff it travels in a manner independent of the distance to the cliff In other wordk, the sound wave travels asi the distance to the cliff ‘wore infinite, That is exactly what happens in the reservoir during the infi- nite acting state, The pressure disturbance travels as if the drainage area ‘were infinite. “The arrival of the pressure disturbance at the drainage boundary marks the end of the early transient state and the beginning of the semi-teady state, Intuitively, one should not expect the change from early transient to semisteady state to occur instantaneously. It actually takes a few seconds in the case of a circular drainage area, and it takes longer for other shapes. ‘This short period of time which separates the early transient state from the semi-steady state is called the late transient state. Due to its complenity and Short duration, the late transient states not used in practical well test analy- ‘Note that during the unsteady state the cumulative production of the well {sa very small fraction of the oil present in its drainage area. Therefore, if ‘we close the well during the early transient state, we should expect the pres- sure in the well to stabilize essentially at the pressure P; measured before ‘opening the well to production. On the other hand, after the semisteady state has been reached, the entire drainage area of the well is gradually temptied. Accordingly, if the well s elosed sometime after the semi-steady State is reached, the pressure in the wellbore should stabilize at an average pressure, & which is lower than the initial pressure, Py. Figure 1-6e shows that during the semi-steady state the pressure transients affect the entire drainage area of the well. Then, if Par declines by one psi, the pressure declines by one psi throughout the drainage area of the well, which means that the average pressure inthe drainage area, P,also declines by one psi. Accordingly, during the semi-steady state the following condition ‘must hold: P ~ Pyg = constant Simple application of material balance during the semi-steady state yields: aBt -p Dt Vp PP) (16) 12 Wall Test Analysis where t= total time since the well was put on produetion, hrs G. total compresbility, Vpst = C8, +, 8, +C, Cy, Go, Gy = oll, water and rock compressibility, respectively Sy Sy = oil and water saturation, respectively V, = reservoir pore volume, RB For a cylindrical drainage area, Vp = wh 615.615 (RB) where f= radius of drainage area, ft From Equation 1-6 we gett = 0.0417 @ Bs on PAP ‘Do find the expression for tgs in Equation 1.2, let us rewrite this equa: ‘Hon: Py~ Per = myo t + Cpe a ‘Now, by subtracting! Equation 1-7 from Equation 1-2 we eliminate P, and obtain: : oat Pace fy OHA t+ Cpe a8) ‘Since we have already established that during the semi-steady state @ — Pu) is a constant, then: 0.07 gl | ree VC py = 2041 a Be as) eV, 0 Accordingly, Equation 1-2 could be written as: O07 ao. BP) (10) PP ‘Equation 1-10 deseribes what is known as the reservoir limit test. To per- form this test, the well is produced at a constant rate, q, until the semi- Fundamentals of Drawdown Testing 13, steady state is reached, then a pressure recorder is lowered into the well, without closing the well or disturbing the flow, and left for 72 hrs or longer. If the well is indeed producing at the semi-steady state, then a plot of AP versus the time t yields a straight line. From the slope of the straight line, itis possible to determine C, V,. If C, could be determined with confidence, it ‘would then be possible to determine V,. ‘With respect to our well, we have already established that ngs ~ 0.2499 psifhz, Let us assume that from PVT and well-og data we determined C, at 15.55 x 10° (psi~!, then we apply Equation 1-9 to caleulate V,: 0.0417 x 100 x 1.35 0.2489 x 15.55 x 10-* = 1.489 10° RB y, and, 6h 489 x 108 x 5.615 x 1058 x20 = 012 ‘To complete the formulation of Equation 1-10, we have to find the expres- sion for (P ~ Py). For this purpose we apply Darcy's law to radial flow and. ‘then find the volumetric average pressure, P. The procedure is explained by Craft and Hawkins (1959), where the equivalent of the following equation hhas been derived: Po Peg = 32522 Bo tl ygg 0.472 nl : poran SBR gate wy ‘Thus, the final form of the semi-steady state equation is as follows: 325.2 q Ba ul), 0472 1 2g Bl tee] au 0.472 “| (413) a 14 Well Test Analysis “To complete the formulation of the transient drawdown equation (Fqus- tion 1-5), we mus find the expression for AP. We could arbitrarily choose to express AP as follows: AP, = m log C’ and proceed to find the expression for the constant, C’. With this choice, Equation 1-5 is written as: AP = mlog (C' ) ren) Since we have established that the dimension of m is that of pressure, then (C’f) nmst be dimensionless, And, from measurement of the slope, m, we are able to caleulate the transmissibility, , defined as, T=khip But the reservoir i also characterized by its storage, S, defined by, ay S-6hG ne Yow “Thus, we may intuitively suspect thatthe dimensionless expression (C’t) n- ‘cludes both T and § as well as the wellbore radius rq, since it has not been included in the expression for m. Indeed, the following expression is dimen- sionless and satisfies our intuition: oy, Gy ; \t aN oe ae wh where N = « number oe is, 4 Ni a J - ‘TodetermineN, we choose any data point from Table 1-1 during the early transient state, i¢., at < 40 hrs. By choosing t = 0.1 hr and the corre- sponding AP = 74.8 psi, and substituting in Equation 1-14, we get: ce N= 5.916 x 10-* Since log N= ~ 3.23, Equation 1-5 can now be written in its complete form as: Fundamentals of Drawdown Testing 15 k ap =m log t+ tog [-* | ~ 3.29 ee 15) fe lecal | uw oe Equations 119 and 1-15 are valid when the following set of ld units is used: psi, STB/D, RB/STB, md, ft, hr, ', and ep. Although the time, tin Equation 1-18 is given in hours, the slope ofthe straight line on the semi-log plot will not change if tis plotted in days, min- tte, or seconds, Let us consider the casein svhich is measured in minutes Equation 1-15 could be written as: " real en Ce arm fg (+o locke ‘This equation shows again that a plot of AP versus log t, where tis in min- utes, yields a straight line of slope m. The same results obtained when tis measured in days or in seconds, From the preceding discussion and refults we conclude that if a well is ‘closed until the pressure stabilizes at P, and then produced at a constant rate, 4, the following sequence of events takes place. uring the early transient state, a plot of AP = P, ~ Puy versus log t gives a straight line of slope m = 162.6 q B, x/k h. From the slope, itis fy possible to determine the transmissibility, T, and if the viseosity 1, s <7 Keown, determine the production eapacity Kh, and ifh is known, de. termine k. If the wel is closed during the transient state, the pressure will stabilize essentially at P 2. Shortly after the pressure transient reaches the drainage boundary, the “emi-steady state begins. During this state the rate of pressure diop as ‘measured at the wel is equal to the rate of pressure drop throughout the drainage area of the well, which means that P ~ Pyris a constant, where P is the pressure at which the well would stabilize if it was closed during the semf-steady state. A plot of AP = P; ~ Pas (or just Py) versus t gives a straight line of slope mgs from which the product V, G.can be determined. However, identification of the beginning of the straight line on the rectangular plot can be difficult and uncertai. Depending on the scale used, it is quite possible to identify a straight line that, in reality falls in the transient state, which would give the {impression thatthe drainage area of the wel is smaller than it really i. 16 Well Test Analysis [dentification ofthe straight line on the rectangular plot fs even more dif- ficult when the production rte, q, is small in comparison to the production capacity ofthe well, kh, Tb see this, let us assume in our example of Table rT thatthe reservoir thickness s 100 f instead of 20 ft. Equations 1-13 and 115 clearly show that both m and mp, are inversely proportional to the net pay thickness, h. Thus, when h ~ 100 ft, the pressure drop is obtained by Iltiplying the pressure data in Table L-l by 0.2. ~~ Figure 17 is a plot of the pressure-time data, assuming that his 100 ft. Tt should be clear ftom this figure that neither the beginning nor the slope of the straight line could be determined with any degree of accuracy. Bearing in mind that these data are ideal, actual field data would be much moro Aiffca to interpret. Le AP=(Pi-Pul) ps 4" cy 7 e810 1040 Producing Time tp (hes) Figure 1-7. Semi-og plot of (Pt ~ Pw) vs producing time, t. Data from Table 1-1 multiplied by 0.2 eee PARK JONES METHOD “This method, introduced by Park Jones (1956), i sometimes helpful in de- tecting the semi-steady state, expecially when Pye fluctuates due to small fluctuations inthe production rate, q. To develop the necessary equations, we first rewrite the transient drawdown equation (Equation 1-15) in terms Fundamentals of Drawdown Testing 17 of he natural ogenth, In, and then take the dervative with respect ot as jn 5918 x 10-4 Inte onan dap _6qBp and 44? 706.9 ae” kht ‘Next, we define the function Y, as follows: y= 24 dar ab dt ‘Then, for the transient state, 1004» Y kht (L186) and tog Y= ~Iog + ng OHA an ‘Equation 1-17 shows that during the transient state a plot of ¥ versus t on Jog-log graph paper should yield a straight line, making an angle of 45°. Itis, ‘possible to determine the permeability, k, by choosing any time, t, and read- ing the corresponding value of Y off the graph and then substituting in Equation 1-16. Differentiation of the semi-teady state drawdown equation (Equation 1-13) with respect to t yields: aap _ 0.0417 By at VG from which, 1 YY (eins) (43) “Thus, during the smtsteady state a plot of ¥ vers ton log og graph pa yields horizontal ine with an intecept equal to 1/V, Cy. OF couse, the 18 Well Test Analysis interpreter does not need to know a priori the beginning of the semi-steady state, All thats needed isto calculate the Y function and then plot ¥ versus ¢ fn log-log graph paper, (This plot will be referred to here as the Park Jones plot.) ‘Dpillustrate the application of the Park Jones method, we will utilize the data of Table 1-1 and Equations 1-13 and 1-15 to fill in additional data points, ‘The results aze shown in Table 1-2, and Figure 1-8 is the correspond- ing Park Jones plot. Table 1-2 / IMustration of the Park Jones Method Field Data ‘alculation of ¥ Function dap 1 ap at en, (osn (ositne) ° es = = o4 var resi" = 10 9233 sat 220 20 9782 30 o78 30 0071 22 047 40 102.80 uv 035 50 104860 geo 12007 02, - 90012027 ons: 0.082 wo 120.42 0.243 04s 0.243 08s + Batteo 495 [98.51 + 62012} = 220 Referring to Figure 1-8 and choosing t ~ 4.5 rs, the corresponding Y = 0.3, Thus, by Equation 1-16, Lots w j= LOLA He Yat 1604.4 x 0.8 Osx Wx45 = 50 md Fundamentals of Drawdown Testing 19 0 10 7 7 toe TOY Producing Time t, hrs Figure 18. Park Jones plo; data from Table +2 From the intercept of the horizontal line which describes the semi-steady state on Figure 1-8, the reservoir pore volume of the drainage area of the ‘well is caleulated by Equation 1-18 as follows: 1 Veeco YG a 0.0432 x (15.55 x 1 = 1.489 x 10 RB Its evident that the Park Jones method is helpful in detecting the begin: ning of the semicdeady state, hut it requires highly arcurate data. Ale, when the rate of pressure decline is very small, and when real field data are ‘being used, itis possible to obtain the horizontal line which is characteristic ‘of the semi-steady state even though the well is stil in the transient state, In ‘other words, the Park Jones method is not always successful in correctly 20, Well Test Analysis {identifying the beginning of the semi-steady state. The main strength of the Park Jones method is in interpreting the transient state. Because it is often difficult to stabilize the production rate, q, from the instant of producing the well, the semi-log plot of AP versus t may not clearly project the straight line as predicted by Equation 1-15, On the Park Jones plot, however, the data could form a scatter, but the lope of the straight ine to be fitted to the seatter is always equal to one. . Tt is important to remember that if several wells afe producing from the same interconnected segment of the reservoir, and if the well being tested is thought to be producing at the semi-steady stato, then all the wells in that segment of the reservoir must be producing at the semi-steady state. Thus, regardless of what the Park Jones plot shows, the interpreter can decide from the production history of the neighboring wells whether all the wells are producing at the semi-steady state. Also, i the Park Jones plot indicates that the semi-steady state has been reached but calculation of V, leads to an un- realistically small r., then the interpreter should realize that the appearance of the semi-steady state on the plot is erroneous, "THE PRESSURE DERIVATIVE METHOD. ‘This method, introduced by Bourdet etal. (1084), is somewhat similar to ‘the Park Jones method. However, in the presure derivative method, the de- rivative s taken with respect tologt, and the results are also plotted on log- Jog graph paper. During the transient drawdown period, the necessary equation is derived as follows: k = m|log t+ - (0-15 AP mllogt + log Es - 9.28 (15) and ———~ wit? m\ as) og 5. ‘Thus, a plot of d AP/d(log t) versus ton log-log graph paper should yield « horizontal line with an intercept equal to m. During the semi-steady state, we refer to the semi equation: LO0017 Be 6 9m fing 2472 - ap = 20872 2m [og A ag weady state drawdown Fundamentals of Drawdown Testing 2 dap _daP) dt Since Fogg” dt * IQoe5 when ASP ONT GB, dt Tog)” VG 0 * Damage Pan <0 —+ Stimulation ‘APan = 0 + No damage or stimulation ‘To account for the skin effect, Equations 1-15 and 1-13 are modified as follows: D ap =m [loge + og 5k ~ 9.95] + Pan o portoves reg 99 2 ana ap = 2079 52m fig A ar aay We could define APjys in any way we please. For example, we could de- fine it as: Pa where s is a dimensionless number called the skin factor. However, in well testing, AP, has been defined as: _ML2 a Be Fh AP ain = 087 ms Py Well Tast Analysis observed prossure vransiont ke > k= stimulation Pan = Pitta ~ Pld ~ He) Figure 1-10. The skin zone. Fundamentals of Drawdown Testing 25 Accordingly, Equations 1-15 and 1-13 are written as follows: k oO ee 42m fg 2 + 04055 o: AP =m [logt+ 3] (29) and AP = img t+ Gye 40.87 ms 2 where Flog K-28 +0875 sm, = 20817 4 Bo ve ViG Ata, Cpe = 2m flog It should be evident thats cannot be determined by Equation 1-28 unless the permeability k, is known Since both k and sare readily obtainable from the transient drawdown equation (Equation 1-22), Equation 1-23 is not used for the determination of either k ors, Its main use is to determine V, 6, ‘By solving Kiguation 1-22 for s, one obtains: kt ai eGR +3.23) 34) ‘To find s, choose any time t on the semi-log straight line and substitute the corresponding AP in Equation 1-24. For example, let us refer to Figure 1-5. ‘We already found that mn = 17.56 pstieyele. Let us choose t = 2.5 hrs; the 26 Will Tost Analysis corresponding pressure drop AP = 99.33 psi, Substituting in Equation 1-24 we get +3.93| [s0.38 30x25 =0.00 ‘Asa special case, we could set ¢ equal to 1 hr in the expression for s (Equa- tion 1-24), Tn this ease the expression for s becomes: Pe k peta Big Bog = ‘Equation 1-25 is the standard equation for determination of s. To use this equation, one must choose t = Ihr, and if there is no straight line at one hhour, one simply extends the straight line to one hour and then reads the corresponding AP on the semi-log plot. ‘Te is important to keep in mind that the skin factor, s, represents the change in the permeability in the vicinity of the wellbore only if the reser- voir is horizontal and homogeneous with respect to permeability, porosity, ‘water saturation, thickness, and compressibility; contains oi of constant vis- cosity and low compressibility; and the well penetrates the entire pay sec- tion, If one or moze of these conditions is not met, or if errors are made in ‘estimating any of the reservoir or fluid parameters, then s will be found dif- ferent from zero even though its real value is zero, “To see how this could happen, let us suppose that we mistakenly took h equal to 30 ft instead of 20 ft, Then, k = 83.33 md, and by substituting in Equation 1-25, we obtain s = 0.21. ‘Although this value of ss not sufficiently large to warrant any action, the result shows that an error made in estimating a reservoir parameter could result ina value ofs that is not zero. Thus, in reality, sis not only a measure of the changes in permeability that occur in the vicinity of the wellbore; itis the sum of all reservoir heterogeneities and errors made in estimating reser- voir and fluid parameters as well a the changes in permeability in the vi ity of the wellbore. Under most circumstances it is not possible to know how much of sis due to inherent reservoir conditions that deviate from the ideal conditions. ‘in the preceding example it was assimed that an error had occurred only in the net thickness, h, However, rock compressibility, fluid saturations, and oil compressibility are usually uncertain quantities. When the reservoir pressure is close to the bubble point pressure of the oil it contains, estimates ‘of oil compressibility under testing conditions could be in error by a factor of Fundamentals of Drawdown Testing 97 10 or more, In our example, substitution of ©, = 15.55 x 10-5 instead of 15.55 x 10° in Equation 1-25 would result in a value of s of 1.20. Tt should be evident that: 520 + Damage <0 ~ Stimulation $= 0 + No damage or stimulation Avalue of s which is different from 0 does not necessarily indicate damage or stimulation unless the reservoir and the oil contained in it are homogene- ‘ous with respect to all parameters, and that the correct values ofall these parameters have been inserted in Equation 1-25. Recently, Villalobos etal. (1989) showed that because of turbulence, a positive skin will he obtained ‘when analyzing tests of high volume oil wells ‘The skin effect ean also be viewed as a modification of the wellbore ra- dius, To see this, Equation 1-22 is written in terms of the natural logarithm, as follows: _ 0.6 Bu {,, 5.916 x 10-# kt ap = DEAR E lig SOC aas ouGre fnssicxio~ i one) a ng = appt wel rad Then, Ta = Fw e°* ‘and the equation can be written in terms of ra, as follows: HC Tay Ls Teshould be evident that for positives, tyy < ty and for negatives, tgs > Tye THE RADIAL FLOW EQUATION Jn the preceding sections we have derived the dravidown equations on the basis of simple mathematics and intuitive reasoning. These equations are valid only when the pressure, Py, is recorded inthe wellbore. Ifthe pressure is measured in an observation well located at a distance r > ry, then the ‘equations we have derived until now are not valid. 28 Well ost Analysis [An equation that is valid at an observation well is derived by solving the radial flow equation, which is @ linear, partial differential equation given (1-26) where S and T are the storage (¢ h C) and transmissibility (K hip), respeo- tively. The pressure, Py 18 that observed at any given time, t, at any point ‘within the drainage area of the well located a distance r = 14. The constant 3,192 (=1/2.64 x 10-4) is necessary when field units (psi, ft, hr, md, STBY D, ep) are used. ‘Derivation of Equation 1-26 is based on the following assumptions: 1, The flow is radial 2, The fluid is slightly compressible. 3. (@PIan}* = 0. Radial flow occurs only if, within the drainage area ofthe well, the zeservoi is Ihorizontal, each of its properties (k, Sey Cy bi the same in all directions, the well penetrates the entire pay section, and the reservoir is thin such that vertical variation of P due to gravity is insignificant. The second requirement ‘means thatthe fluid compressibility is constant, which means that Equation 1- ‘26 isnot suitable for gas reservoirs. The third requirement means that the pres- sure gradient P/ér at any point located at a distance r from the well i small such that (@P/9r)? = 0. For example, if dP/@r = 0.01, then (9P/2:}* = 0. There- fore, if AP atthe wellbore is very large, then Equation 1-26 may not be appica- ble. Also if the skin zone is not infinitesimal but extends a finite distance, ty from the well, then Equation 1-28 may not be applicable inthe strict sense. Its very important to keep these requirements in mind because ignoring them often leads to wrong interpretations. ‘The transient solution to Equation 1-26 is given by: azn | (1-28) \ where AP =P\-Pre : ja presse observed at time € at distance from the well Fundamentals of Drawdown Testing 29 Atte to, Pra = Pos, and Equation 1-28 becomes equal to Equation 1-15. ‘This actually happens a few seconds after producing the well at the constant rate, q, because the exponential integral, ~ Ei {x}, can then be approx- {mated by a logarithmie function. In general, for x < 0.02, the exponential {integral is approximated by: Ei { —x} = ~ 2.908 log x + 0.25) Forr > ry, x could attain values less than 0.02 at large values of producing time, t. Thus, the logarithmic approximation is valid at r= ry when t is ‘greater than a few seconds. For r > rq, the approximation is valid providing that the producing time, , is sufficiently large such that x < 0.02. Forx > 10, the exponential integral is essentially equal to zero. The exponential in- ‘twgral is presented in graphical form in Figures 1-11, 1-12, and 1-13. Tt can also be evaluated on @ computer through the use of the programs given later in this chapter. 10%, 107 4 ‘i t 10 5-9) Figure 1-11. Evaluation of the exponential integral. X = 0.001 - 0.4 Fundamentals of Drawdown Testing 31 : “To see an application of the exponential integral, let us suppose that we c wish to ealeulate the pressure drop that should be observed at a point lo- 7 cated 100 ft from our well. For this purpose we evaluate Equation 1-28 as follows xe (gp 770.8 X 100 1.35 x 0.8 | | = =7.6248 Ei | By evaluating the exponential integral at different values of t, we obtain the | data shown in Table I-4, J Suppose that AP at the producing well is not equal to zero, but the skin word be iH | zone is infinitesimal. What would be the pressure drop at the observation 10 KH) : well? The answer is that the pressure drop atthe observation well will not | ‘change from that shown in Table 1-4. This is because APa occurs only ia Figure 1-12. Evaluation ofthe exponenti | the vicinity ofthe producing wellbore. Thismay not be true, however, if the 10 I H Table 1-4 é Pressure at Observation Well 7 i (r= 100 1) x i ' -5 { = C | (we) oon 7 i 0s Boa x 10-+ = * | 008 Bare to 52x10 | O10 ‘69 x 10-* ox8 O45 oosr 043 7 o20 0333 O88 00 0.49 358 ono O70 B08 ‘00 O96 12 200 528 ree 00 to 1489 “00 2am teva VO Die rita ag t 500 23a iver sooo 3028 aor 8D 20 3705 mas ano ste a8 Figure 1-13, Evaluation of the exponential integral. X = 1.0 - 10.0. 