You are on page 1of 12
SPE 11954 SPE Seco Perna gener AME A Graphical Method for Estimation of Production Response From Cyclic Steam Stimulation Using Past Performance Data by raj Ershaghi, M.S. Al-Adawiya, and V.0. Kagawan, U. of Southern California Members SPE-AIME 2 Tonae 35006 UBM, Tles T8585 SPEBAL ‘Ts 5p was presenta he 8m Arnal Tet Contre at Exon ald in San Fae ABSTRACT A graphical solution method is presented whereby the production and injection data from a number of steam stimulation cycles for a given well are used to estimate recoveries from future cycles. The proposed method was established based on the experimentation with the analytical model, originally proposed by Boberg and tantzl, It requires a log-log plot of cumulative oil production vs. cumulative steam injection, The technique allows the computation of a response factor for individual wells undergoing cyclic stimulation in a given reservoir. INTRODUCTION In recent years considerable progress has been made in numerically simulating thermally enhanced oil recovery processes. However, there are numerous instances where the use of a sophisticated computer model may not be justified, Simple models are often of tremendous help in situations Where field engineers wish to make reservoir and economic evaluation studies on cyclic stimulation jobs without the availability of all pertinent reservoir rock and fluid properties. A review of literature shows various attempts made by many authors to present practical solution methods to this problem. Jan important characteristic of a simple model should be its ability to use the past response data and predict the future Performance. A good example is the use of decline curves in the estimation of reserves for conventional operation, Unfortunately, the decline curve method is not applicable to steam stimulation jobs. A realistic approach should be the one that would consider the effect of successive heat injection into the formation. In 1966 Boberg and Lantz! presented a technique which included many of the essential aspects of the steam stimulation process. Included were the estimation of radius of the heated zone, heat losses and the gradual viscosity changes during the Production cycle. The technique presented in | this paper is based on a modified form of the | Boberg and Lantz model ina form adaptable to the use of the production data from previous cycles to project the recoveries from future cycles. | Description of the Process During the steam soak process a mass of lsteam (N,) is injected into the formation and the well'is shut in and allowed to soak for la specified period of time before being put lback on production. Recoveries from steam stimulation depends on the simultaneous lvariation of temperature, pressure and fluid lsaturations. A sufficient soaking period is lallowed for the pressure to stabilize and for lthe steam to be eliminated from the heated zone. This will allow the steam zone to lexpand and the steam to condense. The response of the well is a function of crude | loi viscosity, the radius of the heated zone, lcormation permeability, initial reservoir lpressure, structural position of the well, RAPE, METHOD FOR ESTIMATION OF PROD, RESPON: STEAM STIMILATTON USING PAST. 1 FROM CYCLU damage removed by heat and invasion efficiency of the steam. The maximum increase in production is obtained at the instant that production begins following the soak period. The improvement in production rate steadily declines reaching a value similar to the unstimulated case. Repeat of the process will result in resumption of production increase. As shown in Figure 1, the area under the production curve for each subsequent cycle gradually decreases. Prediction methods attempted in the past have mostly concentrated on the estimation of the production rate following the steam injection. In this paper, we will focus on the cumulative recoveries from individual cycles. Modified Bobera and Lantz Model ‘A computer code was developed to use the Boberg and Lantz model for various case studies. To account for the gradual drop in oil saturation and its consequence on relative permeability to oil, the following equation was used for estimation of J prior to each cycl 0.00708 hk + Kro a B oA (Sctln The relationship used between Kro and So was either a hypothetical model of the form: kro = (atbsw)™ or data from published literature on thermal recovery operation. Figure 2 shows some typical relative permeabilities used in the study. Average oil saturation before each new cycle was estimated