Вы находитесь на странице: 1из 6
Modeling Variations in Annular Pressure Due to Fluid Injection by Fredric C. Amold” Abstract Monitoring of the liquid pressure in a casing-tubing annulus confined by the wellhead at the surface and by a packer above the injection interval isa simple and inexpensive method of detecting the oss of mechanical integrity in wells used for disposal of liquid wastes. However, variations in annular pressure which occur in response to changes in injection rate and injected fluid temperature complicate the interpretation of annular pressure data. The variation of fluid and formation temperatures depth, the expansion and contraction of fluids and metal tubing with pressure and temperature, and wellbore depth and ‘geometry, have been combined in a model of annular pressure response to fluid injection. The response to changes in injected fluid rate and temperature have been simulated with the model and verified by field experiment. Although the response (oa change i vary with time and a true steady-state was not attained. Introduction Injection into saline aquifersis the primary method for disposal of liquid hazardous wastes. It was estimated that 423 million gal (1.3 million m’) of hazardous materials together with 10 billion gal (38 million m') of water were injected through 181 wells in 1983 (U.S. Environmental Protection Agency, 1985). To protect current and potential sources of ground water, wells for disposal of Class I industrial wastes are required to inject through tubing within an annulus isolated by a packer (Figure 1). Operators must demonstrate the absence of leaks in the casing, tubing, and packer by peri- dic mechanical integrity testing. A generally accepted method to show mechanical integrity is to monitor the hydraulic pressure applied to the liquid-filed annular volume. Ifthe temperature ofthe systems stabilized and the system is not subjected to changes in injection pressure, a constant annular pressure demonstrates mechanical integ- rity. The cost of annulus pressure monitoring is very low and equipment requirements are minimal. In addition, monitor- ing of annulus pressure does not require that the well be removed from service ‘A limitation in utilizing annulus pressure monitoring as ‘4 means of assessing injection well mechanical integrity is that interpretation of monitoring data is complicated by changes in injection pressure and injected fluid temperature. The University of Oklahoma, College of Engineering, School of Petroleum and Geological Engineering, Energy Center, Suite T 301, Norman, Oklahoma 73019-0628, Received August 1990, revised January 1991, accepted February 1991 Discussion open until May 1, 1992 832 injection rate or temperature is greatest inthe first two hours following the change, annulus pressure continues to An increase in the rate of liquid injection is accompanied by increasing pressure within the tubing and will cause the tubing to expand and annular pressure to increase, Con- versely, ifthe rate of injection through the tubing is dimin- ished, annular pressure will decline Increasing annular liquid temperature within a con- stant volume will result in an inerease in annular pressure. Ascalculated by Wilcox and Langlinais (1984), the pressure ina liquid-filled annulus may increase several hundred psi in response toa 10 degree Fahrenheit increase in injected liquid lemperature, Conversely, annular pressure may decrease significantly if a cooler liquid is injected. In their modet of annular pressure response to heated fluid injection, Wilcox and Langlinais (1984) utilized a ‘model of wellbore heat transmission developed by Ramey (1962). The assumptions incorporated in this model, a line heat source, steady wellbore heat flow, and radial flow of heat in the surrounding formations, are approximations which diminish the utility of the model. As noted by Ramey (1962), when representing a finite-diameter well by a line heat source, “The convergence time is on the order of one week in many reservoir problems” (for negligible error due to the line source approximation) Since changesin injection conditions routinely occur in a matter of hours or even ‘minutes at many disposal wells, a model of annular pressure response to heated fluid injection should respond accurately ‘to changes in injection conditions aver short time intervals In this paper, a new model of the annular pressure change resulting from variations in the volumetric flow rate and surface temperature of injected liquids is described and validated by field tests. The model can simulate any depth, wellbore geometry, and geothermal conditions desired and readily incorporates changes in temperatures and flow as they occur. Vol, 29, No, 6 GROUND WATER November-December 1991 ‘aug onniton ning —-s nn oie — Fig. 1. Schematic of a typical Class | industrial waste disposal well Mathematical Model Injection of liquid through tubing isolated from the annulus by the wellhead at the surface and a packer set above the injection interval is shown in Figure I. The tubing, and casing