Final Report Summary - COMVEBONOV (Computational Modelling and Analysis of Automotive Vehicle Body Noise and Vibration)
COMVEBONOV is the acronym for the European Union (EU)-funded Seventh Framework Programme (FP7) Marie Curie Industry-Academia Partnerships and Pathways (IAPP) project on the 'Computational modelling and analysis of vehicle body noise and vibration'. This 48-month 'People' project ran from June 2008 to June 2012, involving the University of Sussex, United Kingdom (UK), and industrial partner CDH AG, Germany. In this summary, the background to COMVEBONOV, the project objectives, the work description, the results achieved, and the expected impact are supplied.
The background and project objectives:
The analysis of vehicle noise, vibration, and harshness (NVH) forms an important part of design and refinement. A particular problem is how to predict the NVH characteristics of production vehicle bodies. Normal manufacture introduces variability into vehicle geometry, material properties, and suspension components. Lightweight vehicle structures are particularly prone to vibration and interior noise problems which unresolved, can adversely affect customer acceptance and driver safety. Measured NVH characteristics of production vehicles exhibit noticeable scatter - nominally identical vehicles can leave the assembly-line with widely different NVH characteristics.
In vibro-acoustics, the frequency range is classed as 'low', 'mid', or 'high' frequency. For high frequency analysis, the method of statistical energy analysis (SEA) is best used to account for structural variability. In the low- and mid-frequency regions, SEA cannot readily be used - alternative methods are needed, particularly for bounding the frequency response function (FRF). Mid-frequency methods are available which combine deterministic and statistical approaches but these are not practical for realistic structures. At the start of the COMVEBONOV project, an asymptotic (EV) extreme value FRF bounding method had been tested on mid-scale structures in the low-frequency region. This was seen as potentially suitable for large-scale structures in the mid- and high-frequency regions. The initial objectives of the project were to explore EV-based FRF bounding methods for realistic vehicle structures in the low frequency region, and to implement them into CDH/VAO software. This involved industrial-scale vibro-acoustic structural models containing stiffened thin-plate-panels, layered with absorption materials, such as foam-filled carpet. The objectives then switched to creating a high frequency modelling and simulation capability to test EV-based FRF bounding on nominally simple structural models. The final objectives focused on developing hybrid and EV concepts for mid-frequency FRF bounding, which if successful, would be implemented in CDH/VAO and verified using measured data obtained from real vehicles.
The work description:
Initial work at Sussex University focused on low frequency modelling, simulation, and prediction for 'academic' structures. Then, an early CDH/Sussex collaboration engendered a good basis on which to proceed with EV-based-bounding of steady-state and transient dynamic responses of realistic uncertain linear structures. CDH then implemented low frequency FRF bounding into CDH/VAO. Methods were developed by CDH to include porosity in current finite element modelling methods, including testing to determine relevant material parameters for a single absorbing carpet. CDH also implemented and tested Hybrid FEM-SEA mid-frequency methods in a prototype version of the CDH/VAO software, and gained valuable experience in the use of state-of-the-art commercial mid-frequency analysis software.
Separate work was undertaken at Sussex on the inverse problem, namely the determination of the statistical properties of uncertain structural parameters from response measurements. A method was developed to create an efficient way to compute failure probabilities for strongly nonlinear uncertain structures. Mainstream work at Sussex, on high frequency modelling, examined EV-based response bounding of structures across the entire frequency range. The discrete singular convolution (DSC) method was used to do the simulation, where predictions were compared with mode superposition, energy flow analysis, and SEA. But because the DSC method is limited to relatively simple structures, attempts were made to develop a fast high-frequency Monte Carlo simulation capability using a wavelet-based finite element method. Work has also explored high and mid-frequency theoretical EV-based prediction, not involving Monte Carlo simulation. The main effort at Sussex has focused on extending and adapting two SEA methods for use in the mid-frequency region. This has focused on more accurate computation of SEA coupling loss factors (CLFs) via average modal mobilities. Both SEA extensions have been tested (in collaboration with CDH) on two structures with random mass-position. One of these tests has involved Nastran-based Monte-Carlo simulation of a realistic car floor model, and to consider further testing, effort has been made by CDH to obtain an industrial partner to allow response measurements from nominally identical production vehicles.
The results achieved and expected impact:
The work involving use of the DSC method for Monte Carlo simulation has shown that EV-based bounding of FRFs is generally applicable across the entire frequency range. But the findings also give considerable impetus to a future theoretical study to create an EV-based bounding method that does not require Monte Carlo simulation. Regarding the extensions of SEA into the mid-frequency region, the research shows that for CLF determination, the use of average modal mobilities appears to give accurate and robust SEA predictions for both high- and mid-frequency analysis. This has resulted in three joint publications with Sussex and CDH. The main potential impact will ultimately be on reducing unwanted vehicle noise and vibration by meeting the need for appropriate practical tools to enable better modelling and prediction at initial design and refinement stage. This will benefit manufacturers by providing the ability to reduce vehicle weight without creating new NVH problems - the realisable economic benefits being through improved fuel economy and product competitiveness. Societal and environmental benefits will also be derived through reduction of unwanted vehicle noise, and the elimination of a major source of pollution. In extreme cases, health and safety benefits will also be accrued through reduction of unwanted noise within a vehicle which can impair driver concentration. Finally, this IAPP has involved very beneficial collaboration between Sussex University UK and CDH AG Germany. For CDH AG, the main result of the project is that know-how has been gained by having a number of staff highly trained in the theory and practical application of mid and high-frequency vibration analysis. It is expected that this situation will lead to commercial benefits in the event that such methods are accepted by industry.
