scholarly journals Prediction of Combustion Noise in a Model Combustor Using a Network Model and a LNSE Approach

Author(s):  
Wolfram C. Ullrich ◽  
Yasser Mahmoudi ◽  
Kilian Lackhove ◽  
André Fischer ◽  
Christoph Hirsch ◽  
...  

The reduction of pollution and noise emissions of modern aero engines represents a key concept to meet the requirements of the future air traffic. This requires an improvement in the understanding of combustion noise and its sources, as well as the development of accurate predictive tools. This is the major goal of the current study where the low-order thermo-acoustic network (LOTAN) solver and a hybrid computational fluid dynamics/computational aeroacoustics approach are applied on a generic premixed and pressurized combustor to evaluate their capabilities for combustion noise predictions. LOTAN solves the linearized Euler equations (LEE) whereas the hybrid approach consists of Reynolds-averaged Navier–Stokes (RANS) mean flow and frequency-domain simulations based on linearized Navier–Stokes equations (LNSE). Both solvers are fed in turn by three different combustion noise source terms which are obtained from the application of a statistical noise model on the RANS simulations and a post-processing of incompressible and compressible large eddy simulations (LES). In this way, the influence of the source model and acoustic solver is identified. The numerical results are compared with experimental data. In general, good agreement with the experiment is found for both the LOTAN and LNSE solvers. The LES source models deliver better results than the statistical noise model with respect to the amplitude and shape of the heat release spectrum. Beyond this, it is demonstrated that the phase relation of the source term does not affect the noise spectrum. Finally, a second simulation based on the inhomogeneous Helmholtz equation indicates the minor importance of the aerodynamic mean flow on the broadband noise spectrum.

Author(s):  
Wolfram C. Ullrich ◽  
Christoph Hirsch ◽  
Thomas Sattelmayer ◽  
Yasser Mahmoudi ◽  
Ann P. Dowling ◽  
...  

The reduction of pollution and noise emissions of modern aero engines represents a key concept to meet the requirements of the future air traffic. This requires an improvement in the understanding of combustion noise and its sources, as well as the development of accurate predictive tools. This is the major goal of the current study where the LOTAN network solver and a hybrid CFD/CAA approach are applied on a generic pre-mixed and pressurized combustor to evaluate their capabilities for combustion noise predictions. LOTAN solves the linearized Euler equations (LEE) whereas the hybrid approach consists of RANS mean flow and frequency-domain simulations based on linearized Navier-Stokes equations (LNSE). Both solvers are fed in turn by three different combustion noise source terms which are obtained from the application of a statistical noise model on the RANS simulations and a postprocessing of an incompressible and compressible LES. In this way the influence of the source model and acoustic solver is identified. The numerical results are compared with experimental data. In general good agreement with the experiment is found for both the LOTAN and LNSE solvers. The LES source models deliver better results than the statistical noise model with respect to the amplitude and shape of the heat release spectrum. Beyond this it is demonstrated that the phase relation of the source term does not affect the noise spectrum. Finally, a second simulation based on the inhomogeneous Helmholtz equation indicates the minor importance of the aerodynamic mean flow on the broadband noise spectrum.


2011 ◽  
Vol 188 ◽  
pp. 398-403 ◽  
Author(s):  
Chun Hui Ji ◽  
Zhan Qiang Liu

Many workers all over the world suffer significant hearing loss as well as psychological and physical stress as a result of exposure to high levels of aeroacoustic noise. Dipole sources are the major noise sources in aeroacoustic noise generation in rotating face milling cutters. A noise model based on the Ffowcs Williams-Hawkings Equation is used to predict aeroacoustic noise; the noise predicted was 2.5dB less than the experimental observations. Flow field of cutter surface was numerically simulated by the resolution of the Navier-Stokes equations (CFD) and five zones on cutter surface were founded to be the important noise source. The broadband noise spreads over a broad range of frequencies and contributes significantly to overall noise, but the discrete noise at the rotational frequency is usually higher and more detrimental to the body.


