Development of a model for the prediction of combustion noise generated by automotive auxiliary heaters

I Motivation

The continuous development of diesel engines during the past two decades has led to highly increased engine efficiencies. Due to reduced excess heat less energy for the thermal conditioning of a vehicle's interior is available. Especially in cold climate conditions this makes an auxiliary heating unit desirable. These heating devices can also operate as an engine independent parking heater, avoiding problems connected to a cold engine start.

A basic configuration of a generic automotive auxiliary heating unit is shown in figure 1.

Heaters are usually located within the engine compartment of a vehicle. In general they have to be placed in areas with very limited installation space available. Therefore auxiliary heating devices are often operated under turbulent conditions in order to produce the needed thermal power density. For safety reasons fuel is usually fed separately to the combustion chamber. One major problem of these systems is the emission of combustion noise, caused by the interaction of chemistry and turbulent flow field.

II Objective

A modeling approach to predict the spectrum of fluctuating heat release within turbulent premixed swirl flames based on variable mean values has been developed and validated² at the Lehrstuhl für Thermodynamik. The model was also constituted to calculate combustion noise based on RANS simulation data³. The influence of turbulence intensity and timescales on the noise generation as well as their interaction with the combustion reaction are included in this model.

The objective of the current research project is the expansion of the existing model in order to capture noise generation in turbulent none premixed flames appropriately. The final model should be able to predict combustion noise from CFD-data independent of the flame type. The model will be included in a calculation package that provides noise prediction capabilities within the design process of general combustion systems or heaters. Emphasis is being placed on including the effects of relevant flow and chemistry specific parameters on the generation of turbulent combustion noise. Alongside with the acquired knowledge during the project the calculation method could provide general principles on how to optimize heater design with respect to a low level of noise generation.

Validation of the model will be achieved through comparison with experimental results from actual series production heating units.

III Theory

In order to calculate turbulent combustion noise one has to model the appropriate source term on the right hand side of Lighthill's wave equation. In case of a low Mach number flow the assumption that the strong monopole source of combustion plays the dominant role is justified¹ and the inhomogeneous wave equation takes the form:

If the pressure is assumed to follow harmonic time behavior with angular frequency &omega=2&pi f

the wave equation gives way to the Helmholtz equation and the time derivative of the heat release fluctuation takes the form of a heat release spectrum (given time harmonic behavior for the thermoacoustic source term as well). In order to determine the spectral heat release fluctuation based on local mean variables of a RANS simulation a model spectrum of a related variable (progress variable, temperature...) is required.

The oscillation of the local heat release is directly connected with the fluctuating turbulent flow field. Thus a model spectrum of turbulence quantities can be coupled with a combustion model (supplying the average heat release rate) to calculate the frequency-dependent heat release fluctuations. The so-determined fluctuations are linked with a local coherence function and appear as a monopole source in the Helmholtz-equation.

IV Simulation of turbulent combustion noise generation and propagation

In order to ensure cost and time efficient applicability the computation of turbulent combustion noise sources is based on CFD-RANS simulation data. The calculation of the resulting sound field is done separately in a second step. The sound field is calculated in the frequency domain either by solving the inhomogeneous Helmholtz-equation or a one-dimensional linear acoustic network equation system. The latter is able to provide results within minutes. The one dimensional treatment of noise propagation is possible because the wavelength of the radiated noise is sufficiently greater than the characteristic geometry parameters. The relevant acoustic geometry of the heater is split up into different acoustic elements. Figure 2 shows an example of a 1D burner model consisting of reflecting ends (R), ducts (D), area changes (AC), and the flame itself represented by its Flame Transfer Matrix (FTM).