Libraries for ModelBased Mechatronic Concept Design in Sys

Libraries for Model-Based Mechatronic Concept Design in Sys. ML Benjamin Kruse Engineering Design and Computing Laboratory, ETH Zurich Benjamin Kruse

Agenda § Motivation & Introduction § Functional Modeling Library in Sys. ML § Library Definition and Usage § User Study for Library Evaluation § § Amesim Simulation Library in Sys. ML Component Modeling Library in Sys. ML Outlook Summary Benjamin Kruse Engineering Design + Computing Laboratory 2

Motivation: Rising Complexity Benjamin Kruse Systems Engineering Vision 2025 Project Team. “A World in Motion – Systems Engineering Vision 2025. ” International Council on Systems Engineering. 2014. Engineering Design + Computing Laboratory 3

State of the Art in Research & Industry § Functional Architectures in Sys. ML 1 (FAS) § Method for obtaining functional architectures for systems in block-oriented form § By using heuristics for grouping functions and allocating them to functional blocks § State of usage of Sys. ML 2 § Functional modeling has the most added value for users § Needed improvements: § Usability of Sys. ML needs to be improved § Modeling methods and guidelines are needed 1) Lamm, J. G. , and Weilkiens, T. “Method for Deriving Functional Architectures from Use Cases. ” Systems Engineering, 2014; 17: 2, p. 225 -236. 2) Albers, A. , and Zingel, C. “Challenges of Model-Based Systems Engineering: A Study Towards Unified Term Understanding and the State of Usage of Sys. ML. ” Smart Product Engineering, Springer, 2013. Benjamin Kruse Engineering Design + Computing Laboratory 4

Vision 2025 & Evolving MBSE § SE Vision 20251 § Formal systems modeling § Composable design: A key to productivity § By combining formal models from libraries § Evolving Sys. ML 2 § Include precise semantics that avoid ambiguity § Be usable for multiple application domains § Integrate across discipline-specific engineering tools § Primary consideration: Usability! 1) Systems Engineering Vision 2025 Project Team. “A World in Motion – Systems Engineering Vision 2025. ” International Council on Systems Engineering. 2014. 2) Friedenthal, S. and Burkhart, R. “Evolving Sys. ML and the System Modeling Environment to Support MBSE. ” in: Insight, INCOSE, 2015, 18: 2, p. 39 -41 Benjamin Kruse Engineering Design + Computing Laboratory 5

Approach for Mechatronic Concept Generation Formal & Standardized Language Task Specification Mechanical Electrical … Mechatronic Concept Models Benjamin Kruse Engineering Design + Computing Laboratory Model Generation Generic Multi -Disciplinary Model Libraries Model Simulation 6

Hirtz, J. , Stone, R. B. , Mcadams, D. A. , Szykman, S. , and Wood, K. L. A: “Functional Basis for Engineering Design: Reconciling and Evolving Previous Efforts. ” No. 1447, USA: NIST. 2002. NIST Functional Basis (FB) § Established collection of operators and flows to be combined into elementary functions of engineering systems Control. Signal Rotational. Energy : Decrease Rotational. Energy i. e. braking of electric car Decrease. To reduce a flow in response to a control signal. Example: Closing the value further decreases the flow of propane to the gas grill. Benjamin Kruse Rotational energy. Energy that results from a rotation or a virtual rotation. Example: … Engineering Design + Computing Laboratory 7

Functional Modeling Library in Sys. ML § Elementary. Function: FB operator as Sys. ML activity § Basic. Flow: Sys. ML block to define object flows Kruse, Münzer, et al. : “Workflow and Modeling Conventions for Function and Product Structure Modeling of Mechatronic Systems in Sys. ML using Libraries”. Mechatronics 2012, Linz, Austria, 2012 Benjamin Kruse Engineering Design + Computing Laboratory 8

Functional Modeling electric car main function s po e m co e d Benjamin Kruse Engineering Design + Computing Laboratory 9

Functional Decomposition Step-by-step functional decomposition, until Overall Function: e. g. Control Motion elementary functions from library are used User-Defined Function: e. g. Recuperate Braking Energy Elementary Function: e. g. Electrical. Energy : Store Benjamin Kruse Engineering Design + Computing Laboratory 10

