This is not my story, so I sure hope the chap who told it to me won't mind my using it. The reason is the story is firstly true, and secondly it highlights the point that we're trying to make! An integrated circuit wafer fabrication plant, comprising of 6 identical lines, each with 500 seperate processes (yes, this is what they take!), was suffering interruptions on a 'once a month' scale. These were simply major dips but would upset the plant so badly that the whole process would have to be stopped, work in progress scrapped, and started up all over again. The obvious solution was to install UPSs on all 3000 parts of the system. The consultancy firm involved specified units with a 10-year Mean Time Between Failure (MTBF). This sounds impressive and by all accounts should have left them up and running. Wrong!!! Those involved with manufacturing will clearly see the fault made by the consultancy firm. For those who are not too familiar, let us explain. 10 years represents approximately 3652 days. The MTBF is just that, a mean. Take 3652 and divide this by 3000 units and the MTBF of the system now falls to 1 failure every 1.2 days! To prove the point, this is exactly how may failures the wafer fab were having. This is what we, in the industry, refer to as a 'classic mistake'! We later investigate "adapt the scenario before employing more", and we cover issues such as the story above.
© 03.08.02 |