INTERNATIONAL JOURNAL OF MATHEMATICS AND COMPUTERS IN SIMULATION Volume 10, 2016

Transcription

1 Transmission Loss Assessment and Optimization of an Intake System for Atomotive Application Siano D. 1, Aiello R., and D Agostino D. Abstract The acostic project of internal combstion engine s intake systems is a very important task aimed to ensre that a tradeoff between noise emissions and perceived sond qality is achieved. In general, the most sed parameter for characterizing the performance of an acostic filter is represented by the so called Transmission Loss which may be nmerically evalated thanks to a three-dimensional FEM code. In this paper, a FEM model of an intake system for a commercial spark ignition engine has been realized in order to simlate its acostic performance in terms of Transmission Loss, withot considering strctral participation as well as the presence of mean flow inside the system. After a model validation, thanks to previos experimental analysis, an acostic optimization procedre of the stdied system has been carried ot by means of geometry modifications. Keywords Transmission Loss assessment, Finite element method, Internal combstion engine, Breathing noise. INTRODUCTION NOWADAYS, the modern concept of client perception qality of a vehicle, which represents the prior target behind the prodction strategies of internal combstion engines, is qite changed with respect to a dozen of years ago. In particlar, the concept of engine perceived sond represents a critical aspects that car manfactrers are focsing on, in order to be more competitive. In fact, if the engine noise is something that may be appreciated by cstomers, on the other hand it strongly redces the global comfort within the cabin. Sch a new project aspect is becoming more and more important as it is testified by the noise emission standards. It may be stated that the acostic project of a vehicle mst satisfy a trade-off between acostic comfort and perceived sond. In a vehicle varios noise sorces are involved which belong to two main categories: strctral and airborne noise. Among these, the most prominent noise sorce, with particlar regard to the low vehicle speed, is represented by the gas-dynamic noise emitted by both the intake and the D. Siano is with the National Research Concil (Istitto Motori- CNR), Italy (corresponding athor to provide phone: ; d.siano@ im.cnr.it). exhast systems. Frthermore, sch systems directly affect the power otpt of internal combstion engines becase they strongly inflence the volmetric efficiency. However, the acostic design of the intake system of modern internal combstion engine represents a more critical aspect with respect to an exhast one[7]. In fact, whilst for this latter it is relatively easy to place a so called proper mffler (or dbkiller) at the end of the acostic transmission path, when dealing with intake systems the gas-dynamic noise de to the pressre waves created at each IVO crank angle positions may only be attenated by the throttle. That s why modern engines eqipped with VVA distribtion system sonds stronger. It is well known that for a system acostic performance characterization the most sed parameters are Insertion Loss, Level Difference and TL (Transmission Loss)[1]. Among these, the Insertion Loss is rarely sed since for its calclation the knowledge of the sorce impedance is reqired[]. At the same time, the Level Difference represents jst a simple measre of the sond pressre level drop which does not take into accont any bondary conditions for the system s otlet. The Transmission Loss is instead the most adopted acostic performance characterization parameter since it is a measre of the sond power level drop a system ensres, by assming an anechoic termination at the otlet section[]. In the presented stdy, a FEM model of an intake system for a commercial spark ignition engine has been created in order to first evalate its Transmission Loss. After a model validation procedre by means of experimental data, available from previos stdies, an acostic optimization of the original device has been carried ot. All the simlations have been performed by sing the commercial 3D code Virtal Performance Soltion, powered by ESI Grop. I. FINITE ELEMENT METHOD IN ACOUSTIC Today, nmerical simlation plays the most important role in stdying and solving real life problems. Among the nmerical acostics method, the Finite Element Method (FEM) offers an advantageos combination of modeling flexibility, comptation efficiency and reslts accracy. In the classical FEM formlation, the soltion of a physical problem relies on the se of a so-called weak formlation of ISSN:

