Twelfth International Congress on Sound and Vibration
NOISE FROM TURBO-CHARGED DIESEL ENGINE EXHAUST SYSTEMS
H. Bodén
1, A. Torregrosa
2, F. Ollivier
3, K. Peat
4, R. Fairbrother
5, B. Henriksson
6, P. Recouvreur
7, O. Poullard
8, R. Glav
9, J. Lavrentjev
101KTH Aeronautical and Vehicle Engineering, Marcus Wallenberg Laboratory for Sound and Vibration Research, SE-100 44, Stockholm, SWEDEN
2Institute CMT, UPV, SPAIN; 3LMP, Universite Pierre et Marie Curie, FRANCE
4Dept. of Aeronautical & Automotive Engineering, Loughborough University, U.K.
5AVL LIST GMBH, Austria; 6Volvo Powertrain Corporation, Sweden
7RENAULT, France; 8FAURECIA Abgastechnik GmbH, Germany
9Scania AB, Sweden; 10Dept. of Machinery, Tallinn Technical University, Estonia (e-mail address of lead author) hansbod@kth.se
Abstract
This paper summarises the main results of an EU-funded research project, ARTEMIS (G3RD-CT-2001-00511), on noise from turbo-charged Diesel engine exhaust systems. The project started in September 2001 and ended in August 2004 and was co-ordinated by KTH.
The project had 10 partners from 6 different European countries, 5 universities and 5 companies including some major truck and car manufacturers. The main objective was to develop new and improved computational tools for predicting noise from exhaust systems.
New models for describing the engine as an acoustic source were developed and experimentally tested. They include a linear time-varying source model and a non-linear frequency domain model. Linear time-invariant source data was also determined both from experiments and using 1-D gas-exchange simulations. New and improved models were developed for the turbo-group including non-linear time domain models and a linear time- varying model. New models were developed and experimentally tested for sound transmission through the Diesel particulate filter included in modern Diesel engine after- treatment devices. Improved models were developed for describing perforate mufflers with high mean flow velocities. Improved experimental techniques for determination of transmission properties of duct system components were developed. Models were developed and coded for sound reflection and radiation from tailpipe openings. Full experimental validation of the Munt theory for radiation from open pipes with flow was produced. In conclusion it can be said that the project was successful and gave many useful results.
INTRODUCTION
Traffic noise is the most important environmental noise source in Europe and in the rest of the world. It has been estimated that 25 % of the population in Europe is exposed to transportation noise with an equivalent sound level over 65 dB(A). At this sound level sleep is seriously disturbed and most people become annoyed. In all transport systems utilising internal combustion (IC) engines, exhaust and intake noise are major contributors to the overall noise pollution and need to be significantly reduced. Simulation programs for predicting the noise generation caused by unsteady flow in exhaust and inlet systems are very important for this purpose as well as for reducing the amount of experimental testing needed to develop a new product. The vast majority of current IC engines used in water-borne, rail and heavy vehicle applications are turbo-charged; it is furthermore projected that within the next 10 years 70% of automobiles will also be turbocharged. The pressing need to reduce air pollution as well as noise pollution will lead to increasing use of after treatment devices such as particulate traps. There was therefore a need for developing models for sound transmission and sound generation in these new elements. In addition the high flow velocities in turbocharged engine exhaust and inlet systems required that the models for the engine as an acoustic source, sound transmission and generation in mufflers and sound radiation and generation at the open ends, had to be improved or re-developed. The main objective of the project was to develop the models needed for computational tools to predict sound generation from turbocharged engine exhaust systems.
DEVELOPMENT OF ENGINE SOURCE MODELS
Two main engine experiments were made to support the development of improved engine source models. The first was on a six cylinder turbo-charged truck Diesel engine and the second on a turbo-charged automotive Diesel engine. These engines were also simulated using a one-dimensional gas-exchange CFD code (AVL BOOST). Acoustic source data was determined using the linear time-invariant source model and an indirect multi-load measurement technique [1, 2], both from the measurements and from numerical simulations.
In [3] a method was suggested to include non-linear effects when using so called direct methods for determining source impedance. This method has been modified for application with indirect or multi-load methods [4-6]. Figure 1 shows a comparison between measured and predicted pressure in the exhaust system. The predicted pressures have been calculated using source data from the linear time- invariant source model and the non-linear source model. It can be seen that the non- linear model gives only marginal improvements over the linear time-invariant model.
This is because the linear time-invariant model already gave sufficiently good results for this engine.
Figure 2 shows a comparison between simulated and predicted sound pressure
(AVL BOOST) simulations. It can be seen that acceptable results can be obtained from source data calculated from numerical simulations. This technique is of interest since source data measurements are expensive and time consuming. Another advantage is that acoustic source data can be calculated at an earlier stage in the engine development process, even before a prototype is produced.
