Quality and manufacturing cost: A case study in the chemical industry

This paper presents a case study where the optimization of a chemical mixture was accomplished. The study included the following steps: problem identification, design of an experiment, model building for each individual response, choice of an objective function, and optimization. In the steps of problem identification and design of the experiment, a matrix was used to gather and to organize the information. Five control factors and ten response variables were included in the experiment. In the step of model building, models for mean and variability for each response variable were built. The objective function used for optimization was an extended version of the Multivariate Quadratic Loss Function proposed by Ribeiro & Elsayed. The optimization was carried out taking into account the losses due to poor quality as well as the losses due to the costs of raw materials and energy. After the optimization it was possible to identify the best set for the control factors. A sensibility analysis was performed and an operational envelop for process control is suggested.