Search

Back to overview




Results: 27
26/10/2021
08:35 conference time (CEST, Berlin)
Future Virtualized Engineering – A Journey From Applied Research Perspective
26/10/2021 08:35 conference time (CEST, Berlin)
Room: A
B. Fachbach (Virtual Vehicle Research GmbH, AUT)
B. Fachbach (Virtual Vehicle Research GmbH, AUT)
Automotive industry is undergoing more or less radical change at different levels. Environmental requirements gain higher interest, mobility concepts influence the traditional business cases, the product is getting smarter, and new partners on the market or from other domains force industry to re-think collaboration. Challenges and fields of action cover a huge range from availability and reliability of data, gaining effectiveness in the early development phase, a more consequent integration of systems engineering into lifecycle processes, the lacking interoperability of data and models, demand for simulation governance and traceability, to the adequate role of data analytics and machine learning within continuous system integration and deployment. The risk is to take decisions and measures without having a big picture or vision what has to be expected within the next 5-10 years. The lecture will analyze major differences of current and future development, give an overview on the future oriented fields of action in the context of advanced virtualized development, and takes a look on some promising methodical and technological approaches.
26/10/2021
08:35 conference time (CEST, Berlin)
Enhanced Correlation for off Highway Vehicle Wheel Loads Using an Integrated Multiphysics Multibody Dynamics Simulation Model
26/10/2021 08:35 conference time (CEST, Berlin)
Room: L
R. Udasi (John Deere India Pvt Ltd - Pune, IND); A. Shah (John Deere, USA)
R. Udasi (John Deere India Pvt Ltd - Pune, IND); A. Shah (John Deere, USA)
Wheel loads are one of the most sought-after inputs for durability evaluation of chassis and suspension system in off-highway agricultural vehicles. Collecting wheel loads through physical test is cumbersome in terms of time, cost, and efforts. Instead, usage of a simulation-based approach to predict accurate wheel loads during early phase of product development has lot of usefulness and benefits. This paper will talk in detail about using an Integrated Multiphysics Multibody Dynamics Simulation Model with Hydrostatic Transmission and Embedded Software for off highway vehicle which is also called CommandDrive powertrain that consists of a single pump which sends flow to all four variable displacement motors separately to drive all four wheels. If one or more wheels loose traction, the system will adjust to slow the wheels that are slipping and direct more flow to the remaining wheels to retain all wheel drive. In this paper, a comparison of simple to advanced level of simulation integrated model for predicting wheel loads on a Self-Propelled Sprayer during a turning operation will be presented. During a turn, rpm of inner wheel will be less than the rpm of outer wheel. This is because the inner wheel travels less distance as compared to outer wheel. To capture the turning phenomenon accurately, different fidelity levels of integrated models including Hydrostatic Transmission or Embedded Software and their pros and cons will be discussed as below: 1. All wheel’s same speed input driven vehicle level Rigid MBD model – Traditional Approach 2. Vehicle speed input driven Hydrostatic transmission equivalent Mechanical drivetrain full vehicle model 3. Hydrostatic transmission integrated vehicle Multiphysics model 4. Integrated Multiphysics Multibody Dynamics Simulation Vehicle Model with Hydrostatic Transmission and Embedded Software The effect and significance of Multiphysics integrated simulation model’s increased fidelity would then illustrate enhanced wheel loads correlation with respect to test data and show importance of integrated modeling approach.
Powertrain, Embedded Software, Integrated, Multiphysics, Multibody Dynamics Model, Dynamic System Modeling, Hydrostatic Transmission, Wheel loads, Enhanced Correlation, Off highway Agricultural vehicle
26/10/2021
08:55 conference time (CEST, Berlin)
Shape Optimization of Tire Tread Pattern to Minimize Water Splashing on Vehicle Body Using Particle Method CFD Simulation
26/10/2021 08:55 conference time (CEST, Berlin)
Room: F
S. Tokura (Prometech Software, Inc., JPN)
S. Tokura (Prometech Software, Inc., JPN)
With the electrification of power, the enhancement of safety equipment, and the sophistication of automatic driving systems, many electrical components such as sensors, radars, and cameras have been installed in automobiles in recent years. It is necessary to take waterproof measures for these electrical components and to take measures to prevent wetness by water as much as possible. For underbody, which requires strength, anti-corrosion coating measures are required to prevent damage caused by water exposure containing chemical substances such as anti-freezing agents. Therefore, the manufacturing cost for the protection from water splashing is not so small. One of the ways to reduce the cost of the damage countermeasures is to reduce the amount of water that hits the lower part of the vehicle body. For this purpose, it is considered that the reduction of the water splashing caused by the tires is effective. In this paper, we applied a shape optimization method based on numerical simulation and tried to minimize the amount of water being splashed on the underbody by optimizing the tread pattern shape of the tire. In simulating such problems, it is necessary to model the water splashing caused by tire rotation from the road surface to the vehicle body surface. As a numerical computation method that can efficiently track the separation, collision, and coalescence of many water droplets in a wide space, the computation method based on the particle method is considered more suitable than the mesh-based conventional CFD computation. Besides, using the mesh-free particle method has the advantage that the remeshing does not be required when changing the shape of the flow path. Therefore, in this paper, we adopted Particleworks, a CFD software that uses MPS (Moving Particle Simulation), which is one of the major particle methods, for numerical simulation. In addition, CAD-based parametric shape optimization software CAESES was used to optimize the shape of complex tread patterns, and by efficiently repeating the optimization computation cycle, the optimum shape could be obtained in a short time.
Shape optimization, Particle method, MPS, CFD, Meshfree
26/10/2021
09:15 conference time (CEST, Berlin)
Data Compression for Simulation results
26/10/2021 09:15 conference time (CEST, Berlin)
Room: K
S. Müller, F. Natter, H. Talaat, C-A. Thole, T. Weinert (SIDACT GmbH, DEU)
S. Müller, F. Natter, H. Talaat, C-A. Thole, T. Weinert (SIDACT GmbH, DEU)
Crash, Noise Vibration Harshness (NVH) and Computational Fluid Dynamics (CFD) are three pillars in Automotive Computer Aided Engineering (CAE). What these areas have in common is that they use numerical simulations to improve an initial design until it meets the specified requirements, and thus justifies the production of a physical prototype. These simulations require, depending on the application area, between a few hours on a multi-core system for NVH simulations, to several days on an HPC in CFD. Assuming a mid single-digit cent amount per core hour, this results in a price of a few euros up to several thousand euros per simulation, without considering the costs for the simulation code. Thus, archiving the simulation result is preferable to re-simulating also in terms of environmental protection. Depending on the field of application, the size of a simulation result varies between a few hundred MB and already reaches the terabyte range, especially in CFD. A proven approach to deal with large amounts of data is compression. Without data compression, for example, today's communication and work from a remote place with video calls and online presentations would be impossible. Lossy compression methods that have been specially developed for audio and video streams, exploit their properties for efficient compression depending on the available bandwidth and the quality required by the user. The outstanding compression rates, while maintaining good quality, can only be achieved by addressing and taking into account the specific characteristics of the data in question. In the present work, we compare different approaches for the compression of simulation results. For this purpose, we model dependencies of data points based on the simulation grid, simulation time and between simulation results. It is shown that the more dependencies are considered, the more effective compression can be implemented. In addition, we specifically address the properties that simulated data contain numerical and modelling errors and therefore the used precision of 32bit or 64bit for simulated post values is window dressing. Hence lossy compression is applicable. Justified by rate-distortion theory, it is possible to achieve higher compression factors by relaxing the requirements on the required precision for reconstruction. Exemplarily, we consider modelling grid dependencies for NVH results, dependencies between time steps for CFD results, and dependencies between simulations for Crash results. Moreover, for these results, we show the relation between the required precision for reconstruction and its compression factor.
