RI 7443 Optimization Of Carbon Deposition Experiments In Thermal Oil Recovery By Statistical Analysis

The Bureau of Mines used regression analysis and applied statistical methods in an effort to rationalize the results of an experiment examining the effects of certain variables on carbon deposition in a reservoir for thermal recovery of petroleum. Although none of the 15 trials of the original experiment resulted in the response of weight-percent of carbon regarded by the experimenter as adequate, values calculated using canonical analysis in two trials yielded responses 33 and 54 percent higher than the minimum acceptable figure. The actual responses differed by less than 3 and 6 percent, respectively, from the predicted responses. Methods are described for defining a path of increasing response based on the minimum point and the response surfaces and for defining a set of values for the parameters in question, which result in increased response, A detailed explanation of the mathematical process used in this analysis is presented in the appendix for ease in computer programing the technique.