NUMERICAL STUDY OF TRANSFER FUNCTION OF COM- BUSTION NOISE ON A HEAVY DUTY DIESEL ENGINE Ibrahim Ciylez Ford OTOSAN A.Ş., Sancaktepe, Istanbul, Turkey email: iciylez@ford.com.tr Haluk Erol Istanbul Technical University, Faculty of Mechanical Engineering, Taksim, Istanbul, Turkey In the automotive sector, diesel engine becomes common engine as much as gasoline engine due to diesel s enourmous torque outcome with more efficient way of fuel consumption. In contrast to excessive torque and fuel efficiency advantages, diesel engine produces cylinder pressure in order to satisfy higher compression ratio results in noise and vibration problem. Extreme cylinder pressure and generated dynamic forces cause excessive vibration and noise problem specifically as known as combustion noise. Researchers study on combustion noise to understand effective parameters and combustion noise have been associated with two essential parameters which are structural transfer function of the combustion noise and cylinder pressure. This study is performed to obtain structural transfer function of the combustion noise from combustion chamber to microphones. Structural FE model is built for a heavy duty diesel engine and surface velocity of the engine is obtained from nodes of the each element on the outer surface. These results are imported into acoustic analysis in order to collect microphone data as SPL in db format. Structural attenuation of the diesel engine has been extracted from each microphone results and combustion noise transfer function is plotted in 3rd octave frequency bands from 500 Hz up to 3150 Hz. Parametric study has been performed in order to satisfy correlation between test and CAE results. Study expresses the opportunity of numerical study on the transfer function of combustion noise for a heavy duty diesel engine can be extracted and transfer function plays a fundamental role to improve weak frequency contents of the structure in aspect of NVH. Keywords: Diesel engine, combustion noise, acoustic analysis, finite element methods 1. Introduction In the recent years, diesel type of engines are being used commonly and it has been a compelling engineering research area in aspect of noise and vibration. Diesel engine has produces dramatically high combustion noise according to its rival gasoline engine [1]. Nature of the diesel engine is to create combustion by adding fuel into pressurized air [2]. Generated cylinder pressure is about two times higher than a conventional gasoline engines. This results in excessive dynamic force on the cylinder side walls, piston top surface and cylinder head hence cylinder pressure contributes to the engine vibration and noise dramatically than other excitations [3]. Combustion noise is expressed with two parameters according to the studies which are cylinder pressure and transfer function of the engine structure. Both parameters are essential to understand and prevent unreasonable noise level. Cylinder pressure is the loading of the engine and the source in the acoustic nomenclature which is transformed into noise by the path of the vibration. This path is the structural transfer function of the engine. 1
Transfer function of the combustion noise is reliant to diesel engine design and structural properties. From combustion chamber to microphone, structure attenuates most of the noise which provides efficient and stabile noise absorption. Early stages of engine design provide straightforward modification on the structure according to predicted problems and transfer function of the combustion noise can be improved at this phase [2]. Generally, low frequencies are strongly attenuated by diesel engine structure that range is between 0 and 500 Hz. Although combustion noise is attenuated in low frequency range by the engine structural attenuation, high frequencies are vulnerable to combustion noise for a range between 500 Hz and 4 khz. Efficient attenuation range of the engine depends on structural modes and design of the components which can reduce the dominant range of combustion noise or can significantly increase [4]. ISVR has a methodology to be able to extract transfer function of the diesel engine by experimentally. Test rig denominated as Banger Rig is a special rig which creates combustion with special fuel and injector to control and measure cylinder pressure is created in the cylinder. The method is built to prevent all mechanical noise of the engine by locking mechanism of the crank train system so any piston motion is blocked and pure combustion noise can be measured by microphones from 1 meter distance [5]. Another method is a statistical method that noise breakdown and regression analysis method is based on noise separation of the engine, correlating combustion noise and cylinder pressure. The method implies that combustion noise and cylinder pressure are linearly dependent and relation between two values is the combustion noise transfer function of the engine [6]. Study includes predominantly numerical methods. Method