from the following material balance equation TMre%n S0;-5.615(Cum.0i1 Prod) (50) teat = — that rene ‘The model was run for various cases of reservoir rock and £luid properties. Sensitivity studies were conducted on the effect of relative permeability curves, rate of injection, drainage radius, colé oil viscosity, reservoir sand thickness and initial Je. Results of the computations were plotted on a log-log scale and are shown in Figures 3 through 8. A unit slope is evident on all the plots. the performance factor which is defined as the intercept of these plots are listed in Table I. Performance factors computed here represent the average oil/steam ratios for the individual cases. ‘The unit slope indicates a relationship of the typer Zo exists between the cumulative oil produced and the cumulative steam injected. The intercept (m) which was defined above as the performance factor is a complex function of reservoir rock and fluid properties mS My o The explanation for the unit slope based on our experimentation with the modified form of the Boberg and Lantz is discussed below. Estimation of Cumulative Recovery Cumulative recovery from the first cycle can be approximated by the following integration of the gn equatio *p Boa JP ao, at ow le Based on Boberg and Lantz formulation: ~ cls An re/ew ~ 0.5 An(re/rh)-0.5 + cas An ce/tw ~ 0.5 0.00708 hrk kro Mo (Sctin — a TRAJ ERSHAGHT, MOSA AL-ADAWTYA, AND VINCENT D. KAGAWAN. ‘and x, is the radius of the heated zone given by the followng equaton: Be © 4K ty - Ty) ty Ny Now consiéering typical fie1a applications where Tyr tyr Be Bggr Xjr Bege her and f, may be assumed confeant, “f becomes edsentially a function of the Mer For most cyclic application, the amount afd duration of injection is relatively constant. Thus one may assume that throughout the life history of a well, approximately the same given distance ty is being heated by Successive injections of stean. The average temperature in the heated zone during a cycle reaches to Tavg, which gradually cools down to Tavgy. The oil viscosity will range from on, to Mohz. Since the purpose of computation is the cumulative production rather than the instantaneous rate, average values for oh may be inserted into equation (2). Furthermore, the average Kro during a cycle is a function of the saturation balance during that cycle. The expression for the incremental oil produced during the second cycle is also similar to equation (1). For a similar radius of heated zone and the sane temperature and oh history during the subsequent production cycles, one would expect similar incremental production as the first cycle, except, for the depletion of oil saturation and static reservoir pressure in the heated area. In order to observe a constant slope, this reduction in Kro must be compensated by the lower “oh during subsequent cycles, because of buildup of residual heat. Eventually, when the oil Saturation drops to some low value, a flattening of the curve may occur because of a drop in Kro. As long as the flattening of the relative permeability has not affected the incremental oil, a constant proportionality exists between the cumulative production and Kro. Kro can be expressed as (a+bSw)" where a, B and n are constants. This means that log 20 = A loglatbSw) where A is a constant. The studies conducted using the Boberg and Lantz model show a linear trend between the S, at the end of each cycle and the cumulstive steam injection at that time. Pigure § shows the relationship for three runs. Runs dand @ correspond to different set sof relative permeability curves. Run ¥ uses the same base data as run ata different injection rate. Assuming negligible Sg and expressing so 1-Sw = Soj- m'. Mg where m is a constant and substituting into the cum.plot equation: log ZO log a+ log (@ #1 ~ 50, + wz) log A+ log (E+ mt Hy) where f= a +1 ~ 80; 2 tog Bg = log A+ log G+ ats Since m'ZM, is much larger than 1, log Z OQ = log A+ log (az Hy) = log A + log at + 10g Z My this ehows a unit slope expected for the Log- log plot of 30 vs. Z My. The plots can be made either vs.2M, or 30, where Proposed Technique Versus The Conventional Oil/Steam Ratio One convenient method of describing the response of a given well to cyclic steam stimulation has been the use of oil/steam ratios. This ratio is sometimes computed for incremental oil production and often for the total oil production during the cycle. As discussed by Burns’, oil/steam ratios per cycle have been reported in a wide range of 0.21 to 5 (for incremental oil) and 0.38 to 