Computational Combustion for the PetroChemical Process Industry Alion Science & Technology ACCES Advanced Combustion and Chemical Engineering Solutions Tulsa, OK
OUTLINE Introduction to ACCES Reactive Flow Analysis Detailed Kinetics and Mixing models Examples Conclusions and Summary Computational Combustion for the Process Industry Page 2
ACCES: Part of the Alion Team Practical Problem Solving for the Process Industry Joseph D. Smith, Ph.D. (Brigham Young) Turbulent reactive flows (Dow Chemical-gasification, incinerators, Cabot-fumed metal oxides, John Zink-process heaters, flares, incinerators) Eric Hixson, P.E. (Iowa State) Inorganic & organic chemical production (Dow Chemical-organic chemicals, incineration, FMC-inorganic chemicals, AE-Staley Staley- biochemicals,, Bayer-biochemicals biochemicals,, Cabot-aerogels aerogels,, fumed metal oxides, John Zink-process heaters, incinerators) Larry Berg, M.S. (MIT) Design/Application of Combustion equipment for Process Industry (John Zink); Engineering Solutions for NOx retrofits of Coal fired Power Generation (RJM) Mike Fard,, M.S. (Univ( of Tulsa) Technology Development/application to LNG, Upstream production Computational Combustion for the Process Industry Page 3
Background: Who are we? Group s Relevant Experience 60+ years Hands-On experience in CPI/HPI Practical, applied solutions using advanced CFD tools Proven track record in Chemical and Petrochemical industry Key Achievements by Group Members Advanced optimization of Industrial Incineration Systems Lead CFD Groups for Dow Chemical and Koch Industries Developed/Patented advanced process equipment for gas flares and flare pilots, industrial drying ovens, HAZ- Waste incinerators, Bio-chemical reactors Computational Combustion for the Process Industry Page 4
We add value by: Providing advanced engineering analysis of Coal-fired Electric Power Generation Gas flares (air assist, steam assist, multi-tip, tip, enclosed) Pyrolysis furnaces Low NO x burners Haz-waste incinerators Specialized Reaction Engineering expertise Coal combustion Chloro-hydrocarbon hydrocarbon chemistry Hydrocarbon Production Fumed metal oxides Bio-Chemical Processing Computational Combustion for the Process Industry Page 5
We Can Help you: Meet tighter environmental regulations NOx, CO, PIC s Identify best design to meet your needs Burners, turbines, nozzles, reactors, etc. De-bottleneck and/or optimize existing systems Yield, Quality, Safety, Production, etc. Reduce downtime when retrofitting new technology into existing system Shorten down time for Plant turnaround when installing new Low NOx Burners into operating furnace Computational Combustion for the Process Industry Page 6
Computational Combustion Growing Impact of CFD Validation critical (Lab and Plant) Start with known Base Case Identify trends vs. exact values Improves Bottom line (operating costs, yield, quality, and safety) Best done hand-in in-hand with Experiment Focus testing on best design options - reduce experimental costs Shorten development cycle/reduce development cost Industry uses what s available & WORKS (e.g., Correlation, spreadsheet, CFD, etc.) Bridges gap between Theory & Practice Computational Combustion for the Process Industry Page 7
Physics of Combustion Analysis Secondary Air w/ Swirl Fuel & Primary Air Cool Walls Quarl Atomizer 2 4 Cool 1 Hot Hot 3 1 Free shear layer -mixing of fuel and oxidizer, hot gas recirculation on centerline 2 Multiphase droplet/ particle dispersion, vaporization 3 Fluid Mechanics- plug flow Cool 4 Heat Transfer - gas recirculation, heat transfer, participating media Computational Combustion for the Process Industry Page 8
Computational Combustion Combines Multi-Physics into CFD Based Tool Gaseous Reactions (Homogeneous) Local Equilibrium Turbulence Coupling PDF Chemistry Gaseous Fluid Mechanics Momentum Equations Energy Equation Turbulence Model Particle Mechanics Turbulent Dispersion Wall Deposition Nucleation/ Agglomeration Multi-Physics Combustion Analysis Pollutant Formation (Trace Chemistry) SOx - Non-Equilibrium NOx - Fuel and Thermal PIC - Incineration Radiation Heat Transfer Discrete Ordinates Data for Comparison Radiative Properties Particle/Droplet/ Surface Reactions (Heterogeneous) Pyrolysis Devolatilization Vaporization Computational Combustion for the Process Industry Page 9
CFD Includes Multi-physics Fluid Mechanics: All flow regimes laminar and turbulent All fluid types Newtonian and non-newtonian Compressible and incompressible Steady state and transient Heat Transfer: Convection, conduction, radiation Conjugate heat transfer Multiphase: Lagrangian, Eulerian,, Free Surface Particles, sprays, droplets, bubbles, cavitation Chemical Reactions Turbulent Chemistry w/ Detailed Kinetics Computational Combustion for the Process Industry Page 10
Coupling Chemistry to Turbulent Flow Computational Combustion for the Process Industry Page 11
