ANALYTICAL TOOL DEVELOPMENT USING INFORMATION THEORY--FINAL REPORT

Data on synthetic fuel byproducts, such as Rock Springs data on oil shale retort byproducts and Hanna data on coal gasification byproducts, has been analyzed by various statistical procedures. It is found that standard statistical procedures, which assume Gaussian probability distributions for each concentration variable, are often inadequate. This investigation is the subject of part I of the enclosed report. A discriminated analysis is then proposed in detail to analyze the data where probability distribution functions (pdf’s) are obtained by using the principle of maximum entropy (see part II). Preliminary numerical results from the new procedure are presented and discussed. Recommendations for further studies are also given.