EXPLORING APPLICABILITY OF LINEAR REGRESSION MODEL FOR PREDICTION OF THE EFFICIENCY OF WATER STABILIZATION TREATMENT USING HEDP AND ATMP

Abstract

Stabilization water treatment is a crucial stage in water preparation for industrial and municipal systems. Its primary goal is to prevent scale formation, which negatively affects the efficiency of water supply and heating systems. This paper addresses the problem of improving the effectiveness of stabilization water treatment and proposes modern approaches to solving it. The study presents the results of stabilization treatment of tap and natural water with a hardness of 5.9–9.9 mg-eq/dm³ using HEDP and ATMP reagents. The impact of various doses of these reagents (0.50–2.00 mg/dm³) on the residual water hardness was investigated, and their stabilization and anti-scaling properties were evaluated. The findings demonstrated that using HEDP and ATMP significantly reduces sediment formation, contributing to the efficient operation of water treatment systems. To predict the efficiency of antiscalants, the linear regression method was applied, enabling the evaluation of their performance based on concentration. Thus, the proposed approaches help optimize stabilization water treatment technologies and reduce water purification costs. The study is primarily related to the use of the linear regression method for modeling the effectiveness of stabilization water treatment. The application of the least squares method provides an accurate determination of the parameters of the dependence between the dose of antiscalant and the stabilization effect. The Python programming language with pandas, numpy, scikit-learn libraries was used to automate the calculations. The modeling process includes data loading, sample separation, model training and accuracy assessment. The LR method is an effective model for statistical analysis and prediction of new values in the field of water stabilization treatment. It was established that these reagents provide stabilization and antiscaling effects at the level of 81-96%. Modeling was carried out using the LR method, the predicted results obtained are close to the experimental ones, in most cases within 3%. The LR method allows you to obtain linear dependencies for antiscaling and stabilization effects, which confirms its prospects for optimizing water stabilization treatment.