A search method for process optimization with designed experiments and some observations

Purpose – The purpose of this paper is to attempt to get more in‐depth understanding of real life processes, for which theoretical models are not available and knowledge about the processes is empirical. The processes are noisy; interaction effects may or may not be present. The paper attempts to formulate a more practical view, related to such processes. This is sought to obtain improved ways of experimentation for process and product optimization. Design/methodology/approach – Functional forms are discussed when there would be no interactions. A conceptual process is considered to highlight some important aspects of interactions. A discrete search method of optimizing a process or a product is suggested. Analysis is done for the method and a numerical experiment is carried out to investigate the performance of the method. This is further verified with a real life case study. Findings – Some conditions are obtained when the search method would yield an optimal solution. The numerical experiment and the case study indicate satisfactory performance of the method. A set of preferable steps, utilizing the method, is suggested to conduct experiments. Research limitations/implications – The paper suggests a set of steps to carry out experiments in order to optimize a process, and give the rationale why these would work for some processes. The steps are not so general to include all kinds of processes which may occur. Practical implications – Improved ways of experimentation would help to arrive at optimal values of parameters in products and processes, with less cost and time. Originality/value – The paper suggests a new discrete search method of optimization and a particular approach, based on the method, to the conduct of experiments.