Issue link: https://iconnect007.uberflip.com/i/340751
80 The PCB Magazine • July 2014 the marginal case for cycle time is the cyan line. I show the result of running the simulation with a WIP of 5. The yellow dot is the marginal throughput and the orange dot is the marginal cycle time. With the marginal case, all process times must be the same, the variation must be exponentially distributed, and there is no pass- ing of product. There are specific formulas that describe the resulting marginal case curves in Figure 2 [1] . By the way, this marginal case should re- mind you of the balanced production line for the CEO described above. What does this graph mean? If your actual performance is below the marginal case line for throughput, chances are you can improve your performance simply by reducing the variation in your process times or improving the varia- tion in your up time on your machines. If your performance is at or better then the marginal case line, then improving your performance may be difficult and you may have to make sys- temic changes to your manufacturing operation (e.g., add machines or people). Consider the impact we have experienced by reducing our lot sizes. What industry has done in decreasing lot sizes is to significantly reduce average cycle times. This is a good thing, but it has unintended consequences. What the indus- try in general does not realize is that too little WIP is as bad as too much WIP. As the curves above show, decreasing WIP decreases cycle time but it also decrease throughput. Here is where managers go astray. They love the shorter WHY REMOvING YOUR BOTTLENECK IS A BAD IDEA continues Figure 3: an unbalanced production line with plenty of capacity. note the second department with two machines has a utilization of over 90%, but the utilizations downstream of the bottleneck are only slightly above 50%. unbalanced lines tend to keep the bottleneck fed and prevent the bottleneck from being blocked.