Data Compression, Simulation results, NVH, CFD, Crash
26/10/2021
09:35 conference time (CEST, Berlin)
CFD Simulation of a Vehicle Driving in Snow
26/10/2021 09:35 conference time (CEST, Berlin)
Room: A
D. Bäder (Audi AG, DEU); A. Oliva, P. Kolar (AVL, DEU)
D. Bäder (Audi AG, DEU); A. Oliva, P. Kolar (AVL, DEU)
Motor vehicles are designed for different purposes. For the development of motor vehicles it is essential that the vehicles function without problems in winter. For this purpose, winter test drives are carried out in elaborate trials in order to determine, for example, how much snow is deposited on which structural components in the wheel arch. For the tests, ready-to-run vehicles must be available, which are often only available at a late stage of the development process and are usually very expensive to purchase. In this publication, a 3D CFD calculation method is presented. The calculation method is based on a SPH method, with which the snow entry into the wheelhouse is simulated. First, the simulated snow in the form of wet snow is validated on simple experiments. Then, the validated snow parameters are used to calculate the snow entry into the wheelhouse of a real, complex vehicle. In a first step the aerodynamics is neglected and the simulation is focused on purely the snow phase, i.e. a very slow driving speed. In a second step, the influence of airflow on snow input is investigated. Herefore a higher speed of the car is assumed. The results of the snow entry with and without the influence of the airflow field are compared and a discussion is given. Finally the simulation results are validated with real experiments. Herefore an Audi Q7 was tested on an ice field covered with new snow. Different cameras attached to the car have filmed the snow entry into the wheel arch. Important components of the vehicle, like for example the tires, have been 3D scanned in order to set up a simulation where the geometry is almost identical to the tested real car. A comparison of the experimental and numerical results is presented. Overall a CFD setup is presented that is able to simulate wet snow on a complex real car wheel arch geometry.
Snow, Simulation, CFD, SPH, Automotive, Vehicle
26/10/2021
15:35 conference time (CEST, Berlin)
Transforming the CAE Result Sharing Process for Automotive Powertrain Mounting Brackets with Automation and 3D Digital Reports
26/10/2021 15:35 conference time (CEST, Berlin)
Room: K
J. Liu, R. Stec ((Ford Motor Company, USA); P. Mandava, S. Brown (Visual Collaboration Technologies, Inc., USA)
J. Liu, R. Stec ((Ford Motor Company, USA); P. Mandava, S. Brown (Visual Collaboration Technologies, Inc., USA)
The competition for the design and manufacturing of the best-in-class automotive products is more intense than ever due to technological advances and increased customer expectations. As a result, companies across the automotive industry are looking for innovations to improve their product design and manufacturing cycle time, cost and quality, and CAE simulation innovations and their effective implementations have long been proven to be a key competitive advantage. CAE organizations are also facing pressure to become more competitive and efficient. CAE Management and technical experts must constantly address an ever-growing demand to deliver more analytical product design and manufacturing evaluations with little or no headcount increase, thus driving them to create and deploy more automated and simplified CAE methods, tools and processes. Many steps in the “end-to-end” simulation processes in using the commercially available CAE tools have been streamlined or automated, with the notable exception of post-processing and report generation. Today, post-processing & report generation are time-consuming due to the involvements of various manual processes and manipulations, and it consumes about 15-25% of highly skilled FEA analyst's time and yet often yields low-fidelity 2D reports to be shared. The purpose of this paper is to explore and demonstrate how CAE report automation using VCollab® - a report automation tool - and its associated 3D "Digital CAE Reports" will enable CAE analysts to speed up and streamline the way they process and share simulation results with the product design and manufacturing stakeholders. In addition, as compared to traditional 2D CAE reports, 3D "Digital CAE Reports" will allow CAE analysts to share a richer, clearer, interactive and animated report to enable engineers and the management to make better and more confident design and manufacturing decisions.
Keywords: Simulation – Automation - CAE Reports – Collaboration – Post-processing – Design Decisions
26/10/2021
15:35 conference time (CEST, Berlin)
Methodology for Automated Designing of Composite Structures in Suspension Systems
26/10/2021 15:35 conference time (CEST, Berlin)
Room: M
T. Grünheid, R.Sturm, O. Deißer (DLR - Deutsches Zentrum für Luft- und Raumfahrt, DEU)
T. Grünheid, R.Sturm, O. Deißer (DLR - Deutsches Zentrum für Luft- und Raumfahrt, DEU)
Composite materials in automotive structures have the potential to reduce CO2 emissions by simultaneously saving weight and improving fatigue performance. Further weight reduction potential can be obtained by integrating additional functions into composite structures. Regarding a suspension system, the control arms and coil springs in a McPherson front axle can be substituted by a transverse composite leaf spring. Designing this composite leaf spring structure with wheel controlling functionality is extremely challenging due to the high number of design parameters influencing the geometry and the laminate layer setup. In particular, large deformations due to bending of the control arm during wheel travel while preserving longitudinal and lateral stiffness and strength complicates the finding of solutions. Meta-model-based optimization methods can be used to identify solutions in the complex design space. Therefore, automation in the modelling process is essential. In the Next Generation Car project (NGC) of the German Aerospace Center (DLR), new development methods are investigated for automated designing and dimensioning by using optimization algorithms. The programmed process chain is connecting CAD, finite element and multi body simulation software is controlled by optimization software. All obtained displacements are cross-checked with the reference trajectory. The parametric CAD model and layer setup is controlled by meta-model-based optimization software. The resulting geometry is meshed and an implicit finite element model is created to directly identify the structural response. The design space of the optimizer includes different geometric designs of the composite structure and each layer of the laminate which are sized according to the load cases. To ensure elastic deformation, the Tsai-Wu criterion is applied as an optimization constraint. The first results show that the selected optimization approach has a high potential for finding structures which can achieve the desired wheel travel. The investigated process chain creates transverse leaf spring geometries with reasonable layer sequences.
