Disrupting Computer Aided Engineering Key Building Blocks of Accurate Interactive Engineering Solutions

Similar documents
How to Optimize Geometric Multigrid Methods on GPUs

Introduction to Multigrid and its Parallelization

Efficient Finite Element Geometric Multigrid Solvers for Unstructured Grids on GPUs

Accelerated ANSYS Fluent: Algebraic Multigrid on a GPU. Robert Strzodka NVAMG Project Lead

Towards a complete FEM-based simulation toolkit on GPUs: Geometric Multigrid solvers

Many Roles and Point Solutions are Used in Product Development

Automatic Generation of Algorithms and Data Structures for Geometric Multigrid. Harald Köstler, Sebastian Kuckuk Siam Parallel Processing 02/21/2014

WHITE PAPER Simulation-Driven Product Development: Will Form Finally Follow Function?

Introduction to C omputational F luid Dynamics. D. Murrin

National Cybersecurity Center of Excellence

The Experience of Prototyping. Jennifer Kalz, CBAP, CUA, CSM March 28, 2017

A Collaborative User-centered Approach to Fine-tune Geospatial

Accelerated tokamak transport simulations

Simcenter Motion 3D. Mechatronics - Improve Design Dynamics Performance: Combine 3D Multi-Body Simulation with 1D Actuation & Controls Simulation

Free-Form Shape Optimization using CAD Models

Multigrid Pattern. I. Problem. II. Driving Forces. III. Solution

UMKC Augmented and Virtual Reality Innovation Academy Spring 2019 Sessions

INTEROPERABILITÀ MODELLAZIONE PARAMETRICA- DIRETTA IN CREO

Accelerating image registration on GPUs

Creo Elements/Pro Advanced XE

computational Fluid Dynamics - Prof. V. Esfahanian

Accelerating Double Precision FEM Simulations with GPUs

User-Centered Analysis & Design

THE BENEFIT OF ANSA TOOLS IN THE DALLARA CFD PROCESS. Simona Invernizzi, Dallara Engineering, Italy,

Recent Approaches of CAD / CAE Product Development. Tools, Innovations, Collaborative Engineering.

ORACLE SERVICES FOR APPLICATION MIGRATIONS TO ORACLE HARDWARE INFRASTRUCTURES

THOMAS LATOZA SWE 621 FALL 2018 DESIGN ECOSYSTEMS

Virtual EM Inc. Ann Arbor, Michigan, USA

Version 11

Fundamentals of Modeling with Simcenter 3D Robin Boeykens

Enabling Efficient Optimization / Sensitivity and Robustness Analysis for Crashworthiness, NVH, and Multi-disciplinary Concept Assessments

Very fast simulation of nonlinear water waves in very large numerical wave tanks on affordable graphics cards

NVIDIA DGX SYSTEMS PURPOSE-BUILT FOR AI

On Level Scheduling for Incomplete LU Factorization Preconditioners on Accelerators

3D Helmholtz Krylov Solver Preconditioned by a Shifted Laplace Multigrid Method on Multi-GPUs

CAD TRENDS & CREO TECHNOLOGY ROADMAP. Paul Sagar VP, CAD Product Management

Accelerating the Business Value of Virtualization

Autodesk Moldflow Adviser. Design plastics confidently.

Center for Computational Science

Multigrid Solvers in CFD. David Emerson. Scientific Computing Department STFC Daresbury Laboratory Daresbury, Warrington, WA4 4AD, UK

Combinatorial meets Convex Leveraging combinatorial structure to obtain faster algorithms, provably

Simulation of In-Cylinder Flow Phenomena with ANSYS Piston Grid An Improved Meshing and Simulation Approach

Paper. Delivering Strong Security in a Hyperconverged Data Center Environment

Design, prototyping and construction

ACCELERATING CFD AND RESERVOIR SIMULATIONS WITH ALGEBRAIC MULTI GRID Chris Gottbrath, Nov 2016

Developments in Topology Optimization for Additive Manufacturing

T6: Position-Based Simulation Methods in Computer Graphics. Jan Bender Miles Macklin Matthias Müller

