Reference Class Forecasting of System Reliability

View at publisher
Authors: Harry Jones
Cited by: 0
References: 6

Abstract

Initial estimates of system reliability are often very unrealistic and much too high, due to excessive optimism and the neglect of past results. A similar problem often occurs with project cost estimates, and the method of reference class forecasting has been used to improve their accuracy. The cost of a planned project is estimated based on the costs of a group of similar projects. Reference class forecasting can also be used to more accurately estimate system reliability. The key prerequisite is to find a set of completed similar projects with data on their achieved reliability and on any system factors that are correlated with reliability. Reliability has been found to depend on system cost and design complexity. The system failure rate usually decreases with higher system cost, due to investments in improving reliability, but it may increase in certain situations. The system failure rate typically increases with measures of system complexity.

Fields of Study & Subjects

Reliability and Maintenance Optimization Software Reliability and Analysis Research Construction Project Management and Performance

Publication details

Journal / source
Annual Reliability and Maintainability Symposium (RAMS)
Pages
1–6
Publication date
26 January 2026
Language
EN