{"id":{"repo_id":"cambridge","oai_identifier":"oai:www.repository.cam.ac.uk:1810/314810"},"canonical_url":"https://search.dev.ndltd.org/etd/cambridge/oai:www.repository.cam.ac.uk:1810/314810","repository":{"repo_id":"cambridge","name":"Cambridge University","base_url":"https://api.repository.cam.ac.uk/server/oai/request"},"display":{"title":"Surrogate Model Optimisation for PWR Fuel Management","abstract":"Pressurised Water Reactor (PWR) fuel management is an operational problem for nuclear operators, requiring solutions on a regular basis throughout the life of the plant. A variety of conflicting factors and changing goals mean that fuel loading pattern design problems are multiobjective and, by design, have many input variables. This causes a combinatorial explosion, known as the ‘curse of dimensionality’, which makes these complex problems difficult to investigate. In this thesis, the method of surrogate model optimisation is adapted to PWR loading pattern generation. Surrogate models are developed based around three approaches: deep learning methods (convolutional neural networks and multi-layer perceptrons), the fission matrix and simulated quantum annealing. The models are used to predict core parameters of reactors in simplified optimisation scenarios for a microcore, a small modular reactor, and a ‘standard’ PWR. The experiments with deep learning models show that competitive results can be obtained for training sets using a much lower number of simulations than direct optimisation. Fission matrix experiments demonstrate the method to predict core parameters for the first time, with interesting preliminary results. Novel experiments using simulated quantum annealing demonstrate the technique is able to generate loading patterns by following heuristic rules and is suitable for application to custom optimisation hardware. The principal contribution of this work is to show that surrogate model optimisation can be used to augment fuel loading pattern optimisation, generating competitive results and providing enormous computational cost reduction and thus permitting more investigation within a given computational budget. These methods can also make use of new computational hardware such as neural chips and quantum annealers. The promising methods developed in this thesis thus provide candidate implementations that can bring the benefits of these innovations to the sphere of nuclear engineering.","abstract_html":"Pressurised Water Reactor (PWR) fuel management is an operational problem for nuclear operators, requiring solutions on a regular basis throughout the life of the plant. A variety of conflicting factors and changing goals mean that fuel loading pattern design problems are multiobjective and, by design, have many input variables. This causes a combinatorial explosion, known as the ‘curse of dimensionality’, which makes these complex problems difficult to investigate. In this thesis, the method of surrogate model optimisation is adapted to PWR loading pattern generation. Surrogate models are developed based around three approaches: deep learning methods (convolutional neural networks and multi-layer perceptrons), the fission matrix and simulated quantum annealing. The models are used to predict core parameters of reactors in simplified optimisation scenarios for a microcore, a small modular reactor, and a ‘standard’ PWR. The experiments with deep learning models show that competitive results can be obtained for training sets using a much lower number of simulations than direct optimisation. Fission matrix experiments demonstrate the method to predict core parameters for the first time, with interesting preliminary results. Novel experiments using simulated quantum annealing demonstrate the technique is able to generate loading patterns by following heuristic rules and is suitable for application to custom optimisation hardware. The principal contribution of this work is to show that surrogate model optimisation can be used to augment fuel loading pattern optimisation, generating competitive results and providing enormous computational cost reduction and thus permitting more investigation within a given computational budget. These methods can also make use of new computational hardware such as neural chips and quantum annealers. The promising methods developed in this thesis thus provide candidate implementations that can bring the benefits of these innovations to the sphere of nuclear engineering.","abstract_has_math":false,"creators":["Whyte, Andrew"],"institution":"University