Ov Over ervie view of R of Reacting Flo acting Flows Outline - - PowerPoint PPT Presentation

Ov Over ervie view of R

• f Reacting Flo

acting Flows

Various Applications rious Applications Ov Over ervie view of a

• f available reacting flo

ailable reacting flow models models Lat Latest additions st additions Exam Example Cases ple Cases Summar Summary

Outline Outline

Reacting Flo acting Flows Applications in S s Applications in STAR-CCM+ AR-CCM+

– Chemical Process Industry (liquid-liquid reactions)

• Finite-rate chemistry model with a flexibility to modify EOS
• EMP inter/intra-phase reactions
• Moment methods
• Surface Chemistry

– Rocket Engines (Solid, Liquid, and Hybrid)

• Particle Reactions in Lagrangian
• Real–Gas model with all Combustion Models
• Coupled Solver

– High-speed jet engines (Ramjet, Scramjet)

• Coupled solver with combustion models

– Oil and Gas

• Multiple-Phase reactions (intraphase and interphase)

Ev Ever er-Expanding application co

• Expanding application coverage

rage

– Gas turbine, process heaters, burners, and furnaces

• Partially-premixed combustion models
• LES
• Soot and Nox models

– Chemical Vapor Deposition

• Detailed/Global Surface chemistry
• multi-component diffusion

– Aftertreatment (Automotive)

• Detailed/Global Surface chemistry
• Coupled with liquid film and porous media

– Energy Industry (Coal and Biomass combustion)

• Multiple Coal Types and cofiring with gas-fuel

Reacting Flo acting Flows Applications in S s Applications in STAR-CCM+ AR-CCM+

Non-P Non-Premix remixed Combustion d Combustion

– EBU

• Standard, Hybrid, Finite-Rate
• User Defined

– PPDF (Multi-stream)

• Equilibrium
• Flamelet

– PVM (Chemistry Table)

Premix Premixed Combustion ed Combustion

– CFM (Choice for Laminar flame speed) – PEBU – TFC

Par Partially ially-Premix

• Premixed Combustion

d Combustion

– PCFM

• Equilibrium
• Flamelet

– EBU – PVM – PTFC

Finit Finite-Rat

• Rate Chemistr

e Chemistry Calculation using D y Calculation using DARS-CFD RS-CFD

– EDC – ISAT – Dynamic Load Balancing

Sur Surface R ace Reactions actions Soo Soot and No and Nox Emission Models mission Models

Reacting Flo acting Flow Models in S Models in STAR-CCM+ AR-CCM+

Lat Latest A st Additions (v 8/9) dditions (v 8/9)

Micr Micromixing

• mixing Models
• dels

Sur Surface Chemistr ace Chemistry

– Global mechanisms Combustion M Models w with Re Real G Gases

– SRK and Peng-Robinson EOS

Soo Soot Model Model

– MBH Model

– Soot absorption properties Eulerian Eulerian Multi-phase R Multi-phase Reaction Model action Model

– Flexibility to add user defined reactions

Com Comple lex Chemist x Chemistry Model (D y Model (DAR ARS-C S-CFD) D) – ISAT – Analytical Jacobian – PaSR

Pe Performance

Ne New Micr w Micromixing Model f

• mixing Model for Liq

r Liquid R id Reactions in 9.02 actions in 9.02

Eddy Contact Micr Eddy Contact Micromixing

• mixing Model (based on F
• del (based on Frone
• ney and

and Naf Nafia, Chem a, Chem Eng ng Sc, 2000) f Sc, 2000) for Liq r Liquid-Liq id-Liquid R uid Reactions. actions. User Users will ha s will have choice of three mixing choice of three mixing scales scales

• Kolmogor
• lmogorov (corresponds t

v (corresponds to Engulfment type of mixing) Engulfment type of mixing)

• Classical Scalar Mixing

lassical Scalar Mixing (consider (considers high Sc s high Sc number) number)

• User Def

ser Defined (ability t ned (ability to do user def do user defined f ined ff)

• Kine

inetics Only (No tics Only (No Micr Micromixing)

• mixing)

The abo The above f e four choices are a ur choices are available f ailable for each reaction r each reaction

Micr Micromixing Model –

• mixing Model – Eddy Contact Micr

ddy Contact Micromixing

• mixing

Eddy Contact Eddy Contact Micr Micromixing

• mixing

Validation Case – lidation Case – Low Re Re flo flow in Coaxial Jet in Coaxial Jet

The results com The results compared t ared to the numerical of the numerical of Forne rney and Naf and Nafia (2000) a (2000) and e and experimental data of perimental data of Li and T Li and Toor (1

