Propagation of boundary and geometrical uncertainties for the aeroacoustics analysis of a side mirror

Edmondo Minisci, Angela Scardigli, Federico Gallizio

Research output: Contribution to conferenceSpeech

2 Downloads (Pure)

Abstract

Uncertainty in the design and operation of engineering systems may arise from various sources. The uncertainties in physical properties of materials and inevitable randomness in boundary conditions and geometries, as well as physical models uncertainties, are a few examples of such uncertainties that can significantly restrict the reliability of deterministic analyses and designs. For a reliable analysis and design process based on computational fluid dynamics (CFD) simulations, including computational aeroacoustics (CAA), all sources of uncertainty must be considered in the analysis and design process. However, CAA analyses requires exceptionally fine 3D computational meshes, very small time-step, and (usually) high-dimensional stochastic space, bringing to very high, and up to now prohibitive, computational costs.
In the literature, various techniques have been proposed for uncertainty quantification (UQ). The Monte Carlo (MC) approach is widely used for UQ given its conceptual simplicity, but, unfortunately, the conventional MC methods converge slowly and often require a large number of samples to achieve reasonable accuracy and thus are impractical for problems with a large number of uncertainties, and/or very high computational costs. During the last decades, some other more efficient UQ approaches have been developed, with some of them being intrusive and others non-intrusive. The intrusive approaches involve some modifications of the implemented code, while non-intrusive methods consider the models as black-box and sample it through the use of meta-modelling techniques.
Aeroacoustics has received great attention in the last years, due to the ever stricter noise regulations, and increased computational capabilities. However, despite the stochastic nature of most aeroacoustics systems and models, non-deterministic investigations in regards to computational aeroacoustics are limited.
In this contribution, some non-intrusive approaches for probabilistic propagation of uncertainties are presented through the use a simple automotive test case, considering boundary and geometrical uncertainties for the aeroacoustics analysis of a side mirror. Obtained results are then used to detail some approaches giving statistical similitude between uncertain numerical performance and (synthetic) uncertain experimental data.
Original languageEnglish
Publication statusPublished - 18 Sep 2019
EventAPG Acoustic Symposium - Bochum, Germany
Duration: 18 Sep 201918 Sep 2019

Other

OtherAPG Acoustic Symposium
CountryGermany
CityBochum
Period18/09/1918/09/19

Fingerprint

Aeroacoustics
Mirrors
Computational aeroacoustics
Uncertainty
Systems engineering
Costs
Computational fluid dynamics
Monte Carlo methods
Physical properties
Boundary conditions

Keywords

  • uncertainties
  • analysis
  • design process
  • computational fluid dynamics
  • computational aeroacoustics
  • aeroacoustics

Cite this

Minisci, E., Scardigli, A., & Gallizio, F. (2019). Propagation of boundary and geometrical uncertainties for the aeroacoustics analysis of a side mirror. APG Acoustic Symposium, Bochum, Germany.
Minisci, Edmondo ; Scardigli, Angela ; Gallizio, Federico. / Propagation of boundary and geometrical uncertainties for the aeroacoustics analysis of a side mirror. APG Acoustic Symposium, Bochum, Germany.
@conference{1e67c645c46a46caba972f3c426b7a6e,
title = "Propagation of boundary and geometrical uncertainties for the aeroacoustics analysis of a side mirror",
abstract = "Uncertainty in the design and operation of engineering systems may arise from various sources. The uncertainties in physical properties of materials and inevitable randomness in boundary conditions and geometries, as well as physical models uncertainties, are a few examples of such uncertainties that can significantly restrict the reliability of deterministic analyses and designs. For a reliable analysis and design process based on computational fluid dynamics (CFD) simulations, including computational aeroacoustics (CAA), all sources of uncertainty must be considered in the analysis and design process. However, CAA analyses requires exceptionally fine 3D computational meshes, very small time-step, and (usually) high-dimensional stochastic space, bringing to very high, and up to now prohibitive, computational costs.In the literature, various techniques have been proposed for uncertainty quantification (UQ). The Monte Carlo (MC) approach is widely used for UQ given its conceptual simplicity, but, unfortunately, the conventional MC methods converge slowly and often require a large number of samples to achieve reasonable accuracy and thus are impractical for problems with a large number of uncertainties, and/or very high computational costs. During the last decades, some other more efficient UQ approaches have been developed, with some of them being intrusive and others non-intrusive. The intrusive approaches involve some modifications of the implemented code, while non-intrusive methods consider the models as black-box and sample it through the use of meta-modelling techniques.Aeroacoustics has received great attention in the last years, due to the ever stricter noise regulations, and increased computational capabilities. However, despite the stochastic nature of most aeroacoustics systems and models, non-deterministic investigations in regards to computational aeroacoustics are limited.In this contribution, some non-intrusive approaches for probabilistic propagation of uncertainties are presented through the use a simple automotive test case, considering boundary and geometrical uncertainties for the aeroacoustics analysis of a side mirror. Obtained results are then used to detail some approaches giving statistical similitude between uncertain numerical performance and (synthetic) uncertain experimental data.",
keywords = "uncertainties, analysis, design process, computational fluid dynamics, computational aeroacoustics, aeroacoustics",
author = "Edmondo Minisci and Angela Scardigli and Federico Gallizio",
year = "2019",
month = "9",
day = "18",
language = "English",
note = "APG Acoustic Symposium ; Conference date: 18-09-2019 Through 18-09-2019",

}

Minisci, E, Scardigli, A & Gallizio, F 2019, 'Propagation of boundary and geometrical uncertainties for the aeroacoustics analysis of a side mirror' APG Acoustic Symposium, Bochum, Germany, 18/09/19 - 18/09/19, .

