Validation studies of CFD codeson hydrogen combustion

Sudarat Worapittayaporn, Luciana Rudolph, Harald DimmelmeierAREVA NP GmbH

ERMSAR 2012, Cologne, March 21 – 23, 2012

- ERMSAR 2012, Cologne March 21 – 23, 2012 - AREVA NP GmbH Proprietary - © AREVA - p.3 All rights are reserved, see liability notice.

Content

Introduction

Validation resultsSlow combustion experiments in THAI facility

Fast combustion experiments in ENACCEF facility

Conclusions

- ERMSAR 2012, Cologne March 21 – 23, 2012 - AREVA NP GmbH Proprietary - © AREVA - p.4 All rights are reserved, see liability notice.

Motivation

During a postulated severe accident large amount of H2 can accumulate in the containment, which exhibits a potential risk to the structure integrity.

CFD tools have been applied in containment analysis to:

Calculate the gas mixture distribution in containment Calculate combustion of a predefined gas mixture Yield dynamic pressure loads on internal walls and

containment shell Assess risk of deflagration-to-detonation transition

Evaluation of the applicability of CFD codes to predict the H2 combustion in nuclear plant containment is an important exercise in this context.

- ERMSAR 2012, Cologne March 21 – 23, 2012 - AREVA NP GmbH Proprietary - © AREVA - p.5 All rights are reserved, see liability notice.

Objective

Three CFD codes used in this study:

ANSYS CFX

ANSYS FLUENT

COM3D (Karlsruhe Institute of Technology)

Selection of most-appropriate models, parameter sensitivity, and calibration of models and correlations are part of this study.

Validation against data of selected experiments with specified conditions relevant to containment analysis, e.g.:

Slow and fast combustion

Negative hydrogen concentration gradient

Upwards and downwards burning direction

Steam and hydrogen concentration gradient

Confined geometry with obstacles

- ERMSAR 2012, Cologne March 21 – 23, 2012 - AREVA NP GmbH Proprietary - © AREVA - p.6 All rights are reserved, see liability notice.

Main differences of codes and models

CFX FLUENT COM3DVersion 12.1 13 4.0

SolverPressure-velocity,

coupledPressure-based,

segregatedCoupled, compressible

Turbulence Model Shear Stress Transport RNG k-epsilon Standard k-epsilon

Combustion ModelPartially premixed:

BVM+EDMPartially premixed model

with PDF tablesKYLCOM

Laminar flame speed Liu and MacFarlane Liu and MacFarlane Exp. DatabaseTurbulent flame speed Zimont, Dinkelacker Zimont SchmidtSlip conditions at walls No-slip No-slip SlipThermal conditions at

wallsHeat flux with radiation Heat flux with radiation Adiabatic

Grid Unstructured Unstructured Cubic structuredTime Step Adaptive Fixed Adaptive

Validation THAI Facility

Experiments

Modeling

Results

- ERMSAR 2012, Cologne March 21 – 23, 2012 - AREVA NP GmbH Proprietary - © AREVA - p.8 All rights are reserved, see liability notice.

Large scale test facility operated by Becker Technologies GmbH

Cylindrical stainless steel vessel of 9.2 m height and 3.2 m diameter with a total volume of 60 m3

Inner cylinder and condensate trays are removed for the hydrogen deflagration (HD) tests

Tests with up- and downwards flame

Fully instrumented

THAI facility: Description

Pressure monitoring

- ERMSAR 2012, Cologne March 21 – 23, 2012 - AREVA NP GmbH Proprietary - © AREVA - p.9 All rights are reserved, see liability notice.

THAI experiments: Mesh sensitivity

Coarse Standard Fine HybridType of grid Tetra,

Prism layerTetra,

Prism layerTetra,

Prism layerHexa,Tetra, Prism layer

Ave. cell size 0.331 m 0.199 m 0.135 m 0.204 m

Number of cells 81,626 452,435 3,223,535 181,323

Grid-independent solution

Standard Tet-mesh is chosen

- ERMSAR 2012, Cologne March 21 – 23, 2012 - AREVA NP GmbH Proprietary - © AREVA - p.10 All rights are reserved, see liability notice.