Figure 1 - Comparison between measured and predicted sound pressure level: measurements
– full line; predicted linear time-invariant model – dashed line; predicted non-linear
model – dashed-dotted line.
Figure 2 - Comparison between simulated and predicted sound pressure level when the source
data was obtained from 1-D CFD simulations:
simulated – full line and stars, predicted – dashed line and squares.
A linear time-dependent model of a general multi-cylinder engine source was also developed [7]. The theoretical model was extended to consider a multi-cylinder engine and a model of the truck engine used within the Artemis project. Predictions have been compared to experimental results. Comparisons between source impedance extracted from the linear time-varying source model using different techniques and experimental results were made. There was a reasonably good agreement especially for the imaginary part, shown in Figure 3.
-3 -2 -1 0 1 2 3
0 100 200 300 400 500 600 700 800 900
Frequency (Hz)
Norm. Source Impedance
Average Doctored Average Experimental SVD
Figure 3 - Comparison
between imaginary
part of normalised
source impedance calculated using the linear time-
varying source model
and experimental
data.
DEVELOPMENT OF TURBO-GROUP MODELS
New turbo compressor models were developed during the project [8, 9]. A turbine model has been improved by taking into account a more realistic geometry of the stator [10, 11]. Both models have been validated with two different turbo groups, an automotive group and a truck group, with quite good results. Figure 4 shows the turbo-compressor model where two volumes represent the space available in the device, and they are connected by means of a boundary condition which is fed from the steady compressor operating chart. Then, under the assumption of adiabatic flow, and imposing conservation of mass and energy, one can compute the influence of the turbo-compressor on the flow.
Figure 4 - Scheme of the turbo-compressor model.
Making use of the operating chart obtained in a turbo-group test facility, a truck engine was modelled, and the results obtained compared with the compressor characterisation measurements performed on that engine. In Figure 5, this comparison is shown in terms of the pressure upstream and downstream of the compressor, both in the time and the frequency domain. It can be observed that, while the results upstream of the compressor are quite acceptable, there are some significant differences in the results obtained downstream; this applies mostly to the amplitude of the pressure fluctuation. It is remarkable that the differences are observed precisely at the downstream side, since this should be essentially governed by the pressure pulses generated by the engine intake, which are supposed to be correctly modelled.
Figure 5 - Truck turbo-
compressor model:
pressure comparison.
DEVELOPMENT OF AFTER-TREAMENT DEVICE MODELS
New time and frequency domain models for describing sound transmission through Diesel particulate traps were developed [12-16]. Figure 6 shows a sketch of the investigated after treatment device including both particulate filter and catalytic converter and Figure 7 shows a comparison between measured and predicted transmission loss.
Figure 6 - Sketch of the
investigated after-treatment
device.
Figure 7 - Transmission loss for
the after-treatment device shown in
Figure 6.
DEVELOPMENT OF MUFFLER MODELS
The objective was the development of low frequency acoustic models for mufflers
with high flow velocities. Studies were made both on basic muffler elements such as
Helmholtz resonators, orifices and perforated elements where the effect of tangential
or through flow was expected to be large [17-18]. A commercial muffler was also
studied. Figure 8 shows an example of a comparison between measured and predicted
transmission loss for the commercial muffler.
Figure 8 - Comparison between measured
and predicted transmission loss
for the studied muffler, M = 0.15.
DEVELOPMENT OF TAILPIPE OPENING MODELS
A numerical boundary element solution to the problem of radiation from a tailpipe with mean flow effects and vortex shedding was developed [19]. The purpose of seeking a numerical solution is that it is readily extendable to complex situations, such as slant-cut tailpipes, unlike the existing analytical solutions. The main problem, as noted from the outset, was to determine the strength of the shed vorticity. A stable method to achieve this was eventually found and the results for the end correction from a baffled pipe showed reasonable agreement with existing measured values, see Figure 9.
The so-called Munt model [20] for calculation of the reflection coefficient of a jet pipe with subsonic mean flow was coded. Figure 10 shows results comparing the end correction obtained using the Munt model compared to experimental results.
These accurate experimental results were obtained using a new technique for full plane wave characterization of acoustic data in flow ducts [21] are actually the first complete experimental verification of Munt´s theory. An experimental study of different tailpipe configurations was also made [22].
VALIDATION MEASUREMENTS
To investigate if the new models developed would be applicable for an egine and an
exhaust system not used in model evelopments a study was made on another truck
exhaust system. Figure 11 shows as an example of the resuls a comparison between
measured and predicted noise reduction loss for a muffler. The best agreement is
End Correction
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
0 0.5 1 1.5 2
Strouhal Number, S δ/a
theory M=.001 theory, M=.02 Expt, M=.01 Expt, M=.02
Figure 9 - Predicted and measured end correction of a baffled pipe outlet with flow.
Figure 10 - Comparison between measurements and theory for the end
correction of a free pipe opening, measurements – lines, predictions –
symbols.
