Evaluation of a Large Eddy Simulation's Applicability to a Worst Case Fire Scenario

Abstract

The applicability of a large eddy simulation to a small scale turbulent flow problem is assessed by comparing modeled results to those recovered from a physical apparatus with the same geometry. The computational domain is that of a rearward facing step with a channel-width to step aspect ratio of2:1. The code utilized is LES-3d, and focus is placed on measuring the discrepancy between the recovered recirculation zone lengths when initial and boundary conditions of the virtual flow and duct are altered. It is found that the modeled results exceed the experimental by a factor of 2. These preliminary results point to the degree to which the user-specified parameters of upstream boundary conditions, inlet length, flow speed, flow profile, and computational domain resolution characterize and affect the simulated flow behavior. LES-3d's treatment of these crucial parameters is tested by performing additional experiments in a constant cross section straight duct with the same dimensions as the inlet to the previously mentioned rearward facing step. After looking into LES-3d's set of assumptions and means of incorporating the user's simulation preferences, a second set of simulations are executed with what are considered the optimal settings to guarantee the greatest degree of convergence between the experimental and modeled results. Findings indicate almost a 25% improvement in the recirculation zone measurements; however, other flow parameters such as the profile and boundary layer thickness are not maintained.This report serves as the computational portion of an ongoing study aimed at engineering a bench-scale apparatus to test the effectiveness of non-halogenated fire suppression agents in aircraft engines. The work is performed in cooperation with the National Institute of Standards and Technology, Gaithersburg, MD and the University of Maryland, College Park, MD.