Research Findings on Ventilation in Tunnels, Car Parks and Atria By

Dr Tarek Beji Ghent University – UGent, Dept. Flow, Heat and Combustion Mechanics, Belgium 1

Outline Introduction 1. Ventilation in tunnels Based on work conducted at the Institute of Mechanics of Marseille with Dr Olivier Vauquelin

2. Ventilation in car parks Based on: (i) Results presented in an International Workshop on “Fire and Explosion Safety in Large Car Parks” Organized by Pr. Bart Merci (Ghent University) + (ii) work conducted by Dr Nele Tilley (Ghent Uni.)

3. Ventilation in atria Based on work conducted by Dr Nele Tilley (Ghent Uni.)

Conclusion

2

Introduction Importance of ventilation: •Provide tenable conditions for: (i) evacuation of people and (ii) intervention of fire fighters •Major cause of death in fires: Smoke inhalation . CO Ù “Chief killer” The design of a ventilation system depends on: • Type of building (tunnels, car parks, atria…etc), • Occupancy and use of the building…etc Example: 2-directional long tunnels + congested one-directional tunnels Î smoke transverse control. 1-directional low traffic tunnelÎ smoke longitudinal control 3

1. Ventilation in Tunnels

4

1.1 Ventilation strategies There are several ventilation strategies: 1. Longitudinal ventilation: Smoke is pushed by jet fans in one direction. The opposite direction remains smoke free.

2. Transverse ventilation: Smoke is removed by extraction ducts

3. Other options (e.g. semi-transverse ventilation system: confine and then extract the fire induced smoke)

5

1.2 Design calculations Fire characteristics: 1. Heat Release Rate (e.g. Van-Bus => 15 MW, Tank => 100 MW) 2. Smoke flow rate

(e.g. Van-Bus => 50 m3/s , Tank=> 200-300 m3/s)

Mechanical ventilation system: 1. Ventilation System Efficiency, 2. Extraction capabilities, 3. Vent shape and locations…etc.

Ventilation system efficiency: e.g. preventing backlayering Flow patterns of smoke longitudinal control [1]

6

1.3 Research Tools 1. Experimental testing 1.1. Large-scale testing (expensive) 1.2. Reduced-scale testing (scaling laws) 2. Computational Fluid Dynamics (CFD)

-Visualization of temperature iso-contours in a tunnel [2]-

7

1.4 Reduced-Scale Helium Experiments (principle: 1/2) From a full-scale fire plume to a small-scale Helium plume

Principle of two-step splitting [3]

1. Fire source – non-thermal buoyant release: fire HRR and smoke density and flow rate used as release conditions for a buoyant fluid. 2. Non-thermal buoyant release – Scale reduction: conservation of reduced density difference and Richardson number 8

1.4 Reduced-Scale Helium Experiments (principle: 2/2) • Convective HRR of a fire source

Q c = ρs c p qs (Ts − Tair )

(1)

• Buoyant flux of a buoyant jet

ρair − ρs B = g qs ρair

(2)

• Density of air-Helium mixture

ρs = χ air ρair + χ hel ρhel

(3)

(2) + (3) Î

B = g qhel

ρair − ρhel ρair

(4)

• The buoyant flux depends only on the Helium volume flow rate [1]. • Air has to be added to increase the mixing flow rate, decrease the density deficit and reach coherent fire plume characteristics [1]. 9

1.4 Reduced-Scale Helium Experiments [1, 4]) (set-up) • Rectangular plexiglass channel (10m × 0.5m × 0.25m) • Circular opening located at the floor level (at the centre) • Injection of a mixture air-Helium • Mixture seeded by ammonium salt particles • Visualization of smoke flow patterns: vertical laser sheet in the medium plan

- Picture of the set-up-

- Sketch of the set-up-

- LASER set-up -

10

1.4 Reduced-Scale Helium Experiments: Natural Ventilation Smoke is extracted through openings placed at the ceiling “Aeraulic transparencies”

- Smoke extraction through openings-

- Set-up -

Objective: Critical opening width to have full extraction of smoke ? 11

1.4 Reduced-Scale Helium Experiments: Natural Ventilation

+ Air

Total extraction + He

Partial extraction

Total/Partial extraction + Air/He

(Intermittent extraction)

Partial extraction (low efficiency)