Author(s):  
Douglas Darling ◽  
Krishnan Radhakrishnan ◽  
Ayo Oyediran

Noise generated in gas turbine combustors can exist in several forms — broadband noise, sharp resonant peaks, and regular or intermittent non-linear pulsing. In the present study, dynamic pressure measurements were made in several JP-5-fueled combustor configurations, at various mean pressures and temperatures. The fluctuating pressure was measured at mean pressures from 6 to 14 atm and inlet temperatures from 550 K to 850 K. The goal of the present work was to study the effect of changes in mean flow conditions on combustor noise: both broadband noise and sharp tones were considered. In general, the shape of the broadband noise spectrum was consistent from one configuration to another. The shape of the spectrum was influenced by the acoustic filtering of the combustion zone. This filtering ensured the basic consistency of the spectra. In general, the trends in broadband noise observed at low mean pressures were also seen at high mean pressures; that is, the total sound level decreased with both increasing equivalence ratio and increasing inlet temperature. The combustor configurations without a central pilot experienced higher broadband noise levels and were more susceptible to narrow peak resonances than configurations with a central pilot. The sharp peaks were more sensitive to the mean flow than was the broadband noise, and the effects were not always the same. In some situations, increasing the equivalence ratio made the sharp peaks grow, while at other conditions, increasing the equivalence ratio made the sharp peaks shrink. Thus, it was difficult to predict when resonances would occur, however, they were reproducible. Noise was also observed near lean blow out. As with other types of noise, lean blow out noise was affected by the combustion chamber acoustics, which apparently maintains the fluctuations at a uniform frequency. However, the actual conditions when this type of noise was experienced appeared to simply follow the lean blow out limit, as it varied with mean temperature and pressure.


2014 ◽  
Vol 16 (5) ◽  
pp. 901-918 ◽  

<div> <p>Three-dimensional calculations were performed to simulate the flow around a cylindrical vegetation element using the Scale Adaptive Simulation (SAS) model; commonly, this is the first step of the modeling of the flow through multiple vegetation elements. SAS solves the Reynolds Averaged Navier-Stokes equations in stable flow regions, while in regions with unstable flow it goes unsteady producing a resolved turbulent spectrum after reducing eddy viscosity according to the locally resolved vortex size represented by the von Karman length scale. A finite volume numerical code was used for the spatial discretisation of the rectangular computational domain with stream-wise, cross-flow and vertical dimensions equal to 30D, 11D and 1D, respectively, which was resolved with unstructured grids. Calculations were compared with experiments and Large Eddy Simulations (LES). Predicted overall flow parameters and mean flow velocities exhibited a very satisfactory agreement with experiments and LES, while the agreement of predicted turbulent stresses was satisfactory. Calculations showed that SAS is an efficient and relatively fast turbulence modeling approach, especially in relevant practical problems, in which the very high accuracy that can be achieved by LES at the expense of large computational times is not required.</p> </div> <p>&nbsp;</p>


2017 ◽  
Vol 826 ◽  
pp. 396-420 ◽  
Author(s):  
M. Bouyges ◽  
F. Chedevergne ◽  
G. Casalis ◽  
J. Majdalani

This work introduces a similarity solution to the problem of a viscous, incompressible and rotational fluid in a right-cylindrical chamber with uniformly porous walls and a non-circular cross-section. The attendant idealization may be used to model the non-reactive internal flow field of a solid rocket motor with a star-shaped grain configuration. By mapping the radial domain to a circular pipe flow, the Navier–Stokes equations are converted to a fourth-order differential equation that is reminiscent of Berman’s classic expression. Then assuming a small radial deviation from a fixed chamber radius, asymptotic expansions of the three-component velocity and pressure fields are systematically pursued to the second order in the radial deviation amplitude. This enables us to derive a set of ordinary differential relations that can be readily solved for the mean flow variables. In the process of characterizing the ensuing flow motion, the axial, radial and tangential velocities are compared and shown to agree favourably with the simulation results of a finite-volume Navier–Stokes solver at different cross-flow Reynolds numbers, deviation amplitudes and circular wavenumbers.


