Evaluation of the efficiency of conventional drinking water treatment plant using machine learning and water quality index in low-income countries: A case study of Adama City, Ethiopia

Ensuring safe drinking water requires monitoring water quality parameters, optimizing plant design, and addressing emerging contaminants. This study evaluates the efficiency of the Adama City Conventional Drinking Water Treatment Plant (CDWTP) using machine learning model and the Water Quality Index (WQI). Water samples were collected from various treatment stages and analyzed for physicochemical and bacteriological parameters. The study employed the VARMAX model to predict water quality trends based on historical data. Results indicate that the CDWTP operated with an average efficiency of 89%, though performance fluctuated due to operational and environmental factors. While the final treated water generally met World Health Organization (WHO) standards, total coliform levels exceeded permissible limits, suggesting the need for improvements in chlorine disinfection. The plant demonstrated high removal efficiencies for key contaminants, including iron (99.82%), turbidity (99.27%), ammonia nitrogen (98.55%) and phosphate (99.89%). The water quality index (WQI) was 26.149 classified Adama City’s treated water as “excellent.” The study further utilized predictive modeling to assess the plant’s ability to maintain water quality over the next five years considering the limitation. The model relied on historical data, limiting its ability to capture sudden water quality changes and non-linear interactions, with results indicating continued compliance with WHO standards when proper operational measures are maintained. To sustain this performance, ongoing monitoring, optimal chemical dosage, and infrastructure improvements are recommended. These findings provide valuable insights for policymakers and water management authorities to enhance treatment efficiency and ensure long-term water security.