With the ever-increasing cooling requirements across industries, from cooling engines to electronic equipment, fan noise has become one of the important factors that drives design decisions in product development. The need for quieter machines has not only been a performance issue but also forms part of the customer experience and branding.

Due to government regulations, industrial equipment needs to adhere to a maximum noise level to protect workers and avoid excessive noise in residential areas. Generally, the noise generated from rotating fans is the dominant contributor of equipment operating noise overall. Not just exclusive to the heavy equipment industry, other industries that are more recently turning to electric power sources also need to have a comprehensive understanding of fan acoustic noise. In the automotive industry for example, the increased electrification of vehicles requires a robust fan noise analysis to help prevent excessive operating noise. While electric motors are far quieter than traditional combustion engines, they still require a lot of cooling. Ultimately, the challenge is that air needs to be moved quietly and efficiently.  

Physical testing in an anechoic chamber is expensive from both a time and cost perspective, potentially delaying product development timelines in an ever competitive and price sensitive environment. Often excessive fan noise is only discovered in a prototype testing scenario and benchmark test results can give misleading information as the acoustic character of a rotating component will change in an installed condition. 

Taking a Simulation-Based Approach
Simulation-based analysis can provide both accurate and quick predictions of aero-acoustic noise levels at different locations over frequency ranges. It can also give additional insights into noise propagation through the flow medium. Unlike physical testing, simulation-based tools can provide both overall noise levels and detailed insights into noise sources and propagation paths. Altair CFD™ brings a solution to the market for installed fan acoustics by directly calculating the convective field and other key points of interest. This virtual method can significantly reduce the risk of failure leading to development cost savings and mitigation of timeline overshoot. Altair’s Lattice Boltzmann (LBM) solver is a perfect fit for understanding fan noise thanks to features such as high-fidelity external aerodynamics, automated meshing, and efficient results analysis. 

Click the video below to see how Altair CFD can also be used in the agricultural industry. 

Comparison of Simulation and Test Data from Real Examples
In the example below, the aero-acoustic performance of a fan was simulated with a similar rotating speed and diameter as a fan commonly used in commercial trucks. A microphone was placed seven meters away from the noise source in a semi-anechoic environment and 0.75 seconds of acoustic data was analyzed. It is imperative that the sample frequency is correct and collected at the right stage in the simulation. When collecting this data, one of the biggest challenges is the complexity of the machine as there are many places from which acoustic waves can propagate and reflect. Directly simulating the convective phenomena of this process is therefore the most accurate process to fully understand the aero-acoustic performance of the fan. As shown in the graph below, the simulation results (shown in blue) track well with the test data from the physical test (shown in grey). 

The same process can be applied to a heat ventilation air conditioner (HVAC) blower. This is a completely different type of fan that is typically used in automotive heating and ventilation systems. In this case, the fan system consists of around 40 blades with a thickness of around 0.5mm. Microphones were placed around the fan assembly as shown and a simulation is set up to replicate the test as accurately as possible. From the analysis shown in the graph, 30 seconds of test data is represented in black and compared to one second of simulation data, shown in red. Because of the discrepancy in the number of averaging blocks due to the significant differences in signal sampling, we expect the simulation data to contain more noise than the test data. This simulation was set up to minimize run time as there was a specific interest in frequencies below 2,500Hz. The blade pass frequency of 750Hz was captured along with the overall sound pressure level.  

Both examples act as validation that Altair CFD can provide users with the tools to accurately analyze the acoustic performance of a machine, making it possible to engineer products in a virtual environment to reduce risk of failure for acoustics. To learn more, watch this presentation.

Harnessing Graphics Processing Unit (GPU) Computing Power

Successfully predicting fan noise involves accurately simulating the noise produced by turbulent flow structures created by the presence of a rotating object. As a result, running a CFD simulation of this nature is a computationally expensive process, sometimes taking several days to complete using CPU computing power. 

However, thanks to Altair’s GPU solutions, it’s possible to solve extremely quickly by relying on superior processing power, equivalent to thousands of CPU cores. Utilizing only eight NVIDIA A100 GPUs, Altair CFD was able to solve typical industrial size fan noise problems in just 10 to 15 hours. As a native GPU-based code, Altair's Lattice Boltzman solver is a perfect fit for the parallel architecture of NVIDIA GPUs, making it possible to perform highly transient aero-acoustic simulations on a single server.

Furthermore, Altair® Unlimited® offers companies the ability to provision GPU hardware on the cloud with services such as Google Cloud Platform. All of Altair’s CFD solvers are optimized for use on GPU clusters for faster and more efficient simulations regardless of the scale and complexity. Thanks to Altair’s turnkey on-premises and cloud-based appliances, users can access this processing power remotely, enabling companies to dynamically scale their infrastructure.

Utilizing GPUs to accelerate numerical simulation delivers significant increases in speed and throughput. This means more opportunities to explore and fine-tune designs, make decisions faster based on more accurate results, and consequently, considerably reduce time-to-market. Harnessing the computational capabilities of the GPU is one of the cornerstones of Altair’s mission to empower designers. To learn about how Altair harnesses the power of GPU, read this article written by our chief technical officer, Uwe Schramm.

Cooling fans play a critical role in product design. Getting it right can save manufacturers from costly redesign iterations and protect a product’s heat-sensitive components over the course of its lifecycle. It can prevent a printed circuit board (PCB) from overheating, extending the life of electronic devices, or keep the engine and hydraulic fluid cool in heavy-duty construction equipment operating under extreme conditions. 

Only a fully resolved CFD-acoustics model with true rotating geometry can capture the transient aero-acoustic phenomenon of rotating fans accurately. With Altair’s aero-acoustic solution offering, Engineers can increase throughput while reducing product development time, hardware, and energy cost. 

For more on aero-acoustic simulation with Altair CFD, watch this webinar.