Application of Real Gas Flow Theory to Well Testing and Deliverability Forecasting 2 ALHUSSAINY JUNIOR MEMBER AME HE RANEY, Je MEMBER Ane ABSTRACT Previous gas well rest analyses have been based mainly ‘upon linearizations of ideal gas flow results, although a ‘method for drawdown analysis based upon real gas flow results has been proposed. Linearizations based upon ideal gas flow require estimation of gas physical properties at Some sort of average pressure, ond implicily involve the assumption that pressure gradients are small everywhere in the reservoir. A new real gas flow equation has been developed by means of a substitution which couples pres sure, viscosity and gas law deviation factor, This sub- stitution has been called the real gas pseudo-pressure, Use of the real gas pseudo-pressure leads ta simple ‘equations describing real gas flow which do not con- tain pressure-dependent gas properties, and which do not require the assumption of small pressure gradients every here in the flow system. Equations required to determine flow capacity, well condition and static formation pressure jrom pressure drawdown and build-up tests with the real gar pieudo- pressure concept are presented. Also shown are applica- tions of the real gas. pseudo-pressure to backpressure testing, the gas materials balance and rigorous determina tion of average gas properties for previous gas flow equations. Included are sample calculations for well test analyse, INTRODUCTION A recent study’ of the flow of the transient flow of real gases in ideal radial systems showed that it is possible to consider gas physical property dependence upon pressure by means of a substitution called the real 25 psoudo-pressure. The substitution has the advantage thet it enables engineering solutions for steady and transient flow of real gases that are moze aceurate and general than those previously available. The solutions are particularly important for the case of gas flow in tight, high-pressure. formations under large drawdowns. Even for more ideal flow conditions, the real gas pseudo- pressure concept leads to important rules for finding average gas physical properties pertinent for older well test analyses. Ref. 1 provides a detailed study of the theory behind the real gas flow correlations to be used in this paper: Oringel manucrint ceed im Soigy of Panaeum Exaineat ofr ogee dbl Reveal psuaton ot SPS. a8 tee Tabs sted af Sa" Antal Fal Mesiog Reda Bowes “Presently onthe fact at Stanford U. "Retorenes given at end of peer. TEXAS Agu U. COUEGE STATION, Tex ‘The purpose of this paper is to provide necessary engi- sneering forms for use of the real gas low results, and to llusiate applications. ‘The following considers flow of real gases in an ideal radial flow system. I is assumed that: (1) formation ‘thickness, porosity, water saturation, absolute permeability, temperature and gas composition are constant; (2) gas compressibility and density are functions of pressure. as described by the gas law py = 2aRT; (3) gis viscosity is a function of pressure; (4) effective permeability 10 ‘gas may be a function of pressure to account for liquid condensation; (5) condensate is immobile; and (6) the formation has no dip. SUMMARY OF REAL GAS FLOW EQUATIONS Combination of the continuity equation and Darcy's law for radial ow of a real gas yields the equation: o_o). 2] e w wR) Or] te If the lefttiand side of Eq. 1 is differentiated, « term jnvolving @y'/2n)" arises which does not occur in ideal 35 flow equations. Thus, modification of ides! gxs flow Solutions to fit real gas flow involves the assumption that presie gradients aro small. An alternate procedure is to make @ change of variable in Ea. 1- We define the reel gas pseudo-pressre m(p) as mm @ As shown in Ref. 1, substitution of m(p) in Eq. 1 leads to: Fmtp) ,. 