We have applied fuzzy set theory to some exposure control problems encountered in the machining and the manufacturing of beryllium parts. A portion of that work is presented in this paper. The major driving force for using fuzzy techniques in this case rather than classical statistical process control is that beryllium exposure is very task dependent and our manufacturing plant is quite atypical. We believe that standard statistical techniques will produce too many false alarms. The beryllium plant produces parts on a daily basis, but every day is different. Some days many parts are produced and some days only a few. Some times the parts are large and sometimes the parts are small. Some machining cuts are rough and some are fine. These factors and others make it hard to define a typical day. The problem of concern, for this study, is the worker beryllium exposure. Even though the plant is new and very modern and the exposure levels are expected to be well below the required levels, the Department of Energy has demanded that the levels for this plant be well below required levels. The control charts used to monitor this process are expected to answer two questions: 1. Is the process out of Control. Do we need to instigate special controls such as requiring workers to use respirators. 2. Are new, previously untested, controls making a difference. The standard Shewhart type control charts, based on consistent plant operating conditions do not adequately answer this question. The approach described here is based upon a fuzzy modification to the Shewhart X Bar- R chart. This approach is expected to yield better results than work based upon the classical probabilistic control chart.
PDF
download
878987797
028-85220240
