One model, many use cases
1 like382 views
The document discusses the benefits of a unified modeling approach that allows for the development and deployment of consistent vehicle models across various execution environments, enhancing efficiency and reducing maintenance efforts. It emphasizes interoperability with open standards like FMI and Modelica, enabling integration with third-party models and facilitating multi-domain simulations. Moreover, the document highlights the importance of customizable design, real-time execution capabilities, and the ability to evaluate designs against requirements across multiple fidelity levels.
![9/13/2016 © Modelon 2016 1313
”[We now have] an intuitive and highly customizable
design [that] runs powerful physical models through FMI.
Vehicle Energy Management Engineering (VEME),
Ford Motor Company](https://image.slidesharecdn.com/modelonrollingavec2016noanim-160913135630/85/One-model-many-use-cases-13-320.jpg)
![9/13/2016 © Modelon 2016 20
”The vehicle model [allows] engineers to replace, modify,
add and refine system blocks such as tires, aerodynamics,
engine, suspension, driveline [and to] link external blocks.
Andrea Toso and Allesandro Moroni, Dallara](https://image.slidesharecdn.com/modelonrollingavec2016noanim-160913135630/85/One-model-many-use-cases-20-320.jpg)
One model, many use cases
- 1. 9/13/2016 © Modelon 2016 1
- 2. ONE MODEL IN MULTIPLE ENVIRONMENTS • Define once Reduces effort to develop and maintain Ensures consistency • Deploy broadly Bring the model to those that can benefit from it Open standards, manageable and affordable usage Matlab/Simulink, MS Excel, Driver-in-the-Loop, .... 9/13/2016 © Modelon 2016 2
- 3. 80+ ONE MODEL 9/13/2016 3 MIL/SIL/HIL © Modelon 2016 Model authoring Model Deployment
- 4. VALUE OF ONE MODEL • Consistent model across organization and use-cases • Choose execution tool depending on purpose • FMI open standard format: model supported by 80+ CAE tools • Modelica open standard: user extensibility and choice of tool vendor • Unlimited model deployment: cost for authoring – not per user You are in control of your model IP! 9/13/2016 © Modelon 2016 4
- 5. 9/13/2016 © Modelon 2016 5 ”Being able to dig into the details and study how each component or subsystem affects the entire vehicle behavior is the key to an efficient and systematic analysis. Bengt Jacobson, Volvo Car Corporation
- 6. • Parameter management is consistent regardless of application • Modify data without regenerating execution code PARAMETERIZATION 9/13/2016 Execution environment Courtesy: Volvo © Modelon 2016 6 • Hardpoints • Spring data • Damper data • Bushing data • Part data • Parameters Authoring environment
- 7. PARAMETERIZATION-INDEPENDENT TOPOLOGY 9/13/2016 © Modelon 2016 7 No parameterization/default parameters
- 8. OVERLAY DATA ON TOPOLOGY 9/13/2016 © Modelon 2016 8 • Hardpoints • Spring data • Damper data • Bushing data • Part data • Parameters Data-aware components Supports different file formats xml, json, mat User extensible Utilities to read and write data
- 9. 9/13/2016 © Modelon 2016 9 ”“The openness of the code essentially gives us the advantages of an in-house tool without having to have a software development team in-house.” Mats Beckman, Volvo Car Corporation
- 10. INTEROPERABILITY • System architecture • Vehicle model with external models for: Brakes Steering Powertrain Tires Controls 9/13/2016 Courtesy: Nissan (©Modelon 2016 10
- 11. INTEROPERABILITY • Combine with 3rd party models, steering example Native Modelica Imported FMU Model for export 9/13/2016 © Modelon 2016 11
- 12. ASSEMBLY • Same/similar complexity parametric geometry • Driven by part data from CAD MULTI FIDELITY 9/13/2016 © Modelon 2016 12 COMMON INTERFACE All model fidelity interchangeable in all supported execution environments BEHAVIORAL • Driven by KnC data • Compatible with many different data formats • Executable faster than realtime on a single core Courtesy: Volvo PARAMETRIC GEOMETRY • Driven by geometry (hardpoints) • Reads existing data format • Executable on realtime on multiple cores REQUIREMENTS • Early phase models • Low fidelity • Very fast execution Requirements verification System design optimization Verification
- 13. 9/13/2016 © Modelon 2016 1313 ”[We now have] an intuitive and highly customizable design [that] runs powerful physical models through FMI. Vehicle Energy Management Engineering (VEME), Ford Motor Company
- 14. 9/13/2016 14© Modelon 2016