1 ame) _ suede) dt) ® acer econ Pore wriec Eq. 3 looks like the diffusivity equation, but is still non-linear because the diffusivity depends upon pressure, ‘or m(p). Eq. 3 also closely resembles the equation for the flow of an ideal gas. Furthermore, the dependence ‘of the diffusivity term upon m(p) for the seal gas is very similar to the dependence of the ideal gas diffusivity upon pressure squared. As is shown in Ref, 1, this leads immediately to practical solutions for Eq. 3 which are similar to the Aronofsky-Jenkins’ ideal gas flow solutions. Engineering units will be used throughout the following For production of a real gas in an ideal radial system At constant rate with a closed outer boundary: 245;1.987 10" kT se aa [: mime.) rey 2n(E) © 0.000264 ke BUC ‘The m(p) is evaluated at an average pressure determined from a materials balance: (2),- (BY. TPG th z wire gS Tae Z(uedelp—m The palte) és the van Everdinger-Hurst? dimensionless pressure-drop function, For large ratios of r./r.: J (ues) dmc) o Elia 44+0,80807] , tor 100 < te << oat fees = I =) torn >t 3) re) 3. Bh, hy WE apie te ete cei 28 drotnaln pod’ T a oie ase o eoektne “he toa tem eampeiyin Ea (e)memaly can be ial pal oth go toon gcse Tn vow ofthe fing tht production of seal pn ean seat tin "end son Bey to iSO Sod tore ste boul eae pa aT Tr toste+ostg] 2... 08 Xf standard conditions are taken to be 14:7 psia and 60: mip) ~ 5.792 x 0° (Pu) [tose + 03513 (Pui) = mp.) ~ 1,637 Tioga ts + 0.3513 + 0.875 i +0879] Gee Oi ‘And for long flowing times: HAD oat rao = 1987 x 104 >a] a2) In the preceding equations, s is the familiar skin effect, and (Dg) is an approximate non-Darcy flow term. The non-Darey flow coefficient D is inversely proportional to 246 gas viscosity, and slightly dependent on time for very short transient flows. However, D ean be considered constant as an adequate engineering approximation. APPLICATION OF REAL GAS FLOW EQUATIONS ‘The following describes application of the real gas flow equations to common cases of gas well flowing testing (drawdown, build-up and back-pressure testing). DRAWDOWN TESTING Eqs, 10 or 11 provide the basis for drawdown test analysis for real gas low, From Eg. 10, it is apparent that 2 plot of mip.,) vs the logarithm of the producing time for constant rate production should produce a straight ine of slope: mbesmax ee ay Or for standard conditions of 14.7 psia and 6OF: -b=167% b= ort 7 - 09 Eq, 14 is analogous to the equation for analyzing build-up and drawdowns for gas wells presented by Tracy, except neither viscosity nor z appear. The striking feature of all ‘of the real gas flow equations presented in the previous section is that pressure-dependent viscosity and gas law eviation factor are absent. Procedure is as follows. Drawdown data are taken as in the past; measure sandface producing pressures as a function of producing time for constant race of pro- duction. Next, produce 2 working plot of m(p), psi'/eps vvs pressure in psi for the reservoir. If specific viscosity and density data are available, this can be done by graphical integration or the Trapezoiéal rule using Eq. 2. If viscosity and zfactor correlations are normally used, ‘m(p) can be found without integration by reading either Table 1 or Fig. 2, Ref. 1, Drawdown pressures are converted to appropriate m(p) values and plotted ys the logarithm of drawdown time, and a straight line is passed through the data. Either Eqs. 13 or 14 can then bbe used to find the flow capacity, An example calulation is given in the Appendix. In general, production of an mip) vs p working plot for a given reservoir requires less effort than squating pressures as in current methods, The plot, of course, muy be used for all wells producing from the same reservoir if formation temperature and gas composition does not vary widely, Eqs. 10 or 11 can also be solved for the skin effect and non-Darey flow coefficient. This result is (p.) ~ mba.) Fare pyenist (MDM) k peace sae a Basan * 323] vee It is necessary 10 have drawdown tests at two flow rates to determine both the skin effects and the non- Darcy flow coefficient D separately, As described by Ramey," wellbore siorage effects or any variable-rate test, can often be used to separate the skin effect and non Darey flow. The real gas potential drop across the skin ccan be found from: : AMP) win = O8T(—BS es 18) Flow efficiency can be found by means of Eqs. 12 or 13:FE Jay oy = ED etd — OP lB) ~ Mle) jn analogy with the expression presented by Matthews.’ ‘The preceding for drawdown analysis for real gas flow should be very accurate. In many cases, conventional plotting of pressure or pressure squared leads to plots which are not linear, although the m(p) plot is linear ~ as it should be. One example given in the Appendix ‘yields reasonable straight lines for both m(p) and pressure- squared plotting (Fig. 2), although a plot of drawdown pressure vs log time has considerable curvature, In other eases, it has been observed that che pressure-log time plot and mip}log time plots are excellent straight lines, although a pressure-squared plot shows considerable curva ture, Tn all eases tried 10 date, the m(p)-tog time plot has produced reasonable straight lines. Often, it is necessary to analyze a variable-rate draw- down—such as Russells! two-rate test, of to account for wellbore storage effects.” This requires. superposition of constant rates to generate variable-rate cases. Ref. 1 indicates that superposition ean be used as a satisfactory approximation in many cases. An important question at this point is whether the real gas flow solutions offer a worthwhile improvement over existing gas flow solutions currently used in well test analysis? In addition to the problem of gas physical property variation, use of the ‘eal gas flow solutions offers an important improvement over existing gus flow theory that is particularly important for low pecmesbility for- ‘mations, That is, the solution in terns of real gas pseudo: Pressure includes the second-degree pressure gradient ferm neglected by methods based on either liquid flow ‘or ideal gas flow solutions, In the ease of low-permeability gas well testing, current well test interpretation methods can lead to formation capacity or deliverability estimates that ean be grossly in error. ‘This can be illustrated by comparison of flow eapacity determinations by various methods for a known capacity eystem. Excellent data for testing various methods are available from Carter's? finite difference solutions for real gas flow. Carter proposed a method to Gnd flow capacity which was based on a correlation with a dimensionless time wherein viscosity and compressibility were poing time functions. Thus, it is possible to compare flow capacities developed from his method, the liquid case analogy and from the real gas flow analysis of this paper. Table 1 presents the comparison. As can be seen, Carter's analysis, leads to flow capacities which aze high, the liquid flow analogy leads to flow capacities which are generally low, while the real gas flow analysis based on mip) values leads to capacity estimates in very close agreement with the actual values. Note particularly the good agreement for the Solution 3 case for 8 md-ft capacity. The dit- ference between Carter's analysis and the liquid case a7) Cattaneo ashe SHE Nees) ceooey i) analogy is a general résult. As flow approaches that of an ides! gas, the two methods and the real gas flow ‘method yields tho same answer BUILD-UP TESTING ‘The pertineot equations for analysis of pressure build-up jn terms of the real gas pseudo-pressure are: low capacity = mtpys) = m(p,) — + DoT Flow capacity = m(p..) = m(p,) — 5.792 X 10" a as) (Or using standard conditions of 14.7 psia and 60F: rc) ie EI at win io) 909 ho ‘Thus, a Homertype’ plot of m(p.s) should yield a straight Tine of slope (—-6), and formation flow capacity can be determined from Eqs. 13 or 14. The skin effect and non-Darey flow coefficient can be desermined from Eq. 415 if the real gas potential difference is replaced by m(Pix).* Qua). The flow efficiency can be determined from Eq. 17° if m@p.,) is substituted for mip.,). The pressure drop across ‘the skia can be computed from Eq, 16 as for dravidovrn. In a completely analogous manner ¢o previous build-up theory, Eqs. 18 and 19 indicate that che plot of mlpe) can be extrapolated to infinite shut-in time (a time rato of unity) to yield m(p*). The Matthews-Brons-Hazebooek™ pressure correction charts can be used 10 comect the extrapolated m(p*) value to m(fI, the average real gas potential, if the ordinates of their pressure correction charts are changed to: 20) 21) where 4 is the drainage area in square feet If producing time 1 is long enovgh that the drainage radius is stabilized before the well is shut in, the best approximation for