6.5 for total oil. Studying the oil/steam ratios for various cycles of a given well may be both a source of confusion and a constructive way of characterizing the individual well's response to steam injection. The oil/steam ratios reported for various cycles show variations which may be because of non-reservoir related factors such as short cycle length or operational problems. Foy example, data reported by Rivero and Heinz® on seven cycles of a well in the Midway Sunset field in California indicated oil/steam ratios ranging from 0.88 to 1.42, Table II. The variation in reported oil/steam ratios from one cycle to the next may hamper the use of this indicator for fieldwide comparison of individual wells. Table III shows data on 3 wells producing from another section of the Midway Sunset field. The changes for the first cycles for all 1195¥ A GRAPH. METHOD FOR IMA N_OF PROD. RESPONSE FROM CYCLIC STEAM STIMULATION USING PAST three wells are quite confusing. To some extent, the duration of the production phase may explain the changes in observed oil-stean ratio. This correlation may be pointed out for cycles 3, 4, 8, 9 of well A, cycles 5, 6, 7, 8, and 9 of well B and cycles 3 and 5 of well C. But there are also a few exceptions such as cycles 2, 7 of well A, cycles 2, 3 of well B and cycle 4 of well c. To smooth out the variations in oil/stean ratios from one cycle to the other, the data for the wells mentioned earlier are plotted on a log-log plot and shown in Figures 10-13. A unit slope is fit through the best points. Points falling on the unit slope are assumed representing the true performance response of the well and the points below the line are termed data affected by non-reservoir type problems. If the points either below or above an established unit slope form a Parallel line, there may be a shifting either due to enhancement or degeneration of the performance depending on the relative Position to the major unit slope line. Figure 10 shows the cumulative oil produced versus cumulative steam plot for well X in the Midway Sunset field. The performance factor for this well is computed as 2 Bbl oi1/Bb1 of steam. For wells A, B, C some interesting observations can be made. Well A in spite of some deviations during the third and seventh cycle maintains a performance factor of 0.429. The established trend for well B is at PF. = 1,9 Well C starts at a trend of P.P. = 0.18 during the first four cycles and then after the corrective actions are taken, it establishes itself at a P.F. = 0.515. The stimulation data for well @ in section 28 of the Kern River field are given in Table IV. A log-log plot of the data is also shown in Figure 14. Establishing the trend of the unit slope on the log-log plot results ina P.F. = 0.033, Data points for the early cycles indicates that the well had not established its performance factor before the seventh cycle. A case corresponding to a well approaching economic limit is well 119A as reported by Yoelin#. A log-log plot of the data shows the curvature formed because of the extreme depletion of the oil saturation. Discussion The use of unit slope on the log-log plot of cumulative oil produced versus cumulative steam injected provides a convenient means of establishing the correct trend for the well performance. The plot also helps in deriving a performance factor from the intercept. The Performance factor for various wells in a Given well can be used to contour areas of the reservoirs with similar response or areas receiving additional benefits from other factors such as structural position (gravity drainage). Te must be emphasized that the extrapolation of the straight line will represent the upper bound of the recovery for late cycles. after numerous cycles and when the oil saturation in the heated area is reduced to the vicinity of So,, the relative Permeability to oi] diminishes and the Cumulative plot will show a flattening of the straight line. The exact onset of this flattening depends on the nature of relative Permeability curves, érainage radius and residual oil saturation. A material balance check on the cumulative production may help to establish the magnitude of the residual oi] in the heated zone. Also, change in injection rate during the subsequent cycles will tend to shift the curve upward (higher injection) or downword (lower injection) and will maintain the similar unit slope. Conclusions A log-log plot of cumulative oil Production vs. cumulative steam injection results in @ unit slope and an intercept defined as the performance factor. ‘The unit slope aids in describing data points which Jag behind the true capability of the well in the projection of future performance. The performance factor is a more reliable estimate of the overall stean/oil ratio than the other definitions currently used in the industry. Acknowledgement The authors wish to thank various companies who supplied production data for testing the application of the proposed method. Also the assistance received from Nga Nguyen, Maria Vargas and Mrs. Senzaki and Taraneh Jifroodi during the preparation of the manuscript is appreciated. 