Feasibility of Deep Neural Network Surrogate Models in Fluid Dynamics

This paper studies reduced-order-models for the fluid flow problem of a digital valve, and whether it may efficiently be formulated by a deep Artificial Neural Network (ANN) to model e.g. the valve flow, flow-induced force, stiction phenomena and steep local pressure gradients that arise before plun...

Full description

Bibliographic Details
Main Authors: Niels C. Bender, Torben Ole Andersen, Henrik C. Pedersen
Format: Article
Language:English
Published: Norwegian Society of Automatic Control 2019-04-01
Series:Modeling, Identification and Control
Subjects:
CFD
Online Access:http://www.mic-journal.no/PDF/2019/MIC-2019-2-1.pdf
id doaj-27e867898645494ab7132cd4d06b2c69
record_format Article
spelling doaj-27e867898645494ab7132cd4d06b2c692020-11-24T20:46:15ZengNorwegian Society of Automatic ControlModeling, Identification and Control0332-73531890-13282019-04-01402718710.4173/mic.2019.2.1Feasibility of Deep Neural Network Surrogate Models in Fluid DynamicsNiels C. BenderTorben Ole AndersenHenrik C. PedersenThis paper studies reduced-order-models for the fluid flow problem of a digital valve, and whether it may efficiently be formulated by a deep Artificial Neural Network (ANN) to model e.g. the valve flow, flow-induced force, stiction phenomena and steep local pressure gradients that arise before plunger impact, which may otherwise require CFD to be accurately modeled. Several methodologies are investigated to evaluate both the required computation time and the accuracy. The accuracy is benchmarked against CFD solutions of flows and forces. As basis for comparison an analytical model is proposed where some fitting parameters are allowed, and the equation is tested outside its fitting range. A similar model is built as a deep ANN which is trained with data from the analytical model to investigate the amount of data required for an ANN and its fitting capabilities. The results show that in higher dimensions the required training data can be maintained low if data is structured by a Latin Hypercube, otherwise the amount becomes infeasible. This makes an ANN surrogate feasible when compared to a look-up table, and may be expanded to higher dimension where dynamical effects are included. However, the required data and computational cost for this is too extensive for the valve design considered as basis for the analysis. Instead, for this specific problem, the derived analytical model is sufficient to describe the valve dynamics and reduces the computation time significantly.http://www.mic-journal.no/PDF/2019/MIC-2019-2-1.pdfArtificial Neural NetworksCFDDigital ValvesFlow-induced ForcesReduced Order ModelsLumped Parameter Models
collection DOAJ
language English
format Article
sources DOAJ
author Niels C. Bender
Torben Ole Andersen
Henrik C. Pedersen
spellingShingle Niels C. Bender
Torben Ole Andersen
Henrik C. Pedersen
Feasibility of Deep Neural Network Surrogate Models in Fluid Dynamics
Modeling, Identification and Control
Artificial Neural Networks
CFD
Digital Valves
Flow-induced Forces
Reduced Order Models
Lumped Parameter Models
author_facet Niels C. Bender
Torben Ole Andersen
Henrik C. Pedersen
author_sort Niels C. Bender
title Feasibility of Deep Neural Network Surrogate Models in Fluid Dynamics
title_short Feasibility of Deep Neural Network Surrogate Models in Fluid Dynamics
title_full Feasibility of Deep Neural Network Surrogate Models in Fluid Dynamics
title_fullStr Feasibility of Deep Neural Network Surrogate Models in Fluid Dynamics
title_full_unstemmed Feasibility of Deep Neural Network Surrogate Models in Fluid Dynamics
title_sort feasibility of deep neural network surrogate models in fluid dynamics
publisher Norwegian Society of Automatic Control
series Modeling, Identification and Control
issn 0332-7353
1890-1328
publishDate 2019-04-01
description This paper studies reduced-order-models for the fluid flow problem of a digital valve, and whether it may efficiently be formulated by a deep Artificial Neural Network (ANN) to model e.g. the valve flow, flow-induced force, stiction phenomena and steep local pressure gradients that arise before plunger impact, which may otherwise require CFD to be accurately modeled. Several methodologies are investigated to evaluate both the required computation time and the accuracy. The accuracy is benchmarked against CFD solutions of flows and forces. As basis for comparison an analytical model is proposed where some fitting parameters are allowed, and the equation is tested outside its fitting range. A similar model is built as a deep ANN which is trained with data from the analytical model to investigate the amount of data required for an ANN and its fitting capabilities. The results show that in higher dimensions the required training data can be maintained low if data is structured by a Latin Hypercube, otherwise the amount becomes infeasible. This makes an ANN surrogate feasible when compared to a look-up table, and may be expanded to higher dimension where dynamical effects are included. However, the required data and computational cost for this is too extensive for the valve design considered as basis for the analysis. Instead, for this specific problem, the derived analytical model is sufficient to describe the valve dynamics and reduces the computation time significantly.
topic Artificial Neural Networks
CFD
Digital Valves
Flow-induced Forces
Reduced Order Models
Lumped Parameter Models
url http://www.mic-journal.no/PDF/2019/MIC-2019-2-1.pdf
work_keys_str_mv AT nielscbender feasibilityofdeepneuralnetworksurrogatemodelsinfluiddynamics
AT torbenoleandersen feasibilityofdeepneuralnetworksurrogatemodelsinfluiddynamics
AT henrikcpedersen feasibilityofdeepneuralnetworksurrogatemodelsinfluiddynamics
_version_ 1716813098825809920