Parameter Estimation for the Arrhenius Model

This paper suggests a procedure of interval estimation for the parameters of the Arrhenius model and the corresponding design of experiments where the failure distribution is assumed to be exponential. The estimation procedure has two steps, a modified maximum likelihood estimation of the failure rate followed by an ordinary least squares regression analysis. Design of experiments is used to achieve estimation with given confidence intervals. The procedure uses the properties of the log gamma distribution. While the estimation procedure is for the Arrhenius model it can be used in other life test situations which can be described by linear regression models with one regressor variable, eg, the inverse power law.