A U-net based reconstruction of high-fidelity simulation results for flow around a ship hull based on low-fidelity inviscid flow simulation

In this study, neural networks are trained to transform inviscid simulation data for flow around a ship hull into data representative of viscous flow simulations. The objective is to provide high-fidelity viscous flow simulation data using machine learning in conjunction with inviscid flow simulatio...

Full description

Saved in:
Bibliographic Details
Main Authors: Dayeon Kim, Jeongbeom Seo, Inwon Lee
Format: Article
Language:English
Published: Elsevier 2025-01-01
Series:International Journal of Naval Architecture and Ocean Engineering
Subjects:
Online Access:http://www.sciencedirect.com/science/article/pii/S2092678225000342
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849254616418484224
author Dayeon Kim
Jeongbeom Seo
Inwon Lee
author_facet Dayeon Kim
Jeongbeom Seo
Inwon Lee
author_sort Dayeon Kim
collection DOAJ
description In this study, neural networks are trained to transform inviscid simulation data for flow around a ship hull into data representative of viscous flow simulations. The objective is to provide high-fidelity viscous flow simulation data using machine learning in conjunction with inviscid flow simulation results, which are significantly less time-consuming to generate. This approach has the potential to accelerate high-fidelity flow simulations by a factor of more than 100, enabling simulation-based design for ship hulls with numerous repetitive cases. To create the training dataset, a variety of hull forms are generated from six baseline hull forms using parametric modification function techniques. Inviscid and viscous flow data for each hull are obtained through potential flow analysis and computational fluid dynamics - simulations, respectively. The neural network structure and hyperparameters are subsequently optimized through parametric studies. The trained neural networks are then employed to predict viscous flow simulation data based on inputs comprising inviscid flow data and hull form geometry. The results demonstrate that the neural networks successfully predicted both the pressure distribution around the hull and the free surface elevation. Notably, the ability to predict the free surface elevation is significant, given that inviscid flow simulations inherently lack this capability. Additionally, the neural network's dimensionality reduction feature is utilized to visualize how the flow and hull form data were clustered within the latent space based on baseline hull forms and ship speed.
format Article
id doaj-art-844e4adfd5eb43e29c52f1f69f7529a5
institution Kabale University
issn 2092-6782
language English
publishDate 2025-01-01
publisher Elsevier
record_format Article
series International Journal of Naval Architecture and Ocean Engineering
spelling doaj-art-844e4adfd5eb43e29c52f1f69f7529a52025-08-20T03:56:04ZengElsevierInternational Journal of Naval Architecture and Ocean Engineering2092-67822025-01-011710067610.1016/j.ijnaoe.2025.100676A U-net based reconstruction of high-fidelity simulation results for flow around a ship hull based on low-fidelity inviscid flow simulationDayeon Kim0Jeongbeom Seo1Inwon Lee2Ship and Offshore Performance Research Center, Samsung Heavy Industries Co. Ltd., Daejon, Republic of KoreaDepartment of Naval Architecture and Ocean Engineering, Pusan National University, Busan, Republic of KoreaDepartment of Naval Architecture and Ocean Engineering, Pusan National University, Busan, Republic of Korea; Global Core Research Center for Ships and Offshore Plants, Pusan National University, Busan, Republic of Korea; Corresponding author.In this study, neural networks are trained to transform inviscid simulation data for flow around a ship hull into data representative of viscous flow simulations. The objective is to provide high-fidelity viscous flow simulation data using machine learning in conjunction with inviscid flow simulation results, which are significantly less time-consuming to generate. This approach has the potential to accelerate high-fidelity flow simulations by a factor of more than 100, enabling simulation-based design for ship hulls with numerous repetitive cases. To create the training dataset, a variety of hull forms are generated from six baseline hull forms using parametric modification function techniques. Inviscid and viscous flow data for each hull are obtained through potential flow analysis and computational fluid dynamics - simulations, respectively. The neural network structure and hyperparameters are subsequently optimized through parametric studies. The trained neural networks are then employed to predict viscous flow simulation data based on inputs comprising inviscid flow data and hull form geometry. The results demonstrate that the neural networks successfully predicted both the pressure distribution around the hull and the free surface elevation. Notably, the ability to predict the free surface elevation is significant, given that inviscid flow simulations inherently lack this capability. Additionally, the neural network's dimensionality reduction feature is utilized to visualize how the flow and hull form data were clustered within the latent space based on baseline hull forms and ship speed.http://www.sciencedirect.com/science/article/pii/S2092678225000342Artificial intelligence (AI)Machine learningU-netPixel-level predictionHull form design
spellingShingle Dayeon Kim
Jeongbeom Seo
Inwon Lee
A U-net based reconstruction of high-fidelity simulation results for flow around a ship hull based on low-fidelity inviscid flow simulation
International Journal of Naval Architecture and Ocean Engineering
Artificial intelligence (AI)
Machine learning
U-net
Pixel-level prediction
Hull form design
title A U-net based reconstruction of high-fidelity simulation results for flow around a ship hull based on low-fidelity inviscid flow simulation
title_full A U-net based reconstruction of high-fidelity simulation results for flow around a ship hull based on low-fidelity inviscid flow simulation
title_fullStr A U-net based reconstruction of high-fidelity simulation results for flow around a ship hull based on low-fidelity inviscid flow simulation
title_full_unstemmed A U-net based reconstruction of high-fidelity simulation results for flow around a ship hull based on low-fidelity inviscid flow simulation
title_short A U-net based reconstruction of high-fidelity simulation results for flow around a ship hull based on low-fidelity inviscid flow simulation
title_sort u net based reconstruction of high fidelity simulation results for flow around a ship hull based on low fidelity inviscid flow simulation
topic Artificial intelligence (AI)
Machine learning
U-net
Pixel-level prediction
Hull form design
url http://www.sciencedirect.com/science/article/pii/S2092678225000342
work_keys_str_mv AT dayeonkim aunetbasedreconstructionofhighfidelitysimulationresultsforflowaroundashiphullbasedonlowfidelityinviscidflowsimulation
AT jeongbeomseo aunetbasedreconstructionofhighfidelitysimulationresultsforflowaroundashiphullbasedonlowfidelityinviscidflowsimulation
AT inwonlee aunetbasedreconstructionofhighfidelitysimulationresultsforflowaroundashiphullbasedonlowfidelityinviscidflowsimulation
AT dayeonkim unetbasedreconstructionofhighfidelitysimulationresultsforflowaroundashiphullbasedonlowfidelityinviscidflowsimulation
AT jeongbeomseo unetbasedreconstructionofhighfidelitysimulationresultsforflowaroundashiphullbasedonlowfidelityinviscidflowsimulation
AT inwonlee unetbasedreconstructionofhighfidelitysimulationresultsforflowaroundashiphullbasedonlowfidelityinviscidflowsimulation