Sensitivity of Production Data To Calculated Original Oil-in-Place Using A One-Dimensional Mathematical Simulator

Abstract

Abstract Errors in oil, gas and water production data are unavoidable. It has always been a matter of concern that these errors may have a serious effect on the results of model studies to determine the original oil-in-place. A one-dimensional, three-phase simulator study was undertaken to investigate the sensitivity of errors in production data on the original oil-in-place calculation. The Rainbow Keg River AA Pool in Alberta; was used as an example. This pool has had both water and gas injection and has experienced excess gas production. The presence of various production mechanisms, therefore, made the example a comprehensive one. After obtaining a suitable pressure history match of the pool using the field-measured oil and gas production data, the monthly production data were random changed by introducing a certain maximum error. The random production data were then used in subsequent model runs to evaluate their effect on the pressure history match. The question of whether or not a change would be required in the originally modelled oil-in-place to again obtain a good match was also evaluated. The present work indicates that the results of a simulator study to determine the original oil-in-place are insensitive to significant random errors in the monthly oil and gas production data. The per cent cumulative error decreases with time due to the random nature of the individual errors. Therefore, costly measures to refine the production data in order to improve on the accuracy of the oil-in-place determination are not warranted. Introduction It has often been argued that accuracy in production data is critical to the results of a model study to determine the original oil-in-place. Errors in the field-measured production data from a battery are introduced due to the practice of allocating the total metered production in a month to individual wells based on one or more well tests and the number of days each well was on stream. Fluctuations in the well rates are normal and thus introduce an error in the production allocated to the well based on only one, two or rarely more tests per month. Errors in the production data can also be introduced due to incorrect gas chart. This paper examines the effect of the inaccuracies in the production data on the calculated original oil-in- place. A one-dimensional, three-phase, compressible fluid mathematical simulator was used for this purpose. Errors in the monthly oil and gas production data were introduced using a subroutine which generates random numbers corresponding to a mean value and a given standard deviation. The approach of using random numbers to evaluate the uncertainties in engineering calculations is not new and has been used in the past in connection with oil and gas reservoir problems(1, 2). Unfortunately, there is no reference in the literature evaluating the sensitivity of the production data on the calculated original oil-in-place, although Muskat recognized this problem, particularly regarding the uncertainty of the gas production data(3). An attempt has been made in this paper to evaluate this sensitivity.