{"id":{"repo_id":"calpoly","oai_identifier":"oai:digitalcommons.calpoly.edu:theses-4862"},"canonical_url":"https://search.ndltd.org/etd/calpoly/oai:digitalcommons.calpoly.edu:theses-4862","repository":{"repo_id":"calpoly","name":"Cal Poly","base_url":"https://digitalcommons.calpoly.edu/do/oai/"},"display":{"title":"Reverse Design of Solid Rocket Grain Geometry","abstract":"The reverse design of solid rocket motors remains a challenging process due to the nonlinear relationship between the grain geometry and the internal ballistics. The reverse design process involves obtaining an optimal grain geometry given a target performance curve set by mission requirements, which is the inverse problem of the forward design. The purpose of this study is to assess the validity and implementation of neural network and simulated annealing algorithms to perform the reverse design process. A modified phase field model of the eikonal equation is derived to simulate the evolution of combustion surfaces over time, enabling burn back analysis as well as internal ballistics calculations. Initial grain geometries are generated using a parametric function called the superformula, which provides a flexible representation of complex surfaces with six parameters. The neural network is trained to predict the optimal grain geometry parameters from a given target chamber performance curve. Separately, simulated annealing is implemented to optimize the superformula parameters utilizing the error between target and calculated performance curves as an objective function. Since simulated annealing requires an initial grain geometry as a starting point for the optimization, a method of combining neural network and simulated annealing is proposed. This method performs simulated annealing on the outputs of the neural network to potentially reduce computational costs and increase practicality. Results from this study shows that a simple neural network is not able to perform the reverse design process alone while the simulated annealing and the combined method are. Additionally, the NN+SA method was able to reduce computational costs, making it the most viable method for reverse design for the specific cases studied. Although the exact reduction in computational costs was not able to be quantified due to the random initialization of SA, this study shows that the NN+SA method produces similar results to SA for worse case scenarios and up to 3000 seconds faster in the best case scenarios specifically explored in this study.","abstract_html":"The reverse design of solid rocket motors remains a challenging process due to the nonlinear relationship between the grain geometry and the internal ballistics. The reverse design process involves obtaining an optimal grain geometry given a target performance curve set by mission requirements, which is the inverse problem of the forward design. The purpose of this study is to assess the validity and implementation of neural network and simulated annealing algorithms to perform the reverse design process. A modified phase field model of the eikonal equation is derived to simulate the evolution of combustion surfaces over time, enabling burn back analysis as well as internal ballistics calculations. Initial grain geometries are generated using a parametric function called the superformula, which provides a flexible representation of complex surfaces with six parameters. The neural network is trained to predict the optimal grain geometry parameters from a given target chamber performance curve. Separately, simulated annealing is implemented to optimize the superformula parameters utilizing the error between target and calculated performance curves as an objective function. Since simulated annealing requires an initial grain geometry as a starting point for the optimization, a method of combining neural network and simulated annealing is proposed. This method performs simulated annealing on the outputs of the neural network to potentially reduce computational costs and increase practicality. Results from this study shows that a simple neural network is not able to perform the reverse design process alone while the simulated annealing and the combined method are. Additionally, the NN+SA method was able to reduce computational costs, making it the most viable method for reverse design for the specific cases studied. Although the exact reduction in computational costs was not able to be quantified due to the random initialization of SA, this study shows that the NN+SA method produces similar results to SA for worse case scenarios and up to 3000 seconds faster in the best case scenarios specifically explored in this study.","abstract_has_math":false,"creators":["Kang, Yechan"],"institution":null,"degree_name":"MS in Aerospace Engineering","degree_level":null,"degree_discipline":"Aerospace Engineering","degree_department":null,"school":null,"contributors":["Amuthan Ramabathiran","Aerospace Engineering","College of Engineering"],"advisors":[],"committee_chairs":[],"committee_members":[],"year":null,"date_issued":"","date_published":null,"updated_at":"2026-11-19T07:12:41Z","subjects":["Reverse design","simulated annealing","neural network","solid rocket motor","grain geometry","burn back analysis","Aerospace Engineering"],"languages":[],"rights":[],"rights_urls":[],"identifier_entries":[],"person_identifier_entries":[]},"links":{"outbound_url":"https://digitalcommons.calpoly.edu/theses/3198","outbound_label":"Repository record","outbound_source":"dc:identifier","record_url":"https://digitalcommons.calpoly.edu/theses/3198","record_label":"Repository record","record_kind":"record","record_source":"dc:identifier","download_url":null,"download_label":null,"download_source":null,"access_status":null,"embargo_until":null,"access_note":null},"source_record":{"url":"https://digitalcommons.calpoly.edu/do/oai/?verb=GetRecord&metadataPrefix=oai_etdms&identifier=oai%3Adigitalcommons.calpoly.edu%3Atheses-4862","prefix":"oai_etdms"},"metadata_groups":[{"id":"people","label":"People","entries":[{"key":"dc:contributor","label":"Contributor","values":["Amuthan Ramabathiran","Aerospace