Hindalco Muri Works, a premier aluminaAlumina refinery under the Aditya Birla Group in Jharkhand, India, excels in producing high-quality aluminaAlumina, including specialty grades for various industrial applications. Hindalco Muri utilizes Aveva Pi Vision for its machine learningMachine learning models, offering real-time data visualization, customizable dashboards, and predictive analytics for comprehensive monitoring and informed decision-making. Several advanced models have been developed and successfully implemented using Aveva Pi Vision. Steam economy improvement model in Evaporator Hindalco Muri operates a four-body falling film evaporatorEvaporator that uses 6 ata steam to concentrate spent (aluminaAlumina stripped) liquor to thick(concentrated) liquor while maintaining the refinery’s water balance. The steam economySteam economy is the amount of steam required to evaporate 1 ton of water from spent liquor. The predictive modelPredictive model, trained on 2 years of data and validated with 3 months of subsequent data, achieved mean absolute percentage accuracy of 97% for steam economySteam economy and 98% for output thick liquor concentration. This model enhances operational efficiency by ensuring accurate predictions and optimal performance. Soda (NaOH) loss control model in washers In aluminaAlumina refinery, caustic soda (NaOH) is used to extract aluminaAlumina from bauxite ore. During aluminaAlumina refining process, caustic soda recoveryRecovery is done in the mud washer circuitWasher circuit. The last washer soda prediction model involves an in-depth analysis of the Deep Cone Washer (DCW) process, optimizing parameters to maximize caustic soda removalRemoval. Utilizing Pi Vision data, the model tailors’ recommendations for each washer, aiming to minimize soda liquor loss to 12–14 gpl with five washers and 6–8 gpl with six washers. This optimization tool empowers users to take corrective actions, significantly enhancing caustic soda recoveryRecovery efficiency.