Turbulent Chemistry Issues What Turbulent Mixing Models Consider mixing time and reaction times: D a = τ t /τ c = (l( t /v')/ (l f /s L ) D a 0 (Frozen); D a (Fast); D a 1 (Coupled) How to couple chemistry & turbulence Resolve flow to micro scale for reactions Couple fluctuations on macro-scale with reaction Require Detailed Kinetic mechanism CH 4 combustion - GRIMECH (50 species, 350 x2 reactions) CH 2 Cl 2 combustion - Bozzelli mechanism (35 species, 170 x2 reactions) Must Reduce Degrees of Freedom for practical problems Chemical state space (r, H, Y 1,...,Y ns ) 2+ns degrees of freedom (dof( dof) For each DOF, 1 Non linear PDE must be solved over ~10 6 grid points Large disparity in reaction time scales (fast vs slow reactions) Computational Combustion for the Process Industry Page 12
Current Modeling Approaches Eddy Break-Up Models (EBU & Combined Kinetics/EBU) Reaction rate approximated by mixing rate Simplified kinetics when reaction time << mixing time Presumed Probability Density Function (PPDF) Mixing Limited (Reaction time << mixing time) Probabilistic mixing for local turbulence effect with local equilibrium Detailed species transport (reduced or global mechanism) Limited # of species described with individual PDE Detailed species transport (N-step mechanism) Full mechanism with partial turbulence effects Others: Manifolds, TPDF, RCCE, etc. Computational Combustion for the Process Industry Page 13
Reaction Analysis: Which Tool? Decreasing Computational Time Thermodynamics Analysis Requires good thermodynamic data Assumes infinite reaction rate and perfect mixing Reaction temperature & products upper limit Detailed Kinetics Analysis Requires consistent reaction mechanism Assumes detailed kinetics with generalized mixing (PFR, CSTR) Reaction temperature/products < thermodynamics estimate CFD Analysis User specifies reactor geometry, operating conditions Approximate chemical kinetics and turbulent mixing process Analyze/interpret results based on assumptions Increasing Complexity Computational Combustion for the Process Industry Page 14
Thermodynamics vs. Kinetics 100 Should I use Thermodynamic equilibrium or kinetic mechanism to estimate HCl/Cl 2 split? 80 ξ Extent of reaction based upon HCl 60 40 Typical Operating Temperatures General thermodynamics Literature elementary reactions (PSR) 20 0 500 1000 1500 2000 2500 3000 Temperature, K Computational Combustion for the Process Industry Page 15
Measured vs.. Predicted [Cl 2 ] 600 500 Cl2 (Exp) ppm Cl2 (Prd) ppm Kinetic effects control chemistry at high flowrates 400 300 200 100 0 1 2 3 4 5 6 7 Experiment No. Computational Combustion for the Process Industry Page 16
Reaction Chemistry in Industrial Reactor Sparger Tube Reaction Zone Constricted Exit Typical Reactor Geometry Computational Combustion for the Process Industry Page 17
Predicted Concentration Profiles: full chemistry vs. reduced chemistry Dashed lines = ILDM Solid Points = Reduced chemistry Measured Concentration s Computational Combustion for the Process Industry Page 18
Predicted Ignition points: different kinetic mechanisms Predicted Ignition point Computational Combustion for the Process Industry Page 19
NO x Chemical Kinetics complex set of elementary reactions wide range of parallel and sequential rate processes NOx routes of formation Thermal NO Zeldovich-NO Prompt NO Fenimore-NO Fuel NO Measured and calculated NOconcentrations in H 2 air flames Computational Combustion for the Process Industry Page 20
Glass Furnace Simulation Highly resolved individual port-necks measurements for boundary conditions & validation calculated port-neck exhaust define b.c. s for furnace Full furnace calculation (with as much resolution as possible) LES in port-necks Computational Combustion for the Process Industry Page 21
Temperature Distribution in the Burner Plane Single port model (Temperature in K) Computational Combustion for the Process Industry Page 22
Velocity Validation Study Hole 2 Hole 5 Single port Three-port Full furnace Experiment Single port Three-port Full furnace Experiment Distance above glass surface(m) Distance above glass surface(m) X Velocity (m/s) X Velocity (m/s) Two distinct regions: High velocities in the flame region and a large recirculation in the crown region Not much variation between different models Good agreement with experiments Computational Combustion for the Process Industry Page 23
Temperature Validation Hole 1 Hole 2 Single port Three-port Full furnace Experiment Single port Three-port Full furnace Experiment Dist above glass surface(m) Dist above glass surface(m) Temperature (K) Temperature (K) Reasonable predictions: Low gradient region (1 m from glass) Flame region very sensitive to mesh resolution Accurate flame temperature predictions Accurate air flow inlets: Profile very important Robust turbulent mixing model : Large Eddy Simulations (LES) High resolution mesh: resolve different scales in the furnace Computational Combustion for the Process Industry Page 24
