1. Characterizing Fine-Grained Associativity Gaps: A Preliminary Study of CAD-CAE Model Interoperability Russell.Peak@marc.gatech.edu http://itimes.marc.gatech.edu/ http://eislab.gatech.edu/projects/ ASME 2003 Design Engineering Technical Conferences and Computers and Information in Engineering Conference September 2-6, 2003 Chicago, Illinois Paper No. CIE-48232 http://eislab.gatech.edu/pubs/conferences/2003-asme-detc-peak/ Copyright © 1992-2003 by Georgia Tech Research Corporation, Atlanta, Georgia 30332-0415 USA. All Rights Reserved. Permission to reproduce and distribute for non-commercial purposes (including internal corporate usage) is hereby granted provided this notice and a proper citation are included.

2. 2 Abstract Characterizing Fine-Grained Associativity Gaps: A Preliminary Study of CAD-CAE Model Interoperability This paper describes an initial study towards characterizing model associativity gaps and other engineering interoperability problems. Drawing on over a decade of X-analysis integration (XAI) research and development, it uses the XAI multi-representation architecture (MRA) as a means to decompose the problem and guide identification of potential key metrics. A few such metrics are highlighted from the aerospace industry. These include number of structural analysis users, number of analysis templates, and identification of computing environment components (e.g., number of CAD and CAE tools used in an example aerospace electronics design environment). One problem, denoted the fine-grained associativity gap, is highlighted in particular. Today such a gap in the CAD-CAE arena typically requires manual effort to connect an attribute in a design model (CAD) with attributes in one of its analysis models (CAE). This study estimates that 1 million such gaps exist in the structural analysis of a complex product like an airframe. The labor cost alone to manually maintain such gaps likely runs in the tens of millions of dollars. Other associativity gap costs have yet to be estimated, including over- and under-design, lack of knowledge capture, and inconsistencies. Narrowing in on fundamental gaps like fine-grained associativity helps both to characterize the cost of today’s problems and to identify basic solution needs. Other studies are recommended to explore such facets further. http://eislab.gatech.edu/pubs/conferences/2003-asme-detc-peak/ X = design, mfg., sustainment, and other lifecycle phases.

3. 3 Design ModelsAnalysis Models Design Models Analysis Models Frame of Reference 10+ years of R&D in CAD-CAE Model Representation & Interoperability Resulting techniques to date: u Architecture with new model abstractions (patterns) –Enables modular, reusable building blocks –Supports diversity: »Product domains and physical behaviors »CAD/E methods and tools –Supports multiple levels of fidelity Other Model Abstractions (Patterns)

4. 4 Frame of Reference (cont.) 10+ years of R&D in CAD-CAE Model Representation & Interoperability u Represent design-analysis model associativity as tool-independent knowledge u Provide methodology –Capture analysis idealization knowledge –Create highly automated analysis templates –Support product design Design Models Analysis Models Other Model Abstractions (Patterns) Idealization & Associativity Relations

5. 5 X-Analysis Integration Techniques for CAD-CAE Interoperability http://eislab.gatech.edu/research/ a. Multi-Representation Architecture (MRA)b. Explicit Design-Analysis Associativity c. Analysis Module Creation Methodology

6. 6 COB-based Constraint Schematic for Multi-Fidelity CAD-CAE Interoperability Flap Link Benchmark Example

7. 7 FEA-based Analysis Template Linkage Plane Stress Model Higher fidelity version vs. Linkage Extensional Model ABB  SMM SMM Template

8. 8 Flexible High Diversity Design-Analysis Integration Electronic Packaging Examples: Chip Packages/Mounting Shinko Electric Project: Phase 1 (production usage) EBGA, PBGA, QFP Analysis Modules (CBAMs) of Diverse Behavior & Fidelity FEA Ansys General Math Mathematica Analyzable Product Model XaiTools ChipPackage Thermal Resistance 3D3D Modular, Reusable Template Libraries Analysis Tools Design Tools PWB DB Materials DB* Prelim/APM Design Tool XaiTools ChipPackage Thermal Stress Basic 3D** ** = Demonstration module Basic Documentation Automation Authoring MS Excel

