14 Fenyes4/14/04 Packaging ? Geometric measures to check requirements – Occupant model – Engine compartment packaging ? Inputs – Occupant position - H-point, etc – Engine/transmission selection and position ? Outputs – Shoulder room, knee room, etc – Roominess measure – Packaging feasibility Fenyes4/14/04 Trunk Volume 15 Fenyes4/14/04 Linear Structures ? FE model associatively linked to UG parametric model - UG Scenario – FE model is built once - automatically updated as geometry changes – Beams/Springs/Shells/Masses ? Locations and properties associated to geometry (M, K, b, h, t, etc) ? Inputs – Parametric geometry specification – Component masses – Initial structures model ? Outputs – Vehicle mass – Structural modes Fenyes4/14/04 Tightly Coupling Representation to Analyses ? Analysis models remain synchronized with representation ? Example – CAD to CAE/structures: UG Modeling and UG Scenario – Automatically update hybrid beam/spring/shell/mass model from CAD model Automatically update from CAD to CAE 16 Fenyes4/14/04 Aerodynamics ? Exterior aero surface linked to underlying structure CAD representation – Aero drag is approximated – Frontal area calculated in UG ? Inputs – Exterior shape ? Outputs – Aero drag – Frontal area Fenyes4/14/04 Energy ? Compute fuel economy, acceleration based on spreadsheet model – Depends on structures, aero, marketing disciplines ? Inputs –Cd –Drag – Frontal area – Powertrain, tires, etc. – Performance requirements ? Outputs – Fuel economy ? (city, highway, combined) – Acceleration 17 Fenyes4/14/04 Business ? Estimate sales, revenue, costs – Link performance to customer value – Link customer value to sales/revenue ? Inputs – Competitors – Performance – Forming and assembly technology – Equipment, tooling costs – BOM/Parts - size, mass, material ? Outputs –Sales – Revenue –Cost – Net income, profit Fenyes4/14/04 One Approach to Link Market Demand, Value, and Performance: S-Model (Ref: H. E. Cook, 1997) Customer-Perceived Value Drives Market Demand Price Sale s Volum e Baseline Value Competitive Advantage Increased Value Product Specifications Drive Customer-Perceived Value Pr oduct Va l u e Product Spec Improved Function Increased Value (e.g. 0-60 Time, Turning Circle) 18 Fenyes4/14/04 Example Problem - Dimensional Flexibility for Vehicle Architecture ? Maximize net income while satisfying performance requirements – Discipline analyses: geometry, aero, fuel economy. packaging, business – Discipline sub-optimization: structures, business ? Specific vehicle configuration: – Body style – Powertrain and components ? Nine high level, architectural design variables – Vehicle width at rocker – Front and rear track width – Front and rear overhang – Front and rear axle location – vertical and horizontal ? Performed automated discipline analyses: – Perturbed representation – Generated analysis models – Exchanged data through database – Computed change in Net Income (“natural” objective) ? Generated sensitivities and optimized Design Representation (Unigraphics) Database (MS Access) Multidisciplinary Design (iSIGHT) Structural Optimization (NASTRAN) Aerodynamics Interior Roominess (Excel) Business Summary of Results (Excel) Energy Custom Custom Custom Fenyes4/14/04 Shoulder Room Design Representation (Unigraphics) Database (MS Access) Multidisciplinary Design (iSIGHT) Structural Optimization (NASTRAN) Aerodynamics Interior Roominess (Excel) Business Summary of Results (Excel) Vehicle Geometry Bo d y St ru ct u r e M ass Frontal Area, C d Value of Roominess Optimized Gauges And Sections Architecture Configuration & Parameterization Net Income Exterior Width Gauges, Areas, Section Sizes Fuel Economy, Performance Energy Overall Width Example – Dimensional Flexibility Data Flow and Analysis Vehicle Width at Rocker 19 Fenyes4/14/04 Framework Illustration: Analysis Outputs Sensitivity of Net Income to Rocker Location Relative Sensitivities for Other Architecture Parameters ? Front and Rear Axle