pressure build-up mp. = mp) — Pon Plucdr? ee aieeeie it oe 0.00266 Fr, @2) Tho slope of a plot of mlpes) vs logs ts yields the flow ‘capacity, while the m(p) can be read on the straight line at a buildup time 4 of; @ ‘This is analogous to the Dietz" method for liquid flow. BACK.PRESSURE TEST ANALYSIS Application of the preceding to back-pressure test anal- ysis is straightforward, but several features deserve men. tion. First, Eq. 4 provides an excellent match to both short-time isochronal flow tests and stabilized flow test, ‘Tho skin effect and the non-Darey flow constant can bbe obtained from short-tims isochronal flow testing if a single drawdown of build-up is available to obtain the 247flow capacity xh. Once this is done, ra can be placed equal to 0.472 r., and either Eqs, 4 or 12 used to generate & plot of logy {m(p) ~ralr~,)] Vs logu @. ‘This is analogous to tho familiar backpressure curve The mip) curve has many of the characteristics of a formal back-pressure curve, but there ate some important diferences, also. At low rates, the non-Darcy term Dg Will be negligible and the slope of the curve will be unity. At high values of flow rate, non-Darey flow may become important and the slope of the curve will ap- proach 2 (n value of 0.5) in & fashion similar ¢o thet described by Carter et al However, the m(p) back pressure curve is only slightly sensitive to sitie pressure level, beeause only the constant D depends upon pressure through dependence upon gas viscosity (Swift and Kiel) Thus, a single, stabilized back-pressre curve in terms of the real gas’ pseudo-pressure difference can be used to sgnerato an eniite family of backepressure curves in terms of pressure squared—or used in terms of mip) with the ‘gis materials balance to provide a simple means of fore: fasting gas well performance. AVERAGE GAS PROPERTIES Current engineering practices in the analysis of gas well tests involves evaluation of an equivalent liquid flow system (Matthews, ‘or modification of expressions for flow of an ideal gas (Tracy* or Carter’), Matthews ree~ ‘ommended that gas properties be evaluated at an arithmetic average pressure, while Carter found empirically that ‘evaluation of physical properties at an average pressure ‘equal to the square root of the average of squared pres- sures gave reasonable agreement with real gas flow soli tions. One important result of the prosent study is that use of the real gas m(p) leads to a solution for the proper average value for gas physical properties, For example, it is ‘not uncommon to assume that gas vsicosity and ‘gas law deviation factor ean be regarded as constant at some average value for the flow region, For a bounded system, this would lead to the flow equation: = 1987x102 al PoT(pz)ong 10 F (24) Where we have neglected skin effect and non-Darey Row. If Eq. 24 is compared with Eq, 12, the result is mF) = ripe) ++ 25 or 6) A similar result will hold if pressures are expressed in terms of p, rather than p. Carter’ suggested the use of an equation similar to Eq. 26 for determination of average Properties. Since. many gas reservoir flow tests can be analyzed by current theory, Eq. 26 may bs of much use. Note however that the (12)... would be used only in determination of flow capacity from the usual equation— not in determination of dimensionless times. Dimension- less times should still be evaluated with physical properties at the initial formation pressare before production. If Matthew's liquid case analogy is used, an average value of (wB,) is required for flow capacity determination. A liquid flow analogy to Eq. 12 is: 248, 4 + @ Comparison of Eqs. 12 and 27 leads to: 2B pos), Te tn). a seegeeepat age a Bd 9) tne) m1 ‘Again, the average value of viscosity-formation volume factor’ product would be used in determination of flow capacity, not in determination of dimensionless. times. Extension of the averages to the transient period is ‘obvious in view of Eq. 4 It should be emphasized that the liquid case approxi- ‘mation recommended by Matthews, and Tracy's method of gas well build-up analysis, will’ give excellent results in many cases where pressure drawdown is not large ‘and permeability is high enough that second-degree pres: ‘sure gradients are not important. Nevertheless, it is rec- ommended that a real gas flow analysis is useful, even in this ease, to determine that the current approximate methods are applicable, Tn any event, it is likely that the real gas flow methods outlined in this paper will be found simpler in application than older methods once a master plot_of m(p) vs p is prepared for any given reservoir. This results mainly because it is no longer necessary to square pressures for plotting, and gas physical Properties appear only in dimensionless times, An ad= ditional benefit is that gas properties are always evaluated at known pressures in the real gas flow method. CONCLUSIONS 1. As a result of a change of vatiable called the reat ‘gas pseudo-pressure, it is possible 40. Write approxi solutions for the production of real gases from ideal radial systems, which are analogous to the Aronofsky- Jenkins ideal gas flow solutions 2. The approximate solutions for transient flow of a eal gas using the real gas psevdo-pressure leads to ‘methods for interpreting pressure drawdown and build-up tests which are similar ¢o current methods, except that gas physical properties either do not appear in the engi- Hering equations or appear at known pressures. Use of the real gas flow analysis indicates there are many well tests which ean be analyzed with current procedures, but there are likely to be many others where current’ pro- cedures are in error, 3. The real gas flow approximation can be used to establish proper average values of physical properties for current procedares, Finally, it is emphasized that the real gas pseudo pressure concept is principally a computing device, Re~ sults of back-pressure, build-up and drawdown tests ean be converted from m(p) solutions 10 solutions in terms of pressure oF pressure squared to provide familiar infor- mation and displays. NOMENCLATURE A= drainage area, 9q ft B= formation volume fastor, res vol/sid vo! b= slope of build-up or drawdown plot, psi/ep- cyclecompressibility psi* non-Darcy flow constant, (Msef/DY* cumulative gas produced, sof net formation thickness Productivity nde, (st/Dy/unit potential dit aaPtn ke = effective permeability ‘m(p) = real gas pseudo-pressure (Eg, 2) psi/ep P= pressure, psi alte) = van Everdingen-Hurst dimensionless pressure drop ‘gas rate, Msof/D radial location van Everdingen-Hurst skin effect, dimensionless apparent skin effect, dimensionless fractional gas saturation temperature, “R time, hours| real gas law deviation factor, dimensionless (pv=neRT) total poresity, fraction pore volume Viscosity of gas, cp SUBSCRIPTS AND SUPERSCRIPTS mensionless drainage exterior boundary gas Fiat fone hour on straight tine base superbar, average standard conditions of pressure and temperature sand face fe OR vad ze tal Ww = well, nner boundary wf = well flowing wp = producing before shutein w8 = shutin well + = extrapolated avg = average ACKNOWLEDGMENTS ‘The authors gratefully acknowledge financial support of the Texas A&M U. and the Texas Engineering Ex- periment Station of Texas A&M. The encouragement of P. B. Crawford and R. L. Whiting is also appreciated, Portions of this work were done by Al-Hussainy in partial fulfliment of the requirements toward a graduate degree at Texas A&M. REFERENCES 1LAbStusciny, Ry Ramey, H. Jo Jes and Cruylord, Ph “The Flow of Heal Gases Through Porous Media, Jour. Pe. Tech (Oy, 1966) 2 2.Aronolcky, J. S. and Jenking, Fes “Unsteady Radial Flow, of Gea" Por Nasa" Jur, pls Mech CD38) 2, Svan Brerdingen, A. P. and Hurt, W. the LaPlace Frinsfornation ip Flow P ‘Frans. AINE: (1939) 186, 205. 4.Swifty G. W. and Kiel, 0. G.: “The Prediction of Gas Well evlormance Including’ the Eitect of Non-Darcy Flow", Jocr. Pet, Tech, aly, 1962) 19 S.Trecy, C. Was “Why Gas, Wells Have Low Productivity Oita Gas Jour (Ave 6, 856) 8. G.Ramney, H. J, Ina *NonDarey Flow and Wellbore Storage “The Application of blens in Reservoin”, Eifecs in. Pressure Buildup aed Drawdown of Gas Wells Jour, Pet, Tech (Feb, 1965) 223. T.Matthows, C S.z “Anelysis of Pressure Buildup and Flow Test Data”, Jour. Pet Tech, (Septy 1961) 852. 