Composite leaf spring ; Meta-model ; Optimization ; Chassis ; Suspension
27/10/2021
08:35 conference time (CEST, Berlin)
What Makes the Integration of an SDM Successful? The Journey and Experiences of an Automotive Supplier
27/10/2021 08:35 conference time (CEST, Berlin)
Room: K
M. Tupy (Brose Fahrzeugteile GmbH & Co. KG, DEU)
M. Tupy (Brose Fahrzeugteile GmbH & Co. KG, DEU)
The successful introduction of a simulation data management system is a complex matter and is subject to a large number of influencing variables and dependencies. Due to the time horizon, some of these variables are in a perpetual state of change and determine the course of the integration. We divided the implementation into four phases. Initially, the needs and potentials of the business units were determined and, based on these, an appropriate SDM tool selection was made in the next step. In phase three, the SDM software was piloted in the working environment and is now being rolled out worldwide. The capability for integration into the existing and future IT landscape, including interfaces to other data management systems, plays a central role in ensuring end-to-end data flows and transparency while reducing manual and administrative efforts to an absolute minimum. This frees up capacities for value-adding activities. In the course of digitalization, automation and connectivity, it is important to question existing CAE processes and adapt them to the new circumstances and align them across teams and business units. Here, internal support and supervision is essential to ensure process understanding and fast maintainable implementations. From the author's point of view, however, the inclusion of users in the change process is the most important building block and the decisive factor for success or failure. While the previously mentioned topics are highly related to the SDM tool, this factor is to be considered absolutely tool-independent. Here, the consultation and active involvement of the users in every phase is decisive. The active involvement of CAE engineers as process designers, pilot users, but also as training providers has a convincing effect and serves as a multiplier for the entire SDM world. The author would like to thank :em engineering methods AG, GNS Systems GmbH and PDTec AG for their support on this journey.
27/10/2021
08:55 conference time (CEST, Berlin)
Hood Fluttering Caused by Unsteady Aerodynamic Loads by On Route Vehicles’ Interaction
27/10/2021 08:55 conference time (CEST, Berlin)
Room: C
A. Pérez Peña (ESI Spain, ESP); Á. Segura Santillana, J. Comas Font,T. Angulo de Diego, V. Cermeño Escobar (SEAT SA. ESP); R. Almenar (ESI, DEU)
A. Pérez Peña (ESI Spain, ESP); Á. Segura Santillana, J. Comas Font,T. Angulo de Diego, V. Cermeño Escobar (SEAT SA. ESP); R. Almenar (ESI, DEU)
Over the last few years, automotive manufacturers have been optimizing the weight of the vehicles by using lighter materials and reducing the thickness of sheet metal panels. As a result of the reduction in thickness, deformation of panels under standard loads has increased in many cases. These higher deformations do not mean that the vehicle quality is lower in terms of functional performance, however they are perceived by the Customers as poor product quality and should be avoided. One of these “standard load cases” appears when one car overtakes another vehicle (truck, pick-up or other). In this case, the turbulent structures in the wake of the car in-front reach the overtaking car, causing time- and space-variant pressure oscillations on the panels which cause vibrations. In the case of the hood, these vibrations can be of several millimeters at the trailing edge and may be visible from the driver point of view. Such issues are typically detected at the late testing stages, leading to costly design improvements. These effects can be virtually predicted by automated chained simulations easy to set up to allow its assessment and necessary design changes along the vehicle development iterations. In collaboration between SEAT and ESI Group, a computational methodology has been developed to detect such issues early in the vehicle engineering process, chaining Computational Fluid Dynamics (OpenFOAM) for the aerodynamic predictions and Finite Elements (Virtual Performance Solution - VPS) for the structural predictions. The application of this process with virtual prototypes enables engineers to detect such problems long before the actual physical prototypes are built and propose solutions to minimize the impact and avoid low perceived quality by the Customers. This paper describes the application of the developed methodology in detecting the vibrations in the early design stage and proposing countermeasures. The process showed in this paper opens the door to many other load cases, in which combinatorial CAE analyses are relevant, that can be evaluated virtually to improve the vehicle quality impacting the design in early stages of the development.
Aerodynamics, Overtaking Vehicles, Fluttering, Lightweight, Computational Fluid Dynamics, Finite Element, Virtual Prototypes
27/10/2021
11:00 conference time (CEST, Berlin)
Two-Way Coupled Thermal-Electric Simulation of a Packaged Laserdiode using Reduced Order Models
27/10/2021 11:00 conference time (CEST, Berlin)
Room: C
T. Moldaschl, G. Grosso (SAL Silicon Austria Labs GmbH, AUT); R. Fuger (CADFEM ,AUT)
T. Moldaschl, G. Grosso (SAL Silicon Austria Labs GmbH, AUT); R. Fuger (CADFEM ,AUT)
Laser diodes have many applications in different areas of research and technology and underly different working principles. Semiconductor Lasers are of most importance as they can be produced using established semiconductor manufacturing processes. A Quantum Well (QW) Laser Diode is used in this report as a powerful light source for automotive LIDAR applications. Especially the extended automotive environment conditions require a special selection of elements and materials and require specific cooling solutions. In order to evaluate a LIDAR system with its high power densities the Laser diode is simulated both electrically and thermally to find crucial elements that are electrical and thermal bottlenecks. The electrical System is analyzed with Ansys Maxwell and the thermal system with Ansys Mechanical and they are both coupled through a transient Ansys TwinBuilder simulation. Both systems comprise the same geometry, a fourfold bonded edge emitting QW Laser with connection pads in a molded package. Each of the four Lasers on a single chip can be addressed separately and is connected through a gold bond wire and a copper pin. In order to obtain 2-way coupling, electrical losses from the Maxwell simulation are coupled into a transient thermal simulation using Mechanical. The resulting temperatures are then fed back into the electrical simulation, where the conductive copper has been assumed to be temperature dependent, and thus, also the losses are temperature dependent. Each simulation in either Maxwell and Mechanical requires a separate FEM simulation to obtain the losses or the temperatures. To obtain coupling in this way would require a very long simulation time for a two-way coupling scheme. In order to speed up the process the electrical simulation is transformed into reduced order model (ROM) using a functional mockup unit comprising a sufficiently large input parameter space and a thermal LTI ROM system using all separate loss inputs. For each parameter combination in both Maxwell and Mechanical a complete FEM simulation must be run. But once the data is available, the ROMs are created and a transient simulation in TwinBuilder can be run with significant reduction of simulation time. We present simulation results of this system using only ROMs, compared to full FEM simulations and discuss the advantages and implications of using only ROMs.
