The noise emitted by axial fans plays an integral role in product design. When conventional design procedures are applied, the aeroacoustic properties are controlled via an extensive trial-and-error process. This involves building physical prototypes and performing acoustic measurements. In general, this procedure makes it difficult for a designer to gain an understanding of the functional relationship between the noise and geometrical parameters of the fan. Hence, it is difficult for a human designer to control the aeroacoustic properties of the fan. To reduce the complexity of this process, we propose an inverse design methodology driven by a genetic algorithm. It aims to find the fan geometry for a set of given objectives. These include, most notably, the sound pressure frequency spectrum, aerodynamic efficiency, and pressure head. Individual bands of the sound pressure frequency spectrum may be controlled implicitly as a function of certain geometric parameters of the fan. In keeping with inverse design theory, we represent the design of axial fans as a multi-objective multiparameter optimization problem. The individual geometric components of the fan (e.g., rotor blades, winglets, guide vanes, shroud, and diffusor) are represented by free-form surfaces. In particular, each blade of the fan is individually parameterized. Hence, the resulting fan is composed of geometrically different blades. This approach is useful when studying noise reduction. For the analysis of the flow field and associated objectives, we utilize a standard Reynolds averaged Navier–Stokes (RANS) solver. However, for the evaluation of the generated noise, a meshless lattice-Boltzmann solver is employed. The method is demonstrated for a small axial fan, for which tonal noise is reduced.

Contra-rotating fans have several advantages over single stage axial fans. If they are well designed, the exit flow field is almost irrotational. This helps to increase the aerodynamic efficiency by up to 16%, when compared to single stage fans. However, since the second stage interacts with the flow disturbances from the first stage, the associated noise generation is a disadvantage. This may be remedied by carefully tuning the design. The optimization of a contra-rotating fan involves a large set of design parameters. These include the geometrical parameters of the fan blades, the winglets, the guide vane as well as the hub diameter. We demonstrate an evolutionary algorithm which helps to automate the optimization process. It is controlled by two objective functions: (1) aerodynamic efficiency and (2) the emitted tonal noise. For the evaluation of the sound pressure, we implemented a new lattice-Boltzmann solver. Due to its algorithmic structure, this is ideally suited for massive parallelization. To leverage this potential, it is designed to run on general-purpose graphics processing units (GPGPUs). To further accelerate the optimization, it is supported by a meta-model based on a radial-basis function network. We demonstrate the method for a small contra-rotating fan. Our numerical results are compared with physical tests. The new algorithmic arrangement has shown to drastically cut development costs and time.

The noise emitted by axial fans plays an integral role in product design. When conventional design procedures are applied, aeroacoustic properties are controlled via an extensive trial-and-error process. This involves building physical prototypes and performing acoustic measurements. In general, this procedure makes it difficult for a designer to gain an understanding of the functional relationship between noise and geometrical parameters of the fan. Hence, it is difficult for a human designer to control the aeroacoustic properties of the fan. To reduce the complexity of this process, we propose an inverse design methodology driven by a genetic algorithm. It aims to find the fan geometry for a set of given objectives. These include, most notably, the sound pressure frequency spectrum, aerodynamic efficiency, pressure head and flow rate. Individual bands of the sound pressure frequency spectrum may be controlled implicitly as a function of certain geometric parameters of the fan. In keeping with inverse design theory, we represent the design of axial fans as a multi-objective, multi-parameter optimization problem. The individual geometric components of the fan (e.g., rotor blades, winglets, guide vanes, shroud and diffusor) are represented by free-form surfaces. In particular, each blade of the fan is parameterized individually. Hence, the resulting fan is composed of geometrically different blades. This approach is useful when studying noise reduction. For the analysis of the flow field and associated objectives, we utilize a standard RANS solver. However, for the evaluation of the generated noise, a meshless Lattice-Boltzmann solver is employed. The method is demonstrated for a small axial fan, for which tonal noise is reduced.

One of the main design decisions in the development of low-speed axial fans is the right choice of the blade loading versus rotational speed, since a target pressure rise could either be achieved with a slow spinning fan and high blade loading or a fast spinning fan with less flow turning in the blade passages. Both the blade loading and the fan speed have an influence on the fan performance and the fan acoustics, and there is a need to find the optimum choice in order to maximize efficiency while minimizing noise emissions. This paper addresses this problem by investigating five different fans with the same pressure rise but different rotational speeds in the design point (DP). In the first part of the numerical study, the fan design is described and steady-state Reynolds-averaged Navier–Stokes (RANS) simulations are conducted in order to identify the performance of the fans in the DP and in off-design conditions. The investigations show the existence of an optimum in rotational speed regarding fan efficiency and identify a flow separation on the hub causing a deflection of the outflow in radial direction as the main loss source for slow spinning fans with high blade loadings. Subsequently, large eddy simulations (LES) along with the acoustic analogy of Ffowcs Williams and Hawkings (FW–H) are performed in the DP to identify the main noise sources and to determine the far-field acoustics. The identification of the noise sources within the fans in the near-field is performed with the help of the power spectral density (PSD) of the pressure. In the far-field, the sound power level (SWL) is computed using different parts of the fan surface as FW–H sources. Both methods show the same trends regarding noise emissions and allow for a localization of the noise sources. The flow separation on the hub is one of the main noise sources along with the tip vortex with an increase in its strength toward lower rotational speeds and higher loading. Furthermore, a horseshoe vortex detaching from the rotor leading edge and impinging on the pressure side as well as the turbulent boundary layer on the suction side represent significant noise sources. In the present investigation, the maximum in efficiency coincides with the minimum in noise emissions.

One of the main design decisions in the development of low-speed axial fans is the right choice of the blade loading versus rotational speed, since a target pressure rise could either be achieved with a slow spinning fan and high blade loading or a fast spinning fan with less flow turning in the blade passages. Both the blade loading and the fan speed have an influence on the fan performance and the fan acoustics and there is a need to find the optimum choice in order to maximize efficiency while minimizing noise emissions. The present paper addresses this problem by investigating five different fans with the same pressure rise but different rotational speeds in the design point. In the first part of the numerical study, the fan design is described and steady-state Reynolds-averaged Navier-Stokes (RANS) simulations are conducted in order to identify the performance of the fans in the design point and in off-design conditions. The investigations show the existence of an optimum in rotational speed regarding fan efficiency and identify a flow separation on the hub causing a deflection of the outflow in radial direction as the main loss source for slow spinning fans with high blade loadings. Subsequently, Large Eddy Simulations (LES) along with the acoustic analogy of Ffowcs Williams and Hawkings (FW-H) are performed in the design point to identify the main noise sources and to determine the far-field acoustics. The identification of the noise sources within the fans in the near-field is performed with the help of the power spectral density of the pressure. In the far-field, the sound power level is computed using different parts of the fan surface as FW-H sources. Both methods show the same trends regarding noise emissions and allow for a localization of the noise sources. The flow separation on the hub is one of the main noise sources along with the tip vortex with an increase in its strength towards lower rotational speeds and higher loading. Furthermore, a horseshoe vortex detaching from the rotor leading edge and impinging on the pressure side as well as the turbulent boundary layer on the suction side represent significant noise sources. In the present investigation, the maximum in efficiency coincides with the minimum in noise emissions.