Author(s):  
Djordje Romanic ◽  
Horia Hangan

Analytical and semi-empirical models are inexpensive to run and can complement experimental and numerical simulations for risk analysis-related applications. Some models are developed by employing simplifying assumptions in the Navier-Stokes equations and searching for exact, but many times inviscid solutions occasionally complemented by boundary layer equations to take surface effects into account. Other use simple superposition of generic, canonical flows for which the individual solutions are known. These solutions are then ensembled together by empirical or semi-empirical fitting procedures. Few models address turbulent or fluctuating flow fields, and all models have a series of constants that are fitted against experiments or numerical simulations. This chapter presents the main models used to provide primarily mean flow solutions for tornadoes and downbursts. The models are organized based on the adopted solution techniques, with an emphasis on their assumptions and validity.


2017 ◽  
Vol 12 (1) ◽  
pp. 43-49
Author(s):  
Egor Palkin ◽  
Rustam Mullyadzhanov

Flows between two closely spaced bounding surfaces are frequently appear in engineering applications and natural flows. In current paper the flow over a cylinder in a narrow rectangular duct was investigated by numerical computations of Navier-Stokes equations using Large eddy simulations (LES) at ReD = 3 750 based on cylinder diameter and the bulk velocity at inflow boundary. The influence of the bounding walls was demonstrated by comparing mean flow streamlines with the flow over an infinite cylinder at close Reynolds numbers. A comparison between the time-averaged velocity field in front and past the cylinder with experimental from the literature data showed good agreement although the characteristic horseshoe vortex structures are highly sensitive to Reynolds number and turbulence level at inflow boundary. Most energetic modes in recirculating region were revealed by spectral analysis. These low-frequency modulations were characterized by the pair of dominating vortices which are expected to have high influence on the heat transfer in near wake of the cylinder.


2014 ◽  
Vol 136 (6) ◽  
Author(s):  
B. A. Younis ◽  
A. Abrishamchi

The paper reports on the prediction of the turbulent flow field around a three-dimensional, surface mounted, square-sectioned cylinder at Reynolds numbers in the range 104–105. The effects of turbulence are accounted for in two different ways: by performing large-eddy simulations (LES) with a Smagorinsky model for the subgrid-scale motions and by solving the unsteady form of the Reynolds-averaged Navier–Stokes equations (URANS) together with a turbulence model to determine the resulting Reynolds stresses. The turbulence model used is a two-equation, eddy-viscosity closure that incorporates a term designed to account for the interactions between the organized mean-flow periodicity and the random turbulent motions. Comparisons with experimental data show that the two approaches yield results that are generally comparable and in good accord with the experimental data. The main conclusion of this work is that the URANS approach, which is considerably less demanding in terms of computer resources than LES, can reliably be used for the prediction of unsteady separated flows provided that the effects of organized mean-flow unsteadiness on the turbulence are properly accounted for in the turbulence model.


Author(s):  
S M Fraser ◽  
Y Zhang

Three-dimensional turbulent flow through the impeller passage of a model mixed-flow pump has been simulated by solving the Navier-Stokes equations with an improved κ-ɛ model. The standard κ-ɛ model was found to be unsatisfactory for solving the off-design impeller flow and a converged solution could not be obtained at 49 per cent design flowrate. After careful analysis, it was decided to modify the standard κ-ɛ model by including the extra rates of strain due to the acceleration of impeller rotation and geometrical curvature and removing the mathematical ill-posedness between the mean flow turbulence modelling and the logarithmic wall function.


Author(s):  
Tanya S. Stanko ◽  
Derek B. Ingham ◽  
Michael Fairweather ◽  
Mohamed Pourkashanian

Numerical solutions of a turbulent jet flow are used to provide velocity information throughout a simple cold turbulent propane jet at a Reynolds number of 68,000. Predictions provided by the Reynolds-averaged Navier-Stokes simulations, based on a Reynolds stress turbulence model, are compared with experimental data available in the literature. The effect of the modelled inlet boundary conditions on the predicted flow field is described, and the discrepancy between the simulation results and experiment measurements is found to be less than the corresponding variations due to uncertainness in the experimental boundary conditions. In addition, these solutions are used as the basis for noise predictions for the jet based on Lighthill’s theory using the Goldstein broadband noise source formalization that postulates axisymmetric turbulence superposed on the mean flow. The latter model provides an aeroacoustic tool that is reasonable in identifying components or surfaces that generate significant amounts of noise, thereby providing opportunities for early design changes to aircraft and gas turbine components.


Sign in / Sign up

Export Citation Format

Share Document