Search Results

The structural excitation by turbulent flow pressure fluctuations continues to be an important part of vehicle NVH analysis and design. Much of the analytical work on turbulent flow has been limited to flow over smooth, flat surfaces. Measurements and CFD analyses show a significantly different pressure spectrum for flows over an obstruction. The coupling of the turbulent pressure excitation to the structural vibration and transmitted sound depends on the spatial matching between the source and response fields. This paper develops an analytical model for the spatial correlation of turbulent flow pressure fluctuations downstream of an obstruction. Such a model is useful in the early design stage of a vehicle and is compatible with SEA models of the fluid structure interaction.

Vehicle interior wind noise is typically managed through the overall exterior geometry of the vehicle, mirror shape and mounting location, sealing features and glass thickness and damping. Prior research has distinguished between contribution of fluctuating pressure due to air turbulence as compared to acoustic pressure to a passenger vehicles exterior at highway speeds. Because of the large difference in propagation speed between turbulent and acoustic pressure for on-road passenger vehicles, the structural response of the glass to turbulent versus acoustic pressure is not the same. The acoustic coincidence frequency of door glass is typically in the 2-3 kHz range. Turbulent coincidence frequency is much lower, and the effective transmission loss (TL) of the glass depends on the mix of turbulent and acoustic pressure on the exterior surface of the glass.

Wind noise is becoming a higher priority in the automotive industry. Several past studies investigated whether Statistical Energy Analysis (SEA) can be utilized to predict wind noise. Because wind noise analysis requires both radiation and transmission modeling in a wide frequency band, turbulent-structure-acoustic-coupled-SEA is being used. Past research investigated coupled-SEA’s benefit, but the model is usually simplified to enable easier consideration on the input side. However, the vehicle is composed of multiple interior parts and possible interior countermeasure consideration is needed. To enable this, at first, a more detailed coupled-SEA model is built from the acoustic-SEA model which has a larger number of degrees of freedom for the interior side. Then, the model is modified to account for sound radiation effects induced by turbulent and acoustic pressure.

The transmission of turbulent flow pressures through panels to the interior noise depends on the spatial matching of the pressure and vibration fields. Since the exterior pressure field on a moving vehicle includes both turbulent pressure and acoustic pressure, both need to be factored into a noise transmission loss calculation. However, these two exterior pressure fields have very different spatial patterns. This is further complicated when the exterior flow is separated from the surface due to an obstruction. This study uses wind tunnel and road tests to measure and model the wind noise transmission loss through the side glass of a vehicle. The results are seen to be very different from the traditional sound transmission loss curves for an acoustic pressure source.

The magnitude of the turbulent pressure spectrum in fluid flow over an obstruction is usually much larger than in attached flow over a smooth surface. External features on a vehicle, such as windshield wipers, side mirrors and pillars which cause flow separation, are a major source of wind noise. The modeling of the pressure spectrum in separated flow is important for designing quiet vehicles. In this study wind tunnel tests have been performed with different shaped obstructions to measure and correlate the surface pressure spectra with flow parameters such as the pressure coefficient and separation size. The model by Chase for attached turbulent boundary layer pressures is generalized to apply to separated flow conditions.

A four element wind noise transducer has been designed with surface mounted electret microphones in an array pattern which allows for the separate determination of the acoustic and turbulent pressures in wind noise. Three closely spaced transducers, defining an x-y coordinate system, are positioned to determine the velocity and direction of the turbulent flow. A fourth transducer is positioned at a greater distance such that the correlation of the turbulent flow will be diminished while the correlation of the acoustic pressure remains due to its longer wavelength. By averaging the cross-spectral densities of the pressure signals over time, the two contributors to wind noise can be differentiated. In addition, a wireless interface has been designed to minimize the flow disturbance of the transducer array.

From 1983 to 1995, Richard H. Lyon published several papers on Statistical Phase Analysis, showing that the average phase of the transfer functions in complex systems grows with frequency in proportion to the modal density of the system. In one dimensional systems this phase growth is the same as that of freely propagating waves. However, in two and three dimensional systems this phase growth is much larger than the corresponding freely propagating wave. Recent work has shown that these phase growth functions can be used as mode shape functions in discrete system models to obtain results consistent with Statistical Energy Analysis. This paper reviews these results and proposes naming the statistical mode shape functions in honor of Lyon.

