˘ˇ ˆ - IEAGHG 5_C/3_090910... · 2013. 7. 25. · ˇ ˆ ˙ Full Model (Doosan Babock, FLUENT/Gambit): • CCTF & wind box (including coal inlet & SA/TA swirlers) • Mesh Type:

��

� ��

��

��

Oliver Stein, Andreas Kempf

[email protected] ; [email protected]

��

• Background

• Inlet Conditions

• Geometric Modelling

• Turbulence Modelling

• Results: Non-Reacting / Coal Combustion

• Conclusions

• Future Work

�� !�� "#$% &��'

�� "#(

Thermal Input: 40MW

Fuel Versatility:

• Coal

• Heavy Fuel Oil

• Natural Gas

• Orimulsion

Facility Usage:

• New Burner Development

• Contract Burner Testing

• Third Party Burner Testing

• OxyCoal II

)��

Core Primary

Secondary Tertiary

*� ��

Full Model (Doosan Babock, FLUENT/Gambit):

• CCTF & wind box (including coal inlet & SA/TA swirlers)�

• Mesh Type: Polyhedral, non-conformal interfaces

• 3.6M cells, local refinement to resolve geometry

• Boundary conditions: 3 inlets, 2x Overfire Air, 1 outlet

• Typical run parameters: 4 CPU cores, ~1 week (Coal)

• Limited control over SA/TA swirler mass fluxes

Compact Model (Imperial, StarCD/CCM+):

• CCTD w/o wind box, downstream of SA/TA swirlers

• Mesh Type: Polyhedral, conformal mesh

• 2.1M cells, local refinement where required

• Boundary conditions: 4 inlets, 2x Overfire Air, 1 outlet

• Inlet conditions derived from averages of Full Model

• Typical run parameters: 4-8 CPU cores, 3-7 days

• Nominal mass fluxes

�� +� �",��-�.

RANS Modelling Parameters:• Software: StarCD/CCM+

• Turbulence models:

• Standard k/eps (StarCD)

• Realisable k/eps (CCM+)�

• Order of convection terms:

• 1st Order (StarCD)

• 2nd Order (CCM+)�

• Thermal model: Enthalpy transport

• Solver: Steady State, SIMPLE

• Convergence: Residual Tolerance: 1x10-3 or

better, visual inspection, data comparison

• Grid: 2.1M cells, polyhedral

LES Modelling Parameters:• Software: In-house Code “PsiPhi”

• Turbulence model:

Smagorinsky Model

• Order of convection terms:

2nd Order

• Thermal model: Conserved Scalar Mixing

• Solver: Low Mach Predictor/Corrector

• Convergence: Statistical Averaging

• Grid: 45.4M cells, hexahedral + immersed BCs

(shorter furnace model)

��(� ��/0��-�1��

Radial Probe, z/D = 0.0




Axial Probe

• Vertical 2D YZ slice: Contours

• 4 Radial line probe locations (horizontal)�

• 1 Axial line probe (burner axis)�

LES Domain

� �",��,��%23��1��4� ��

� �",��,��4� �� +,1��.

Mean Axial Velocity

Mean Circumferential Velocity

� �",��,��%,�� #��5#6

IC-CCM+ (RANS) vs. IC-StarCD (RANS) DB-Fluent (RANS) vs. IC-PsiPhi (LES)

� �",��,��%,�� #��5#6


� �",��,��%,�� 5�6


� �",��,��%1��


,�� +��.

• Modelling of Coal Particle Transport:

• Injection of coal particles of measured size distribution into the primary air stream

• Lagrangian particle tracking until (near) complete combustion

• Modelling of Devolatilisation:

• Constant rate model: stable flame for start-up, low devolatilisation temperature

• Single step model: higher accuracy, realistic devolatilisation temperature

• Modelling of Char Combustion:

• 1st Order effect model (kinetic parameters from lab analysis)

• Modelling of Gas Phase Combustion:

• EBU based on 2-step combustion mechanism

• Modelling of Radiative Heat Transfer:

• Discrete Ordinate Method (DOM): Ordinate set S2+, participating media/particles/walls

,��0��-�,��)

Constant Rate Devolatilisation: Temperature Contour

,��0��-�,��))

Constant Rate Devolatilisation: Temperature Contour, Effect of Radiation (DOM-S2)

,��0��-�,��)))

1st Order Effect Devolatilisation: Temperature Contour

� ��

• Carried out non-reacting (RANS & LES) and reacting (RANS) calculations of Doosan Babcock’s CCTF.

• Non-reacting Simulations (Quantitative Comparisons):

• IC’s RANS results from StarCD and CCM+ for the nominal mass flow rates agree well.

• Lower turbulence levels in StarCD likely stem from the lower order of convective discretisation.

• IC’s LES results (first shot) are comparable to Doosan Babcock’s RANS FLUENT model.

• LES data additionally contains time-resolved turbulence information (flow dynamics, length/time scales,

Re stress, etc.)

• Reacting Simulations (Preliminary Qualitative Results):

• Constant rate devolatilisation (low Tdev) results in a long, hot flame stabilising near the flame holder.

•1st order devolatilisation results in a cooler flame stabilising near the burner quarl.

• Radiative heat transfer modelling using DOM with coarse S2 discretisation has a slight cooling effect.

��7 ��

• Short Term:

• Continue coal combustion analysis & validate preliminary findings

• Study parameter sensitivity of coal combustion model

• Carry out quantitative comparisons (air firing)

• Compare CFD results to experimental data

• Perform realistic oxyfuel coal combustion simulations

• Mid to Long Term:

• Carry out detailed simulations of oxy-fuel firing and compare results to air firing

• Investigate the applicability of LES to coal combustion

�&�� 8

1�� 9��

Sponsors

Prime SponsorLead Company

University Participants

UK Government Support

˘ˇ ˆ - IEAGHG 5_C/3_090910... · 2013. 7. 25. · ˇ ˆ ˙ Full Model (Doosan Babock, FLUENT/Gambit): • CCTF & wind box (including coal inlet & SA/TA swirlers) • Mesh Type:

Documents