Predicting electronic meter reliability

The parts count method of predicting reliability described in the paper is probably the most simple method of reliability prediction and generally produces worst case results. Its basic assumption is that all meter components are ascribed a similar criticality failure and a failure of any component is assumed to cause a system failure. In practice of course this is not true. It is also assumed that all components fail at a constant rate for the period being considered. However, an exponential failure distribution for all electronic components is known to be realistic. A prediction of reliability assists in proving a design and also enables meter operators to plan a more efficient meter change programme. Nevertheless, it is important to acknowledge that any prediction is a best guess which may prove to be off target in use. For this reason OFFER intends to continue sample surveys of meter types at 5 year intervals. Depending on the results of these surveys the certification life will be revised as necessary