The SEA model of wind noise requires the quantification of both the acoustic as well as the turbulent flow contributions to the exterior pressure. The acoustic pressure is difficult to measure because it is usually much lower in amplitude than the turbulent pressure. However, the coupling of the acoustic pressure to the surface vibration is usually much stronger than the turbulent pressure, especially in the acoustic coincidence frequency range. The coupling is determined by the spatial matching between the pressure and the vibration which can be described by the wavenumber spectra. This paper uses measured vibration modes of a vehicle window to determine the coupling to both acoustic and turbulent pressure fields and compares these to the results from an SEA model. The interior acoustic intensity radiating from the window during road tests is also used to validate the results.

A 1/4 scale model vehicle profile has been tested in a wind tunnel with speeds up to 360 km/h. In order to simulate the free field flow over the vehicle, the top surface of the wind tunnel is contoured. A CFD simulation of the free field flow at various speeds is used to identify the desired top streamline. Then the boundary layer growth on the top surface is calculated and the top contour is adjusted accordingly. Since this contour changes very little with flow speeds of interest, an average contour is used for a fixed top surface of the wind tunnel. Pressure drop measurements are used to verify the flow similarity to the CFD model. Wind noise measurements using surface mounted pressure transducer arrays are used to determine the acoustic loads on the vehicle surfaces.

Road tests on a pickup truck have been conducted to determine the acoustic loads on the back panel surfaces of the vehicle. Surface mounted pressure transducers arrays are used to measure both the turbulent flow pressures and the acoustic pressures. These measurements are used to determine the spatial excitation parameters used in an SEA model of the transmission loss through the back panel surfaces. Comparisons are made between tests on different road surfaces and at different speeds to identify the relative contributions of acoustic and wind noise.

Methods for conducting accelerated vibration fatigue testing of structures, such as MIL-STD-810G, allow for the non-linear scaling of the test time with the inverse of the rms vibration amplitude based on the slope of the material S-N curve obtained from cyclic fatigue tests. The Fatigue Damage Spectrum (FDS) is used as a method to allow for different level scalings at different frequencies in a broadband vibration environment using the relative responses of resonances in the structure. A recent development in industry has been to mix impulses with random excitations to increase the vibration peak levels (as measured by the kurtosis), thereby accelerating the fatigue even more than would occur with a Gaussian excitation. This paper presents results from a study to determine the conditions under which high kurtosis, impulsive excitations actually produce high kurtosis responses in structural resonances thus increasing the level of the FDS.

A new methodology to predict vehicle interior wind noise using CFD results has been developed. The CFD simulation replaces wind tunnel testing for providing flow field information around vehicle greenhouse. A loadcase model based on the CFD results is used to excite an SEA vehicle model. This new approach has been demonstrated on a production vehicle with success for the frequency range of 250-10K Hz. The CAE prediction of interior wind noise agrees within 0.2 sones from wind tunnel testing. The model has been used to evaluate wind noise performance with different door glass design parameters. A glass thickness change from 3.8 mm to 4.8 mm results in 1.1 sones improvement, which agrees well to 1.4 sones improvement from testing. Laminated glass with about 3 times higher damping results in 2.5 sones improvement. This methodology using CFD results can be used in the early stage of product development to impact designs.

A new formulation for dynamic analysis of the response of vibro-acoustic systems is developed. The method is based on a discrete element formulation similar in geometry to a finite element model. However, the Dynamic Element Analysis uses transcendental functions for the response interpolation functions. The phase of the functions converges at high frequencies to the Statistical Phase. At low frequencies the interpolation functions converge to the polynomials used in finite elements. Thus, the Dynamic Element Analysis covers a wide frequency range without requiring a refinement of the mesh, and it provides a deterministic response in the mid-frequency range before converging to a statistically correct response at high frequencies. Examples are shown of the response of structures and acoustic radiation.

A wind noise model of a vehicle has been developed using Statistical Energy Analysis (SEA) with measured turbulent pressure data as the source input. Empirical formulas are used to scale the input data for changes in flow and design parameters. Wind tunnel tests have been conducted on a standard and modified vehicle to validate the SEA model and the input scaling. The results show good correlation with both the exterior turbulent pressure levels and the interior sound pressure levels across the audio frequency range.

A series of tests have been performed on a production vehicle to determine the characteristics of the external turbulent flow field in wind tunnel and road conditions. Empirical formulas are developed to use the measured data as source levels for a Statistical Energy Analysis (SEA) model of the vehicle structural and acoustical responses. Exterior turbulent flow and acoustical subsystems are used to receive power from the source excitations. This allows for both the magnitudes and wavelengths of the exterior excitations to be taken into account - a necessary condition for consistently accurate results. Comparisons of measured and calculated interior sound levels show good correlation.