• or (1986) at lo

986) at low high R w high Reynolds ynolds number

• number. R

. Reactants A and B react in aq actants A and B react in aqueous solution ueous solution

Mass F Mass Fractions of A actions of A, B, R B, R (Desired) and S (Desired) and S

Results (% Yield of R) sults (% Yield of R)

 Micromixing Models predictions are much better than EBU  Classical Scalar Mixing time scale appears to give best match

Comparison with Experimental Data

50 60 70 80 90 100 2000 3000 4000 5000 6000

Reynolds Number % Yield of R

Experiment (Li & Toor, 1986) Hybrid EBU Eddy Contact Micromxing (Kolmogorov) Eddy Contact Micromxing (Classical Scalar Mixing) Froney & Nafia Paper, 2000

No D No DARS-CFD licenses are req RS-CFD licenses are required ired User Users can s can add their o add their own reactions n reactions Same int Same interface f ace for adding r adding reactions as reactions as gas-phase reactions gas-phase reactions

Sur Surface R ace Reactions actions

Applications Applications

Tools

• ols

– Porous chemistry – Global Surface chemistry – Detailed Surface Chemistry

CVD R CVD React actors rs Af After-treatment de treatment devices: vices:

• Three-Way Catalytic Converters (TWC)
• Diesel Oxidation Catalyst (DOC)
• Diesel Particulate Filter (DPF)
• Selective Catalytic Reduction (SCR)

Results – sults – NOx R Ox Reduction Com duction Comparison arison

Two-Step Model Detailed Surface Chemistry

Real Fluid Modeling in STAR-CCM+

• Real Fluid Physics in STAR-CCM+
• Van der Waals
• Redlich-Kwong (RK)
• Peng-Robinson (PR)
• Soave-Redlich-Kwong (SRK, available in 8.02)
• Modified Soave-Redlich-Kwong (MSRK, available in 8.02)
• All above Equation of Sates are Cubic

Real Fluid Thermodynamic Departures

Enthalpy : Enthalpy : Specific Heat Specific Heat : Entropy : Entropy : Speed of Speed of Sound Sound :

Results (Density Comparison)

PR PR SRK SRK Ideal Ideal Gas Gas

Soot modeling

Two-Equation Soot Model

Transport equations are solved for two soot variables – Transport equations are solved for two soot variables – Soot

• ot

number density (N) and Soot Mass density (M) number density (N) and Soot Mass density (M) Key ph y physical pr ysical processes are :

• cesses are :

– Nucleation cleation – Coagul Coagulati ation – Soo Soot gr growth th – Soo Soot o

• xidati

idation

Nucleation PAH inception Acetylene inception

C2H2, C , C6H6, C , C6H5, H , H2 C2H2 Compute from: Compute from:

• 1. Species list
• 1. Species list
• 2. Empirical (non-
• 2. Empirical (non-

premixed) premixed)

Available Soot Models

Two-equation model Two-equation model without radiation without radiation Moments model with Moments model with radiation radiation Two-equation model Two-equation model with radiation with radiation All the All the scaling scaling facto factors fo rs for r source terms source terms are 1.0 are 1.0

Int Inter-Phase R Phase Reactions with EMP actions with EMP

23

• Following Options are

Provided

• First-order combined rate
• Half-order combined rate
• Second-order combined

rate

• User reaction rate

Gas phase reaction setup in S Gas phase reaction setup in STAR-CCM+ AR-CCM+

24

• When using the built-in

reaction rate expression, input

• the temperature exponent
• activation energy
• pre-exponent, and
• the diffusion coefficient.

General Ov General Over ervie view of F

• f Furnace Flo

rnace Flow

Ore / Coke Layer Ore / Coke Layer

• Falls

lls d down n very s ry slowly.

• wly.

Gas Gas

• Hot gas in

gas inject jection ion

• Flow upward t

upward through rough ore /

• re /

coke coke l laye yers rs

• Lost of heat i

Lost of heat into ore to ore / coke / coke la layers rs

• Chemic

Chemical r al reactions wit eactions with

• re /
• re / coke

coke

Cohesive Zone Cohesive Zone

• Ore la

e layer temper erat atur ure e inc increases eases

• Bloc

Blocked gas passage ked gas passage due due to m melted or lted ore

• Cohesive z

Cohesive zone of ne of lar large e vol volume me

Fe2O3 Fe3O4 FeO Fe

Eulerian por Eulerian porous media appr us media approach

• ach

26 26

• Gas Phase

 Three components: CO/CO2/N2

• Porous media

 Three components: Fe/Ore/Coke

• Boundary conditions:

 Outlet boundary: Pressure outlet  Inlet boundary:

• Mass fraction of the gas phase: CO/N2= 0.8/0.2.
• Velocity = 15 m/s, Temperature = 2000K

Chemical reactions Chemical reactions

27 27

• Two reactions

 C + CO2 -> 2CO  Fe2O3 + 3CO -> 2Fe + 3CO2  Time step: 1 sec  The model is stable and fast:  32 processors, one hour,

simulated around 3000 seconds in the physical time.

