Metamodels for Interphase Heat Transfer from Mesoscale Simulations of Shock–Cylinder Interactions

摘要

Macroscale computations of shocked particle-laden flows rely on closure laws to model the heat transfer between the fluid and particle phases. Typically, closure models are semiempirical and obtained for a limited range of parameters because experiments can be difficult and expensive to perform. This paper describes an approach to obtain closures for heat and momentum exchanges from ensembles of high-fidelity mesoscale computations of shock-cylinder interactions. The simulations are performed for flow over a single cylinder for a wide range of Reynolds Re-D and Mach numbers M-s. The results are used to construct a metamodel for the drag coefficient C-D and the Nusselt number Nu correlation using a modified Bayesian kriging method. To study the effects of the particle volume fraction phi, mesoscale computations are performed for cylinder clusters and the Nu and C-D are calculated. The metamodel shows that, although the Nusselt number Nu is primarily a function of the Re-D, the M-s and phi also significantly affect the interphase heat transfer. In particular, the Nusselt number Nu first decreases until M-s similar to 1.5-1.8 and increases for values of M-s 1.8. The results show that compressibility and viscous effects must be taken into account to provide accurate closure laws for interphase heat transfer in shocked particle-laden flows.