9. 9 Chip Package Thermal Resistance Analysis Template (FEA-based CBAM)

10. 10 Circuit Board Design-Analysis Integration Electronic Packaging Examples: PWA/B Analysis Modules (CBAMs) of Diverse Mode & Fidelity Design Tools Laminates DB FEA Ansys General Math Mathematica Analyzable Product Model XaiTools PWA-B XaiTools PWA-B Solder Joint Deformation* PTH Deformation & Fatigue** 1D, 2D 1D, 2D, 3D Modular, Reusable Template Libraries ECAD Tools Mentor Graphics, Zuken, … Analysis Tools PWB Warpage 1D, 2D Materials DB PWB Stackup Tool XaiTools PWA-B STEP AP210 ‡ GenCAM**, PDIF* ‡ AP210 Ed2 WD8 * = Item not yet available in toolkit (all others have working examples) ** = Item available via U-Engineer.com

11. 11 PWB Warpage Modules a.k.a. CBAMs: COB-based analysis templates PWB Thermal Bending Model (1D formula-based CBAM) PWB Plane Strain Model (2D FEA-based CBAM) APM Usage of Rich Product Models

12. 12 Purpose of this Paper Preliminary Characterization of CAD-CAE Interoperability Problem Number of Subsystems (Models, Tools, Relations) Design Models Analysis Models Other Model Abstractions (Patterns) Idealization & Associativity Relations

13. 13 Multi-Representation Architecture (MRA) Summary Characteristics of Component Representations u Solution Method Models (SMMs) –Packages solution tool inputs, outputs, and control as integrated objects –Automates solution tool access and results retrieval via tool agents and wrappers u Analysis Building Blocks (ABBs) –Represents analysis concepts using object and constraint graph techniques –Acts as a semantically rich 'pre-preprocessor' and 'post- postprocessor' model. »ABB instances create SMM instances based on solution method considerations and receive results after automated solution tool execution

14. 14 Multi-Representation Architecture (MRA) Summary Characteristics of Component Representations (cont.) u Analyzable Product Models (APMs) –Represent design aspects of products and enables connections with design tools –Support idealizations usable in numerous analysis models –Have possibly many associated CBAMs u Context-Based Analysis Models (CBAMs) –Contain linkages explicitly representing design-analysis associativity, indicating usage of APM idealizations –Create analysis models from ABBs and automatically connects them to APM attributes –Represent common analysis models as automated, predefined templates –Support interaction of analysis models of varying complexity and solution method –Enable parametric design studies via multi-directional input/output (in some cases)

15. 15 Multi-Representation Architecture (MRA) Summary Overall Characteristics u Addresses information-intensive nature of CAD-CAE integration u Breaks design-analysis integration gap into smaller subproblems (patterns) u Flexibly supports different design and analysis methods and tools u Based on modular, reusable information building blocks u Defines methodology for creating specialized, highly automated analysis tools to support product design u Represents analysis intent as tool-independent knowledge

16. 16 Multi-Representation Architecture (MRA) Summary Overall Characteristics (cont.) u Multiple representations required by: –Many:Many cardinality –Reusability & modularity Self-Test: Consider impact of removing a representation u Similar to “software design patterns” for CAD-CAE domain –Identifies patterns between CAD and CAE (including new types of objects) –Captures explicit associativity –Other needs: conditions, requirements, next-higher analysis u Distinctive CAD-CAE associativity needs –Multi-fidelity, multi-directional capabilities

17. 17 Multi-Representation Architecture for Design-Analysis Integration O(100) tools

