Politecnico di Torino
Porto Institutional Repository
[Other] Multiscale analysis of tunnel ventilation flows and fires
Original Citation:
Colella F., Rein G., Verda V., Borchiellini R., Torero J.L. (2010). Multiscale analysis of tunnel
ventilation flows and fires. .
Availability:
This version is available at : http://porto.polito.it/2370566/ since: June 2010
Terms of use:
This article is made available under terms and conditions applicable to Open Access Policy Article
("Public - All rights reserved") , as described at http://porto.polito.it/terms_and_conditions.
html
Porto, the institutional repository of the Politecnico di Torino, is provided by the University Library
and the IT-Services. The aim is to enable open access to all the world. Please share with us how
this access benefits you. Your story matters.
(Article begins on next page)
GDR CNRS n
GDR
CNRS n°2864
2864 17‐18 June 2010 Multiscale analysis of tunnel u sca e a a ys s o u e
ventilation flows and fires
Tunnel Portal
CFD 3D fire module
1-dimensional network
Tunnel Portal
Dr. Francesco Colella
Dr Guillermo Rein
Dr. Guillermo Rein Prof. Vittorio Verda
Prof. Romano Borchiellini
Prof. Jose L. Torero
Prof. Jose L. Torero
Dipartimento di Energetica – Politecnico di Torino
1-dimensional network
Introduction ‐ Ventilation system
The most widespread safety system in tunnels is the ventilation system
ventilation system
¾Normal operating conditions
• visibility i ibili
• pollutant concentrations ¾Emergency conditions
• Smoke management
• safe evacuations
fire fighting
• fire fighting Dipartimento di Energetica – Politecnico di Torino
Introduction ‐ Ventilation system
The flow conditions within tunnels are dependent on the combined i fl
influence of:
f
¾Ventilation devices (axial and jet fans)
¾Tunnel layout (slopes)
ll
(l
)
¾Boundary conditions at the portals
¾Blockages within the domain
¾Fire size and Location ¾Comprehensive analysis has to consider the whole system
y
and not simply a part of it
py p
Dipartimento di Energetica – Politecnico di Torino
1D Network model ‐ characteristics
¾Fast simulations
¾Straightforward
St i htf
d definition
d fi iti off boundary
b
d
conditions
diti
¾Predict the global behaviour of coupled systems
¾Well
W ll established
bli h d for
f design
d i purposes
BUT
¾Need calibration constants
¾Provides only ballpark figures
Dipartimento di Energetica – Politecnico di Torino
CFD model ‐ characteristics ¾Detailed flow field representation
¾Well established for design verification
BUT
¾Large
g computational
p
time
¾Not affordable for long tunnels or for
analysis of ventilation strategies
¾Still
ll limited
l
d results
l due
d to uncertainties
in modelling turbulence, combustion,
radiation and py
pyrolysis
y of condensed fuels
¾“Acceptable”
accuracy for global
quantities (i.e. back‐layering distance)
¾“Poor”
“
” predictions
di i
off local
l l flow
fl field
fi ld data
d
Dipartimento di Energetica – Politecnico di Torino
Com
mputattional time
CFD vs. 1D computing time
CFD models
Accurate
Is there Is
there
anything in y
g
between?
1D model
Rough
Tunnel Length
Dipartimento di Energetica – Politecnico di Torino
Multiscale model ‐ introduction
Typical velocity contours in presence of jet fans
1D region
T
Tunnel
lP
Portal
t l
1-dimensional network
3D region
1D region
CFD model
1D model
1D model
CFD 3D jet fan module
Tunnel Portal
1-dimensional network
1D model
Dipartimento di Energetica – Politecnico di Torino
Multiscale model ‐ introduction
Typical temperature contours in presence of fire
1D region
3D region
1D region
1D model
CFD model
1D model
Dipartimento di Energetica – Politecnico di Torino
Multiscale – coupling procedure
Physical decomposition of the domain
1D Regions
3D Regions
¾
Low velocity/temperature gradients
¾
High velocity/temperature gradients
¾
Mono‐dimensional models can be used
¾
CFD models must be adopted
1D region
3D region
1D region
Dipartimento di Energetica – Politecnico di Torino
Multiscale – coupling procedure
•Based on Dirichlet‐Neumann methods
•Use non‐overlapping domain decomposition •Neumann type boundary condition •Neumann‐type
boundary condition
applied to the CFD sub‐domain interfaces •Dirichlet‐type boundary condition applied to the 1D sub‐domain interfaces. •The model guarantees the continuity of average pressure, velocity and
average pressure, velocity and temperature values at the 1D‐CFD interfaces
x=a
Ω3D
Γa
Ω1D
Dipartimento di Energetica – Politecnico di Torino
x
Multiscale – coupling procedure
Coupling stages
1 ‐ Full 1D model
2 CFD model near field
2 ‐
CFD model near field
3.a – 1D model far field
K‐times
3.b ‐ CFD model near field
Proceed to the next time step when some global convergence is reached Dipartimento di Energetica – Politecnico di Torino
Multiscale ‐ characteristics
¾ Dramatic reduction of the computational time (up to 2 orders
of magnitude for a 1.2 km long tunnel)
¾ As accuracy as CFD
¾ Successfully used for parametric studies
BUT
¾ The placement of the 1D‐3D interfaces is an issue
¾ Higher set up time than CFD
¾ The CFD solver has to be accessed by the user during the
iteration algorithm (i.e. compiled UDF for FLUENT)
Dipartimento di Energetica – Politecnico di Torino
Simulation of tunnel ventilation flows
East & West Dartford tunnels (UK)
250m
1055m
130m
Essex
(north)
Kent
((south))
133m
1280m
Colella, F., et al (2010), Tunnel. Underg. Space Technol.
