Simulating Backfill Operations for Underground Mining

Problem

Sibanye-Stillwater is a multinational mining company based in South Africa. It operates platinum group metals (PGM) and gold mines in the Americas and Africa. The company’s processes involve mining, transporting ore and waste to the surface, milling ore into powder, and backfilling mined-out cavities with tailings from milling operations.

The circular dependency between mining and backfilling operations, along with variability for each task, makes it difficult to develop a realistic short-term schedules and efficiently deploy equipment and people to various tasks.

Sibanye needed to develop a high-level annual mine plan for budgeting purposes and then detailed monthly plans for executing operations. The detailed plans should include all the constraints related to equipment, staffing, and ore body.

Sibanye’s immediate goal for a process digital twin was to understand the interdependence between underground mine operations, backfill plant capabilities, and related logistical operations. The model was designed to help the mining company understand how bottlenecks move through their operations, to help identify which resources are constraining underground mining production increases, and to understand where capital investments are needed in backfill operations.

Solution

For this project, Sibanye-Stillwater involved engineers from MOSIMTEC – an American company that provides technology consulting services. They used AnyLogic’s capabilities for mining simulation and developed a flexible discrete-event simulation model so that Sibanye engineers could generate a backfill schedule given different round-level mine plans.

The model allowed engineers to test the impact of daily changes in the tailings supply rate.

The model contained objects that represented mining, backfilling, milling, and logistics operations. The AnyLogic Fluid library was used to model the flow of tailings, sand, and paste through a network of underground pipes. Preliminary analysis indicated that the flow representation in the model closely matched real-world flow behavior.

The model could be run via an Excel user interface. It allowed engineers to change mine plans, processing times, monthly resource schedules, and backfill types. The interface also allowed engineers to understand key model outputs in the Excel environment.

Results

The digital twin allowed engineers to plan weekly schedules, periodic maintenance, and long-term planning more promptly and accurate. Since constraints and real-world operations are accounted for from a site-wide perspective, Stillwater can now plan the distribution of tailings from the mill to each backfill area. The system replaces average estimates for backfilling with real values and constraints from blackfilling operations so that mining teams now have more accurate completion estimates.

Postes connexes