32 Well st Analysis skin zone is large and the observation well s very close to the producing well. DETECTION OF A LINEAR BOUNDARY [A linear boundary is « physical boundary such as a sealing fault, the oi ‘water contact, permeability pinchout, or an abrupt change in the diffusi ity, 9, defined by: 2.64 x10. 5 2.64 x 10-4 k on ‘Suppose that the tested well is located near a linear boundary, as shown in Figure 1-14, Intuitively, one would expect the pressure drop as observed in ‘the tested well to be greater than the pressure drop that would occur if the boundary did not exist. To account for the additional pressure drop due to the presence of the linear boundary, an image wel is placed on the other side of the boundary at a distance equal to the distance between the original well and the boundary (Figure 1-14), and the boundary itself is removed. The pressure drop at the original well would be the sum of the pressure drop as caloulated by Equation 1-22 and the pressure drop due to the image well as wel! ‘mage Figure 1-14, Well near a linear bound- ary, Fundamentals of Drawdown Testing 33, calculated by Equation 1-28. Thus, the pressure drop at the tested well is given by: G (2d) AP =m flog t +5] = (1-29) where AP =P; ~ Pye = distance between well and barrier (Figure 1-14), ft At the beginning of the test, when tis very sinall, the exponential integral is ‘essentially equal to zero and the pressure drop at the well is represented by Equation 1-29. As t increases, the exponential integral is represented by the logarithmic approximation, and thus the pressure drop is given by: arm flogt 3}~m og (M82#G- 28 «05 =m toge +5} fot te gE apa 8.25} = 2m log t + constant (1-30) According to Equations 1-29 and 1-80, the slope of the straight line on the semi-log plot doubles as t becomes large. Thus, in the presence of a linear boundary, a plot of AP versus t on semi-log graph paper produces the usual straight line of lope m, and as t becomes large, the slope gradually increases ‘to 2m. The skin factor, s, can be determined by Equation 1-95 by finding Phir on the straight line of slope m. It cannot be determined from the straight line of slope 2m, beceuse, as Equation 1-30 shows, the constant term (the intercept) includes two unknown quantities: the skin factor, , and the dis- tance, d, to the linear boundary. Once the skin factor is determined from the early straight line, it would then be possible to solve Equation 1-80 for the distance, d. “Table 1-5 presents a listing ofthe drawdown data plotted on Figure 1-15. Note the presence of the two straight lines as predicted by Equation 1-30, Note also that the change from m to 2m occurs over more than one cycle of time, Krom the slope, m = 17.56 psi/eyele, and using the same flow, PV'I, and reservoir description parameters of our well, the permeability is found equal to 50 md. From the value of APjy, = 122.8 psi and Equation 1-25, the skin factor, s, is estimated at 2. ‘M4 Well Test Analysis Fundamentals of Drawdown Testing 35, Table 1-5 ‘Toestimate the distance, d, between the well and the linear boundary, we Detection ofa Linear Boundary make use of Equation 1-20: eee 58 Aor =m +o -3: om om (es) log t+ ogee nae 20 ‘so29 0.87 s+ log_—_* sx 20 * +0.87 s+ log* __, - 008 20 14589 ; P85 uC, a ox0 40 os ‘ ; 2 0 09 qeast ofetore 2ts k— — 640 +0875 2g} an O80 20 traa0 oEGh 930 soo ‘e000 : me ee Thus, d = 0.5 x 10-7 039) ar k where y= fa log ~ Io + 8.28 ~ 0.495 , i BE EG | ‘To determine d in our example, we choose any point on the straight line of ° slope 2m and substitute in Equation 1-32. Choosing t = 10 hrs and AP = 165.5 psi, \ 4205 x 10-(0e0-1-t+332-089 g =48 ft ‘The distance to the linear boundary may also be estimated by finding t,, g the time at which the straight lines of slopes m and 2m intersect (Figure a 1.15), Then by equating Equations 1-22 and 1-30 we get: 7 si + k é milo som[s2s—t taal & id * 570, Ga a8 kt f - or d= 0.012 | =| ow Cl 7 In our example, t, = 0.45. Thus, the distanee, d, is given by: ‘ see 0 i 10 10" 90 x 0.9 Figure 1-15. Semi-log plot showing change in slope due to a linear boundary; data from Table 1-5. sar it 36° Well st Analysis [At this polnt itis necessary to stress the following: 1. ‘The first part of Equation 1-29 Is a solution of the radial flow Equa- tion 1-28 at r = rq. The second part of Equation 1-29 is a solution of Equation 1-26 atr ~ 2d. The summation of the two solutions, as given by Equation 1-29, is known as superposition with respect to distance, ‘Later we will see the other form of superposition: superposition with respect to time. 2, The presence of a linear boundary causes the slope of the straight line on the semi-log plot to change from m to 2m, But, the occurrence of ‘ovo steaight lines of slopes m and 2m does not by itself imply the pres- cence of a linear boundary. Later we will see that a change in the slope of the straight line could occur in response to other forms of reservoir hheterogeneities. Thus, unless there is supporting evidence from gecl- ogy and/or seismic interpretation, well tests can only infer the exist- cence of the boundary, {tis possible to extend the method of images to account for intersecting ‘and parallel boundaries. For example suppose our wel is located near two linear, mutually perpendicular boundaries, as shown in Figure 1-16, In this Image Wt oS ° nage imaoe & ‘ Figure 1-16, Well near two perpendicular boundaries. Fundamentals of Drawdown Testing 97 ‘case we place three images as shown in the figure. To account for images 1 and 3, we simply add the data of Tables 1-{ and 1-5. To account for the secondl image, we evaluate Equation 1-28 with r = 2 V2 d. The results are shown in Table 1-6 and plotted on Figure 1-17. Note that in this case the slope of the second straight line is 4 times the first slope. In general, the fol- owing rule applies: ope of won ara ie = 22 where m = slope of the first straight line, psileycle {= angle between the two boundaries If the two boundaries are parallel, then the number of images would be infinite, in much the same way as when one views himself/herself between ‘two parallel mirrors. Of course one need only account for a finite number of & i 2g ISN z 2 zd = 144 3 g Yo* 107 10 10% 1 Producing Time tp, hrs Figure 1-17. SemHiog plot showing change in slope dus to two perpendicular boundaries; data from Table 16. : perpen 38 Well Test Analysis mages, since the distant images would have a negligible effect, and the more distant the image isthe longer it will take for its effect to be felt at the producing well. ‘COMPUTER PROGRAMS TO EVALUATE THE EXPONENTIAL INTEGRAL ‘Programs 1-1 and 1- were written in GWBASIC, For values of x = 1.5, ‘evaluation of ~ Ei (—x) was based on the following expression: -Bi(-a)- faa tne 9) — Ms (u/3!) + I 1 was found sufficient to carry the integration to the upper limit of 10.5 instead of o» and to evaluate the series to x", For x > 1.5, itis best to inte- tate by Simpson's Rule over the interval x and x + 10.5. But to obtain good accuracy, the interval was divided into 2,000 intervals. Because GWBASIC has no direct means of calculating nl, it was calculated by evaluating the ‘Gamma Function: Ti+) Although, in this particular program, the factorial could have been easily ‘evaluated inside the loop (lines 60-100), the Gamma Funetion was given to show an alternate method of calculation. Fundamentals of Drawdown Testing 39) Program 1-1 Evaluation of Exponential Integral Given the Value of x CLSDEFDEL AHOZINPUT XA IFA > 15 THEN GOTO 210 REM INTEGRAL(EXP( ~ X90) = LNOQ = X + SUM( = 1) ~ Nex ~ NWN) Y4 = L0a(10.8) ~ LOGI) Y= 0va~0 FORN = 17031 osus 140 B= YB + ((~ 1)" Nw (10.5) ” NY GAMMA) YS.= 3+ ((= 1)“ N)(A~ NVINSeGAMMA) NEXT Ya= Yi + Y2— YOPRINT “THE INTEGRAL IS "v4 oro 330 EM GAMMA + 1) =X! XeNet Pim 316150068 42, G2 = SAR(2 PUY) xEXPLY xLOGQY) + (1 ~ 190" RYZE Y) ~ Y) GAMMA = G2IK#(K + 1) RETURN REM INTEGRATION BY SIMPSON'S RULE DEF FNFOQ = EXP( ~ XK: B= A+ 105 KK = 2007DIM YC): D = (B ~ AYIKK = 1: X= A FOR J = 1 TO KK: YW) = FNEQO: X= X + DINEXT J ono) TO KK = 2 STEP 2 (000 = 000+ 2m YY) NeXT ev-9 FORJ = 270 kk -1 STEP2 EV~EV+ 44 YY) NEXT Dise(vV(t)+ EV + 00D + YK) PRINT "THE INTEGRALS ":¥ END Program 1-2 Determination of x Given the Value of the Exponential Integral ‘OLS: DEFDBL AH,OZDIM Zi-001): DEF FNDOX = EXP — 70% INPUT "WHAT IS THE GIVEN VALUE OF ~ El(~)":¥INPUT "YOUR ESTIMATE OF x" Ax FOR = 1 70 100 GosuB 110 IPARSYY ~ ¥4) < ont THEN 410 X12 x= (¥ — YayENODH) =x sgagese Bs (orvram cotnuad on nx page) 40 Wall Tost Analysis Program 1-2 5 Continued 400 NEXT 110 IFA > 1.8 THEN GOSUB 290 :RETURN, 420 EM INTEGRAL(EXP( ~ XV) = LIX) ~ X + SUM((— 1) “ NeX” NINEND) 490 ¥1 = LOG(t0.5) ~ LOGrA) 140 ¥2.=0¥3 = 450 FORN=1 7031 460 GOSUE 370 YB ¥2 + (= 1) * Npx((105) * NN GAMMA) 180 Y8.= 8 + (= 1)" N)w(A” NVINRGAMNA) 390 NEXT N 200 Ya= Yi + ¥2—Y9 210 RETURN 220 REM GANMAQ + 1) =X! 230 Xt= Net 240 P= a.14180204 250 YO= XI +2 250 G2 = SOR(2xPIYO}+EXP(YO*LOGNO) + (1 ~ 11{90%YOx¥0)/(12#Y0) ~ YO) 270 GAMMA = G2IRK1#(K1 + 1) 280 RETURN 290 REM INTEGRATION BY SIMPSON'S ULE 300 B=A+ 105 10 KK = 2001: D = (8 — AMKK ~ 9): XO = A 820 FOR J = 1 TO KK: 2) = FNO(K2) X2 = X2 + DINEXT J 300 00D = 0: FOR J ='STO KK ~2 STEP 2 {340.000 = ODD + 24210) 350 NEXT J 380 EV=0 370 FOR J = 270 KK 1 STEP 2 280 EV =EV + 4424) 290 NEXT J 400 Y4 = Did (Z(t) + EV + ODD + ZIKK):RETURN 4310 PRINT “THE VALUE OF X I 420 END Program 1-2 is used when the value of ~ Ei (~x) is given and it is desr~ able to determine x. The program is based on the Newton-Raphson method and makes use of the entire Program I-1 as a subroutine. RADIUS OF INVESTIGATION “The radius of investigation, rin, of a given test is the effective distance travelled by the pressure transients, as measured from the tested well. Intul- tively, one would expect riy to be proportional to the duration of the pro- duction time, ty, of the test. Thus, ray of a test that lasts for 24 hrs is larger Fondamentals of Drawdown Testing 41 than the fey of atest that only lasts for 6 hrs providing, of course, that the semi-steady state is not reached during the first 6 hrs of the test. Ifthe se steady state is reached at t, = 5 hs, for example, then the radius of invest gation for any t, > 5 hrs is equal to the radius of drainage, re ‘The radius of investigation depends on the speed by which the pressure ‘waves (transients) propagate through the reservoir rock. This speed is deter- 1mined by the reservoir and fluid properties, such as: porosity, fluid vis- cosity, a; total compressibility, Cs and permeability, k, However, like sound ‘waves, the speed of pressure waves must not vary due to variations in ampli- ‘tude. Since, for a given reservoir, pressure amplitude variations are directly related to variations in the production rate, q, the speed of the pressure ‘waves must be independent of q. ‘To find the radius of investigation, we have to seek an expression for the speed of the presure waves through the reservoir rock. To do that, we must first become familiar with the meaning of an impulse; second, we must re- view the rules of differentiation under the sign of integration; and, third, the reader should review the beginning of Chapter 2 regarding the principle of superposition with respect to time. ‘A.unit impulse, or an impulse, is simply a pulse of a very short duration, At, such that qB, At ~ 1 RB. If, for example, we inject in a well for At = T minute, then to create an impulse, we must inject at the constant rate, q By, ‘of 1,440 RBID, because with this injection rate q B, At = 1 RB. If we choose to inject for only 15 seconds, we must inject at the rate of 5.760 RB/D in ‘order to initiate the impulse. Although creating an impulse cannot be achieved in practice, the impulse i a significant concept which forms the basis of impulse testing (see Chapter 8). ‘Having defined the impulse, let us refresh our memories with some of the clementary aspects of calculus. In particular, # uy and ty are functions of both r and t, then, eH au, uy ot eau, a at 42 Well Test Analysis a then 8 {ew} ext bet Huj=brit, wee, 2 {mous} ao te t “Let us imagine that at t = 0 we initiate an impulse in the reservoir by {injecting in a well while observing the pressure in an observation well lo- ‘cated a distance r froin the injection well. Our objective isto determine the time of artival of the impulse at the observation well, and thus be able to determine the speed of the presure wave through the reservoir, Let, = qBs ~ injection rate, RB/D qB, + production rate, RB/D "Tofind the pressure drop AP, = Pj ~ P,x at the observation well, we super- pose Equation 1-28 as follows: 70.6 (~ Bt py Kb (aPeade= and (AP.Je™ ‘The sum is given by: AP. = (APradi GBs 70.6 a Bus fp | —248 6 nC _ -msam Ia kt a Fundamentals of Drawdown Testing 43 rh 45 fb which, in the limit as At — -bs t Denes (1-33) Equation, 1.38 is known as the unit impulse response. We could ignore the negative signin this equation and keep in mind that an impulse induced by injection causes a pressure rise atthe observation well P,, > P), whereas an impulse induced by production causes a pressure drop (P, > P,,). Without the requirement that q B At/24 = 1 RB, the derivation would have led to: 1,604.4 g B Ate APs MEAT kt Let Q = total production or injection, RB, then Q = q B At/24 and the ‘above equation becomes: 1,604.4 Qu kRt (1-34) 948 6 Cl kt Equation 1-34 is valid when the elapsed time, ¢, is much greater than the uration of the impulse, At. ‘Equation 1-33 has been evaluated, without the 24 multiplier, atr = 10ft for a reservoir of the following parameters: @ = 0.12; 4 = 0.8 ep; Deserta Coenen aseciencenatie Ladson 44 Well st Analysis ©, = 15.55 x 10-Ppsi; k= 50 ma; and h = 20 ft. The results are given in ‘Table 1-7 and displayed on Figure 1-18. Note that the maximum pressure response occurs at taux Which is given by: fossec. ‘| te ‘This result could have also been obtained by differentiating Equation 1-33 ‘with respect to t and equating to 0. "The equation for tna ives the relationship between the time of arrival of the pressure wave and the distance z. Thus, in general, for a drawdown test that lass for period of time ty the rads of investigation i given by a ry = 0.0525 el 1.35) x6 : Although the radius of investigation is independent of q, our ability to measure the change in pressure, AP, at an investigation well depends on q- Table 1-7 Radius of Investigation (r= 10 ft) (ts) AP & esas 04 {aeH0E-08 ONO ‘5.660705 08 ee7e9e-02 020 BaO10SE=08 «2 7OGSELOD © S.z8RAOE-O1 OD {asvi4e 03 —«4.00806E +00 «5 S7BRSE-01 040 {aist7e—08 —«SADIZTE+0O —«7SS7SQE-O1 0.60 {enezie-03 —«B2B17E+00 —«BSSERSE-01 0.60 Lose 0s —««GESIGTEY0O © 9.S0627E-01 0.70 22628603 —«“714055E +00 © B.7SS01E-01 0.80 2ee7sie-0s —«729888E +00 -—«.BHaSBE-01 0.00 2ea0s5 08 —«7-34108E +00 ———*1.00000E+00 1.00, 511330603 7. 80885E+00 —«9.95809E-01 1.10 300e2E 08 «7. 22706E +00 © S.BAABTE-OV 1.20 Beradse—08 © 7NZT7E+00 © .88809E-01 10, 3962495-03 «GO7776Es00—«GSNSBE-O1 140. G2assoe 03 —«G.88021E 400 98NK0BE-01 180 “52050608 —«GB7S7OE400 —«9.08069E-01 1.60 ‘arsse—0a «GS 1006E+00 © B.S7OZNE-01 1.70 SopiesE—o9 —«G 0074E +00 .REASTE-O1 1.80 SS77r6eE—03 —«G2047SEF00 —«BASIOE-01 190 Bee070E-03 —-«G.0S170E+00 —«S.2ABEDE-01 2.00 Sous7se 0s ——-S902H0E400 ——BOKUAZE-O1 210 “By md athe remand prosute op and ns, reapsciveh Fundamentals of Drawdown ‘esting 45 J] ©. [~~ Ge : ° LJ °F 1 2 4 5 6 & (ve x €-03) 3. Impulse response at r = 10 f; data from Table 1-7 ‘The higher is, the lees the pressure drop andthe eer rw measure it. Even with the use of very sensitive pressure gauges, the pressure Grop aan obervation sel st be ager han the background noe gener. ated by lunar-solar tides and other activities in the field in order for us to detect it, and thus q must be sufficiently large. ‘Table 1-1 shows two sets of data. The first set consists of time, t, and AP as calculated by Equation 1-33. This set of data is plotted on Figure 118. The second set of data is obtained by normalizing the data of the firs set. The normalization is made by dividing each time increment in the fist set by tian and each AP by AP. These data are shown on Figure 1-19. The pur- pose of Figure 1-19 is to show the form ofthe response at the observ ‘well. While the amplitude ofthe response shown on Figure 1-18 will dim: ish as the distance to the observation well, s, increases, the form of the re sponse as shown on Figure 1-19 will not change. The reader can easily verify this conclusion by placing the observation well at r = 100 ft, for example, substituting in Equation 1-33, and then calculating the normalized time and pressure "The form of the impulke response is rather strange: it appears as if t eon. sits of diferent presuze components. Some of these components travel faster than others, which makes the response somewhat diffused even though it was generated by a sharp pulse. For this reason, its not possible to obtain the radius of investigation by simply substituting into Equation 1-28, 46 Well Test Analysis 10 09 oo OS TO 1S 20 25 SO Normalized Time Figure 1-19. Normalized impulse response; data from Table 1-7. as this procedure could conceivably give a radius of investigation which is ‘much larger than that obtained by Equation 1-35 due to the fact that some pressure components do travel faster than others, as shown by the norma- lized pressure response of Figure 1-19. TEqustion 1-95 could have also been derived by a totally different method {tis reasonable to expect the slope of the transient drawdown to be equal to the slope of the semi-steady state drawdown at r= r,, Thus, by equating Equations 1-16 and 1-18 we get: 1,604.4 4 5.615, Elite eho 948 6 Ce K Thus, Ge 0:36) Fundamentals of Drawdown Testing 47 where tps = time to reach the semt-steady state, hrs For any producing time, tp, Equation 1-36 can be written as: te MBS EC ee " k which is the same result obtained through the impulse response. However, as we will ee later, if the shape ofthe drainage area isa square rather than & circle, Equation 1-36 is not entirely correct because it takes a little longer to reach the semi-steady state, and the equation for tis given by: eo HMRC, aan "THE STEADY STATE GASE, So far ‘ve have discussed both the transient and semi-steady state cases. Now we introduce the steady state case that occurs if thé pressure at the boundary of the drainage area is maintained constant. This happens ifthe boundary of the drainage area of the wel isthe oil-water contact and the ‘water drive is active, or ifthe well is surrounded by injection wells as in a ‘water-flooding situation. It can also happen if the gas-oil contact between large gas cap and the oil zone forms the boundary of the drainage area of the ‘well. In all these situations, the tested well experiences the usual transient flow period which is followed by the steady state. tis seldom, if ever, that a true steady state can be reached except under flooding conditions. But, as we will se, there is a major advantage to the study of the steady state case When the pressure transients reach the drainage boundary where the pressure is maintained constant, the flow through the boundary must equal the production rate ofthe well q. Asa result, no pressure decline in Pu is further observed at the producing well, and the pressure at the external boundary, ,, must remain equal tothe initial presse, ,. Under these con Alitions, the drawdown equation is given by: less asm ty” 3303] 48 Well Tet Analysis Equation 1-98 is derived in Chapter 11. The equation shows that during the steady stat the presure drop, P. ~ Pag is not dependent on time. Ace cordingly, when testing # wel that produces from a drainage ares bounded bya constant presse boundary, the observed pressure atthe well wll ni tally decline as deeribed by the transient flow: equation (Equation 1-22) find once the presure transients have reached the drainage boundary, the presi drop will remain at the constant value given by Equation 1-38, “Thus, a plot of AP = P, ~ Pa (or just Pg) versus time on rectangular graph ‘paper wil give the usual curve, whichis characteristic of the transient flow period, followed by a horizontal line, On semi-log araph paper the data will Show the usual straight Line of slope, m, followed by a horizontal line. “The results stated above can also occur in practice during the transient state the case ofa high productivity well which produces from a reservoir af high permeability, To see this, consider the following reservoir, produe- tion, and fluid data: q= 2,500 STB/D B, = 1.5 RB/STB po = L2ep fy = 0.33 ft = 18x 10-%psi b= 2005 6 = 0.08 = 4,600 psi s=5 k= 3,000 ma With these data, we can eastly evaluate Equation 1-22 and plot the results ‘on rectangular graph paper as shown in Figure 1-20. Figure 1-20 shows that between 300 and 1,670 hours, the pressure dropped by only one psi. Under actual field conditions, it would be very difficult to discern this very small pressure drop over such a long period of time, and therefore one could easily conclude that the steady state has been reached during the test when in fact the well is producing in the transient state. “The error can sometimes be discovered when one attempts to caleulate r, by Equation 1-38. Thus, assuming steady state conditions were reached at += 300 hours, we can substitute in Equation 1-38 as follows: 16 925.2250 %15x12 [0 4 5. “——“zo00xam «(°F 0.33 * 2.303, From which r, = 8,081 ft Fundamentals of Drawdown Testing 49 Pi = Pet, psi ap. 1“ of By 204 & 200400 ‘abo 1000 1260 1400 1600 1800 the Figure 1-20, Rectangular plot of drawdown data of a high productivity well, But by Equation 1-35, ry is calculated as follows: ,000 x 300 I Faw = 0.0825 | 5:00 S00 ___ 0.x 12x 18 x 10-*| = 12,597 ft Since ry must be less than or equal to r,, one may conclude that neither the steady state nor the semi-steady state has been reached at t = 300 hours, ‘CONGLUSIONS In this chapter we have uncovered the subject of drawdown testing under ideal reservoir, fluid, and flow conditions. Before we leave this chapter, however, we must bring the following points into perspective: 50 Well Test Analysis 1. Ifthe reservoir is homogeneous with respect to all description parame ters and the production rate fs held constant from the instant of pro- ducing the well, then a plot of AP = P, ~ Py. (or just Pj) versus t on Semi-log graph paper ofthe transient state data gives a straight line. From the slope of the straight line and its intereept, we could deter- rine the transmissibility, T, andthe skin factor, s. But we must be cog- nizant of the fact thatthe mere presence ofthe straight line does not indicate that the reservoir is homogeneous. : 2, The vale of th kn facto 5, meaning ony he reser homogeneous with respect to all description parameters and in caleu- lating, the correct, representative parameters have been substituted in the appropriate equation. 