Robust planning and scheduling for assembly lines

This tool support the robust production planning based on a proactive, simulation-based method that applies regression functions in the optimization models to capture the effects of the dynamic underlying processes. That method relies on up-to-date production MES and SCADA data, in order to provide executable results. Moreover, it also determines the best operator control policies (operator-task assignment) for each product variants that provides the desired throughput besides the least losses possible. The tool is based on a simulation-based optimization method to predict the effect of disturbances during the production, and consider them already in the production planning phase (in a proactive way). The tool consists of three modules. The Simulation module includes stochastic and unpredictable effects (e.g. machine breakdowns, varying processing times, missing material etc.), to predict the best operator control policies (operator-task assignment), to maintain the target level of the KPI-s. Input to this module are processing and testing times (SCADA), possible operator-task assignments, historical reject rates. Outputs are best operator-task assignments for each products, capacity requirements. The Data analysis detects the patterns in the production data, and understand the relationship between the stochastic parameters (mainly the rework) and the true capacity requirements. Inputs to this module are from simulation (the capacity requirements and production volumes), while the outputs are regression models to predict the real capacity requirements considering the production batches and stochastic parameters (rework). The Optimization module applies the pre-calculated regression models, and plans the production (production batch sizes) accordingly. Inputs to this module are order volumes and capacity requirement functions (from the data analysis module), while its outputs are production lot sizes, time horizon, and resolution. The simulation module enables flexibility by offering the best operator task assignment for each product, enabling the assembly of different product with the same number of operators besides minimal losses (switch from one product to another in the same shift, with the same staff but different assignment). The data analysis module provides regression functions that predict the actual capacity requirements including the effects of possible reworks. Lastly, by integrating the regression models in the optimization constraints, the robustness of the production plans can be increased, as the effect of stochastic events are handled already in the planning phase.