A Computational Aeroacoustic Method for near and Far Field Vehicle Noise Predictions

This paper describes an approach to predict the acoustic nearand farfields with minimal computational requirements. The near field flow time dependent pressure history in the flow is predicted from computational fluid dynamics either using readily available finite difference codes or using in house finite element research code based on Gradient Adaptive Transfinite Elements. This pressure history information then used as input for a Kirchhoff formulation for acoustic far field predictions. A driven cavity problem numerical solution using commercial codes for the near field and in house developed code for far field predictions is presented which demonstrates the effectiveness of this method. The results are presented for cavity aspect ratios of 0.5 to 2.5 and flow Mach numbers of 0.26 to 0.672. INTRODUCTION Aerodynamic component of noise is one of the major contributors to overall transportation noise levels that affect the environment,. Accurate prediction of aerodynamic noise component with minimal computational requirements is desirable. This paper describes a computational methodology to predict the acoustic nearand farfields. The near field flow time dependent pressure history in the flow is generated using computational fluid dynamics either using readily available finite difference codes or using in house finite element code underdevelopment, based on Gradient Adaptive Transfinite Element (GATE) Family of elements, . It is possible to use computational fluid dynamics everywhere if the region of fluid concerned is not too large in terms of acoustic wavelength. GATE Finite elements, formulated by the first author, are different than the traditional ones in that they satisfy the field variable and its derivatives up to a required order along the boundaries. Confidence gained from successful applications of the GATE family to steep gradient problems and aeroelasticity, another form of fluid-structure interaction problem has initiated the methodology presented here for acoustic predictions. In this paper, the finite elements are used in terms of GATE-Domains only around the sources, although it is possible, with these special elements, to use finite elements everywhere by varying their dimensions and order. The far field noise levels are then predicted by 1 Professor and Co-Director, Graduate Research Assistant *Contact Author: N. Sarigul-Klijn, Associate Fellow, [email protected] (530)-752-0682 Copyright © 2001 The American institute of Aeronautics and Astronautics Inc. All rights reserved. using Kirchhoff integral formulation. A numerical scheme is developed for acoustic far field predictions which uses the near field solution as input to Kirchhoff formulation. This method has advantages in that nonlinear acoustic propagation is accurately modeled using computational fluid dynamics either using finite elements or readily available finite difference codes within the near field where the compressibility effects are important. THEORETICAL BACKGROUND Noise generated by turbulent flow in the presence of a solid boundary is the primary noise generation mechanism of interest in this study. Turbulence may be generated in the boundary layer of the vehicle and in regions of separation, or may already exist in the incoming flow. The problem of determining the sound transmitted to the far-field is divided into two steps. First, the sound source is determined. Navier-Stokes computations of the flow field are conducted with fine resolution of the boundary layer to determine the timedependant pressure near the solid boundaries. Second, the transmission from the known sound source is treated as an independent problem. GOVERNING EQUATIONS OF FLUID DYNAMICS Determining the pressure fluctuations near the surface of a vehicle requires accurate knowledge of the behavior of the boundary layer. Potential codes using boundary layer corrections are often inadequate to describe the time-dependent behavior in the boundary layer and separated regions. Full Navier-Stokes computations of the fluid flow are then required. This method allows the boundary layer to be resolved with great accuracy. The Navier-Stokes equations consist