Multifidelity modeling of irradiated particle-laden turbulence subject to uncertainty
The study of thermal radiation interacting with particle-laden turbulence is of great importance in a wide range of scientific and engineering applications. The computational study of such systems is challenging as a result of the large number of thermo-fluid mechanisms governing the underlying phys...
| Autores: | , |
|---|---|
| Tipo de recurso: | artículo |
| Fecha de publicación: | 2020 |
| País: | España |
| Institución: | Universitat Politècnica de Catalunya (UPC) |
| Repositorio: | UPCommons. Portal del coneixement obert de la UPC |
| Idioma: | inglés |
| OAI Identifier: | oai:upcommons.upc.edu:2117/330571 |
| Acceso en línea: | https://hdl.handle.net/2117/330571 https://dx.doi.org/10.1615/Int.J.UncertaintyQuantification.2020032236 |
| Access Level: | acceso abierto |
| Palabra clave: | Turbulence Multifidelity Monte Carlo Particle-laden flow Thermal radiation Uncertainty quantification Turbulència Àrees temàtiques de la UPC::Enginyeria mecànica::Mecànica de fluids |
| Sumario: | The study of thermal radiation interacting with particle-laden turbulence is of great importance in a wide range of scientific and engineering applications. The computational study of such systems is challenging as a result of the large number of thermo-fluid mechanisms governing the underlying physics. To build confidence and improve the prediction accuracy of such simulations, the impact of uncertainties on the quantities of interest must be measured. This, however, requires a computational budget that is typically a large multiple of the cost of a single calculation, and thus may become infeasible for expensive simulation models featuring a large number of uncertain inputs and highly nonlinear behavior. In this regard, multifidelity methods have become increasingly popular in recent years as acceleration strategies to reduce the computational cost. These methods are based on a hierarchy of generalized numerical resolutions, or model fidelities, and attempt to leverage the correlation between high- and low-fidelity models to obtain a more accurate statistical estimator with a relatively small number of high-fidelity calculations. In this work, the performance of a collection of different multifidelity strategies and modeling approaches is assessed to propagate the uncertainties encountered in the simulation of irradiated particle-laden turbulence relevant to volumetric solar energy receivers. The results obtained indicate that multifidelity methods provide speedups on the order of 10-1000x with respect to straightforward Monte Carlo approaches, resulting in remarkable reductions in computational cost. |
|---|