12

1.4 Reduced-Scale Helium Experiments: Natural Ventilation Non-dimensional mixing flow rate

qm* =

qm gH 5

Critical opening width (minimum width for total extraction), Lcrit

Correlation [4] 1/ 2

⎛ q − 0.036 ⎞ Lcrit ⎟⎟ ≈ 1.8 ⎜⎜ H ⎝ Δ ρ m /ρ m ⎠ * m

- Correspondence between mixing relative density difference and mixing flow rate at critical condition for several opening lengths -

- General correspondence between mixture density and flow rate for a given opening length at critical condition -

13

1.4 Reduced-Scale Helium Experiments: Natural Ventilation

• Design experiments • Derive results

• Apply results derived from reduced-scale experiments to real-scale fires

The required opening length Lcrit as a function of Qc [4] 14

1.5 Numerical simulations (1/3) CFD code: The Fire Dynamics Simulator (FDS), designed specifically for fire calculations

- Geometry and meshing in FDS for the simulation of the reduced-scale Helium model-

- Visualization of the flow at the opening -

15

1.5 Numerical simulations (2/3) Calculation of the extraction efficiency from CFD results

W

L

H Slice plane downstream

Q c ,inlet = c p T0 qm Δ ρ m

• Simulated convective HRR (FDS input) • Non-extracted convective HRR Q c ,non −extracted

⎛H ⎞ ⎜ = c p T0 ⎜ ∫ u x (z ) Δ ρ (z ) dz ⎟⎟W ⎝0 ⎠

• Extraction efficiency

Q c ,non −extracted

e extraction

⎞ ⎛ K −1 u kΔ ρ k + u k +1Δρk +1 Δ z ⎟⎟W = c p T0 ⎜⎜ ∑ 2 ⎠ ⎝ k =1

Q c ,non − extracted =1− Q c ,inlet

16

1.5 Numerical simulations (3/3)

Opening length (L) in the reduced-scale model (in cm)

Comparison: Experimental correlation

Vs FDS results

10 8 6 4 Exp. correlation FDS (full extraction) FDS (partial extraction)

2 0

0

2

4

6

8

10

Simulated convective HRR (MW) 17

2. Car parks (References [5-8])

18

2.1 Full-Scale Experiments: Hexane Pool Fire Conducted at Warrington Fire - GENT

• Set-up: – Car park 30 m x 28.6 m x 2.7 m – Inlet opening modular – Pool fire: tray of 3 m x 1.5 m – Fuel height: 45 cm above floor level – 4 extraction fans (4 x 50000m3/h) – 2 induction type jet fans (50 N each) Modular inlet opening 19

2.1 Full-Scale Experiments: Hexane Pool Fire

- Picture of the car park -

- Sketch of the exp. set-up 20

2.1 Full-Scale Experiments: Hexane Pool Fire Results (1/3) Different flow patterns depending on the inlet opening configuration

- Flow patterns near the ceiling -

Once smoke is trapped in a recirculation region, it is hard to get it out! 21

2.1 Full-Scale Experiments: Hexane Pool Fire Results (2/3) Smoke backlayering distance determined from mean temperature profiles along centerline under ceiling

Smoke direction in backlayering

22

2.1 Full-Scale Experiments: Hexane Pool Fire Results (3/3) • Activation of jet fans for the case of full open inlet (OOOO) hardly affects results on the centreline. • For the case XXOXX the exact position of extraction fans is not crucial. •Transversal beam: blocking of smoke backlayering (momentum is broken).

23

2.2 Full-Scale Experiments: Real-car fires Collaboration with Brandweer Gent Research purposes: • Effect of SHC system on fire spread to next car. • Effect of fire service intervention on temperatures. • Use of video data. • Use of thermal camera. Practical purpose: Development of SOPs for Fire Service interventions in case of car park fire. 24

2.2 Full-Scale Experiments: Real-car fires Experimental set-up

25

2.2 Full-Scale Experiments: Real-car fires Results (3 cars, configuration XXOX)

26

2.3 Numerical study Objective: design a system with a necessary extraction velocity to guarantee smoke clearance in the car parks

- Principle of critical horizontal velocity in a car park -

Similarity with tunnels for the ‘critical velocity’ to keep all the smoke in the downstream direction But Different ratio height/width 27

2.3 Numerical study ‘Numerical experiments’ in order to depict relationships between: 1. Convective HRR per unit area, 2. Fire source area. Domain of validation: 1. Large closed car parks, 2. Flat ceiling, 3. Uni-directional smoke and ventilation flow pattern