3, Under the most favorable condition, determination of the beginning Of the sem-steady state is highly uncertain. 4. Detection of linear boundaries from well tests is only feasible if tis made with a good understanding of the geology (stratigraphy and structure) ofthe reservoir. 5. In Darey units (Darey, om, se, ep, atm, color) the diffusivity, n, i defined as: arey-atm ey But, 1 Darey = 9.869 x 10-8 em* ep = 0.01 (dyne-see)iem* Latm = 1.01325 x 10° dymelom? ‘Thus, by substituting in the expression for 9, we get: i cem*lsee on Darey-atm _em® In other words, S60 Since 1 atm = 14,696006 pst and 1 ft ~ 30.4801 em 2.637 x 10-*k mest, ft Then en pir 1 2. Fundamentals of Drawdown Testing 51 REFERENCES Jones, Park,“ Reservoir Limit Test,” Oil and Gas Journal, June 18, 1956, pp. 184-196, Bourdet, D., Ayoub, J. A., and Pirard, Y. M., “Use of Pressure Derivative in Well Test Interpretation,” SPE paper 12777, 1984, presented at Call- fornia Regional meeting, Long Beach, CA. |. Craft, B.C. and Hawkins, M. F, Applied Petroloum Reserooir Engi- neering. Englewood Clifis: Prentice Hall, Ine. 1958, pp. 287-288. . Villalobos, H. L. et al., “Pressure Transient Behavior of Naturally Frac- ‘tured Reservoirs Considering the Effect of High-Velocity Flow,” SPE pa- per 19787, presented at the 6th Annual Technical Conference and Exhi- ition, San Antonio, Texas, Oct. 8-11, 1989, 2 Fundamentals of Pressure Buildup ‘Testing Preeure buildup testing the industry most favored method of well test tng. One reason for this preference stat presre buildup tess do nat e- ings e the close sypervion demanded by oer methods of testing. Another Ras hat when th wel sshutn during the trazset stato, the buildup Tepes, the nial reservoir presure atte level of measurement Ifthe aT Shatin daring the somisteady stato, then es posble to estate {he average tacrvolr pressure inthe drainage area ofthe wall. Although tous denedown snd butldup tet yield the average permeeblity, ky and {En tactor a Gravidown tt doesnot yea Py feast be estimated frm ati wal surveys before running the tot. “hese butts performed a follows dhe wel s produce at « constant rte fora period of tine, apres recorder Is lowered into the fal soy before cosng the well, nd then the wells lowed. I 1 were Yesbletoprodo the well ta constant rat, fom the instant of open- igthe wall to production, then there would be no restsltions onthe length Teeing tine, before shutting the well in However, when ty very SEMID the radios of investigation ofthe drawdown period would be very al lo and the permeability as etimated onthe basi ofthe buildup test Srould only represent the reservoir rock In the immediate vicinity of the Vellboe: Thi could be mileadng Ue permeabity ofthe reservoir rock Wik nunediste vii af the wellbore has bee altered in the course of ling and completing tho well, Usa snot posible to stablize the flow rte from the Instant the wells pot on production, Therefore itis ne teary to produce the well at a constant rate fran extended period of tine, ‘This ensures thatthe test results ane not heavily affected by Muctuations in 52 Fundamentals of Pressure Buildup Testing 53 the flow rate which normally occur when the well is first opened to flow, and extends the radius of investigation of the test beyond the immediate vie cinity of the wellbore, Later on we will discuss the effects of rate fluctuation ‘on the test interpretation results ‘The theory of well test analysis is based on the assumption that wells are closed for buildup or opened for drawdown instantaneously. This is not really true; it takes some time to open or close a well. The theory is also based on the assumption that upon shutting the well in no fluid enters the ‘wellbore, and upon opening the well to produce in a drawdown test, all the production comes from the reservoir, not from the expansion of the fluid contained in the wellbore. In reality, however, the fluids we are dealing with are compresible. Therefore, the reservoir fluid must enter the wellbore upon shutting the well in, so as to transmit the pressure buildup. Likewise, ‘upon opening the well for production, part of the production must come from the expansion of the wellbore fluids, atleast for a short period of time. (On the other hand, if the well is equipped with a packer and tubing, it ‘would be possible to use downhole opening and closing devices to reduce the volume of fluid in the wellbore and thus reduce the effects of wellbore fluid compressibility, In mast cases, however, the well is simply opened or closed at the wellhead. As a result, an important condition of the theory is viola- ted. Later on we will study wellbore effects which develop as a consequence of closing the wel at the surface rather than near the sand face. In this chap- ter we will not consider wellbore effects. ‘A pressure buildup testis performed when the well is producing either in the transient or semi-steady state. In both cases, formulation of the pressure buildup equations is accomplished by superposing the drawdown equations (Equations 1-22 and 1-23) with respect to time. Therefore, the principle of superposition with respect to time must be clearly understood in order to understand the theory of pressure buildup testing. PRINCIPLE OF SUPERPOSITION Suppose that after producing a wel at a constant rate for a certain length of time we decide to shut it in for pressure buildup. Intuitively, we expect fluid movement to continue in the reservoir after shutting the well in, The fact that we shut-in the well means that q=0 at the wellhead. We account for the fluid movement which continues in the reservoir after the well is shut-in as follows: we let the well produce indefinitely at the same constant 4, aun al the instant of shutting the wel in we inject in the same well at the same q, and then we sum the pressure drawdown due to producing q and the same pressure data multiplied by ~ 1 and shifted to the instant of closing the well. The result would be the pressure buildup. i | I 54 Well Test Analysis Table 241 ‘The Principle of Superposition ae Time Shifted Drawdown Data Drawdown Data Multiplied By - 1 Summation + =P P ate on, si) (os ys o 2 ° 1 29 1229 2 82 182 3 3313 3913 4 1935 1335. 5 1952 ° 0 1352 ° 1368 - 1229 1 137 7 1978 = 1282 2 98 8 1388 Tiaa 3 73 a 1307 = 1935 3 62 0 1405 = 1952 5 53. " wate = 1968 8 48 2 sa19 = 1978 7 a 3 12s = 1388 8 a7 4 143.0 = 1307 8 33 15 vaa6 = 1408 10 a 2 1458 | 1438 6 22 25 1075 1488 2 w ‘Asan example, Table 2-1 shows the pressure data obtained by producing ‘well at 100 STB/D, To close this well after, say, 5 hrs of production, we mul- tiply these pressure data by — 1 to designate injection, and then shift the data by 5 rs and sum the pressures as shown in Table 2-1 and on Figure 2-1. ‘This is the principle of superposition used to derive the pressure buildup equations. a DERIVATION OF HORNER'S EQUATION, Derivation of the pressure buildup equation, known as Horner's equa- tion, follows exactly the same superposition procedure discussed earlier. Be- fore we go through the derivation, however, let us refer to Figure 2-2 to get acquainted with the following nomenclature: ‘ty = production time prior to shut-in, hrs [At = shut-in time, measured from the instant of shutting the well in, hs. ‘Thus, At = 0 at the instant of closing the well. Fundamentals of Pressure Buildup Testing 55 150 " 50. 109: 303 6 iOS Time, hours Figure 2-1. Graphical ilustration of superposition, + 27] a a aw o 3 7 1012 Time t, hrs Figure 2-2. Relationship between t,t, and At 56 Well Test Analysis ‘t= time measured from the instant of opening the well to produce at a ‘constant q, hrs The fence ten andy sold be mad cle. na gen es the production time ty before shut-in isa fixed number. For example, we may produce the well for 50 hrs before shutting it in, then t, for that test would be 50 his. In another test we may only produce the well for one hour and then shut it infor buildup. Tn this eae f, = Ihe. The time, t, is measured from the instant of opening the well for production. Thus, t > 0. In particu- lar, after the well is shut-in, the following relationship holds: tat tat en "To derive Horner's equation, we recall Equation 1-22, the transient drav- down equation. We then add to it the same equation shifted in time by t, and multiplied by ~ 1 to designate injection. Thus, 162.6 q Baw 80PeA fog t+ 5] (23) 162.6 (-@) Bo tog 4) +5 AES (ED Bet [log (tt) +53] ‘Then AP yy = AP, + APs 162.6 4 Bs tog ¢— tog (t~ EGA EH tog t— log - GI] By substituting Equation 2-1 in the above expression, we get: AP, = 162:6.4 Bot flog (4, + at) - log At] kb i] ox a= fone] J” eo where Pay = P.= Boy the presure as measured inthe wellbore from the stant of shutting the well in t+ At ~ Horner's time at Vv Fundamentals of Pressure Buildup Testing 57 Equation 2-2 is known as Horner's equation. According to this equation, aplot of AP, (or just P.) versuslog(t, + At)/At should give a straight line of slope m. This plot is referred to here as Homer's plot ‘Note that Horner's equation (Equation 2-2) does not include any informa- tion about the skin factor, s, That is because in the process of superposition the term cancels out. Thus the pressure buildup not dependent on the “which implies that in the caae Of positive San Tre, dasa) the pres ‘sure in the wellbore must instantaneously rise, upon shutting the well in, by the amount of APau, before it begins to build up as if there were no skin. In the case of negative skin, the pressure must instantaneously drop by the amount APais before it begins to build up. ‘To obtain an expression for the skin factor, s, we rewrite Equations 1-22 and 2-2 as follows: Py Pagar eg = ™ [og ty + 5] (1-29) +at Py = Pep = m log tt @2) where Pafgy,-9 = flowing bottom-hole pressure at the instant of shutting the well in By subtracting Equation 2-2 from 1-22, we obtain: e +a Par Patgag =m floes, +5 -tog | Ths, Fe Fetou 4 Jog tA tog at Since log —K_, ~ 3.23 40.875 SoG [Poe — Pos kat = 1st [2 “ener _ Jog Kat | ™ om ‘When ty is large as compared to At, then t, + At = ty, and log & + At! ‘be 4%) =0. Also, by choosing At = 1 hr, the above equation becomes: Pete-d jog Kk saa Poas ig a a where Prir = Pospsecie 58 Well Test Analysis Equation 2-3 can also be written as: JAP _ jog —_k Se GH + 3.23) (23) where AP = Pega ~ Pogues es) Asan example, Table 2.2 shows the buildup data a taken fom Table 21, and Figure 2-3 is the resulting Horner's plot of these data. From Figure 2-3 and Table 2-2, we get: m = 17.56 psileycle APgueo = Pi Petgae.g 7 135.2 psi ap. Progarei ” 17 pst ‘Thus, AP, = 121.5 psi For q = 100 STBID, B, = 1.35 RBISTB, «= 0.8 ep , h= 20 ft, 6 = 0.19, C= 15.55 x 10-%psi, and ry » 0.39 ft, 162.6 q By « mh | 168.6 x 100 x 1.35 x 0.8 17.56 x 20 wast 225 50 so 1451 T7536 | piaxdax sas x10 exoa* 2 =19 ‘Thus, APyue = 0.87 ms = 0.87 x 17.56 x 1.9 = 29 pst Note that this well was produced for only 5 hrs before it was shut-in. Intui- tively, therefore, one expects that after shutting the well in for 5 hrs, the initial reservoir pressure, P,, would be reached. However, this is one of the instances in which intuition fails. As the data of Table 2.2 shows, at At = 5 hs, AP,, = 5.3 psi In fact, by Equation 2-2 one can verify that at At = 100 hts, Pyy = 0.37 psi, and after shutting the well in for 1,000 hs Py, = 0.04 psi. The reason, of course, is that buildup involves the superposition of the pressure drawdown due to production and the pressure buildup due to injec- at oon, Pws, psi 3004 3000: 2996. 2992 2988: Fundamentals of Pressure Buildup Testing 59 Table 22 Pressure Buildup Data from Table 2-1 Sepent78 palace (tp + At)/ at Figure 2-8. Horner's plot; data from Table 2-2. 60 Well Tet Analysis tion. Upon shutting the well in, the pressure transients created by the draw- down continue to propagate through the reservoir in much the same way as sound waves continue to propagate in the air after one stops hollering. The ‘transients created by the injection will always lag behind those created by the drawdown. It should not be surprising then that the pressure continues +o build up in the well for days or even for weeks following a short produe- tion period. Theoretically, the transients due to production and injection should meet at infinity after shutting the well in for an infinite period of time. ‘When a pressure buildup test is performed by closing the well during the transient state, then one of the objectives of the testis to determine P,, the initial reservoir pressure within the drainage area of the well. We have seen that it takes a very long shut-in time for the pressure to approach P,, and it is not economically feasible to keep a well shut in just to obtain P,. However, Equation 2-2 provides us with the means of estimating P, without having to shut the well in for a long period of time. Note that: beat ‘Thus, as At becomes large, W/At approaches zero. And since log(l) = 0, APs, = 0, 0. F, = Puy Therelore, to find P,, itis necessary to plot Py, ver sus Horner's time as shown in Figure 2-3 and extend the straight line to Horner's time equal to 1 to obtain P. It should be emphasized that this procedure is only valid when the flow is radial, which means that the reservoir is homogeneous with respect to all description parameters and the fluid contained init s slightly compressible and has constant propertics, and the well was shut-in while producing in ‘the transient state, IF any ofthese conditions is not met, then extension ofthe straight ine on Horner's plot will not yield P, For example, ifthe reservoirs naturally fractured, then extension of the straight lineon Horner’ plot may rot necessarily yield P, unless the well has been shut infor a very long time, almost twice the producing time. Also, if at closing, the well was producing in the semi-steady tate then extrapolation of the straight Line will not yield B, DERIVATION OF MDH EQUATIO ‘There is another buildup equation known as the MDH (Miller, Dyes, and ‘Hutchinson, 1950) equation which represents the pressure buildup when the well is closed while producing in the semi-steady state, To derive the MDH Fundamentals of Pressure Buildup Testing 61 equation, we superpose the semi-steady state drawdown equation (Equa: tion 1-28) and the transient drawdown equation (Equation 1-22) as follows: AP, = mat + Cpe + 0.87 ms (1.23) and AP; =~ m (log (t=) + 3] ‘Note that the transient drawdown equation is multiplied by ~1 to designate Injection, and it is shifted in time by the duration of the producing time t, in accordance with the principle of superposition. Note also that although the well was producing in the semi-steady state before shut-in, the injection be- sins at the instant the well is closed. Therefore, the injection must take place in the transient state. Eventually, of course, the injection will proceed into the semi-steady state and the superposition then must only involve the semi- steady state equation By summing the previous two equations, we get AP, = AP; + APs = tye t ~ m log (tt) + Ce + 0.87 msm where APy, =P, — Poe Since t= t, + at AP y= ~m log At + sips (fp + 48) + Cpu ~ m flog —* = and i eG By evaluating Equation 1.28 at t = ty, Le, at At = 0, we obtain: My ty + Cpe + 0.87 ms By eliminating P, from the preceding two equations and cancelling like terms, we get Pag ~ Pes dgueog 7 BIOLAL~ tgg At + m5 es) [Equation 25 is known as the MDH equation. For small values oft, a plot ‘of Pe, Versus Aton semi-log graph paper should gave a straight lin of slope m. By determining AP, from the pt, it would be possible to calulate the skin factor, , by Equation 2.3. As At inereases, however, the term Ms At i Equation 2-5 becomes significant, and the semi-log straight line gradually 62 Well Test Analysis changes into a curve concave downward. When At becomes equal t0 tps Pz ‘becomes equal to B the average reservoir ‘Tosee these results, consider the data of ‘Table I-1. We know that the semi- steady states reached at t = 40 brs, and tips = 0.2429 psi/hr. Let us assume that the well is produced for 80 hus and then closed for 60 hrs. We can per- form the superposition as shown in Table 2-3 and plot the results on an MDH plot as shown in Figure 2-4, To verify that for At © tye, APy, is indeed equal to (P, ~ P) = 19.44 pai, swe neod to apply material balance. We already know from Chapter 1 that YV, of the 80-acre drainage area is equal to 1.489 x 10° RB. Then, (q/24) t) By = GV, (Pi - PY ‘or (100/24) x 80 x 1.35 = (15.55 x 10-8) (1.489 x 10°) (P, - P) ‘Then (P; - B) = 19.44 psi In general, if the well is producing in the semi-steady state and then closed for At & tym, the observed shut-in pressure, Pye, will equal the aver- age reservoir pressure, P ‘This result seems rather strange, because it shows that if a well has been producing in the semi-steady state for any length of time and then closed, the pressure will stabilize at P as soon as At becomes equal t0 ty, Whereas ‘we have seen that it takes an infinite amount of time for the pressure to reach P; when the well is closed while producing in the transient state fol- lowing a short produetion period. However, during the semi-steady state Table 2-3 Buildup Test While the Well Is Producing ih the Semi-Steady State 1 BP, =P Pas Ps ‘at toe) si) os on 80, 13022 ° a 19048 = 9259 4 2 sso 9762 33.09 2 83 19095 S007 3024 3 2 131.19 = 10200 28.29 4 8 19143 = 10460 28.83 5 0 1926 = 109.89 zen 10 100 135.08 S168 19.00 20 120 139.98 = 12050 19.46 0 190 4237 = 12293 19.48 50 +40 14479 = 125.38 19.46 60 Fundamentals of Pressure Buildup Testing 63 16 20: 24 28 (aope = 1758 pace APws, 32 36: 40: 10 10* ‘Shut-in Time, At, hrs Figure 2-4. MDH plot; data from Tablo 2-3. there are no outward propagating transients, the whole drainage area is ‘equally drained, and P ~ Pyremains constant, Asa result, the stabilization time is finite and is equal t0 tp: ‘Suppose that onthe basis of a presure buildup test of our wel, following avery short transient flow period and a long shut-in period, we determined that P, = 3,000 psi. We then put the well on production atthe rate of 100 STBID for a period of 3 months (t, = 2,160 hrs) before we closed it for a second pressure buildup that yielded the data shown in Table 2-4. Tt is re- aquired to determine k, s, and B ‘We know that sinee the well was producing in the semi-steady state before shut-in, we could interpret the test by the MDH method, calculate k and s (igure 2-4) and calculate P by material balance. However, Table 2-4 in- ‘cludes a tabulation of Horner's time so we might try a Horner's plot, even though the well was closed while producing in the semi-steady state, and Homer's equation was derived by superposing the transient drawdown equation (Equation 1-28). 64 Wall Tost Analysis Table 24 Pressure Buildup Test at ty = 2,160 hrs, P, = 9,000 psi at Pa (ore) oH (+ stat G00 2088 = 010 24093 zreo10 015 pases ‘ago 02 2ades togor0 025 2403 3.41.0 030 2aars 7200 040 2408 54010 os 24895 S281 080 24850 27010 400 2486.7 2610 150 2588 vast 200 24987 10st 250 2653 2050 00 Zao 719 00 2805 Seo 500 2488.0 #0 650 24098 x03, 5.00 24708 ano 0.00 2are0 2170 1500, 2ars9 ‘aso 20.01 2araa 4090 25.00 Zanes ors 3000 2arse 730 40.00 Bars so 50.00 2ars2 ue 100.00 Bare 2s ‘Figure 2-5 is the Horner's plot ofthe data of Table 2-4, It is obvious that this plot is remarkably similar to the MDH plot (Figure 24). In fact, from the slope ofthe straight line on Horner's plot and the value of AP, we ob- ‘ain the same values of k and s which we obtained from the MDH plot. By applying material balance, we can easily verify that P = 2,473.1 pst By ex- ‘apolating the straight lin in Figure 25 to the Horner's time of one, we ‘obtain a pressure of 2,515.3 psi. This pressure is usually given the symbol, P+, and is known as P star. Obviously, the value of P*, in this particular example, i less than P; and greater than P “The results obtained from this exercise cause us to raise the following ‘questions: 1. Why ist that Horner's plot is applicable to the case of closing the well in the semi-steady state even though Horner's equation was derived by superposing only the transient drawdown equation? Fundamentals of Pressure Buildup Testing 65 2520. 2500) 2480) 2420. 