Unexpected Results

“I’m running MiningMath, but it’s mining a less profitable area, or even waste, or it’s not achieving the specific constraint I’ve set… Why is that?”

To understand this, it’s important to recognize that not everything that is aimed to be incorporate into a mining project is mathematically feasible when attempting to respect all constraints simultaneously. Handling multiple, complex constraints increases the likelihood of either not finding or not having feasible solutions.

However, MiningMath is flexible enough to accommodate all types of constraints and always delivers a solution, even if it requires relaxing some constraints, reaching different areas, or reducing the NPV to achieve feasibility.

When an infeasible solution is detected, the algorithm determines which less critical constraints should be relaxed, and generates warnings in the report. Moreover, MiningMath allows you to maintain a global perspective, helping you identify potential bottlenecks that may arise later.

It’s also crucial to confirm that all blocks have coherent values for the variables used in the constraints. For instance, MiningMath might target waste if an average constraint is set in a way that selects undesirable blocks to satisfy the imposed restriction.

To properly identify the root cause of such situations, we recommend performing a bottleneck analysis, adding one constraint at a time to evaluate its impact on the project. Alternatively, you can remove constraints one by one, starting with Sum and Average constraints, until the issue no longer occurs.

For a more in-depth analysis, review the generated charts and check if any constraints are hitting their maximum or minimum limits, indicating a bottleneck. This suggests that the algorithm might choose a better solution if that specific constraint were relaxed.

To address these issues, consider the following options:

• Relax constraints where possible.
• Use the restrict mining option to focus the solution on a specific area.
• Adjust block values for the variables used in constraints to facilitate meeting them.

Therefore, MiningMath can accommodate all types of constraints in its Single-Step Optimization approach. The only requirement is that the modeling of values and constraints is carefully evaluated.

If you have similar issues, feel free to share your experiences in this topic.