6. RESULTS AND DISCUSSION A key requirement for the Operator Assistant is its ability to propose actionable and realistic setpoints that align with plant behavior. This experiment compares actual operator-selected setpoints against the optimizer-generated recommendations for two critical control levers (i) pH Setpoint Recommendation (Fig 8), (ii) Organic-to-Aqueous (O/A) Flow Ratio Recommendation. During each operating window, the optimizer generates candidate setpoints, evaluates them using the hybrid digital twin, and filters infeasible options through the constraint validator. The remaining setpoints are ranked by KPI performance. The recommended pH setpoints demonstrated close agreement with plant-applied values, generally remaining within ±0.05–0.10 units. Larger deviations were observed primarily during intervals affected by delayed laboratory assays or elevated Fe loading, conditions under which operator heuristics tended to diverge from model-inferred optima. The recommended O/A ratios showed strong alignment with operator adjustments while providing smoother operational trajectories and improved anticipation of Fe co-extraction peaks. Across all evaluation periods, the optimized setpoints either matched or outperformed historical operator selections and remained fully compliant with equipment, chemical, and hydraulic constraints. The final evaluation quantified the operational impact of the recommendation engine by comparing plant-measured KPIs with optimizer-predicted KPIs across each experimental interval. The analysis incorporated actual process data covering copper recovery, solvent extraction (SX) efficiency, iron concentration in the rich electrolyte, raffinate pH stability, organic phase utilization, and, when available, predicted electrowinning (EW) current efficiency. Although representative subsets of the data are shown in this paper, identical performance trends were Figure 8 - PLS pH - Actual Plant Vs Operator Assistance Recommendation 8 - PLS pH - Actual Plant Vs Operator Assistance Recommendation Figure 9 - O/A - Actual Plant Vs Operator Assistance Recommendation 9 - O/A - Actual Plant Vs Operator Assistance Recommendation Figure 10 - Cu Recovery - Actual Plant Vs Operator Assistance Recommendation 10 - Cu Recovery - Actual Plant Vs Operator Assistance Recommendation Figure 11 - SX – Efficiency - Actual Plant Vs Operator Assistance Recommendation Figure 12 - Iron in rich electrolyte - Actual Plant Vs Operator Assistance Recommendation
RkJQdWJsaXNoZXIy MTM0Mzk2