Prediction of Selected Jet Fuel Test Results Using ASTM Test Method D2887Data with Multiple Linear Regression Analysis.

摘要

At the present time, to analyze a five gallon sample of JP-4 requires approximately 24 hours, with over thirty separate tests to determine overall sample acceptability. The time required to ship the sample from bases and distribution points to central laboratory facilities and to disseminate the results is considerable. This research attempted to identify those boiling points from the simulated distillation curve which could be used to predict other JP-4 test results such as API Gravity, Aniline Gravity Product, RVP, Water Separometer Index, Modified, without the requirement to perform these tests. The research used multiple linear regression techniques to obtain the algorithms for the above JP-4 tests. Models developed by these techniques did not attain the level of accuracy needed to delete the requirement for physically completing the above tests. However, several ideas are presented for use by future teams performing research in predicting fuel properties from simulated distillation data. (Author)