Autodesk Alias Design Drives Success

Modernizing Healthcare IT for the Data-driven Cognitive Era Storage and Software-Defined Infrastructure

Process of Interaction Design and Design Languages

Towards ParadisEO-MO-GPU: a Framework for GPU-based Local Search Metaheuristics

Experiences with GPGPUs at HLRS

The Power to Prototype

Topology-Optimization for internal Fluid Flow: TOSCA Fluid FE-DESIGN the optimization company

Design of Finite Element Software for Modeling Bone Deformation and Failure

Hardware Modeling using Verilog Prof. Indranil Sengupta Department of Computer Science and Engineering Indian Institute of Technology, Kharagpur

Introduction to Computational Fluid Dynamics Mech 122 D. Fabris, K. Lynch, D. Rich

Why Use the GPU? How to Exploit? New Hardware Features. Sparse Matrix Solvers on the GPU: Conjugate Gradients and Multigrid. Semiconductor trends

Introduction. Chapter 1. What Is Visual Modeling? The Triangle for Success. The Role of Notation. History of the UML. The Role of Process

Oracle and Tangosol Acquisition Announcement

ANSYS HPC Technology Leadership

Available online at ScienceDirect. Procedia CIRP 60 (2017 ) th CIRP Design 2017

1 Past Research and Achievements

SEMANTIC NETWORK AND SEARCH IN VEHICLE ENGINEERING

Linear Algebraic Representation of Big Geometric Data

Efficient AMG on Hybrid GPU Clusters. ScicomP Jiri Kraus, Malte Förster, Thomas Brandes, Thomas Soddemann. Fraunhofer SCAI

Connected Intelligence

Performance and accuracy of hardware-oriented. native-, solvers in FEM simulations

GRIDS INTRODUCTION TO GRID INFRASTRUCTURES. Fabrizio Gagliardi

Outline. follows the structure of the report

The LUCID Design Framework (Logical User Centered Interaction Design)

Package on Board Simulation with 3-D Electromagnetic Simulation

The Finite Element Method

Digital Enablement bridging the digital divide

The Infinite Network. Built on the Science of Simplicity. Everywhere. Always. Instantly.

Interaction design. The process of interaction design. Requirements. Data gathering. Interpretation and data analysis. Conceptual design.

HPC and IT Issues Session Agenda. Deployment of Simulation (Trends and Issues Impacting IT) Mapping HPC to Performance (Scaling, Technology Advances)

IMECE FUNCTIONAL INTERFACE-BASED ASSEMBLY MODELING

CSE 591: GPU Programming. Introduction. Entertainment Graphics: Virtual Realism for the Masses. Computer games need to have: Klaus Mueller

Generic finite element capabilities for forest-of-octrees AMR

Recent Advances in Modelling Wind Parks in STAR CCM+ Steve Evans

OzenCloud Case Studies

DICOM Educational Conference Brisbane, Australia

ECE 8823: GPU Architectures. Objectives

Design Rules. increasing generality. increasing authority. Guide lines. Standards. increasing authority. increasing generality

pre- & post-processing f o r p o w e r t r a i n

Exploring unstructured Poisson solvers for FDS

Supporting Simulations to Guide Engineering Design

Future Grid Initiative Technology Challenges in Designing the Future Grid to Enable Sustainable Energy Systems

What s inside: What is deep learning Why is deep learning taking off now? Multiple applications How to implement a system.

T A C. Total. Taking Design Visualization to the Next Level

Resilient geometric finite-element multigrid algorithms using minimised checkpointing

Solving Large Complex Problems. Efficient and Smart Solutions for Large Models

Numerical Algorithms on Multi-GPU Architectures

Adaptive Mesh Astrophysical Fluid Simulations on GPU. San Jose 10/2/2009 Peng Wang, NVIDIA

RBF Morph An Add-on Module for Mesh Morphing in ANSYS Fluent

Reconstruction of Trees from Laser Scan Data and further Simulation Topics

Space Filling Curves and Hierarchical Basis. Klaus Speer

Multigrid Algorithms for Three-Dimensional RANS Calculations - The SUmb Solver

HPC Computer Aided CINECA

Transcription:

Disrupting Computer Aided Engineering Key Building Blocks of Accurate Interactive Engineering Solutions

Simulation is evolving from a trouble shooting tool to adding customer value in the form of digital twins Model Pioneers CAx Sunrise PLM Pervasion Digital Twin Era Timeline ~1985 ~2000 ~2015 Scientific experts use models Understanding of phenomena Computer technology aid in product design and engineering Model-based Systems Engineering Key for communication Combining the virtual and physical world Bridge value chains Page 2

Accurate Interactive Engineering Solutions Table of content CAE as we know it today 3 Are Interactive Tools the Future of CAE? 7 Key Building Blocks of Interactive CAE Tools 12 Interactive CAE Experience it yourself! 21 Interactive CAE Conclusions 28 Page 3

Computer aided engineering (CAE) solutions are a key tool in any R&D industrial department User Base More users Today CAE solutions are used by 10-20% of all engineers 1 Model Fidelity Computational Efficiency Highly integrated multi-physics simulations can accurately predict any phenomena 3. Bigger & more accurate Faster Moore s law for hardware and algorithms have led to exponential growth in efficiency 2. Page 4 1) Whitepaper of Mentor Graphics: Democratization of CFD (link) 2) SIAM Research and Education in Computational Science and Engineering (link) 3) NAFEMS The NAFEMS Simulation Capability Survey 2015 (link)

World of engineering is transforming and computer aided design and engineering is facing new challenges Need to run "bigger", "faster", and "more accurate" CAE simulations by engineers who have less deep CAE expertise The adoption of computer-aided engineering (CAE) is marred by complex user interfaces, legacy codes, large capital investments and more. 2 Democratization of CFD is indeed possible, but to get from 5-10% market penetration to 80-90% in the next 30-40 years will require more thinking out of the current box and probably radical rethinking of CFD code structures with more fundamental research. 3 Democratization of CFD from Analysts to Designers Page 5 1) NAFEMS Simulation 20/20 Vision: Democratization - Accessibility (link) 2) Webinar of the ASSESS Initiative (link) 3) Whitepaper of Mentor Graphics: Democratization of CFD (link)

Today s complexity requires a new attitude as well as new computer aided engineering paradigms Example: Ideation for Additive Manufacturing (AM) along the GenePlore model 1 We need to redesign a burner tip for AM Product Constraints Simulation Experiment Generation of Preinventive Structures Focus or Expand Concept Preinventive Exploration and Interpretation 1) T. Ward, S. Smith, R. Finke: Creative Cognition: Theory, Research, and Applications. Bradford Books. 189-21, 1996 Page 6

Accurate Interactive Engineering Solutions Table of content CAE as we know it today 3 Are Interactive Tools the Future of CAE? 7 Key Building Blocks of Interactive CAE Tools 13 Interactive CAE Experience it yourself! 22 Interactive CAE Conclusions 29 Page 7

New innovations are driving CAE from an expert centric approach to true democratization of CAE Product complexity is exploding Democratization of CAE is taking place Early analysis is getting momentum Computational power is exploding Virtual reality is entering professional market Interactive collaborative?????? worlds??? are? key?? to? innovation????? Simulation is becoming challenging attractive accessible interactive immersive 2000 2012 2015 2016 2017 Page 8

Are interactive tools the future of computer aided engineering? Today Simulation is restricted to dedicated simulation experts in late R&D phases Tomorrow Boost simulation in early phases by intuitive virtual design worlds for everyone Research question Are virtual worlds suitable for conceptual mechatronic design, analysis and testing? Page 9

Today s complexity requires new computer aided engineering paradigms: Interactive Virtual Design Worlds Sketch 3D CAD models in virtual spaces instead of lengthy CAD operations Meet, design and try out. Together, in virtual reality. With real physics, inside your design. Computer aided creativity instead of computer aided documentation and validation Interactive and concurrent physics: Immediate evaluation in virtual reality instead of waiting for hours classical CAE tools to finish Page 10

We enhance virtual worlds from a pure visual experience to where one can create and analyze designs by realtime solvers Innovation Enabler Benefit Ultra-fast Solvers Physics Engines Out of the box and great for 0 order 3D physics multigrid solver on GPUs more than 100 x faster Large FE usable for immediate feedback Realtime results for large FE models Gamification Approach Industrialization Take the best you find User Experience Interaction concepts CAD world combined with computer graphics voxel world Inspirations from fast moving market Proven concepts Create reuseable geometries Immersive Technologies Latest Virtual Reality HTC Vive Latest Augmented Reality MS HoloLens Intuitive 3D display of 3D data Reuse for stakeholders, marketing Cheap HW Page 11

Accurate Interactive Engineering Solutions Table of content CAE as we know it today 3 Are Interactive Tools the Future of CAE? 7 Key Building Blocks of Interactive CAE Tools 12 Interactive CAE Experience it yourself! 21 Interactive CAE Conclusions 28 Page 12

Ultra-fast background simulation is the key technology for the next generation of virtual representation. High fidelity 3D models More than1.000.000 degrees of freedom Focus: Design & Engineering Challenge: State-of-use vs. state-of-art technologies Established tools rely on general purpose solvers Exploitation of novel computing architectures Our Innovation: Interactive computer aided engineering tools 500x faster than state-of-art solvers Key Building Blocks: 1 Efficient voxel-based discretizations 2 Cut-cell methods Novel integration schemes 3 Multi-grid solvers 4 5 Hardware-aware algorithms Page 13

Structured hexahedral discretizations (Voxels) originating from computer graphics 1 offer highly efficient implementations 1 Replace conformal elements by simple structured elements vs + + + - Efficient implementation Very little overhead for data-structures Highly efficient memory layouts possible Less accurate due to non-conformity 1) C. Dick, J. Georgii, R. Westermann (2011): A Real-Time Multigrid Finite Hexahedra Method for Elasticity Simulation using CUDA (preprint) Page 14

Cut cell methods 1 relying on Nitsche s method allow to efficiently handle non-conforming geometries 2 Realize boundary conditions weakly via Nitsche s method Enforce Dirichlet boundary conditions via an additional potential where Classical approach yielding the following FEM discretization + + - Little computational overhead Accuracy close to conformal approaches Surface integrals are difficult to evaluate efficiently Additional terms due to Nitsche s method P. Hansbo, M. G. Larson, K. Larsson (2017) Cut Finite Element Methods for Linear Elasticity Problems (preprint) Page 15 Nitsche method

Multi-level surface integrations allow efficient realization of cut-cell methods on GPUs 3 Voxel-based surface integration One finite element cell vs Instead of classical methods, e.g. marching cubes, multi-level voxels are used to determine volume integrals. Appropriate choice of voxels and integration weights ensure convergence Absolute Error + + Fast evaluation (also on GPUs) First order convergence Grid Size Dennis Hechler (2017): Voxel-basierte Cut-Cell Verfahren für lineare Elastizität. Master Thesis, U Münster Page 16

Simple iterative matrix-free solvers efficiently damp high frequency errors Iterative matrix-free solvers Solutions of r x = r b can be approximated iteratively by r x = f ( r x i + 1 i ) e.g. the Jacobi scheme with D r x i + 1 = - ( L + U ) r x i + r b + - Efficient smoothing of high frequency errors Slow convergence Page 17

Geometric multi-grid methods have an optimal complexity 4 Geometric multi-grid method Algorithm: Hierarchy of grids, starting on finest level Few iteration of a simple iterative scheme Error is restricted to the coarser grid On the coarsest level a direct solver is used Solutions are prolonged back to finest grid + + Straight forward to implement Optimal complexity (linear in # unknowns) Page 18

Graphics processing units (GPU) allow additional speedup 5 Hardware-aware solvers Adapt algorithm and memory-layout to specific hardware architecture to exploit massive parallelism - CPU Multi-grid solver - Direct solver - GPU Multi-grid Solver CPU - efficient memory layout error GPU - efficient memory layout Speed up by multigrid (100x) Speed up by GPU (5-10x) time (s) + Interactive 3D simulations with engineering accuracies Page 19 S. Gavranovic, D. Hartmann, U Wever (2015): Topology Optimization using GPGPU, EUROGEN 2015

Voxel-based multi-grid methods leveraging cut cell methods open up the vista of interactive computational engineering error 2 Nitsche s Method - CPU Multi-grid solver - GPU Multi-grid solver - CPU Multi-grid with Nitsche - Direct solver 1 Voxel- Discretization accuracy 3 4 Speed up by multi-grid (100x, optimal complexity) 5 Speed up by GPU (5-10x) Conformal Discretization accuracy time (s) S. Gavranovic, D. Hartmann, P. Stelzig (2017): Accurate Interactive 3D Engineering Simulations Accelerated by GPU. NAFEMS WC 2017 Page 20

Accurate Interactive Engineering Solutions Table of content CAE as we know it today 3 Are Interactive Tools the Future of CAE? 7 Key Building Blocks of Interactive CAE Tools 13 Interactive CAE Experience it yourself! 21 Interactive CAE Conclusions 28 Page 21

FEM solver case study Geometry FEM with146k cells Interactive FEM with 1500k cells Page 22

FEM solver case study 8% deviation from FEM 300x faster FEM with146k cells Interactive FEM with 1500k cells FEM with146k cells Interactive FEM with 1500k cells Page 23

Today s complexity requires new computer aided engineering paradigms: Interactive Virtual Design Worlds Design 3D CAD models in virtual spaces instead of lengthy CAD operations Meet, design and try out. Together, in virtual reality. With real physics, inside your design. Computer aided creativity instead of computer aided documentation and validation Interactive and concurrent physics: Immediate evaluation in virtual reality instead of waiting for hours classical CAE tools to finish Page 24

Challenge: FEM for non-experts Interactive FEM in Virtual Reality Interactive and concurrent physics: Immediate evaluation in virtual reality instead of waiting for hours classical CAE tools to finish Page 25

Challenge: Optimal design structures within seconds Interactive Design Assistance Computer aided creativity instead of computer aided documentation and validation Page 26

Challenge: Conceptual design task in a virtual world Interactive Collaborative Spaces Meet, design and try out. Together, in virtual reality. With real physics, inside your design. Page 27

Accurate Interactive Engineering Solutions Table of content CAE as we know it today 3 Are Interactive Tools the Future of CAE? 7 Key Building Blocks of Interactive CAE Tools 12 Interactive CAE Experience it yourself! 21 Interactive CAE Conclusions 28 Page 28

Interactive simulation is not meant to replace today s solution but to expand simulation assistance along the whole life cycle Established simulation tools to validate and tune designs Autonomous simulation based design assistance without any user inputs to boost innovative concept designs Machines with build in simulations / digital twin to make informed decisions autonomously Ideation Realization Utilzation Page 29

If you want to know more Dirk Hartmann CT RDA AUC Phone: +49 173 2537709 E-mail: hartmann.dirk@siemens.com Stefan Gavranovic Theo Papadopoulos Philipp Stelzig Utz Wever Internet siemens.com/corporate-technology Page 30