are radially concentric with respect to the well centerline. Liquid enters the top ofthe tubing ata specified temperature, To, and exits through the perforations into the disposal zone. In the absence of fluid injection, the temperature in a liquid-flled annulus will equilibrate with the surrounding formations and increase with depth at a more or ess con- stant rate, the geothermal gradient. When liquid is injected through the tubing, thermal energy is conducted between the injected liquid and the surrounding formations, heating ‘or cooling the intervening annular liquid Ifthe temperature ‘of the confined annular liquid increases in response to injec- tion of liquid through the tubing, the annular pressure will increase; conversely, cooling the annular liquid reduces the pressure in the annulus ‘Changes in annular temperature and pressure will cause the injection tubing to expand or contract lightly and alter the annular volume, Compression ofthe tubing due to increased annular pressure or reduced injection pressure and expansion of the tubing caused by decreased annular pressure or increased injection pressure result in radial and circumferential stresses. The magnitudes ofthe stresses may be calculated from appropriate equations for the mechani- cal properties of metal tubing and combined to quantify the change in tubing diameter and annular volume. Inthe following paragraphs, the formulation and solu- tion of the energy balances used to calculate changes in injected fluid, metal tubing, annular fuid, and formation temperatures are presented first. The calculated change in annular liquid temperature is subsequently used to quantify the magnitude of the annular pressure change. Lastly, the pressure change is corrected for changes in the tubing diameter and annular volume which occur in response to radial and circumferential stresses induced by annular and injection pressure variations. Temperature Variation Accompanying Liquid Injection To model the variation of temperature with time and position, it was assumed that there are no radial temperature gradients in the injected and annular liquids and that trans- mission of heat is controlled by film coefficients at the interfaces with the metal tubing and casing, The correlation of Seider and Tate (1936) was employed to model heat transmission from the injected liquid tothe tubing; Dropkin and Sommerscales’ (1965) correlation was used to model heat transmission between the annular lguid and the tubing and surrounding formations. Liquid properties (heat capac- ity, density, and thermal conductivity) were assumed to vary insignificantly with temperature; heat generation by viscous dissipation and chemical reaction is absent. With these assumptions the equation describing the variation of injected liquid temperature with time and depth is, at; at meer + Arke e+ a oF at T)= Are wo meu : on Similar equations describe the temperatures of the ‘metal tubing, annular fluid, and surrounding formations as functions of time and depth. at tks Gar + 2atiohio(Ts — 1) + g 2mruha(Tr~ Te) = Awrer = @ at Aaks SS + 2nriohio(Ti = Ta) + oe Anrche(Tin ~ Ta) = Aapata @ atk a at ar beet 2) = pc a oe ae Ca) meme ae ‘The variables and parameters appearing in the preceding equations are as follows: Ay, Ac, and Ay are cross-sectional ‘areas normal to flow of the fluid within the tubing, metal tubing, and tubing-casing annulus respectively, fts¢F, 1, and cim are the specific heats of the injected liquid, metal tubing, annular liquid, and formation, respectively, Beu/Ib- °F; bi, hio, and he are the heat transfer coefficients at the 833 interior of the tubing, tubing exterior, and formation face, respectively, Btuyhr-ft”°F; ki, kr, ka, kr, and ke are the thermal conductivities of the injected liquid, metal tubing, annular liquid, formation inthe radial direction, and forma tion in the vertical direction, respectively, Btu/hr-f-°F: my is the mass flow rate of injected fluid, Ib/ he isthe distance from the wellbore centerline, ft; i, ttoand rare the interior tubing radius, exterior tubing radius, and radius to the formation, respectively, ft; Tr, Ti, Ta, and Tim are the temperatures of the injected liquid, metal tubing, annular liquid, and formation, respectively, °F; ts the time, hr; zis the depth below the surface, ft: and pr. pt. pa. and pimare the densities ofthe injected liquid, metal tubing, annular liquid, and formation, respectively, Ib/f. ‘To calculate the change in temperature over a small time inerement, equations (1) through (3) were represented as numerical analogues by substituting second-order implicit finite differences for the vertical derivatives, second-order explicit finite differences for the radial derivatives, and a first-order finite difference for the temporal derivative. This results in transformation of each of equations (1) through 3) into a set of simultaneous linear equations whose tri diagonal coefficient matrix is easily inverted. The equation for variation of formation temperatures, equation (4), was transformed to a logarithmic radial