Noise Reduction, 25% load, 1000 rpm
-20 -10 0 10 20 30 40 50 60 70
0 100 200 300 400 500
Frequency [Hz]
NR [dB]
Lamps Expt
Figure 11 - Comparison between measured and predicted sound reduction.
SUMMARY
This paper has presented an overview of some of the most important results from a recently concluded EU-funded research project, ARTEMIS (G3RD-CT-2001- 00511). Progress has been made in development of engine source models, turbo- group models, models for after-treatment devices including new models for Diesel particulate filters, models for mufflers with constrictions and perforates with high flow velocities and model for exhaust tailpipe openings.
M=0.2
M=0.05
M=0.15 M=0.1
REFERENCES
[1] H. Bodén and M. Åbom, Modelling of fluid machines as sources of sound in duct and pipe systems, Acta Acustica 3, 549-560, 1995.
[2] H. Bodén, Characterisation of fluid-borne sound sources and structure-borne sound sources, Proceedings of ICSV9 ,2002. Orlando, Florida, USA.
[3] S-H. Jang and J.G. Ih, A measurement method for the nonlinear time-variant source characteristics of intake and exhaust systems in fluid machines, Proceedings of ICSV10, 2003, Stockholm, Sweden.
[4] H. Rämmal and H. Bodén, Modified multi-load method for non-linear source characterization.
Proceedings of ICSV11, 2004, St. Petersburg, Russia
[5] H. Bodén, R. Fairbrother and M. Tonsa, On extraction of IC-engine linear acoustic source data from non-linear simulations, Proceedings of ICSV1, 2004, St Petersburg, Russia.
[6] H. Bodén, Acoustic source characterisation for Diesel engine exhaust systems, Proceedings of THIESEL 2004, Valencia, Spain.
[7] J.F. Dowling and K.S. Peat, Source impedance model, Proceedings of ICSV11, St Petersburg, Russia. 2004.
[8] J.M. Desantes, A. J. Torregrosa, A. Broatch and T. Fernández, Study of the acoustical behaviour of turbo-compressors for supercharged Diesel engines, submitted to Applied Acoustics.
[9] K.S. Peat, A.J. Torregrosa, A. Broatch and T. Fernández, An investigation into the passive acoustic effect of an automotive turbocharger, submitted to Journal of Sound and Vibration, 2004.
[10] K.S. Peat, M. Elsari and S. Dequand, A linear acoustic model of the passive effect of the turbine of an automotive turbocharger, Proceedings of ICSV10, 2003, Stockholm, Sweden.
[11] A.J. Torregrosa, A. Broatch, H. Climent and T. Fernández , Time-domain modelling of turbo- compressors in DI Diesel engines, submitted to the Proceedings of the Institution of Mechanical Engineers Part D - Journal of Automobile Engineering.
[12] S. Allam and M. Åbom, On Acoustic Modelling and Testing of Diesel particulate Filters, Proceedings of INTER-NOISE 2002 –August 19-21, Dearborn, Michigan,USA, paper No. 250.
[13] S. Allam and M. Åbom, Acoustic Modelling of an After-Treatment Device (ATD), Proceedings of EURONOISE 2003, May 19-21, Naples.
[14] S. Allam and M. Åbom, On Transmission of Sound in Diesel Particulate Filters (DPF:s), Proceedings of ICSV10, 2003, Stockholm, Sweden.
[15] Robert Fairbrother and M. Tonsa, Acoustic simulation of after treatment devices using linear and non linear methods, Proceedings of ICSV10, Stockholm, Sweden.
[16]S. Allam and M. Åbom, Acoustic Modelling and Testing of Diesel Particulate Filters, accepted for publication in the Journal of Sound and Vibration, 2004.
[17] K.S. Peat, The influence of plate thickness on the acoustic impedance of an aperture in grazing flow, Proceedings of ICSV9, Florida, USA. 2002.
[18] K.S. Peat, The acoustic impedance of a circular orifice in grazing mean flow: Comparison with theory, J.Acoust.Soc.Am., 2003, 114, pp.3076-3086.
[19] K.S. Peat, Effect Of The Jet Shear Layer On Tailpipe Radiation, Proceedings of INTERNOISE 2003, Cheju, Korea.
[20] R. M. Munt, Acoustic Transmission Properties of A Jet Pipe with Subsonic Jet Flow: I. The Cold Jet Reflection Coefficient, 1990 Journal of Sound and Vibration 142, 413– 436.
[21] S. Allam Acoustic Modelling and Testing of Advanced Exhaust System Components for Automotive Engines. Paper 6, 2004 Ph.D Thesis KTH.
[22] F. Ducret and M. Åbom, Aeroacoustic behaviour of different exhaust pipe geometries, Proceedings of ICSV11, 2004, St. Petersburg, Russia.