Pilot-scale production optimization based on ore blending digital twin

Abstract

The variations within an ore deposit and a processing plant’s operating variables significantly affect the metallurgical performance and overall economic value of metal concentrates produced. This study was conducted to develop the Oulu Mining School Minipilot plant’s Ore Blend Advisor (OBA) HSC simulation model based on pilot-scale flotation experiments conducted using laboratory-scale flotation results. The OBA was modeled in the context of production optimization to simulate the effects of ore blending on flotation outcomes under various plant configurations and optimized airflow rates in flotation banks. The optimized OBA HSC simulation model is realized as a digital twin cloud-based simulation application. The validation of simulated results by experimental results was also an integral part of this work.

The raw material used in two pilot-scale flotation experiments was from polymetallic Pyhäsalmi Mine Oy (Cu-Zn-S). The first experiment (Run/Ore Type 1) was conducted on sulfide-rich Pyhäsalmi ore (1.63% Cu, 2.3% Zn, 39.6% S) and the second experiment (Run/Ore Type 2) was conducted on blend of sulfide-rich Pyhäsalmi ore and low-grade Pyhäsalmi ore (0.8% Cu, 1.1% Zn, 19.4% S). The X-ray Fluorescence, X-ray Diffraction, Particle Size Distribution, and Total Sulfur analyses were performed on collected samples. The mass-balanced data showed that in Run 1, the final copper concentrate had a Cu grade of 8.48% with a 91% recovery, and the zinc concentrate had a Zn grade of 32.4% with a 55% recovery. In Run 2, the final copper concentrate had a Cu grade of 5.37% with a 98.9% recovery, while the zinc concentrate had a Zn grade of 19.97% with an 89.9% recovery. The principal mineral phases identified in the feed samples were Chalcopyrite, Pyrite, Sphalerite, and Quartz. The D50s of the flotation feed sample for Run 1 and Run 2 were 25.24μm and 17.60μm, respectively. An excellent correlation was observed between experimental and mass balanced data from both pilot-scale flotation experiments, with an R² value of 0.998. The Route Mean Square Deviation (RMSD) for Cu%, Zn%, and S% for Run 1 is 0.45, 0.90, and 2.41, and for Run 2 is 0.10, 0.06, and 0.42, respectively.

In OBA, the static scenario editor tool of HSC Sim was used to simulate eight scenarios from A to H, to find the optimal flotation circuit configurations for both copper and zinc against various ore blending ratios. The OreMet optimizer unit models were used to compute the product value of the concentrates. Increasing ore type 1 in the blend consistently improves both metallurgical performance and total product value. The optimal flotation circuit configurations (scenario ‘I’) for Cu and Zn were identified in scenarios 'F' and H', respectively. Further improvement was achieved by mass pull optimization (scenario 'J') by optimizing airflow rates in flotation banks using the Monte Carlo algorithm. Specifically, for ore type 1, scenario 'J' increased Cu recovery by 11.6% and Zn recovery by 7.5%, leading to a 15.3% increase in total product value. For ore type 2, scenario 'J' increased Cu recovery by 18.4% and Zn recovery by 5%, resulting in a 45% increase in total product value. This approach validates the applicability of this research work to full-scale metallurgical plants.