is developed to illustrate transfer function can be obtained from computer aided engineering. Traditional method of acoustic calculation, boundary element method efficiently predicts radiated noise from the engine. Finite element method is used to calculate normal modes of the engine and accordingly surface velocity under specific loading. Surface velocity is measured via MSC Nastran and transferred into LMS Virtual Lab. Acoustic model is built to model test medium. Predicted structural surface velocity, structural modes and calculation of the acoustic transfer vectors in LMS Virtual Lab. are integrated to calculate response at microphone location [7]. In this paper, the aim of the study is to achieve combustion noise transfer function and satisfy correlation of transfer function between experimental study and CAE prediction. Transfer function of a heavy duty diesel engine is obtained by the finite element and boundary element method. Existing test method of the ISVR is adapted by these numerical methods and transfer function of the combustion noise is obtained. Comparison of the experiment and CAE results showed deterioration and parametric study is conducted to enhance the confidence of the CAE method. 2. Experimental Study Diesel engine under working condition creates various type of noise and generated noises can be listed as combustion noise, mechanical noise and flow induced noise. On a diesel engine, combustion noise is the most dominant noise type. Experiment was conducted to separate noise contributions and combustion noise is extracted from experiment [6]. Eq. (1) shows total noise and contributors to engine radiated noise. SPL Total Noise = SPL Combustion Noise + SPL Mechanical Noise + SPL Flow Noise (1) A heavy duty diesel engine is used to conduct experiment. Diesel engine is placed into semianechoic acoustic room and microphones are set to 1 meter distance from engine. Figure 1. shows experiment layout in the acoustic room and study includes only measurement of the engine. Transmission is excluded in the dynamometer. 2 ICSV24, London, 23-27 July 2017
Figure 1: Engine position and microphone locations Data acquisition is performed as load sweep condition and gas load is increased so that every data is acquired from the system. At the end of the measurement, data is stored from each microphone as load dependent and speed. On the other hand, cylinder pressure measurement is carried out from each cylinder at the same time. Every data is turned into frequency domain by FFT and data examination is conducted in 3 rd octave bands. Each microphone data is linked with cylinder pressures and transfer function is obtained in all conditions. H(ω) = SPL(ω) CP(ω) (2) As given in the eq. (2), transfer function which can be found as frequency response function in the literature is the difference between sound pressure level measured at each microphone and cylinder pressure at determined speed of the crankshaft [1]. According to given eq. (2), results are expected as negative db values. Experimental results obtained at each microphone and as 4 microphone average in the figure 2. Figure 2: Transfer function of combustion noise in 3 rd Octave Bands measured and calculated from experiment. ICSV24, London, 23-27 July 2017 3
Results show that attenuation at low frequencies are significantly better than high frequencies up to 3150 Hz. Combustion noise becomes dominant from 500 Hz up to 3150 Hz. These results can be explained with that engine components have high natural frequencies and transfer function is weak at high frequencies and vulnerable to combustion noise. In this study, CAE limits the maximum frequency and study has been conducted up to 3150 Hz. 3. Numerical Study 3.1 Structural FE Model Build and Dynamic Analyses Recent methodology to calculate sound pressure level is with finite element and boundary element methods which are coupling the structural and acoustic medium. The study requires a FE model of the engine, modal results and surface velocity from the outer surface of the engine. Figure 3. shows the FE model of subjected engine and performed surface velocity results. All structural FE analyses are conducted up to 3550 Hz which are normal mode analysis and dynamic surface velocity analysis. Figure 3: An illustration of heavy duty diesel engine FE model & Structural surface velocity results Structural results of the model have been used in the acoustic model to transfer structural results from FE model to BE model. 3.2 Acoustic Model Acoustic model includes two step. First of them is acoustic model preparation and ATV (Acoustic Transfer Vector) calculation. Second step is that structural results will be transferred into acoustic mesh and results will be calculated with eq. (3). P(ω) = ATV(ω) V(ω) (3) P is the pressure value as Pa with S-I unit system and V is the surface velocity value extracted from structural model as m/sec. ATV is a transfer vector which provides to transform velocity values into pressure at specific microphone locations. Figure 4. illustrates the method of the ATV and how to calculate microphone response from boundary element mesh. Figure 5. shows the boundary element mesh and acoustic model with microphone locations. Model is built to calculate sound pressure levels at same condition and same microphone locations. Figure 4: Illustration of ATV method and utilization of the ATVs for acoustic calculation 4 ICSV24, London, 23-27 July 2017
