A three-dimensional, three-phase numerical steam-displacement model was used to history match 5 1/2 years of field data from a representative five-spot pattern in Kern River, Calif. The model was used further to effect optimization of steam-injection rates for typical five-spot patterns. Introduction The three-dimensional, three-phase numerical model for steam displacement, described by Coats et al., was previously tested with three sets of laboratory previously tested with three sets of laboratory experimental data and was used in various applications, including a representative field-scale problem and a steam-stimulation example. This paper concerns the same model and consists of two parts. The first part is a further validation of the numerical model by history matching the performance of a single pattern in the Kern "A" project in Kern River field, Calif. (Fig. 1). Related project in Kern River field, Calif. (Fig. 1). Related studies are included on the simulation of steam stimulation, the effects of grid orientation, upstream weighting of viscosities, and the temperature dependence of relative permeabilities. In the second part, the model is used to optimize an objective function for the cases of constant and varying steam rates. While the constant-steam-rate cases maintained the same rates for the entire project life, the varying-steam cases used decreasing sequences of steam rates, with each maintained for a prespecified length of time. Performance of a Single Pattern Performance of a Single Pattern Field Pattern Description A five-spot steam-displacement pattern was selected in the Kern "A" project (Fig. 2) for history-matching purposes. The basis for this selection was primarily the purposes. The basis for this selection was primarily the following criteria: the field operation typifies Kern River, the availability of good reservoir data, the nearly complete cycle life of the displacement zone, and near symmetry of the pattern. The pattern selected is about 430 ft in the east-west direction and 270 ft in the north-south direction, and covers an area of 2.7 acres. East-west and north-south cross-sections through this pattern are given in Fig. 3. The displacement sand shown pattern are given in Fig. 3. The displacement sand shown in the cross-sections is locally referred to as the "R1" interval. Note the existence of a tight streak in a part of this interval. Core-hole data are available from Well 503, which was cored before displacement of the R1 zone, and Well C. H. 1, which was cored near the depletion of displacement in the R, sand. The patterns including Wells 503 and C. H. 1 were not chosen because the wells are not as symmetrically spaced as those in the pattern around Well 68. Data from Well 503, shown in Table 1, were used for the vertical distribution of permeability and saturation. The permeability of Layer 2 in Table 1 was assumed to be only 1 percent of that of Layer 1, which reflects the existence of the tight streak shown in Fig. 3. The fluid volumes, as indicated by core analysis, were assumed to completely fill the available pore space, and normalized saturations were calculated. This method tends to restore the core data of the unconsolidated sand to reservoir conditions, as suggested by Elkins. Calculations were also made to correct core porosities to values closer to reservoir conditions. Additional Fluid and Rock Data This type of reservoir simulation requires a wide range of input data. Where possible, field data were used to supplement data obtained from the literature. JPT P. 765