Economics of diagnosis

Abstract

Detecting the existence of a fault in complex systems is neither sufficient nor economical without diagnostics assisting in fault isolation and cost-effective repairs. This work attempts to put in economical terms the technical decisions involving diagnostics. It looks at the cost factors of poor diagnostics in terms of the accuracy and completeness of fault identification and the time and effort it takes to come to a final (accurate) repair decision. No Problems Found (NPF), Retest OK (RTOK), False Alarms, Cannot Duplicates (CND) and other diagnostic deficiencies can range from 30% to 60% of all repair actions. According to a 1995 survey run by the IEEE Reliability Society, the Air Transport Association (ATA) has determined that 4500 NPF events cost ATE member airlines $100 million annually. A U.S. Army study has shown that maintenance costs can be reduced by 25% if 70-80% of the items if had been repairing were to be discarded. Many of these situations can be overcome by investing in emerging technologies, such as Built-in (Self) Test (BIST) and expert diagnostic tools. The role of these tools is to minimize dependence on the skills, knowledge and experience of individuals, and thus overcome costs of inaccurate, inefficient, and incomplete diagnostics. Use of BIST can also directly reduce costs. This paper presents technical solutions and economic analyses showing to what extent such solutions provide a sufficient return on investment.