18. 18 Holding Account Billing & Payable E-CAE Toolsmiths and Workstations DNP Operations Toolsmiths WorkstationsSystemCAE DIVISION 31 Servers & Sys Admin M-CAE Servers & Sys Admin Servers & SA TMOD Severs * Not DNP Operations Design, Build, Assembly, Test (DBAT) Process CAE Cost Centers JPL Projects and Technical Divisions Customers Robin Moncada Management and Administration Soap Sat Took Kit SDK Doors ApGen Fast Flight Ansoft HPEE Sof Sonnet Mentor Graphics Cadence Mathworks Matlab Synopsys Synplicity Ilogix Statemate Orcad AutoCad Relex Avant! Place & Route - Actel - Xilinx - Atmel (and many more) PTC Computer Vision PTC Pro-E SDRC Ideas SDRC Femap Solid Works Cosmos NASTRAN Adams Sinda/Fluent PDMS - EDMG SDRC Metaphase Sherpa Visual ToolSets Cool Jex Perceps Rational Rose Ruify Harlequin LISP I-Logix Rhapsody Code V LensView TracePro Zemax Software ToolsElectronicsCAE DIVISION 34 RF & EM CAE DIVISION 33 MechanicalCAE DIVISION 35 SoftwareCAE DIVISION 36 OpticalCAE DIVISION 38 M-CAE Toolsmiths and Workstations Adapted from “Computer Aided Engineering Tool Service at JPL” - 2001-07-22 - Mike Dickerson -NASA-JPL Example CAD/E/X Toolset (JPL) ~100 tools

19. 19 Multi-Representation Architecture for Design-Analysis Integration O(100) types

20. 20 Analysis Building Blocks (ABBs) Analysis Primitives Beam q(x) Distributed Load Rigid Support Cantilever Beam System Analysis Systems - Primitive building blocks- Containers of ABB "assemblies" Material Models     Specialized General - Predefined templates - User-defined systems Analysis VariablesDiscrete Elements Interconnections Continua Plane Strain Body Linear- Elastic Bilinear Plastic Plate Low Cycle Fatigue  N MassSpringDamper x y q(x) Beam Distributed Load Rigid Support No-Slip body 1 body 2 Temperature, Stress, Strain,   T Geometry Object representation of product-independent analytical engineering concepts

21. 21 Example Industrial Needs: Common Structures Workstation (CSW) Request for Information Publicly available document (see http://eislab.gatech.edu/projects/boeing-psi/2000-06-csw-rfi/ )http://eislab.gatech.edu/projects/boeing-psi/2000-06-csw-rfi/

22. 22 Common Structures Workstation (CSW) Request for Information June 2000, The Boeing Company. Appendix B: Required Standard Analysis Methods Available at http://eislab.gatech.edu/projects/boeing-psi/2000-06-csw-rfi/ ~110 generic template groupings

23. 23 Appendix B: Required Standard Analysis Methods (continued)

24. 24 COB-based Libraries of Analysis Building Blocks (ABBs) Material Model and Continuum ABBs - Constraint Schematic-S Material Model ABB Continuum ABBs modular re-usage Torsional Rod Extensional Rod 1D Linear Elastic Model

25. 25 COB-based Libraries of Analysis Building Blocks (ABBs) Material Model and Continuum ABBs - COB Structure-S COB one_D_linear_elastic_model SUBTYPE_OF elastic_model; youngs_modulus, E : REAL; poissons_ratio, : REAL; cte,  : REAL; shear_modulus, G : REAL; strain,  : REAL; stress,  : REAL; shear_stress,  : REAL; shear_strain,  : REAL; thermal_strain,  t : REAL; elastic_strain,  e : REAL; temperature_change,  T : REAL; RELATIONS r1 : " * ( 2 * (1 + ) ) == "; r2 : " == + "; r3 : " == / "; r4 : " == * "; r5 : " == / "; END_COB; COB one_D_linear_elastic_model_isothermal SUBTYPE_OF one_D_linear_elastic_model; RELATIONS r6 : " == 0"; END_COB; COB slender_body SUBTYPE_OF deformable_body; undeformed_length, L 0 : REAL; reference_temperature, T 0 : REAL; temperature, T : REAL; RELATIONS sb1 : " == - "; END_COB; COB extensional_rod SUBTYPE_OF slender_body; start, x 1 : REAL; end, x 2 : REAL; length, L : REAL; total_elongation, ΔL : REAL; force, F : REAL; area, A : REAL; material_model : one_D_linear_elastic_model_noShear; RELATIONS er1 : " == - "; er2 : " == - "; er3 : " == / "; er4 : " == / "; END_COB;

