Bayesian model-scenario averaged predictions of compressor cascade flows under uncertain turbulence models

Zordo-Banliat, M. de; Merle, Xavier; Dergham, G.; Cinnella, Paola

doi:10.1016/j.compfluid.2020.104473

Cited by 11 publications

(15 citation statements)

References 35 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Calibrated parameter values larger than nominal, for both c b1 and κ, were also reported for a similar study focused on SA turbulence model calibration for compressor cascades (de Zordo-Banliat et al, 2020). While the optimal parameter values presented herein are larger, de Zordo-Banliat et al indicated that numerical instabilities limited the use of larger parameter distribution intervals and as such restricting the posterior distribution of the calibration.…”

Section: Calibrated Resultssupporting

confidence: 68%

Calibration of Spalart-Allmaras model for simulation of corner flow separation in linear compressor cascade

Matsui¹,

Perez

Kelly

et al. 2021

J. Glob. Power Propuls. Soc.

View full text Add to dashboard Cite

This study focuses on the calibration of Spalart--Allmaras turbulence model parameters using the Bayesian inference approach to reproduce experimental measurements of corner flow separation in linear compressor cascade. The quantity of interest selected for the calibration process is the pitchwise distribution of Mach number in the wake of the linear compressor cascade. The model parameters are assumed to be random variables obeying uniform prior probability distributions. Sensitivity analysis is used to rank the importance and select the most influential turbulence model parameters for the calibration process. The sensitivity ranking indicates that two model parameters cb1 and kappa are the most influential random variables resulting in a two--parameter Bayesian calibration process. The likelihood distribution is specified in the form of the Gauss distribution to include the experimental uncertainty. The likelihood distribution is used together with prior distribution to compute posterior probabilities of selected model parameters. The polynomial chaos expansion is employed as a surrogate model to reduce the cost of posterior calculation. Numerical simulations with calibrated turbulence parameters show a significant increase in the accuracy of Mach number profile prediction for separated flows in linear compressor cascade. Numerical simulations also demonstrate that the calibrated set of model coefficients produce accurate predictions of the total pressure and Mach number profiles for the range of incidence angles that were not part of the calibration process.

show abstract

Section: Calibrated Resultssupporting

confidence: 68%

Calibration of Spalart-Allmaras model for simulation of corner flow separation in linear compressor cascade

Matsui¹,

Perez

Kelly

et al. 2021

J. Glob. Power Propuls. Soc.

View full text Add to dashboard Cite

show abstract

“…The approach used in this paper has been described in our previous work (de Zordo-Banliat, 2020; Merle and Cinnella, 2019). For the sake of self-containedness, we recall hereafter the principles of Bayesian model calibration and model mixures, with specific focus on BMSA.…”

Section: Bayesian Inference Frameworkmentioning

confidence: 99%

“… K=(σe2+ση2)I (with I a unit matrix) being a diagonal matrix representative of observation errors (moodelled as zero mean Gaussian noise with standard deviation σ e ) and model-inadequacy errors (modelled as Gaussian noise with standard deviation σ η ). The reader is referred to de Zordo-Banliat (2020) for more details about the statistical model.…”

Section: Bayesian Inference Frameworkmentioning

confidence: 99%

“…In total, 4,800 CFD calculations were necessary to build all the required surrogate models. The procedure used to assess surrogate modeling errors is described in de Zordo-Banliat (2020).…”

Section: Flow Configuration and Numerical Modelmentioning

confidence: 99%

“…In this paper, we build on previous work (de Zordo-Banliat, 2020) and we further develop a Bayesian model averaging (BMA) method (Draper, 1995; Hoeting, 1999) to perform stochastic predictions of turbomachinery flows from a predefined set of RANS models. More precisely, we consider an extended version of BMA referred to as Bayesian model-scenario averaging (BMSA) (Edeling et al , 2014b; Edeling, 2018; Rojas, 2010).…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Estimates of turbulence modeling uncertainties in NACA65 cascade flow predictions by Bayesian model-scenario averaging

Zordo-Banliat

Merle

Dergham

et al. 2022

HFF

View full text Add to dashboard Cite

Purpose The Reynolds-averaged Navier–Stokes (RANS) equations represent the computational workhorse for engineering design, despite their numerous flaws. Improving and quantifying the uncertainties associated with RANS models is particularly critical in view of the analysis and optimization of complex turbomachinery flows. Design/methodology/approach First, an efficient strategy is introduced for calibrating turbulence model coefficients from high-fidelity data. The results are highly sensitive to the flow configuration (called a calibration scenario) used to inform the coefficients. Second, the bias introduced by the choice of a specific turbulence model is reduced by constructing a mixture model by means of Bayesian model-scenario averaging (BMSA). The BMSA model makes predictions of flows not included in the calibration scenarios as a probability-weighted average of a set of competing turbulence models, each supplemented with multiple sets of closure coefficients inferred from alternative calibration scenarios. Findings Different choices for the scenario probabilities are assessed for the prediction of the NACA65 V103 cascade at off-design conditions. In all cases, BMSA improves the solution accuracy with respect to the baseline turbulence models, and the estimated uncertainty intervals encompass reasonably well the reference data. The BMSA results were found to be little sensitive to the user-defined scenario-weighting criterion, both in terms of average prediction and of estimated confidence intervals. Originality/value A delicate step in the BMSA is the selection of suitable scenario-weighting criteria, i.e. suitable prior probability mass functions (PMFs) for the calibration scenarios. The role of such PMFs is to assign higher probability to calibration scenarios more likely to provide an accurate estimate of model coefficients for the new flow. In this paper, three mixture models are constructed, based on alternative choices of the scenario probabilities. The authors then compare the capabilities of three different criteria.

show abstract

Calibration of Modified Spalart-Allmaras Model Parameters for Linear Compressor Cascade Corner Flow

et al. 2022

View full text Add to dashboard Cite

Bayesian model-scenario averaged predictions of compressor cascade flows under uncertain turbulence models

Cited by 11 publications

References 35 publications

Calibration of Spalart-Allmaras model for simulation of corner flow separation in linear compressor cascade

Calibration of Spalart-Allmaras model for simulation of corner flow separation in linear compressor cascade

Estimates of turbulence modeling uncertainties in NACA65 cascade flow predictions by Bayesian model-scenario averaging

Calibration of Modified Spalart-Allmaras Model Parameters for Linear Compressor Cascade Corner Flow

Contact Info

Product

Resources

About