Cok Coke and Ore par and Ore particle icle area area

28 28

Con Conver ersion of Ore sion of Ore int into F Fe

29 29

Eulerian multiphase: 2-phase model Eulerian multiphase: 2-phase model

30 30

• Full size furnace:
• 25m height
• 7.2m hearth diameter
• 2D axisymmetric model
• Multi-component Eulerian phases:

 Gas phase: CO, CO2, N2  Solid phase: Ore, Coke, Fe, Fe2O3, C

• Two reactions:

 Fe2O3 + 3CO -> 2Fe + 3CO2  C + CO2 -> 2CO

Vo Volume Fr Fractions

31

Te Temperatures

32

Com Comple lex Chemistr x Chemistry

• Can read

Can read Chemkin Chemkin format rmat and no limit and no limit on number

• n number of
• f species

species

• Online tabulation using ISA

Online tabulation using ISAT is a is available ailable

– Factor of 2-5 speedup is commonly observed

• Dynamic load balancing is a

Dynamic load balancing is available ailable to achie achieve e scalability scalability for chemistr r chemistry y calculation with large number of pr calculation with large number of processors.

• cessors.
• DARS-Basic pr

RS-Basic provides t ides tool t

• ol to reduce th

reduce the chemistr e chemistry that can that can be be im impor ported in ed in STAR AR-CCM+ f

• CCM+ for fur

r further speedup f her speedup for com r comple lex chemistr x chemistry calculations. y calculations.

Par Partially ially-Stirred R

• Stirred React

actor (PaSR) in S r (PaSR) in STAR-CCM+ 9.02 AR-CCM+ 9.02

• PaSR representative of real

turbulent combustion simulations

• Chemistry represented by a finite

number of particles

• Particle composition evolves by

pair-wise mixing and chemical reaction

• Can be used to test different

combustion models for representing detailed chemistry

• Computationally inexpensive

compared to real simulations – ideal for initial testing and setup

Analytical Jacobian f Analytical Jacobian for Dars-CFD r Dars-CFD

• Dars-CFD ODE integrator used for computing reaction solution
• ODE integration involves the use of Jacobian
• Computing Jacobian numerically using divided-difference is

expensive and inaccurate

• Additional benefits when used with ISAT due to increased accuracy
• f sensitivity matrix and reduced CPU time per add
• Collaborating with Prof. Lu, University of Connecticut, to integrate

use of analytical Jacobian in Dars-CFD ODE integrator

Performance

• rmance

Test Case Description

Baseline v8.02

• 49 Species
• Cp and Mw use Mixture Methods

Case 1 v8.06

• 49 Species
• Cp and Mw use Table Methods

Case 2 v9.01 dev

• 49 Species
• Cp and Mw use Table Methods
• Uniform mixture fraction spacing
• Efficient retrieval of Table entries

Performance/Speed Com

• rmance/Speed Comparison

arison

Test Case Solver CPU Time (Serial Run) sec Baseline v8.02 19078 s Case 1 v8.06 13040 s

• 31.6% faster than Baseline

Case 2 v9.01 dev 8940 s

• 53% faster than Baseline
• 31% faster than Case 2

Performance/Speed Com

• rmance/Speed Comparison

arison

Speed Comparision (Flamelet Model)

5000 10000 15000 20000 25000

STAR-CCM+ Version Solver CPU time (sec)

8.02 8.02 8.06 8.06 9.02 9.02

Conclusions Conclusions

Expanding Application Co Expanding Application Coverage rage Eulerian Eulerian Multi-Phase with R Multi-Phase with Reactions actions LES ef LES effectiv ctive but e e but expensiv pensive

– High fidelity calculations

Finit Finite-rat

• rate kine

e kinetics tics

– Library-based – Direct chemistry coupling (no additional license requirement)

Speedup (Detailed Chemistr Speedup (Detailed Chemistry) y)

– Load balancing – Analytical Jacobian – ISAT

Performance Im

• rmance Impr

provement ement

02/01/12 adapco Report 531-0015-001 Page 40

More Exam More Examples ples

Note that close burner spacing in the center of the furnace leads to

• xygen-starved

areas up into the convection section