Job shop control: In search of the key to delivery improvements

Abstract

The last major performance breakthroughs in job shop control stem from the 1980s and 1990s. We generate a new search direction for designing job shop control policies, providing a key to delivery improvements. Based on a common characteristic shared by the most effective job shop control policies, we posit that control should have a specific focus during high load periods. A probability analysis reveals that substantial periods of high load are common, and even occur under assumptions of stationarity and moderate utilization. Subsequent simulations show nearly all tardy deliveries can be attributed to high load periods; and that the success of the best control policies can be explained by their ability to switch focus specifically during these periods, from reducing the dispersion of lateness to speeding up the average throughput time. Building on this, we demonstrate that for example small capacity adjustments targeted at handling high load periods can improve the percentage tardy and other delivery-related performance measures to a much greater extent than the best existing policies. Sensitivity analysis confirms the robustness of this approach and identifies a performance frontier reflecting the trade-off between capacity resources used and delivery performance realized. We conclude that a paradigm shift in job shop research is required: instead of developing single policies for application under all conditions, new policies are needed that respond differently to temporary high load periods. The new paradigm can be used as a design principle for realizing improvements across a range of planning and control decisions relevant to job shops.