reduced order models, FEM Simulation, Ansys Maxwell, Ansys Mechanical, Coupled Multi-Physics Simulation
27/10/2021
13:20 conference time (CEST, Berlin)
Simplification of FE-Crash-Models for Optimization of Vehicle Structures
27/10/2021 13:20 conference time (CEST, Berlin)
Room: A
M. Schäffer (DLR - Deutsches Zentrum für Luft- und Raumfahrt, DEU); M. Totzke, R. Sturm (German Aerospace Center, DEU)
M. Schäffer (DLR - Deutsches Zentrum für Luft- und Raumfahrt, DEU); M. Totzke, R. Sturm (German Aerospace Center, DEU)
New simplification strategies for crash simulations are developed at the German Aerospace Center (DLR) as part of the Next Generation Car META-project. The computational costs required for the various crash load cases are challenging for optimization of vehicle structures. In this paper an automated model generation for simplified vehicle crash models is presented by using global deformation characteristics of structures which are obtained from the global crash model. Due to the obtained reduction of computational costs, structural crash optimizations can be performed. The development of a light and crash safe body-in-white structure is currently supported by lots of crash simulations. As part of these developments the increased application of virtual methods the product development process has been more than halved leading to shorter product life cycles. [1], [2] However, due to the high computational costs required for crash simulations, structural optimizations cannot be carried out with a full vehicle model. Therefore, there is a strong need for an automated generation of simplified vehicle crash models to include structural optimizations into the shorted product development process. For reduction of computational time, an automated four-step approach for the generation of a simplified vehicle model was developed in [3] and [4]. For the verification of the automated approach, crash simulations were carried out investigating the US NCAP (100%, 56 km/h) crash load case. In this load case, the vehicle is driven against a rigid wall at a speed of 56 km/h. The crash model of the Toyota Yaris [5] was taken, since the available crash model is validated for this crash load case. In comparing the global deformation of the simplified vehicle model and full vehicle model no major differences can be identified. Besides, the same overall crash kinematics between the models, also the same buckling and folding of the bonnet, deformation of the exhaust system and deformation to the chassis is obtained for the two models. The main target of simplified crash models is the reduction of computational time with a minimum influence on the simulation results. For the assessment of the time reduction potentials, all crash simulations were performed on the same CASE-2 Cluster on 24 CPUs. Through the application of the modelling approach, the number of components in the model could be reduced by 50%, and the computational time can be reduced by approximately 65% in this load case. REFERENCES [1] Duddeck, F.: Multidisciplinary optimization of car bodies. https://doi. org/10.1007/s00158- 007-0130-6; Struct. Multidisc. Optim. 35, 375–389, 2008. [2] Klaiber, M.: Use of Innovative 3D Printing Technologies for Flexible Process Chaining, 2. Technologietag Hybrider Leichtbau, Stuttgart, 2015. [3] Schäffer M, et. al, Methodological approach for reducing computational costs of vehicle frontal crashworthiness analysis by using simplified structural modelling; https://doi.org/ 10.1080/13588265.2017.1389631; International journal of Crashworthiness, 2017. [4] Schäffer M, et. al, Automated generation of physical surrogate vehicle models for crash optimization; https://doi.org/10.1007/s10999-018-9407-8; Int J Mech Mater Des, 15, 43-60, 2019. [5] NCAC, „Extended Validation of the Finite Element Model for the 2010 Toyota Yaris Passenger Sedan“. The George Washington University, 2012.
crashworthiness, safety, optimization, simplification strategies
27/10/2021
13:20 conference time (CEST, Berlin)
Pressure Drop and Thermal Field Prediction of Car Heat Exchangers Using CFD Submodeling Techniques
27/10/2021 13:20 conference time (CEST, Berlin)
Room: L
T. Płusa, N-Y.Francois (Valeo Thermal Systems, FRA)
T. Płusa, N-Y.Francois (Valeo Thermal Systems, FRA)
In recent years, R&D has undergone a real transformation and is being rationalized in order to increase its efficiency, reduce its development cycles and costs, and improve its capacity to innovate. Turned towards digitalization, Valeo Thermal Systems relies in particular on digital simulation such as CFD (Computational Fluid Dynamics) to optimize concepts, orient its technical choices, make decisions and engage with customers. Its use is becoming a major asset to offer innovative and competitive automotive heat exchangers. Cost of computational power decreases from year to year and it causes resources to be more accessible for different entities like: corporations, research institutes, universities, smaller companies etc... Despite this fact, car heat exchangers like radiators or intercoolers cooled by water or air, are still very challenging in terms of simulation. These heat exchangers have hundreds or even thousands of very small tiny parts called turbulators or fins increasing the area of heat exchange in a unit of volume. It causes that it is nearly impossible in reasonable time to mesh and simulate detailed geometry. As a simplification, special replacement models are used. Pressure drop is calculated using the so-called porous medium while heat transfer is carried out by the 3D heat exchanger interface model. Both models need input like porous medium pressure drop coefficients or local heat transfer coefficients. This article presents a new numerical method based on a CFD submodeling technique to characterize the pressure losses and the heat transfer coefficients from a small periodic pattern of the heat exchanger. Then, based on CFD submodeling results, special laws are built using dimensionless numbers such as Reynolds number, Nusselt number, Prandtl and Bejan number. Finally, these laws of pressure drop and heat transfer are applied to a full CFD model of the heat exchanger to predict its fields of temperature and flow with its performances.
heat exchangers, heat transfer, pressure drop, cfd, submodeling
27/10/2021
13:20 conference time (CEST, Berlin)
Analysis of Numerical Crash Simulation Data Using Dimensionality Reduction and Machine Learning
27/10/2021 13:20 conference time (CEST, Berlin)
Room: M
N. Ballal, M. Dlugosch (Fraunhofer EMI, DEU)
N. Ballal, M. Dlugosch (Fraunhofer EMI, DEU)
With the increasing virtualization of automotive R&D processes, analyzing the growing amounts of numerical simulation data produced is becoming more and more challenging. A considerable amount of resources are spent to extract underlying knowledge about the crash behavior under certain input parameters. This research proposes data science methods to semiautomatically analyze numerical simulation data. The goals of this research include comparing different dimensionality reduction algorithms to represent simulation data as lower-dimensional embeddings, clustering algorithms to cluster simulations displaying similar crash behavioral patterns, finding causes for a certain behavioral pattern, and discovering design rules to avoid undesired behavioral patterns. The lightweight lower-dimensional embedding of the simulations is represented using feature extraction dimensionality reduction methods like Principal Component Analysis (PCA) and t-distributed stochastic neighbor embedding (t-SNE). The simulations’ underlying knowledge is extracted from the lower-dimensional embedding and similar simulations are clustered based on their behavioral patterns by using unsupervised clustering algorithms like k-means, Balanced Iterative Reducing and clustering using Hierarchies (BIRCH). These behavioral patterns can be analyzed and the importance of the input parameters causing a certain behavioral pattern is obtained using random forests. Finally, the design rules to avoid an undesired behavioral pattern are extracted using decision tree algorithms and associative rule mining. The workflow is applied to a simple side pole impact simulation model, where a generic vehicle floor structure is impacted by a pole. To create a dataset, extensive simulations are carried out by varying the values of the input parameters using a Latin-Hypercube DoE scheme. Best results were obtained using a combination of a linear dimensionality reduction algorithm and a hybrid clustering algorithm, which yielded three different behavioral patterns containing 90 % of the original information respresented in 50 dimensions. The obtained results from rule extraction confirm the rules anticipated by domain experts to avoid a non-desired cluster. These design rules can assist engineers in efficiently defining successful future designs.
