MHiVec is the acronym for the European Union (EU)-funded Seventh Framework Programme (FP7) Marie Curie Industry-Academia Partnerships and Pathways (IAPP) project on Mid-to-High Frequency Modelling of Vehicle Noise and Vibration. MHiVec is concerned with a new simulation tool for noise, vibration and harshness (NVH) predictions in mechanical structures. The tool is based on an innovative method called Discrete Flow Mapping (DFM) which has been invented in 2011 by the coordinating partners of this consortium, Nottingham Trent University (NTU) and University of Nottingham (UoN), and has been tested on small scale bench-marking projects with the car maker Jaguar Land Rover. As design and construction decisions are increasingly made by means of virtual prototyping and as part of Computer Aided Engineering (CAE), efficient simulation tools in all areas of engineering are prevalent and sought after. A whole simulation industry has emerged serving the needs of the Original Equipment Manufacturers (OEMs). NVH software developers and consultants, as well as their OEM end-user customers are therefore the target groups for this research.
DFM makes it possible to efficiently compute vibrational energy distributions on structures of arbitrary complexity, making it ideal for an NVH software tool to be used by the non-specialist engineer. Vibrational energy flow can be computed on meshes harnessing the full flexibility of mesh based approaches, including conventional CAE methods such as the Finite Element Method (FEM). DFM has the potential to reduce NVH simulation costs considerably and provide a reliable everyday tool for the engineer, here with a focus on applications in the car manufacturing industry. The main objective of the project was to develop a modelling and simulation tool for a complete car body over the full frequency range up to 20 kHz.
The project consortium was led by NTU and UoN and included further academic expertise from the world-leading Institute of Sound and Vibration Research (ISVR) at the University of Southampton. The academic knowledge was channelled via two expert SMEs to OEMs. CDH AG (CDH) provided software and modelling expertise with a track record in NVH simulation for the automotive industry. inuTech GmbH (iT) is an expert in numerical solutions and algorithm development with expert knowledge on DFM. The consortium was completed by the associate partner (AP) Jaguar Land Rover (JLR), who provided benchmark problems and measurements.
In the first 24 months the majority of the work took place within the academic partners, including contributions from members of the industrial partners whilst on secondment. Work at NTU focussed on the understanding how the curvature of a structure affects the vibrational energy transport, the development of three-dimensional DFM elements and constructing global interface models between two-dimensional substructures and three-dimensional acoustic elements. The modelling of uncertainties has also been implemented at NTU, in collaboration with UoN, by designing a propagation operator that interpolates between random and deterministic propagation, which can be used in place of the standard DFM propagation operator. In addition, UoN have developed acoustic radiation and fluid structure interaction models leading to predictions of both the sound pressure levels in an interior region and the vibrational energy on the surrounding structure. The main focus of work at UoN was on the wider development of transfer operator approaches to facilitate efficient DFM modelling on multi-substructure meshes by providing point to point connections and interface relations without requiring explicit geometric connections. Work at ISVR has concentrated on local models for small sub-structures, using the Wave Finite Element Method (WFEM) where wave effects are captured. ISVR have led the subsequent development of local interface models that can be implemented along with DFM in a hybrid fashion, including the connections most commonly found on vehicle structures. Research at ISVR has also contributed to the modelling of composite elements, which can again be implemented in DFM as part of a hybrid DFM/WFEM scheme.
In the second 24 months the majority of the work has taken place within the industrial partner inuTech GmbH, including contributions from members of the academic partners whilst on secondment. One of the central tasks has been to construct global DFM meshes for complete BiB structures. This has been performed for two distinct industrial examples to demonstrate the versatility of the methodology: (i) a glazed Body in White (BiW) mesh from JLR (the closest testing stage to BiB that is used at JLR; note that BiW is the BiB without moving parts and glass) as well as (ii) a BiB structure from tractor manufacturer Yanmar Co Ltd. An infrastructure for large-scale computations on full vehicle structures with realistic NVH sources has been developed. In order to deal with large matrix systems, the computation architecture has been based on a parallel multiprocessor environment with shared memory computing and iterative solvers. The DFM software developed within inuTech GmbH has been augmented with an automated meshing tool, which can transform traditional FEM-meshes from popular software tools such as NASTRAN and ANSYS into DFM meshes. As such, links from the global DFM solver to standard low-frequency solvers have been developed, making it possible to run simulations across the full frequency range. A graphical user interface and analysis tool has been set up including post-processing via third party software packages such as VTK. A number of validation tests have been performed in both CDH and inuTech. CDH have contributed data and performed tests for a car floor, a car floor with rails, the acoustic space inside a BiB and a spot welded panel. InuTech have performed further tests for a spot welded panel as well as for larger structures including, a quarter car, a BiW and finally a BiB.
Results, Conclusions and Impact:
The capability of DFM has been extended considerably to make it both viable for modelling a full vehicle BiB structure and marketable as a software product within a user-friendly package with interfaces to other popular software tools. A major area of progress has been the development of several techniques whereby meshes of substructures can be assembled into a full structural model. These techniques include the development of a DFM counterpart to the Rigid Body Elements (RBEs) used in FEM models. A second approach has also been implemented and involves updating the mesh geometry, including the addition of new nodes and edges, so that direct geometric connections are formed between all substructures. The enhancements to DFM modelling achieved within the academic partners have been transferred to the SME commercial partners, and testing and validation has taken place using a range of substructure problems as well as full BiB structures. A comparison of DFM with results from other methods including FEM and Statistical Energy Analysis (SEA) has been performed for the substructures described above. The results showed that DFM can handle the high frequency range of SEA and gives more detailed results like FEM, but without the large computational expense associated with FEM for high-frequency simulations. Tests on full BIB type structures showed great promise, even down to relatively low frequencies, where the DFM results can be readily supplemented with efficient FEM calculations of the resonant peaks. Associate partner JLR have been involved in the testing process and have provided measurement data for low to mid-frequencies. Higher frequency test data from Yanmar showed even better agreement with the DFM simulations.
In the short term, the principle impacts of the project are: 1. a substantially increased level knowledge and training on DFM within all project partners, and 2. the development of a marketable DFM software tool by inuTech GmbH. All four recruited researchers remain employed within the consortium partners at present, where the knowledge and experience gained during the MHiVec project has proved valuable in helping them secure longer term positions, and enhancing their hosting research groups in mid and high frequency NVH simulation methods. Two of the researchers are employed at inuTech where they are involved in development and consultancy activities for the DFM software tool, which is already being marketed commercially. The long-term impact of the project is envisaged to be that DFM technology will become one of the main methods for the NVH simulation market. The successful introduction of these software tools will stimulate further R&D efforts, mainly in a European context, turning Europe into a focus of the NVH simulation world