26. 26 Multi-Representation Architecture for Design-Analysis Integration O(10,000)

27. 27 Analysis Tools Flexible High Diversity Design-Analysis Integration Phases 1-3 Airframe Examples: “Bike Frame” / Flap Support Inboard Beam Analysis Modules (CBAMs) of Diverse Feature:Mode, & Fidelity Design Tools Materials DB FEA Elfini* MATDB-like Analyzable Product Model XaiTools Fitting: Bending/Shear 3D 1.5D Modular, Reusable Template Libraries MCAD Tools CATIA v4, v5 Lug: Axial/Oblique; Ultimate/Shear 1.5D Assembly: Ultimate/ FailSafe/Fatigue* * = Item not yet available in toolkit (all others have working examples) Fasteners DB FASTDB-like General Math Mathematica In-House Codes Image API (CATGEO); VBScript

28. 28 Lug Template (CBAM) Applied to an Airframe Problem Solution Tool Interaction Boundary Condition Objects (links to other analyses) CAD-CAE Associativity (idealization usage) Material Models Model-based Documentation Geometry Requirements Legend: Annotations highlight model knowledge capture capabilities. Other notation is COB constraint schematics notation.

29. 29 Appendix B: Required Standard Analysis Methods (continued)

30. 30 “Bike Frame” Bulkhead Fitting Analysis Template Using Constrained Object (COB) Knowledge/Info Representation e se tr P f 0 2   2 1 e be ht P Cf  ),,( 13 hbrfK  18 associativity relations

31. 31 Quantity estimates by MRA representation type O(100) tools O(100) types O(10,000)

32. 32 CAD-CAE associativity relations are represented as APM-ABB relations (in CBAMs) O(1,000,000) relations An associativity gap is a computer-insensible relation

33. 33 e se tr P f 0 2   2 1 e be ht P Cf  ),,( 13 hbrfK  Channel Fitting Analysis Associativity Gaps between CAD and CAE Models Analysis Model (with Idealized Features) Detailed Design Model Idealizations   1 : b = cavity3.inner_width + rib8.thickness/2 + rib9.thickness/2  “It is no secret that CAD models are driving more of today’s product development processes... With the growing number of design tools on the market, however, the interoperability gap with downstream applications, such as finite element analysis, is a very real problem. As a result, CAD models are being recreated at unprecedented levels.” Ansys/ITI press Release, July 6 1999 http://www.ansys.com/webdocs/VisitAnsys/CorpInfo/PR/pr-060799.html No explicit fine-grained CAD-CAE associativity

34. 34 “Bike Frame” Bulkhead Fitting Analysis Template Using Constrained Object (COB) Knowledge/Info Representation e se tr P f 0 2   2 1 e be ht P Cf  ),,( 13 hbrfK  18 CAD-CAE associativity relations

35. 35 ~1M Associativity Gaps Reference: http://eislab.gatech.edu/pubs/conferences/2003-asme-detc-peak/ Categories of Gap Costs Associativity time & labor - Manual maintenance - Little re-use - Lost knowledge Inconsistencies Limited analysis usage - Fewer parts analyzed - Fewer iterations per part “Wrong” values - Too conservative: Extra part costs and performance inefficiencies - Too loose: Re-work, failures, law suits Initial Cost Estimate per Complex Product (only for manual maintenance costs of structural analysis problems)

36. 36 Cost Estimate per Complex Product p.1/2 Manual Maintenance of Associativity Gaps in Structural Analysis Problems Reference:http://eislab.gatech.edu/pubs/reports/EL004/http://eislab.gatech.edu/pubs/reports/EL004/

37. 37 Cost Estimate per Complex Product p.2/2 Manual Maintenance of Associativity Gaps in Structural Analysis Problems Reference:http://eislab.gatech.edu/pubs/reports/EL004/http://eislab.gatech.edu/pubs/reports/EL004/

38. 38 Characterizing Complex Model Interoperability Using the Multi-Representation Architecture (MRA) u MRA: Similar to “software design patterns” for CAD-CAE domain –Identifies patterns between CAD and CAE (including new types of objects) –Captures multi-fidelity explicit associativity u Provides hybrid top-down & bottom-up methodology for characterizing problems & solutions –O(1,000,000) CAD-CAE gap estimate