1196Y RAJ ERSHAGHT, MOSA AL-ADAWTYA, AND VINCENT D. KAGAWAN Nomenclature Bo b be htg ber bfs Ms Pe ow gon rh 011 formation volume factor thickness of the sand, £t enthalpy of liquid water at Tavg, Btu/1b specific enthalpy of vaporization of| water at Tavg, Btu/ib specific enthalpy of liguid water at| Ty, Btu/ib specific enthalpy of liguid water at| Ty, Btu/1b unstimulated productivity index, STB 7D/Psi stimulated productivity index, STB/D/ Psi dimensionless (Ip/Ie productivity index thermal conductivity formation permeability, ma total mass of steam plus condensate| injected, Ibs. static formation pressure, psia Producing bottom hole pressure, psia volume of steam injected, Bb1. oil production rate during stimulated| production. s7B/1b radius of the heated zone, ft skin factor dimensionless. of the well, residual oil saturation original reservoir temprature, °F steam temperature at the sand face, °F time of injection oil viscosity, op eerfe (fy y+ J = porosity, fraction + ake - n? (Pe), (PO), = volumetric capacity of reservoir Btu/cu £t - OF Reference 1. Boberg, %. C. and Lantz, R. Bey "Calculation of the Production Rate of a Thermally Stimulated Well." Jour. of Petr. Tech. (Dec. 1966), 1613. 2. Burns, J. "A Review of Steam Soak Operation in California," Jour. of Petr. Tech. (Jan. 69) 25-34 3. Rivero, R. 7 and Heintz, R. Cc. , "Resteaming Time Determination - Case History of @ Steam-Soak Well in Midway Sunset," Jour. of Petr. Tech. (June, 1975) ‘p. 688. 4. Yoelin, s. 4. "Phe TM Sand steam Stimulation Project", Jour. of Petr. Tech. (Aug. 1971), 991 Tage 1 Penrommance Factors AS A Fi ResenvoiR Avo F.uip Prorenrtes de Pa 105; 1.028 05 01S 9% BE 500 1.266 3000 1.150 2000 950 027 730 11954 Table II MIDWAY SUNSET PIELD (Data from Referce 2) Cycle Oi Production, Cumulative Of] Bb1 Production, abl 5,585 3,585 2 2,784 14,369 3 5,702 20,071 4 5,898 25,969 5 7,218 33,187 6 5,137 38,324 7 7,393 45,717 table II cont. MIDWAY SUNSET FIELD (Data from Reference 2) cycle Steam Cumulative Oil Steam Injection, Injection, Ratio Bb1 Bb1 Y 6,332 6/332 —~*~*~*~SCS 2 6,179 12,511 2.421 3 5,170 17,681 1,102 4 4,592 22,273 1.284 5 5,618 27,891, 1.284 6 5.420 33,311 +947 7 5,664 38,975 1.305 Table 111 Steam Oil Ratio per cycle cycle Well A well Bowel c 2 0.425 1.25 0.087 2 0.4L 2.04 0.52 3 0.66 2.38 0.195 4 0.35 leet 0.59 5 0.37 3.44 1.47 6 0.38 1.29 0.92 7 0.25 3.70 0.52 8 0.83 1.75 0.47 9 0.958 0.98 0.65 Talbe IV Water for Well 8 Section 25 Kern River Field cycle oi cum. Steam Cum. Prod. Oil, Inj. steam, Bb1 Bb1 Bel. BbL r 143 1a) 8,278 6,278 2 294 437 12,451 20,724 3 317 734 9,265 29,994 4 208 962 7,181 37,175 5 261 1,223 7,251 44,426 6 466 1,689 12,338 56,764 7 353 2,040 11,984 68,748 8 304 2,344 6,753 75,501 9 414 2,758 12,961 88,462 i0 448 3,206 12,298 100,760 1 195, 3,401 13,225 113,985 12 47 3/447 13,641 127,626 13 542 3,990 © 4,391 132,017 14 538 4,528 9,792 141,809 1195¢ 110.00) T T T agg" 50 rag 00 one 000 xeoe 7 a 8 reco0}- 3 soooLtL \ \ ‘350 {25000 iar woop rep erp ssw Neo a sc gmooof- 3 1 L stat 1 ‘to Soe tha 100,000 -ry- T1417 T1191 T © c+ 008 2x 0.03 0 1C+ 0015, 10,000} 2 § a mo et 1,000 10,000 100,000 1x10® CUMULATIVE STEAM, 8 10 oa Joe| fos 3 02 20000 “40.000 ~ 60,000 ~ 80,000 ~ 100,000" 120,000 140,000" 160000 180,000 CUM, STEAM, BbL Fig, Relationship between Sand Mt, 100,000 abl 10,000 6 z o 3 unit sire 10865 1,000 101000 CUM. STEAM, BbI Fig, 10—Cumave po for Wal Xin Misway Sunset fel 100000 ry rd T1111 [WELL A MIDWAY-SUNSET FIELD é 4 wy 10.000-— & E 3 3 yoolul ed 1,000 10,000 100,000 CUMULATIVE STEAM, Bol Fig. 11—Cumuative pot for Wel Ain Rtway Sunset ‘ee 100.000, CUMULATIVE O14, BH S 8 WEL 8 MOWAICSUNSET FELD CUMKAIE ot, To rr we ¢ mIDWAY-SUNET Ftd Ly ge | 10.005 ‘co at CCOMRATIVE STEAM, Bh + 1 — scours: ya send SSUES 4 ome ] 2 B rool Z 4 8 ‘ See J iz | ot ws J ot J Li low ald soa spelen CCUMUAATIVE STEAM, BI hg omit a We @ Scan Ham fe 1900000 ry r i cuMULAriVE 04 008 ‘ear ‘eaoor CUMULATIVE STEAM, 861

You might also like