2400) Peay ee Figure 2-5, Horner's plot; data from Table 2-4 2. What is the relationship between P* and each of P, and PP 3. Isit posible to utilize the relationships between P*, P,, and Ptodeter- mine P without having to lose the well for @ long period of time ? ‘To respond to these questions, let us frst substitute in Equation 1-22, the transient drawdown equation, and calculate Py: at the instant of closing the wel, asif the well were producing in the transient state for the entire 2,160 hours of production. Thus, P,— Puc = m [log t+ 5] (2-29) 3,000 ~ Pye = 17.56 [log 2,160 + 5.26] from which, (Pus)ranint = 2.849 psi Now, Py ~ (Peieaniens = 3,000 ~ 2,849 = 151 psi and from Table 2-4, P, ~ Pag. ~ 9,000 ~ 2,364.6 = 635.4 psi Since, P, - P* = 3,000 ~ 2,515.3 = 484.7 psi 6G Well ast Analysis ‘Then by ignoring roundoff errors, the above calculations lead to the follow- {ng conclusions: = PY = AP — Prfgasag) ~ (Pi Paden} Thus, P,~ P* = (QPpa)oa-0 ~ (APranieeat-0 es) Figure 2-6 is a plot of the transient drawdown as evaluated by Equation 1.22 and the semi-steady state drawdown as copied from the data of Table 1-1 and our knowledge that mys = 0.2429 pail. "This plot suggests thatthe semi-steady state pressure drop, APj., can be represented as follows: AP = APranstent + Z(t) where AP pe = P,~ Prins te APrrandieat ™ Pi ~ (Pot)iracsioet Z(t) = the difference between AP, and APyanient at any given time t > be AP=(PI-Put) psi 0600 Bd0«1000-—«1200 Producing Time, tp (hrs) Figure 2-6. Comparison between actual and alliransiont drawdown data, Fundamentals of Presure Buildup Testing 67 Thus AP, = m [log t + 3] + Z(t) oF AP ya = m [log t +3 + Y(9] en where Y(0) = Z(ty/m It isnot difficult to verify that for a small At, Y(tp + At) = Y(t). For examn- ple, at t, = 2,160 hrs, Y(2,160) = (635.4 ~ 151)/17.56 = 37.6, and at At = 20 his, ¥(2,180) = (640.3 ~ 151)/17.56 = 27.9. Also, for a small at, Y(At) = 0. For example, Z(2) = SP pgp ~ APramin, By substituting in Equations 1-23 and 1-92 we obtain: (20) = 116.5 ~ 115.2 3 psi 3117.56 =0.07 and, ¥(20) ‘We can now derive another expression for the buildup when the well is closed while producing at the semi-steady state, by superposing Equation 2-7 as follows: (APp)y = m flog +3 + (9) en (Pade = ~ mlog (t=) + +¥ (C49) ‘The sum is given by: PaRanm og #L AY a0 -Y 0] a Since for a small At, Y(ty + At) = Y(t), and Y(at) = 0, then P teat, Poem fos tatey ea] es) Eiquation 2-9 shows that a plot of Py, versus Homers time on semtlog graph paper should be a straight line of slope m. As At increases, however, the approximation Y(t + At) = Y(t) becomes invalid, and the straight line 6$ Wall Tet Analysis sradually changes into a curve in the same way as it does on the MDH plot Ultimately Py, atains the value of F By Equation 2.9 we can oaleulate P* at Horner's time of unity as follows: Pt =P,- mY) = 8,000 ~ 17.56 x 97.6 = 2,515.3 psi which is essentially the same value obtained graphically on Figure 2.5. To describe the straight line portion on Horner's plot, we write: Pay = P* — m log (ty + Aty/at (210) We will make use of this equation later, to determine the average reservoir pressure, B ‘We have now seen that Horner's plot is valid whether the wel is closed in the transient or semi-steady state. We have also seen that the MDH equation is-valid when the well is shut-in while producing in the semi-steady state. It ‘would be of interest to find out if the MDH equation is also valid when the well is closed during the transient state. For this purpose, we rewrite Horner's equation (Equation 2-2) as follows: P,— Py, = m [log (tp + AU) ~ log At) In the event that ty is much larger than At, then (ty + At) ~ t,, and the equation becomes: = Pos = m [log t ~ log At] But, at at 0, the transient drawdown equation is written as follows: P,~ Poigateo ™ m [log ty + 5] Fy climinatingP, from the previous two equations, we get Pas ~ Patgar ao = m [og At + 5] en) Equation 2-1 shows that during the transient state, the MDH plot is valid sslongas i, + At) = ty Le, when tislarge compared to At, From a prac- tral point of view, itis sufficient that ty = 5 Bt. In conclusion, Horner’ plot is always valid whether we close the wel ‘transient or semi-steady state. If the well i closed in the semi-steady state for Fundamentals of Pressure Buildup Testing large At, then the semi-log straight line will change into a curve conc: downward, The MDH plots valid ifthe wel i closed in the transient st providing that t, > 5 At. The MDH plot is valid if the wll is closed in semi-steady state, and inthis case the semi-log straight line will change a curve, concave downward, as At becomes large. For At = th, the pressure will be equal to B Wie summarize our conclusions as follows: 1. Horner's method: a. well closed in the transient state: t+ at] a », well closed in the semi-steady state, and AC is small: n-Paemfogtetate yg] e 2. MDH method: a. well closed in the semfsteady state Pas ~ Prigarag = 1 log At ~ my At +m 5 e@ b, well closed in the transient state and t, > 5 At: Poa Poigar eo =m [log At + 5] @ 3, Pressure relationships: 8. - Pt = AP gy @ b.Pt ~m¥ () (2-0 evaluated at At = 4, Skin Factor: In all cases the skin factor is given by: kK ~ lon eet 8.3 @ i= Pe s= 1151 10 Well Test Analysis DETERMINATION OF P Reservoir performance monitoring by material balance requires input of periodically determined average pressure of the enti reservoir together ‘with cumulative production and injection data. For each time period, the average pressure of the entire reservoir is determined by running buildup tests on a selected number of wells. Each buildup test yields the average res- exvoir pressure, B within the drainage area ofthe tested well. The weighted average ofthese P values constitutes the average pressure ofthe entire reser- voir for the specific time period during which the buildup tests were made. ‘Therefore, determination of P is one of the main objectives of a pressure buildup test - “To determine P from a buildup test, we must seek a relationship between P and P*, since P* is readily determined on Horners plot. Recalling Equation 1-7, —p- ab AE BoPe gat 7) Aoh where Vo S15? Ac drainage area, sa ft Equation 1-7 can be written as: 2.64 x 10-* k OuGA ty (213) Let us define the dimensionless producing time, tyna, as follows: 2.64 x 10> k tog BSE, @13) ‘Then, Equation 212 is written as P,- B= 5.45 m toa eu) Since by Equation 2-9, P.-Pt=mY () (215) ‘Then by eliminating P, from Equations 2-14 and 2-15, we get: Fundamentals of Pressure Buildup Testing TL PtP 2.903 HF = 12.55 tuo, ~ 2.909 ¥(4) or Pp = 4r toa ~ 2.803 YG @16) where Pp = dimensionless pressure PtP = 2.303 ere tee 06 qB, wkh Equation 2-16 is dimensionless but not unitless. The usual set of oilfield units (STBID, ep, RBID, md, ft, hr, psi) must be used in evaluating Pp and ‘yon, The advantage of having Equation 2-16 in « dimensionless form is that ifcan be used for any set of reservoir description parameters. This will be evident in the following paragraphs From a practical point of view, a plot of Pp versus ty is quite useful. We could generate this plot by substituting the reservoir parameters of our well in Equation 2-16. Table 2-5 displays the results. As an example, for tyoa = 0.05, tp = 19.7 hrs, which is less than tya. Thus, Y(19.7) = 0, since from the definition of Z(}), Z(t) = O during the transient state, which im- plies that Y(t,) = 0 for ty < ta. Substituting into Equation 2-16, we get: Py = 4 rx 0.05 = 0.698, When t > tur we frst calculate Y(t) and then substitute in Equation 2-16, For example, at tyoy = J, ty = 3941 hrs. Then, (AP pesos = (APp ove + 0.2429 (894.1 = 40) 1205 $86.01 = 206.51 (BP rangi ~ 17.56 [log (294.1) + 5.258} = 157.91 psi Z (394.1) = 206.51 ~ 137.91 68.6 pst ¥ (@04.1) = 68.6/17.56 2391 Substituting into Equation 2-16, Pp = 3.57 72 Well Tet Analysis. Table 25 Calculation of Po Versus tox . 2) ey si) ta Po 0.07 7 396 0.90 0.00 033 0.02 788, 0.00 0.00 02s. 0.08 182 0.00, 0.00 038, 0.04 3576 0.00 0.00 050 005, 3970 0.00 0.00 oss. 0.06 2365 0.00 0.00 07s. 007 2789 0.00 0.00 08s, 008 3153 9.00 0.00 101 008 3547 0.00 0.00 413 010 saat 0.00 0.00 126 020 7.82 429 O28 195, 030 11025 1078 ost 238 040 157.64 1815 103 265 050 197.05 25.03 148. 287 280 23646 aaa 198. 305 0.70 21587 #260 243 321 080 31528, 5116 2o1 34 090 35669 59.83 341 346 4.00 994.10 18.60 31 357 2.00 788.20 159.08 3.08 428 3.00 418230 251.68 1433 489 4.00 1576.40 aes21 1986 499 5.00 197050 430.24 2501 523 6.00 2.364.60 sa087 20.8 542 7.90 756.69, 26.12 387 559 8.00 152.79 7283 4116 574 9.00 Sist6 80 21786 4656 587 10.00 3.040.99 o1288 S197 598 ‘Figure 27 isa plot of the data of Table 2-5 on semi-log graph paper. This plot consists of a curve followed by a straight line of slope equal to 2.303. Note thatthe straight lin is associated with the high values of tana ie. the hhigh values oft. This observation leads to the conclusion that the straight line portion of the curve is associated with the semi-steady state, Therefore, the beginning of the semi-steady state should be determined by the begin ning of the straight line portion of the plot. ‘To find the beginning of the straight line in Figure 2-7, we rewrite Equa- tion 2-16 for the transient state. Fundamentals of Pressure Buildup Tasting 73 D=2.303(Pavg-Pr)/m 0. 107 10" 1 10 tpDA Figure 2-7. Approximate MBH curve for a circular or square drainage area. Pym 4m hon Py Then, Fon ‘The semi-steady state on Figure 2-7 can be represented by: Pp = 2.308 log typ, + Constant (217) ap 1 Htyon toa At the beginning of the semi-log straight line on Figure 2-7, the two slopes ‘must bo equal. Thus, 1 nan TA Well Test Analysis euler) 4. “which is the same result obtained earlier (Equation 1-87) when we discussed ‘the radius of investigation. Returning to Equation 2-17, Po = 2.908 log tung + constant ery Equation 2-17 can be written as: Pp = 2.903 log (Ca toon) e18) where log Cy = constant From Table 25, Pp = 3.57 when typ, = 1. However, because of small roundoff errors let us accept that Pp = 3.45 when tgo, = 1. Then, Cy = 102.90 Z = 316 ‘where C, is known as the shape factor. Thus, the shape factor of a circular rainage area fs equal to 31.6. Equation 2-18 could be written as: peor Jog (Ca tyoa) (2-18) ‘Recalling Equation 2-10: Bi Pers jog et At (10) at minating P* from Equations 2-18 and 2-10, we get: PaPre a tog ttt - tog Cx tn @19) Fundamentals of Pressure Buildup Testing 75 From Equation 2:19, P = Py, when + (QD, ® Gabe = Ca on (2-90) Because during the semi-steady state the approximation t, + At = ty is valid, then, (Atp= (ay ee Trtyoa Equations 2-20 and 2-21 and Figure 2-7 provide us with three interrelated methods of estimating Ps 1, Determine k, s, and P* from Horner's plot and then calculate tn. From Figure 27, find the corresponding Pp and solve for P. This ‘method will be referred to here as the MBH (Matthews, Brons, Haze- broek) method. 2. If the MDH plot is applicable, determine k and s and then calculate tong. Use Equation 2-21 to calculate (At)y then locate P on the extra~ polated portion of the MDH straight line. This method will be referred to here as the Dietz, method. - 3, Caleulate Cx tana and locate this value on Horner’ time axis. Then P ‘would be equal to the corresponding P,.,on the extrapolated portion of ‘Horner's straight line. Clearly, this method is merely an application of the Dietz method to Horners plot. As an illustration of the MBH method, let us refer to Horner's plot (Figure 2.5). We know how to calculate k from the slope, m. We also know how to calculate s. However, itis very important to remember that Py, must be determined graphically on the semi-log straight line, not from the given data, This s becanse real field data could considerably deviate from the the- oretically correct straight line, With k determined to be 50 md, and our Inowledge that the producing time was 2,160 hrs, toa is found equal to 5.48. From Figure 9-7, we find that the corresponding Pp is approximately equal to 5.4. Thus, pe -P Pp = 2.903 54 = 2.308 & 17.56 P= 2,471.3 psi 16 Well Test Analysis Which is very close to the value of 2,473.1 psi found earlier by material bal- ‘As an illustration of the Dietz method, we refer to the MDH plot (Figure 2-4), From the slope and AP, we can determine k and sand then calculate toa: (Fort = BO hus, to, = 0.2.) By Equation 2-21 wo caleulate At, as fol lows: ° oe “6” 316x000 = 12.68 hrs From Figure 2-4, the corresponding AP,, = 20 psi. Thus, (Paap = Pi~ F 20 psi For the given value of P, of 3,000 psi, B would be equal to 2,980 psi, which is very close to the value obtained by material balance. “Application of the Dietz method to Horner’ plot may be illustrated as From Homer's plot (Figure 2-5) the value of P is approximately 2,476 ps — Horner's plot (Figure 2-5) is based on a t= 2,160 hrs” Suppose that we repeat this plot on the basis of t, = 60 hrs. Thus, Py, remains the same as given in Table 2-4 and the only change would be in the values of Horner's time, For example, for At = 10 hrs, we ealeulate Horner's time at 7 instead. of 217 as given in Table 2-4. The question now is: Would this erroneous value of ty affect our conelusions in regard to k, s, and P? “To answer this question, we refer to Figure 2.8 which is a Horners plot of, the same P,, data of Table 2-4 and the new values of Horner's time, Iti evident that the two plots are quite similar. In fact, we still got the same slope, m, and the same Py, which means that Figure 2-8 gives the same k ‘and s obtained earlier on the basis of Figure 25. The only difference be- toveen Figures 25 and 28 is thatthe plot onthe latter figure has been later- ally displaced with respect to the plot on the former figure. As a result, the value of P* has changed from 2,515.9 psi on Figure 9-5 ta 9,487.8 pt on Figure 2-8. Whether this has any effect on P can be seen by repeating the calculation according to MBH method. For t, = 60 hrs, equal to toa = 0.152 Fundamentals of Pressure Buildup Testing 77 2520: ae ; 2480. _ Basso ; 2440- we a Bade Figure 2-8. Horner's plot; data trom Table 2-4 with ty = 69 hrs. rom Figure 27, the corresponding Pp is approximately equal to 1.6, Sol ing for B, we get: P= 2,475.6 psi which is esentially the same value obtained earlier. ‘We have thus verified a very important conclusion: if the well is produe- {ng in the semt-steady state before shut-in, then t, can be assigned any value greater than tye. In our case, ty = 40 hrs. Therefore, since the chosen ty of (60 hrs greater than 40 hrs, we would obtain exactly the same results which wwe obtained with t, = 2,160 hrs. If, however, the chosen ty is less than ty, then the resulting Horner's plot ‘will be distorted. Figure 2-9 is the Horner's plot of the data of Table 2-4 with t, = 5 hrs. Note the break in the slope which could be misinterpreted as an Indication of a linear boundary. We have been discussing the caleulation of P for the ease of a eylindrical Grainage area. Now we want to discuss the determination of P for other shapes of drainage. But before we do that, we must enquire about the reser- voir conditions which could result in shapes of drainage that aro substan- tially different from circular, and we should also become familiar with the ‘means of estimating shapes of drainage. 78 Well Test Analysis 2520. 2500. Hi 7 Caaeee io meee ee To" p te vaysat Figure 2-9. Horner's plot; data from Table 2-4 with t, = 5 hrs. If the reservoir is homogeneous with respect to all description parameters and the wells are drilled on a regular pattern, e.g., Figure 1-1, then it is reasonable to expect that the shapes and sizes of the drainage areas be deter- ‘ined by the relative production rates. For example, consider well A in Fig- ure 2-10, the well pattern surrounding it, and the given produetion rates of the wells. The drainage area of well A can be outlined by multiplying the production ratio by the distance between well A and the neighboring well, Thus, according to this method, the drainage boundary between wells A and B is calculated as follows: 100 da eae Geran ai [ 288.6» r= Ky +A) {988.9 nC -En (-28gec.4] +m og (223) {tis not difficult to verily that for the closest image to our well and for At < 10 hours, the expression: ey - enc g kat ‘and that itis negligible for 10 < At < 20 hrs. Thus, for practical purposes Equation 2-23 could be written as; " Le _ pe a8 a Beale 948 6 «Ceri rere smtghal fn ges nfo eo 82 Well Test Analysis _ Since P= Pe at 88, abate Fp) | 8 ee eh Ot Ke pa. 4B, 06) By eliminating P, from the above two equations, we get: oo ee Dean ER WaGRRAY Ge Agh 5615 Since V, 7 pe-P P and Po~ =e qe, kh the Fo=4 tot Sti [ES] 225 ‘To generate the MBH curve for the square drainage, we assume a value for ty, calculate x = 11(4 A tyos) and the corresponding Bi (~) for each image and then sum the results, The calculations are carried out on the com- puter by Program 2-1, written in GWBASIC, which makes use of Program LL ‘Although the number of images is theoretically infinite, the actual num- ber is small (Figure 2-11) since Ei (—x) =0 for x = 10. “Table 2-6 shows the Pp data as given by Earlougher et al. and some data “obtained by Program 2-1 for comparison. Figure 2-12 is a plot of data from Earlougher at al. on somiclog graph paper. ‘Teshould be noted that Program 2-1 in ts present form is not suitable for calculating Pp for values of tan > 1. The program can easily be modified, however, to include a larger number of images. | i | | | | Fundamentals of Pressure Buildup Testing 83 Program 2-1 Calculation of MBH ‘Curve for a Square Drainage 10 ‘380 a0 CCLEARDEFDBL AKOZK. AREA = 3484500871 ~ 9.14 ‘SUM = O:INPUT "TPOA Xie = 1120058 31993,808 FORL= 17011 FORI=17011 Xba Xt + 1856.76 IF ABS (x) < 11 THEN IF ABS (¥I) < 11 THEN GOTO 320 F2= XI" 2+ YI" 2,TERM = 4*AREAYTPDAX » ROITERM IF X> 101 THEN G = 01:G0T0 220 A=KIF A>1.5 THEN GOTO 200 Yr = LOG(I0.5) - LOGIA).Y2 = O1:Y8 = O1 FOR N= 17031 GOSUB 380 Y2 = Y2 + ((~ 1)" N)*((10.5) - NVINGAMMA) Y3=¥9 + (= 1)" NMA” NYIN“GAMMA) NEXT N Yan Vi + ¥2—Y8G = ~ Ye-Goro 810 DEF FNFOO = EXP(~ XVKB =A + 10.5 Da (B-AViKK - )x=A, FOR J = 1 TO KKGYY() = FNFOOX = X + D:NEXT J (00D = O:FORJ = 3 TO KK - 2STEP 2 ‘OD = ODD + 2-¥¥4) NEXT ev-0 FORJ = 270 kk -1 STEP 2 EV=EV 4 4¥YYU) NeXT = ~ Di9s(v¥{t) + EV + 000 + YYIKK) SUM = SUM +0. NEXT Xt= — 1120058 Vie vi— 1006.76 NEXT PD = a¥PISTPOA + SUM RINT "TPDA = ““TPDAPRINT "PO = X=Net vexe2 G2 = SORR*PIY)*EXPLY*LOGKY + (1 — 1KB0"Y>YHIEI2NY) = Y) GAMIA = CBIR OC 4) RETURN END 84 Well Tost Analysis ble 2-6 Pp Versus toa fora Square Drainage, Well at Center toon Pot Pot tron Po’ Po oor oxear oa Bow cos Oats os Bama cm baste oa Bate Coes ba oata2 cs rma Oost os aves co aoe Oy Sams ase cos aera oe a0 cos nats os azo cor baser to Same gare coo Osser zo faa oo sme ico atee oro hasae ze ac Same cis tose 1000 aH oz taete 7 Fon Boag oT Fron Pram ° 5 ‘ gs 2 1 = 10, 1 10 a a> ° (t+ avsat Figure 2-12. Plot of Earlougher's data, Tablo 26. Fundamentals of Pressure Buildup Testing 85 ‘The data by Earlougher et al. show that between 1.0 < tio, = 10, the slope is equal to 2.903. Extension of this slope to tpn = O.1 gives a value of Pp = (3.4302 ~ 2.903) = 1.1972. Thus, the beginning of the straight line on ‘the MBH curve (Figure 2-12) must occur at 0.09 < tn, < 0.1. Through the use of Program 2-1 it can be verified that the straight line begins at 304 = 0.0973, which for ease of reference should be considered 0.10. The Deginning of the semi-steady state is calculated as follows: (opipa = O41 or 284X104 k Ou (re) tye = OL where 1, = equivalent external radius of drainage, ft Then, te BOOSH fey Recalling Equation 2-18, Pp = 2.90258 log (Cx tox) 18) From the data of Table 26, at ty = 1, Pp = 9.4902, thus, Cy = 1o-omn es or Cy = 30.88 (2-26) ‘To derive the MBH curve for the case of a 2 by 1 rectangular drainage with the well at one-eighth of the length away from side, we first place the {mages as shown in Figure 2-13, and then modify Program 2-1 as shown in Program 2-2. The results are given in Table 2-7. These results should be com- pared with the appropriate MBH curve on Figure 2-14. Comparison between the data of Table 2-7 and the MBH curve on Figure 2.14 shows a close agreement except when tng = 4. This indicates that for ‘ops values greater than 3, Program 2-2 has to be modified to include more images. (tes comin on og 8) 8G Well Test Analysis Fundamentals of Pressure Buildup Testing 87 Program 2-2 Program 2.2 Derivation of MBH Curve for a 2 by 1 Rectangular Drainage with the Well ‘continued ‘at One-eighth of the Length from the Side 40 CLEARDEFDBL A— 1,0 - ZKK = 2001:D1M YK ‘370 PRINT “TPOA = “TPDA."PD = °PD,"SUM = ";SUM:GOTO 490 11 REM A PROGRAM TO CALCULATE PO FOR A2 BY + RECTANGLE WITH WELL 300 X=NS4 42 REM AT ONE-EIGHTH OF LENGTH FROM SIDE 900 YoX42 20 AREA = o4a40004-P1 = 3.1415906¢ 400 G2 = SORI2+PIY)*EXP(Y*LOGCY) + (1 ~ UIBO*Y*YHNLI24Y) — 90 SUM = O:INEUT “TPOA = "sTPDA 410. GANIMA = G2I0*0K + 1) 40 Xx1= ~ 10560! ‘420 RETURN 150 ¥i~ 145201 430 END 60 FOR L= 170 10 7 FORI=17021 20 Yi~Yi- 1320 90 IF ABS (X) < 1! THEN IF ABS (¥) < 1! THEN GOTO 320 100 Re= x1" 2+ VI" 2TERM = 4*AREA*TPDAX = ROTTERM 140 IFX > 101 THEN G = OLGOTO a20 12) A= Xi A > 1.5 THEN GOTO 200 ee : os 190 Y1 = LOG(10.5) ~ LOGIA}Y2 = O13 = 01 fearon Heiiito s! meme oo a ed 150 GOSUB 380 weet ae 160 Y2=Y2 + (— 1)” N)*(10.5) ~ NVIN*GANMEA) 170 Y= 9+ (= 1)“ NA” NYIN-GAMMA) roe 6 480 NEXT N 190 Y4= Yi 4 Y2- Yas = - Ys-GoTO 310 wee 8 200 OF FNFDX) = EXP XIKB =A 105 20 Dee ANNR NSA | aries 220 FOR J = 1 TO KKYY() = FNFQQR = X + DINEXT J ames 220 O00 O1FORI= 8 TOKK~2STEP 2 a 240 ODD = ODD + 2*¥¥(J) . sof sof 250 NEXT eee bo eV=0 : : 270 FOR S=27OKK—1 STEP? ei ad bao EV = EV YY) ee ee 350 NEXT S00 G = ~ DIS*(VYN) + EV + 000 + YY) Figure 2-13. images ofa wellinside a rectangle, Copyright © 1964 SPE, Tans, 0 SuM=SuM re SIE, 198i 9 NET 0 Vi= 14500 St IF LF THEN XI = x1 660 Be IPL = THEN X= xl +4820 933 IF L = $ THEN XI = XI + 660 Table 2-7 St (EL = A THEN I= x1 4829 urput From Program 2:2 5 (FL =O THEN XI = xs 600, eee 0 (FL = THEN XI~ x1 4620 i ben Pe en Pe 7 (FL 7 THEN XI = x1 660, | om aces = anes 808 IF L = 8 THEN XI ~ XI ¥ 4620 | 0.02 0.15951 1.00 = 0.51603 390 IF =O THEN X1 = XI + 680 0.08 9.00132 2.00 0.13165, 960 NEXT L 0.10 = 0.09999 3.00 0.54894, 280 PD 4+ TDA + 0.20 0.51916 4.00 0.96141 a 85 Well Test Analysis — _ oe ‘Dinansonigs Proton Time, ox Figure 2-14. MBH curves for citferent wal locations in a 2:1 rectangular drain- age area, Copyright © 1977 SPE, Earlougher, 1977. (continue rom page 85) ‘Now that we know how to generate MBH curves, the next step isto learn how to generate the pressure buildup curve for a given reservotr and a given ‘drainage area. For this purpose we recall Equations 2-23 and 2-25: ee =Reabe Sal- 948 6 x Cot? kG +a) 9486 1 Coif] on tog + At eatvag|| ieeeeagat a # Prdrtot Dy [st (225) Equation 2-23 could be written as: # aoe aa -aatesl-E ye oem smog 2 At ea Fundamentals of Presture Buildup Testing 89 2.64 x 10-* where (A9aa= a 2.64 x 10-* ke (ot ae SOK (+H axetil y # and Sum — a (-pp8ae| ‘Then Equation 2-27 can be written: tat n t = Ppp = gh [Sum ~ Sum] +m log e+ At 7 Pr = 5-apg (Sum, ~ Sum,] + m log 5 (2-28) ter sum = = wi ae ‘To obtain Sum:, we simply substitute in Equation 2-25 the value of (tp + Atos instead of tgp, then refer to the appropriate MBH curve to find te corresponding Pp. By a similar procedure we can calculate Sums ‘As an example, suppose that somehow we determined that our well is Io- cated one-eighth of the length away from the side in a 2 by 1 rectangular ‘area and that we produced the well for 60 brs at a constant q of 100 STB/D before shutting it in for buildup. We want to find Py, at At ~ 5 hrs, given that P, = 3,000 pl, and the following reservoir and fluid parameters: = 0.12, «= 0.8 ep, B, = 1.35 RB/STB, O, = 15.55 x 10-"ipsi, k= 50 md, h = 20 ft, and ry = 0.33 ft. Then, 0.12 x 0.8 x 15.55 x 10-® (80 x 43,560) + AB on = (00 +5) = 0.00254 (65) = 0.16493, From the appropriate MBH curve of Figure 2-14, the corresponding Pp is, approximately equal to ~ 0.35. Substituting these values in Equation 2-25, we get = 0.95 = 4 0.16493 ~ Sum, ‘Thus, Sum; = 2.42 Also, (At)py = 0.00254 x 5 = 0.0127 90 Will st Analysis ‘From the same MBH curve, the corresponding Pp is approximately equal to 0.16. Thos, - 016 x 0.0127 ~ Sumy or Sumy = ~ 0.00041 Substituting in Equation 2-98, we get 17.56. +s = Pyy = 25819. 