39. 39 For Further Information... u Contact: Russell.Peak@marc.gatech.eduRussell.Peak@marc.gatech.edu u Web site: http://eislab.gatech.edu/http://eislab.gatech.edu/ –Publications, project overviews, tools, etc. –See: X-Analysis Integration (XAI) Central http://eislab.gatech.edu/research/XAI_Central.doc http://eislab.gatech.edu/research/XAI_Central.doc u XaiTools ™ home page: http://eislab.gatech.edu/tools/XaiTools/http://eislab.gatech.edu/tools/XaiTools/ u Pilot commercial ESB: http://www.u-engineer.com/http://www.u-engineer.com/ –Internet-based self-serve analysis –Analysis module catalog for electronic packaging –Highly automated front-ends to general FEA & math tools

40. Backup Slides

41. 41 An Introduction to X-Analysis Integration (XAI) Short Course Outline Part 1: Constrained Objects (COBs) Primer –Nomenclature Part 2: Multi-Representation Architecture (MRA) Primer –Analysis Integration Challenges –Overview of COB-based XAI –Ubiquitization Methodology Part 3: Example Applications »Airframe Structural Analysis (Boeing) »Circuit Board Thermomechanical Analysis (DoD: ProAM; JPL/NASA) »Chip Package Thermal Analysis (Shinko) –Summary Part 4: Advanced Topics & Current Research

42. 42 Constrained Object (COB) Representation Current Technical Capabilities - Generation 2 u Capabilities & features: –Various forms: computable lexical forms, graphical forms, etc. »Enables both computer automation and human comprehension –Sub/supertypes, basic aggregates, multi-fidelity objects –Multi-directionality (I/O changes) –Reuses external programs as white box relations –Advanced associativity added to COTS frameworks & wrappers u Analysis module/template applications (XAI/MRA): –Analysis template languages –Product model idealizations –Explicit associativity relations with design models & other analyses –White box reuse of existing tools (e.g., FEA, in-house codes) –Reusable, adaptable analysis building blocks –Synthesis (sizing) and verification (analysis)

43. 43 Constrained Objects (cont.) Representation Characteristics & Advantages - Gen. 2 u Overall characteristics –Declarative knowledge representation (non-causal) –Combining object & constraint graph techniques –COBs = (STEP EXPRESS subset) + (constraint graph concepts & views) u Advantages over traditional analysis representations –Greater solution control –Richer semantics (e.g., equations wrapped in engineering context) –Unified views of diverse capabilities (tool-independent) –Capture of reusable knowledge –Enhanced development of complex analysis models u Toolkit status (XaiTools v0.4) –Basic framework, single user-oriented, file-based See Advanced Topics for Gen.3 Extensions

44. 44 COB Modeling Languages Lexical and Graphical Formulations Structure Level (Template) Instance Level OWL, XML, and UML formulations are envisioned extensions

45. 45 COB Structure: Graphical Forms Tutorial: Triangle Primitive Basic Constraint Schematic-S Notation c. Constraint Schematic-S a. Shape Schematic-S b. Relations-S d. Subsystem-S (for reuse by other COBs) Aside: This is a “usage view” in AP210 terminology (vs. the above “design views”)

46. 46 COBs as Building Blocks Tutorial: Triangular Prism COB Structure c. Constraint Schematic-S a. Shape Schematic-S b. Relations-S d. Subsystem-S (for reuse by other COBs) e. Lexical COB Structure (COS) COB triangular_prism SUBTYPE_OF geometric_shape; length, l : REAL; cross-section : triangle; volume, V : REAL; RELATIONS r1 : " == * "; END_COB;

47. 47 Example COB Instance Tutorial: Triangular Prism Constraint Schematic-ILexical COB Instance (COI) state 1.0 (unsolved): INSTANCE_OF triangular_prism; cross-section.base : 2.0; cross-section.height : 3.0; length : 5.0; volume : ?; END_INSTANCE; state 1.1 (solved): INSTANCE_OF triangular_prism; cross-section.base : 2.0; cross-section.height : 3.0; cross-section.area : 3.0; length : 5.0; volume : 15.0; END_INSTANCE; Basic Constraint Schematic-I Notation example 1, state 1.1

