Abstract
Future performance is very important to evaluate whether an enhanced oil recovery (EOR) is successful or not. The most important step in prediction is to determine the amount of oil that can be recovered after applying an EOR process. Polymer flooding is one of the most important EOR techniques used to improve the mobility ratio and, therefore, sweep efficiency. Multiple linear regression techniques were used to develop the equation that can be used to predict the oil recovery based on rock and fluid properties in field data set. A dataset was created by collecting information from EOR surveys of Oil and Gas Journal (1974 - 2012). A total of 481 field projects was considered to construct the dataset. Unfortunately, this data contained a number of problems (duplicate, missing, and inconsistent data) that affected the dataset’s quality. To ensure the quality of the dataset before running any analyses, box plots and cross plots were used to identify data problems. Graphical and statistical methods were used to analyze and describe the results of the dataset. After enhancing the data quality, only 82 fields were used for the predictions.75 fields were used to build the model. The remaining fields were selected to validate the models. Parameters were chosen for the models: area, oil gravity, oil viscosity, porosity, saturation start, permeability, depth, and temperature. The stepwise technique was used to establish the independent model that affects the response variable significantly. Two models were constructed; one to predict oil recovery and another to predict oil saturation (So(end)) after polymer flooding. Equations for both models were presented in this paper. The equation for So(end) appears to represent the best model based on R2 values.