49 — (—0,00041)] + 17.56 lo 3,000 ~ Py = FEE L2 42 — ( y+ ee (or Py = 2,961.99 psi ‘We can also calculate Sum, and Sum: by Program 2-2 and then manually cealeulate Pye, or Write a new program that uses Program 2-2 as a subrou- tine. ‘Through the use of Program 2-2, the buildup data shown in Table 2-8 ‘were obtained. These data are plotted as Horner's plot, Figure 2-15, Figure 2-15 is rather strange. It displays three straight line segments of Aifferent slopes. The first straight line, which occurs for a very short time after shut-in, has a slope that is very close to the usual slope of 17.56 psi! Table 28 Pressure Buildup Data, t) = 60 hrs at Pee Homers at a on wai, Time ney (os ° 20592 Pwd SStSCSC*~S~*S corer 28204 8601 2 29670 6 oss 2.9026 rat 8 2.968:3 5 025 2.9408 2a 20 29608 ‘ 050 2048.1 ‘21 20 pares 3 19 2981.1 et 50 29757 22 20 2.3560 3 100 2080.4 60 30 2.9588 2 120 29815 150 40 23605 16 150 2.9827 140 50 2,961.8 13 300 29852 120 60 2.0629 " 70 20699 997 80 20687 85 90 2.9654 787 10.0 2,966.0 7 no 2.9685 eas, Fundamentals of Presure Buildup Testing 91 supe = 5002 payotse 23804 * 2965. 2920: : 7 70 TO" v0” 70° To" (o sansa Figure 2-15. Horners plot; data from Table 28, cycle. The second slope of 15.62 psileycle is somewhat close to the correct slope, but a strict interpretation ofthis slope would give a permeability of 56 id instead of the usual 50 md. The third slope is more than twice the cor- rect slope of 17.56 psi/eyele. By material balance, P,~P 5 100 x 1.35, Bx TAB x 1x 15S IOS * = 2,985.4 psi ‘calculate P by the MBH method, we fist calculate tyox: yon = 0.00254 (60) = 0.1524 92 Wall Test Analysis ‘The corresponding Pp from the appropriate curve on Figure 2-14 is approxi- mately equal to ~ 0.3, Thus, a PtP 2,988.3 ~P oe 17.56 B= 2,000.6 psi ‘Thus there isa difference of about 5 psi between the true as determined by ‘material balance and P as determined by MBH method on the basis of the Pt shown on Figure 2-15. On the other hand, if we extend the straight line of slope equal to 17.46 psileycle, shown in Figure 2-15, to Horner's time ‘equal to 1 we will obtain P* = 2,982.9 psi, from which P by the MBH ‘method equals 2,985.2 psi. And if we extend the straight line of slope 15.62 palleyele, we will obtain P* = 2,979.2 psi, from which P by the MBH rethod would be equal to 2,981.5 psi. In other words, only by extending the very early straight lin it is posible to correctly estimate P by the MBH method if, and only if, the true shape of drainage is known, ‘Suppose, however, that we did not know the exact shape of drainage and wwe terminated the buildup test after 12 hrs of shut-in. Then, by assuming that the drainage area is a square, P would be estimated om the basis of Fig- ure 27 as follows: Py = 16 2,979.2 -P 2.309 29792-P 1.6 17.56 P= 2,967 psi ‘which differs from the correct P by 18.5 psi. The most important conse- quence of this erroneous pressure is that we would be tempted to conclude that the reservoir is much smaller than it really is. If this were a develop- rment well and the buildup test was made as part of a fieldwide survey to ‘estimate the average pressure ofthe entire reservoir for material balance cal- culations, then on the basis of lower than-actual average pressure, material balance will give an original oil-in-place which is mach lower than the oll- in-place originally present in the reservoir. And if this were an exploration well, we could, on the basis ofthe test, declare the reservoir uneconomical, ‘Note that the P* of 2,988.3 psi obtained on the basis ofthe third straight Jing is reasonably close to the correct P. For this reason, iti best to close the Fundamentals of Pressure Buildup Testing 98 ‘well for a sufficient length of time to allow the third straight line to develop in the event the drainage area is not a circle or a square, If the well is pro- ‘ducing in the transient state at shut-in, then the well must be closed until Horner's time is equal to 1.5, or for a At = 2. This condition, however, does not apply if the well is producing in the sem-steady state before shut- in, We will study this case next, These conclusions can be stated differently: in the case ofa heterogeneous reservoir (and all reservoirs should be consid- cred heterogeneous unless proven otherwise), and with the well closed in the tuansient state, then P will always be incorrectly estimated unless the well hhas been shut-in for twice the producing time. The magnitude of the error depends on the actual shape of drainage. ‘The example which we have just concluded shows the difficulties in- volved inthe determination of P when the well att, = 60 hrs was producing in te transient or infinite acting state. The question now is: what additional difficulties should we expect ifthe tested well were producing in the semic steady state atthe time of shut-in? To answer this question, we fist refer to the set of MBH curves in the book by Earlougher (1977). Bach curve is la- belled with an asterisk which indicates the beginning of the semi-steady state for the corresponding shape of drainage. Thus, for a by 1 rectangular 7 Wwrdtya ° Figure 2-18. Horner's plot. Detection of linear boundary; data from Table 210. i From Figure 2-16 the following information is obtained: 17.56 patleycle i Pre = 2,950 psi P= 3,000 psi 1 (lp + At)fAt, = 33 Substitute the given reservoir and fluid parameters in the expression for m, and by Equation 2-3, the following results are obtained: k= 50md 52 c P,= 8,000 psi ‘To find the distance to the linear boundary, we substitute in Equation 294, Thus, Ei (~) = 2.908 log 33 =3.497 Fundamentals of Presture Buildup Testing 108, or see can = 0.017 From which, d = 54.8 ft ‘However, because of the numerous approximations and roundboffs, it is bet- ter to report a= sof ‘CONGLUSIONS 1. Under ideal reservoir, fluid, and flow conditions, determination of P from a buildup test is approximate unless the shape of the drainage area isa square or a circle, or unless the pressure buildup is monitored for At > 2ty. 2. Under ideal reservoir, fluid, and flow conditions a breakin the slope of the Horner's straight line can occur due to the use af the wrong value Of ty in calculating Horner’s time; deviation of the shape of drainage from that of a circle, square, or hexagon; and, presence of a linear ‘boundary. In Chapter 3, we will see that a break in the slope can also occur if the flow rate varies considerably before shutting the well in, 8. Except for a limited number of geologic conditions, the shape of the drainage area cannot be determined to any degree of certainty. Reser- voir heterogeneities of seemingly homogeneous reservoirs can create drainage areas of unpredictable shapes. Natural fractures, multilay- ering, permeability pinchouts, directional permeability, proximi oil-water or gas-oil contacts, and steep dips are but a few examples of reservoir heterogeneity. 4, Determination of the radius of investigation of a buildup test can be made through the use of Equation 1-35 upon replacing t, by the maxi- ‘mum At. Thus, for buildup tests, Equation 1-5 is written as follows: Ke Ata” o-oo wos Note that Atnar must be less than ty and/or t. In other words, the radius of investigation of a buildup test cannot exceed the radius of investigation of the drawdown which preceded it, nor the radius of drainage, r. 104 Well Test Analysis TREFERENGES __ (Cheng, M. C. Aaron, “A Simple Technique to Determine the Unknown Constant in y or x for Producing y vs. x as a Straight Line in Reservoir Engineering Problems,” Jour. Pet, Tech., May 1982, pp. 1140-1142. Bossie-Codrean, D., “A Simple Buildup Analysis Method to Determine Well Drainage Area and Drawdown Pressure for a Stabilized Well,” SPE Formation Evaluation, Sept. 1989, pp. 418-420. Dietz, D. N., “Determination of Average Reservoir Pressure from Buildup Surveys,” Jour. Pet. Teck., Aug. 1965, pp. 955-959: Trans. AIME, Vol. 234. Barlougher, R. G., Jr, Ramey, H. J, Jr, Mille, F.G., and Mueller, T D., “Pressure Distributions In Restangular Reservoirs,” Jour. Pet. Tech, Feb, 1968, pp. 199-208; Trans. AIME, Vol. 243, Earlougher, R. C., Jr, Adoances In Well Test Analysis, Dallas: Soe. of Pet. Eng. of AIME, Monograph Series, Vol. 5, 1977, pp. 61-64 Hassan, A. R. and Kabir, C. 5, “Pressure Buildup Analysis: A Simplified Approach,” Jour, Pet. Téch., Jan, 1983, pp. 178-188. Homer, D. R., “Pressure Buildup in Wells.” Proc. Third World Pet Cong., The Hague, Sec. Il, 1951, pp. 503-523. Also in SPE Reprint Se- ries No. 8. Matthews, C. S., Brons, F, and Hazebroek, P, “A Method for Determi- nation of Average Pressure in a Bounded Reservoir.” Trans. AIME, Vol. 201, 1954, pp. 182-191. Also in'SPE Reprint Series No. 8, Mead, H.'N., “A Practical Approach to Transient Pressure Behavior,” SPE Paper 9901, 1981, California Regional Meeting, Bakersfield. Miller, C. C., Dyes, A. B. and Hutchinson, C.'A., Jz, “The Estimation of Permeability and Reservoir Presture from Bottom Hole Pressure Buildup Characteristics,” Trans. AIME, Vol. 189, 1950, pp. 91-104 Also in SPE Reprint Series No. 9. 3 Fundamentals of Multirate Flow Testing In Chapters 1 and 2 we developed interpretation methods that are appli cable when q is constant. In the event that q cannot be stabilized, the meth- ods of Chapters 1 and 2 are not applicable and we must resort to the meth- ‘ods developed in this chapter. In addition, the theory of multirate flow ‘esting is needed to study wellbore storage effects, which will be discussed in the next chapter. Furthermore, the advent of high resolution bottomhole flow meters could revolutionize well testing procedures such that constant q ‘may become a preference rather than a requirement, and with future ad- vances in systems identification the traditional buildup test could be re- placed with some form of multirate testing. We will frst study two-rate flow testing and then expand the theory to the case of multirate flow testing. We will rely on our knowledge of and familiarity with superposition to develop the governing equations, TE FLOW TESTING Atovo-rate flow testis supposed to yield the same information that ean be obtained from a buildup test: the permeability, k; and the skin factor, s. The test can be performed whether the well is producing in the transient or seri steady state, Inthe latter case, the test should also provide an estimate of P with the same margin of exor of a buildup test ‘To perform a two-rate flow test, the well must be producing ata constant rate, qy, for a reasonable length of time. A pressure recorder is lowered into the well toa level above the perforations where it is left for @ few hours 105; 106 Well Tet Analysis before the rate is suddenly changed from qj to qs. The theory we are about to develop is applicable whether quis les or greater than qs. But itis best to require q: to be less than a, forthe simple reason that i is always possible to reduce the rate, Also, if q; were greater than qy, Pg could fall below the Dubble point and thus gas could evolve from the ofl; which adds comples- ities to tet interpretation, Economically, a two-rate flow test is more desirable than a buildup test the well continues to produce during the test, abet ata diminished ratey thus production losses are reduced. On the other hand, a tworrate test fuires closer attention than a buildup test, and itis quite often the case iat by the time the second rate is stabilized, the transient period has lapsed and the opportunity to obtain k and s has vanished. ‘According to the principle of superposition, changing the rate from q, to qp 4s equivalent to the simaltancous injection at the rate qy and production at the rate gp. The net rest is the injection of (q, ~ qa). Recall that in the case of a pressure buildup test we close the well by injecting atthe rate q only, which is ‘equivalent tothe simultaneous injection of and production of qs ~ 0. Thus, buildup test could be viewed as a two-rate flow test in which the second rate is set equal to zero. For the ease of changing the rate while the wel is producing in the tran- sient state, the two-rate flow test equation is derived by superposing Equa- tion 1-82. However, before, we perform the superposition, let us define the following terms 162.6 Bee, 1626 a Bx kh * kh and mm ~my~ 028-4) Bee nd m,~ mp i Thus, my = m, 2 a = producing time at the rate of gy, hrs ime measured from the instant of changing the rate from q; to bs ime measured from the instant of producing the well at qi hrs In particular, after changing the rate from qj to qe, the following relation holds (Figure 3-1): =hrat Fundamentals of Multirate Flow Testing 107 3 1 Flow Rate, q STB/D Time t hrs: Figure 9-1. Relationship between t,t, and At. With these definitions, we can derive the two-rate flow test efuation by superposing Equation 1-22 as follows: AP, = m Qogt +5) (0.29) AP, = = (m~ my [log (6) +3] The sum, AP = AP; + APs FP t= = t+ [2 (og (=) + Ios gk ~ 8.29 +0875] rh -Pav moreso (0.87 s+ log —* - 3.29} @n TOE 108 Well Test Analysis According to Equation 3-1, a plot of AP = P, ~ Pyr, or just Py, versus the ratectime-function, RTF, where eae =log +B log av’ ‘on rectangular graph paper should give a straight line of slope m,. However, from the practical point of view, itis almost always difficult to stabilize the second flow rate from the instant of changing the rate. Therefore, the plot ‘may actually be in the form of two straight lines or a curve followed by a straight line. The correct straight line is that which corresponds tothe higher values of at’. ‘Todetermines, we wste Equation 1-22tthe instant of changing the rate co ~ millog ty +3] (0.22 @at’ = 0) By evaluating Equation 3-1 at At’ = Ih, and then eliminating P, from Equations 3-1 and 1-22, we get: |8_} ~ tog —k__ 4. 3.931 2) onGe I the rate is changed while the wells producing in the semi-steady state, then the two-rate flow test equation is derived by superposing Equations 1-23 and 1-22 as follows: AP, = (ingah t+ (Cpa) + 0.87 my s AP, = ~ (mm ~ my) [log (t~ t) + 3] ‘The sum, AP = AP, + APs Thus, P,~ Pye m, B—2 Jog at’ + (gay (tp + At”) + (Cyult w= afiog__k +m, B23 ho ~ 3.034087. a [Sonar +0.87 ms 3) Fundamentals of Multivate Flow Testing 109 we = 0.472 te (Cyn) = 2 my log. Equation 9-3 shows that for small At’, t, + At’ = ty, and a plot of AP, or Py, versus At” on semi-log graph paper (MDH plot) gives a straight line of slope m (4h ~ «2)/ay. As At” increases, however, the straight line changes into a curve which is concave downwards To find s, let us write Equation 3-3 at At’ = Ihr and Equation 1-23 at At’ = 0. Thus, ~ Phe (paity + (Cpa) + m 2B) +0.87m 5 (p01 tp + (Cahn + 0.87 my s By subtracting these two equations, we get: sn Esa) ag Ag val which is the same as Equation 3.2, Note that at the instant of changing the rate from q, to da, Le at At’ = 0, the pressure drop due to the skin is given by: AP ae = 0.87 my 5 If the well is producing at the semi-steady state when the rate is changed, then until the instant of changing the rate, Equation 1-11 is valid. Thus, recalling Equation 1-11 in terms of q; and adding APyie, we get: = 995.2 1 Boi joy 0472 ne a SS +087 ms P~Patgarno ‘This equation enables us to determine the average reservoir pressre, Brom a bvorate flow test, asuming a circular drainage area, because eis obtain- able from the slope of the straight line on the MDH plot, sis ealeulated by Equation 3-2 and Prigse, 5 is measured. 110 Well Test Analysis "To account for drainage areas of different shapes, the Dietz shape factor is {introduced in the previous equation as follows, noting that Cy fora circular drainage area is equal to 31.6 (see Figure 2-17) 0.472 55 B~Petgar ag 2 log +087 ms +0.495 s) - ult 0.293 x 31.6 x ef : wmf ABESLECHE oo, or, 1 Ce Pa2m fo I + 0.495 | + Potgarea ey where A = drainage area, sq. ft In conclusion, itis necessary to monitor Py for afew hours before chang ‘ing the rate from q; to qz,'mainly to make sure that there are no severe pres- sure fluctuations, Tf the well were producing in the transient state at the instant of changing the rate, then k can be calculated by measuring m; on a rectangular plot in accordance with Equation 3-1 and sis calculated with Equation 3-2. If the well were producing in the semi-steady state then k is determined by measuring the slope of the straight line on the MDH plot in accordance with Equation 3-3; s and P are then determined by Equations 3.2 ind 3-4, respectively ‘As an example, suppose we rin a two-rate test and obtain the pressure- ‘time data shown in Table 3-1, To interpret the test, the following reservoir, PVT, and flow parameters are given h= 208 6 = 015 Figure 8-2 is a plot of the pressure data before changing the rate. Since this plot exhibits a straight line, one might tentatively suspect that before the rate change the well was producing in the semi-steady state. Nonetheless, Fundamentals of Multiate Flow Testing LLL Table 3-1 ‘Two-Rate Flow Test Example Data Before Rate Change Data After Rate Change t Pas Pa RateTine (os) (ws, (ws) Function 208 e075 P 2a7s are 269 16072 2.402 230 1906.8 2487 aa 1808.4 age 22 1808.4 2407 23 1505.7 2.500 24 1508.4 2.502 235 11605. 21505 236 49086 2508 27 16043 2ar4 298 2520 209 2505 ‘300 Bast 2582 252 224 2508 238 2n2 213 2.590 210 2.600 206 Put, psi ‘io 280 202i 282 233 2a wo ae aby abe ae Producing ‘Time, tp Figure 3-2. (Pi - Pw) vs producing time before two-ate test; data from Table $4 112 Well Test Analysts the data of Table 3-1 have been plotted twice: according to Equation 3-1, Figure 3-3; and according to Equation 3-3, Figure 3-4 (MDH plot) Figure 3-3 shows two straight lines instead of just one. The question now is: Which one of these lines should we use to obtain the slope? To make our choice, we reason as follows: it is very unlikely that the second rate has ta bilized from the instant of changing the rate, and therefore the data ob- tained immediately following the rate change should not be considered. Based on this argument, the straight lin is fitted as shown on Figure 33. From Figure 3-3, the slope m, = 274 paifeyele. Thus k is approximately 4.5 md. Also from the same figure, Pry. = 2,485 psi, and by substituting into Equation 3-2 we estimate s approximately at 3.0. From Figure 3-4, the slope mi(qi ~ qs)/q) = 193 psifeycle, ‘Thus ‘m; = 266 psileycle, The permeability k, is approximately 4.5 md, and from the value of Pix, = 2,488 psi, essentially the same value of s is obtained by Equation 3-2. “To check on whether the well was producing in the semi-steady state ‘when the rate was changed, we substitute in Equation 1-36 or 1-37, How- ever, if by using Equation 1-36 the well isnot found to be in the semi-steady state then there would be no need to repeat the calculations by Equation 1-37. By substituting into Equation 1-36, we obtain tye = 378 hrs, and we conclude that if the shape of the drainage area isa circle or a square with the ‘well atthe center, then the well was not producing in the semi-steady state atthe time of changing the rate, despite the appearance ofa straight line on Figure 3.2, Since from the MBH curves we know that for any other shape of drainage it takes longer to reach the semt-steady state, we conclude that at t, = 300 hns the well was still in the transient state MULTIRATE FLOW TESTING ‘The theory of multirate flow testing is an extension of the theory of two- rate flow testing. The fundamental equation is derived by superposing the transient drawdown equation (Equation 1-29) n times. The procedure is il- lustrated by referring to Figure 3-5 which exhibits a hypothetical case of varying q versus time. ‘To perform the superposition, we first divide the curve of Figure 3-5 into small time increments, and for each increment we calculate the mean flow rate and consider each mean to represent the flovr rate that corresponds to the time increment. By this procedure, we simply convert the variable flow rate into a sequence of constant flow rates as shown in Figure 3.5, Put, ps Fundamentals of Multirate Flow Testing 113 2625: 2600 ae ope 2206 peer “R2ss0 - sos. 2500. 2475. 2450, 2 ads, 230 2s, Rote—Time—Function, RTF Figure 9-3. (Pi ~ Pw) vs ratectime function (RTF); data from Table 3-1 2608. Ue 2600. 2575. 2505 2500 2478. 268049 pe i at ne Figure 3-4. MDH plot; data from Tablo 3-1, Ld Well Toot Analysts Figure 3-5. Hypothetical case of varying q versus time, Oand &=0 the mean flow rate from t = 0 tot = t, the mean flow rate from t = t to t = te the mean flow rate from t = t to t= te 4-1 = the mean flow rate from t ‘a, = the mean flow rate from then, Equation 1-22 can be superposed as follows: AP; = m’ ay flog t+ 5) AP, = =m’ (a) ~ a) flog (t- &) + 5] APs = — m’ (ap ~ as) [log (ty ~ te) + 3) AP, = =m’ (qy-1 ~ 4 [log (te ~ tao) #3) ‘The sum AP = P, ~ Pys is given by: 5) Fundamentals of Multirte Flow Testing 15 = Pas ceo 99 ee ew +a) oe + (a5 ~ a3) log (ta — ty) +--+ (Qa ~ dns) 10g (ba ~ te] +m'5 6 Equation 3-6 is the multirate flow equation. In this equation, the rate-time funetion, RTF, is defined by: TF = 2 [a ~ 9) log (4) * (= a) log 8) + (a) 10g (ty ~ ta) + nee + (Qp — Gat) 108 (ta — ti] 7 According to Equation 3-6, a plot of (P, ~ Puy)/qy versus RTF isa straight line of slope m’ and intercept equal to m’ 5, from which s is determined, ‘Equation 8-6 ean also be expressed as follows PaPw ye Th = =m’ [D2 tq — 9-2) log te ~ ah] +m (4 rs La, 91-2) 10g ft ~ t | 66) where gp = 0 b=0 AAs an example, suppose that while performing a drawdown test on a ‘well, the flow rate increased from 0 to 105 STBID over 5.25 hrs as shown in Figure 3-6 and then continued at the stabilized rate of 105 STBID. The pres sure data are given in Table 3-2. Tp determine k and s, we proceed as fol- lows. The first step isto calculate RTF by Equation 3-7 Forn = 1, = 0.75 hn, q = 7.5 STB/D RIF = 2 {(a,~ a Jog ty ~ 8) Fg lt5~0) log (0.75 ~ 0)] = — 0.195 ue Well Test Analysis Flow Rate, q S1B/0 ‘Muttrate Flow Test Data (for the first 5. Figure 3-6. Variation of q versus time. ‘pressure drop (P; ~ Px) at the middle of each time interval) 2 13 7 ors) 076 150 225 3.00 375 450 525 725 30.28 2025 5025 100.25, 360.25 ‘ean @ STD 78 228 375. 25 ors. 225. 975. 