Colella F et al (2009) Building and Environment
Dipartimento di Energetica – Politecnico di Torino
157m
Simulation of tunnel ventilation flows
East & West Dartford tunnels (UK)
¾≈1500 m long with hybrid ventilation system
y
¾East tunnel diam 9.5 m (1980) ¾West tunnel diam 8.6 m (1963)
(
)
¾2 supply & extraction ventilation stations in the vicinity of the portals
¾28 jet fans in the West Tunnel
¾11 reversible jet fans in the East tunnell
Colella, F., et al (2010), Tunnel. Underg. Space Technol.
Colella F et al (2009), Building and Environment
Dipartimento di Energetica – Politecnico di Torino
Simulation of tunnel ventilation flows
West Tunnel Jet fan near field
6
Multiscale arrangement
5
h [m
m]
h [m
m]
K+3
0
0
8
i-1
i-2
South portal
1
3
0
10
k
Γi
Far field 4
6
horizontal velocity [m/s]
10
0
j
6
10m
10
Γj
3D Sub‐domain Ω3D
3
Near field
0
2
j+1
j+2
4
6
t+1
horizontal velocity [m/s]
8
10
North portal
N
1D Sub‐domain Ω1,1D 4
3
2
1
section 100 m
4
6
horizontal velocity [m/s]
2
t
5
100m
0
8
8
5
1
section 80 m
2
6
7
4
2
0
4
6
2
1
2
section 60m
t+2
0
horizontal velocity [m/s]
i
1D Sub‐domain Ω1,1D 4
section 40 m
t+3
North Extraction shaft 1
0
2K+1
4
6
horizontal velocity [m/s]
5
2
South Extraction 1
shaft section 20 m
Tunnel Reference Section 1
K+2
3
Tunnel Reference Section 2 h [m]
1
h [m]
LD
2
2
0
4
L3D
3
h [m]
h [[m]
4
3
6
5
5
4
‐2
6
6
Far field
0
8
Dipartimento di Energetica – Politecnico di Torino
10
0
2
section 120 m
4
6
horizontal velocity [m/s]
8
10
Simulation of tunnel fire scenarios
¾1200 m long longitudinally ventilated
¾10 pairs of jet fans (50m spaced, 34 m/s discharge air velocity)
¾Fire located in the centre of the tunnel
Fire located in the centre of the tunnel
¾Fire max HRR: from 10 MW to 100 MW
Dipartimento di Energetica – Politecnico di Torino
Simulation of tunnel fire scenarios
¾Ventilation scenarios
¾Scenario 1: 3 jet fan pairs
¾Scenario 2: 5 jet fan pairs
j
p
¾Scenario 3: 10 jet fan pairs
Dipartimento di Energetica – Politecnico di Torino
Simulation of tunnel fire scenarios
Fire near field
Multiscale arrangement
North portal
1
K
K+1
K+3
K+2
K+5
K+4
i
t
t+3
t+2
t+5
t+4
1D Sub‐domain Ω2,1D 1D Sub‐domain Ω1,1D Γi
South portal
t+1
i+1
L3D
Γi+1
3D Sub‐domain Ω3D
Far field
Near field
Dipartimento di Energetica – Politecnico di Torino
Far field
N
Steady state results
Comparison to full CFD solutions (30 MW fire)
Scenario 1: 3 Jet fan pairs
full CFD
Multiscale
Highly accurate results are achieved!
Colella F et al Fire Technology (in press)
Colella F., et al., Fire Technology, (in press). Dipartimento di Energetica – Politecnico di Torino
Time dependent results
Velocity
HRR
Deteection
30 MW
2 min
15 MW/min
4 min
Dipartimento di Energetica – Politecnico di Torino
time
Time dependent results
temperature: 300 320 340 360 380 400
330
6
320
310
4
2
0
50
100
x-velocity:
6
-0 .
