Computational Aero-acoustic Analysis of Low-noise Airframe Devices with the Aid of Stochastic Method

Executive Summary:

In summary, the CALAS project had been successful and has given new insights and justification concerning aero-acoustic performance of high-lift and landing-gear airframe configurations with a variety of proposed low-noise concepts and, furthermore, gained experiences, as well as contributed to the data base, in CAA analysis and assessment of these airframe configurations aiming at noise reduction and using the stochastic method for flow-generated noise-source modelling.

In addition to the project management and coordination in WP 1, the project consisted of two technical work packages, WP 2 and WP 3, addressing respectively the aero-acoustic performance of a double-flap wing section with flap side-edge (FSE) and of a main landing gear (MLG) configuration, as well as of conceptual low-noise measures added on these configurations aiming at noise reduction. In recognition of the fact that, for its favoured computational efficiency, RANS modelling remains the mainstay in industrial aeronautic applications, the CAA analysis in evaluating the FSE and the MLG configurations has been carried out on the basis of RANS computations with the aid of a stochastic method for modelling the noise sources.

With the commercial package, VNoise, the use of the SNGR method was thus first introduced and demonstrated in CAA analysis for the baseline FSE configuration based on RANS computations. It is shown that the stochastic method was effective to model the flow-generated noise source of broadband type for predicting far-field noise radiation. The same method incorporated in the VNoise package had formed the major tool for CAA analysis in the CALAS project.

To provide a reference, an additional acoustic analysis was carried out using different acoustic analogies based on a turbulence-resolving simulation with a hybrid RANS-LES model. The analysis was conducted for the baseline-WT FSE configuration. It was shown that the flap side-edge has induced flow separation generating extensive aerodynamic fluctuations and vortex motions. The region around wing-flap junction and the flap side edge was the most significant area of generating noise. The acoustic performance of the baseline-WT configuration was then investigated using, in addition to the SNGR method, acoustic analogy methods, including the Kirchhoff, the FW-H, and the Curle method. It was shown that the Kirchhoff and FW-H methods predict higher noise levels comparing with the SNGR method in the frequency band of 40 Hz and 800 Hz. This may suggest that, for the baseline FSE configuration, the anisotropic and inhomogeneous flow dynamics in association to noise generation has happened at relatively low frequencies. In this frequency band, nonetheless, the noise level computed with the Curle method is more comparable with the SNGR method. The SPL predicted with the SNGR method hardly decays with respect to increasing frequencies, with magnitudes consistent and comparable with those obtained with acoustic analogy methods at high frequencies. The OASPL predicted using the SNGR method was comparable to the level obtained with the Kirchhoff towards the upstream side, on the downstream side with the FW-H method. It suggested that the SNGR method was able to provide a reasonable estimation for the broad-band noise, and thus for relevant assessment of airframe configuration with the major concern of broadband noise.

As compared to the CAA analysis using acoustic analogy based on a turbulence-resolving simulation, the RANS-based CAA analysis using the SNGR method was more computationally efficient and able to produce comparable trend in the SPL spectrum. The turbulent flow taken as the noise sources were synthetically constructed using the stochastic method. The SNGR method cannot thus provide an accurate estimation of the tonal noise. However, the CALAS work had demonstrated that the SNGR method is effective for predicting broadband noise in industrial applications.

The acoustic assessment for the full-scale fence-on FSE configuration was undertaken in comparison with the baseline configuration (with no add-on fence). It was found that the add-on fence attached on the flap side-edge had enabled effective reduction of noise emission at frequencies between 55 Hz and 5000 Hz. The add-on fence had led to an overall noise reduction of 5-7 dB in the far-field and up to 6-8 dB in the near-field. Furthermore, it was shown that the noise radiates with a directivity of a nearly monopole pattern. It suggested that the noise source in the wake had been represented by monopole type as a whole with synthetic turbulence based on a RANS solution, which failed to distinguish quadruple and dipole sources that would have otherwise been reflected in the directivity of noise radiation obtained with turbulence-resolving simulations. The downstream noise was more intensive than in the upstream direction, due to the noise source embedded in the trailing wake of the wing section.

In the acoustic analysis of the Main Landing Gear (MLG) configurations in WP 3, the configurations were all of full scale. The door of the bay was at the “open” position to investigate the scattering effect of the bay-door on the noise radiation and to mimic the landing or take-off situations.

The CAA analysis of all four MLG configurations had shown that relatively high noise levels have occurred in the range of frequencies between 300Hz to 3000Hz. Moreover, for all the configurations assessed, the directivity presents in general a peak in OASPLA at a polar angle of about 100o in relation to the after-LG wake with significant generation of turbulent kinetic energy (and thus intensive modelled noise sources). The noise-scattering effect of the open-door was significant, mainly at the frequencies higher than 1600 Hz at the flight condition considered. For the baseline configuration (Configuration 1), Configuration 2, Configuration 3 and Configuration 4, respectively, the RANS-based CAA analysis had predicted an averaged A-weighted OASPL (OASPLA) of 148.355 dBA, 149.239 dBA, 153.582 dBA and 146.538 dBA. This suggested that Configuration 2 (with a fairing attached in front of the LG strut) and Configuration 3 (with a shallow LG-bay cavity) were not effective to introduce noise reduction, as compared to the baseline configuration. However, the acoustic liner installed in the bay rear wall (Configuration 4) had effectively led to noise level. The averaged OASPLA with Configuration 4 was about 1.82 dBA lower than with the baseline configuration. It was further revealed that, with Configuration 4, the most efficient noise reduction presents in the downward arc from 90 to 115 degree, which was in the range of interest in industry designs. It would be interesting to further investigate the proposed concepts introduced in Configurations 2 and 3 to explore their possible impacts on suppressing noise sources of tonal type, since the noise-source modelling using the SNGR method had been based on a steady RANS solution and focusing on noise sources of broadband type.

In close collaboration among the project partners (FOI and Chalmers) and with the topic manager (CIRA), the CALAS project had been successfully completed and had achieved its objectives in line with the project work plan and with the requirement of GRA program. The project consortium had extensively analysed and evaluated the double-flap wing-section configurations and the main landing gear configurations with a variety of proposed concepts aiming at airframe noise reduction, specified by the topic manager in collaboration with CALAS and with involved JTI GRA members. By the completion of this final summary report, along with all the submitted project deliverables, the project had reported all the technical work conducted.