105, 405, 108 108 +108 405 P (esp 1923 4023 oan 905.4 1.258:7 1545.2 4139.5 2.0482 21004 2203.6 2a14s 2,998.8 2,440.4 fd 0.128 = 0.025 0.091 0.188 0.250 ozs 0.380 0638 ogee 268 1878 41980 2370 Pian (psiTB/0) Te 1788 1816 1339 1857 1873 1887 y951 20.00 2099 2208 22.80 23.26 Fundamentals of Multirate Flow Testing 117 For n = 2, ty= Lh, qp = 22.5 STB/D RTF = = f(a, ~ Joe a ~t) + (42~ a) J to ~ 10] 1 - = Bg [M7S~ 0) log (1.50) + (22.5 ~7.5) log (1.5 ~ 0.75) = - 0.025 Forn = 7, t= 5.25 hrs, q, = 97.5 STBD 1 RTF = gt [7.5 log (5.25) + 15 log (5.25 ~ 0.75) + 15 log (6.25 ~ 1.5) + 15 log (6.25 ~ 2.95) + 15 log (6.25 ~ 3) + 15 log (5.25 ~ 8.75) + 15 log (6.25 ~ 4.5)) =0.38 For n = 12, t= 100.25 hrs, qa = 105 STB/D RIF = 71 (7.5 log 100.25 + 15 log 99.5 + 15 log 98.75 + 15 log 98 + 15 log 97.25 + 15 log 96.5 + 15 log 95.75 + 7.5 log 05) = 1.99 A plot of AP/q, versus RTF is given in Figure 3-7. From this plot, the fol- lowing results are obtained: rm’ = 2.430 pailoyele 1b’ = 17.95 psiiSTBID where b’ = interoept M8 Wall Test Analysis 2s. ‘Sopen2 438 psi/eyee O50 Os bdo” Vdo 1d0 2 20 Rote-Time Function (RIF) Figure 3-7. Interpretation of multirate flow test. Data from Table 3-2 102.6 x 1.2 x 1.95 2.439 x BD =5md and s= L151} + 3.23] 8) an “4 Tshould be noted, however, that to interpret multirate flow tests, the flow rate must be measured with precision, This was not possible until very re- cently when high resolution, bottomhole spinners were introduced. ‘The new spinners have a linear response over a large range. Therefore, i is not necessary to enter the flow rate in STBID in the above equations. Its suffi ty Figure 3-12. Schematic diagram showing constant q; until t= Ly; variable rate thereatter Pundamental of Multirate Flow Testing 125 For the transient case, we rewrite Equation 3.6 as fllows: P,—Por= m’ DY (a) ~ 44-1) Jog (ta ~ 2) + aa mF 8) Since qy = ty = 0, Equation 3-6 can be written as follows, for ty > ty P,~ Potgs, =m’ qs log te +m’ 7 (ay — 44-1) log (te ~ 1) tems . (@12) and at t, = ty, Le, at At, = 0, Equation 1-22 can be written as follows: Pi = Prigay-0 =m’ gi flog t + 5] (1-28) By ublrating the above two equations, and noting that = ty + Ate we Pooon Potgayeo m8 a Tog Bae +m! Ye fas -a) Hog (ta ~ t-2) + (ua) | Since forty > ty, t= ty + At, the above equation can be written as follows: Pram sas Pet, | 1 bo ae a=) "r= an [ME Saag By tog - 44-3] | +5 (19 where gp = 2 a AP + a1 ~ Pwtgaiged (= ap) (=p), Tog leetg EE ean os Bea att (at, = At) (14) 126 Wel Test Analysis Equation 3-13 shows that a plot of AP/(1 ~ gp) versus Z,/(U~ qo) is a straight line of slope m,. From the intercept, scan be calculated by Equa tion 3-8 upon substituting my for m’. Note that by using gp, only q, is needed in RB/D together with the corre- sponding spinner reading. For all the rates that follow q, only the spinner readings ere needed. 1 the wel is producing in the semi-steady state when the rate is changed, then the first term on the right-hand side of Equation 3-12 is written as mm’ q [log ty + Y(ty)] where ¥() has been defined in Chapter 2, We then replace Equation 1-22 by P,—Por= m’ ay [log ty + Vet) + 5] and subtract the two equations. The result is the same as Equation 3-13, since for small At,, Y (At) = 0 and ¥ (tp + At) = Y(t). As an example, Table 3-4 shows the test data and computation results for 1 well which was producing for several days at the stabilized rate of 1,103 RBID, from a reservoir with the following rock and fluid characteristics; h= 18 ft =001 = Mx 10-Fipst n= 100p Tu = 0.26 ft te = 3,000 ft Table 3-4 ‘Two-Rate Flow Test, Variable Rate Case nah oe 0, APS = ae) East ~ a) yo 4108 2 4 36H 00 O75 97.9863 o.6021 3 6 3880 Bao 07821258175 0.8507 4 8 3865 Mo 0782 148.7871 0.9489 593.866 80 0708 178.0628 aan 6103865 800798 187.9552 1852 ‘pte tor Odah ana ones 187 Since the well was producing ata stabilized rate for several days, and the ‘maximum value of At, = 10 hrs, then it is reasonable to set ty + Aty = tp, ‘and thus the first term in Equation 3-13 becomes equal to zero. Fundamentals of Multirate Flow Testing 127 Forn = 5, At, » 9 hrs, qo = 880/1,108 = 0.798, AP ~ 3,666 ~ 3,630 = 36 psi. Thus, AP. __36___ 178.0998 Ta)” T= 0.705) x. 1a O= ap)” T= 07708 1108 [aos ~~ 800) log {9 - 0) + (800 - 840) log (9 —4) + ($40 ~ 840) log @ ~ 6) + (840 ~ 860) log @- 5] =1amg Figure 3-13 is a plot of AP/(1 — gp) versus ©,/(1 ~ gp) for the data shown fon Table 8-4, From the slope of 170.3 psiRBID, the permeability, kis found a follows: 102.6 4 Bae mb __ 162.6 (1108) 1.0 “Tax 18 k =58 md 200.0 Sope=170.3psifecte Intrepte—i8.2 Figure 3-19. Two-rate, variable rate case; data from Table 9-4 128 Well Test Analysis From Figure 3-13, the intercept is ~ 16.2, Thus by Equation 8-8, modi- fied for this case by substituting m, for m’, we obtain: binos ~!8 paper eta w oe * 99 Ifthe drainage ara is circular, we ean easily ealeulate by Equation 1-36 that it takes about 9.5 days for the well to reach the semi steady state, and wwe know from the MBH curves that it takes longer for other shapes. Since the duration ofa isnot given, i isnot possible to decide whether the semi- steady state was reached before changing the rate, Bu ifthe well were pro- ducing inthe semi-steady state, then P can be calculated by Equation 94. Before leaving tis example, let us evaluate Equation 1-22 (the drawdown equation) for this well after one minute of production atthe rate of 1,103 KBD, P= Per m flog t+ 5] (1-28) 162.6 x 1,103 x 1 BBX 18 = 178 palleyele k —*_,-s23+0875 SPOR 58 ~ be xT xx eet =-01 ‘Thus, P, ~ Pyr = 171.8 [log 0.0167 ~ 0.1) = ~ 392.7 psi ‘The above result shows that by opening this well to flow, the flowing bot- tombole pressure rises above the reservoir pressure by 322.7 psi during the first minute of production. This, of course, is absurd, and there must be something wrong. Fundementals of Multirate Flow ‘esting 129 ‘What is wrong here is that we applied the logarithmic approximation of the Ei function when itis not applicable, If the reservoir is homogeneous, ¢tc,, then a negative skin indicates that the well is stimulated. When 5 = ~ 645, the well is most probably hydraulically fractured. Thus, the effective radius of the well, ra, is given by: 0.26 x er = 164.5 ft ~ 948 9 w Cor, Thus, Fi(—3) He i i F St (hour) Figure 3-15. Presaue buildup tet, Data trom Tablo 36 Sep ee 2300.00 ; - 2450.00 e000 4 4 2380.00 * 2300.00 +e . 2280.00 Fro 2200.00 = : 2 sa é E ay00.00 ‘ a 2080.00 ie 2000.00 re 1350.00 : 120000 4 | 1850.00 1200.00 ; : : ' : 70 18. aol? 10 10 i Whe a %b Figure 3-16. Homers plot. Data from Table $6. Logiiog plot of (Pws ~ 1844.65) vs At; data from Table $6, 140 Well Test Analysis coma rom page 128) ‘whereas the last data point which falls on the straight line of unft slope on Figure 3-18 is at At = 0.09 hr. This means that fluid influx into the wellbore has affected not only the data that fall on the straight line of unt slope, but also the data between At = 0.09 and 1-4 hrs, which is a ratio (1.4/0.08) of about 15. We will rettim to this subject in Chapter 4 ‘To determine B. we first caleulate tay, by Equation 2-13, and then use Equation 2-21 and Figure 3-17. Thus, fap = 0.167 ‘At, = 29.94 hrs B= 2,370 psi Itshould be clear by now that the assumption of a homogeneous reservoir {implicit in the process of extending the straight line on Figure 3-15 or the semi-log straight line on Figure 3-16 to obtain P*. The author has exp. enced several cases in which the interpreters of buildup tests of wells that ‘were shut-in during the transient state extended the semi-log straight lines and obtained P* < P,, and concluded that the tested reservoirs were too small and uneconomie, when in fact the reservoirs were large but heteroge- ‘SUPERPOSITION, CONVOLUTION, AND DECONVOLUTION Itis evident that the multirate drawdown equation (Equation 3-10) was derived by superposition. It has been shown that the convolution integral equation (Equation 3-11) is an equivalent expression of Equation 3-10. ‘Therefore, we conclude that convolution and superposition are the same. ‘We could write Equation 3-11 as follows: P= Pa(t) =m! [880 og ¢—9) drt m’ GF eu) Since q(t) and log (t) are continuous functions of time, the above equation could also be written as: BP (AL 9 ar’ & Note that the occurrence of m’ log(t ~ 2) and 5 in Equation 3-11 is due to the fact that we have adopted the approximate logarithmic solution to the radial flow Equation 1-26 for a homogeneous, infinite reservoir and con- Fundamentals of Multirate Flow Testing AL stant flow rate. Furthermore, in deriving Equation 3-11 we have assumed that at t= 0, (0) = 0. "To generalize Equation 3-11, we first remove the requirement that .q(0) = 0, and we let P(t) be the unit influence function, Itis the solution of Equation 1-26, at r= ry, which corresponds to a unit rate of production, Then, AP) = GO) PO) and AP(t) = [a(t) — a(0)] Pt ~ 4) AP) = [at = Ke] POE ~ ) P(t) = Lats) ~ a(t] Pit - ts) By superposition, the total pressure drop at the wellbore, AP = P, ~ Pu(t), {s simply the sum of the above expressions, Thus, AP = a(0) PO) + fats) ~ af] PO ~ 8) + fate) ~ afti)] P(t — te) + on aP =a) + B= @ <0) M4) =a) (4) PEW) + ‘hich inthe limit becomes: ap = qo) Pt) + f° 9 pe a dr (820) Wie now inteprate by part. Let P69 hon d= PE a ant r= 290 th v= M42 Well Test Analysis Ths, f) 88 Re a= ay PE 2] + fa) PEED ar = a(9P(0) - q(ORW) + f° ate) arte) rs dr Equation 3-20 can nowt be written ap PO) + J; at) PER ay @a Equations 3-20 and 3-21 do not account for the skin effect, To add the skin effect, we replace P(t) in the above derivations by P,(t), where P(t) = P(®) +8" "= pressure drop due to the skin, for a unit rate of production Let AP, = P, ~ Peis then, when q(0) = 0, Equation 3-20 becomes 8,~ [9 nee oer ea or, AP, at) + f 82-9 2 ‘and Equation 3-21 becomes: AP, = af) PO) + fat) AE Day ‘This equation is valid since s” is a constant, and therefore, ae t — dr aP(e- 7) ra APU) = 0, then B,(0) and the above equation becomes: =aty s+ f° a(R 2 Fundamentals of Multirate Flow Testing 43 Since g(t) and P,(t) are continuous functions oft, the following result applies to all the convolution integrals: a9 aR) Ja P= AO ar arm J" t= 9) Tn these equations, AP i= at) Thus, ql) s"= 0.87 ms _ML2 a Be, kh Now let q, = a reference g, STB/D and multiply both sides of Equations 3-22, 9-23, and 3-24 by (kh)/(Idl.2q.Bu)s ‘then apply the following definitions: ‘ty = dimensionless time = 2.637 10-4 kt Ou Cty off) = dimensionless presure drop in the wellbore Poltp) = dimensionless presssure drop in the reservoir per unit flow rate, fas obtained from the solution to the radial flow equation (Equation 1 Md Well Test Analysis With these definitions, Equations 3-22, 3.28, and 9-24 are, respectively, ‘written in dimensionless forms as follows: Pyolte) = f° S421 Pty ~ 10) dry 25) edn Prolto) <5 (to) + J" Sapte) Polto~ 7) dro 626 Prolto) = ato) + J ane) Pelo= "day ox ‘The problem now may be stated: Given Pai(t) and q(t), is it possible to ‘obtain the pressure response P,(t) that would have been recorded ifthe rate ‘were kept constant? The answer is yes, we can do that by deconvolution of Equation 3-25, 3-26, or 3.27. ‘The problem with deconvolution, however, is that it requires highly aocu- rate data, and all the pressure variations in the recorded Pyt) must be caused solely by the rate variation, which is not always the case. Pyg can vary due to the formation of gas bubbles in the tubing, and in response to density and temperature variations, mar-slar tides, and closing and open- ing of nearby wells. Also, part of the variation in the recorded Py; could be caused by the fact that opening or closing ofthe tested well does not oceur instantaneously; rather, it proceeds over a measurable length of time. Fur- thermore, the available high resolution, bottom-hole flow meters are inac- curate in & two-phase flow environment, and they are also inaccurate below a certain rate threshold. Thus, in the presence of oil and gas or water and ‘gas, and below the rate threshold, the available flow meters are inaccurate, For these reasons, deconvolution, although theoretically viable (Kucuk and Ayestaran, 1985), is not practically possible with present-day technology. ‘And although the literature abounds with deconvolution schemes and some service companies are offering deconvolution, it is not really possible except perhaps in those rare cases that involve strictly single phase flow, ie, in gas wells, But, as we will see later on, in gas well testing there is a non-Darcy flow coefficient to contend with, which is not a constant, but a function of the flow rate, Therefore, when the rate varies, part ofthe observed pressure variation is caused by non-Darcy flow effects which cannot be accurately estimated. Tn spite of these present-day difficulties, we will stady deconvolution be- ‘cause this author anticipates that near-future improvements in instrumenta- tion and monitoring will make it posible to succesfully apply deconvolu- Fundamental of Multirate Flow Testing 145 tion. However, deconvolution is easier to apply in the Laplace domain rather than in the time domain. For this reason, the subject of deconvolution is presented in Chapter 11. REFERENCES 1, Odeh, A. S. and Jones, L. G., “Iwo-Rate Flow Test, Variable Rate Case—-Application to Gas-Lift and Pumping Wells,” Jour, Pet. Tech., Jan. 1974, pp. 93-99 Trans AIME, 257, 2. Meunier, D., Wittmann, M. J., and Stewart, G.,“Interpretation of Pres- sure Buildup Test Using In-Situ Measarement of Afterflow," Jour. Pet Tech., Jan, 1985, pp. 143-152. 3. Kucuk, F. and Ayestaran, L.,“Analysis of Simultaneously Measured Pres- sure and Sandface Flow Rate in Transient Well Testing,” Jour. Pet. Tech., Feb, 1985, pp. 323-334. 4 Wellbore Effects Inthe first three chapters it was assumed that the conditions atthe well. bore were ideal such thatthe test data were only influenced by the reservoir characteristics. In reality, however the conditions at the wellbore are not ideal, and the test data are affected by the conditions a the wellbore, espe- cially during the early part of the test. Willbore storage, hydraulic fracturing, partial penetration or perfora- tion, and phase separation are among the wellbore conditions tha influence the test results Accordingly, i is very important to learn how to recognize the presence of these effects inthe test data, and to estimate the duration of thr influence on the observed pressure-time data In this chapter we will begin th discussion with wellbore storage effects and then discuss the other wellbore effects in the following order: hydraulic fracturing: partial penetration or perforation; and, briefly, discuss the ef facts of phase segregation inthe wellbore. Inthe process, we will introduce the subject of type-curve analysis and become closely familiar with certain specific type-curves. WELLBORE STORAGE EFFECTS In the introduction to Chapter 2, the problem of wellbore storage is stated. In brief, wellbore storage effects arise beeause upon opening a well for a drawdown test, part ofthe flow comes from the expansion of the fluids {n the wellbore, In a buildup test, wellbore storage effects arise because fluid influx into the wellbore continues after shutting the well in Fundamentals of Multivate Flow Testing 147 In Chapter 3 we dealt with the problem of wellbore storage effets assoc ated with a buildup test (Tables 3-5 and 3-6). We have seen that continued ‘lui influx after shut-in causes distortions in the buildup curve which delay ‘the development ofthe semi-log straight line on Horner's or MDH plots. We haave also observed that wellbore storage effects cause the development of a straight line of lope of 1 on the log-log plot, and we learned how to treat ‘wellbore storage effects when bottomhole flow meter data are available. In this chapter we will learn how to treat wellbore storage effects in the absence of bottomhole low meter data, since many wells will continue to be tested without the benefit of these new flow meters. Qur objectives are rec- cognition ofthe presence of wollbore storage effects, and interpretation ofthe test data in the presence of these effects. The removal of wellbore storage effects from the data can only be done through deconvolution, which is pre- sented in Chapter 11. However, simplified deconvolution methods which are only applicable to homogeneous reservoirs have boon presented by Bour- det and Alagoa (1984) and expanded upon by Yeh et al, (1988) to include the case of changing wellbore storage coefficient High resolution flow meters are now routinely included among the well testing tools because they are helpful in detection of the beginning of the semi-log straight line while the testis being ran (je. in realtime), which could shorten the duration ofthe test, and they provide the necessary data for deconvolution ‘The wellbore storage problem is stated as follows: Let = surface flow rate, STB/D B = formation volume factor, RB/STB @B = surface flow rate, RB/D ‘qy.B = sandface flow rate, RB/D (q~ qa) B = rate of unloading from the wellbore, RB/D ‘Then the pressure buildup theory requires that from the instant of shutting the well in, q B= qq B= 0, and the pressure drawdown theory requires ‘hat from the instant of opening the well to production, qB = qu B. If these requirements are not met, then the early part of the test would be domi- nated by wellbore storage effects, ‘These requirements cannot be met in practice for two reasons the fluid in ‘the wellbore is compressible; and, wells are usually opened and closed at the surface, not at the sandface. Since the fluid is compressible, some fluid must continue to enter the wellbore after the well is shut-in in order for the bot- tomshole pressure (BHP) to iso, In the ease of a drawdown test, production ‘comes from expansion of fluid contained in the wellbore and from the reser- voir. In other words, (q ~ q,)B is not equal to 0 as demanded by the as sumptions upon which the theory of well testing has been developed. In ease 148 Well Test Analysis of pressure buildup, q B= qe B= 0 only when Py, = P. In case of a-draw- down, q B= gy B only when dPyq/dt becomes sufficiently small such that the change in fluid compressibility in the wellbore becomes negligible. ‘To alleviate the problem of wellbore storage, itis necessary to install a production packer immediately above the perforations, and to use bottom- hhole opening and clesing devices. Generally, this is not done, and in the ease of a naturally fractured reservoir wellbore storage effects will occur even ‘when these devices are used, since the fractures act as an extension to the wellbore. Wellbore storage is caused by fluid compressibility, or by changes in liq- uid level in the wellbore. Ifthe wellbore is completely full of incompressible fluid, there will be no storage effects. But, ifthe wellbore is partially full of compressible or incompressible liquid, then wellbore storage ocours due to changes in the liquid level in the wellbore. ‘The compressibility, cy, of the fluid contained in the wellbore is defined by: =14Vy “Vo oF oy where cy = compressibility of the fluid in the wellbore at wellbore conditions, I/psi Ve = toal wellbore volume, tubing ad annulus if the ‘well is unpacked, or tubing only when the well is packed and the packer is set above the perforations, RB AV, AP = change in wellbore volume and pressuze, respectively ‘The negative sign makes ¢; postive, since Vy increases as P decreases. ‘According to the above definition of compressibility, AVQIAP = Ve ‘When the well is completely filled with Fluid, the wellbore storage con- stant, C, is defined as: C= ~ Ava/AP = Ve, blips. 