Fourth International Symposium on Tunnel Safety and Security, Frankfurt am Main, Germany, March 17-19, 2010

Design of Tunnel Ventilations Systems for Fire

Emergencies
Using Multiscale Modelling
F. Colella1,2 *, G. Rein2, V. Verda1, R. Borchiellini1, R. Carvel2, T. Steinhaus2 J. L. Torero2
2

1
Politecnico di Torino, Dipartimento di Energetica, Italy.
University of Edinburgh, BRE Centre for Fire Safety Engineering, UK
* corresponding author: e-mail address: f.colella@ed.ac.uk

ABSTRACT
This paper presents a novel and fast modelling approach to simulate tunnel ventilation flows during fire
emergencies. The complexity and high cost of full CFD models and the inaccuracies of simplistic zone or
analytical models are avoided by efficiently combining mono-dimensional (1D) and CFD (3D) modelling
techniques. A simple 1D network approach is used to model tunnel regions where the flow is fully
developed (far field), and a detailed CFD representation is used where flow conditions require 3D
resolution (near field). This multi-scale method has previously been applied to simulate tunnel ventilation
systems including jet fans, vertical shafts and portals (Colella et al 2009, Build. Environ. 44(12): 23572367) and it is applied here to include the effect of fire both in steady state and transient situations. The
methodology has been applied to a modern tunnel of 7 m diameter section and 1.2 km in length. Different
fire scenarios ranging from 10 MW to 100 MW are investigated with a variable number of operating jet
fans. Emphasis has been given to the discussion of the different coupling procedures for steady state and
transient calculations. An accurate discussion on the computational cost reduction as well as on the control
of the numerical error is also presented. Compared to the full CFD solution, the maximum flow field error
can be reduced below 2%, but providing a reduction of two orders of magnitude in computational time.
The much lower computational cost is of great engineering value, especially for parametric and sensitivity
studies required in the design or assessment of ventilation and fire safety systems.
KEYWORDS: multiscale modelling, jet fans, longitudinal ventilation system
INTRODUCTION
The response of the ventilation system during a fire is a complex problem. The resulting air flow within a
tunnel is dependent on the combination of the fire-induced flows and the active ventilation devices (jet
fans, axial fans), tunnel layout, atmospheric conditions at the portals and the presence of vehicles.
Depending on the accuracy required and the resources available, a solution to the problem can be reached
using different numerical tools.
The overall behaviour of the ventilation system can be approximated using 1D fluid dynamics models
under the assumptions that all the fluid-dynamic quantities are uniform in each tunnel cross section and
gradients are only present in the longitudinal direction. 1D models have low computational requirements
and are specially attractive for parametric studies where a large number of simulations have to be
conducted. In the last two decades several contributions on the application of 1D models to tunnel flows
have been published [1-5]. All the contributions underline the strength of 1D models to investigate
ventilation scenarios within tunnels. However, this modelling approach cannot predict the characteristic of
complex three-dimensional (3D) flow regions typically encountered close to operating ventilation devices
(jet fans), intersection of galleries (shafts) or close to the fire, where air entrainment, plume formation and
thermal stratification dominate the flow movement. Thus, in order to account for these important elements,
1D models must rely on approximated overall aerodynamics coefficients calculated somewhere else [5].
Computational fluid dynamics (CFD) remains the most powerful method to predict the flow behaviour due
to ventilation devices, large obstructions or fire but its use requires much larger computational resources
427