of Cambridge","degree_name":"Doctor of Philosophy (PhD)","degree_level":"Doctoral","degree_discipline":null,"degree_department":null,"school":null,"contributors":[],"advisors":["Parks, Geoff"],"committee_chairs":[],"committee_members":[],"year":2020,"date_issued":"2020-07-31","date_published":"2020-07-31","updated_at":"2026-07-22T22:24:20Z","subjects":["PWR","deep learning","quantum annealing","surrogate model optimisation","optimisation","fuel loading pattern","Multiobjective optimisation","fission matrix","NSGA2","iterative optimisation","optimization","neural network","convolutional neural network","CNN","python","nuclear engineering","nuclear","nuclear fuel","evolutionary algorithm"],"languages":["eng"],"rights":[],"rights_urls":["https://apollo8-f-pro.lib.cam.ac.uk/bitstreams/b38a25d1-3fb1-4023-b19d-5d451e742ed6/download","https://creativecommons.org/licenses/by-nc-sa/4.0/"],"identifier_entries":[{"key":"dc:creator.authoridentifier","label":"Author Identifier","values":["0000000181885047"],"render_values":[{"text":"0000-0001-8188-5047","href":"https://orcid.org/0000-0001-8188-5047","code":true}]}]},"links":{"outbound_url":"https://doi.org/10.17863/CAM.61917","outbound_label":"DOI","outbound_source":"dc:identifier.doi"},"metadata_groups":[{"id":"people","label":"People","entries":[{"key":"dc:contributor.advisor","label":"Advisor","values":["Parks, Geoff"]},{"key":"dc:creator","label":"Author","values":["Whyte, Andrew"]},{"key":"dc:creator.authoridentifier","label":"Author Identifier","values":["0000000181885047"]}]},{"id":"academic_context","label":"Academic Context","entries":[{"key":"dc:date.issued","label":"Date","values":["2020-07-31"]},{"key":"dc:publisher.institution","label":"Dc Publisher Institution","values":["University of Cambridge"]},{"key":"dc:relation.isreferencedby.uri","label":"Dc Relation Isreferencedby URI","values":["https://www.repository.cam.ac.uk/handle/1810/314810"]},{"key":"dc:type","label":"Dc Type","values":["Thesis"]},{"key":"dc:type.qualificationlevel","label":"Dc Type Qualificationlevel","values":["Doctoral"]},{"key":"dc:type.qualificationname","label":"Dc Type Qualificationname","values":["Doctor of Philosophy (PhD)"]}]},{"id":"subjects_keywords","label":"Subjects and Keywords","entries":[{"key":"dc:subject","label":"Dc Subject","values":["PWR","deep learning","quantum annealing","surrogate model optimisation","optimisation","fuel loading pattern","Multiobjective optimisation","fission matrix","NSGA2","iterative optimisation","optimization","neural network","convolutional neural network","CNN","python","nuclear engineering","nuclear","nuclear fuel","evolutionary algorithm"]}]},{"id":"language_rights","label":"Language and Rights","entries":[{"key":"dc:language","label":"Dc Language","values":["eng"]},{"key":"dc:rights","label":"Dc Rights","values":["https://apollo8-f-pro.lib.cam.ac.uk/bitstreams/b38a25d1-3fb1-4023-b19d-5d451e742ed6/download","https://creativecommons.org/licenses/by-nc-sa/4.0/"]}]},{"id":"identifiers","label":"Identifiers","entries":[{"key":"dc:identifier.doi","label":"DOI","values":["10.17863/CAM.61917"]},{"key":"dc:identifier.uri","label":"Identifier URI","values":["https://apollo8-f-pro.lib.cam.ac.uk/bitstreams/85860a0f-07c3-41ea-9fcc-12695575f43c/download"]}]},{"id":"additional","label":"Additional Metadata","entries":[{"key":"dc:description.abstract","label":"Abstract","values":["Pressurised Water Reactor (PWR) fuel management is an operational problem for nuclear operators, requiring solutions on a regular basis throughout the life of the plant. A variety of conflicting factors and changing goals mean that fuel loading pattern design problems are multiobjective and, by design, have many input variables. This causes a combinatorial explosion, known as the ‘curse of dimensionality’, which makes these complex problems difficult to investigate. In this thesis, the method of surrogate model optimisation is adapted to PWR loading pattern generation. Surrogate models are developed based around three approaches: deep learning methods (convolutional neural networks and multi-layer perceptrons), the fission matrix and simulated quantum annealing. The models are used to predict core parameters of reactors in simplified optimisation scenarios for a microcore, a small modular reactor, and a ‘standard’ PWR. The experiments with deep learning models show that competitive results can be obtained for training sets using a much lower number of simulations than direct optimisation. Fission matrix experiments