Análise do potencial de oxirredução do hipoclorito de sódio: desenvolvimento e aplicação de um método de inferência

Abstract

The Chlorine-Soda Units produce sodium hypochlorite (NaClO) through a batch process, employing periodic titrimetric analyses for quality control, with a focus on sodium hypochlorite and caustic soda (NaOH) concentrations. These analyses demand significant time, operational effort, and extensive consumption of reagents and materials, becoming more frequent toward the end of the batch to ensure product stability. As a more efficient alternative, this study explores potentiometric methods based on Oxidation-Reduction Potential (ORP) measurements, allowing for the indirect control of NaClO and Cl₂ concentrations. Inferential ORP measurements can provide greater operational efficiency, cost reduction, and safety, minimizing operators' exposure to process-related risks. The primary objective of this study is to establish a sampling plan based on standardized technical instructions and construct regression models, both linear and non-linear, using Artificial Neural Networks (ANN). Experimental process data were utilized, applying statistical techniques to accurately describe the relationship between chlorine concentration and Oxidation-Reduction Potential (ORP). The results indicated an inverse exponential relationship between ORP and active chlorine concentration, with a coefficient of determination (R² = 0.985) in the model adjusted via Excel. Additionally, Artificial Neural Network models with configurations of 20 and 36 neurons demonstrated high performance, with accuracy values above 96% and reduced mean absolute errors (MSE) compared to the Excel model. Statistical testing confirmed the significance of the relationship between the variables, with p-value < 0.05, validating the robustness of the developed models.