Automotive, Crash Simulation, Data Analysis, Machine Learning, Dimensionality Reduction
27/10/2021
13:40 conference time (CEST, Berlin)
Application of Physical and Mathematical Surrogate Models to Optimize the Crashworthiness of Vehicle Front Structures
27/10/2021 13:40 conference time (CEST, Berlin)
Room: A
P. Lualdi (DLR - Deutsches Zentrum für Luft- und Raumfahrt, DEU); M. Schäffer, R. Sturm (German Aerospace Center (DLR), DEU)
P. Lualdi (DLR - Deutsches Zentrum für Luft- und Raumfahrt, DEU); M. Schäffer, R. Sturm (German Aerospace Center (DLR), DEU)
Exploiting the full potential of lightweight materials and weight reduction techniques while preserving the high safety standards of a vehicle still remains one of the major challenges in the field of crashworthiness design. Such a problem, already challenging from a qualitative point of view, is also hard to be solved numerically due to the computational costs related to crash simulations of full FEM vehicle models. This factor represents also a real limitation in the field of optimization of vehicle crash structures. It is especially challenging, if load cases are investigated, which additionally involve complex impactor models such as deformable impact barriers, which are required for the certification of vehicles. Only by the use of simplification strategies and physical surrogate models the computational costs can be reduced in a way to make crashworthiness problems for cars suitable for structural optimization. Additionally, the application of mathematical surrogate models can efficiently improve the possibility to find suitable structural solutions in the respect of given design requirements compared to conventional optimization approaches. In this paper, the Urban Modular Vehicle (UMV), i.e. a battery-driven modular car concept developed by the German Aerospace Center (DLR) is investigated to improve its crash performance in the event of a front crash using physical and mathematical surrogate models. The NCAP - Mobile Progressive Deformable Barrier (MPDB) crash load case is considered as frontal crash scenario. A physical surrogate model will be used to replace the detailed crushing behavior of the MPDB model by means of kinematic descriptions. The finite element vehicle model is also simplified in a way that structural components which are assumed to only bear elastic deformations are replaced by kinematic numerical representations describing the substituted structures. Finally, a design optimization strategy based on mathematical surrogate models is successfully applied to optimize the absorption properties of the crash relevant vehicle structures. For the optimization appropriate constraint functions are included to ensure that any applied structural change does not undermine the safety requirements of the passengers. Only with these applied simplification strategies optimization can be applied to such a complex crash load case.
crashworthiness design, design optimization, surrogate models, structural optimization
27/10/2021
13:40 conference time (CEST, Berlin)
Simulation of Alpha Cabin Reverberent Room to Estimate Absorption Coefficient under Diffuse Sound Field
27/10/2021 13:40 conference time (CEST, Berlin)
Room: J
X. Robin, E. Richard, M. Raskin, T. Poulos (Hexagon, BEL)
X. Robin, E. Richard, M. Raskin, T. Poulos (Hexagon, BEL)
The acoustic efficiency of components used in the automotive industry for noise insulation can be characterized by their absorption coefficient. It can be measured either based on the normal acoustic incidence or a diffuse field. The diffuse field approach is more realistic for the car interior since the sound field inside the car is diffuse. Automotive OEMs and suppliers are usually applying experimental approaches to determine the absorption coefficient of a given porous material utilizing setups that act as small reverberant rooms called “alpha cabins”. This paper discusses a methodology based on a frequency-domain Finite Element (FE) method to simulate an alpha cabin in the range of 400-10000 Hz and determine the absorption coefficient in a computationally efficient way. To overcome computational challenges emanating from the size of the cabin, an innovative three-step approach is proposed. Below the Schroeder frequency (~1500 Hz), where the alpha cabin is not large enough to be considered as reverberant, the absorption coefficient is determined in the same way as in the experimental approach. This involves the reconstruction of the time-domain signal from the response spectrum on a series of microphones for estimating the sound-decay values inside the cabin (RT60). For the mid-range frequencies, between 1500 and 5000 Hz, energetical quantities calculated via the finite element method in the fluid and porous parts of the setup are used to determine the RT60 values without the need of time-domain reconstruction. Finally, at high frequencies, between 5000 and 10000 Hz, the model is scaled so as to reduce its size while still keeping the cabin sufficiently diffuse, and the same post-processing method as in the mid-frequency range is used. The results from this methodology are compared with previously published results in order to evaluate the viability of the method, which provides the possibility to reduce expensive experimental work.
alpha cabin, trim, porous material, absorption coefficient, diffuse field, finite element
27/10/2021
13:40 conference time (CEST, Berlin)
A Streamlined Approach to Couple In-cylinder and Conjugate Heat Transfer Engine Models
27/10/2021 13:40 conference time (CEST, Berlin)
Room: L
J. Fernandes, W. Seeley (Siemens Digital Industries Software, GBR)
J. Fernandes, W. Seeley (Siemens Digital Industries Software, GBR)
Increasing regulatory and commercial requirements have led to widespread adoption of digital analysis in the IC engine design process. This has helped to improve fuel efficiency, reduce emissions, as well as reduce reliance on physical testing and the development of costly prototypes. CFD is one of the key types of analysis employed in generating a digital twin of the IC engine. It is it able to model a wide range of physical processes that include gas and coolant flow, combustion, emissions and conjugate heat transfer (CHT). When using CFD to evaluate the thermal field of engine components, the most common approach is to generate separate models for the gas flow and combustion (the in-cylinder model), and the coolant flow and metal temperatures (the CHT model). The in-cylinder model is used to estimate spatially varying heat flux which are averaged over a cycle and then applied as thermal boundary conditions to the CHT model. Spatially varying temperatures on the combustion boundaries from the CHT simulation can be then passed back to the in-cylinder simulation and data exchanged between the two simulations until convergence is achieved. Traditionally the in-cylinder and CHT models have been generated using separate software packages. Within the digital environment, this has created a few challenges relating to efficient and accurate data exchange between the two models as well as overall process automation. This paper presents a coupled in-cylinder/CHT approach carried out using a single software package that employs state of the art physics models for both the in-cylinder and CHT analysis. It highlights how integrating both analyses into a single user environment can help to make the process more streamlined, efficient and automated. The approach is applied on a 4-stroke gasoline engine and comparison of the predicted in-cylinder combustion performance and component temperatures are made with experimental measurements.
CFD, ICE, In-cylinder, Conjugate Heat Transfer
27/10/2021
14:00 conference time (CEST, Berlin)
Analytical Prediction of Whoosh Noise & Blade Pass Frequency Noise at AIS Orifice
27/10/2021 14:00 conference time (CEST, Berlin)
Room: J
S. Mishra, A. Karim (Ford Motor Company, USA)
S. Mishra, A. Karim (Ford Motor Company, USA)
During acceleration of a vehicle the turbocharger operates very close to the surge line. At this operating regime of the turbocharger, the pressure ratio increases substantially in comparison to the increase in airflow. This leads to a NVH error state generically known as whoosh noise which is a broadband noise between 4000 Hz to 11000 Hz. A test for whoosh noise is usually performed late on the design stages and usually performed on the vehicle. This late testing in the design phase, increases risk of late changes if whoosh is detected. This also inhibits any design changes that could have mitigated the issue and requiring to take only palliative measures such as silencers or damping pads. These palliative measures to contain whoosh noise can be significantly expensive. Therefore, an analytical method becomes more desirable that could potentially save cost if whoosh noise is predicted early in the design process. These in-vehicle tests are performed with a microphone at the air induction system orifice location. So, the analytical model would not only need to predict the source accurately, but also it must predict the propagation of acoustic energy through the AIS without numerical loss to predict sound pressure level at the orifice location accurately. This CFD investigation aims to achieve the above; asses sound pressure level at different location in the air induction system including at the orifice. The results are then validated with actual gas turbine lab test results. The CFD uses transient aeroacoustics (CAA) along with rigid body motion to assess sound pressure level. The simulation demands a high level of spatial and temporal accuracy because the speed of rotation of turbocharger could be more than 100,000 RPM and to minimize numerical loss. The current investigation takes 6 weeks of CPU time on 384 Processors. The software StarCCM+ is used for CFD analyses & visual post-processing and LMS Testlab is used for NVH post-processing.