demonstrate the method to predict core parameters for the first time, with interesting preliminary results. Novel experiments using simulated quantum annealing demonstrate the technique is able to generate loading patterns by following heuristic rules and is suitable for application to custom optimisation hardware. The principal contribution of this work is to show that surrogate model optimisation can be used to augment fuel loading pattern optimisation, generating competitive results and providing enormous computational cost reduction and thus permitting more investigation within a given computational budget. These methods can also make use of new computational hardware such as neural chips and quantum annealers. The promising methods developed in this thesis thus provide candidate implementations that can bring the benefits of these innovations to the sphere of nuclear engineering."]},{"key":"dc:format.checksum.md5","label":"Dc Format Checksum Md5","values":["53ef74878017861109bd2d18f85f16c7","353adac0d1ebdfd65ab16480263c3c87"]},{"key":"dc:title","label":"Title","values":["Surrogate Model Optimisation for PWR Fuel Management"]}]}],"canonical_facts":{"dc:contributor.advisor":["Parks, Geoff"],"dc:creator":["Whyte, Andrew"],"dc:creator.authoridentifier":["0000000181885047"],"dc:date.issued":["2020-07-31"],"dc:description.abstract":["Pressurised Water Reactor (PWR) fuel management is an operational problem for nuclear operators, requiring solutions on a regular basis throughout the life of the plant. A variety of conflicting factors and changing goals mean that fuel loading pattern design problems are multiobjective and, by design, have many input variables. This causes a combinatorial explosion, known as the ‘curse of dimensionality’, which makes these complex problems difficult to investigate. In this thesis, the method of surrogate model optimisation is adapted to PWR loading pattern generation. Surrogate models are developed based around three approaches: deep learning methods (convolutional neural networks and multi-layer perceptrons), the fission matrix and simulated quantum annealing. The models are used to predict core parameters of reactors in simplified optimisation scenarios for a microcore, a small modular reactor, and a ‘standard’ PWR. The experiments with deep learning models show that competitive results can be obtained for training sets using a much lower number of simulations than direct optimisation. Fission matrix experiments demonstrate the method to predict core parameters for the first time, with interesting preliminary results. Novel experiments using simulated quantum annealing demonstrate the technique is able to generate loading patterns by following heuristic rules and is suitable for application to custom optimisation hardware. The principal contribution of this work is to show that surrogate model optimisation can be used to augment fuel loading pattern optimisation, generating competitive results and providing enormous computational cost reduction and thus permitting more investigation within a given computational budget. These methods can also make use of new computational hardware such as neural chips and quantum annealers. The promising methods developed in this thesis thus provide candidate implementations that can bring the benefits of these innovations to the sphere of nuclear engineering."],"dc:format.checksum.md5":["53ef74878017861109bd2d18f85f16c7","353adac0d1ebdfd65ab16480263c3c87"],"dc:identifier.doi":["10.17863/CAM.61917"],"dc:identifier.uri":["https://apollo8-f-pro.lib.cam.ac.uk/bitstreams/85860a0f-07c3-41ea-9fcc-12695575f43c/download"],"dc:language":["eng"],"dc:publisher.institution":["University of Cambridge"],"dc:relation.isreferencedby.uri":["https://www.repository.cam.ac.uk/handle/1810/314810"],"dc:rights":["https://apollo8-f-pro.lib.cam.ac.uk/bitstreams/b38a25d1-3fb1-4023-b19d-5d451e742ed6/download","https://creativecommons.org/licenses/by-nc-sa/4.0/"],"dc:subject":["PWR","deep learning","quantum annealing","surrogate model optimisation","optimisation","fuel loading pattern","Multiobjective optimisation","fission matrix","NSGA2","iterative optimisation","optimization","neural network","convolutional neural network","CNN","python","nuclear engineering","nuclear","nuclear fuel","evolutionary algorithm"],"dc:title":["Surrogate Model Optimisation for PWR Fuel Management"],"dc:type":["Thesis"],"dc:type.qualificationlevel":["Doctoral"],"dc:type.qualificationname":["Doctor of Philosophy (PhD)"]},"updated_at":"2026-07-22T22:24:20Z"}