2,

1min

2.31

4.41)00002.0step( 22 HO

l

CC

kA

Sk

ProgressReactionR

Coupled BVM-EDM model in CFX:Parameter dependence

Parameters in the Burning Velocity Model (BVM)

Laminar burning velocity Sl

Liu and MacFarlane correlation Model by Szabo (KIT): a function of

temperature, pressure, mixture composition

Turbulence burning velocity St

BVM reaction progressClassic EDM reaction rateSaid-Borghi

Factor

Parameters in the Eddy Dissipation Model (EDM)

Empirical coefficient A in the reaction rate: A = 8, A = 16

Modification with Said-Borghi factor

Zimont Model A=0.6 (default), A=1.7

Dinkelacker

|~| cSS tuc

4/14/12/14/3' tult lSuGAS

2.0

0

3.0

25.0 'Re

46.01

P

P

S

u

LeSS

ltlt

Sensitive to laminar and turbulent burning velocities

Not much influenced by EDM-A

Improvement by using Said-Borghi factor (very minor in slow combustion)

- ERMSAR 2012, Cologne March 21 – 23, 2012 - AREVA NP GmbH Proprietary - © AREVA - p.11 All rights are reserved, see liability notice.

THAI experiments:Parameter sensitivity study

St SlLiu and

MacFarlaneKIT

Zimont A=1.7 OK OKZimont A=0.6 - Too slow!Dinkelacker OK Too slow!

HD-8: better predicted by Dinkelacker correlation

- ERMSAR 2012, Cologne March 21 – 23, 2012 - AREVA NP GmbH Proprietary - © AREVA - p.12 All rights are reserved, see liability notice.

THAI experiments: Comparison of CFX, FLUENT and COM3D

HD-7 HD-8 HD-27Pressure 1.480 bar 1.487 bar 1.497 bar

Temperature 30-90°C 30-90°C 30-90°CH2 concentration 10 vol% 10 vol% 6-12 vol%

H2O concentration - - 3-47 vol%Burning direction upwards downwards upwards

Slower pressure development predicted

Combustion calculations sensitive to initial turbulence level (unknown in experiments)

- ERMSAR 2012, Cologne March 21 – 23, 2012 - AREVA NP GmbH Proprietary - © AREVA - p.13 All rights are reserved, see liability notice.

THAI experiments: Comparison of CFX, FLUENT and COM3D

HD-7 HD-8 HD-27Pressure 1.480 bar 1.487 bar 1.497 bar

Temperature 30-90°C 30-90°C 30-90°CH2 concentration 10 vol% 10 vol% 6-12 vol%

H2O concentration - - 3-47 vol%Burning direction Upwards downwards upwards

Combustion progress involves two regimes (slow and fast)

None of the codes can completely reproduce entire combustion progress(slow and fast)

- ERMSAR 2012, Cologne March 21 – 23, 2012 - AREVA NP GmbH Proprietary - © AREVA - p.14 All rights are reserved, see liability notice.

THAI experiments: Comparison of CFX, FLUENT and COM3D

HD-7 HD-8 HD-27Pressure 1.480 bar 1.487 bar 1.497 bar

Temperature 30-90°C 30-90°C 30-90°CH2 concentration 10 vol% 10 vol% 6-12 vol%

H2O concentration - - 3-47 vol%Burning direction upwards downwards upwards

Well predicted by all codes

Maximum pressure overestimated due to assumption of combustion completeness

Validation ENACCEF Facility

Experiment

Modeling

Results

- ERMSAR 2012, Cologne March 21 – 23, 2012 - AREVA NP GmbH Proprietary - © AREVA - p.16 All rights are reserved, see liability notice.

1.7

mi.

d.

