Publications

Evaluation of factor effects on material balance calculations using statistical design methods

Proceedings Title : Proc. Indon. Petrol. Assoc., 22nd Ann. Conv., 1993

Material balance techniques have been widely used as essential engineering tools to estimate original hydrocarbon in place and analyze past and predict future reservoir performance. However, the results of these calculations are only as accurate as the input reservoir data which is often uncertain and difficult to define. This paper describes a procedure using statistical screening design to determine the factor effects of reservoir properties on the results of material balance calculation. The screening design techniques permits estimation of the influence of several factors simultaneously, thus yielding the maximum amount of information from a minimum number of trials.Ten factors representing reservoir fluid and rock properties and historic field production and pressure data were evaluated for their effects on the material balance calculations. The magnitude of main factor effects and two factor interactions were evaluated first to identify the most significant variables. Ten factor effects were determined for the three types of material balance models. In the analysis of the single factor effects it was found that the order of significance of the factors varied depending upon the model used for the material balance calculation, Schiltuis, Hurst or infinite linear aquifer model.A statistical surface design procedure was then used to determine the functional dkpendence of the most significant factors. This surface design technique was employed to, again, minimize the number of trials required. The results from this procedure compared favorably to the functional responses determined by a typical one-factor-at-a-time approach.Depending upon the material balance model used, either 37 or 65 trials were required to evaluate the factor effects.In order to evaluate all ten factors with a 2 level factorial approach, 1024 trials would have been required. As few as 30 trials could have been used for d one factor at a time approach, but no information regarding factor interactions, and little regarding nonlinear effects would have been obtained.

Log In as an IPA Member to Download Publication for Free.
or
Purchase from AAPG Datapages.