Aeroacoustic Study of an Axial Ring Fan Using Lattice- Boltzmann Simulations

Transcription

1 Aeroacoustic Study of an Axial Ring Fan Using Lattice- Boltzmann Simulations Dominic Lallier-Daniels, Department of Mechanical Enginering, Université de Sherbrooke, Sherbrooke, Quebec, Canada Mélanie Piellard, Delphi Thermal Systems, Bascharage, Luxembourg Bruno Coutty, Delphi Thermal Systems, Bascharage, Luxembourg Stéphane Moreau, Department of Mechanical Enginering, Université de Sherbrooke, Sherbrooke, Quebec, Canada SYMPOSIA ON ROTATING MACHINERY Long Abstract Introduction In the context of turbomachinery design, more and more emphasis is being put on the aeroacoustic performance of novel systems in addition to the constraints regarding aerodynamic efficiency. Designers must therefore take steps to be able to correctly predict the acoustic performance of the product and be able to identify the noise mechanisms involved to eventually control them. Achieving this through experiments in the early stages of design, while desirable, is often out of the question as the expense of both money and time required to acquire detailed data outweighs the potential benefits. On the other hand, in recent years numerical simulations (Computational Fluid Dynamics, CFD) have proven useful in realizing early detailed analysis of both the aerodynamic and aeroacoustic performance of turbomachine systems, helping the designers to ascertain and optimize their design s performance early in the development process. The current paper presents the foundation of one such study using the numerical solver PowerFLOW to simulate a complete radiator cooling module geometry (Condenser, Radiator and Fan Module, CRFM) typical of one found in the automotive industry in engine cooling applications to investigate the aerodynamic and aeroacoustic performance of the system. This study is a continuation of the work done on similar geometries and shown in previous publications [1, 2]. The concurrent aerodynamic/aeroacoustic study is made possible by the use of lattice-boltzmann simulations, which are inherently unsteady and compressible, allowing for direct acoustic simulation [3, 4, 5]. An experimental acoustic study was led in a semi-anechoic environment in parallel with the simulations to provide data for the comparison and validation of numerical results. The current paper aims at presenting the results of the comparison of the direct acoustic simulation with the available experimental data before further expanding processing of the simulation results to identify and quantify the noise sources on an automotive engine cooling module and postulate as to the noise mechanisms via the use of the Ffowcs-Williams and Hawkings analogy to propagate the nearfield phenomena into the farfield. Simulated Geometry The geometry studied in this paper consists of a complete engine cooling module constituted of a fan, an electrical motor, a shroud as well as the various heat exchangers of the radiator it is mounted on. The simulated geometry is shown in Fig. 1. It is possible to see that the heat exchangers (usually located upstream of the fan) are not represented due to their complexity the grid refinement that would be required to properly mesh such a geometry. They were in fact reproduced in the simulation using porous media by setting the inertial and viscous resistance of the media in such a way that it reproduces the pressure loss incurred in the physical heat exchangers. A blockage plate located downstream of the module (outlined in Fig. 1) is also added to the simulation in order to reproduce the effect of the

2 proximity of the engine block when mounted on a car. It is possible to see that even the electric motor is represented in the model on the view from downstream. Figure 1. Illustration of the studied fan module In order to reproduce the semi-anechoic environment from the parallel experiment, the modeled CRFM was included in a large prismatic domain, as shown in Fig. 2 The ground is shown in solid grey. High viscosity is artificially imposed in the outer layers of the fluid domain to prevent the reflection of acoustic waves back into the simulation domain. Figure 2. Illustration of the semi-anechoic simulation domain. [2] Regarding the operational conditions of the CRFM, they are driven by the modeled geometry as well as the rotational speed of 2535 RPM imparted to the fan rotor, as the simulation is run at the free-flow condition (uniform pressure in the room). In the simulation, a series of 20 probes located 0.5 m upstream of the CRFM was included in the simulation setup to mimic the experimental microphone array used to collect acoustic data in the semi-anechoic chamber as shown in Fig.3. Of note also is the presence, as in the simulation, of the downstream blockage plate in the experimental setup on the right side of Fig. 3. It is to be noted that the possible interference of the microphone array in the experiment with fan operation is not represented in the simulations, as the probe locations are non-physical but rather represent data extraction locations. Simulation Results In the course of this study, the direct acoustic performance as recorded in the simulationhas been compared to the experimental data gathered in the semi-anechoic room on the abovementioned microphone array. An example from an arbitrary microphone is shown in Fig.??. The abscissa shows the frequency normalized by the rotational frequency of the fan, hereafter dubbed order (abbreviated by O ). The PSD level increments in the figure are 5 db.

