[Paper Review] Electronics Cooling Fan Noise Prediction

Antoine Dozolme, Hossam Metwally+1 more|ArXiv.org|Sep 12, 2007

Aerodynamics and Fluid Dynamics Research2 references3 citations

TL;DR

This study uses steady-state finite volume CFD simulations with the MRF model in FLUENT to predict broadband noise and cooling performance of an electronics cooling fan across three rotational speeds (2,000–2,700 rpm). The method enables noise-performance trade-off analysis without requiring fan curve inputs, showing qualitative agreement with catalog data for fan design optimization.

ABSTRACT

Using the finite volume CFD software FLUENT, one fan was studied at a given flow rate (1.5m3/min) for three different operational rotating speeds : 2,000, 2,350 and 2,700 rpm. The turbulent air flow analysis predicts the acoustic behavior of the fan. The best fan operating window, i.e. the one giving the best ratio between noise emissions and cooling performance, can then be determined. The broadband noise acoustic model is used. As the computation is steady state, a simple Multiple Reference Frame model (MRF, also known as stationary rotor approach) is used to represent the fan. This approach is able to capture the effects of the flow non-uniformity at the fan inlet together with their impact on the fan performance. Furthermore, it is not requiring a fan curve as an input to the model. When compared to the available catalog data the simulation results show promising qualitative agreement that may be used for fan design and selection purposes.

Motivation & Objective

• To predict fan noise and cooling performance under varying rotational speeds for electronics thermal management.
• To evaluate the feasibility of using steady-state CFD with MRF modeling for noise prediction without fan curve inputs.
• To identify the optimal operating window balancing noise emissions and cooling efficiency.
• To validate simulation results against catalog data for practical fan design and selection.

Proposed method

• Finite volume CFD software FLUENT was used to simulate turbulent airflow through a fan at 1.5 m³/min flow rate.
• The Multiple Reference Frame (MRF) model was applied to simulate rotating fan effects in a steady-state framework.
• A broadband noise acoustic model was employed to predict noise emissions from the simulated flow field.
• Three rotational speeds—2,000, 2,350, and 2,700 rpm—were analyzed to assess performance and noise trade-offs.
• The method captures inlet flow non-uniformities and their impact on fan performance without requiring empirical fan curves.

Experimental results

Research questions

• RQ1 How accurately can steady-state CFD with MRF modeling predict fan noise levels across varying rotational speeds?
• RQ2 What is the relationship between fan rotational speed, noise emission, and cooling performance in electronics cooling applications?
• RQ3 Can the MRF approach effectively model fan aerodynamics and noise without fan curve input data?
• RQ4 How does the simulated noise-performance trade-off compare to published catalog data?

Key findings

• The simulation results showed promising qualitative agreement with catalog data for noise and performance metrics.
• The MRF model successfully captured flow non-uniformities at the fan inlet and their effects on performance.
• The method enabled identification of an optimal operating window balancing noise and cooling efficiency.
• The absence of required fan curve inputs simplifies the modeling process for fan design applications.