6
150
200
250
Longitudinal
coordinate
70 m back –layering -1 -0.8-0.6-0.4-0.2
0 0.2
0.4 0.6 0.8 1[m]
0.4
-0.4
0.6
1
4
2
0
50
100
150
0
200
Longitudinal coordinate [m]
Dipartimento di Energetica – Politecnico di Torino
0
elevatio
on [m]
e
elevation
n [m]
Multiscale results: 2 minutes
250
Time dependent results
Multiscale results: 3 minutes
S
Scenario 1: 3 Jet fan pairs
i 1 3J tf
i
5
0
400
100 m back –layering 500
600
700
Scenario 2: 5 Jet fan pairs
5
70 m back –layering 0
400
500
600
700
Scenario 3: 10 Jet fan pairs
5
0
400
0 m back –layering 500≈ 0 m back –layering 600
Longitudinal coordinate [m]
Dipartimento di Energetica – Politecnico di Torino
700
Time dependent results
Multiscale results: 5 minutes
S
Scenario 1: 3 Jet fan pairs
i 1 3J tf
i
5
0
400
500
600
700
800
900
1000
1100
1200
Scenario 2: 5 Jet fan pairs
5
0
400
500
600
700
800
900
1000
1100
1200
Scenario 3: 10 Jet fan pairs
5
0
400
500
600
700
800
900
Dipartimento di Energetica – Politecnico di Torino
1000
1100
1200
Conclusions (1)
Several numerical techniques have been used to
deal with ventilation and fire induced flows in tunnels
deal with ventilation and fire induced flows in tunnels
• 1D and CFD models
• Multiscale models
¾
Dramatic reduction of the computational time (up to 2 orders of
magnitude)
¾
As accurate as full CFD
¾
Results agree well with experimental measurements
¾
Allow for full coupling
p g of fire and ventilation system
y
and complete
p
analysis of ventilation system response
¾
Well suited to conduct parametrical studies, sensitivity analysis,
d d
studies
t di and
d Detection/Activation/propagation
D t ti /A ti ti /
ti problems
bl
redundancy
Dipartimento di Energetica – Politecnico di Torino
Conclusions (2)
Evaluation of Fire throttling effect
((…. the additional fire induced pressure losses due to sudden air expansion, higher p
p
, g
velocities, buoyant effects and localized losses in the plume region. …. amplified for larger fires and longer tunnels…)
6
# jett fan pairs
•Require
R
i a global
l b l simulation
i l i
of tunnel and ventilation
system to be evaluated
5
4
•Can be significant for large
fires (>100 MW) and long
tunnels!
3
•The number of jet fan pairs
required to achieve critical
velocity is highly dependent
of the fire size
1
2
Fire size [MW]
0
0
20
Dipartimento di Energetica – Politecnico di Torino
40
60
80
100
120
Conclusions (3)
Evaluation of the ventilation system response (Time required to remove back‐layering)
elapsed time from fire outbreaak [s]
400
Scenarios 1, 2, 3; TD=2min
Detection time 2.5 min
Scenarios 4 5 6; TD=2 5min
Scenarios 4, 5, 6; TD=2.5min
Detection time 2.0 min
Scenarios 7, 8, 9; TD=1.5min
350
300
Detection time 1.5 min
250
200
150
100
50
0
2
3
4
5
6
7
8
9 10
# active jet fan pairs
11
12
13
14
Significant impact of detection time on ventilation system response!
Dipartimento di Energetica – Politecnico di Torino
List of publications
Any questions?
yq
Dipartimento di Energetica – Politecnico di Torino
Tunnel ventilation models ‐ Network
Dipartimento di Energetica – Politecnico di Torino
Modelling techniques – 1D Network
¾Computational domain splitted in branches and nodes
¾Mass conservation in each node
¾Momentum conservation in each branch
¾Constitutive equations for components
• Fire Modelled as source of heat ((variable in time))
• Fans modelled using a fan characteristic curve
Dipartimento di Energetica – Politecnico di Torino
Modelling techniques – 1D Network
Jet fan characteristic curve
12
pressure rise [pa]
10
8
6
4
2
avrg velocity [m/s]
0
‐2
0
2
4
6
8
‐4
‐6
Dipartimento di Energetica – Politecnico di Torino
10
12
Multiscale – coupling procedure
Time step:
Time step: N N
Time step: N+1
Multiscale iterations ‐
Multiscale
iterations
K Time step: N+1 3.50
0.000
[kg/s]
‐0.010
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19
3.00
‐0.020
‐0.030
[Pa]
2.50
Total pressure at inlet
2 00
2.00
‐0.040
Introduction: three‐stage coupling
‐0.050
Mass flow rate at inlet
Mass flow rate at inlet
‐0.060
1.50
1.00
‐0.070
0.50
‐0.080
0.00
Dipartimento di Energetica – Politecnico di Torino