Comparison of Stopping-Rule Methods in the Process Optimization Strategy Using Steepest Ascent or Descent in a Tension Measuring Machine

Abstract

Nowadays, process improvement is already an essential competitive mechanism for companies around the world. Production processes require establishing improvement strategies to obtain greater productivity. In modern literature, there are a series of statistical experimentation schemes that allow the establishment of methodologies permitting such improvement to be carried out efficiently through a previously established analysis. The Steepest Ascent or Descent Method (SADM) is an example of this. SADM consists of the development of an experimental design that yields a linear model that follows a straightline of ascent or descent path towards a target point within the process. Fundamentally, it is a procedure that builds a sequential experimentation model where two-level factorial designs are highlighted, in which a series of iterations are made to follow a line towards a region where optimization is feasible. These iterations or individual experimentations follow certain criteria called “Stopping rules”. Series of rules to know when no more iterations are required because the desired region would have already been reached. This paper presents the implementation of the SADM in a case study based on a fitted linear obtained from a factorial design. The Myers-Khuri method and the Parabolic-Recursive method are applied to proceed with the stop. Both methodologies are intended to create a decisive and efficient stopping strategy. The objective is to make a comparison between both methods to show the predominant one in relation to a better use of resources. Results obtained and a conclusive analysis are disclosed at the end of this document.