Automation of water treatment and industrial wastewater disposal systems at oil refineries

Abstract

The article is devoted to the development and validation of a comprehensive approach to automating water treatment and industrial wastewater disposal systems at oil refineries, with the dual objective of increasing environmental reliability and reducing operating costs; the research object is a typical refinery treatment train (mechanical, physico-chemical, biological, and filtration stages) subject to sharp fluctuations in influent composition and hydraulic loading. The methodological basis includes analysis of a 24-month APCS archive with hourly resolution across more than 35 parameters (flow, pH, temperature, petroleum hydrocarbons, phenols, suspended solids, COD, BOD5, ammonium nitrogen, reagent doses, aeration, energy consumption), application of correlation - regression and variance analysis, time-series methods, as well as the construction of multivariate models and adaptive control algorithms in DCS/SCADA; the applicability of fuzzy logic and neural-network approaches to nonlinear regimes of biological treatment is theoretically substantiated. The results demonstrate a substantial environmental effect of automation: at the plant outlet, mean concentrations of petroleum hydrocarbons and phenols decrease by a factor of 7-15 with a multiple reduction in variance, shifting the process into guaranteed regulatory compliance; simultaneously, resource savings are achieved through dynamic dosing and aeration optimization - specific consumption of coagulant and flocculant decreases by roughly a quarter, and specific energy use by nearly 25%. Under stress scenarios (slug discharges, 30-50% flow surges), the system exhibits high robustness: stabilization time for quality parameters is reduced by more than fivefold owing to anticipatory adjustments of reagent dosing, blower throughput, and flexible flow management (including the engagement of buffer tanks). The discussion highlights the synergistic effect of end-to-end automation of all stages, whereby stabilization of flotation improves the inlet to the biological circuit, and aeration optimization reduces the loading on polishing filters; a transition to a digital twin of the treatment facilities is proposed for scenario-based optimization and staff training. It is concluded that the methodology is suitable for scaling to plants with comparable process schemes and wastewater compositions and that there are prospects for deploying self-learning control systems.