Otimização econômico-estatística no controle estatístico de processos contínuos por bateladas

Abstract

Purpose: This research has as general objective to propose a cost optimization and statistical performance model for Statistical Process Control in a continuous batch processing operation. Design/methodology/approach: Design Science Research was chosen as the research method. Following its steps, in the Awareness of the Problem it was observed the need presented and the environment in which the problem was inserted. In the Suggestion stage, the identification of related studies and analysis of the cost models identified in the literature was carried out. In the Development stage, the results of the search in the previous stage were the basis for decisions in the development of a conceptual map and in the construction of the artifact. The optimization heuristic chosen for the case was a Genetic Algorithm, which was built with the aid of the statistical software R. The validation of the artifact occurred in two stages. First, the operation of the algorithm was validated, quantitatively, by comparing results with another research already published. Then, in a qualitative way, specialists in the field of continuous batch processing were consulted. This consultation aimed to validate the artifact as a whole, observing whether the inputs of the model are consistent with the reality of the industries and whether the outputs were in accordance with the needs of the companies. The Evaluation stage also occurred in two stages, with the aid of computer simulation. First, a comparative assessment of the economic performance of the artifact was carried out against traditional economic optimization processes. Then, the statistical performance of the control chart from the application of the artifact was analyzed by checking its detection speed. Finally, in the Conclusion phase, analyzes about the results obtained were performed. Findings: The results achieved with the implantation of the artifact through the computer simulation pointed to a reduction of 21.75% to 51.97% in the costs of the statistical control of the process, when compared to the use of the artifact with the traditional optimization process, in a horizon of 20 simulated batches. In the evaluation of the statistical performance of the control chart from the artifact, a constant increase in the detection speed was observed as the process deviation increased, regardless of which deviation the control chart was optimized for, which reinforces the robustness control chart statistics. Originality/value: This research contributes to the more precise evaluation of a type of process common in some industries, the continuous batch process. It was observed that this process has peculiar characteristics that make it different, for example, from a continuous flow process. This led to the need for a more specific approach to this type of process, highlighting significant gains.