Rapid Assessment of Pattern Waterflooding Uncertainty in a Giant Oil Reservoir

Abstract

Abstract This paper presents a sensitivity study of pattern waterflooding in a giant carbonate oil reservoir in North Kuwait, currently being produced by natural depletion. Fine grid models for different pattern types and well spacing were constructed and used to estimate time to breakthrough, ultimate recovery and production rate for a range of reservoir descriptions and fluid properties derived from core, log and lab studies. The total number of simulation runs necessary to cover all possible combinations was impractically large (close to 800). Consequently, a reduced selection (144) was chosen based on the theory of experimental design to maximise the information obtained. This theory identifies what combinations of input parameters should be chosen in order to provide accurate estimates of the effects of individual factors. Results were analysed using analysis of variance, which indicated the pattern, spacing and description factors which were important, and their effect on recovery. It also gave a prediction of recovery for all combinations of input parameters. Overall the results suggest that recovery is relatively insensitive to pattern type and spacing which implies that the choice of pattern and spacing can be guided by injectivity productivity and the required offtake rate. The results of this study will be used to guide future well placement for the waterflood development of this reservoir. The methodology presented in the paper illustrates the value of experimental design theory for planning and rapidly extrapolating the results of a reservoir simulation sensitivity study to cover additional combinations of input parameters. 1 Introduction A pattern waterflood is being designed for a giant platform-carbonate reservoir in North Kuwait, currently being produced by natural depletion. The waterflood development requires a substantial investment in wells and facilities and optimising the number and location of wells will be key to the success of this project. To assist the design of the development a simulation sensitivity study was conducted to understand the impact of the choice of well pattern and spacing and of uncertainties in description on recovery efficiency and rate. Pattern element models were developed for different well spacing and patterns for simulation using ECLIPSE. A range of generic descriptions was used, representative of the crest and mid-flank areas of the two fields in which the reservoir occurs. These were based on pvt, core and log data and were guided by the available dynamic data. The range of models reflected uncertainties in the description. A sensitivity study of waterflood performance was then conducted. Techniques of experimental design were used in planning and analysing the sensitivities so as to maximise information. 2 Alternative well patterns and spacing Four different patterns are considered: inverted 5, 7 and 9 spot and normal 7 spot. For each of these patterns a range of well spacing is evaluated: 100, 200 and 300 acres/well. Consideration of the number of injection and production wells required to produce a field suggests the following approach to selecting the optimum pattern and spacing:–From a knowledge of the relative injectivity and productivity of wells select the pattern with the appropriate injector : producer ratio (see table 1). Different patterns have different relative numbers of injection and production wells. P. 483^