3 Figure 3. Illustration of the experimental setup in the semi-anechoic room for the CRFM. Flow from left to right. [2] (a) (b) Figure 4. Acoustic pressure PSD for two separate microphone locations (a) Microphone 2 (b) Microphone 14 Overall, observing the spectra, a good correlation between the experiment and simulation is shown, with some variance between the different evaluation locations; for example, it can be observed that the correlation for microphone 14 is noticeably better than in the case of microphone 2. The spectra furthermore shows that the configuration produces consistent tonal noise around the 1st BPF (O7), with the level varying with the considered microphone (not shown here for brevity), indicating an uneven directivity pattern, which is to be expected. There is also a frequency range at and below the 2nd BPF (O12-14). This behavior is telling of the presence of rotating turbulent structures in the flow interacting with the fan blades, causing sub-harmonic humps. This specificity was covered in a previous publication for similar geometries [2, 6, 7, 8]. The 3rd BPF also emerges as a discrete tone, although again its emergence level differs from microphone to microphone. A discrete tone at 760 Hz (O18) is also seen to emerge from the spectra but is not always well represented by the simulation. It can also be seen that its amplitude varies from microphone to microphone in the experiment. This particular peak was subjected to additional investigation in the course of the study to ascertain its source and will be covered in more length in the final article. Given that the heat exchanger, modeled by porous media in the simulation, acts like a flow conditioner, the inflow should be relatively free of any large distortions from upstream. However, performing an analysis of the vortical structures in the volume surrounding the fan using visualizations

4 of the λ 2 factor [9] in Fig. 5 highlights the presence of a turbulent field rich in vortical structures. The structures identified are colored by their rotational speed normalized by the fan rotational speed. The image yields a plethora of pertinent information regarding the flowfield in close proximity to the fan. First of all, the appearance of an annulus of structures of varying shapes and sizes is seen close to the ring-shroud, originating from the tip gap and wrapping around it towards the tip of the blades. Larger, more radially penetrating structures are seen appearing at locations where the CRFM frame is closest to the fan along the circumference; this occurrence might be linked to the appearance of strong BPF harmonics as it creates static disturbances in the flowfield at regular intervals. These structure s influence on the velocity field can be seen in Fig. 6, which shows the 3 components of the instantaneous velocity field in a plane just upstream of the fan. Figure 5. λ 2 isosurfaces colored by normalized azimuthal velocity A string of coherent vortical structures originating near the trailing edge of the blade at 70% span is also seen appearing. A ring of vortical structures is also seen to form near the hub at the root of the blades. Generally, the structures identified are seen to be rotating at a fraction of the fan speed and thus would interact with the following blades and lead to an increased generation of tonal and broadband noise by the configuration. The filtered pressure fluctuations on the surface of the fan, stator and motor elements of the CRFM were also analyzed to try and gain an understanding of the possible sources of high tonal noise in the frequency bands identified on the spectra. Fig. 7 shows the PSD of pressure fluctuations for two frequency ranges corresponding to the 1st BPF and O18 tones. From these, it can be observed that the main areas of high pressure fluctuations are located on the fan blades and stator arms, with generally more elevated levels near the tip, which is clearer in the case of O18. In the case of the fan blades, there are specific hotspots that can be seen on the suction side of the blades near the leading edge at the hub and tip of the blades, as well as near the trailing edge near the fan ring and at approximately 70% of the span. These locations correspond well to the appearance of vortical structures in Fig. 5. However, these wall pressure fluctuation levels, while providing a certain indication as to the

5 (a) (b) (c) Figure 6. Instantaneous velocity field directly upstream of the fan (a) Azimuthal velocity normalized by fan rotational speed at the tip (b) Axial velocity (c) Radial velocity provenance of the noise, do not necessarily radiate into the farfield, as some (or most) of these sources could very well be evanescent or be drowned out by other sources in the farfield. The only way to ascertain the provenance of the noise is to employ a hybrid method in order to propagate these pressure fluctuations into the farfield, which was done via the use of the Ffowcs Williams and Hawkings analogy in order to identify and quantify the noise sources in presence. The sources were decomposed in a way as to allow for a precise identification and localisation of the dominant noise sources. The detailed analysis will be presented in the final version of the paper. References [1] M. Piellard, B. Coutty, V. Le Goff, Pérot F., and V. Vidal. Direct aeroacoustics simulation of automotive cooling fan system: An application study. In Aachen Acoustic Colloquium 2013, Aachen, Germany, [2] M. Piellard, B. Coutty, V. Le Goff, Pérot F., and V. Vidal. Direct aeroacoustics simulation of automotive cooling fan system: Effect of upstream geometry on broadband noise. In AIAA-CEAS 2014, Atlanta, Georgia, USA, [3] G. Brès, F. Pérot, and D. Freed. Properties of the lattice Boltzmann method for acoustics. In 15th AIAA Aeracoustics Conference, Miami, Florida, United States of America, 2009.

6 (a) Figure 7. PSD of wall pressure fluctuations on the fan, stator and electrical motor surfaces for (a) Hz range (O12-14) (b) Hz range (O18) (b) [4] S. Marié. Étude de la méthode Boltzmann sur Réseau pour les simulations en aéroacoustique. PhD thesis, Université Pierre et Marie Curie, Paris, France, [5] S. Marié, D. Ricot, and P. Sagaut. Comparison between lattice Boltzmann method and Navier- Stokes high order schemes for computational aeroacoustics. Journal of Computational Physics, 228: , [6] S. Moreau. Panel 2: Numerical methods for the prediction of fan aerodynamic and acoustic performances - where are we today? In Fan International Conference on Fan Noise, Technology and Numerical Methods. IMechE, [7] S. Moreau. Keynote 3: Numerical and analytical predictions of low-speed fan aeroacoustics. In Fan International Conference on Fan Noise, Technology and Numerical Methods. IMechE, [8] S Magne, S Moreau, and A. Berry. On the identification of a vortex. Journal of the Acoustical Society of America, 137: , [9] J. Jeong and F. Hussain. On the identification of a vortex. Journal of Fluid Mechanics, 285:69 94, 1995.