University of Southampton
Model validation and uncertainty qualification for the preliminary aero-engine design process
Abstract
dc:description.abstractThis thesis investigates the design decisions taken during the preliminary aero-engine design process where the amount of knowledge is limited, although deciding on fuel efficiency, noise, emissions, weight and overall performance occurs within this stage. In order to not commit all resources during this phase, those decisions are made using low fidelity models. Unfortunately, the results from low fidelity models lack accuracy, so there is a natural need to take this into account. Improving those low fidelity methods via uncertainty quantification methods is the main theme of this thesis. <br/><br/>In order to create accurate models for the preliminary design stage of the aero engine, a probabilistic framework was created and implemented. This framework is based upon suggestions from the literature and was constructed from two main components: expert systems as well as Bayesian inference. The way in which this was developed will be shown in detail later. This framework was applied to three aero-engine related case studies which reflect the need to have more knowledge available during the preliminary stage. <br/><br/>The results obtained from the framework have the form of predictions which offer information which was not available otherwise. For the fan-blade off case study, the posterior predictive distributions show what the characteristics of the most likely events are, and this information can be used to make the detailed design stage less expensive by doing fewer finite element analysis simulations. For the grain growth case study, the results show what probability distributions the manufacturing inputs should have in order to keep the size of the grain within certain limits in order to maximize the life cycles. Finally, the results obtained with regards to fatigue failure due to non-metallic particle inclusions show how component life as well as failure cause can be obtained. This type of knowledge was not previously available in the literature, and making use of it can avoid removing components from their service too early.
Degree
thesis:*- Name dc:type.qualificationname
- Ph.D.
- Level dc:type.qualificationlevel
- doctoral
- Grantor dc:publisher.institution
- University of Southampton
- Year dc:date.issued
- 2019
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Profir, Bogdan
- Advisor dc:contributor.advisor
-
- Eres, Murat
- ORCID person:identifier.orcid
Rights
- Language dc:language
- English, eng