Using OpenFOAM to model of complex industrial devices Christophe Duwig 1, Henrik Hassing 2, Elisabeth Akoh Hove 2 1 R&D division - Haldor Topsøe A/S & 2 Force Technology
Outline of the talk Shortly about Haldor Topsøe A/S What does CFD do for us Replacing commercial CFD with OpenFOAM OSFRI initiative in Denmark How does OpenFOAM fit in the context? LES with detailed chemistry: example of hydrogen auto-ignition LES of a swirling flow
Haldor Topsøe A/S www.topsoe.com Founded by Dr. Haldor Topsøe in 1940 selling high quality catalyst since 1944 selling high quality processes since 1957 Business areas: Syngas Hydrogen & gasification-based tech. Ammonia Methanol/ DME Refinery Environmental / gas cleaning denox applications sulfuric acid (WSA) CatOx Automotive
In real life, it may look like
CFD applications in HTAS Reforming (hydrogen, ammonia, ) furnaces simulations Reformer simulation (air and oxygen rich flames) Turbulent heat transfer Environmental Ammonia/ Urea mixing in SCR devices Heat transfer in sulfuric acid condensers Particle depositions in SCR catalysts Refinery Multiphase flows Heat & Mass transfer Traditional fields of applications for commercial codes
OSFRI initiative in Denmark http://www.osfri.dk
When did it start? A DANSIS meeting in October 2007 on new trends in CFD and with a special focus on OpenFOAM revealed a strong interest 110 participants!
What do we do? Open Source CFD has a clear potential but also barriers and dangers High energy of activation However the benefits and dangers are almost the same for all new users! Need of the same catalyst So why not create national competences, share knowledge, test cases, user guides and not do the same mistakes? Need a critical mass Create the OSFRI consortium!
The need of funding An application for public funding was tailored by H. Hassing at FORCE Technology. A a total budget of 7.4 mill. DKK including 3.4 mill. DKK in public funding (1.00 =7.43895 DKK) The results should be made available to all interested Danish companies (and others!) FORCE Technology Grundfos Management Niro A/S MAN Diesel A/S Ødegaard & Danneskiold-Samsøe A/S Babcock & Wilcox Vølund A/S Haldor Topsøe A/S FLSmidth Airtech A/S FLSmidth A/S Aalborg University Technical University of Denmark
What is the outcome? Between the primary participants, the expected outcome are more or less like: Organize training sessions Use Open Source for standard cases Particles modeling Combustion modeling Swirling flow modeling Open Source will be used for 50-70% of the volume which will be increased drastically
However OpenFOAM also opens new avenues Detailed modeling of state-of-the-art devices Application to flames about the physics an example what is new with OpenFOAM Application to Large Eddy Simulation about the physics an example what is new with OpenFOAM
Example of turbulent flame
How does a flame front really look like? what we see and what we hear OH PLIF from Buschmann et al., 26th Symp. On Combustion, pp. 437-445, 1996 Even simple a Bunsen flame is complex as shown beside! There is a need of advanced investigation techniques (experimental and numerical), here laser based OH-PLIF Does it happen only in the lab.??? NO! But traditional CFD oversimplifies the story -> need to do better!!
How do we address these problems? Large Eddy Simulation Fine mesh Low dissipation numerical schemes Small time steps Large computers Advanced models for combustion Accurate thermo-chemistry description realistic kinetics Reads ChemKin format! Turbulence/chemistry: ILES/EDC/ PaSR OpenFOAM is suitable because Open source = flexibility Fair choice of numerical schemes that can be used for LES/ I-LES Runs in // without additional costs!
Cabra s lifted flame Cold high-speed H2 jet in Hot oxidant with 17% O2 H2 /N 2 T~305 K T~1045 K H2 O/O 2 /N 2 XIG Lift-off hight X T~1045 K H2 O/O 2 /N 2 Flame location OBS: not on scale!
Large Eddy Simulation using OpenFOAM Customized solver based on the OpenFOAM library solving filtered continuity and Navier-Stokes low-mach number & ideal gas assumption pressure coupling is PISO Smagorinski SGS closure second order discretization solver CG with AMG-preconditioning for pressure Numerical grid of cell size in the jet region h~d/24 (~0.6 millions Hex) Costs: ~24h for one jet-flow-through on 8 procs. Filtered specie- and enthalpy-equations: ~ ( ρyi ) + ( uy ~ ~ ρ ) t ~ ( ρh ) + ( ρ uh ~~ ) t i = ρ D = ρ D h i υ + Sc + υ Pr Y ~ ~ i + ωt (Yj h ~ ) ILES : We resolve the reaction layer in the LES grid!!
Complex chemistry is needed Example: 1. H 2 oxidation by Juan Li et al., 2004; 9 species & 19 reactions 2. CH4, GRI3.0, 53 species 3..
Flame turbulence interaction Cold H2/N2 jet issuing into a hot coflow Tcoflow=1045K O2=17%
Radical species : OH & HO2 fields Cold H2/N2 jet issuing into a hot coflow - Tcoflow=1045K O2=17% HO2 precedes OH: ignition precursor!!
Example of turbulent swirling flow
Example: incompressible swirling jet Filtered continuity and Navier-Stokes u u t = 0 + I-LES 1 ρ ( uu ) + ( uu uu ) = p + ν ( u) Numerical methods (solver oodles) spatial discretization: cubic limited temporal discretization: 2nd order implicit Numerical grid typical cell size in the shearlayer h=d/50 number of cells: 1.5 millions (hex) Costs: 3-4 days on 32 procs. For 50000 time steps
Instantaneous flow field - 1 Streamwise velocity Azimuthal velocity
Instantaneous flow field - 2 Streamwise velocity
Summary & future work OSFRI initiative is a success Open-source CFD need catalysis to enter the industry Critical mass is needed OpenFOAM is suitable when dealing with complex turbulent flows and flames = industrial problems! Open source = flexibility Fit the code to the problem and not the other way around Massive parallel runs without additional costs! Future actions Development of new capabilities To the OpenFOAM community: thank you for your work Phase II for the OSFRI initiative
Thank you for your attention!