Fourth International Symposium on Tunnel Safety and Security, Frankfurt am Main, Germany, March 17-19, 2010

than simpler 1D models. In the last two decades, the application of CFD as a predictive tool for fire safety
engineering is becoming widespread and great achievements have been made. CFD simulations of full and
small scale tunnel fires have confirmed the capability of CFD tools as instruments to predict critical
velocity or temperature fields with a certain degree of accuracy [6-10].
Only a limited number of CFD studies directly focus on the performance of tunnel ventilation systems in
normal operating conditions or in case of fire [11-15]. This kind of analysis usually requires the adoption of
a very large computational domain including the tunnel regions where the operating ventilation devices are
located.
CFD analysis of fire phenomena within tunnels suffers from the limitations set by the size of the
computational domains. The high aspect ratio between longitudinal and transversal length scales leads to
very large meshes. The number of grid points escalates with the tunnel length and often becomes
impractical for engineering purposes, even for short tunnels domains less than 500 m long. An assessment
of the mesh requirements for tunnel flows is made by Colella et al. [16-17] for active ventilation devices
and for fire-induced flows. Grid independent solutions could be achieved only for mesh density larger that
4000 cells/m and 2500 cells/m for ventilation and fire induced flow respectively.
The high computational cost leads to the practical problem that arises when the CFD model has to consider
boundary conditions or flow characteristics in locations far away from the region of interest. This is the
case of tunnel portals, ventilation stations or jet fan series located long distances away from the fire. In
these cases, even if only a limited region of the tunnel has to be investigated for the fire, an accurate
solution of the flow movement requires that the numerical model includes all the active ventilation devices
and the whole tunnel layout. For typical tunnels, this could mean that the computational domain is several
kilometres long.
MULTISCALE MODEL
The study of ventilation and fire-induced flows in tunnels [8-10, 13, 15, 16] provides the evidence that in
the vicinity of operating jet fans or close to the fire source the flow field has a complex 3D behaviour with
large transversal and longitudinal temperature and velocity gradients. The flow in these regions needs to be
calculated using CFD tools since any other simpler approach would only lead to inaccurate results. These
regions are hereafter named as the near field [16]. However, it has been demonstrated for cold flow
scenarios [16] and for fire scenarios [10,17] that some distance downstream of these regions, the
temperature and velocity gradients in the transversal direction tend to disappear and the flow becomes 1D.
In this portion of the domain the transversal components of the flow velocity can be up to two orders of
magnitude smaller than the longitudinal components. These regions are hereafter named as the far field
[16]. The use of CFD models to simulate the fluid behaviour in the far field leads to large increases in the
computational requirements but very small improvements in the accuracy of the results.
The adoption of multiscale models represents a way to avoid the large computational cost of the full CFD
and the inaccuracies of simplistic assumptions of 1D models. A multiscale method uses different levels of
detail when describing the fluid dynamic behaviour of near field and far field. The behaviour of far field
and near field regions is modelled by using a 1D model and CFD model respectively. The 1D and the CFD
models exchange information at the 1D-3D interfaces and thus run in parallel.
This approach allows a significant reduction of the computational time. At the moment multiscale
modelling techniques have been applied to design tunnel ventilation systems in no-fire operating conditions
[16]. The application of multiscale techniques coupling the fire to the ventilations systems is the subject of
this paper. In this work, the prediction capabilities of full CFD and multiscale models are compared in
terms of accuracy and computational time using a modern tunnel as case study.
1D model of the far field
The general methodology of the 1D network model for tunnels is presented in [1 and 16] and only a brief
overview is given here. The developed 1D model is based on a generalized Bernoulli formulation. It is
designed to account for buoyancy effects, transient fluid-dynamic and thermal phenomena, piston effects
428

Fourth International Symposium on Tunnel Safety and Security, Frankfurt am Main, Germany, March 17-19, 2010

and transport of pollutant species. It is developed to handle complex layouts typical of modern tunnel
ventilation system (especially true for transverse ventilated tunnel) on the basis of a topological
representation of the tunnel network. The fluid dynamic model needs the mass and momentum
conservation equations to be solved in the whole domain (Eq. 1 and Eq. 2).