The background and project objectives:
The analysis of vehicle noise, vibration, and harshness (NVH) forms an important part of design and refinement. A particular problem is how to predict the NVH characteristics of production vehicle bodies. Normal manufacture introduces variability into vehicle geometry, material properties, and suspension components. Lightweight vehicle structures are particularly prone to vibration and interior noise problems which unresolved, can adversely affect customer acceptance and driver safety. Measured NVH characteristics of production vehicles exhibit noticeable scatter - nominally identical vehicles can leave the assembly-line with widely different NVH characteristics.
In vibro-acoustics, the frequency range is classed as 'low', 'mid', or 'high' frequency. For high frequency analysis, the method of statistical energy analysis (SEA) is best used to account for structural variability. In the low- and mid-frequency regions, SEA cannot readily be used - alternative methods are needed, particularly for bounding the frequency response function (FRF). Mid-frequency methods are available which combine deterministic and statistical approaches but these are not practical for realistic structures. At the start of the COMVEBONOV project, an asymptotic (EV) extreme value FRF bounding method had been tested on mid-scale structures in the low-frequency region. This was seen as potentially suitable for large-scale structures in the mid- and high-frequency regions. The initial objectives of the project were to explore EV-based FRF bounding methods for realistic vehicle structures in the low frequency region, and to implement them into CDH/VAO software. This involved industrial-scale vibro-acoustic structural models containing stiffened thin-plate-panels, layered with absorption materials, such as foam-filled carpet. The objectives then switched to creating a high frequency modelling and simulation capability to test EV-based FRF bounding on nominally simple structural models. The final objectives focused on developing hybrid and EV concepts for mid-frequency FRF bounding, which if successful, would be implemented in CDH/VAO and verified using measured data obtained from real vehicles.
The work description:
Initial work at Sussex University focused on low frequency modelling, simulation, and prediction for 'academic' structures. Then, an early CDH/Sussex collaboration engendered a good basis on which to proceed with EV-based-bounding of steady-state and transient dynamic responses of realistic uncertain linear structures. CDH then implemented low frequency FRF bounding into CDH/VAO. Methods were developed by CDH to include porosity in current finite element modelling methods, including testing to determine relevant material parameters for a single absorbing carpet. CDH also implemented and tested Hybrid FEM-SEA mid-frequency methods in a prototype version of the CDH/VAO software, and gained valuable experience in the use of state-of-the-art commercial mid-frequency analysis software.
Separate work was undertaken at Sussex on the inverse problem, namely the determination of the statistical properties of uncertain structural parameters from response measurements. A method was developed to create an efficient way to compute failure probabilities for strongly nonlinear uncertain structures. Mainstream work at Sussex, on high frequency modelling, examined EV-based response bounding of structures across the entire frequency range. The discrete singular convolution (DSC) method was used to do the simulation, where predictions were compared with mode superposition, energy flow analysis, and SEA. But because the DSC method is limited to relatively simple structures, attempts were made to develop a fast high-frequency Monte Carlo simulation capability using a wavelet-based finite element method. Work has also explored high and mid-frequency theoretical EV-based prediction, not involving Monte Carlo simulation. The main effort at Sussex has focused on extending and adapting two SEA methods for use in the mid-frequency region. This has focused on more accurate computation of SEA coupling loss factors (CLFs) via average modal mobilities. Both SEA extensions have been tested (in collaboration with CDH) on two structures with random mass-position. One of these tests has involved Nastran-based Monte-Carlo simulation of a realistic car floor model, and to consider further testing, effort has been made by CDH to obtain an industrial partner to allow response measurements from nominally identical production vehicles.
The results achieved and expected impact:
The work involving use of the DSC method for Monte Carlo simulation has shown that EV-based bounding of FRFs is generally applicable across the entire frequency range. But the findings also give considerable impetus to a future theoretical study to create an EV-based bounding method that does not require Monte Carlo simulation. Regarding the extensions of SEA into the mid-frequency region, the research shows that for CLF determination, the use of average modal mobilities appears to give accurate and robust SEA predictions for both high- and mid-frequency analysis. This has resulted in three joint publications with Sussex and CDH. The main potential impact will ultimately be on reducing unwanted vehicle noise and vibration by meeting the need for appropriate practical tools to enable better modelling and prediction at initial design and refinement stage. This will benefit manufacturers by providing the ability to reduce vehicle weight without creating new NVH problems - the realisable economic benefits being through improved fuel economy and product competitiveness. Societal and environmental benefits will also be derived through reduction of unwanted vehicle noise, and the elimination of a major source of pollution. In extreme cases, health and safety benefits will also be accrued through reduction of unwanted noise within a vehicle which can impair driver concentration. Finally, this IAPP has involved very beneficial collaboration between Sussex University UK and CDH AG Germany. For CDH AG, the main result of the project is that know-how has been gained by having a number of staff highly trained in the theory and practical application of mid and high-frequency vibration analysis. It is expected that this situation will lead to commercial benefits in the event that such methods are accepted by industry.