Figure 5. One MCTS flow assignment during the representative simulation. In comparison, the same plot for the greedy benchmark (not shown here) sends all mine outputs to one pile, blending them together, and then sends that material almost exclusively to flotation. Given that the greedy algorithm can solve deterministic problems to optimality, this result implies that the optimal one-day solution in this scenario is to blend all feedstocks together (since there is an even and equal distribution of low and high quality feedstocks) and primarily float them, which is largely consistent with industry approaches. 5. DISCUSSION The two scenarios demonstrate the importance of planning ahead given the uncertainty that arises from feedstock and process variability. When the feedstock changes day-by-day, as in Scenario A, the blending formulas and flow assignments must be able to adapt. Figure 3 and Figure 4 together clearly demonstrate the capability of an MCTS solver to maintain consistent profits even as the feedstock composition varies dramatically through stockpiling and smart blending decisions. Even when the feedstock composition only varies slightly, and a set of different feedstock qualities are available (as in Scenario B), slight deviations in process conditions can make it difficult to anticipate product quality in advance. A stochastic, long-horizon approach such as MCTS is able to better account for these uncertainties in fulfilling client contracts while achieving higher profits at the same time, as demonstrated in Tables 6 and 7. The performance of greedy-guided MCTS relative to MCTS is informative in illustrating to what degree the greedy solution is a good prior versus a local optimum and potential trapping state. ggMCTS is underperforms MCTS in Scenario A because the greedy prior is actively detrimental, biasing the solver towards solutions that deprioritize long-term planning. However, since the feedstock varies less day-to-day in Scenario B, the greedy solution is a better starting point than nothing, so ggMCTS is almost always able to find a better solution than MCTS with the same computational budget. These results indicate that a greedy prior is useful in scenarios with
RkJQdWJsaXNoZXIy MTM0Mzk2