Engineering","College of Engineering"]},{"key":"dc:creator","label":"Author","values":["Kang, Yechan"]}]},{"id":"academic_context","label":"Academic Context","entries":[{"key":"dc:date.available","label":"Dc Date Available","values":["2025-12-11T08:00:00Z"]},{"key":"dc:date.created","label":"Dc Date Created","values":["2025-12-01T08:00:00Z"]},{"key":"thesis:degree_discipline","label":"Discipline","values":["Aerospace Engineering"]},{"key":"thesis:degree_name","label":"Degree Name","values":["MS in Aerospace Engineering"]}]},{"id":"subjects_keywords","label":"Subjects and Keywords","entries":[{"key":"dc:subject","label":"Dc Subject","values":["Reverse design","simulated annealing","neural network","solid rocket motor","grain geometry","burn back analysis","Aerospace Engineering"]}]},{"id":"identifiers","label":"Identifiers","entries":[{"key":"dc:identifier","label":"Identifier","values":["https://digitalcommons.calpoly.edu/theses/3198"]}]},{"id":"additional","label":"Additional Metadata","entries":[{"key":"dc:description.abstract","label":"Abstract","values":["The reverse design of solid rocket motors remains a challenging process due to the nonlinear relationship between the grain geometry and the internal ballistics. The reverse design process involves obtaining an optimal grain geometry given a target performance curve set by mission requirements, which is the inverse problem of the forward design. The purpose of this study is to assess the validity and implementation of neural network and simulated annealing algorithms to perform the reverse design process. A modified phase field model of the eikonal equation is derived to simulate the evolution of combustion surfaces over time, enabling burn back analysis as well as internal ballistics calculations. Initial grain geometries are generated using a parametric function called the superformula, which provides a flexible representation of complex surfaces with six parameters. The neural network is trained to predict the optimal grain geometry parameters from a given target chamber performance curve. Separately, simulated annealing is implemented to optimize the superformula parameters utilizing the error between target and calculated performance curves as an objective function. Since simulated annealing requires an initial grain geometry as a starting point for the optimization, a method of combining neural network and simulated annealing is proposed. This method performs simulated annealing on the outputs of the neural network to potentially reduce computational costs and increase practicality. Results from this study shows that a simple neural network is not able to perform the reverse design process alone while the simulated annealing and the combined method are. Additionally, the NN+SA method was able to reduce computational costs, making it the most viable method for reverse design for the specific cases studied. Although the exact reduction in computational costs was not able to be quantified due to the random initialization of SA, this study shows that the NN+SA method produces similar results to SA for worse case scenarios and up to 3000 seconds faster in the best case scenarios specifically explored in this study."]},{"key":"dc:title","label":"Title","values":["Reverse Design of Solid Rocket Grain Geometry"]}]}],"canonical_facts":{"dc:contributor":["Amuthan Ramabathiran","Aerospace Engineering","College of Engineering"],"dc:creator":["Kang, Yechan"],"dc:date.available":["2025-12-11T08:00:00Z"],"dc:date.created":["2025-12-01T08:00:00Z"],"dc:description.abstract":["The reverse design of solid rocket motors remains a challenging process due to the nonlinear relationship between the grain geometry and the internal ballistics. The reverse design process involves obtaining an optimal grain geometry given a target performance curve set by mission requirements, which is the inverse problem of the forward design. The purpose of this study is to assess the validity and implementation of neural network and simulated annealing algorithms to perform the reverse design process. A modified phase field model of the eikonal equation is derived to simulate the evolution of combustion surfaces over time, enabling burn back analysis as well as internal ballistics calculations. Initial grain geometries are generated using a parametric function called the superformula, which provides a flexible representation of complex surfaces with six parameters. The neural network is trained to predict the optimal grain geometry parameters from a given target chamber performance curve. Separately, simulated annealing is implemented to optimize the superformula parameters utilizing the error between target and calculated performance curves as an objective function. Since simulated annealing requires an initial grain geometry as a starting point for the optimization, a method of combining neural network and simulated annealing is proposed. This method performs simulated annealing on the outputs of the neural network to potentially reduce computational costs and increase practicality. Results from this study shows that a simple neural network is not able to perform the reverse design process alone while the simulated annealing and the combined method are. Additionally, the NN+SA method was able to reduce computational costs, making it the most viable method for reverse design for the specific cases studied. Although the exact reduction in computational costs was not able to be quantified due to the random initialization of SA, this study shows that the NN+SA method produces similar results to SA for worse case scenarios and up to 3000 seconds faster in the best case scenarios specifically explored in this study."],"dc:identifier":["https://digitalcommons.calpoly.edu/theses/3198"],"dc:subject":["Reverse design","simulated annealing","neural network","solid rocket motor","grain geometry","burn back analysis","Aerospace Engineering"],"dc:title":["Reverse Design of Solid Rocket Grain Geometry"],"thesis:degree_discipline":["Aerospace Engineering"],"thesis:degree_name":["MS in Aerospace Engineering"]},"updated_at":"2026-11-19T07:12:41Z"}