Position - L, H ? Front and Rear Overhang ? Front and Rear Track -8% -6% -4% -2% 0% 2% 4% 6% 8% -4% -2% 0% 2% 4% Change in Rocker Center Y Coordinate C han ge i n N e t I n c o m e -15% -10% -5% 0% 5% 10% 15% Rocker Ctr. Y Coord. + 5%- 5% RR Axle X Coord. - 5%+ 5% FRT Overhang + 5% - 5% RR Overhang 0%- 5% / + 5% FRT Axle X Coord. + 5% - 5% RR Track + 5%- 5% RR Axle Z Coord. + 5% - 5% FRT Axle Z Coord. - 5% + 5% Fenyes4/14/04 96.00% 98.00% 100.00% 102.00% 104.00% 106.00% 108.00% 110.00% 112.00% 0 2 4 6 8 10 12 14 16 18 RunCounter N e t In c o m e % Series1 Iteration History - Dimensional Flexibility ? Sequential discipline analyses, sub-optimizations for structures and business ? 10 iterations for gradients, 6 iterations for convergence 20 Fenyes4/14/04 Reporting Analysis and Optimization Results: View Database through Web Interface ? Iteration history may be reviewed Fenyes4/14/04 Other Automotive MAO Applications ? Crash, linear analysis, robust design ? Aero and acoustics 21 Fenyes4/14/04 General MAO Challenges ? Problem Formulation – Determine key drivers and responses – MDO formulation – “natural” objective, multi-objective, preference modeling, etc. ? Consistent parametric representation – Consistent information shared by all disciplines ? geometric data ? non-geometric data (BOM, configuration, material properties, …) ? analysis results history, gradients, approximations ? Discipline analysis to support tradeoffs – Analysis tightly coupled to representation – Key disciplines are supported – Balance analysis detail against design knowledge ? Support design and analysis strategies through quality, commercial software – Approximation strategies – Design approaches ? DOE, optimization, decision support, Pareto frontiers, Fenyes4/14/04 Challenges to Widespread MAO Application ? Educational challenges – Educating corporations on optimization, then MAO – Educating the next generation of users and teachers ? Corporate cultural challenges – Organizing work for MAO ? Software challenges – Simpler to use, better GUI – More capable to handle distributed computation with broad range of analyses, database interaction, interactive data visualization, report generation, … 22 Fenyes4/14/04 MAO Software Challenges ? Data Storage, Management, Communication – Consistent information requires database storage and communication with disciplines – Standards will be required to drive this (e.g. http://www.omg.org/ ) ? Vehicle and Results databases used for: – Model building – Storing analysis results, history, gradients, approximations – Communication with commercial systems (ODBC, SQL) a must ? Full support for user defined design strategies, algorithms – Approximation strategies ? DOE, neural net, response surfaces, etc. ? Use gradients and Hessians as available – Simplify use of proprietary or other algorithms within commercial frameworks ? Data Transformations – Units, coordinate systems – Geometric relationships – Variable relationships ? Parametric, DV linking Fenyes4/14/04 Challenges in Automotive MAO ? Level of detail and complexity – When and how should MAO be used in the vehicle and component design processes? ? Inclusion of more disciplines that have a strong linkage to the vehicle or component design problem – Vehicle design ? Safety, reliability, aesthetics, vehicle dynamics, … 23 Fenyes4/14/04 Summary We have developed an MAO system for coarse balance and integration during the early vehicle development process which – Enables use of math based decision tools for vehicle and architecture design – Facilitates multidisciplinary analysis with consistent data ? Extends math based beyond engineering to manufacturing and business ? Provides consistent sharing of representation and analysis data through database ? Simplifies storage and access of analysis results through database and GUI – Quantifies discipline consequences of design and architectural changes Much work remains ! Fenyes4/14/04 Designing Great GM Cars and Trucks !