2 the governing eqation (which represents the so-called strong form of the physical phenomenon) throgh two main steps: 1) The Partial Differential Eqation (PDE) is mltiplied by a test fnction and integrated over the whole comptational domain ) An integration by parts makes possible to transfer some of the partial derivate to sch test fnction In sond propagation problems, the pressre distrbance at a generic point Q(x,y,z) and at the specified time t, is given by the so called wave eqation whose homogeneos version is: sorces and it does not take into accont the strctral participation, as considered in this early stage. II. TRANSMISSION LOSS ANALYSIS The Transmission Loss represents a measre of the sond power level drop between the pstream and the downstream position of an acostic filter, when the downstream section is anechoic, as schematized in Fig. 1. Generally, dring experiments, both inlet and otlet tbes are sch that plane waves propagation is ensred for a wide freqency range. a 1 d dt p = (1) in which represents the sond velocity propagating in the ndistrbed medim. Eqation 1 represents the strong formlation of the problem which, if solved, gives the exact soltion. In order to perform sch a Finite Element Analysis, the freqency version of the wave eqation is generally preferred, which is known as Helmholtz eqation: p ˆ + k pˆ = () in which k is the spatial freqency or wave nmber and eqals to ω a. By mltiplying the Helmholtz eqation by a test fnction o and integrating over all the domain, it becomes: pdv ˆ + k pˆ dv (3) Then, integrating by parts, it is possible to obtain the following weak formlation of the sond propagation PDE, as reported in the following: pdv ˆ + k pdv ˆ + pˆ ds n (4) = Then the discretization process of (4) is accomplished throgh a standard Galerkin formlation[3], which leads to a system of linear eqations with the pressre nodal nknowns, whose matrix formlation is represented by (5). [ K + i C ω M ]{ pˆ } = ω (5) In eqation 5 K, C and M represents the acostic stiffness, damping and mass matrices, respectively, whilst { pˆ } is the nknowns vector. The above process does not change in presence of acostic Fig. 1 Transmission Loss definition In terms of standing wave components of the sond filed, anechoic termination means that there are no back moving waves within the otlet tbe, being the reflection coefficient eqal to nity. Formalizing, the pressre and particle velocity at the pstream and downstream position may be expressed as: p = A + B (6) 1 = ρa d A d A B (7) p = (8) 1 = (9) d A d ρa where A and B are the amplitdes of the forward and back moving waves, the sbscripts and d stand for pstream and downstream tbes respectively, whilst S and the prodct ρ a stand for the cross section of the pstream and downstream connection tbes and the characteristic impedance of the medim respectively. Mathematically, the Transmission Loss is defined as the ratio of the incident and transmitted power into an anechoic termination, namely Win TL = 1 Log (1) Wot Sch parameter is the most adopted for characterizing the acostic performance of a system becase it only depends pon its geometry and material. By making se of Eq. (6-9), the incident and transmitted acostic powers may be expressed in the plane wave range as W in A = ps = S (11) ρa ISSN:

3 W ot Ad = ps = S (1) ρa may be sed to calclate the desired Transmission Loss, as indicated by (14): Ths the expression for the Transmission Loss is A TL = Log (13) Ad Its experimental calclation may be performed by calclating A and A d thanks to the Kndt s tbe techniqe[]. A. Software Implementation The model developed in Virtal Performance Soltion, in particlar within the NVH-Interior Acostics environment of VPS [6], follows the definition of TL, i.e. the difference between incident and transmitted power into an anechoic termination. Therefore, the following bondary conditions have been applied to the acostic mesh: 1) a constant velocity of 1 m/s at the inlet section as well as a D absorber acostic in order to ensre only incident acostic waves, ) D absorber acostic at the otlet section in order to simlate the anechoic termination. The D absorber acostic BC is typically applied in VPS environment in order to simlate a semi-infinite flid (impedance characteristic given by the prodct of air density and sond velocity). In Fig., the above described bondary conditions are depicted. TL = Log ρcv p (14) The implementation logic of nmerical models in VPS is based on the