Future Plans Michael T Heath Professor and Director

Future Plans Michael T. Heath Professor and Director 12 September 2001 © 2001 Board of Trustees of the University of Illinois

2 Y 6 -10 n Multiscale simulations n Improved physical modeling n Multiphysics coupling n Verification and validation © 2001 Board of Trustees of the University of Illinois

3 Fluid Dynamics © 2001 Board of Trustees of the University of Illinois

Two-Phase Flow Burning Aluminum with Smoke Crude droplet burning model n No support for slag build-up n No subgrid effects for particles, combustion, etc. n © 2001 Board of Trustees of the University of Illinois 4

5 Validation of Rocturb Ad hoc turbulence injection model n No two-phase flow effects in model n Simple subgrid models n Inadequate agreement with experimental turbulence data n © 2001 Board of Trustees of the University of Illinois

6 Fluids — Multiscale Simulations n Spatial multiscale: Generalized LES l Further “Optimal LES” turbulence development l Combustion “LES” l Radiation “LES” l Lagrangian particles and super particles n Temporal multiscale l Time zooming formulations n High order, high resolution, optimal numerics (for LES) l Adaptive grids for turbulence simulation © 2001 Board of Trustees of the University of Illinois

7 Fluids — Improved Physical Modeling n Real gas effects l Thermal and chemical non-equilibrium l Nozzle and plume physics n Aluminum particle combustion l Based on detailed single-particle simulations n Acoustic coupling l Hydrodynamic instabilities l Combustion instabilities n Multi-phase flow l Slag accumulation l Agglomeration and break-up © 2001 Board of Trustees of the University of Illinois

8 Fluids — Multiphysics Coupling n Injection from combustion layer l Turbulent fluctuations l Aluminum droplets Particle combustion n Fluid-structure in deforming/complex geometry n l Inhibitor and crack propagation l Hybrid structured/unstructured grid Inhibitor and nozzle scouring and ablation n Grid generation, motion, and remeshing n © 2001 Board of Trustees of the University of Illinois

9 Fluids — V&V n Automated verification test suite l Regular periodic testing of developing code l Expanding domain of testing n Component model validation against DNS l LES & CPR l Particle combustion and single-particle DNS n Characterizing solution uncertainties l Uncertainties from modeling, input data, numerics l Integration of physical data n Simulation validation against available data l NASA, Thiokol, ONERA, … © 2001 Board of Trustees of the University of Illinois

10 Fluids — Experiments Detailed fluid measurements in combustion-driven flows Optical flow measurements in propellant combustion flow n Turbulence and particle distribution n © 2001 Board of Trustees of the University of Illinois

11 Combustion and Energetic Materials © 2001 Board of Trustees of the University of Illinois

12 CEM – Surface Propagation Current: single-valued Future: multivalued © 2001 Board of Trustees of the University of Illinois Level sets/ghost fluid? (collaboration with CS)

13 CEM – Turbulence Seeding Time-dependent mass flux from propellant provides boundary conditions for turbulent chamber flow © 2001 Board of Trustees of the University of Illinois

14 CEM — Multiscale Simulations n Average 3 -D condensed phase equation to get 1 -D model l Use Rocfire to modify Rocburn l Use DNS to construct models for averaged terms Turbulence seeding n Erosive burning n © 2001 Board of Trustees of the University of Illinois

15 CEM — Improved Physical Modeling n Complex kinetics l Gas-phase l Condensed-phase n Level-set methods for steep or over-hanging propellant surfaces l Fuel-rich packs l Aluminum pooling and injection © 2001 Board of Trustees of the University of Illinois

16 CEM — Multiphysics Coupling n Effects of flow on burning l Erosive burning l Ignition transients and flame spread Turbulence seeding n Injection of igniting aluminum droplets n Burning in cracks n Surface propagation n © 2001 Board of Trustees of the University of Illinois

17 CEM — V&V n Sandwich-propellant experiments l Flame structure l Surface structure l Regression rate n Composite-propellant data l Average regression rate l Unsteady response, pressure coupling © 2001 Board of Trustees of the University of Illinois

18 CEM — Multiscale Simulations Path integral Monte Carlo for thermal simulations of quantum systems n Ab initio molecular dynamics (Car. Parrinello) simulations of materials n Excited states and quantum reaction dynamics in molecules n Combining the methods n l Thermal simulations of matter using quantum Monte Carlo l First principles calculations of reaction dynamics l Density functional simulations for large systems © 2001 Board of Trustees of the University of Illinois

CEM — Multiscale Simulations Decomposition pathways of ethylene Shock data for hydrogen comparison of PIMC simulations with NOVA laser shock experiments and SESAME EOS. Longer time scales Science 281, 1178, 1998. Embedding High pressure nonmolecular phase of nitrogen - density functional simulation - compared with experiments Nature 411, 170 (2001) Extend to heavier atoms Force models Continuum mechanics Longer time scales

20 CEM — Improved Physical Modeling n Improved methods l Improved functionals using quantum Monte Carlo methods l Extending quantum Monte Carlo to heavier atoms n Embedding of methods for multiscale simulations l Reaction dynamics calculations embedded in regions simulated by density functionals and continuum mechanics n Time scale - the hardest problem l Hyperdynamics and beyond for atomic scale simulations l Integration into kinetic Monte Carlo for reactions © 2001 Board of Trustees of the University of Illinois

21 Structures and Materials © 2001 Board of Trustees of the University of Illinois

Constitutive Modeling of SP n Currently: Arruda-Boyce l 8 -chain rubber elasticity model 30 rubbery materials l Nonlinear kinematics l Material stiffening l Simplicity n Still missing l Rate dependence (Bergstrom- Boyce) l Microstructural damage and fracture mechanisms Ø Dewetting Ø Void growth Ø Binder tearing l Integration of multiscale models Nominal stress (MPa) l Developed for unfilled and filled 20 10 1. 0 2. 0 3. 0 Stretch 4. 0 5. 0

23 Structures — Multiscale Simulations Molecular, granular, and structural modeling of case and SP n Adaptivity in Rocsolid and Rocfrac n l Spatial adaptivity through mesh refinement l Dynamic remeshing and repair l Dynamic insertion of cohesive elements n Implementation of 4 -D space-time FE scheme © 2001 Board of Trustees of the University of Illinois

Structures Improved Physical Modeling Implement historydependent constitutive models in ALE formulation n Extend SP constitutive theory to failure regime n Modeling of nozzle ablation and scouring n © 2001 Board of Trustees of the University of Illinois 24

25 Structures — Multiphysics Coupling Mesoscale structure/combustion modeling of advancing crack n Macroscale modeling of fracture events with Rocflu/Rocfrac n l Mesh generation n Implicit/explicit coupling strategies © 2001 Board of Trustees of the University of Illinois

26 Structures — V&V n Benchmark verification problems l Expand library of known benchmark problems l Develop methods to create new benchmark problems to test complicated modules (e. g. , ALE) l Verification of coupled codes n Validate codes l Thiokol failure data on model SP motors l Lab-scale rocket data l Quasi-static and dynamic properties of SP and case (D 6 AC steel) under tension and compression n Methods to deal with model uncertainty and sensitivity © 2001 Board of Trustees of the University of Illinois

27 Computer Science © 2001 Board of Trustees of the University of Illinois

28 Software Framework Challenges n Full exploitation of object decomposition l Load balance within individual phases l Source code changes for AMPI n 10 K-100 K processors l Inadequate load balancing strategies l Automatic optimization of object communication n Component frameworks l Dynamic insertion support for FEM l Component frameworks for other common abstractions n Integration of modules l Complex orchestration code l Optimized parallel data exchange © 2001 Board of Trustees of the University of Illinois

29 CS — Frameworks n Software integration l Intelligent interface between modules l Intelligent handling of dynamic behavior Ø Adaptivity Ø Load-balancing l Support for choice of coupling strategies l Support for spawning subscale simulations n Solid Combustion Fluid HDF IO Roccom Interface Components l Application-aware data structures Ø FEM, multiblock, AMR, particles, … © 2001 Board of Trustees of the University of Illinois Orchestration

30 CS — Computational Infrastructure n Computational steering l Autopilot n On-the-fly performance analysis and visualization l n n n © 2001 Board of Trustees of the University of Illinois Pablo Panda/Globus integration for Grid Remote visualization of ultra large data sets Scalability issues for 100 K processors

Challenges in Computational Mathematics and Geometry n n n Independent meshing capability Mesh adaptation and repair Robust and efficient propagation of moving interface in 3 -D Interpolation extended to 3 -D for remeshing, etc. Robust and scalable solvers for ill-conditioned, nonsymmetric systems © 2001 Board of Trustees of the University of Illinois 31

CS Improved Computational Techniques n Computational geometry l Dynamic mesh repair l On-the-fly remeshing l Interpolation between 3 -D meshes l Interface surface propagation in 3 -D n Computational mathematics l Iterative solvers for ill-conditioned systems l Application-specific preconditioners l Grid-based solvers © 2001 Board of Trustees of the University of Illinois 32

33 Integrated Simulations © 2001 Board of Trustees of the University of Illinois

34 GEN 3 n Multiscale simulations l Homogenization l Time zooming l Dynamically-spawned subscale simulations n Improved physical modeling l Propellant ignition and tailoff l Real gas effects l Slag formation l Nozzle and inhibitor ablation and scouring l Plume l Environmental effects © 2001 Board of Trustees of the University of Illinois

35 GEN 3 n Multiphysics coupling l Modules coupled using SWIFT l External modules (inbound) l Technology transfer (outbound) n V&V l Normal burns l Accident scenarios l Thrust vector control (3 -D effects) l Additional rocket designs Ø NASA five-segment Shuttle booster © 2001 Board of Trustees of the University of Illinois

© http: //www. csar. uiuc. edu © 2001 Board of Trustees of the University of Illinois

37 Michael T. Heath, Director Center for Simulation of Advanced Rockets University of Illinois at Urbana-Champaign 2262 Digital Computer Laboratory 1304 West Springfield Avenue Urbana, IL 61801 USA m-heath@uiuc. edu http: //www. csar. uiuc. edu telephone: 217 -333 -6268 fax: 217 -333 -1910 © 2001 Board of Trustees of the University of Illinois