History Match of Multiwell Simulation Models of the Cyclic Steam Stimulation Process at Cold Lake
- T.C. Boberg (Exxon Production Research Co.) | M.B. Rotter (Exxon Production Research Co.) | S.D. Stark (Imperial Oil Ltd.)
- Document ID
- Society of Petroleum Engineers
- SPE Reservoir Engineering
- Publication Date
- August 1992
- Document Type
- Journal Paper
- 321 - 328
- 1992. Society of Petroleum Engineers
- 5.5 Reservoir Simulation, 4.1.2 Separation and Treating, 5.2 Reservoir Fluid Dynamics, 5.3.4 Integration of geomechanics in models, 5.1 Reservoir Characterisation, 5.2.1 Phase Behavior and PVT Measurements, 5.4.6 Thermal Methods, 1.2.3 Rock properties, 2.2.2 Perforating, 4.3.4 Scale, 1.6 Drilling Operations, 5.5.8 History Matching, 5.1.5 Geologic Modeling, 2.4.3 Sand/Solids Control, 5.8.5 Oil Sand, Oil Shale, Bitumen
- 2 in the last 30 days
- 499 since 2007
- Show more detail
- View rights & permissions
|SPE Member Price:||USD 10.00|
|SPE Non-Member Price:||USD 30.00|
Various simulation model elements were examined in an effort to matchobserved behavior of a selected steam stimulation well at Cold Lake. The bestmatch required a large model encompassing eight wells, but a three-well modelprovided an adequate match. Well location, fracture angle, and geology must berepresented accurately.
Background. The Cold Lake project, located 258 km [160 miles] northeast ofEdmonton, Alta., currently produces 14 000 m/d [88,000 B/D] of bitumen from1,900 wells and is the largest steam-injection project in Canada. In late 1987,Exxon Production Research Co. began a multiwell reservoir simulation modelingeffort to guide future simulation of cyclic steam stimulation (CSS) operationsand potential follow-up processes at Cold Lake. Appropriate simulator modelsare needed for use in reservoir management of the pilot and commercialoperations at Cold Lake. A prior simulation study of the L Pad area in theLeming pilot greatly enhanced the understanding of how best to simulate the CSSprocess. Because of simulator limitations at the time, the L Pad process.Because of simulator limitations at the time, the L Pad study was performedwith a single-well radial model. The L Pad study showed the importance ofgeomechanical effects by including a porosity expansion/compaction model thatsimulates the variation in porosity expansion/compaction model that simulatesthe variation in porosity as a function of pore pressure. 1.2 Modeling changesin porosity as a function of pore pressure. 1.2 Modeling changes inpermeability as a function of pore pressure in a thin fracture layerpermeability as a function of pore pressure in a thin fracture layer oppositethe perforation interval also was found to be necessary to achieve reasonablesteam injectivity and a proper areal extent of the heated zone. This feature isdiscussed in Refs. 1 and 2, as is the importance of including relativepermeability hysteresis in a model of the CSS process. The L Pad study hadconsiderable value in providing experience in describing early CSS behavior fora situation where well interactions were not important. However, observationsfrom the pilot and commercial CSS operations at Cold Lake indicated thatinterwell communication was quite common, often occurring during the firstcycle of steaming. It was apparent that larger, multiwell models would benecessary to simulate these interactions adequately. The study reported herewas performed with Exxon's newly developed, fully implicit thermal version ofthe MARS simulator. While there is no theoretical restriction on the size ofmodel that can be simulated with MARS, the practicality of long running timeswith large thermal models remained as a limiting factor. Models containing upto 4,500 gridblocks and up to eight wells were developed for this study. Thefocus of this study was shifted to the W Pad in Leming from the L Pad for thefollowing reasons. 1. The W Pad was more representative of the commercialoperating areas, having the same pad orientation and well pattern configurationand similar geology to that in the commercial projects. projects. 2. Many ofthe wells in the W Pad experienced some degree of interwell communication inthe early cycles of CSS, as did wells in the commercial operation. Developingmodels that could match such communication behavior was a principal objectiveof this study. 3. The W Pad had completed five cycles of CSS by late 1987. Thequality of the measured injection and production data at the W Pad was good andprovided a reasonable basis for history matching.
Overall Study Approach. Fig. 1 shows a well map for the W Pad, which islocated in the northwestern sector of the Leming pilot area. Well W-08 wasselected as the focus of the history match for several reasons. First, WellW-08 is an interior well in the W Pad and is affected mainly by wells withinthis pad and not by wells external to the W Pad. Second, Well W-08 is betweentwo cored wells. Thus, the initial saturation distribution in its vicinitycould be determined more accurately. Third, Well W-08 and its neighboringon-trend wells had obvious communication events as early as the first cycle.Because an objective of the study was to develop models that could match earlyinterwell communication behavior, Well W-08 was a good choice. Finally, WellW-08 has good-quality injection and production data-necessary criteria formatching purposes. purposes. A geological study was undertaken to develop adetailed reservoir description. Computerized geological models describingformation thickness, structure, bitumen saturation, permeability and thelocation of vertical flow barriers or "tight streaks" were developed.The detailed reservoir description provided by these geological modelsconsiderably aided obtaining a reasonable history match. The history-matchapproach taken was first to obtain as good a match as possible in a two-welldetailed model that included Well W-08. Much was learned from the two-wellmodel about the effects of the various deformation, fracture, andrelative-permeability-hysteresis parameters on model behavior. It soon becameclear however, that, at a minimum, the influence of the adjacent rows of wellson both sides of Well W-08 would have to be included in the model to obtain areasonable history match. As a result, the effort soon progressed to three-wellmodels centered around Well W-08. The bulk of the progressed to three-wellmodels centered around Well W-08. The bulk of the history-matching effort wasthen conducted in three-well models. After an acceptable history match wasobtained in a three-well model, larger models were used in which wells fromfour adjacent rows were included. Wider models also were examined that includedtwo wells from multiple rows of wells. Of this latter type, an eightwell modelrepresented the largest model size (4,500 gridblocks) that was practical to runon available computers. These models were used to evaluate the impact of modelsize on the quality and accuracy of the history match. The field data historymatched were the total fluid and oil production rates (and cumulativeproduction) from individual wells production rates (and cumulative production)from individual wells at W Pad that were included in the respective models. Inaddition, the temperatures observed upon initial drilling of the infill wellswere matched. No reservoir pressure or gas production data were available. Theprincipal matching parameters were the orientation of the fracture zone, theoriginal formation permeability, and the multipliers in the equation describingthe variation of permeability as a function of porosity in the fracture layer.In addition, variables affecting the shape of the relative permeability curves,particularly the irreducible water saturation, were examined. The particularlythe irreducible water saturation, were examined. The effect of including or notincluding tight streaks was also studied.
W Pad History Match
Reservoir Description. Thermal recovery operations at Cold Lake occur in theClearwater formation. A description of the regional geology of the Clearwateris contained in Ref. 6. The Clearwater is a thick, unconsolidated sand withporosities up to 35 % and with net pay thickness often in excess of 35 m [115ft].
|File Size||1 MB||Number of Pages||8|