1. Simulering av granulära material GRAM – Gruvprogram inom ProcessIT Dr Martin Servin UMIT Research Lab / Department of Physics 2010-06-03

2. The challenge simulation tool for -understanding -optimization -re-design requires -fast large-scale simulations -suitable models -analysis of large data sets

3. The challenge

6. Teaser

7. 900.000 particles, water, ~3m, 10/1 sim ratio

8. Teaser

10. Overview Background The processes Project status – results Outlook People

11. Background UmU: research on high-performance physics based visual simulation Oryx Simulations: making heavy machinery training simulators (1998) Algoryx Simulations: spin-off company -> AgX Multiphysics Toolkit (2006) Volvo CE: optimization/re-design of loading / digging LKAB: optimization/re-design of balling plant ProcessIT pre-study (PellIT - 2009) Research projects: Algoryx, Fraunhofer, Univ Kaiserslauten, LKAB, Oryx, ProcessIT, UmU, Volvo CE (2010 - 2013)

12. The processes Balling plant (LKAB) Wheel loader (Volvo CE / Oryx)

13. Balling plant Mono-sized spherical pellets Understand flow patterns and forces Outlet design Control variables Agglomeration process 100K – 100M particles Time scales 1ms – 5min In: - fines (ore + binding) - undersized pellets Out: - green pellets Return: - undersized pellets - oversized pellets (crush) 100 ton/h

14. Wheel loading Optimal loading, digging, tillage Minimize time, fuel, wear on tool and wheels Maximize operator comfort and support Re-design of load bucket and motion control Coupling AgX simulations, operator VE with Volvo CEs Simulink drive train models Shmulevich et al, Terramechanics (2007)

15. Project status – pellet part Fast large-scale simulations stable DAE integrator at large timestep (100%) parallel solver for large sparse MLCPs (25%) merge/split acceleration (50%) adaptive resolution (10%) contact reduction (25%) Suitable models Analysis Optimization

16. Project status – pellet part Fast large-scale simulations stable DAE integrator at large timestep (100%) parallel solver for large sparse MLCPs (25%) merge/split acceleration (50%) adaptive resolution (10%) contact reduction (25%) Suitable models rigid body discrete elements (75%) particle fluid / SPH (25%) hybrid (5%) simplified dry friction model (10%) Analysis Optimization

17. Project status – pellet part Fast large-scale simulations stable DAE integrator at large timestep (100%) parallel solver for large sparse MLCPs (25%) merge/split acceleration (50%) adaptive resolution (10%) contact reduction (25%) Suitable models rigid body discrete elements (75%) particle fluid / SPH (25%) hybrid (5%) simplified dry friction model (10%) Analysis Model identification (25%) Simulation validation (5%) Flow patterns and forces (5%) Optimization

18. Project status – pellet part Fast large-scale simulations stable DAE integrator at large timestep (100%) parallel solver for large sparse MLCPs (25%) merge/split acceleration (50%) adaptive resolution (10%) contact reduction (25%) Suitable models rigid body discrete elements (75%) particle fluid / SPH (25%) hybrid (5%) simplified dry friction model (10%) Analysis Model identification (25%) Simulation validation (5%) Flow patterns and forces (5%) Optimization Outlet design (0%) Control variables (0%)

19. Project status – pellet part

28. Outlook Market for dynamical simulation tools for granular matter and machines? New projects? optimal design of off-road vehicle boggies and tracks - SLU dynamic load forces and mechanical wear on tools for excavation, mining, forestry – LTU

29. Staff Researchers Dr Martin Servin, UMIT / Department of Physics Dr Claude Lacoursière, UMIT / HPC2N PhLic Kenneth Bodin, UMIT / HPC2N Prof Mats G Larson, UMIT / Department of Mathematics Prof Bo Kågström, UMIT / Department of Computing Science / HPC2N PhD, project assistants and master thesis Pengfei Tian, Mona Forsman, Stefan Hedman, Olof Sabelström, Adam Sernheim, John Nordberg Algoryx, Oryx, Volvo CE LKAB: Kent Tano, Kjell-Ove Michelsson

30. Thank you!

31. UMIT Research Lab En strategisk satsning inom beräkningsteknik, visuell simulering och optimering bas i framstående grundforskning teknikvetenskaplig forskning tvärdisciplinära frågeställningar industriella tillämpningar och innovationer 40 MSEK på 5 år Umeå universitet, Balticgruppen, Umeå kommun, EU Mål-2 och samarbete med ProcessIT Innovations,

32. UMIT Research Lab - Relevans och aktualitet Problem ModelSimulationResultsOptimization Konvergens i metoder och verktyg Parallella revolutionen Dynamiskt skalbar IT-infrastruktur BeräkningIT-Infrastruktur Programvara & hårdvara Visualisering & interaktion

33. UMIT Research Lab - Satsningens komponenter Fysisk och organisatorisk etablering Rekrytering – 2 tenure track Ass. Prof. 6-10 postdoc/doktorandpaket Projektmedel & utrustning Samtidigt rekrytering av forskningsledare inom datorgrafik och visualisering Utrustning: display, interaktion, simulatorer, programvaror, datorer, 3d-skrivare, 3d-scanners mm

34. UMIT Research Lab - delområden idag Computational design optimization – M Berggren Computational mathematics – M Larson Control system – A Shiriaev/L Freidovich Flexible and scalable IT infrastructures (Grids & clouds) – E Elmroth Interactive multiphysics and complex mechanical systems – M Servin/C Lacoursière IT management – J Holmström Parallel and scientific computing – B Kågström

35. UMIT Research Lab - Tillämpningsprojekt - ett urval Modeling and simulation of granular matter and machines Volvo CE, LKAB, Algoryx, Oryx Inverse Problems for Electromagnetics Valutec AB, SP Trätek Optimal Control and Design DAS Audio SA Numerical Algorithms for Stabilization of Linear Systems with Periodic Coefficients - German Aerospace Center (DLR) Simulator Based Design Komatsu Forest Effektivt nyttjande av muticore och parallellism IBM, Intel, Microsoft, Nvidia, Sony Ericson Migration of large-scale virtual machines SAP Research Simulation of fluids in bearings SKF

36. Recruitments – 10 Associate professor with tenure track Industrial and applied mathematics Parallel and multicore computing Industriell ekonomi – logistik, kombinatorisk optimering, riskanalys. Energiteknik – energieffektivisering, drivmedel. Teknisk fysik - tillämpad spektroskopi och detektion. Interaktion och design Teknisk datavetenskap - industriell programvaruteknik, flexibel och skalbar IT-infrastruktur Bioteknik Miljösystemteknik – miljökemi/systembiologi, livscykelanalys Automation/Reglerteknik