Alessandro Mariotti

Uncertainty Quantification in Large-Eddy Simulations of the Flow Around a 5:1 Rectangular Cylinder

The Benchmark on the Aerodynamics of a Rectangular Cylinder (BARC) with chord-to-depth ratio equal to 5, sketched in Fig. 1a, was launched in 2008 with the aim of collecting a large number of flow realizations obtained both in wind tunnels and in numerical simulations.

Validation of Numerical Simulations of Thoracic Aorta Hemodynamics: Comparison with In Vivo Measurements and Stochastic Sensitivity Analysis

PurposeComputational fluid dynamics (CFD) and 4D-flow magnetic resonance imaging (MRI) are synergically used for the simulation and the analysis of the flow in a patient-specific geometry of a healthy thoracic aorta.MethodsCFD simulations are carried out through the open-source code SimVascular. The MRI data are used, first, to provide patient-specific boundary conditions. In particular, the experimentally acquired flow rate waveform is imposed at the inlet, while at the outlets the RCR parameters of the Windkessel model are tuned in order to match the experimentally measured fractions of flow rate exiting each domain outlet during an entire cardiac cycle. Then, the MRI data are used to validate the results of the hemodynamic simulations. As expected, with a rigid-wall model the computed flow rate waveforms at the outlets do not show the time lag respect to the inlet waveform conversely found in MRI data. We therefore evaluate the effect of wall compliance by using a linear elastic model with homogeneous and isotropic properties and changing the value of the Young’s modulus. A stochastic analysis based on the polynomial chaos approach is adopted, which allows continuous response surfaces to be obtained in the parameter space starting from a few deterministic simulations.ResultsThe flow rate waveform can be accurately reproduced by the compliant simulations in the ascending aorta; on the other hand, in the aortic arch and in the descending aorta, the experimental time delay can be matched with low values of the Young’s modulus, close to the average value estimated from experiments. However, by decreasing the Young’s modulus the underestimation of the peak flow rate becomes more significant. As for the velocity maps, we found a generally good qualitative agreement of simulations with MRI data. The main difference is that the simulations overestimate the extent of reverse flow regions or predict reverse flow when it is absent in the experimental data. Finally, a significant sensitivity to wall compliance of instantaneous shear stresses during large part of the cardiac cycle period is observed; the variability of the time-averaged wall shear stresses remains however very low.ConclusionsIn summary, a successful integration of hemodynamic simulations and of MRI data for a patient-specific simulation has been shown. The wall compliance seems to have a significant impact on the numerical predictions; a larger wall elasticity generally improves the agreement with experimental data.

Variational multiscale LES investigation of drag and near-wake flow of an axisymmetric blunt-based body.

The characterization of the relationship between the base pressure of bluff bodies and the geometrical and fluid dynamical parameters defining a certain configuration is a complex and still open problem, of interest for both fundamental research and practical applications.

Uncertainty quantification of a rectangular 5:1 cylinder

Uncertainty plays a significant role in the Benchmark on the Aerodynamics of a Rectangular Cylinder (BARC) with a chord-to-depth ratio of 5. In particular, besides modeling uncertainties and numerical errors, in numerical simulations it is difficult to exactly reproduce the experimental conditions due to uncertainties in the set up parameters, which sometimes cannot be exactly controlled or characterized. In this study, a computational sensitivity analysis and uncertainty quantification (UQ) study to the parametric uncertainties is carried out using probabilistic methods. The following uncertain set-up parameters are investigated in the subsequent uniform ranges: the angle of incidence α (-1o −1o), the free-stream longitudinal turbulence intensity Ix(0.001−0.03), and the free-stream turbulence length scale L (0.1D −5D), D being the cylinder depth. The Stochastic Collocation (SC) method is employed to perform the probabilistic uncertainty propagation of the three set-up parameters. This results in 25 URANS simulations based on the Smolyak sparse grid extension of the level-2 Clenshaw-Curtis quadrature points. The numerical error is estimated by comparing the results from computations on different mesh sizes. The findings are that the considered parameters have the largest impact on the time root mean square (t-rms) of the vertical force coefficient t-rms(cy) and of the surface pressure coefficient t-rms(Cp). These sensitivities are in agreement with previous BARC findings which indicate that those are the most dispersed parameters. There is a very good convergence of the UQ procedure and level 2 is sufficient. This additional UQ information can lead to new insights into the unresolved issues of the BARC flow test case.