Burr, John T. (1989, ASQC) Rochester Quality Associates, Churchville, NY
One of the questions most often asked in conferences lately has been how statistical process control can be applied to many short production runs. This is most commonly asked by machine or tool and die companies. It is just as important, however, in chemical manufacturing processes. In this latter case, several runs of one chemical may be manufactured in a reactor system followed by a sequence of several runs of another chemical. It may be weeks, months or years before the first chemical is made again. If a lack of control is in evidence on a control chart of the first chemical, seeking a cause is very difficult. In this paper, several simple statistical techniques that look at the process regardless of what chemical is being manufactured at the time, will be discussed. The use of deviation control charts require that the specification limits be close to the same width for all the chemicals manufactured in the reactor system. The Q-statistic, a transformation of the data, which includes the aim and the tolerance width, will be discussed. The Q-statistica can be used with data from chemicals that have specification limits of different widths as wellas for different tests. Out-of-control conditions can be spotted immediately while it is most easy to identify whether the cause is in the raw material or in the manufacturing system. Trends and runs can be used to detect subtle system changes. It is further interesting to note that the Q-statistic also results in a very simple determination of process capability for the manufacturing system. Several chemical manufacturing examples will be shown.
Short runs,Chemical and process industries