Propagation of boundary and geometrical uncertainties for the aeroacoustics analysis of a side mirror. / Minisci, Edmondo; Scardigli, Angela; Gallizio, Federico.

2019. APG Acoustic Symposium, Bochum, Germany.

Research output: Contribution to conferenceSpeech

TY - CONF

T1 - Propagation of boundary and geometrical uncertainties for the aeroacoustics analysis of a side mirror

AU - Minisci, Edmondo

AU - Scardigli, Angela

AU - Gallizio, Federico

PY - 2019/9/18

Y1 - 2019/9/18

N2 - Uncertainty in the design and operation of engineering systems may arise from various sources. The uncertainties in physical properties of materials and inevitable randomness in boundary conditions and geometries, as well as physical models uncertainties, are a few examples of such uncertainties that can significantly restrict the reliability of deterministic analyses and designs. For a reliable analysis and design process based on computational fluid dynamics (CFD) simulations, including computational aeroacoustics (CAA), all sources of uncertainty must be considered in the analysis and design process. However, CAA analyses requires exceptionally fine 3D computational meshes, very small time-step, and (usually) high-dimensional stochastic space, bringing to very high, and up to now prohibitive, computational costs.In the literature, various techniques have been proposed for uncertainty quantification (UQ). The Monte Carlo (MC) approach is widely used for UQ given its conceptual simplicity, but, unfortunately, the conventional MC methods converge slowly and often require a large number of samples to achieve reasonable accuracy and thus are impractical for problems with a large number of uncertainties, and/or very high computational costs. During the last decades, some other more efficient UQ approaches have been developed, with some of them being intrusive and others non-intrusive. The intrusive approaches involve some modifications of the implemented code, while non-intrusive methods consider the models as black-box and sample it through the use of meta-modelling techniques.Aeroacoustics has received great attention in the last years, due to the ever stricter noise regulations, and increased computational capabilities. However, despite the stochastic nature of most aeroacoustics systems and models, non-deterministic investigations in regards to computational aeroacoustics are limited.In this contribution, some non-intrusive approaches for probabilistic propagation of uncertainties are presented through the use a simple automotive test case, considering boundary and geometrical uncertainties for the aeroacoustics analysis of a side mirror. Obtained results are then used to detail some approaches giving statistical similitude between uncertain numerical performance and (synthetic) uncertain experimental data.

AB - Uncertainty in the design and operation of engineering systems may arise from various sources. The uncertainties in physical properties of materials and inevitable randomness in boundary conditions and geometries, as well as physical models uncertainties, are a few examples of such uncertainties that can significantly restrict the reliability of deterministic analyses and designs. For a reliable analysis and design process based on computational fluid dynamics (CFD) simulations, including computational aeroacoustics (CAA), all sources of uncertainty must be considered in the analysis and design process. However, CAA analyses requires exceptionally fine 3D computational meshes, very small time-step, and (usually) high-dimensional stochastic space, bringing to very high, and up to now prohibitive, computational costs.In the literature, various techniques have been proposed for uncertainty quantification (UQ). The Monte Carlo (MC) approach is widely used for UQ given its conceptual simplicity, but, unfortunately, the conventional MC methods converge slowly and often require a large number of samples to achieve reasonable accuracy and thus are impractical for problems with a large number of uncertainties, and/or very high computational costs. During the last decades, some other more efficient UQ approaches have been developed, with some of them being intrusive and others non-intrusive. The intrusive approaches involve some modifications of the implemented code, while non-intrusive methods consider the models as black-box and sample it through the use of meta-modelling techniques.Aeroacoustics has received great attention in the last years, due to the ever stricter noise regulations, and increased computational capabilities. However, despite the stochastic nature of most aeroacoustics systems and models, non-deterministic investigations in regards to computational aeroacoustics are limited.In this contribution, some non-intrusive approaches for probabilistic propagation of uncertainties are presented through the use a simple automotive test case, considering boundary and geometrical uncertainties for the aeroacoustics analysis of a side mirror. Obtained results are then used to detail some approaches giving statistical similitude between uncertain numerical performance and (synthetic) uncertain experimental data.

KW - uncertainties

KW - analysis

KW - design process

KW - computational fluid dynamics

KW - computational aeroacoustics

KW - aeroacoustics

M3 - Speech

ER -

Minisci E, Scardigli A, Gallizio F. Propagation of boundary and geometrical uncertainties for the aeroacoustics analysis of a side mirror. 2019. APG Acoustic Symposium, Bochum, Germany.