Estimation and inference in inverse methods for kinetic models of hydrocarbon generation

Abstract

Procedures based on ordinary and generalized least of squares for the estimation of kinetic parameters in laboratory pyrolysis experiments are compared. When the data consist of measurements on the degree of chemical reaction through time for a number of temperatures, a single regression procedure is preferred to the two-stage regression alternative. Methods are developed for the construction of confidence intervals for calibration curves used to infer formation temperature on the basis of the unreacted fraction of source material. © 1995 International Association for Mathematical Geology.