CFD, LES, Whoosh, blade pass frequency, nvh, aeroacoustic, turbocharger, caa
27/10/2021
14:20 conference time (CEST, Berlin)
A Reduced Order Model Optimization Method for Spot-weld Position and Vehicle Structure Crash Performance
27/10/2021 14:20 conference time (CEST, Berlin)
Room: A
M. Couste (Renault Technocentre, FRA); Y. Tourbier (Groupe Renault, FRA)
M. Couste (Renault Technocentre, FRA); Y. Tourbier (Groupe Renault, FRA)
In automotive industry, optimisation on vehicle structure has been a constant focus in projects. The goal is to find the best compromise between mass, cost, and performances like crash, acoustics, overall stiffness... The structure of the vehicle body is made up of 200 metal parts mainly joined with thousands of spot-welds. Each performance is simulated with increasingly accurate finite element model to reduce the number of physical tests that are still time-consuming and expensive. Despite computational power are in constant progression, simulations remain time challenging to be used in optimisation studies. For example, crash simulation takes around ten hours on HPC. Groupe RENAULT has a rich experience in the use of design of experiments (DOE) for vehicle optimisation. Furthermore, DOE will be completed or replaced with Model Order Reduction (MOR) to improve efficiency, duration, and cost. “Classical” MOR allow to lead two-steps optimisation studies: first a lot of simulations to create an abacus, then uses it to quickly determine the results for new sets of parameters. However, our studies vary tens of parameters, so classical MOR is as costly as DOE. Therefore, RENAULT developed the Regression-CUR (ReCUR) method to reduce the number of simulations. The parametrized reduced order model gives an estimation of the high-fidelity result for a new set of parameters. The accuracy of the reduced model can be balanced with less simulations. Indeed, an overall improved design is searched instead of demonstrating the global optimality. The article will focus on a combinatorial optimisation problem by finding the right balance between assembly process and crash performance (with ReCUR). The first part treats about the optimisation of the main crash criterion from certification rules and media scoring derived from structural deformations. The second part deals with the optimisation of the spot-weld risk requirements used to avoid successive spot-welds failure by applying on a simple test case and on a side-impact industrial case.
optimization, model order reduction, crashworthiness, machine learning, vehicle body
27/10/2021
17:35 conference time (CEST, Berlin)
Using Superelement Approach to Improve Automotive FEA
27/10/2021 17:35 conference time (CEST, Berlin)
Room: A
E. Czerlunczakiewicz (Valeo, POL)
E. Czerlunczakiewicz (Valeo, POL)
As product development time in the automotive industry becomes shorter, the customers require more rapid and accurate FEA results. Simultaneously, FE models are frequently getting larger and more complex, in order to represent the engineering problem closely. It is becoming more and more challenging to meet these requirements. One of the possible solutions, to avoid a significant increase of computing resources and calculation time without losing accuracy in the results of the analyzed fields, is substructuring. Authors try to show that implementation of superelements in FE models is neither more complicated nor too time consuming in comparison to achievable benefits. To demonstrate the validity of superelement usage some practical examples, from the automotive industry, are gathered in this paper. In the presented cases, substructures improved FE results and in some, helped to meet project timing deadlines. Particular case studies are presented in the area of linear dynamics. Substructure replaced parts that are not in the area of FEA engineer interest and are not covered in the data post-processing. However, these components should not be skipped for reflecting real physical behavior of the analysed system. Applying superelements in linear dynamic analysis allowed to obtain better results without significantly increasing computing resources. A part of the paper is dedicated to handling substructures for optimization purposes. Optimization step is an iterative process, and in a lot of cases to gain acceptable results, hundreds or thousands of increments have to be calculated. Calculating a high complexity model with numerous variables can result in an unsolvable task. Fortunately, with implementing superelements, it is possible to focus only on components to be optimized while taking into account global model behavior. The last part of the presentation will present an example of new opportunities to use superelements in our applications.
FEA, superelement, substructure, structural simulations, optimization, linear dynamic, CAE
27/10/2021
17:55 conference time (CEST, Berlin)
Co-simulation of Semi-active Dampers for Durability Road Load Simulations
27/10/2021 17:55 conference time (CEST, Berlin)
Room: A
H. Kolera-Gokula (MSC Software, USA); J. Zakrisson (Volvo, SWE); T. Nygards (Hexagon Manufacturing Intelligience, SWE)
H. Kolera-Gokula (MSC Software, USA); J. Zakrisson (Volvo, SWE); T. Nygards (Hexagon Manufacturing Intelligience, SWE)
Active chassis systems are becoming the norm today, especially in the premium car segments. These systems provide the ability to adapt to varying road conditions. It is critical to have CAE methods that model active systems since they significantly influence the accuracy of computed durability road loads used in vehicle design. There is a certain amount of complexity associated with creating simulation models of active vehicle systems. To enhance prediction accuracy, full-fidelity models of the controls guiding the active systems have to be included. These control models are often delivered as black boxes from the supplier. The first part of the presentation outlines the CAE strategy for active systems at Volvo Cars Durability Centre, with a particular focus on co-simulation of semi-active dampers. The co-simulation strategy forms the basis for simulation-led predictions of strength, endurance, durability, and road loads characteristics of vehicles with active chassis systems. The co-simulation approach is compared and contrasted against a simpler approach involving passive simulations. The incremental improvement in the accuracy of the computed road loads using the co-simulation approach is detailed under various road conditions and vehicle events. These include single-sided vertical impact, driving through a pothole, drop-off rebound, and the Belgian Pave. The second part of the presentation describes the implementation of a customized simulation process at Volvo Cars Durability Centre that enables efficient co-simulation, at scale, with minimal computational overhead. Processes and methods have been developed previously at Volvo Cars Durability Centre to automatically submit a whole suite of fatigue simulation events. Still, for co-simulation methods, the only option had been primarily manual. A customized process enhancement is presented in the new implementation, which seamlessly integrates cross-platform co-simulation into Volvo Cars' current automated simulation processes with minimal additional computational overhead. This process implementation will serve as a framework for similar co-simulation implementations in the future at Volvo.