28

2.3 Numerical study Correlations between the critical velocity, the convective HRR per unit area and the fuel surface Dependence on

'' conv .

q

Dependence on A F General correlation

'' 1/ 4 conv .

v cr ,in ∝ q

v cr ,in ∝ A F

1/ 5

'' 1/ 4 conv .

v cr ,in = 0.26 q

AF

1/ 5

Required ventilation velocity when a backlayering distance d is allowed

d = 40 (vcr − v )

29

3. Atria (References [7-8])

30

3.1Objective Design a mechanical ventilation system which allows a given smoke-free height in case of a fire • M(z): mechanical mass flow rate (kg/s) • Mb: emerging smoke flow rate • Qconv.: convective HRR (kW) • z: height above spill edge (m) • z0: virtual origin (m) • W: width of atrium (m) • Db: smoke depth (m) - Small-scale atrium configuration and spill plume-

1/ 3 M (z ) = C (z + z 0 ) Q conv .

C = 0.3Cm ρ 0 W

2/3

z 0 = Db +

Mb 1/ 3 C Q conv .

31

3.2 Performance of CFD • Validation of CFD results for small-scale atria • Can we perform “numerical experiments” to capture trends?

- Experimental and CFD results of smoke mass flow extraction rate as a function or rise height32

3.3 Scaling issues ƒ Froude number

Fr =

u charcteristic velocity = g h gravitational velocity

ƒ Densimetric Froude number

Fr =

u

ƒ Richardson number

Ri =

g'h

;

g' = g

ρ1 −ρ 2 ρ

g h potential energy = u2 kinetic energy

• Nele Tilley: “Scaling based on only the Froude number is allowed as long as all configurations are sufficiently turbulent”. • Olivier Vauquelin: “The most important aspect is to represent the duality between inertial and buoyant forces. This imposes a strict conservation of the Froude and Richardson numbers. The Reynolds number cannot be preserved, but it has to be high enough in order to ensure turbulent flows”.

33

3.4 Multi-dimensional smoke layer effect (1/2) Reduced-scale calculations 1-D smoke layer

Multi-D smoke layer Temp (°C) 50

Hs

zs,ave zs,min ≈ zs,ave

m = 0.37kg / s

zs,min

m = 0.69 kg / s

20 34

3.4 Multi-dimensional smoke layer effect (2/2) New expression

When multi-dimensional smoke layer is present, a new slope is found 35

Conclusion • The design of efficient ventilation systems is challenging. • Research tools: Experiments and CFD modelling • Experiments must be carefully designed (e.g. scaling issues) • CFD modelling could be used for “numerical experiments” ==> capture trends

36

References [1] O. Vauquelin, Experimental simulations of fire-induced smoke control in tunnels using an ‘‘air–helium reduced scale model’’: Principle, limitations, results and future, Tunnel and Underground Space Technology 23 (2008) 171178 [2] H.Y. Wang, Numerical and theoretical evaluations of the propagation of smoke and fire in a full-scale tunnel, Fire Safety Journal 49 (2012) 10-21 [3] O. Vauquelin, G. Michaux, C. Lucchesi, Scaling laws for a buoyant release used to simulate fire-induced smoke in laboratory experiments, Fire Safety Journal 44 (2009) 665667 [4] O. Vauquelin, , T. Beji, G. Michaux, Laboratory experiments on smoke natural extraction in case of tunnel fire. Thriteenth International Symposium on Aerodynamics and Ventilation of Vehicle Tunnel, BHRGroup, New Brunswick, NJ, USA, May 2009 [5] N. Tilley, X. Deckers, B. Merci, CFD study of relation between ventilation velocity and smoke backlayering distance in large closed car parks, Fire Safety Journal 48 (2012) 11-20 [6] X. Deckers, S. Haga, N. Tilley, B. Merci Smoke control in case of fire in a large car park: CFD simulations of full-scale configurations, Fire Safety Journal (2012) [7] N. Tilley, P. Rauwoens, B. Merci, Verification of the accuracy of CFD simulations in small-scale tunnel and atrium fire configurations, Fire Safety Journal 46 (2011) 186-193 [8] N. Tilley Numerical study on Fire Smoke Extraction in Large Complex Buildings. PhD Dissertation, Ghent University (2011).

37