grid and resolved by the alternating direction implicit technique at each time step. Pressure Change Due to Heating of a Liquid Ina Confined Space When the temperature of a liquid confined within a fixed volume rises, its pressure will increase; conversely, ‘when the temperature declines, the pressure will decrease. ‘The magnitude of the pressure change is calculated in to steps. First, the change in volume of the confined annular {uid that would result from the change in annular tempera- tureat constant pressure is calculated, Second, the change in pressure required to return the annular liquid to its original volume while maintaining the temperature constant is determined. ‘The volume ofthe liquid relative to that ata standard temperature is correlated in terms ofthe cubical expansion coelficients of the liquid (Perry eta, 1984), a= Vo(I — 6427 10° ATs + 8.5053 X 10 ATE = 6.79 10 ATs) 0 ‘where vo and v, are the annular fiquid volumes, 1, at 32°F and the annular liquid temperature, respectively; and ATs is the annular fluid temperature, °F, minus 32°F. When, asi confined annulus, the liquid is unable to expand, the pressure must increase to offset an increase in temperature. If heating of the liquid from temperature Ts; to Ta would result in an increase in liquid volume from vs to var, the corresponding change in the pressure of a confined liquid, AP, bv/in?, is I APs = — lav va) © 834 where the liquid compressibility, cw, 1 /psi,is given for water bby (Osif, 1984) ew = 10/(ay* pat ars Tatayese tas) (1) where a= 7.033 as a= 403.3 X 10" (8) and ps, Ts, and sare the pressure, temperature, and salinity of the annular liquid, respectively, in Iby/in2, °F, and gi Expansion and Contraction of the Tubing ‘The expansion and contraction ofthe tubing will result in small changes of the annular volume. Since even small variations in confined liquid volume will resutin substantial changes in liquid pressure, these changes are taken into account in the model, The isothermal changes in the radial and hoop stresses are given by Lames equations (Hearn, 1985) oy =~ oaPe Friapy ” (oA) don = ~ {fodPo + rap) _ oon) where oy anddos ar the radial and hoop stresses, respec- tively, tby in Apo and Ap, arc the changes in pressure on the external and internal surfaces ofthe wbing, respectively, lbylin-s and ro and rare the external and internal radi of tubing respectively, inches. For nonisothermal changes, the radial and hoop streses are increased proportionally to the temperature change, ATs, in the metal tubing i Baal Bo = bo, + SOOT ay ») Boy = Ao, + ST (1%) 21») where Boy, and Sop are the nonisothermal radial and hoop stresses, respectively, Iby in.*; E is Young's modulus of elas- ticity, Ibs jin; » is Poisson's ratio of strains, dimensionless; «avis the coefficient of thermal expansion of the metal tubing, 1/°F; and AT; is the change in temperature of the tubing, °F. ‘The radial and hoop stresses are used to calculate the diametral strain, en, the change in diameter divided by the unstrained diameter. 1 — = = Flos v1 + B97} (3) CSP) ar 2b — ») aa) Aon + «aay The pressure and temperature induced changes in the annu- lar volume are proportional to twice the diametral strain, Change in Internal Flowing Tubing Pressure ‘Due to Injection Rate Changes ‘When liquid is injected through tubing uid pressure is required to attain the desired injection rate. Changes in injection pressure will alter the radial and hoop stresses given by equations (9) and (10). ‘The change in pressure inside tubing at depth zfor a well of total depth D when the flow rate is changed from q; to qris(Wileox and Langlinais, 1984) Ap. = fA(qi,D) ~ A(quailae ~ tAG@iD)~ Aza? + Bir) 5) where A foe 6 (92) ~ 3593550" i) and B «in In equation (15), the change in injected fluid pressure, Pi, psi, is funetion ofthe injected fluid viscosity, centipoises: fluid density, p, Ib/f1; well radius, rf; formation perme ability, k, millidarcies: formation thickness, hf and equi- librium radius re, ft. The Fanning friction factor, f, dimen- sionless, at flow rate, q,f¢"/hr, in smooth tubing of diameter 4, ft, was determined by use of the Colebrook equation (Colebrook, 1962). Combination of Calculation Steps To commence the calculations, the annular pressure and the temperature and flow rate of the injected liquid were specified; the temperature distributions of the injected liquid Within the tubing, metal tubing, annular liguid, and sur- rounding formations were also needed. At each time inter- val, 0.01 br was a usual time-step size, the fnite-ifference analogues of the thermal energy balances, equations (1) through (4), were solved to obtain the temperature distribu- tions at the end of the time step. The annular temperatures at the beginning and end of the time step were used to calculate the annular liquid volumes, equation (5) at each nodal point in the annulus; the nodal values ofthe liquid volume were integrated over the length of the annulus to calculate the annular liquid volumeat the beginningandend ofthe timestep. The change in injection pressure