Validation of numerical prediction method of BPF noise for industrial centrifugal fans

摘要

Blade passing frequency (BPF) noise is the dominating component of the flow induced noise of centrifugal fans. The numerical methods for BPF noise prediction, based on the computational aeroacoustics (CAA), have been published for decades. However, there are a couple of challenges for accurately predicting noise for industrial centrifugal fans. The first arises from the fact that the free field hypothesis, adopted in the numerical model, has not yet been carefully studied. The second challenge stems from the current criteria for which the prediction results are compared to the measurement data. Because the test conditions do not always satisfy the requirements of the numerical model, inaccurate predictions occasionally resulted. Therefore, since the prediction results may deviate largely from the test data, the applicability of these methods is severely limited. In this article, more realistic criteria are used to numerically predict the BPF noises of three different types of industrial centrifugal fans, and the results are compared with the experimental data. Using this improved model, the predicted BPF noise only deviates from the experimental data by an average of 0.4 dB. These results also indicate that the free field hypothesis is still appropriate in real test environments. This may be due to the exclusion of the shading effects of the volute being compensated by the noise reflection from the surroundings. The reflections always exists in noise measurements in industrial production sites. Moreover, the contributions of sound sources on the blade surface and on the volute tongue are also analyzed in detail.