u
+
= SM
t
x
1
u

+ uS M + u 2 + p + gz = S Mx
x 2
t

(1)
(2)

where, represents the fluid density, u the longitudinal velocity, SM the mass source term per unit volume,
p the static pressure, and SMx the momentum source terms per unit volume acting along the longitudinal
direction, z the vertical elevation and g the gravity acceleration; x and t represent the longitudinal and
temporal coordinates respectively. The momentum source term contains all the terms related to wall
friction, losses due to flow separation at the portals and after obstacles. Eventually, pressure rise due to fan
operation and piston effect are also accounted for.
The calculation of the temperature profile within the tunnel domain needs the energy conservation equation
to be solved. A generic formulation for a tunnel portion of length dx states

T
T
2T

= 2 U (T Te ) + q v
+ c u
t
x
S
x

(3)

where is the fluid conductivity, U the global heat transfer coefficient between the fluid and the wall, is
the surface perimeter, S the cross section, Te the external temperature, c the specific heat and v the fluid
velocity along x. The term qv accounts for heat generation in the control volume (i.e due to fire). In general
the first term, representing the heat conduction along the longitudinal coordinate can be neglected if
compared to the other terms of Eq. 3.
In the case of steady state computations all the temporal derivatives in the mass, momentum and energy
conservation equations must be neglected.
The problem can be solved only after discretizing the computational domain in control volumes which
allows the integration of momentum, continuity and energy equations. The tunnel domain is discretized in
oriented elements called branches, interconnected by nodes. An example of a general network layout is
presented in Figure 1 where the nodes are numbered as i and branches as j.

Figure 1

Example of the network representation of a tunnel showing branches between nodes

CFD model of the near field

The CFD modelling of the near field has been conducted using the commercial CFD code FLUENT.
This code has been extensively used for simulating duct flows and it has demonstrated its capability to
model ventilation flows within tunnels as well as fire induced flows [8-10, 15, 16].
The turbulent fluctuations of the fluid-dynamic quantities have been modelled by means of Reynoldsaveraged Navier-stokes equations (RANS). Among all the RANS turbulence models, the k- model has
been applied in this work. The production and the destruction of turbulence kinetic due to the buoyancy
429

Fourth International Symposium on Tunnel Safety and Security, Frankfurt am Main, Germany, March 17-19, 2010

have been also accounted for. RANS models have been extensively used and largely validated by the
scientific community to simulate fire induced flows [6-8, 9-10, 13, 15].
The fire has been modelled as a volumetric source of energy at a constant rate without using a dedicated
combustion model. This simplified approach, previously used to model fire induced flows in tunnels [13,
15], is the most practical given its low computational cost. It avoids the burden and the complexity of
combustion and radiation models and the large uncertainty associated to the burning of condensed-phase
fuels. Greater details on the fire model are provided in [17].
The tunnel walls have been assumed as adiabatic. Other heat transfer boundary conditions to the walls
could be used, but the adiabatic condition represents the worst case in terms of buoyancy strength, threat to
people and damage to the structure.
The CFD model requires also a representation of the jet fans. The methodology used here simulates the
real construction of the jet fans as a cylindrical fluid region delimitated by walls and containing an internal
cross surface where a constant positive pressure jump is enforced
The simulations have been considered to be converged when the scaled residuals were lower than 10-5
with the exception of the energy equation where the maximum allowed value was 10-7. For time dependent
calculation time steps ranging from 0.1 s to 1 s have been used.
The effect of the interface location
One of the most important issues on the use of multiscale model for tunnel ventilation and fires is
related to the location of the interfaces between 1D and CFD models. These boundaries must be located in
regions of the domain where the temperature or velocity gradients are negligible and the flow behaves
largely as 1D. These dictate the length of the CFD domain. This required length is case specific and in
general depends on tunnel geometry, installation details of ventilation devices, presence of obstacle, etc.
Only broad guidelines are given here.
In previous work [16], the multiscale model was used to study the discharge velocity cone generated by
a jet fan installed in a modern tunnel and 1.5 km long. The analysis of the effect of the interface location
showed that the multiscale solution was within 1% of the full CFD solution when the length of the CFD
domain containing the jet fan was larger than 20 times the tunnel hydraulic diameter.
In another work [17], the multiscale model was used to simulate a several fire scenarios in the same
tunnel whose sizes ranged from 10 MW to 100MW. The analysis of the effect of the interface location
showed that the multiscale solution deviates from the full CFD solution by few percents when the length of
the CFD domain containing the fire was larger than 13 times the tunnel hydraulic diameter.
The simulation of tunnel fire scenarios when the ventilation flow is below the critical velocity must be
performed ensuring that there the back-layering does not extend upstream the inlet boundary. If these
conditions do not apply, the hypothesis of 1D flow pattern at the interfaces would not be valid, leading to
inaccuracies.
Coupling strategies
The solution to the multiscale problem requires the coupling of the 1D and CFD models. The iterative
solver procedure requires a continuous exchange of information between the models at the interface during
the computations. A three stage coupling has been adopted for the scope. A full 1D model of the system is
solved during the first stage. A CFD model of the near field is solved during the second stage. The
boundary conditions are provided by the full 1D model run at the first stage. The global multiscale
convergence can only be reached during the third stage when the 1D model of the far field and CFD model
of the near field are run sequentially k-times exchanging periodically the boundary conditions at the
interfaces (see Fig. 3). In comparison to more traditional coupling approaches, a three stage coupling
allows a significant reduction of the multiscale iterations needed to reach a global convergence.
The complete sequence of operations to be conducted during the solving procedure is described hereafter.
stage 1.

