The Suitability of Fire-Field Modelling for Enclosure ...fire-research.group.shef.ac.uk/steelinfire/downloads/SWinter_06.pdf · Structures and Fire Research Group School of Mechanical,

The Suitability of Fire-Field

Modelling for Enclosure Fires involving Complex Solid Fuel

Loads

STIFF Meeting Sept 2006

Stuart Winter

Suitability of CFD Modelling for Enclosure Fires S. Winter

Structures and Fire Research Group School of Mechanical, Aerospace and Civil Engineering

Introduction

• Two types of solid fuel source: Cars and

Wooden Cribs.

• Discussion of European Commission report on

steel in closed car park fires.

• Comparison of FDS with other CFD

simulations for car fires.

• Findings of simulations from compartment fire

tests using cribs.

• Alternative modelling approaches for cars and

wooden cribs.



European Commission Report

“Development of design rules for steel

structures subjected to natural fires in closed

car parks.” EUR 18867 EN

• Closed car park tests.

• Suggested HRR

curve (Figure 1).

• Fire-field simulations

in ‘representative’closed car parking

floor.

0

1000

2000

3000

4000

5000

6000

0 10 20 30 40 50 60 70

Time (min)

Heat

Rele

ase R

ate

(kW

)

Q_car Q_opencar

Figure 1: European Commission HRR curve comparison.



Car Fire Observations

• Open car fires: ~2 MW peak fire, over 1-2 hours.

• Closed car fires: higher intensity (> 4 MW) peak, shorter duration (< 1 hour).

• Car type and age: More combustibles in modern cars greater fire intensity (over 8 MW).

• Fire spread: 12 – 30 mins to adjacent cars, via external plastics and tyres.

→



Fire Dynamics Simulator

• Computational Fluid Dynamics (CFD) model from NIST.

• Mass, momentum, energy and species conservation and turbulence model.

• Large Eddy Simulation: Small-scale turbulence is modelled (sub-grid), as larger eddies contain most turbulent energy.

• Combustion model: Infinitely fast reaction.

• Radiation Model: Based on local intensities.



European Commission Simulations

• “Representative” closed car park configuration.

• Car location under steel beam at point of max. bending moment.

• No combustion modelling used by European Commission.

Figure 2: Detail of European Commission simulation domain.



Fire Source Modelling Approaches

• HRR curve from older, open car fire (Figure 3).

• Radiative flux: Adiabatic condition.

• Fine grid needed for combustion modelled simulations.

• FDS extended to supply purely convective heat sources.

Figure 3: HRR curve for European Commission Simulations.

0

500

1000

1500

2000

2500

0 30 60 90 120

Time (min)

Hea

t R

ele

as

e R

ate

(kW

)

Q_total Q_front Q_rear



Comparison of Results

Figure 4: Above rear heat/fuel source temperatures, ceiling height.

0

100

200

300

400

500

600

700

0 600 1200 1800 2400 3000 3600

Time (s)

Te

mp

(o

C)

FDS HFlux FDS Combust FLUENT (Euro) VESTA (Euro)



Comparison of Results

Figure 5: Temp. adjacent to lower flange of beam, heat flux sim.

Figure 6: Simulated temperatures along car centre-line, 960s.

0

100

200

300

400

500

600

0 5 10 15 20 25 30

Distance Along Beam (m)

Air

Tem

p a

dja

ce

nt

to

be

am

(o

C)

t = 300 s t = 720 s t = 960 s t = 1320 s t = 2040 s t = 3000 s



Mixture Fraction Combustion

Figure 7: Temps adjacent to lower flange of beam, combustion simulation.



European HRR Curve

0

1000

2000

3000

4000

5000

6000

0 10 20 30 40 50 60 70

Time (min)

He

at

Re

lea

se

Ra

te (

kW

)Q_car Q_opencar

Figure 1: HRR curve for enclosed single car fires, for application to CFD codes, compared to previous curve (from European Commission, 1999).



