Planning and Analysis of Reliability Data Influnced by Two Factors in Self Relocating Designs
by
Amita A. Dharmadhikari
University of Poona, Pune, India

We consider a situation wherein performance (as measured by life expectancy) of m types (brands) of components have to be compared in n different environments. Rather than carrying out independent experiments for each brand in each environment (according to type I or type II censoring) one can consider an association between these mn experiments . Self Relocating Designs(SRD) , introduced by J. N. srivastava (1985) is a method to plan mn experiments where no. of components to be tested for a specific combination of a brand and environment would depend upon how failures have occured in other experiments. Srivastava , in the above paper, has developed analysis of a one way experiment where data are collected by SRD . Dharmadhikari, Dharmadhikari , Shah (1999), developed analysis of two way experiment withoyt interaction. In this paper, we generalize the methodology of data analysis of the above experiment to those with brand X environment interactionand with some relaxed assumption. Data in eachcell are generated according to a specific SRD and units in each cell follow an exponential distribution with different scale parameters.The scale parameter is assumed to be a multiplicative function of two factors and their interaction. MLE of scale parameters are derived and LR test procedure is proposed to test whether brands , environments differ significantly and whether interaction is present.

Date received: October 30, 2000