9) "Tp derive an expression for the wellbore storage constant, C, when the well is unpacked and partially filled with liquid, we first define the follow- ing parameter: V, = annular volume, BbU/ft Fundamentals of Multirate low Testing 149 ‘Upon starting the pump, the bottom-hole pressure (BHP) drops, and the liq- uid level in the annulus also drops as the liquid moves into the tubing. Let the drop in liquid level be a distance h (ft), then the change in the liquid volume, AV., is equal to the volume produced from the annulus through the tubing, ie., AVy ‘h, bbl and the change in pressure in the annulus, AP, is given by: AP = p h/l44, pst where p = average liquid density, Ib/eu.ft then, AV,/AP = 144 Vilp, bblipst and, C = 144 Vilp, bbl/pst 3) Note that the expression AV/AP is positive here, since the pressure dimin- ‘shes as the volume in the annulus diminishes, Note also that c; and p are ‘assumed constant and thus C is assumed constant. This, of course, is not ‘rue unless P,is very close to P. In reality Py; is usually mach lower than P, ‘and both ¢ and p are functions of pressure, This means that C is a function of pressure, not constant, Therefore, the theory presented here may not ap- ly when Py is much lower than P, = the above assumptions, the rate of unloading from the wellbore is siven (a- 4) B= cA Ra Pad Pea) Ca P-Pad on gy q~ G ELS (44) In Darey units (cclse, em, sec, atm, darey), the dimensionless pressure, Py, and dimensionless time, tp, are given by; 150. Wall Test Analysis CAP-Py) CyB kPa, C__ Pan Thus FB Brkh eGR diy TSrohG aly In Darey units, Cis expressed in res cfatm, Cs the total system compress- ‘bility at prevailing reservoir conditions in atm, and hy and ry ate ex- pressed in em. Note, however, that the expression, C/rehCart), is both di- Iensionless and unless If Cis given in reservoir cu. ftps, and both hand Tw are given in ft, and C, is in l/psi, the expression C/(2m@hCyry)) will re- main unchanged. Therefore, we can define the dimensionless wellbore stor- ‘age, Cp, as follows: 5.615 C OSes 5) res bbl/psi # ‘Accordingly, Equation 4-4 can be written as: Pao) ae-9 b 7 | 48) ‘where qy starts at zero and increases to the constant rate q. ‘The question now is: If the wel is produced at a constant rate at the sur- face, but because of the compressibility of the fluid contained in the well- bore, the sandface flow rate, qg, increases from zero to the constant rate q, then what is the pressure drawdown (P, ~ Py)? ‘To answer this question, we recall Equation 3.27, the convolution inte- gral. With s = 0, Equation (3-27) is: Pav J” ap (ro) oleate) dn 4 where Pp = pressure response of the reservoir to constant flow rate eng By Equation 4-6, qy may be written: Pan (to) ty ptt) = (1 - Co Wellbore Effects 151 By inserting the above expression in Equation 4-7, Pyy would appear on Doth sides of the equation. ‘Thus Equation 4-7 is an integrodifferential equa- tion. This equation was solved by van Everdingen and Hurst (1949). The solution was presented in the form of type-curves. A similar solution was obtained by Agarwal etal. (1970) and their solution is given in Figure 4-1, Agarwal et al, also found out that during early time, the solution can be approximated as follows: Poa e 68) * = [Oat e [J a GC oe ia = o = = w . Figure 4-1. Dimensionless pressure fora single well in an infinite system; well ‘bore storage and skin included. Copyright © 1970 SPE, Agarwal et al., SPEJ, ‘Sept. 1970, cf Earlougher, 1977 [1,7]. By substituting for Pp, tp, and Cp, in Equation 4.8, we ol Qwkh(P— Pus) Kt 2aghCuk ‘uB cs (9) and log(?,~ Pi) = log t~ log $B The above equation shows that during the early time a plot of (P, ~ Pu) versus t on log-log graph paper would be a straight line making an angle of 45°. Att = 1, C = qB/(P, ~ Py). At any other value oft, C.ean be obtained 152 Well Test Analysis by substituting in Equation 4-9 the corresponding value of (P; ~ Pu). I tis plotted in hours and q is given in STBYD, the wellbore storage constant, C, is given by: © (Res bblips) = 2B 410) BE = Ped Figure 4.1 shows that forall values of Cp, all the curves consist of straight, lines making an angle of 45°, as predicted by Equation 4-8. These straight lines gradually change into curves. The curves then merge with other curves representing flow with skin but without wellbore storage. ‘These latter curves are labelled s = — 5, s = 0, $= 5, 8 = 10, and s = 20; s here is the skin factor. If, for example, we consider the group of curves labelled Cp = E4, we soe that this group merges with the s-labelled curves approxi- nately at tp = (60 Cp + 3.5). The same conclusion could be reached with, respect to all other values of Gp, which means that regardless of the value of {Cp all the data beyond to = (B0 Cp + 3.5 s) are fre from wellbore storage effects, (It should be noted that the negative s here must not be caused by hydraulic fracturing, because fracturing causes linear flow in the vicinity of the wellbore). ‘Another observation is that for each curve the distance between the point of deviation from the 45° line and the point of merger with any one of the storage-free curves is approximately 1.5 cycles. Since Hs eye is equal to VID = 8.16, 1.5 eycles approximately represent a factor of $0. Accordingly, if in a given test the data begins to deviate from the 45 line after 2 minutes, then the duration of wellbore storage effects is 2 minutes, and by the 1.5 cycle rule all the data up to 60 minutes are considered to be contaminated ‘with wellbore storage effects and these data are not amenable to semi-log analysis (MDH, Horner, Py vs log t). Presentation of the solution to Equation 4-7 in the form of type-curves (Figure 4-1) means that we have to rely on these type-curves for interpreta tion of field data. Note, however, that if the fied date extend beyond the limits dictated by the 1.5 eycle rule then we would not be in need of the type-curves of Figure 4-1, but in the event that the field data fall short, we rust rely on the type-curves for interpretation. The type-curves, however, show a very important characteristic: regardless of the magnitude of Cp and regardless ofthe duration of wellbore storage effects, eventually the pressure ‘will behave as if there were no storage effects, Te should be noted that the solution of Agarwal et al. (Figure 4-1) is a drawdown solution. In fact, all type-curve solutions are obtained for the drawdown case. Therefore, these type-ourves eamiot be used for buildup ‘ests without restriction. The only restriction is that the flow period before shut-in must be somewhat large. Recall that buildup involves the superposi- tion of drawdown and injection. Thus, with a large producing time, the Wollbore Effects 153 rate of pressure decline would be small such that it can reasonably be as- sumed nil during the shutin period. With this assumption, the buildup ‘curve would be a mirror image of the drawdown curve, and the use of the ‘type-curves would be justified. However, Agarwal (1980) empirically found that by plotting the buildup data (Pus — Peigy,_,) Versus At, instead of At, ‘on log-log graph paper, the type-curve analysis can be macle without the re- ‘quirement of a Tong drawdown period. Agarwal defined At, as fllows: AG = x ANI + a8) (1) Accordingly, it is recommended to always use Agarwal’s At, in the interpre- tation of buildup tests by type-curve matching, regardless of the duration of te "TYPE-CURVE ANALYSIS From the definitions of Pyp and tp, we ean write: Jog Pao = log (P,~ Pu) + log 228 and, log tp = log t + lo, k ee OE iC ‘The above equations show that a plot of Pap versus toon log-log graph pa- per is equivalent toa plot of AP = P, ~ Py, versus ton log-log graph paper, except that the Pyo versus tp plot i shifted along the pressure and time axes respectively by the constants 2rki/qxB and kigy«Cu. We could, of course, define Pyo and ty as follows: _ e-P tess E quB Gee 2.64 x 10-*k and, ty = BAI : Soran 2 With these detinitions, the plot of Pyo versus toon log-log graph paper will not differ from the one discussed earlier except in the amount of shift along the pressure and time axes. Therefore, to use type-curves, the observed AP = P, ~ Py in case ofa drawdown, Or AP = Py, ~ Pufgy gy incase of a 154 Wall Test Analysis buildup, is plotted, respectively, versus tor At, on log-log tracing graph pa- pper of the same scale of the type-curves. Ifthe plot of the field data begins ‘With a straight line making an angle of 45°, then according to the assump- tions stated earlier, the field data are contaminated with wellbore storage effects and the interpretation ean be made according to the steps given be- low. However, it should be noted that the field data could be affected by ‘wellbore storage but the 45° line does not develop because the actual field conditions deviate significantly from the assumptions made in developing. the present theory (such as when ey or p are not constants but functions of pressure), This can happen when Py: is much lower than P,, or when gas evolves out of solution due to the decline of Py; below the bubble point of the oll Step 1. Find and record the time t or At, at which the field curve deviates significantly from the 45° line. Apply the 1.5 eyeles rule, choose the data beyond that, and plot them on semi-log graph paper. Interpret it in the usual way to determine k and s, and skip the remaining steps. If there are no data beyond the 1.5 eyeles, then the interpretation ean only be made by ‘curve matching as described in the following steps. Step 2. Calculate the wellbore storage constant, C, by Equation 4-10 and the dimensionless wellbore storage constant, Cp, by Equation 4-5, Step 3. Place the tracing paper, which contains the plot ofthe field data, on the type-curves of Figure 4-1, Move the tracing paper up or down or side- ‘ways until you match the field curve with one of the type-curves which is labelled with the same Cp calculated in step 2. For example, ifthe caleu- lated Cp is approximately equal to E5, then the field data plot should only bbe matched with a member of the family of curves labelled Cp = ES. Step 4. Once a match has been obtained, make sure that the coordinates on ‘he tracing paper and on the type-curve are parallel, then choose a conve- ‘ent point on the tracing paper and call it “match point.” From the match point, draw two lines parallel to the coordinates. Extend these lines to inter- sect the coordinates AP and Pyp, and t and tp, Record APayets Pap mas temas aNd tp pate Step 5. From the values Pyp mu and APauua calculate the permeability k by Equation 4-12. The skin factor, s, isnot calculated; itis read off the matched type-curve. The values tm and tous are not used in any calculations. ‘They are only used to identify the coordinates of the match point. There is, however, a strong temptation to use tax ad th muh and Equation 4-13 to Wellbore Effects 155 calculate the porosity, ¢. But this should not be done, because both ry and G. are not known with certainty. Sometimes the calculated Gp value falls somewhere between the Cp val- ues given in Figure 4-1. For example, suppose that Cy is found equal to 6E: In this ease, draw a 45° line at ty = 6E3, then determine the points of ‘merger with $= 5, 0, 5, 10, and 20 according to the formula ty = (60 Cp + 3.55). From each merger point so identified, draw a smooth eurve to join the 45° line as guided by the curves of Gp = E4 and Cp = 5. Use the constructed curve for matching, Another way is to generate Agarwal’s type- curves by the Stehfest algorithm (explained in Chapter 11) ‘As an example, the pressure-time data of a drawdown test are given in Table 4-1. Additional pertinent data are given: q = 190 STBID B = 1.52 RBISTB n= 1.35 ep Py 3,215 psi C= 18.6 x 10-8 pst Table 4-1 Drawdown Test with Wellbore Storage 7 aP=Pi= Pa 1 on c vo) 005 3 3 0.10 6 5 015, 24 0 0.20 a 2 050 fa 30 080 105 50 11.00 120 80 4150 180 100 2.00 190 200 Figure 42 is a plot of the data of Table 4-1 on log-log graph paper of the samme scale of the type-curves of Figure 4-1. From the 45° line, we apply the 1.5 oycle rule. The las six data points fall beyond the 1.5 cycles. These data are plotted on the semi-log plot of Figure 4-3. From the slope of 29 patleycle and APs, of SA5 psi, K ands are determined at d2.md and 4.6, respectively. Although, in this erample, we do not need type-curve matching because there are sufficient data points beyond the 1.5 eyeles, we will proceed with ‘type-curve matching to demonstrate the method and compare the results 156 Well Test Analysis Wellbore Effects 187 From Figure 42, we calculate C by Equation 4-10 and Cp by Equation re 4.5. Choosing t = 1 hr, the corresponding AP ~ 150 psi. Thus, i c 190 x 1.52 x 1/(24 x 150) = 0.08 BB/psi 10° and, Cp = 5.615 x 0,08/(2r x 0.12 x 48 x 18.68 ~ 6 X 0.958) = 10,706 (F-Pt) eis now evident that we should match the field data with one ofthe type- ‘curves labelled Cp = E4. Following the steps outlined above, the following ‘atch point has been chosen: 1a = 10 hrs Panes = 10 psi 107 107 1 10 107 107 ‘tpmat = TES t 1 Pwo mach = 0.4 Figure 4-2. Logg plot. Data from Table 4-1 MAL.2 190 x 1.95 x 1.52 x 04 ‘Then by Equation 4-19, k= a = 42.5 md 500 | | From the matched curve, s = 5 “0 Although, in this example, the results obtained by type-curve matching are comparable to the results obtained by the semi-log analysis, the reader a 7 ‘who works this example would realize that this will not necessarily be the ‘case. There is an element of subjectivity in the matching. For this reason 3 ‘type-curve matching will yield approximate results. This is better than get- a lfof a ‘ting no results at all, as when the testis terminated before the end of the 1.5 § soo. cycles. On the other hand, if it were possible to detect in real time the begin- ring of the seml-log straight line, then the test can be interpreted by the standard methods without having to resort to type-curves. This became pos- sible only recently with the introduetion of the high resolution bottom-hole flow meters el REAL TIME DETECTION OF END OF WELLBORE STORAGE 100: 1 10 10% 10” Previously, C was defined as the wellbore storage constant. However, © (rou) { with the advent of bottom-hole flow meters we are able to continuously Figure 4-3. Semiiog plot ofthe last six points using data from Table 4-1 monitor ge. Thus, we do not need to assume that Cis constant. In fac, we 158 Wall Test Analysis can acknowledge that C is a function of time. Accordingly, we can modify Equation 4 as follows =) we oB ' co 5 ee ay) Equation 4-14 is valid for a drawdown test. In the ease of a buildup, q = 0, and the following equation applies: 1 qB . Cla) = Fe apa (15) ‘To see how to detect the beginning of the semi-log straight line in real time, we will refer to the last example in Chapter 3 (Tables 3-5 and 3-6) ‘The q data in Table 3-5 can be considered ay by the nomenclature in this chapter. Thus, we can caleulate C() by Equation 4-15, The results are given in Table 42 and plotted in Figure 44. Note that not al the data of Table 3-5 hhave been utilized. This is because the threshold of the flow meter was reached at At = 0.671 hr. os os ss Cll), Res ebi/esh on 10” 107 107 1 ‘At (hou) Figure 4-4. Wellbore storage coetficient versus shut.in time. Data from Table 4-2. Wellbore Effects 150 Table 4-2 Pressure Buildup Test with Aftertiow at a Pa Slope aPasat ct) the) (STB) est) (pstmey sin) RBIs ‘00000-92000 184485 1.787.500 1.988607 0.949 ons 90257 1.85087 ‘a8. 714 1.188.788 0.300 coors sae 1.85451 1381818 az8BK0 0.328 0107 8700.1 1,95007 475802 1.481204 0.07 orgs gears 186838 © ty406.887 1.502980 © 0.209, ores «asec 186781 579s 156728 © 0282 Oz 8480187697 555.172 1500129 0.82 ozs 9,908 1,985.99 1.505.085 1522570 0278 Oost; 791458 1.90867 11590.655 1.808703 0.266 ooa9 77008 © 1,908.80 1.887.981 1.500.102 082 00867 74969 t.atz72 1802273 489.642 0.260, 0054 71953 1.92584 © ta7rot1 1.482.389 0.258 006s © 6a002 99879 tas7.727asea3 OB O72 Ges2 1.96158 tyazeta8 © 1'388.527 © Ode 0.0807 8.3050 vgvest7 1geasto ote 0.0002 s.904.5, sonra 1273609 «0248 otto 5,880.8 2.00848 1.223077 1 188.823 0.258, o1si0 48172 202859 © 1.114.170 1oea.a8s © 08s 0.1557 41847 2086.41 11014800 952.733, O27 0.1807 35859 208178 © we0'8E7 2.27’ OS o2to7 276s 210850 758.187 sa2.245 ODP o2Ko 23788 213449 © 6283S SSN.27B ORS oes 18077 215087 «76.252 98.950 O24 oes 12511218534 sei.es8 | sB.7es 0280 oaos 7228 2ati49 = te5.91B 95.008 ©0205, ose 4708 © 2asze1 195.350 © 118.900 0208, dso ©5003] = 101.250 © 8B.80S 202 oerio 24a 2.24021 719550719 0.208 ‘ie rom Meu ta, 186, Figure 4-4 differs from the one given by Meunier et al. (1985) probably because of differences in the differentiation algorithms used to calculate 108 ‘The following definitions apply to Gringarten’s type-curves Wellbore Effects. 161 (Pi — Pas) for drawdown Pa. ~ Pata, for buildup 2.64 x 10~*k At_ 2xghCry* bur 5.615 2.95 x 10~kh at *C 5.615 Ce Coe" = Bron Gre In the above expressions, At isthe drawdown period. For buildup tests it is neoessary to replace the shut-in period, At, by At. Dimensionless pest, po chrviog evant le on + = 108 7006 7.000 Dimeroeane bm Se Figure 4-6. Type-curves for a well with wellbore storage and skin in @ reservoir with homogeneous behavior. Copyright © 1983 Werld Oil, Bourdet ot al., May 3983 [3 162 Wall Test Analysis Wellbore Effects 163, {Asan example ofthe use of Gringarten's type-curves, Table 43 shows the Table 43 presure-time data ofa buildup test. Additional reservoir and flow parame- Continued ters are: VS ——— i ae P baat b=lore = 174 STBID Ca = 2 a 6 = 0.95 B = 1.06 RBISTB ‘sa389 347s rT noe n= 250 | 50 serra 3500 a.co000 = ‘ Ss | 985 2350 0.63830, Cy = 4.2 x 10% pst t= 15 brs | 0.70833 995.64 2218 0.67639, 075000 907.38 2100 nies ayy osi50 © asiesa aa 20 1943077078 Pressure Bull 7 087500 aas73¢ ast 1316 ogasz7 ee eee eee nec 0.93750 3559.55 467.22 17.00 0.88295 00000 = S7i75 saa 1600 085750 fa Pow ap beat i 1195050 35082349990 is12obaaze ” co) (os at au ‘tes ganas eee 1433 toast (ccoomn aa tees oo = oon ters Sars; s.08, 1363 1.0009, 00m? 3000387 424 360071 0087 333000 gaits saa02 1300115368 000858 3083.81 748 e010? 900003, ‘ 131280 Sia88 S559 w43 120000 00125 3.00655 x0.22, 20100 «aia 1.37600 3.55285 585.52 Hor 125954 ootes7 3100031370 "S008? 001068, was7s0 3.86432 577.90 143131179 002083 «310327 «== teas 7ai12 aaa 50000 9.67881 Heo .98364 Gay pee ee Ge e500 3.00227 aas.ae 102346617 neigh erm teen teed cer 17000 3705521019 987 1 se716 ngessoh @ crane © cere ee oiecee etre 1s7s00 3719.28 ga2g0 900s seer oars © gtt642 308400 asras ‘ 200000 575223645. 90 so 176471 ee 225000 574071888 787 198652 0.0500 312248 ©8500 aaae, zarsoo 37874967088 732 208006 0.05830 «3128964269 258.20 «005007 250000 376344 77-1 700 214286 0.08657 33592 4050228908807 2ysooo 37740502 B45 2ea04 00700 Stai47 Sas 20100 «rasa ! 300000 3785.11 9878 800 2.50000 oes = ata7se tat eio1 —Ooaaer 25000 a7e408 70773 see 267125 000563 Stiss = 75e2— 8783 (Gee 350000 see 71947 520 2asTes 010683 517088 AS taba? O06 375000 Sango 728.17 500 3.0000 cies | surest | teem | | eee toe 00000 91597 Fang 47331680 oissss —Sta712 007s 1950 «tae 42so00 382020 7an87 483 331160 15s = ateuze§—orat jones —Oaaeo , som 962195 e562 490 gots o1g2s0 3208991089, 33106078 475000 8428.70 79787 416 360759 oie” 3216889035, 72017706 00000 gaz84s T4012 400 375000 019563822789 iat 38 79 O08 525000 92309 743.96 386 3.88869 O25 azar sae 7159 0.2053 550000 gas26e 48.31 373 4.02639 ozeoi7 — Sz4007 6274 esas 02572 ' 575000 9804.70.87 a1 4.15069, 9.2500 «32817917548 6100 0.24500 00000 © gas7.18 750.88 350 426571 o2ete7 328721 20.88 5243028011 S200 gaa 75261 340 4asi76 bee fam cas 675000 aaeoe—Fe.69 322 aesi7 037500 © aaaese aos dion Sanwa 728000 34078 saas Ror aavea ey ane e ooa aro ae T7000 38301808 204 5 10960 oes erage eee Br oats 825000 © Saease 758.19 2s 5.258, 050000 339804 081 310048307 ' ——875000_sete27_7soss e768 Ossie? 3,813.90 2757 zaso 052270 (table coined on net pgs) Table 43 Continued at Pre 3P erat oon (ws) si) at ES ‘925000 SBA7S1 761.18 Bee«S 72185 97500 36852 782.19 25e = 5.90900 1025000 sas001 765.68, 248 S0ott 1075000 385075 764.2 240 8.28214 1125000 38517676549, 233842087 4175000 385250 766.17 228858870 1225000 © sass 767.18 222 8 SI2 1275000385425 (767.82 218 680T89 1325000 385507 768.74, 213703540 1375000 3855.50 760.17 200 7.7301 145000 3.85850 770.17 203737268 325000 3.85725 770.92 498756198 100000 385788 771.68 tes 774104 1675000 «3858.74 772.41 19 791390 yrso00 3850.48 773.15, 196 Ba7es2 1825000 © 3e5988 778.68 ae2 823308 yon «3.86073 774.40, 179938255 3975000 © saso99 77488, 176 asa518 2050000 © 388149 775.18. 173 866107 2125000 © 38e22¢ 778.81 7 878310 222500 © 3.88274 778.41 er 898073 2325000 © 3.86522 778.88, 46581785 2425000 © 386348 777.18, tsa 9.26782 252500 © 38639977786 19 9.40004 2.25000 © e449 778.16. 387854605 272500 3864739 778.40 155 867486 2eso0 © 3865.23 778.90 183 9.82750 0.00000 3865.74 779.41 4150 __10,00000 ‘esp tor Boards ota, 1983 Figure 4-7 is Horner's plot. From the slope, m = 65.62 psileycle, and Pie = 3,797 psi, kh = 1,142 md-ft, k = 10.7 md, and s = 7.4. Figure 4-8 shows the buildup data plotted on log-log graph paper and ‘matched to the curve labelled Cye® = 10", Note that in Figure 48, Pesca and APna:s have been chosen to give the same kh value obtained from Horner's plot. The choice was made as follows: PpfAP = khi (141.208) 1,149/(141.2 174 x 1.06 x 2.5) = 0.0175 Wellbore Effects 165 T rapa ine paar ‘Sopem = 6582 piece Inept pt = 3870 pt Parse = B95 pL pate] he seams estat Figure 4-7. Horner's plot; data from Table 4-3. Copyright © 1983 World Ol, Bour- dot etal, May 1983 {3}. . Sang sagt ine site = me oe | ee ee ee Figure 4-8. Buildup data plotted on log-log graph paper and matched to type- carve by Gringarton etl. Conyrit © 4889 Were OF, Bord et a May 1868 166 Wall Test Analysis ‘This means that if AP is set at 100, then Pp = 1.75, which establishes the vertical match. The graph paper which contains the field data was then shifted horizontally until the early straight lines of the field data and the type-curves were aligned. A time match was then chosen at Atos = 1, (to! Co)aueh = 14.8, and (Coe*)mane: = 10, From the time match we get: (= 9.000295Kh — Atawn c ColCo)maa 0.000998 > 1,142 Snore tte ass) = 9.1058 ~ 3 5.615 x 9.105E ~ 3 Co SBE x IOT x 428 = 861 8 = 0.5 In(Cye/Cp) = 05 In (10")861) =81 Figure 4.0 loglog plot of AP vs At and AP vs At. Iti evident that the ‘owo plots are coincident, although the AP vs At, plots shorter than the plot of AP vs at, as the data of Table 4-3 show: Therefore, in this particular ex ample the producing time of 15 hours before shut-in was sufficient to allevi- ate the problem asoviated with the use of drawdown type-curves fr inter- pretation of buildup tests. In general, however, itis advantageous to always plot aP vs at, BOURDETS TYPE-CURVES (PRESSURE DERIVATIVE METHOD) Bourdet etal. (1964) introduced type-curves in terms ofthe pressure de- rivative, Their type-curves are, agin, largely based on the solutions ob- tained by Agarwal et al. (1970). During easy time, Agarwal et al. showed thet Pyp = toICp @s) Thus, dPyp/d(to/Co) Wellbore Effects 167 10° 10” é 10 wr Oe a 1 10 10° a Figure 48. Logog plot. Data from Table 4, 0, Pap = 1 16) ‘Equation 4-16 shows that during early time a plot of Pj vs (to!Gp) isa hor zontal line and the intercept is equal to 1. In a reservoir that is homogenous with respect to both rock and fluid properties and where the wells flly penetrating the pay sction, the flow must become radial after wellbore storage effects have subsided. Therefore, during late time, Equation 1-22 is applicable, assuming, of course, that the flow remains in the transient (infinite acting) state. Thus, rage SBE og 3 th 262.608 Nog ¢ + tog Ky — 8.9975 + os] 1860 age log gokg 0075 +0 160 6a |, 5:28 - 4 ke 2308 th |! ayta, +3] 168 Wall Test Analysis HPP) 9.5 fn Cp +10 Cp + In eM STET gb 7 0 Un 2.2858 to/Go + In Co + In) Pup = 0.5 [In (to/Cp) + 0.80907 + In (Coe*)] “an Thus, dPxp/d(to/Co) = 0.51(tp/Cp) ip = 0.5/(to/Cp) (4.18) and, Jog Pin = ~ log (to!Cp) + log 0.5 19) Equation 4-19 shows that during the transient state and after the wellbore storage effects have subsided, a plot of Pip versus (to/Cp) on log-log graph paper will be a straight line of slope ~ I. Equations 4-16 and 4-19 show ‘that during early time and during the transient state Pip is independent of Gye, At intermediate times Pip is dependent on Cpe" and could be evalu- ated from the solution of Agarwal et al. The resulting type-curves as given by Bourdet et al. are shown in Figure 4-10, where Pj is equivalent to Pi. Bourdet etal. then defined the derivative of the dimensionless pressure as follows: PiltpIGp) = At AP’Kh/(141.2qB)) (4-20) where At = producing time, hrs de, - Pupiat With the definition (Equation 4-20), Bourdet et al. redrew thelr type-curves and presented them as shown in Figure 4-11. ‘Tose the type-curves of Figure 4-11, the field drawdown data must be plotted on log-log graph paper as At AP” versus At. At early time, the data ‘wil fall on the straight line of unit slope and at late time during the tran- sient state the data wil fall on the horizontal line P’ (to/Cp) In the case of a buildup test, At represents the shutin time, and the buildup data are plotted as At AP’ (t, + At)/ty versus At Wellbore Effects 169 co SANNA _ SNe | ta 1 a eensionles tne, Ca Figure 4-10. Pressure derivative type-curve. Copyright © 1989 World Of, Bour- dot et al., May 1983 3}, Pelots) Dinersnies tne, Ca Figure 4-11. Pressure derivative type-curve in terms of P”y(elen). Copyright © 1983 World Oi, Bourdet et al., May 1983 [3]. 