Miguel Aguilo, Ted Blacker, Andre Claudet, Brett Clark, Ryan Rickerson, Josh Robbins, Louis Vaught, and Tom Voth Sandia National Laboratories

Transcription

1 Miguel Aguilo, Ted Blacker, Andre Claudet, Brett Clark, Ryan Rickerson, Josh Robbins, Louis Vaught, and Tom Voth (as told by Corbett Battaile) Sandia National Laboratories Sandia National Laboratories is a multi-program laboratory managed and operated by Sandia Corporation, a wholly owned subsidiary of Lockheed Martin Corporation, for the U.S. Department of Energy s National Nuclear Security Administration under contract DE-AC04-94AL85000.

2 Example: Lantern Bracket 3

3 A Revolution in Design and Fabrication

4 Introduction to This Revolution Affordable Agile Assured

5 Enormous Opportunity (with Challenges) In-Situ Material Qualification Profoundly New Design Flexibility

6 Inverting Conventional Design Paradigms Embedded Uncertainty Quantification Powder Bed Metal Additive Topology Optimization with Material Locality

7 Inverting Conventional Design Paradigms Specify Form Design Verify Function Using FEA Specify Design Domain and Function Use Topology Optimization (FEA) to Determine Form that Meets Function Optimized Design (Form)

8 Mechanical Performance Design by Performance Prioritization Pareto Suite of Topologies Input Infeasible Horizontal Thermal Vertical Feasible Displacement Response Thermal Performance

9 Inverting Manufacture Complexity is Free Preferred Minimal Waste Fast Design To Manufacture Mixed and Graded Materials

10 Inverting Qualification Point-Wise Material Variability Design Must Adjust Accordingly Cost (Compute) Is a Critical Issue In-Situ Metrology to Validate

11 Making Resource Requirements Reasonable - Linear Statics - 1.5M elements - 1 Objective (Maximize Stiffness) - 1 Loading Condition CPU TIME: Hours 12 Minutes 2000 Weeks! Hours 1 High Fidelity FEM 12 Minutes on 4196 Processors X 2 Evaluations Per Iteration Optimization X 50 Iterations Uncertainty Quantification X 100 Samples 12

12 Objective Function Value Physics-Based Reduced Order Modeling 8 p* = arg min z J(p) s.t. g(p)=0; h(p) 0 (pk)u=f Create ROM φ T (ψp)kϕu=ϕ T f ψp* = arg min ψp J(ψp) s.t. g(ψp)=0; h(ψp) 0 Until error > tolerance (pk)u=f 1 Create ROM φ T (ψp)kϕu=ϕ T f (pk)u=f ψp* = arg min ψp J(ψp) s.t. g(ψp)=0; h(ψp) 0 Until error > tolerance Stiffness 7/10

13 Smart Sampling Techniques p* = arg min p E[U T (pk(g,b))u] s.t. (pk(g,b))u f = 0; V(p) V 0 8 Density of Z Voronoi Cells {Z} 1 {Z} 2 {Z} 3 Build SROM Given Available Resources (# FEM Runs That Can Be Performed) {Z} i = {Shear Modulus (G), Bulk Modulus (B)} 1 i M, M = #Samples /10

14 Benefits of Surrogate-Based Optimization - Linear Statics - 1.5M elements - 1 Objective (Maximize Stiffness) - 1 Loading Condition 5 Hours CPU TIME 1 High Fidelity FEM 3 Minutes on 4196 Processors X 2 Evaluations Per Iteration Optimization X 50 Iterations Uncertainty Quantification X 5 Samples 15

15 How Does This All Tie Together? 16

16 Diverse Tools and Expertise Necessary Sandia Analysis Workbench DOORS, Genesys, SysML Primary Standards Laboratory Product Definition & Config. Control Dakota QMU Toolkit Predictive Perfomance AM Topology Margins (ICME) Optimization Component Project Designers Sandia Analysis Workbench CUBIT Geometry & Meshing Tools SCULPT Meshing Technology Predictive Perfomance Margins (ICME) SIERRA Simulation Tools (ICME) Advanced AM Hardware & Diagnostics (ICME) 18

17 Optimization in Action

18 Entire Bracket-Mass Assembly Example: Bracket Redesign Bracket System to be Optimized 26

19 Example: Bracket Redesign Rapid Design Response to Source Requirement Change 27

20 Example: Bracket Redesign Refinements on Design with Options on Mass Mounting Tall Bracket Option High Mass Mount Tall Bracket Option Low Mass Mount 28

21 Moving Forward: PLATO Technology Plan Careful Crafting Expected Outcomes: Modern Design/ Analysis Environment Clean, Smooth, Connected Shapes Fast Convergence Interactive Speed & Control Robust designs Directly Printable Output Interface to CAD 29

22 Inverting Conventional Design Paradigms In-Situ Validation of Materials and Processes Powder Bed Metal Additive Topology Optimization with Material Locality Disruptive Change for NW Disruptive Change for SNL Disruptive Change for the World of Engineering 30