Aeroacoustics Simulation of Electric Component with Compressive Large Eddy Simulation

2.5 Noise Prediction by Analytical or Numerical Models

Electronic devices with air cooling systems are becoming smaller and lighter in recent times. As the size of devices has been reduced, the air flow rate tends to be increased to cope with the increased heat density, accompanied with the relatively increased fluid noise. Reducing the noise from equipment is becoming important from the environmental aspect. With air-cooled equipment, fan noise is a dominant noise source caused by the highest velocity region. However, fan noise is influenced not only by the operating point but also by the positional relation between the fan and obstacles which cause resistance to the flow. Therefore, in order to reduce the noise from equipment, both the fan and obstacles should be taken into account. There are two methods to predict aeroacoustics noise, which are the Direct Method and Coupling Method, depending on the issues. The Direct Method uses Direct Numerical Simulation (DNS) and compressive Large Eddy Simulation (LES) to compute flow and acoustic field directory under Navier Stokes Equations. However, processing this method requires significant computer resources. The other method uses several turbulent models e.g. LES, Detached Eddy Simulation (DES), Reynolds Averaged Navier Stokes (RANS), and acoustic simulation to compute the incompressible flow and acoustic field separately.
This paper presents the Direct Method using the compressive LES approach to predict the aeroacoustics phenomena of the axial fan and obstacle interaction. The experimental set up was designed to mimic an electric equipment structure. Electric parts of the electric equipment, which cause resistance to the flow field, were simplified to a one block obstacle. The computational grid was made fine enough to solve the turbulent and aeroacoustics phenomena. The fan rotation is performed by Arbitrary Mesh Interface (AMI). Time step is defined to conduct the frequency analysis of the sound pressure level (SPL) and to achieve computational stability. In terms of the sound pressure spectrum, Blade Passing Frequency (BPF) was calculated as dominant frequencies. Predicted overall SPL was close to experimental data. Predicted SPL, which depends on the distance of the fan and obstacle, was also close to the experimental data. It was found that the simulation results will be useful in developing low noise products.