430

a. Run the full 1D model of the system until convergence is reached

b. Pressure and temperature values at the nodes corresponding to the interfaces i and i+1
must be recorded (to be used as boundary conditions of the CFD model in the next stage)

Fourth International Symposium on Tunnel Safety and Security, Frankfurt am Main, Germany, March 17-19, 2010

stage 2.

stage 3.

a. Run the CFD model of the near field until a certain degree of convergence is reached
b. Integrate the flow velocities at the interfaces i and i+1 to calculate the global mass flow
rate (to be used as boundary conditions for the 1D model of the far field in the next stage)
c. Calculate average temperature values at the interfaces i and i+1(to be used as boundary
conditions for the 1D model of the far field in the next stage)
a. Run the 1D model of the far field until convergence is reached
b. Pressure and temperature values at the nodes corresponding to the interfaces i and i+1
must be recorded and used as boundary conditions of the CFD model
c. Run the CFD model of the near field until a certain degree of convergence is reached
d. Integrate the flow velocities at the interfaces i and i+1 to calculate the global mass flow
rate (to be used as boundary conditions for the 1D model of the far field in the next stage)
e. Calculate average temperature values at the interfaces i and i+1(to be used as boundary
conditions for the 1D model of the far field in the next stage)
f. Check global convergence
i. If global convergence is not reached go back to point a (eventually a relaxation
step can be added)
ii. If global convergence is reached quit the calculation or proceed to the next time
step for time dependent calculation
Ti =

T v dA

Ai

v dA

Gext ,i = v dA

Ai

i-1

1D domain

Figure 2

'

Ti +1 =

Ai

v dA

Gext ,i = v dA
'

Ai

Ai +1

i
ptoti Ti

T v dA

Ai +1

i+1
CFD domain

ptoti +1 Ti+1

i+2 N-1

Coupling procedure between 1D and CFD models during stage 3

This way of coupling is called direct coupling. It allows a significant reduction in the computational time if
compared to the full CFD calculation of the same scenario. However the timescale of the direct coupling
calculations is limited by the computational speed solving the CFD portion of model, the near field. This
can take from some minutes to up to many hours to solve depending on the complexity of the scenario.
The global convergence check is performed by monitoring the evolution of any fluid-dynamic quantity at
the 1D-CFD interfaces during the inner k-iterations performed during step 3. In particular, the model
checks whether or not the deviation of a certain fluid-dynamic quantity computed in two sequential
multiscale iterations is lower than a fixed tolerance. Fig. 3 shows the evolution of total pressure and mass
flow rate computed at a 1D-CFD interface during a multiscale calculation. The maximum deviation
allowed was 10-6 which was reached after around 20 multiscale iterations. It is worth to note that, given the
high uncertainty characterizing tunnel ventilation flow calculations, lower accuracy (i.e. 10-3) can be used
shortening significantly the computing time (~10 multiscale iterations).