Multi Body Dynamics, MBSE, Durability,Co-Simulation, Systems Modeling
27/10/2021
17:55 conference time (CEST, Berlin)
Simulation of Automotive in the New Context of Validations
27/10/2021 17:55 conference time (CEST, Berlin)
Room: B
E. Landel (ELC, FRA)
E. Landel (ELC, FRA)
Validation and certification of vehicle tend to be extended to numerous situations, in a way to ensure the robustness of critical performances as safety, fuel consumption or autonomy for electrical vehicle. The diversity of vehicle (engine, chassis, various options) must be checked to ensure a good Representative of evaluation for all individual vehicle. The scope of this kind of validation plan become huge. The classical way using testing on proving ground or driving on open road become not adapted regarding cost and time. So, simulation is introduced to be combined with testing activities To perform an efficient validation plan, the fidelity of models of vehicle need to be controlled to ensure a high level, roughly equivalent to testing results. A vehicle is an assembly of many parts and complex sub-systems: powertrain, ADAS systems, thermal management, … These subsystems are made of mechanical parts and controllers with exchanges of information between them. The architecture is complex in a sense that many connections have to be taken into account to provide a realistic description of the vehicle. At the same time, many teams are involved in the development of each system. Developing a model at high level of fidelity request a new organization of the activities of modeling. First, a new actor must be introduced: “simulation architect”. He will be responsible of defining the needs of simulation, designing the model of the vehicle and sending requests of models for all the parts and systems. He will used new methods and tools for developing the models of the vehicle by transformation of the initial model defined in the tools of MBSE. Metamodel like Model Identity Cars (MIC) will ease the communication with model providers. This is very important to ensure delivery of elementary models at the right time and with the right level of fidelity.
automotive, validation, systems, complex, architecture, fidelity
27/10/2021
18:15 conference time (CEST, Berlin)
Design Exploration and Prediction of Automotive Hood Designs based on Non-Uniform Feature Parameters
27/10/2021 18:15 conference time (CEST, Berlin)
Room: A
S. Ramnath, A. Li, J. Shah (The Ohio State University, USA); D. Detwiler (Honda R&D Americas, USA)
S. Ramnath, A. Li, J. Shah (The Ohio State University, USA); D. Detwiler (Honda R&D Americas, USA)
The design of automotive structures is a multi-objective problem that includes, light weighting, manufacturability and overall performance. Light weighting, while not compromising crash worthiness, requires careful placement of features in automotive body components. The current design process of automotive hoods involve the use of designs from previous generations as benchmark upon which new designs are evolved. The design obtained from this evolution process must meet the required performance targets. However, in this process of design by evolution, it becomes impossible to generate new designs or adapt design ideas from other models. In addition, the lack of information on performance for a set of mix and match of features from other designs, make it difficult to adapt features by cross designing. There is a gap needed to be filled to assist designers with cross designing of hood models so that the features from one hood can be used on another. The uniqueness of each design and presence of non-uniform parameters makes it difficult to compare two or more designs and extract useful feature information. It is necessary to use unconventional methods to compare the performance and pick the best suitable design. This paper aims to fill this gap by introducing an innovative approach to use a non-uniform parametric study for design exploration in order to make valuable suggestions to the designer. The proposed method uses data sets produced from finite element analysis (FEA), for a given set of loads. Based on designer preference, the response data generated from this FEA can be processed in three ways: 1) analyze for a specific hood model 2) analyze for a larger set that includes features from multiple hoods at the same time 3) analyze based on specific hood attributes (area, curvature, etc.) instead of individual feature parameters. The final predictions will provide the designer with parameterized surface models with potentially new designs adapted from a range of models. This method can be extended to other components and domains that use feature-based parametric designs.
Design Exploration, Design Automation, Feature based Design
27/10/2021
18:15 conference time (CEST, Berlin)
Application of Machine Learning and CFD to Model the Flow in an Internal Combustion Engine
27/10/2021 18:15 conference time (CEST, Berlin)
Room: M
J. Hodges (Siemens Digital Industries Software, USA); M. Emmanuelli, S. Sathyanandha (Monolith AI, GBR); J. Fernandes (Siemens Digital Industries Software, GBR)
J. Hodges (Siemens Digital Industries Software, USA); M. Emmanuelli, S. Sathyanandha (Monolith AI, GBR); J. Fernandes (Siemens Digital Industries Software, GBR)
As in many engineering industries, production timelines for internal combustion engines are too strict to allow for full (multi-disciplinary) exploration of design permutations through large volumes of simulation and physical test. This study combines machine learning and CFD simulation for accelerated and intelligent design of an internal combustion engine (ICE) to accommodate such a challenge. The specimen investigated is a parameterized cylinder port design, in a 4-stroke gasoline engine, which whereby a number of simulations are generated to partially cover the design space. The focus is an inlet port design which creates favorable developments in the turbulent flow-field for more ideal combustion. With such simulation data generated, neural networks are created to capture the relationship between the design parameters and the performance results (in 1D, 2D, and 3D). For the one-dimensional data, predictions are made for the transient evolution of important scalar performance metrics over an engine cycle, such as turbulent kinetic energy, tumble, and other thermodynamic variables. For the two-dimensional data, predictions are made for local values, similar to the one-dimensional data predictions, in the center plane of the cylinder. Since the design objective is focused on the turbulent flow-field, the three-dimensional data predictions focus on predicting the turbulent kinetic energy in the highly turbulent sections of the flow-field. These predictions prove to be quite accurate and reveal that neural networks are effective at modeling simulation data for predictive design exploration. Their mathematical structure allows them to capture highly non-linear and multi-variable physical behavior. With such a simulation-machine learning approach, design exploration with a greater concentration on more attractive designs is possible. These trained neural networks can also be used in design cycles of subsequent similar products, which could expedite early-stage design via transfer learning. To evaluate the transfer learning capabilities for this problem, the simulation data was split for training and validation in such a way that both focused on different flow-field characteristics. Specifically, the training data was comprised of simulation data with port designs that were acute, which created ‘sharp’ angles that resulted in large flow separation upon entry to the cylinder. For the validation dataset, the simulations had intake port designs which were less steep and therefore significantly different in terms of the resulting flow patterns. Since these flow patterns greatly affect the resulting turbulence and therefore the combustion behavior, it is encouraging that the neural networks were able to accurately predict the ICE simulation results and additionally provides confidence that they can provide further value throughout the design process.
Machine learning, AI, Neural Networks, CFD, ICE, Simulation, Automotive
27/10/2021
18:35 conference time (CEST, Berlin)
Intrusion Vehicle Body Optimization Combining Frontal and Side Crash Responses
27/10/2021 18:35 conference time (CEST, Berlin)
Room: A
F. Leonov S. López (LURI Engineering, MEX)
F. Leonov S. López (LURI Engineering, MEX)
Automobile Manufacturers are required to design their vehicles for safety so that the occupants will survive of a variety of crash scenarios. Computer simulated crash analysis evolved over the years to help augment the crash rest programs and to give engineers to insight into the crash events. These nonlinear simulations have become commonplace during the design phase to save time and design cost. Designing an automobile for compliance with these safety standards along with fuel efficiency standards is hard because of some contradiction between these requirements. Today, to improve the design, analysts and engineers are using analyses and coupling them with general purposes optimization packages, hence, several crash conditions simultaneously can be defined to obtain the last response of the structure and utilizing optimization tools. Crash simulations typically require a significant amount of computational time and resources, as a result, there is an important interest in using approximate models to deal with optimization processes. An approximated model is constructed from a limited set of simulations and used during the optimization to obtain an improved design. But in normal conditions, there is a limitation on the number of independent design parameters that can be used because of the accuracy of the approximate models. This research describes a design system to optimize the non-linear responses computed from a Nonlinear software and General Purpose Optimization packages using various optimization techniques, especially with large-scale (large number of design variables) optimization. With the Nonlinear package software will be implemented to perform equivalent static loads (ESL Method) based on a Nonlinear analysis responses. The optimization package will be used to optimize the structure under frontal and side crash while decreasing intrusions at nodal outputblocks. Equivalent Static load method will require multiple iterations process of non linear structural analysis and Optimization. Large scale optimization techniques, such as, sizing, topometry and topography will be implemented easily.
