Project: Computer Aided Designed Development in Silencers (LAMPS3)
Reference: CSU 28/4/60
Last update: 27/11/2003 11:32:53
To increase the accuracy and reliability of mathematical/computational modelling of the acoustic effectiveness of intake and exhaust silencers of internal combustion engines.
Deficiencies in existing models will be identified and improvements made while new models will be developed for noise generating factors currently ignored. A programme of laboratory and on-engine tests will be conducted to validate and refine the new and improved models.
Watling Street, Nuneaton, Warwickshire, CV10 0TU
+44 (0)2476 355000
University of Loughborough
Dept of Mathematical Sciences, Loughborough, Leicestershire, LE11 3TU
University of Hull
Dept of Engineering Design & Manufacture, Cottingham Road, Hull, HU6 7RX
Arvin Exhausts, Research & Development
Hillock Lane, Warton, Preston, PR4 1TP
North Warren Road, Gainsborough, Lincolnshire, DN21 2TU
Ford Motor Co
Research and Engineering Centre, Laindon, Basildon, Essex, SS15 6EE
Tenneco-Walker (UK) Ltd
Liverpool Road, Burnley, Lancashire, BB12 6HJ
Cost to the Department: £102,000.00
Actual start date: 16 June 1999
Actual completion date: 17 October 2001
Computer Aided Design Development in Silencers (LAMPS3)
Author: Dr K S Peat
Publication date: 28/01/2003
Source: Dr K S Peat, Department of Aeronautical & Automotive Eng. Loughborough University
Summary of results
- Fundamental work was undertaken on the analysis offlow-generated noise in the first 6 months. Thefollowing 18 months, at Hull University, involvedexperimental work on an acoustic flow rig.Experiments were designed to characterise andmeasure the impedance of leakage through bafflesand through orifices with associated mean flow andthen performed the actual experiments on a variety ofleaky baffles. It was believed, or hoped, that steadyflow tests might not only characterise the leakage, butalso enable the low frequency impedance to beevaluated. Sufficient acoustic tests were conducted toprove that this was indeed the case. The quick andsimple steady flow tests alone were then employedover the majority of the wide variety of leakageproblems that were investigated. It has alsodetermined a way to infer the impedance at higherfrequencies from the low frequency data, such thatresults are now available over the frequency rangepertinent to plane-wave propagation in silencersystems. These results were presented in a paper atInterNoise 2001 at The Hague, Netherlands.
Work has also been completed on a set of transmissionloss tests on a range of silencers from the variousIndustrial Partners, all of which contained features forthe current software either gave poor predictions, or could not properly represent. Once again theseexperiments were done on a flow rig, under a varietyof flow conditions, using a simple acoustic source.
Work has since been undertaken on the prediction oftransmission loss for comparison with the experimentalresults completed at Hull, and the subsequentmodelling and development required to improve thecorrelation between theory and experiment. A moregeneralised form of Lamps software has beendeveloped which means that any form of silencer cannow be properly represented within the software. Notsurprisingly, this has increased the accuracy ofprediction in all cases for which the former softwarecould not properly represent the actual geometry. Thedownside of the new, generalised, code is that theanalysis time is considerably extended. More recentlywork has commenced on merging the best features ofthe former and the generalised Lamps codes.
The accuracy of the current Lamps model for theimpedance of louvered tubes, both with and withoutflow, was found to be poor following tests byTenneco-Walker. The current model has been refinedin the light of fresh understanding about thespecification of louvres. Tenneco-Walker undertookfurther experiments to determine whether thechanges had cured the problem. Currently, the bestmodel seems to result from treating a louvre as anequivalent circular hole.
In further developments it has been identified thatcertain geometrical features of silencers must bemodelled much more accurately than is generally thecase, to give good predictions. These are the plaintube sections between perforations in a tube and aflow return chamber. It has been found that theybehave like the neck of a Helmholtz resonator, andtherefore the precise length is vital, such that theprecise position of orifices in the perforate must be modelled. It was also discovered that non-planar wavescould cause significant errors in transmission loss atfrequencies less than half the cut-on frequency of thefirst non-planar mode. This work entailed multi-modefinite element modelling. The second mode wassufficient to eliminate this error, but there are no plansto alter the Lamps software from its current form ofplane-wave analysis, as inclusion of even one moremode would cause a fundamental shift in complexityof usage and time for analysis.
A simple model has been implemented within thesoftware for a new type of element, namely aparticulate filter (or CRT). Experimental tests byArvinMeritor and Eminox have given mixed results,and thus a more comprehensive model is now underdevelopment. A new model for the flow distributionwithin perforates, and hence the effects on perforateimpedance, is also under development. Also completedis the study on flow-generated noise, a paper on this isto be presented at ICSV9, Florida, USA in 2002 andseveral journal papers related to Lamps3 work havealso been published or are in press.
Eminox completed their set of on-engine tests ofsilencers, using a large diesel engine, in December2001. MIRA were unable to provide an engine test cellfor their re-scheduled test programme, and this workwas undertaken by ArvinMeritor instead andcompleted in March 2002. They used a medium petrolengine source. Perkins Engines completed somefurther test work, using a small diesel engine source,by the end of April 2002. Evaluation is currentlyunderway on these earlier test results.