Buoyancy Driven Plumes: Flares and Fires from Tieszen, Nicolette, Gritzo, Holen, Murray, Moya, 1996 Multiphysics has strong coupling buoyancy (density variations) combustion & radiation affect density gradients soot dominates radiation Modeling radiation & soot empirical soot correlations using optically thin approximation empirical radiation model w/ no soot model (eg., 20-30% heat loss) Computational Combustion for the Process Industry Page 25
Soot Prediction (add something from ISIS) Thick layers of soot not predicted by empirical models Computational Combustion for the Process Industry Page 26
Applications: Spray Combustion Spray Combustion Facility (NIST) * swirl burner with a movable 12-vane swirl cascade * profile droplet size, number density, velocity, and gas species concentrations http://www.cstl.nist.gov/div836/836.02/sprays.html Combustion Research Facility (Sandia) Advanced Combustion Center (ACERC) Others (MIT, School of Mines, etc.) Computational Combustion for the Process Industry Page 27
Applications: Spray Pyrolysis Production of Fumed Ceria Oxide Vaporization region Metal Oxide particle formation in Flame Particle agglomeration region Computational Combustion for the Process Industry Page 28
Applications: Metal Oxides Production H 2 Diffusion flame Particle formation Computational Combustion for the Process Industry Page 29
Simulation of Fumed Metal-Oxide Flame Gas Temperature Computational Combustion for the Process Industry Page 30
Applications: Plant Optimization CFD used to Fix existing equipment problems Ethylene furnace had flame impingement on process tubes De-rated capacity - significant impact on Bottom-Line Customer asked for help fixing problem Team formed to solve problem CFD primary tool to evaluate various options Team identified most promising solution Implemented in field - Worked First Time! Computational Combustion for the Process Industry Page 31
Initial problem: Flame roll over into tubes Problem Fixed! Computational Combustion for the Process Industry Page 32
Applications: Equipment Sales CFD helps develop new equipment/technology Customer system must destroy CF 4 (>80% DRE) Kinetics analysis indicated reaction temperatures must be >3300 F to achieve required DRE Alumina based refractory melts <3000 F CFD used to design new system able to meet customer needs Computational Combustion for the Process Industry Page 33
Combustion temperatures >3000 F required to oxidize CF 4 Quench inlets designed w/ CFD to keep refractory surfaces cool Computational Combustion for the Process Industry Page 34
Applications: Implementing New Technology CFD used to Optimize new equipment performance Flames from New Low NO x Burners in Vertical Cylindrical furnace much longer than original ones Flames merged together and extended into convection section Resulted in lower operating capacity/high emissions CFD used to identify most likely solution Implemented in field - Worked First Time! Computational Combustion for the Process Industry Page 35
Before After Computational Combustion for the Process Industry Page 36
Applications: Hazardous Vent Incineration 6.1722 0.0762 0.6096 0.0762 water wall (1st Tube Pass) 0.3048 30 swirl vanes 0.1778 30.256 Combustion Air 0.2032 Fuel Gas/VCM Vapor Vent Ring Centerline 1.067 Organic Vents Gaseous) Fuel Gas Reactor Geometry Computational Combustion for the Process Industry Page 37
Hazardous Vent Incineration Expected DRE? Burning Non-Design Hazardous Wastes Find local maximum and average exit concentrations to improve efficiency Computational Combustion for the Process Industry Page 38
Hazardous Vent Incineration - Performance Optimization Case 5 Simulation (4% Excess 2) O 950 950 400 1250 500 1050 1200 1350 650 850 1100 1300 1300 1400 1500 1550 1600 Case 4 Simulation (Near Stoichiometric Conditions) 1300 950 1300 1050 1150 1250 1000 1200 1250 1350 1450 1400 1350 Gas temperature profile (K) Computational Combustion for the Process Industry Page 39
Conclusions Industrial Problem Solving Many opportunities in HPI/CPI for advanced problem solving skills ACCES uses experience to solve problems for HPI/CPI Tough Problems Solved: Vent Incineration Hydrocarbon Production Plant retrofit Coal Combustion Process Optimization Computational Combustion for the Process Industry Page 40
Conclusions Industrial Problem Solving Capabilities and Limitations: CFD can examine phenomena not previously considered due to safety, cost, time, etc. Advances in analysis of reacting flow with detailed kinetics CFD is not a black box that blindly & unerringly reproduces physics Computational combustion can evaluate new fuels, new designs, and help optimize ACCES can impact your bottom line! Computational Combustion for the Process Industry Page 41