Function Library Evaluation § Common advantages of reuse in software development 1 and engineering design 2: § Better understanding of a system built from building blocks § Higher model and system quality § Faster development Achievable through Function Library? Experiment Hypotheses: § Using the library leads to better models § Using the library leads to reduced workload 1) Chughtai, A. , and Oliver V. "Software-Wiederverwendung-Theoretische Grundlagen, Vorteile und realistische Beurteilung. " in: Software-Management: Beherrschung des Lifecycles. Editor: Versteegen, G. , Springer, 2002. 2) Duffy, A. H. B. , and Ferns, A. F. “An Analysis of Design Reuse Benefits. ” Proceedings of the 12 th International Conference on Engineering Design (ICED ’ 99). 1998; p. 799 -804. Benjamin Kruse Engineering Design + Computing Laboratory 11

User Study for Library Evaluation § Measurements for model quality: § More functionalities covered (compared to master models) to access task completion § Higher relative number of functions of the Functional Basis § Not: bigger model size (because not corresponding to model quality) § Measurements for modeling workload (Better Usability? ): § TLX 1 test to measure the perceived workload of participants (Established test by NASA) § ETH tools course for functional modeling & Sys. ML: § 11 participants (inexperienced students) § 3 afternoon sessions 1) Hart, S. G. & Staveland, L. E. : “Development of NASA-TLX (Task Load Index): Results of empirical and theoretical research. ” in: Human Mental Workload. Amsterdam: North Holland Press. 1988. Benjamin Kruse Engineering Design + Computing Laboratory 12

Experiment Plan § Factor: § With Sys. ML library: Functional Basis incorporated in Sys. ML § Without Sys. ML library: Functional Basis given on paper § Tasks: Creating functional models of a coffee maker § Task 1: Brewing coffee § Task 2: Grinding coffee beans Benjamin Kruse Engineering Design + Computing Laboratory 13

Results: TLX (NASA) Overall Perceived Workload 100 90 80 70 60 10 3 -2: with library 20 3 -1: no library 30 2: with library 40 1: no library 50 0 Day 1 (without Sys. ML library) Day 2 (with Sys. ML library) Day 3 -1 (without Sys. ML library) Day 3 -2 (with Sys. ML library) § Results: § Learning effect § Library increased workload! (0 = minimum workload, 100 = maximum workload) Benjamin Kruse Engineering Design + Computing Laboratory 14

Results: Questionnaire & Comments § Questionnaire: § “Modeling with the library in Sys. ML improved the resulting model / modeling process compared to not having the library. ” Approval rating: 59 – 60 (0 = completely disagree, 100 = completely agree) § Comments about library usage: “library forces you to break down the activity further” Productivity “triggers my thought process” Increased workload & comments: Results of using Functional Basis “resulting model is “having to look up. NOT functions, and directly of using the library decide which are appropriate” more fundamental” “very limited by the functions of the library” Benjamin Kruse Engineering Design + Computing Laboratory Time S. Rifkin: “Why new software processes are not adopted”, Advances in Computers, (59). 2003 15

Results: Library & FB Acceptance § Library & FB acceptance: Library used when available (good user acceptance) Functional Basis barely used without library Benjamin Kruse Relative # of FB functions § Significant correlation (p < 0. 001) between availability of library and the number of functions from library § Significant correlation (p < 0. 001) between availability of library and relative number of FB functions No Sys. ML Library Engineering Design + Computing Laboratory With Sys. ML Library 16

Results: Task Completion § Significant correlation (p < 0. 018) between availability of library and ratio of covered functionalities § Significant correlation (p < 0. 021) between ratio of FB functions and ratio of covered functionalities Relative # of functionalities § Grade of task completion 1: No Sys. ML Library With Sys. ML Library Having the library (and therefore using the Functional Basis) leads to a broader coverage of the necessary functionalities 1) Annett, J. : “Hierarchical Task Analysis”. In: Hollnagel, E. , Ed. , Handbook of Cognitive Task Design, Lawrence Erlbaum Assoc. Inc. , Mahwah. 2003. 17 -35. Benjamin Kruse Engineering Design + Computing Laboratory 17