48. 48 COB-based Constraint Schematic for Multi-Fidelity CAD-CAE Interoperability Flap Link Benchmark Example

49. 49 (1) Extension Analysis a. 1D Extensional Rod 1. Behavior: Shaft Tension 2. Conditions: Flaps down : F = 3. Part Features: (idealized) 4. Analysis Calculations: 1020 HR Steel E= 30e6 psi L eff = 5.0 in 10000 lbs 5. Conclusion: A = 1.125 in 2  allowable   18000 psi 1.025 (2) Torsion Analysis Flap Link Analysis Documentation b. 2D Plane Stress FEA... (1a) Analysis Template: Flap Link Extensional Model APM ABB CBAM SMM Tutorial Example: Flap Link Analysis Template (CBAM) * Boundary condition objects & pullable views are WIP concepts* Solution Tool Interaction Boundary Condition Objects (links to other analyses)* CAD-CAE Associativity (idealization usage) Material Models Pullable Views* Geometry

50. 50 material effective length, L eff deformation model linear elastic model L o Extensional Rod (isothermal) F  L  A L  E x 2 x 1 youngs modulus, E shaft critical_cross _section al1 al3 al2 linkage mode: shaft tension condition reaction allowable stress stressmosmodel Margin of Safety (> case) allowable actual MS description area, A basic example 1, state 1 steel 10000 lbs flaps mid position 1.125 in 2 18000 psi 30e6 psi 1.025 5.0 in 8888 psi 1.43e-3 in Flap Link #3 Flap Linkage Instance with Multi-Directional I/O States Design Verification - Input: design details - Output: i) idealized design parameters ii) physical response criteria Design Synthesis - Input: desired physical response criteria - Output: i) idealized design parameters (e.g., for sizing), or ii) detailed design parameters

51. 51 Flap Link Extensional Model (CBAM) Example COB Instance in XaiTools (object-oriented spreadsheet) Detailed CAD data from CATIA Idealized analysis features in APM Explicit multi-directional associativity between design & analysis Modular generic analysis templates (ABBs) Library data for materials Focus Point of CAD-CAE Integration example 1, state 1

52. 52 Using Internet/Intranet-based Analysis Solvers Web Services Architecture Client PCs XaiTools Thick Client Users Internet June’99-Present: EIS Lab - Regular internal use U-Engineer.com - Demo usage: - US - Japan Nov.’00-Present: Electronics Co. - Production usage (dept. Intranet) Future: Company Intranet and/or U-Engineer.com (commercial) - Other solvers Iona orbixdj Mathematica Ansys Internet/Intranet XaiTools Ansys Solver Server XaiTools Ansys Solver Server XaiTools Math. Solver Server CORBA Daemon XaiTools Ansys Solver Server FEA Solvers Math Solvers CORBA Servers CORBA IIOP... Engineering Service Bureau Host Machines Current Version: XaiTools Web Services - SOAP (adds Patran & Abaqus, ACIS, …)

53. 53 Convergence of Representations Database Techniques (data structure, storage …) Software Development (algorithms …) Artificial Intelligence & Knowledge-Based Techniques (structure combined with algorithms/relations/behavior) EER STEP Express ER UML Flow Charts OMT Objects Rules Constraint graphs Constrained Object - like Representations COBs, OCL,...

54. 54 Short Course: Using Standards-based Engineering Frameworks for Electronics Product Design and Life Cycle Support

55. 55 Technique Summary u Tool independent model interoperability –Application focus: analysis template methodology u Multi-representation architecture (MRA) & constrained objects (COBs): –Addresses fundamental gaps: »Idealizations & CAD-CAE associativity: multi-fidelity, multi-directional, fine-grained –Based on information & knowledge theory –Structured, flexible, and extensible u Improved quality, cost, time: –Capture engineering knowledge in a reusable form –Reduce information inconsistencies –Increase analysis intensity & effectiveness »Reducing modeling cycle time by 75% (production usage)