8 Russell, D: Gut “Detepnination of Formation Characteristics from TworRate Flow Tess", Jour, Pet, Pech, (Duy 1963) BR. 9. Carter, R. Ds *Soluions of Unsteady State Radial Gos Flow", Jour, Bet Tech (ay, 1962) 543 ~ 210.Matthene, CS. Brons, F. and, Hanebeork, Paz “A, Method for Detetwlaation “of Average Presaure ia’. Bounded Res trvoie™ Trans, ATME, (1958) 201, 122 1. Diets, D. Ne: “Determination of ‘Average Reservoir Presse From Baildsip Survey. Jour. Pet Teck (Asp. 1365) 585. W.Canes, RD, Mille, S.C and Riley, H. G.: "Determination of ‘Stubiland "Gas Well Perfomance from Short Flow Tost, Tour: Pet. Peck. (Tune, 1963) 831. 1B Gorter, R. Dr Supplemental Appeniix to. Determination of Stabiled’ Gas_ Well Performance from Shore Flow "Toss, ADE Doc, No. 227, Library of Congtes, Wathingto, D.C APPENDIX DRAWDOWN ANALYSIS EXAMPLE An isochronal flow test is performed on a gas well at two different rates. Given the reservoir data and fluid properties below, determine the flow capacity, kin effect and non-Darcy flow coefficient for this well, The well is completed with tubing-annulus packer. Reservoir and Gas Data 2,300 psia tore osit 2,980 ft (640-acre spacing) 130F Out av 77 per cent PY Gas Properties Phosia) 400 800, 1,200 1600 2,000 2400 Flow No. 1 Flow No. 2 Flowing Time _(@=1,600 Msef/D) (q=3,200 Msef/1) __thours) ne (Bsia) Pov (Psi) | 0232 Tass 7105 04 1836 1020 06 ale 954 08 1/806 906 10 1797 860 20 1758 700 40 1723 338 60 1703 387 SOLUTION It can be assumed that wellbore storage effects ure negligible, since the well is completed with a down-hole packer. The first step is to find sntp) vs p for this gas. 289T T T T eer mes Fool. -q Erol. ron 0 Bo _| a Poi Fro, Lm(p) xo 2(p/ys) v5 p ron Sanerix Daawoows Proaiets, This can be done with the gas properties tabulated above and Eq. 2 in the main text. The quantity 2 (p/,2) can be calculated and plotted vs pressure, ax shown below and on Fig. 1. Integratiod can be performed in a tabu lar calculation by reading mid-point values of 2(p/p2) from the graph and multiplying by Ap. The’ computed ‘m(p), psit/ep, is also shown on Fig. 1. This curve can bbe used for future tests. with this weil or other wells producing the same gas at the same formation temperature, Often, only gas gravity is available, In this case mie) can be found without integration from Ref. I. 2, Sfp), Meme Om afar) mit » 2 eke? wl 2 2h Gh FRE) Ply foo 099 oorzs saa.a0a sarees deo Waextor araxtor NEO Ge 00102. zhteis Wee 2D Fora I TaROx toe too 0.20 ores soma Zeaase 40> Wiss ior aiaoelee 200 O81 cow Saa.nie Singco 400 Iarextor aaroctee In the following, Flow 1 will be analyzed in detail. Results for Flow 2 will also be given to illustrate the importance of a second flow test at a different rate, Fig. 2 presents the drawdown data plotted in the con: ventional manner, and in terms of the m(p.,) ‘The flow capacity for Flow 1 can be estimated from Eq. 14 in the main text: ar (1,600) (590) _ ar GEX 1) ‘The total of skin effect and non-Darey flow resistance can be estimated from Bq. 15. Kh 48.4 mat. 637! nlp. ¢ “+ Da uustf 7 ~ hee(sgtios) 305] = 250 Aa i | ee. ours ss 0, aa scooamrammrasy) * 323] ~ 90 ‘The flow capacity determined from Flow 2 is 45.8 mé-ft, and the skin effest plus non-Darcy flow com- ponent is 1.36, Thus Flow 1: 6+ Dg, = 0.657 1.36. Bs Flow 2: s+ Das These two equations can be solved for 9 = 1,600 Mscf/D, and q,= 3,200 Mscf/D to yield 3 —0.03, and D * 4.39 x’ 10*"(Mscf/D)*, Thus, «he skin effect is negligible and all of the resistance near the well is ‘caused by non-Darey flow. The difference in flow capacities found above for the two rates is not significant. This sample problem was taken from ¢wo computer solutions by Carter” because it provides a good example of an apparent skin effect caused by non-Darey flow. The (rue flow capacity used by Carter in the solution was $0 md-ft. The difference between flow capacities determined above and the true value of 50 mdft results because Carter approximated the effect of non-Darey flow as a constant pressure-squared difference for his solutions. In real gas flow, a better approximation would be a constant difference in m(p) ‘The next step would be to substitute the values of kh, $ and D into Eq. 12 in the main text to provide a general equation for generation of stabilized deliverabili