User Study – Summary Higher usage of formal FB terms More functionalities covered Reuse benefits? Better understanding of a system Higher model / system quality Increased workload (Due to usage of FB, not due to the library) Experiment assumption of general usage of FB not applicable Benjamin Kruse Faster development Better & more formal models Engineering Design + Computing Laboratory 18

Further development process … Structure Behavior Functions Function Library Use Cases Behavior Library Requirements Task Benjamin Kruse (FBS) Engineering Design + Computing Laboratory Component Library 19

Amesim Simulation Library in Sys. ML § Incorporation of Amesim library into Sys. ML Library (Sys. ML) Library (Amesim) Model (Sys. ML) Model (Amesim) § Model transformations between Sys. ML IBDs & Amesim models Benjamin Kruse Engineering Design + Computing Laboratory 20

Component Modeling Library in Sys. ML § Structural Components: Library § (Mostly) Physical entity or module § Following e. Cl@ss standard § Model: § Block with attributes (e. g. weight, etc) § Interfaces: § Ports with additional information (e. g. type, etc) and flow types corresponding to function library Model Image source: http: //www. cross-morse. co. uk/timing_belt. asp Benjamin Kruse Engineering Design + Computing Laboratory 21

Allocation Matrix: Usage-to-Definition column: elementary functions as actions (operator & flow) row: (mostly physical) components from library Electrical. Energy: Store (for recuperation) is allocated to the component “Car Battery” Benjamin Kruse Engineering Design + Computing Laboratory 22

Library Summary § Function Library § § Based on the defined terms of the Functional Basis Increased workload for inexperienced users Higher usage of FB terms Better & more formal More functionalities covered functional models § Amesim Simulation Library § Corresponding to Amesim simulation elements Composing traceable partial simulation models in Sys. ML § Component Modeling Library § Based on e. Cl@ss standard § For reusing common elements Benjamin Kruse Composing configurations Engineering Design + Computing Laboratory 23

Outlook § Combination of model-based libraries in Sys. ML and automated design synthesis § Model transformation for solution space exploration using boolean satisfiability 1 § Model transformations for system behavior simulation Automated synthesis & evaluation of Sys. ML models § Further testing and validation with industry 1) Münzer, C. , Helms, B. , and Shea, K. “Automatically Transforming Object-Oriented Graph-Based Representations into Boolean Satisfiability Problems for Computational Design Synthesis. ” Journal of Mechanical Design, 2013; 135: 10 Benjamin Kruse Engineering Design + Computing Laboratory 24

Automated Design Synthesis Gear Engine rpm Acceleration Motion Brake Münzer, C. , Helms, B. , and Shea, K. “Automatically Transforming Object-Oriented Graph-Based Representations into Boolean Satisfiability Problems for Computational Design Synthesis. ” Journal of Mechanical Design, 2013; 135: 10 Benjamin Kruse Engineering Design + Computing Laboratory 25

Automated Simulation § Automated map between elements and corresponding simulation models § Enables quantitative feedback on generated concepts and optimization Münzer, C. and Shea, K. : “A Simulation-based CDS Approach: Automated Generation Of Simulation Models Based From Generated Concept Model Graphs”, Proceedings of the ASME IDETC/CIE 2015, DETC 2015 -47353 Benjamin Kruse Engineering Design + Computing Laboratory 26

integrated error ei [m] Simulated vs. Optimized Solutions Münzer, C. and Shea, K. : ASME IDETC/CIE 2015 Benjamin Kruse Engineering Design + Computing Laboratory total CO 2 emissions [g] 27

Summary Sys. ML Model (FBS) Function Library • • Operator & Flow Behavior Library • Amesim component Component • Library Functional Basis Physical entity or module Image source: http: //www. cross-morse. co. uk/timing_belt. asp Solution Space Explorer using Boolean Satisfiability Benjamin Kruse Simulation Model Engineering Design + Computing Laboratory 28

Thank you for your attention! Benjamin Kruse Engineering Design and Computing Laboratory, ETH Zurich CLA F 32. 2, Tannenstrasse 3, 8092 Zurich, Switzerland http: //www. edac. ethz. ch bkruse@ethz. ch Benjamin Kruse Engineering Design + Computing Laboratory