European HRR Curve

Figure 8: Estimated temperatures produced using the recommended curve for

closed car-park fires, from FDS and European Commission report.

0

200

400

600

800

1000

1200

0 600 1200 1800 2400 3000 3600 4200

Time (s)

Te

mp

era

ture

(o

C)

Rear (FDS) Front (FDS) Rear (Euro) Front (Euro)



Issues Raised

• FDS successfully compared to FLUENT and TNO Vesta.

• European Commission report implies use of unprotected steel for closed car parks, based on simulations, despite tests (1269oC near ceiling).

• No validation of simulations with experiment.

• Higher peaks produced with combustion modelling than without.

• Questionable results for recommended design curve.



Suggested Car model

• Multi-zone model: one for each ‘compartment’.

• Balances for heat and mass transfer between zones and main simulation.

• Model fire spread through car.

• Supplies fuel and combustion product outputs.

• Dependence on heat feedback and oxygen availability.

• Direct modelling of major external plastics

(thermoplastic model).



Compartment Fire Tests

• Much data on wooden crib

fires, since early 60’s.

• Simulations are dependant

on HRR curves and do not

represent solid burning

processes.

• Wooden cribs often used, as self-sustaining and

representative of office/residential fire loads.

• Can be mixed with plastics.

• Example Cardington 12m X 12m X 3.4m.



Principles of Wood Combustion

z

Cool Air In

Hot Gases Out

Wooden

Shaft

Walls

Gas Fuel

+

Products

+ HEAT

CO2 +

HEAT

O2

•Wood

•Char

•Char

Shaft

Interior

Thermal Decomposition

Surface Oxidation

x

Reactant

Product

Solid



Crib Model Construction

• 1-D fluid-flow equations with time dependence.

• Mass and chemical species equations include

source/sink terms for 3 processes.

• Energy equation includes heat sources from

combustion.

• Compatible with Mixture Fraction combustion

model: outputs fuel OR oxygen at top, not both.

• Crib is collection of parallel shafts.

• Re-evaluates crib combustion processes based

on conditions outside the crib.



Aims of Models

• Model fires of given load and distribution in any compartment design.

• Allows calculation of fire-driving mechanisms.

• Maintain sufficiently course grids for large compartment simulations.

• Dynamically integrated – no HRR curve needed for fire.



Conclusions

1. FDS achieves comparable performance to

other CFD packages.

2. Fire-field simulations highly dependant on HRR

curves.

3. Combustion modelling produces higher air

temperature peaks.

4. Need to develop complex solid fuel combustion

models to improve adaptability and

performance of fire-field models.



Acknowledgements

This work was produced as part of EPSRC-

Corus case award No. 04300406. This financial

contribution, together with the guidance provided

by Dr Brian Kirby of Corus, is gratefully

acknowledged. The knowledge, experience and

time of Prof. Colin Bailey and Dr David Apsley of

The University of Manchester is also greatly

appreciated.





Appendix



Summary of previous progress

Cardington 12m X 12m Compartment Fire Tests:

• Dependence on

Heat Release

Rate curves.

• Need for time-

dependent modelling

solutions.

• Limited practical

use.

Figure A1: Temperatures for Test 3: Rear, 600mm from ceiling.

0

250

500

750

1000

1250

0 30 60 90 120

Time (min)

Tc

Te

mp

. (C

)

FDS Tc 2/3 Lennon & Moore Tc 2, 600mm



Summary of previous progress

Car Fires:

• Combustion in

car interior not considered.

• Underestimates heat transfer

from car body.

• Limited consideration of

fuel release about car. Figure A2: Open car fire temperatures at first

peak burning period: Centre plane.



Heat Flux Vs. Combustion

• No net radiative losses in heat flux case.

• Double conductive losses in combustion case.

Figure A3: Total heat losses from FDS domain, for pure convective and combustion simulations.



Documents

The Suitability of Fire-Field Modelling for Enclosure ...fire-research.group.shef.ac.uk/steelinfire/downloads/SWinter_06.pdf · Structures and Fire Research Group School of Mechanical,