CCM+ in Chemical Process Industry Ravindra Aglave Director, - PowerPoint PPT Presentation

Advanced Applications of STAR- CCM+ in Chemical Process Industry Ravindra Aglave Director, Chemical Process Industry

Outline Notable features released in 2013 Gas – Liquid Flows with STAR-CCM+ Packed Bed Reactors: Beyond porous media approach Optimization: A paradigm shift

Eulerian Multiphase Multiple Granular phases Reynolds Stress Model with EMP – Simulation of mixtures with 2 or – Rotating, swirling and anisotropic more granular phases flows Granular temperature model Multicomponent Boiling Model for extended EMP – Previously algebraic equation – Calculates the mass, energy and solved momentum transfer between a – Solving full transport equation continuous and a dispersed multicomponent phase Chemical reactions Interface Momentum Dissipation – Int ra phase reactions Model – Int er phase reactions – Reduces unphysical parasitic currents

Lagrangian/DEM Stochastic Secondary Droplet (SSD) Forces breakup model – Drag torque – Efficient and accurate method – Spin lift force compared to other approaches Choice of rolling friction models Passive Scalars – Force proportional – Passive scalars may now be used – Constant torque with Lagrangian/DEM – Displacement damping – Scalars may transfer between Lattice and random injectors can particles continuous phase use geometry parts • New multiphase interaction method – Improved speed, convenience Particle-wall conductive heat transfer

Reacting Flow  Diesel engines, boilers, coal-powered plants Soot Two-Equation Model for non-premixed combustion – aka the Moss Brookes Hall soot model – Two additional transport equations solved for increased accuracy Surface Chemistry Model – Chemical reactions on surfaces without requiring DARS-CFD add-on. • The Homogenous Reactor • The Eddy Break-Up (EBU) model • The Non-reacting model with Segregated Species

Reacting Flow Three stream PVM Threaded PPDF table construction – Enhanced user experience and performance – GUI can still be used during operation Progress Variable Model – Can now model two fuel streams and one oxidizer stream – Previously only one fuel stream allowed Sandia Flame EBU Soot Two Equation Model Soot Volume Fraction – Moss-Brookes-Hall soot model can now work with the Eddy Break Up (EBU) model widening applicability to non-premixed flames – Addition of PAH sub-model for nucleation for soot prediction with higher hydrocarbon fuels such as kerosene User Defined Char Oxidation Model – User defined char oxidation rate for coal combustion  Soot Modeling  Coal Combustion

Gas – Liquid Flows

General Setup Gas Outlet 3D Model – 0.45m x 0.2m x 0.05m – 40.000 hexahedral cells – Water does not enter or leave domain Velocity inlet – K-e turbulence model – Time step size = 1e-3 - 0.1 s – Bubble size dp = 2 mm – monodisperse Three Different Set-up – I : Degassing boundary – II: Degassing boundary wih additional forces – III: Flow split /gas pocket at top Gas Inlet

Case I: Pfleger Setup Outlet: Degassing BC Drag Force (Cd = 0.66) Turb. Disp. Force Vgas = 48 l/h v sup =0.00133 m/s

Case I: Results: Plume after 1 sec

Case I: Plume Oscillation

Case II: Enhanced Pfleger Setup Drage Force: Tomiyama Lift Force: Tomiyama Turb. Disp. Force Bubble Induced Turbulence (Troshko&Hassan) Virtual Mass Force Diaz et al. (2008), Chem. Eng. J. 139, 363-379 Ziegenhein (2013), CIT, accepted manuscript

Case II: Results

Case II: Results

Case II: Results Snapshot at t = 220s Averaged over 100s

Case II: Results

Case III: Air Buffer Setup • Setup like Case I • Flow-split outlet • dt ~ 0.001 - 0.01 s • Inner Iteration = 40 - 200 Reaching convergence within each timestep is important !

Conclusion Simulation with degassing BC: – Robust and accurate – All kind of forces can be considered Simulation with air buffer: – Startup has to be monitored carefully (each time step has to be converged) – Lift force can not be taken into account

Power of Optimization: A paradigm shift

Problem Statement To design an Heater ducting for furnaces for use in the refining/petrochemical industry – Goal is to minimize the mass flow variation through burner throats – With the minimal Pressure drop possible – A variety of geometric parameters can be changed The Heater consists of a central duct connected to the burners via short cylindrical legs 20 20

Parameters Radius of connector Height of duct Width of duct Connector Dia Taper Taper

Base Case Results 22 22

Parametric CAD Robustness Study CAD variations explored – 148 evaluations performed – 40 mins on 8 cores for baseline – 32 hrs for entire project on 40 cores – CD-adapco PowerTokens provide ultimate flexibility for DSE by allowing the user to decide what combination of parallel evaluations and solver cores is most efficient for them Metrics used 𝑅 𝑛𝑏𝑦 −𝑅 𝑛𝑗𝑜 – Delta Mass Flow = (Performance) 𝑅 𝑗𝑒𝑓𝑏𝑚 – Delta Pressure = ∆𝑄 𝑛𝑏𝑦 in the system (Fan/Damper limit) 23 23

Meshing Surface Remesher, Polyhedral Mesh Continuum Models Mesher, Prism Layer Mesher Base Size 10.0 mm 4.0 mm / 10.0 mm Surface Size ( min / target ) Block: 1.6 m / 1.6 m Prism Layer Mesher 3 / 1.3 / 2.5 mm (layers / stretching / total thickness) Block Floor: 5 / 1.3 / 100 mm 24 24

Results Design 158 Design 40 25 25

Process Automation Parametric CAD Geometry STAR-CCM+ CFD Analysis Design Simulation Variables Responses • Input & Output Files Are Defined • Program Execution is Automated • Design Variable are Identified and Tagged in Files • Complete Process is Executed from 1 Button or Script 26

Mixing tank geometry • Geometry created within 3D CAD • Specific dimensions set as design parameters

Optimization setup: Pareto front Objectives o Maximize volume averaged turbulent kinetic energy (proportional to mixing) o Minimize moment on impeller blades and shaft (indicative of torque/power consumption) • Variables Variable name Minimum Maximum Increment Baffle length 0.005 m 0.012 m 0.0005 m Baffle numbers 0 9 1 90 o 5 o Impeller blade pitch angle 0 Number of impellers 1 5 1

Computational Summary Single Phase, Water # of Cells = 200K (varies with geometry) # Possible designs ~ 16000 # of Designs = 153 Parametric geometry creation = 2-3 hrs Optimate setup time = 30 mins 5 simultaneous on 12 cores (60 cores) = 10 hrs clock time Total compute hours = 5 x 10 = 600 hrs # of power tokens = 5x12 = 60

Results: Pareto Front (# of Designs 20) Turbulent kinetic energy Pressure on impeller blades

Pareto Front (# of Designs = 20)

THANK YOU