- 15. MULTI-DOMAIN: INTEGRATED VEHICLE SIMULATION 9/13/2016 © Modelon 2016 15 Lateral stability Drivability Launch performance Fuel economy Emissions Thermal management Transmission Prius Alison Dual clutch Manual Engine I4 Downsizing Cooling Engine Transmission Battery Motor Air Conditioning circuit Electric drive Electric motor and front diff Chassis McPherson Multi-link Battery Lithium-Ion Modelica and FMI technologies are used to investigate hybridization: • From: Conventional Rear Wheel Drive (RWD) • To: Hybrid Electric Drive by • Downsizing engine • Change of transmission • Adding front axle electric drive Key enablers of the technologies: • Cover multiple domains and their interaction in the same system model • Plug-and-play compatible component and system models, • System architectures designs • Component based validation • Multiple fidelity levels within the same framework • Export of models to other tools and applications
- 16. 9/13/2016 © Modelon 2016 16 ”Modelica was very suited to make such a multi-discipline and multi-domain investigation by model-based development Yutaka Hirano, Toyota Motor Corporation
- 17. 9/13/2016 © Modelon 2016 17
- 18. REAL TEST RIG (EXISTING) ADDITION CUSTOM ENVIRONMENT: VIRTUAL TEST RIG 9/13/2016 © Modelon 2016 18 Presentation at 1st Japanese Modelica Conference, Tokyo, May 23-24
- 19. CUSTOM ENVIRONMENT: ADAS 9/13/2016 © Modelon 2016 19 Presentation at 1st Japanese Modelica Conference, Tokyo, May 23-24
- 20. 9/13/2016 © Modelon 2016 20 ”The vehicle model [allows] engineers to replace, modify, add and refine system blocks such as tires, aerodynamics, engine, suspension, driveline [and to] link external blocks. Andrea Toso and Allesandro Moroni, Dallara
- 21. REALTIME • High Immersion Realism • Track testing too slow and expensive • Even at later development stages • Detailed models required early in design 9/13/2016 © Modelon 2016 21 SAE PAPER 2014-01-0099 PROFESSIONAL DRIVING SIMULATORS TO DESIGN FIRST-TIME RIGHT RACE CARS
- 22. REALTIME • High fidelty • High accuracy • Prepared for paralellization 9/13/2016 © Modelon 2016 22 Presentation at 1st Japanese Modelica Conference, Tokyo, May 23-24
- 23. Multi-Domain Multi-domain by design • Interaction and cross-dependencies between subsystems and physical domains captured • Facilitates simultaneous engineering • Allow for integrated design and optimization Shift control Cooling Hybrid electric Gearbox Hydraulics Engine A/C SUMMARY: KEYS TO MODEL-BASED SYSTEMS DESIGN Multi-Fidelity Inherent support for working with mixed fidelities • Get the architecture right • Keep the design connected and consistent • Continuously evaluate design against requirements • Executable specs <-> Detailed design BEHAVIORAL – PARAMETRIC – VERIFICATION SYSTEM–SUBSYSTEM–PART Multi-Simulation Formal and open description • Physics capture • Constraint and cost definition • Analysis and decision support • Reduction/elemination of real tests Examples: • Dynamic simulation • Steady-state • Optimization • Realtime/XIL • Controls design • Robust design • Requirements definition • Formal analysis9/13/2016 © Modelon 2016 Multi-Access Models accessible for use by everyone that needs it, on their conditions.
- 24. 9/13/2016 © Modelon 2016 24 www.modelon.com info@modelon.com
Editor's Notes
- #3: Remedy is separation of model and execution environment: so that one model supports multiple execution environments and one execution environment can be used by multiple models, these models can be multi fidelity Define one model (modeling = cost), and deploy it broadly (deployment = benefit) One model is good to reduce effort in maintenance and to ensure consistency Cost for model consists of maintenance of model, and data base (cost for maintaining data base is Klomps cost). After having one model, having one data set is to support multiple models is second best Modelon can support by delivering models in various formats Executable, with or without execution environment FMIE FMIT FMI .NET or FMIL CarMaker Batch simulations etc. Here we should mention jenkins and others OTT Model authoring environment, for generation of executables
- #4: If one OEM has two models, one high fidelity and one medium fidelity, they can both be applied in Matlab/Simulink. What would it take to make them available broader and what would the value be? Other applications include: Execute models in Microsoft Excel, can easily be customizable to support various engineers standard work Batch/cluster simulations for DoE, variant management and more Connect to CarMaker for virtual test driving Run in Matlab for custom analysis control design Steady state/performance optimization Driver-in-the-loop Virtual test (KnC and others)
- #8: Topology built in VDL based on standard components Default data
- #9: Swap in components with data access Data read at runtime, no recompile needed to change data