Figure 5: Illustration of BE mesh and Acoustic model with microphone locations 3.3 Transfer Function of Combustion Noise and Parametric Study to Improve Correlation with Experiment Structural normal mode analysis and dynamic analysis for surface velocity has been completed then, acoustic model is coupled with structural results. After ATV calculation, each microphone response has been calculated as seen in figure 6. Results are obtained in Pascal as pressure and transformed into sound pressure levels as db. Sound pressure levels are processed with post processing tool. According to results, transfer function with numerical calculation shows deterioration at high frequencies such as 1 khz and 2 khz 3 rd octave bands. Comparison of experiment and CAE results show that numerical study has the similar gradient to experiment curves. ICSV24, London, 23-27 July 2017 5
Figure 6: Experiment vs. CAE comparison of the combustion noise transfer function in 3 rd octave band Transfer function extracted from numerical calculation is deteriorated at determined frequencies. Root causes of the deteriorations are identified by normal mode analysis to understand which natural frequency affects excessively. Main driver natural frequencies are comprehended and effective parameters are updated according to narrow band results as shown in the figure 7. Modal results indicate parameters which should be modified and several iterations have been conducted to see parameters effect on transfer function. Improved engine model response have been shown in the figure 8. Each microphone results have been correlated with experimental study. Figure 7: Root cause investigation on narrow band and 3 rd octave band sound pressure levels in db 6 ICSV24, London, 23-27 July 2017
Figure 8: Noise transfer function of improved engine FE model in comparison with experimental results Figure 8. illustrates that transfer function of the combustion noise can be extracted with FE and BE tools to able to predict any transfer function weakness at certain frequencies. At design phase of a diesel engine, CAE tools can be used to avoid from abnormal combustion noise by using transfer function and weakness of the structure can be improved at resonance frequencies of the components and combustion noise can be reduced at early stages of the engine design phase. 4. Conclusions In this paper, numerical method has been developed to investigate transfer function of combustion noise with FE and BE tools, to improve engine design in aspect of NVH with the help of transfer function, study has been conducted. The aim was to build a methodology on combustion noise transfer function for a heavy duty diesel engine. Results indicate that correlation of experimental and numerical results are satisfied with improved engine model. Opportunity of early design improvements can be realized by the CAE method for transfer function of combustion noise. In this way, from the initial results to improved results, parametric study has been conducted to improve correlation between experimental and numerical studies. Improved engine model s transfer function has correlation with test as seen in the figures and there are still numerical data shows deterioration according to experiemtn. To reduce numerical error and make prediction magnitude level, enlarged experimental studies will be conducted to acquire natural frequencies and frequency response functions of the each component. ICSV24, London, 23-27 July 2017 7
REFERENCES 1 S. G.-q. W. Hai-qiao and H. Rui, The Transfer Function of Combustion Noise in DI- Diesel Engine, in SAE 2005 Noise and Vibration Conference and Exhibition, Traverse City, Michigan, 2005. 2 B. Challen and R. Baranescu, Diesel Engine Reference Book, Butterworth Heinemann, 1999. 3 K. A. Krishna, P. Bhardwaj, S. Patil and M. Kumbhar, Concept Phase Powertrain Development for NVH Using the Optimization Technique, SAE International, 2015. 4 K. Mollenhauer and H. Tschoeke, Handbook of Diesel Engines, Berlin: Springer, 2010. 5 University of Southampton, Institute of Sound and Vibration Research, ISVR Consulting, 17 September 2014. [Online]. Available: http://www.isvr.co.uk/automarine/bangerrig.htm. [Accessed 23 July 2016]. 6 I. Hirano, M. Kondo, Y. Uraki and Y. Asahara, Using Multiple Regression Analysis to Estimate the Contributions of Engine Radiated Noise, in Seoul 2000 FISITA World Automotive Congress, Seoul, 2000. 7 R. Feng, D. Zhang, J. Liu, J. FU and B. DENG, Efficient and Accurate Evaluation and Improvement of Engine Radiated Noise, SAE International, 04 08 2013. 8 ICSV24, London, 23-27 July 2017