Sizing, Topographic, Large scale optimization, Equivalent Static Loads, Nonlinear
28/10/2021
08:55 conference time (CEST, Berlin)
Improving Collaboration in Vehicle Performance Simulation Process Through the Introduction of a System Centric Digital Thread
28/10/2021 08:55 conference time (CEST, Berlin)
Room: A
E. Mottola (Toyota Motor Europe NV/SA, BEL)
E. Mottola (Toyota Motor Europe NV/SA, BEL)
Automotive OEMs have established, over the years, standardized processes to coordinate and optimize the development schedule, with clear roles and responsibilities in each function. This constant refinement of requirements and processes has proven effective to develop attractive products for the customers. However, the increasing complexity of product development scenarios, such as Connected, Autonomous, Shared and Electrified (CASE) poses new challenges: the requirements of the systems and their interactions grow to a size which is challenging to manage with traditional approaches. New mobility business models require thinking about a vehicle not just as a product, but rather as a system within a complex system-of-systems. Model Based Development provides the framework to develop the vehicle using holistic systems thinking and to better manage the risks from that complexity using simulation for continuous exploration and validation. Recently, there has been a growing interest in technologies like co-simulation and engineering data management. However, practical industrial implementation still needs to deal with fundamental issues such as domain and subsystem "silos". Often, it is difficult to maintain up-to-date models with high quality data, validating the simulation scenarios against realistic operating conditions of the vehicle. Silos are a barrier to adopting Systems Thinking and to reusing knowledge across the enterprise. The authors propose approaches and platform characteristics that increase the smoothness of the information flow across domains, and between system designers and simulation engineers, through the realization of a custom, domain-independent, system-centric digital thread. This flexible data model provides confidence in up to date system specification data and it enables the simulation engineers to generate simulation models automatically, including variants; it can be further expanded to connect related data and processes; for example, it can trace the link between new requirements, affected systems, their evolutions, and overall product performance. Such digital thread not only enables re-use of previous data; it also provides the foundation for future data mining and AI applications to further accelerate the development process.
MBD, MBSE, Simulation, Collaboration, Digitalisation
28/10/2021
09:15 conference time (CEST, Berlin)
Central Management of Virtual Prototypes from a Lifecycle Perspective
28/10/2021 09:15 conference time (CEST, Berlin)
Room: B
M. Elbs (IPG Automotive GmbH, DEU)
M. Elbs (IPG Automotive GmbH, DEU)
This presentation illustrates an approach for establishing simulation-based development processes for software-driven functions by using CI/CD methods for virtual prototype (VP) fleets, symbiotically supporting physical prototypes. The central management of VPs from a lifecycle perspective and ensuring validity and quality are obstacles to overcome. This work demonstrates methods and tools to build an infrastructure for VPs, enabling an agile and efficient development process. The automotive industry is currently implementing the infrastructure required to increase development efficiency: Functional data (i.e., non-geometric data, NGD) and its management are in focus. These data describe attributes, characteristics or the function of components up to complete systems. For a solution that provides centralized and intuitive access to NGD from all R&D divisions, data from different CAE tools and measurements must be combined, saved, quality-tested and exchanged. Integrating existing databases simplifies the workflow and allows visualizing, downloading and searching data. Standardized metadata enable clear labeling and access to structured data, which leads to an overview of all NGD at any time during vehicle development. Achieved improvements, even using different simulation environments, become traceable and quantifiable due to the single data source. This allows the buildup and use of VP fleets enabling cross-domain cooperation regarding full vehicle testing. The collaboration of engineers from all development stages enables continuous improvement of vehicle models. VPs are used to test the latest status of ECU software with relevant vehicle variants overnight, allowing for automated testing and evaluation against quantifiable criteria. AI-driven methods support the correlation of VPs with real measurement data, thereby building broader engineer-level confidence in simulation. Combined with a unified environment, VPs can be used throughout the complete vehicle lifecycle up to validation and support (OTA) in the field. Publications: Kowalski, K.; Stehle, J.: Industrialization of the CarMaker suspension model parameterization and validation processes. In: ATZ 123 (2021), no. 1, p. 62-66 Schaffnit, J.; Holzmann, H.; Lauer, J.; Klemmer, J.: Efficient and sustainable management of vehicle models. In: ATZ 119 (2017), no. 4, p. 42-46 Niederbrucker, G.; Pfaff, A.; Donn, C.; Kochem, M.: Clouds Ahead – The Transformation of Vehicle Development and Data Management Processes, ATZ Automatisiertes Fahren, 2020
Virtual Prototype Management, Simulation-based Development Processes, Cross-domain Cooperation, OTA
28/10/2021
16:25 conference time (CEST, Berlin)
New Concept of Geometry-Based Finite Element Model Generation for Crash Simulation within the Graph- and Heuristic-Based Topology Optimisation
28/10/2021 16:25 conference time (CEST, Berlin)
Room: H
T. Pohl (Stellantis, DEU); F. Beyer, A. Schumacher (Bergische Universität Wuppertal, DEU)
T. Pohl (Stellantis, DEU); F. Beyer, A. Schumacher (Bergische Universität Wuppertal, DEU)
In today’s engineering processes, optimisation plays a key role in finding good solutions upfront in the concept definition. Especially topology optimisation can drive the design to find new layouts. This is important in the automotive industry these days as new trends as electrification and autonomous driving offer the opportunity to heavily influence the vehicles. For the design of structures, topology optimisation methods are well established for linear elastic loadcases. However, for vehicle structures, a good performance in case of crash events is extremely important and amongst the driving criteria for the concept selection. The extension of the topology optimisation to these highly non-linear loadcases is a fairly recent development. In recent work, the work use of heuristics – i.e. design rules – for the optimisation procedure of 3-D structures has been explored. In order to allow a topological modification, the structure is being represented by a graph description, giving the information about the location of structural members and their connectivity. This graph is turned into a finite element model of the structure. After the simulation run, a number of criteria are evaluated, such as the distribution of the internal energy density or the relative velocity of the deforming members. Based on these, the design rules are activated, e.g. to support a buckling structure with a new member. This new structure is added in the graph description, which then again gives a new finite element model. In order to keep the mass of the structure constant, the thickness of the parts can be varied. In addition to this outer loop, for each structure a shape optimisation is added in an inner loop. Several alternative and competing topological modifications are being evaluated in each iteration, and the best is used for the subsequent iteration. This method has proven to be successful for different loadcases, e.g. to reduce the maximum intrusion of an impactor, or to reduce the maximum acceleration of the structure – which is relevant for the protection of the occupants of a vehicle. One criterion for a reliable optimisation result is the adequate finite element representation of the structure. In addition to the existing tool set, a method is being developed to create a model that contains typical features of vehicle body-in-white sheet metal structures such as flanges and spot weld connections, in addition to more detailed structural joints. A library of members and joints has been created, out of which a new external tool will convert the graph description into an SFE-Concept model. These models can be included in the optimisation loop of the topology optimisation, resulting in more realistic final structures.
Topology Optimisation, Heuristic, FE Model Generation
×

[TITLE]

[LISTING

[ABSTRACT]

[DATE]

[ROOM]

[KEYWORDS]