Thesis
Evaluation of Bayes linear modelling to support reliability assessment during procurement
- Creator
- Rights statement
- Awarding institution
- University of Strathclyde
- Date of award
- 2008
- Thesis identifier
- T12027
- Department, School or Faculty
- Abstract
- A major task facing a number of different procuring agencies is the assessment of the developmental reliability of a product as contractors incrementally provide information. These agencies would like to use structured quantitative methodologies, as opposed to unstructured assessment currently adopted, to evaluate reliability throughout the life of a development programme as additional information becomes available. Due to resource constraints, any developed methodology must be cost and time efficient. This research attempts to develop a methodology for customers that is capable of assessing the information presented by a contractor in a reliability case to support decision making throughout the procurement process. Bayesian Belief Networks (BBN’s) have many features that make them useful for modelling complex problems as they allow for the combination of multiple sources of information, graphically represent decision problems and can formally incorporate subjective beliefs into the modelling. The Bayes linear methodology, although similar in spirit, has advantages over full probabilistic approaches, such as BBN’s, when modelling complex problems. In contrast to these “traditional” Bayesian approaches, the Bayes linear methodology offers a quick, simple method to perform inference. The primitive used in a Bayes linear analysis is expectation rather than probability and as such, the elicitation burden is reduced. Sensitivity analysis is difficult and time-consuming to carry out within a probabilistic framework. However, it is an important and necessary task to validate a model. Due to the way in which the Bayes linear inputs are elicited and the ease with which adjusting beliefs can be carried out, it is possible to carry out extensive sensitivity analysis to assess whether the decision taken is robust to small changes in the decision maker’s prior beliefs. This thesis investigates the different ways in which processes, such as the elicitation protocol, the construction of Bayes linear networks and sensitivity analysis, can be developed to build a Bayes linear model on real-life problems. The United Kingdom Ministry of Defence (MOD) is one such organisation that requires a structured model to track reliability. The usefulness of the methodology was demonstrated on two ‘live’ projects through the use of Action Research. The first project focused on supporting a MOD decision maker in deciding which of three contractors to choose from. The model assessed which of the systems the decision maker believed was currently the most reliable and which reliability programme was reducing her uncertainty of the reliability by the most. The second project focused on supporting a MOD decision maker in assessing the in-service reliability of two systems prior to them entering service. For both systems, test data was used along with expert opinion to assess the reliability of each system. In both projects, extensive sensitivity analysis was carried out to assess how changing the prior beliefs of the decision maker impacts on the final decision taken. The Bayes linear approach developed in this thesis provided excellent decision support to the decision makers in both projects.
- Advisor / supervisor
- Walls, Lesley
- Bedford, T.
- Resource Type
- EThOS ID
- uk.bl.ethos.487866
Relations
Items
| Thumbnail | Title | Date Uploaded | Visibility | Actions |
|---|---|---|---|---|
|
|
PDF of thesis T12027 | 2026-08-13 | Public | Download |