170 Well Test Analysis Wellbore Esfects IL Note that sinoe APye = P, ~ Pye and APyy ~ Pyy ~ Pega, @ functions of At and (tp + AD/At, respectively, then: aPay daar, dat | : Tinay dat * Trav vd a1 AR TT a : = AP jy At 7 2 AAP os dat 1 Tin (e+ avaa) Tin Ge + SHIRA] = ~ apy, at feta ' % ‘These results show that a plot of (APiqAt) versus At in the case of a draw- am oven is equivalent to a plot of d (AP.)/d (In At) versus At, where At here i Figure 4-12. Type-curve matching. Data om Tablo 43. Copyright © 1969 Wor the producing time. Likewise, in the case of a buildap, a plot of (at Of, Bourdet eta, May 1989 [3]. Polly + At)/G] versus Ati equivalent toa plot of d APw/A fin {(fy + At) 10” ‘At)] versus At, where At is the shut-in time, j ‘san example, we refer to the same buildup test data of Table 4-3, The dure is shown in Figure 4-12, ‘ In Figure 4-12 the data were plotted twice on the same log-log graph pa- per: AP vs At; and [At AP’ (fp + At)/t] vs At. There are two sets of type- ccurves on the same graph: the one by Gringarten et al. and the one by Bour- det etal. The pressure match is determined as before according to the results obtained from Horner's plot. ‘The data plot was then shifted horizontally until the late time data matched the horizontal line on Bourdet’s type-curves ‘and the early time data matched the 45° line on Gringarten’s type curves. ‘The calculations are carried out as explained before, but the Gpe* value is better read from the Bourdet type-curves. ‘Even though the pressure-time data of Table 43 are unusually precise and very closely spaced, the derivative function, plotted in Figure 4-13, shows ‘an appreciable amount of scatter, and the horizontal line which signifies the radial, infinite acting state is not clear. In the case of real ficld data which are subject to spurious pressure effects, the derivative function will show se- eS ‘ere oscillations unless the data are smoothed before taking the derivative. ' (text conta on pos 174) 8 pressure derivative fantion i given in Table 4-4 and the matching proos- | 1 i ] 72 Well Test Analysis Table 4.8 Wellbore Effects 173 Table 4-4 Pressure Derivative Method Continued Data of Table 4-3 a oP Slope Pe tar ‘at aPCSlope SPP BLP Gy + aN cc) (5) (psihy) (psi) oo) ee ws) ‘000000 ~—~000~« toe 192500 —«S1ee2 aN. HO Ms ‘so1e7 000817 424 77e gre Vie7s0 531.08 iges 25.60 ze. ones = 74869785 717.84 25000 58.82 15598 187.860 254.04 corso 02283483 748.04 191250 554.53 To1es 173.860, Ber oor6s7 1370 77aa5 808.60 137300 565.52 1835218788 e172 0.02063 «1884 30.33 08.00 yaa7s0 877.996.167.882 Deana 0.02500 04477808 08.15 150000 587.48 14769 (140.76 2ar10 0017 2308788577791 1162500 605.94 10800 126.84 22848 one 5927789877791 175000 619.19 1089210796 21087 003750016 5894 56798 4197500 6329310378 08.88 225.7 oases 53571042 50.18 punto; enane | rans | anes oss 0.0500 «381573072 75007 225000 663.98 Ss Gad 16788 005820 ©4263 631.54 906.19 237500 67086 5000582 151.09 Coss? 435963025 720.0 250000 © S771 4ah 7D 15831 007500 «5404 7767170548 275000 © 6aas2 41.84) 43.04 141.04 oesss 61s) 1144.80 960.76 3.00000 698783580 58.82 19975 ooses 756260840 921.60 325000 70773 2206 20.8, 116.7 0x08 88.95 61680 657.00 350000 7134738805088 13330 0.1208 9205 , 69840 857.80 375000 72317 saa aa 15150 3353 © s0078| 56860 634.00 400000 72968 = 18922140, 10843 04563 1079170808 68.38 4.25000 733.87 700 11.96 65.23 oes © 1186364077307 ‘450000 735.62 700 7.00 40.95 ox7e17 19095 o7e87 65797 475000 © 737371400 9.00, 56.20 oe583 14156 62867 650.77 00000 740.12 1298 11.08 7387 ots 15208 ge.87 69527 52500 743361480 12.38, 8778 022017 1627461085 626.26 550000 746.31 26 10.02 75.82 025000 17545 «610031034 375000 2310 qo38 029167 200.88 «55085 500.34 6.00000 7.00 348, m3 033s 223.82 «58051 565.58 6.25000 108 258 pose 0.37500 248.01 52628 853.40 8.75000 552 494 yor Oates? 26098 © aa. 487.09 7.25000, 546 499) 53.08 0.45803 288.65. 467.00 458.08 775000 756.68 302 374 43.98 50000 s08.11 467.00 467.00 825000 758.19 350 926 44.09 oseis7 S757 47840 472.70 875000 759.98 240. 209 ana 058393 947.50 94.25 409.82 925000 761.18 202 225 3365 ozs © 9617249701 399.19 975000 76219 «28 250 4022 068067 97993 «377.10 407.05 3025000 763.58, 148, 223 38.48 070893 305.64 2125 8.18 30.7500 764.42 202 175 3220 07500 40736 398.04 340.15, 125000 765.49, 148 175 3445 ogi2so 4323020088 40.20 4178000 766.17 200 475 2e87 og7s00 451.01 259.98 279.8 "225000 767.18 148. 175 3808 993750 $6722 201227528 1275000 767.92 184 188 26.80 90000 485.42 281.68 28t.48 1825000 768.74 0.86 125 3119 1196250 49.90 26752 240.60 tae contin on next pe) 174 Well Tet Analysis Table 4-4 Continued at ap ‘Slope BLaP + NM (oy (os) si) osimy co) 3375000 76037 138 10 28.80 #0000 770.317 1.00 a7 3327 15.25000 os oa 3055 16.00000 190 (O99 seas 1675000 772.41 0.9 0.99) 35.22 1750000 73:15 088 oss. 31.80 1825000 773.88, 099 (O83 sar 9.00000 774.40, 035 as? aan 1975000 77465087 st 2318 20.5000 775.16 190 oa 4043 2125000 77581 050s 38.82 2225000 776.1 04308 2r07 2325000 77688028 037 2198 242500 ©7715 Ot 038 244s 2525000 777.66, 050 050 3az2 2825000 778.16, 028 os par 2725000 © 77840040" 32h Base 250000 ©7890 os7 9088 o.o000 774i 2598 19:16 eae = (7a - T7e.fess 27.25) = 04 140+ 02g =038 27283082 (18 © 27.25)15 = 2456 (ood fom page 10) Smoothing of any time series, such as the pressure-time data, is not an ‘easy task, and unless itis done with care and know-how, @ portion of the data which is representative of the reservoir (signal) could be lost. Smoothing is, in effect, a filtering operation. The filter must be properly de- signed to remove spurious pressures (noise), but in the mean time leave the signal intact. Therefore, the use of commorly available moving average op- erators, although producing smooth pressure data, could be detrimental to ‘the information originally contained in these data. In addition, since any given moving average operator is finite in length, it could introduce other subtle forms of noise into the data even though the result would look smooth to the eye, Signal filtering, smoothing, and interpolation isa very advanced subject of science and engineering, and unless the proper smoothing tech- ‘niques are applied to the field data, the results could be utterly misleading. Wellbore Effects 17S HYDRAULICALLY FRACTURED WELLS Ac depths below 3,000 ft, te sbbelieved that hydraulic fracturing results in the formation of vertical fractures. At depths shallower than 3,000 ft, the likelihood is that horizontal fractures will be induced. Methods of analysis of ‘well test data in the presence of both horizontal and vertical hydraulic frac- tures have been developed. However, we will focus here on vertically frac- ‘tured wells. The reader who is interested in the treatment of horizontally fractured wells should consult Advances in Well Test Analysis by Earlougher as, ‘Two solutions have been obtained for vertically fractured wells. In the first solution, the infinite conductivity solution, it was assumed that there is no pressure drop along the plane of the fracture at any time. In the second solution, the uniform flux solution, it was assumed that the fluid entered the fracture at a uniform flow rate per unit area of fracture face, which implied that the pressure varied along the fracture plane. Both solutions were based ‘on the assumptions that the fracture cut across the entire pay section and ‘extended equally on both sides of the well. tis believed that the infinite conductivity solution often matches the be- havior of hydraulically fractured wells. In this case, the drawdown in the infinite acting state is given by: =u | at | BP ow 21 ‘where x; = fracture half length (see Figure 4-17) m1 4-21 is known as the linear flow equation. It shows that during 2, a plotof AP = Py ~ Py versus VF would bea straight line of slope my given by: sng = #06408 | an) letal a ‘Thus, Equation 4-21 can be written as: P= Pa= my VE (£23) As the time, , increases, however, the flow gradually changes from linear to radial. After the change, the usual semi-log analysis applies. 176 Well Test Analysis By taking the log of both sides of Equation 4-23 we get: Jog AP = 0.5 log t + log ma ‘and thus during the linear flow period a plot of AP versus ton log-log graph paper would be a straight line with a slope of one half (~27°). But the lope ‘of one half will not develop in the presence of wellbore storage, and in most cases by the time wellbore storage has subsided, the linear flow period ‘would have ended Inthe absence of wellbore storage, it was found that the semi-log straight line which indicates radial flow, develops at to,, = 3, 2.64 10-4 vi tg = 2OAX1O-t (2 oa gaat and hs t= HT 5 twas also found that at the beginning of the semi-log straight line the fol- owing condition must be met: AP ig = 24P a, (4-26) where AP, = AP at the beginning of the semi-log straight line ‘AP = AP at the end of the linear flow period Asan example, Table 45 lists the pressure-time data for a drawdown test, and Figures 414, 4-15, and 416, respectively, are the log-log, linear, and semi-log plots. Additonal reservoir parameters are: h=ger oe 012 C= 21 x 10-Fipsi = 0.65 op. B, = 1.26 RB/STB P= 3,770 psi ‘The objective isto find ks, and x. Figure 4-1 shows a straight line with a slope of one half, ehus indicating 8 linear flow that lasted les than 0.3 he, and no wellbore storage effects. From the slope of the straight line of Figure 4-15, my = 36 psfhr!®, and ‘AP, = 30 ps. Substituting in Equation 4.36, we get APs 2 60 psi. Now we are able to examine Figure 4-16 and fit the semi-log straight line as shown, cc) ‘0.088 01670 0.2500 ‘5000 07500 1.0000 2.0000 3.0000 ‘4.0000 ‘5.0000 6.0000 7.0000 ‘8.0000 9.0000 10.0000 12,0000 24.0000 48.0000 96.0000 20,0000, ‘aga on Gagaren a 1075 weeren, a Wellbore Effects 0.08 ats 2.440 2.086 2028 3.000 3.162 3468 4.950 6.928 9.708 wee Figure 4-14. Log-og plot, dravidown test; data from Table 4-5, a7 178 Wall Test Analysis Wellbore Effects 179 beginning at AP = 60 pst. The slope of the semi-log straight line is found to 200.00 = 1 be 94.1 poeycle, Thus, So ) c= 102.6419 x 1.26 x 0.65, ‘o.oo 41x BE : =7.28 md z 1 [6.0646 $00.00 By Equation 4-22, x? iI el a 1 [4.064 x 419 x1.96F 0.65 | . 78 [86x RB Tix ax 50.00 — 18,851.3 Ft w=i373 it oa Note, however, thatthe extmation of xis very sensitive tothe reservoir pa- oO 5 10 at 6 a cae rameters. For example, if B, is taken equal to 1 instead of 1.26, x would be ‘equal to 108 ft instead of 137 ft. ‘Thus, one has to keep in mind that mis Figure 4-15. Linear plot, drawdown test; data from Table 4-6 ‘correct only if all the reservoir and fluid parameters are correct, { Having calculated k and x, we substitute in Equation 4-15 to calculate t, 200 i the time of the beginning of the semi-log straight line, which is 48.6 hrs. i ‘Therefore, the beginning of the semi-log straight line must meet the follow: | ing criteria: 150 | AP = 60 psi t= 48.6 his Since the straight line tn Figure 4-16 satisfies these two criteria, it must be { ‘concluded that itis the correct straight cree era : ‘To estimate the skin factor, s, we determine APy., on the semi-log straight line, not from the data table. This is done by: LT ~ AP ie Togl0= log (Pan, psi = 94.1 50) AP, = - 224 pst By substituting in Equation 1-95, we find s = ~ 5.5. a a fi fl TF TP ‘Note that the analytical procedure which we followed to determize k and. Le 1% is based on the assumption that the well was producing in the infinite ‘Somiiog plot, drawdown test; data from Table 45. ‘acting state. In the event that the semi-steady state has been reached or even. 180 Well Tet Analysis approached, the analytical procedure cannot be applied, and one must re- sort to solutions presented in the form of type-curves. Gringarten et al. (1974) presented the type-curves shown in Figure 4-17. ‘These typo-curves are based on a square drainage area with the well at the Wellbore Effects 181 Table 46 Pressure Buildup Test (Game Well as in Previous Example) center, and the fracture is parallel fo one of the sides. What happens ifthe A fa Seas drainage area is not «square and ifthe fracture s not paralll to one ofthe — FO sides? Unfortunately the literature does not provide any answers. Therefore, ; Sess ena 7 cara ‘we eould use Gringarten’stype-curves, bt we must be cognizant of the as- oer Sasso 1 erat somptions implied in their construction. 2500 Saceo ® sree s000 Saees des erat 7500 Saws Es eras ‘0000 aaceo @ erase ca reer 2060 Sass % east 3 Daanage | nate cond ) ‘000 cara @ toot é os — 30000 aca & fet ee 80000 Basso eso & 2000 S00 % fet a eo ) 0000 Sas60 % fess alg eed tr a= 20000 3000 % ear ig cS 7 10.0000 S000 & 21 2 opine sar 120000 S060 & ese ‘ ‘of semiog 20000 S200 18 2355 é Edo near tow ° ‘some asso 188 3386 peed ors , 2400000 e200 m0 har om : pe ‘oe - pe 200teae rex? ) 0 Figure 4:17. Type-curves for an ifnite-conduetvty vertical ration. Copyright F —S (©1974 SPE, Gringarten eta SPEY. Aug, 1974 (10) Watch pine joo = 100 Sh pn = 1.20 4 312 TOR, to = 088 J Asan campson the we of te ypmcunve of Figure it, wenferto =) ‘Table 4-6 which shows buildup data for the same wel of the preceding ex. ey : ample. The wells located in an undeveloped reservoir, and was closed after 5 Cae Naan ‘roctcng for 7,600 hs, Figure 8 shows the mating poets ood é ata at ote the coordinates of the match point. From the definitions of ty and Py, ) ‘eproe. start which are given on Figure 4-17, and from the coordinates of the match eFsemiog | point, t ean easly be verified thatthe same reals obtained froma the draw 7 az tows 10 [Sine 1 - own test are also obtained from the buildup test by typecurve matching. = Ttis clear from Figure 4-18 that the well remained in the transient stato a i 4 0 0 throughout the 7,800 hes of production. Accordingly, we could analyze the ® test the same way we analyzed the drawdown test Inthe case of a buildup, ) Figure 4-18. Typo-curve matching, Data trom Table 46. Copyright © 1674 SPE, however, we must modify Equation 4-21 by superposition to make it suit- Gringarten et al., SPEJ, Aug, 1974 [10} 182 Wall Tet Analysis able for buildup test analysis, Accordingly, for buildup th following equa- tion applies: 4.064g8 [1 - r Pan mia a) [Fat VE] rr) But, P,~ Paty, toe et G = constant ‘Thus, by subtracting Equation 4-27 from the above equation, we get: Pee TH 6HB | Mast0 bh keCa Pera vm + constant (4-28) According to Equation 4-28, during the linear flow period, a plot of [P= Priggaol Versus [Vt + At - Vt] should be a straight line of slope mj If, however, tyislaige such thatt, + At ~ t, then the plot can be made versus Vt only, sinoe in this ease Equation 4-28 reduces to: Pas ~ Bafgaay = Bie VEE + constant (4-29) Equation 4-29 shows that a plot of AP = Py, ~ Pigg. Versus VAt on log- log graph paper gives a straight line of lope of one hal. PARTIAL PENETRATION AND PARTIAL PERFORATION If the well penetrates the reservoir fora very short distance below the cap rock, then the flow would be hemispherical. When the well iscased through ‘thick pay zone and only a small part of the casing is perforated, the flow in the immediate vicinity of the wellbore will be spherical. Away from the ‘wellbore, the flow will again become radial. But if the testi of short dura- tion, the flow will remain spherical throughout. Tin the case of a pressure buildup test of a partially completed well, the representative equations are given by Culham (1974) as follows: Weltbore Rffects 183 (4-80) 14 a ie Fal (4.31) b = thickness of zone open to flow Pa: = the pressure at any At, as read on the straight line Equation 430 shows that a plot of P, ~ Pyy versus [LAVBt ~ ip 8] would be a straight line of slope, m, where, In case of hemispherical flow, m = 1,226.5qBulk°*, and ty = blla(2bir,). ‘OTHER WELLBORE EFFECTS In packed oil wells producing from reservoirs of medium permeability (les than 50 md), gas segregation inthe tubing during shut-in can cause & pressure rise over and above the reservoir pressure. ‘This pressure rise is usu- ally followed by a gradual pressure decline, which is then followed by the proper semi-log straight line on Horners plot. This phenomenon docs not ‘ecur when the permeability of the reservoir is high, because the ol can eas- lly be pushed back into the reservoir. It also does not occu im low permea- bility reservoirs because the production rate is usually small and thas there is ample space for the searegated gs to move into and expand, Likewise, it ddoes not occur in unpacked wells because the annulus provides the needed space forthe gas to expand. “Tubing leakage and leakage around the packer which separates two zones of different pressures can also cause pressure distortions which reult in a buildup curve that is similar to one obtained from a heterogeneous rservoir. ‘Therefore, rather than attempting to estimate and remove the effeds of the leakage from the pressure data, the test should be repeated after cerrecting the situation in the wellbore. 184 Well Test Analysis ‘CONCLUSIONS ‘The main conclusion that could be made is that for given test data in which q remained constant, the following set of plots should be generated automatically without regard to the usefulness of each plot. This way the interpreter would have the full view of the test readily available, 1, Sem-log (AP vst or AP vs log(t, + At)/At, or AP vs At) 2, Logelog (AP vs tor At) 3. Linear (AP ws vi, or vs vi + At- VAR) 4, Spherical (AP vs L/v/t or vs [L/V&t- (vt + 38) 5. Rectangular (AP vs ton rectangular graph paper) ‘The reason for using Agarwal's At, for type-curve matching of buildup data is that type-curves were made for drawdown and not for buildup, and unless the rate of decline of Pye at the time of shutting the well in is very smnall, the interpretation by type-curve matching would be erroneous. On the other hand, if the producing time becomes large, the error will dimin- ish. But in general it is good practice to use At, when curve matching a buildup test without regard to the duration of ty If the 45? straight line develops on the log-log plot, it strongly suggests the presence of wellbore storage effects in the data, hence application of the 1.5 ‘eycle rule becomes necessary. If, within the range of the 1.5 eyele, the log- log plot shows a straight line with a slope of one half, that straight line can- not be interpreted as an indication of linear flow. Its illogical to apply the 1,5 cycle rule on one hand and then attempt to interpret the data contained ‘within the 1.5 eyele on the other hand. This is a common error which the author has seen made repeatedly even by trained interpreters. ‘The removal of wellbore storage effects from a given drawdown or buildup data is attainable, in theory, by deconvolution (see Chapter 11). ‘Any other methods like those proposed by Bourdet and Alagoa (1984) and by Yeh et al, (1980) are only valid if the reservoir consists of a single layer which is homogeneous with respect to all description parameters Finally, in spite of the theoretical attractiveness and recent popularity of the pressure derivative method, it should not be used in the interpretation of field data without extreme caution, and the reliability of the results should be considered quite low. Since the derivative tends to amplify any pressure noise contained in the data, smoothing of the pressure versus time data prior to the taking of the derivative, is necessary: Such smoothing, however, can ‘add distortions to the field data, and can lead to erroneous conclusions. Fur- ‘thermore, in the case of high permeability reservoirs the rate of pressure change during a given test can be very small, such that the derivative be- comes meaningless. 10, LL 12, Wellbore Effects 185 REFERENCES Agarwal, R.G., Al-Hussainy, R, and Ramey, H. J, Jy “An Investiga- tion of Wellbore Storage and Skin Effect in Unsteady Liquid Flow: T. Analytical Treatment,” Soc. Pet. Eng. J., Sept. 1970, pp. 279-290, ‘Trans. AIME, 249. Agarwal, R. G., “A New Method to Account for Producing Time Ef- fects When Drawdown Type Curves Are Used to Analyze Pressure Buildup and Other Test Data,” SPE Paper $289, presented at SPE- AIME 55th Annual Technical Conference, Dallas, Texas, Sept. 21-24, 1989. Bourdet, D., Whittle, T. M,, Douglas, A. A, and Pirard, Set of Type Curves Simplifies Well Test Analysis,” World Of pp. 95-106, Bourdet, D., Alagoa, A., Ayoub, J. A. and Picard, Y. M., ‘New Type Curves Aid Analysis of Fissured Zone Well Tests,” World Oil, Apr. 1984, pp. 111-124 Bourdet, D. and Alagoa, A., "New Method Enhances Well Test Inter. pretation,” World Oil, Sept. 1984, pp. 37-44 |. Culham, W. E., “Pressure Buildup Equations for Spherical Flow Re- gime Problems,” Soc. Pet. Eng. J., Dee. 1974, pp. 545-555, /. Earlougher, R. C., Je, Advances in Well Test Analysis, Dallas: Soc. of Pet. Eng, of AIME Monograph Series, Vol. 5, 1877, pp. 155-150. Gringarten, A. C., Ramey, H. J, Jr. and Raghavan, T., “Applied Pres- sure Analysis for Fractured Wells” Jour. Pet. Tech., Jul. 197, pp. 887- 802. Gringarten, A. C., Bourdet, D. P, Landel, P.A., and Kniazeff,V. J, Comparison Between Different Skin and Wellbore Storage Type- (Curves for Early Time Transient Analysis,” SPE Paper 8206, presented at SPE-AIME Sith Annual Technical Conference, Las Vegas, Nevada, Sept. 23-25, 1979, Gringarten, A. C., Ramey, H. J. Jr, and Raghavan, B., “Unsteady- State Pressure Distributions Created by a Well with a Single Infinite- Conductivity Vertical Fracture,” Sac. Pet. Eng, J. Aug. 1974, pp. 347— 360, Meunier, D., Wittmann, M. J., and Stewart, C., “Interpretation of Presure Buildup Test Using In-Situ Measurement of Afterfow:” Jour. Pet. Tech, Jan. 1985, pp. 143-152 Yeh, N-S., Agarwal, R. G., and Fuss, D. D., “Method to Correct Pressure Transient Data Influenced by Wellbore Storage, Effect,” PC No, 89, Presented at the EGPC Production and Exploration Confer ence, Cairo, Egypt, Nov. 2-24, 1988. 5 Gas Well Testing In this chapter we will study gas well testing, We will ee that with a sim- ple modification practically all of what we have learned in the preceding chapters is applicable tothe interpretation of gas well test data, providing that we are testing dry gas wells o condensate reservoirs in which the con- densate remains immobile. If the condensate is mobil, then the methods ‘presented here ae not applicable, and well test interpretation raust rey on reservoir simulation in which the relative permeability is changed until @ match is obtained between the test data and the performance of the reser- voir model. Another way of testing condensate reservoirs in which the con-

Das könnte Ihnen auch gefallen