0.7

4 m

3.3

mi.

d.

0.1

54

m

ENACCEF facility: Description

Located in France and operated by CNRS

Consists of two parts:

Acceleration tube and dome

Acceleration tube with 9 annular obstacles

Blockage ratio of 0.63

Flame develops in the upwards direction

Initial negative H2 concentration gradient:

from 11.6% vol. in the lower part of the facility to 8.0% vol. in the upper part

Instrumented to measure flame position, pressure build-up and gas composition

9x obstacles

Ignition point

Pressure monitoring

- ERMSAR 2012, Cologne March 21 – 23, 2012 - AREVA NP GmbH Proprietary - © AREVA - p.17 All rights are reserved, see liability notice.

ENACCEF – ISP 49 test run 765: Geometrical model

CFX&FLUENT Hex-coarse Hex-fine COM3D

Type of cell Hexa Hexa HexaAve. cell size [mm] 13.22 8.60 15.4 - dome [mm] 21.97 14.30 15.4 - acc.tube [mm] 8.55 5.58 15.4Number of cells 568,583 2,109,126 950,616

COM3D

ObstaclesBR=0.57 BR=0.63

coarse fine

- ERMSAR 2012, Cologne March 21 – 23, 2012 - AREVA NP GmbH Proprietary - © AREVA - p.18 All rights are reserved, see liability notice.

ENACCEF – ISP 49 test run 765: Mesh and timestep sensitivity

Grid-independent solution in CFX achieved

CFX FLUENT

Influence of time step in FLUENT noticeable:

Coarse grid + large time step = insufficient

Coarse grid + small time step = sufficient

Fine grid + small time step = acceptable

Time step too large

- ERMSAR 2012, Cologne March 21 – 23, 2012 - AREVA NP GmbH Proprietary - © AREVA - p.19 All rights are reserved, see liability notice.

ENACCEF – ISP 49 test run 765: Comparison of CFX, FLUENT and COM3D

Pressure evolutions: good agreement by all codes

Delay in pressure rise

Partially flame quenching? Or only transition of combustion regimes?

All codes failed to predict this behavior Further model development and validation needed!

- ERMSAR 2012, Cologne March 21 – 23, 2012 - AREVA NP GmbH Proprietary - © AREVA - p.20 All rights are reserved, see liability notice.

ENACCEF – ISP 49 test run 765: Comparison of CFX, FLUENT and COM3D

Slow flame propagation after the last obstacle (?)

Flame position and velocity: good agreement by CFX and FLUENT, underestimated by COM3D (BR=0.57 instead of 0.63, too coarse mesh)

- ERMSAR 2012, Cologne March 21 – 23, 2012 - AREVA NP GmbH Proprietary - © AREVA - p.21 All rights are reserved, see liability notice.

Conclusions

Simulations results and their direct comparison to measured data show importance of using appropriate correlations for laminar and turbulent burning velocity.

All codes capture main process features with reasonable adequacy(predictions of global parameters, e.g. maximum pressure in slow and fast turbulent combustion regimes, are consistent and in fair agreement with experimental data)

Combustion models are very sensitive to initial turbulence

►Apparently, some phenomena (like flame quenching, transition in combustion regime from fast to slow deflagration) are still a challenging situation for codes

►Calculation results demonstrate that hydrogen safety analysis in containments using commercial CFD codes such as CFX or FLUENT is possible in near future

- ERMSAR 2012, Cologne March 21 – 23, 2012 - AREVA NP GmbH Proprietary - © AREVA - p.22 All rights are reserved, see liability notice.

Any reproduction, alteration or transmission of this document or its content to any third party or its publication, in whole or in part, are specifically prohibited, unless AREVA has provided its prior written consent.

This document and any information it contains shall not be used for any other purpose than the one for which they were provided. 

Legal action may be taken against any infringer and/or any person breaching the aforementioned obligations.

End of presentationValidation study of CFD codeson hydrogen combustion

AREVA NP GmbH