constrction of modles and stages. The modle represents the model, namely the acostic mesh and the bondary conditions, whilst the stage represents the chosen analysis. For the investigated system, one modle and two stages, have been created. More precisely, in the first stage acostic modes of the model are extracted and then sed as inpt to the second stage for freqency response analysis calclation. Both modles and stages are edited as cards which represent the command lines which will be read from the solver, PAM- NVH. III. INTAKE SYSTEM ACOUSTIC ANALYSIS The so-called engine breathing noise is de to both intake and exhast systems and it is the most prominent vehicle noise sorce when the engine s load is high and the vehicle itself is still or moving at low speeds. Generally speaking, this kind of noise may be seen as made of two components which are shell noise and tail pipe noise. The first type is de to the strctral vibrations exited by the pressre distribtion inside the system and may be very distrbing in correspondence of the strctral resonant freqencies. On the other hand, the second type of noise (so-called airborne noise), which is radiated by the otlet moth of sch systems, consists of primary noise sorces and secondary noise sorces. The primary sorces are de to the travelling pressre waves cased by the plsating flow at each IVO crank angle positions, and their spectral components consists of mltiple of the firing freqency. The secondary noise sorces represent the trblence and other flow indced aeroacostics sorces, e.g. vortex separation, whose spectral components represent the broadband noise. In Fig. 3, a typical intake noise spectrm (sond pressre level expressed in db) of a spark ignition engine is depicted as fnction of the freqency. Fig. Bondary conditions applied within the VPS model: a) inlet velocity and b)anechoic termination In this way, the sond pressre at the otlet termination, ISSN:

4 Spectrm [db] Fig. 4 Intake system nder investigation Fig. 3 Tail pipe noise of a for cylinder SI engine at 8 rpm As it is possible to appreciate from the above figre, the tonal spectral component (mltiple of the firing freqency or engine orders) as well as the broadband noise are clearly visible as main responsible for the tail pipe noise. If the focs is on sedan car (and not for example on motorbike), there are two main reasons why it is important to control the intake breathing noise. Firstly, as mentioned before, it represents one of the most prominent noise sorces at high load and low vehicle speeds[6]. Therefore, redcing intake noise represents a benefit with respect to the overall noise radiated by a vehicle, allowing to respect the emission standards. Besides, a lot of attention mst be paid within the low freqency range where the tonal spectral components may excite the acostic closed space resonant freqencies, redcing the acostic comfort of passengers. Sch copling effects may become very prominent ntil 4 Hz, depending on the specific cabin geometry, which affects the resonant freqencies. From what above, the acostic project of intake systems is a very complicated task, more than for exhast systems, becase the prior goal mst be to maximize the cylinders filling and not to attenate the sond radiation. The global system is composed by five sbsystems: inlet (yellow), filter box (green), air-box (red), otlet (white), engine cover (black) as it is highlighted in Fig. 5 a) and b). Within this early stdy, the presence of the air filter has not be considered as well as the presence of mean flow and the strctral participation. This latter hypothesis has been confirmed by previos analysis, at least in the investigated constraint condition[6]. Ths, apart from the above mentioned bondary conditions, the frther following conditions have been imposed. p n wall = (15) v = (16) Thanks to the momentm balance, (15) simply means that the particle velocity at the wall eqals zero m/s. As previosly stated, this work deals with the acostic performance assessment and optimization of an intake system for atomotive applications. The considered acostic performance parameter is the Transmission Loss and the tested geometry is shown in Fig. 4. ISSN:

5 db and only the region [4; 1] Hz has been shown becase, as it has been highlighted in previos stdy[7], in this range a perfect agreement has been fond between the two analysis. Fig. 5 a) Top and down b) views of the intake system In Fig. 6 and Fig. 7 the CAD model of the air volme embedded within the system and the corresponding acostic mesh are shown. It is worth noting that all the acostic mesh which will be shown allow accrate reslts ntil more or less 3 Hz, choosing six point per wavelength as spatial resoltion. TL [db] Fig. 8 Transmission loss assessment: model validation Fig. 6 CAD model of the acostic domain Moreover, both the inlet and the otlet dcts have been modified with respect to the original ones in order to reprodce the experimental layot[6]. In the graph, the ble dashed line represents the experimental data, whilst the black rhombs line represents VPS otcomes. Dring the experimental facilities, the system has been inserted within a big box fll of sand in order to ensre no strctral participation[6]. As it is possible to appreciate by the examination of Fig. 8, the VPS soltion globally nderestimates the experimental findings bt a perfect freqency agreement has been fond, meaning that no freqency shift hold[7]. So, the main reslts coming from the above TL analysis are that the VPS model perfectly reprodces the acostic performance of the actal system and, secondly, the investigated system does not ensre a good acostic attenation within the low freqency range[6]. Sch behavior is not acceptable since the exiting freqencies of an internal combstion engine are mltiple of the firing freqency. This latter is given by (17): n z f c = (17) 6 ε Fig. 7 Air volme s acostic mesh: 1833 elements (CTETRA 1-noded) and 931 Dring all the simlations, the air density and the sond velocity have been set eqal to 1, kg/m 3 and 34 m/s, respectively, whilst the freqency step has been fixed eqal to 5 Hz. Finally, the experimental/nmerical Transmission Loss assessment is depicted in Fig. 8. Here, the TL is expressed in where n is the revoltion per minte, z is the nmber of cylinders and ε represents the nmber of revoltion per thermodynamic cycle. Ths, for example, considering a for cylinders spark ignition engine, the firing freqency at 1 and rpm eqals 33 and 66 Hz respectively. Besides, in the freqency region where the flow noise sorces are prominent, namely from 3 to 6 Hz, no practical attenation holds. Therefore a geometrical change aimed to improve the noise attenation p to 6Hz shold be performed. For this prpose, it is necessary to associate the varios regions of the TL profile to each sbcomponents the whole intake system consists of. ISSN:

6 Hence, from the theory of the acostic filters[], it is obvios that the smooth profile which ranges from 4 p to 6 Hz is de to the main expansion chamber[6]. It is well known that, considering the transmission loss of sch element, the amplitde of the peaks depend on the ratio between chamber s cross section and that of both the connecting pstream and downstream dcts[14]. In particlar the greater the area ratio, the higher the peaks amplitde. Therefore, in order to increase the TL of the considered system, a CAD modification is reqired. To the aim of increase the above mentioned area ratio, the inlet dct of the intake system has been modified as depicted in Fig. 9. Fig. 11 Transmission loss comparison Fig. 9 CAD modification of the inlet dct As it is possible to appreciate from Fig. 9, the otlet section of the inlet tbe has been redced as mch as possible, meaning that a very small cross sectional area wold accelerate too mch the air flowing into the system with high distribted pressre drops (this last being proportional to the sqare of the mean flow velocity). The corresponding acostic mesh is depicted in Fig. 1. Of corse, the first peak of the crve is still located at the same freqency, abot 69 Hz, becase it is de to the resonant cavity, highlighted in the above figre, which has not been modified. Moreover, the increased TL vales arond 34 and 8 Hz are explained by changes of the acostic modes arond sch freqencies. In fact, considering the original geometry configration, the acostic modes at 3 and 866 Hz are sch that the expected sond pressre at the otlet is higher than at the inlet[6]. Vice versa, thanks to the inlet modification, the pressre distribtion corresponding to the acostic modes at 37 and 811 Hz are shown in Fig. 1. Here, the hotter the color the higher the sond pressre. Fig. 1 Air volme s acostic mesh with modified inlet dct: elements (CTETRA 1-noded) and nodes The effects of the modified inlet dct on the Transmission Loss are shown in Fig. 11. In the pictre, the black and ble lines refer to the original and modified geometry respectively. Fig. 1 Cavity modes at: a) 37 Hz and b) 811 Hz ISSN:

7 Ths, as expected, increasing the cross sectional area ratio reslts in an higher amplitde of the TL profile in almost the whole investigated range. More precisely, the main benefits have been achieved within the freqency window [; 6] Hz, e.g. 6dB at 465 Hz, whilst no sensible improvements have been fond before Hz. However, at lower freqencies where a copling between breathing noise and acostic cavity modes may appear, i.e. below 4 Hz, the reached TL enhancement cold be not satisfactory. For this reason, in order to frther increase the TL profile at low freqencies a filter box enlargement has been foreseen. The modified geometry and the corresponding acostic mesh are depicted in Fig. 13 and Fig. 14 respectively. Fig. 13 CAD modification of the filter box As it is possible to appreciate by the examination of Fig. 13, sch increased volme has been achieved throgh an additional height of the filter box. More precisely, the additional height eqals cm. TL [db] Fig. 15 Transmission loss comparison As it is highlighted in Fig. 15, the effect of filter box enlargement is analogos to the inlet dct modification, namely the TL profile has been simply shifted in amplitde. Sch circmstance is not srprising becase this latter modification relies on a cross sectional area increase too. From the TL trends, it comes ot that this frther geometry change ensre the best acostic performance for the intake system except arond 74 Hz, as ptted in evidence by the green circle in the above figre. The low vale of sond attenation at sch freqency (abot 3 db), is certain de to the inflence that the acostic mode depicted in Fig. 16 has on the system s response. Fig. 14 CAD modification of the filter box: elements (CTETRA 1-noded) and 1845 nodes The corresponding Transmission Loss simlation otcomes are shown in Fig. 15 where the red crve refers to the new configration. Fig. 16 Acostic modes at 78 Hz corresponding to the last CAD modification Nevertheless, this low TL vale wold not affect the overall acostic performance since, as previosly mentioned, the amplitde of the spectral components corresponding to both the primary and secondary noise sorces, which constitte the tail pipe breathing noise, start to rapidly decrease after 6 Hz. In conclsion, within the investigated freqency range, a great improvements have been achieved thanks to the analyzed modified geometry. In fact, the average sond attenation of the original device eqals,5 db whilst it wold be eqal to 9,7 db if the two stdied CAD modifications wold be realized. Obviosly, here no attention has been paid on the ISSN:

8 possibility that the investigated changes are not allowed becase of no available space when the intake system is located on the engine head. A more accrate analysis shold take into accont the degrees of freedom in terms of available space for the geometry modification. IV. CONCLUSION In this work, the acostic performance of an intake system for atomotive applications has been nmerically evalated in terms of its Transmission Loss, in absence mean flow, strctral participation and presence of the air filter. For this prpose, a 3D finite element model of the system has been realized in Virtal Performance Soltion software environment, powered by ESI Grop. First of all, the model has been validated by means of a comparison with experimental analysis coming from previos stdies. From this evalation stage, the system has trned ot to be not well acostically projected. Therefore, thanks to the very good agreement between experimental and nmerical findings, two geometry modifications have been tested on the nmerical model to the aim of performing an acostic performance enhancement procedre. No limits have been fixed for the degrees of freedom in terms of available space for the geometry changes. The foreseen new configrations ensre, together, a great improvements of the Transmission Loss, being the average gain of abot 7 db within the whole investigated freqency range. Frther analysis mst be addressed to confirm sch reslts when modeling the real working condition of the intake system, namely with the presence of mean flow inside the system, taking into accont the strctral participation (in real constraint conditions) and the presence of the air filter. ACKNOWLEDGMENT The athors are gratefl to ESI Grop in the person of G. Pagnozzi for provision of expertise, for software release and technical spport in the implementation. REFERENCES [1] Siano, D., Teodosio, L., De Bellis, V., and Bozza, F., "Analysis and Design of an Intake Filter Box for a Downsized VVA Engine," SAE Technical Paper , 14, doi:1.471/ [] M. L. Mnjal, Acostics of Dcts and Mfflers, John Wiley & Sons, [3] O. C. Zienkievicz and R. L. Taylor. The Finite Element Method, volme 1, The Basis. Btterworth-Heinemann,.. [4] E. B. Magrab. Environmental Noise Control. John Wiley & Sons, New York, [5] VPS 14 Solver Reference Manal. [6] Siano, D., Ferrara, G., Lenzi, G., D Agostino, D., and Fioravanti A., Experimental and Nmerical Comparison of the Acostic Performance of the Air Filter Box of a SI-ICE, to be pblished at 1th International Conference on Engines & Vehicles, 15 [7] Siano, D., Aiello, R.., D Agostino, D., On the evalation of commercial FEA software for acostic performance of complex system, proceedings of WSEAS conference-recent Researches in Mechanical and Transportation Systems, pp , 7-9 Jly 15, Salerno, Italy, ISBN: [8] Siano D., M. Viscardi, and Aiello R Sensitivity analysis and correlation Experimental/Nmerical FEM-BEM for Noise Redction assessment of an engine beaty cover, to be pblished in Energy Proceedia, 15. [9] R. Citarella, and D. Siano, Transmission loss assessment of an air indction system by BEM, In: ICAD6, The Forth International Conference on Axiomatic Design, Firenze, Italy, Jne 6, pp [1] Siano, D., Ariemma F., Bozza, F., Pros and cons of sing different nmerical techniqes for transmission loss evalation of a small engine mffler, SAE Paper / 198, doi: 1.471/ Small Engine Technology Conference (SETC) 1 Conference, Linz, September 1, doi:1.471/1-3-8, ISSN [11] Siano, D., Bozza, F., Teodosio, L., and De Bellis, V., "A Comparison Between External and Internal Resonators Employment to Redce the Gas-Dynamic Noise of a SI Engine," SAE Int. J. Engines 8(1):4-5, 15, doi:1.471/ [1] D. Siano, Transmission loss and gas-dynamic noise analyses of a downsized spark ignition engine, Proceedings of the Instittion of Mechanical Engineers, Part C: Jornal of Mechanical Engineering Science, Volme 9, Isse 4, 15 March 15, Pages , ISSN: 95446, DOI: / [13] Siano, D. and Bozza, F., "CFD Gas-Dynamic Noise Prediction of a VVA Engine Intake System," SAE Int. J. Engines 6(): , 13, doi:1.471/ [14] Abom M., An Introdction to Flow Acostics [15] Y. A. Rossikhin, M. V. Shitikova, Mhammed Salih Khalid J. M., The Impact Indced 3:1 Internal Resonance in Nonlinear Dobly Crved Shallow Panels with Rectanglar Platform, WSEAS Transactions on Applied and Theoretical Mechanics, ISSN / E-ISSN: / 4-349, Volme 1, 15, Art. #1, pp [16] Mohd. Ahmed, Adaptive Finite Element Simlation of Sheet Forming Operations Using Continm Elements, WSEAS Transactions on Applied and Theoretical Mechanics, ISSN / E-ISSN: / 4-349, Volme 1, 15, Art. #9, pp [17] O. Scharda, D. Mikolasek, J. Brozovsky, Finite Element Analysis and Modeling of Details Timber Strctre, international jornal of mathematical models and methods in applied sciences, ISSN: , Pages: , Volme 9, 15 [18] J. El-Mekkaoi, A. Elkhalfi, A Mixed Finite Element Method with new Bondary Condition, international jornal of mathematical models and methods in applied sciences, ISSN: , Pages: , Volme 8, 14. D. Siano was born in Naples in 6/1/1969 Italy, and gradated in Aeronatical Engineering at the University of Naples Federico II, Italy in Until 1, she was researcher in acostic and vibration department at C.I.R.A. (Italian aerospace Research Center). From 1 ntil now,she is a Researcher at National Research Concil of Italy (CNR) in the field of Acostic and Vibration in transport field. She is responsible of Acostic and Vibration Laboratory in her Instittion. Expert evalator within the EU 6th and 7th Framework Research Programme, in Transport-Aeronatics in 6 and 7. Project expert evalator in Ministry Economic Development, Italy. Referee for some International Jornals and session organizers collaborating with SAE conferences. She is athor of abot 75 Scientific Papers pblished on International Jornals and Conferences Proceedings and editor of two scientific books. She is ttor of several thesis and PhD thesis, as well. R. Aiello was born in Naples Italy in 4/8/1987, and gradated in Mechanical Engineering at the University of Naples "Federico II", Italy in 13. He is crrently a PhD Stdent in Indstrial Engineering at University of Naples Federico II in collaboration with Istitto Motori CNR, Italy. He is dealing with acostic and vibration in atomotive field. D. D'Agostino was born in Naples, Italy in 1/11/199 and gradated in Mechanical Engineering at the University of Naples Federico II, Italy in 13. He is crrently a PhD stdent in Indstrial Engineering at the University of Naples Federico II in collaboration with Istitto Motori CNR, in the field of acostic and vibration in atomotive. ISSN: