S IMULATIONS O F 1.5 MW O XY -C OAL F IRED F URNACE U SING STAR-CCM+ AND ARCHES C OUPLING A PPROACH O SCAR H. D IAZ -I BARRA - M ICHAL H RADISKY - J ENNIFER S PINTI - P HILIP J. S MITH STAR Global Conference Prague March 7, 2016
Mission of CCMSC - Demonstrate exascale computing with the use of Verification, Validation and Uncertainty Quantification (VV&UQ) to accelerate deployment of technology for reduced cost and emission of electric power generation to meet US and world’s energy needs Motivation - Cessation of nuclear testing by Department of Energy’s (DoE) National Nuclear Security Administration (NNSA) - Need to use simulation and modeling tools and capabilities to support assessment and certification of the nuclear weapons stockpile - NNSA’s Predictive Science Academic Alliance Program (PSAAP II) - Develop the science and engineering models and software for large-scale simulations utilizing methods of verification and validation and uncertainty quantification - Additional focus on extreme-scale computing - Goal: enable scientists to make precise statements about the degree of confidence in simulation-based predictions where experimental data is not feasible
objectives science of simulation Demonstration-scale prediction 1000 MWe 500 MWe USC oxy-AUSC V&V/UQ DOE - NNSA - PSAAP II 8-corner Design • $25 million over 5 years Boiler Boiler Phil Sean • exascale tools creation Pilot-scale validation 15 MWth oxy-coal boiler validation / UQ surrogate model experimental P Jeremy data r verification & sim o v i d Lab-scale e 1.5 MWth oxy- validation 5 MWth oxy- e Predictive Exascale coal furnace gas furnace v i Science d Computing e n c radiation e 91m Jennifer o f Figure 6. Size comparison p of coal fired boilers for DMAV r particle TASC Bench-scale ash e 550 MW output soot formation radiation validation d EDSL combustion transformation 63m i c t i AUSC- v multiphase i t y flow exascale runtime LES 31m environment multiphase devolatilization char oxidation flow Tom Sean Andy David Jeremy Tower Two pass Horizontal Michael
physical scale
why coal? U.S. Electric Power Generation (%) 4677Mt (global) Coal 7794Mt 7823Mt Diesel / Heating Oil (global) (global) Gasoline Propane Natural Gas 0 55 110 165 220 Pounds of CO2 emitted per million Btu of energy coal gas nuclear hydro, renewable & other
CO2 Emissions (metric tons per capita) - 2011-2015 all fuel sources (solid, liquid, gas) 12.5 37.5 25 50 0 Qatar Trinidad and Tobago Kuwait Brunei Darussalam Aruba Luxembourg United Arab Emirates Oman Saudi Arabia Bahrain United States Australia Kazakhstan New Caledonia Canada Estonia Russian Federation Greenland 23 Czech Republic 49 China
L1500 Test facility - NOx control strategies - Low NOx burners/staging - Oxygen injection - Corrosion / deposition - Coal blends / cofiring - Fuel characterization - Turbulent mixing scales - Swirl e ff ects (0 - 100%) - Radiative conditions
Simulation Strategy STAR-CCM+ARCHES • Simulation tool to reach exascale Hand-off planes • Finite-volume LES code Resolve detailed • Build on top of Uintah framework • Set of software components and geometry and flow field libraries to solve PDE using using STAR-CCM+ 100,000’s of processors • Task-graph parallelization approach • In-house code Create Averaged • Flexibility to test and implement new Boundary models not currently available/ feasible for commercial programs Condition • Free to download • Structured mesh • Detailed geometry more Input to ARCHES for complicated to capture and resolve computation
L1500 ST AR-CCM+ ARCHES
Primary - coal feed
Gas feed
Inner secondary
Outer secondary
0% Swirl Gas Phase 100% Swirl
0% Swirl Gas Phase Instantaneous Velocity Averaged Velocity
100% Swirl Gas Phase Instantaneous Velocity Averaged Velocity
100% Swirl • LES • Dynamic Smagorinsky Subgrid • 30 million trimmer mesh • 1e-5 second time step • 3,600 cores • 100,000 CPU hours / simulation • 170 days on 24-core workstation 0% Swirl
Solid Phase Modeling only center pipe - Obtained averaged flow field - Lagrangian method Rosin-Rammler Distribution D ref = 53.1 𝜈 m q = 0.87 Particle size range 15 - 200 𝜈 m
• Model details ARCHES Simulation • LES w/ Dynamic Smagorinsky SGS Geometry • DOM/S8 every 10 iterations • Direct Quadrature Method of Moments (DQMOM) • u, v, w, raw coal mass, char mass, particle enthalpy • Shadix/Murphy char oxidation model • Gas-phase reactions • Three stream mixture fraction approach • RFG • O 2 • Coal o ff gas Sufco Utah Coal Burner tip Hand-off plane from ST AR-CCM+
T emperature 0% Swirl 100% Swirl
Large Particle Distribution 0% Swirl 100% Swirl
Particle Distribution 0% Swirl Case small large
Comparison with Experimental Data / Consistency Analysis L1500 Experimental Campaign - Repeated every year for 5 years - 1 to 2 week long - Study di ff erent conditions/models/ coal types - Integrate with simulations - Complete full VUQ cycle every year
Demonstration-scale STAR-CCM+ARCHES prediction 1000 MWe 500 MWe USC oxy-AUSC 8-corner Design Boiler Boiler Phil Sean Pilot-scale 15 MWth oxy-coal boiler validation Jeremy Lab-scale 1.5 MWth oxy- validation 5 MWth oxy- coal furnace gas furnace Jennifer Bench-scale ash soot formation radiation validation transformation multiphase devolatilization char oxidation flow Tom Sean Andy David Jeremy Michael
Acknowledgements
Recommend
More recommend
