Optimization in 2m 3n Factorial Experiments

G. S. R. Murthy, D. K. Manna


The need for adopting efficient designs in industrial experiments is well understood. Often situations arise where the existing designs such as orthogonal arrays are not suitable for designing required experiments. This paper deals with one such situation where there was a need for designing an asymmetrical factorial experiment involving interactions. Failing to get a satisfactory answer to this problem from the literature, the authors have developed an ad hoc method of constructing a design. It is transparent from the method of construction that the design provides efficient estimates for all the required main effects and interactions. The later part of this paper deals with the issues of how this method is extended to more general situations and how this ad hoc method is translated into a systematic approach. The method consists of formulating the construction problem as certain integer programming problems. It is believed that this method will be very useful in practical applications. The ideas are illustrated with a number of examples.

Full Text:


Algorithmic Operations Research. ISSN: 1718-3235