with depth as calculated using equation (15) and this change was wed to calculate an initial estimate ofthe diametral strain, equa tion (14). An iterative loop was used to calculate liguid compressibility, equation (7), annular pressure change, equation (6), and diametral strain until successive iterates agreed within a prescribed relative tolerance, This com- pleted the caleulations for a single time step; subsequent time steps repeated the calculations. Fleld Testing Several demonstrations ofthe buildup of annular pres- sure accompanying heated fluid injection were conducted at the East Central University/ U.S. Environmental Protection Agency test well located II miles west of Ada, Oklahoma. ‘The well is 1,080 ft deep from surface to packer. Fluid is injected through 2)-inch stainless steel tubing within 5)5- inch steel casing. Prior to the tests, the annulus was filled with water, allowed 10 equilibrate for several days and a temperature survey was run, Geothermal temperatures extrapolated to the surface gave a surface temperature of (60°F and an average temperature gradient of 0.008°F /f Annulus Pressure Response to an Increase In Injected Fluid Temperature ‘To commence the test, the annulus was shut in and a hydraulic pressure of 100 psi was applied. Water for injec- tion was heated to 83°F in a 4,200 gal fiberglass tank. Injection commenced at a steady rate of 22.9 gpm. Annulus pressure was monitored at the wellhead by Bourdon gages encompassing pressure ranges of 0-200 psig, 0-300 psig, and (0-600 psig. Injection temperature, injection pressure, and injection rate were continuously indicated by gages located near the pump outlet, Readings were made at Sto 1S minute intervals. ‘The injected water temperature was maintained at 83°F during the initial phase of the test. After injecting for 2.5 hours, the injection water tank was nearly empty. A second 4,200 gal tank was filled with water, heated to 88°F, and flow was redirected from the second tank to the injec- tion well. The changeover time between injection tanks was negligible in this test. When the second water tank was nearly empty, flow was stopped. The annulus was shut in, and the decline in annulus pressure was monitored for an additional 40 minutes The annular pressure profile observed during the field test is shown in Figure 2. Annular pressure rose continu- ‘ously during the first three hours of injection of water at ‘83°F, The pressure increase was greatest during the first wo hours of the test, 183 psi; during the next hour, annular pressure continued to rise although more slowly, increasing an additional 30 psi. When the injected water temperature ‘was increased to 88°F, the rapid rise in annular pressure resumed reaching a maximum of 461 psi at the end of injection. Although the rate of pressure increase slowed during the last hour of injection, pressure continued to tise throughout the test and steady-state was never attained ‘When injection eased, annular pressure declined as accumu lated heat was dissipated to the surrounding formations. The simulation of the field test with the numerical model is also shown in Figure 2. The model accurately tracked the pressure increase in the annulus during both of | the periods of heating by the injected fluid, When injection ceased, the predicted decline of annulus pressure closely followed the field observations, A second test of annulus pressure response to changes in injected fluid temperature was conducted by Thornhill (1989), Water heated to 72°F was injected for one hour and 835 00 naoresr 300 ANNULUS PRESSURE, PSIG 200 O00 rewres: [ANNULUS PRESS 400 200 a ee CLOCK TIME, HAS Fig.2. Response of annulus pressureto.a change in injected liquid temperature. the annulus pressure was monitored at 1S minute intervals. Following anoninjection period of [5 minutes during which the injection water was heated to 82°F, injection was resumed for an additional hour, Additional eycles of heating \ere followed by injection of water at temperatures of 92°F, 102°F, and 139°F. The annular pressure measurements are plotted in Figure 3, Annulus pressure increased rapidly with each increase of injected Muid temperature. The short cycles of heated fluid injection did not permit the annulus pressure to stabi- lize during any of the cycles. When injection was stopped to heat the water, annulus pressure declined prior to recom- ‘mencing injection, ‘The agreement between the model pre- dietions of annulus pressure and the field observations is again excellent Annulus Pressure Response to a Decrea: in Injection Rate A third field test was undertaken to demonstrate the effect ofa change in injection rate on annular pressure. Prior 1 the tes, two 4200 gallon tanks were filled with water and allowed t6 equilibrate fortwo days. The temperature of the injection water was 87°F atthe start of the test. ‘The annulus was pressurized with water to 51 psig and injection commenced at arate of 22.6 gpm. The buildup of annulus pressure due to heated liquid injection was abetted by ambient heating ofthe water in