Copper production is a great example of facility with multiple constraints and interdependencies that requires simulation to support decisions. The conventional process-oriented approach implies using multiple local optima rules to simulate manufacturing operations. For example, every time the model needs to decide where to transport a ladle with hot metal, it looks for the closest available converter that can take the metal. Using such rules to model the copper plant leads to deadlocks such as having too much white metal and no available converters to process it.
To address this problem, we developed a scheduler as a separate module that generates a well-balanced schedule of the main production process using an iterative algorithm with stepwise constraints relaxation. The schedule is passed to the model that simulates plant operations trying to adhere to the schedule. The simulation considers cranes logistics, limited number of ladles, fluctuations of processing cycles and multiple other aspects. If the model starts falling behind the schedule, it requests the scheduler to generate a new schedule starting from the current state of the plant. This interaction between the scheduler and simulation model produces a realistic schedule that accounts for most constraints and interdependencies of copper production process.