Colloquia

Samenvatting

Broadband flow-induced sound is an unwanted side effect of HVAC systems. Experiments are used to ensure that the noise levels remain under the 30 dB Dutch-law limit. However, this is onerous. Therefore,  there is a want to use computational aeroacoustics, to investigate flow-induced sound systematically in the design process. Here, a computational aeroacoustics strategy is explored. Its ultimate aim was to determine in-duct broadband noise within the commercially available STAR-CCM+. It combines an LES simulation with the Perturbed Convective Wave Model that calculates sound sources and acoustic propagation. Instead of a complicated HVAC geometry a rectangular duct obstructed by a rectangular orifice was investigated.

From the literature calculation and measurement data are available. The model setup proposed in the literature was used as a benchmark for our model. Two differences in our approach, with respect to what was found in the literature, are noted: (1) A constant inlet velocity was applied rather than an interpolated field. (2) The acoustic solver within STAR-CCM+ was used. Comparing our results to this benchmark, we found that our sound pressure level spectrum in the range 200 Hz to 3500 Hz has significant deviations. This is true for both the simulation and measurement data from the literature.

As a means of checking, if with our simulation strategy the flow field was adequately resolved: the discharge and pressure loss coefficient for a cylindrically symmetric configuration was determined. For this tube (ISO standards 5167 and 15377), the discharge and pressure loss coefficients were determined to judge the quality of the solution. We found that using the meshing strategy described in the literature: the discharge and pressure loss coefficient cannot be resolved properly.

Thus, an improved meshing strategy was applied. Thus used a mesh requirement, which states that at least 80% of the turbulent kinetic energy should be resolved within the grid. The discharge and pressure loss coefficients were calculated for the above-described cylindrically symmetric system. We found that both coefficients can now be determined within the uncertainty range of the ISO standards. We concluded that the flow field is solved correctly. However, when it comes to the noise-level predictions there are still significant differences when compared to measurement data found in the literature.

We conjecture that these discrepancies might be due to: (1) Ill-posed non-reflective boundary conditions within STAR-CCM+. (2) The non-matching inflow conditions. (3) The unquantified quality of the measurement data.