the frst injection tank; as the test continued, the temperature of the injection water climbed t0 94°F. After two and one-half hours, the source of the injected water was changed tothe second tank. This was 826 — > ee CLOCK TIME, HAS Fig. 3. Effect of injection temperature on annulus pressure; simulation and field test results. accompanied by a decrease ofthe injected fluid temperature 10 87°F and a decline in annulus pressure. Flow was con- ‘tinued at the original rate, 22.6 gpm, for 45 minutes before reducing the injection rate to 6.5 epm. ‘As expected, a reduction in the rate of heated liquid injection resulted in a decrease in annulus pressure. As the quantity of heat introduced by the injected liquid decreased, hheat was lost from the liquid inthe annulus to the surround ing formations at a rate exceeding that at which it was replenished. The dectine in temperature of the water in the ‘annulus within a confined volume was responsible for the decrease in pressure; a decline in the stress on the tubulars at the lower injection rate also contributed to the decline in pressure in the annulus. The annulus pressure data are plotted in Figure 4 along, with the pressures predicted by modeling. The simulation adequately predicted the increase in annulus pressure build- up and the decline when changing tanks. The prediction following the reduction in injection rate was less satis- factory; since the ambient temperature exceeded 100°F uring this segment of the test, heating of the injected water may have occurred between the temperature indicator and the wellhead, Conclusions ‘A model of the pressure response of a liquid-filled annulus to changes in injection conditions has been devel ‘oped. The model accounts for changes in injected liquid temperature, injection rate, annular fluid compressibility, and expansion and contraction of the injection tubing to soo 400 ‘00 ANNULUS PRESSURE, PSIG 200 O00 re.otes 100 Lock TIME, HRS Fig. 4. Effect of injection rate on annulus pressure; simula. tion and field test results. calculate changes in annular pressure. The following conclu- sions resulted from field tests and simulations with the model I. Increases or decreases in injected fluid temperatures resulted in rapid increases or decreases in annulus pressures. 2. Changes in injection rate, particularly commencing for ceasing injection, resulted in a change in annulus pressure. 3. The change in annular pressure was greatest in the first two hours following a change in injection conditions. However, temperatures and pressures within the injection system continued to change and a true steady-state was never attained, 4. The model can simulate any depth, wellbore geome try, and geothermal conditions desired. The model was validated for changes in injection rate and injected liquid temperature by field testing in an experimental wel, ‘Acknowledgment ‘The work contained in this paper was partially sup- ported by a grant from East Central University under @ cooperative agreement with the U.S. Environmental Pro- tection Agency, The author appreciates the assistance of Bobby Bennefield and Robert Caldwell of East Central University and Jerry T. Thornhill of the U.S. Environmental Protection Agency in the conduct of the field tests. This ‘manuscript has not been subjected to the peer review pro- cess of the U.S, Environmental Protection Agency and does not represent the policy of this or any other agency of the Federal Government References Colebrook, C. F. 1962. Turbulent flow in pipes. J. Inst, Civil Engineers, London, England. Dropkin, D.and E. Sommerscales, 1968. Heat transfer by natural convection in liquids confined by two parallel plates which are inclined at various anges with respect othe horizontal J. Heat Transfer. pp. 7784, Hearn, E. J. 1985, Mechanics of Materials. nd ed., Pergamon Press, Oxford. Osif,T.L. 1984, The effects of salt, gas, temperature, and pressure ‘on the compressibility of water. Society of Petroleum Engineers Paper 13174, Presented at the 59th Annual Con- ference and Exhibition, Houston, Perry, R. H., D, W. Green, and J. O. Maloney, editors. 1984 rs Handbook, 6th ed. McGraw-Hill few York. pp. 3101-3116, Ramey, H. J, Jr. 1962, Wellbore heat transmission, Journal of Petroleum Technology. v.14, pp. 73-81 Seider, E. N. and G. E. Tate. 1936, Heat transfer and pressure ‘rop of liquids in tubes. industrial ané Engineering Chemistry. v, 28, pp. 1429-1438, Thorakill J.T. 1989, Personal communication. U.S. Environmental Protection Agency. 1985. Report to Congress ‘on the injection of hazardous waste. Office of Drinking ‘Water, Washington, DC, EPA 570/9-85-008, Wilcox, T.C. and J.P. Langlinais. 1984. Waste injection: testing for mechanical integrity. Water Resources Bulletin. v.20, pp. 53-60. Fredric C, Arnold is Assistant Professor of Petroleum and Geological Engineering at the University of Oklahoma. He received his Bachelor's and Master's degrees from the Ohio State University and his Doctorate from the University of Minnesota, all Chemical Engineering. 837

Вам также может понравиться