431

Fourth International Symposium on Tunnel Safety and Security, Frankfurt am Main, Germany, March 17-19, 2010

Multiscale iteration - k
0.00
2

3.00

1.E-01

9 10 11 12 13 14 15 16

mass flow rate

total pressure

-0.02
-0.03

2.50
2.00

-0.04
1.50

-0.05

1.00

-0.06
-0.07
-0.08

Figure 3

Deviation [kg/s or Pa]

1.E+00

Total pressure [Pa]

Mass flow rate [kg/s]

-0.01

3.50

Multiscale iteration - k
1

9 10 11 12 13 14 15

mass flow rate

total pressure

1.E-02
1.E-03
1.E-04
1.E-05
1.E-06

0.50

1.E-07

0.00

1.E-08

a). Evolution of total pressure and mass flow rate at a 1D-3D interface during a multiscale
calculation. The maximum deviation allowed was 10-6.
b). Deviation of the mass flow rate and total pressure at a 1D-CFD interface during a
multiscale calculation

CASE STUDY
The previously discussed multiscale model has been used to simulate a 1200 m long tunnel longitudinally
ventilated. This layout is realistic and typical of a modern generic uni-directional road tunnel. A schematic
of the tunnel layout is presented in Fig. 4. The tunnel is 6.5 m high with a standard horseshoe cross section
of around 53 m2. The tunnel is equipped with two groups of 5 jet fans pairs spaced 50 m, each group
installed near a tunnel portal. The jet fans are rated by the manufacturer at the volumetric flow rate of 8.9
m3/s with a discharge flow velocity of 34 m/s.
South
portal
North
portal
#1
125m

#2

#3

#4

#5

#6

#7

#8

#9 #10

50m
600m
north portal jet fan pairs #1 - #5
(approximate position)

south portal jet fan pairs #6 - #10

Fire source

(approximate position)

1200m

Figure 4

Layout of the tunnel used as case study showing the relative position of the fire, jet fans
and portals (not to scale).

The fires are located in middle of the tunnel and 4 different sizes ranging from 10 MW to 100 MW are
considered. For sake of simplicity, only results relative to 30 MW fires will be presented, while a more
detailed list of results is available in [17].
The emergency ventilation strategy, as for most longitudinally ventilated tunnels, requires the ventilation
system to push all the smoke downstream from the incident region in the same direction as the road traffic
flow, thus avoiding the smoke spreading against the ventilation flow (back-layering effect). The vehicles
downstream the fire zone are assumed to leave the tunnel safely. All the studies on back layering show that
the maximum critical velocity is in the range from 2.5 m/s to 3 m/s [8-9, 18-19]. Thus, an adequate
ventilation system has to provide air velocities higher than this range in the region of the fire incident.
Steady state simulation of fires
The HRR is assumed to be constant and that steady state conditions are reached within the tunnel. In this
section only 7 different ventilation scenarios (see Table 1) were considered. The number of operating jet
fans has been chosen in order to guarantee supercritical ventilation velocities in each fire scenario.

432

Fourth International Symposium on Tunnel Safety and Security, Frankfurt am Main, Germany, March 17-19, 2010

Scenario 1
Scenario 2
Scenario 3
Scenario 4
Scenario 5
Scenario 6
Scenario 7
Table 1

Jet fan
pairs #1- 2
x
x
x
x
x
x

Fire Size
30 MW
30 MW
30 MW
30 MW
10 MW
50 MW
100 MW

Jet fan
pair #3
x
x
x
x
x

Jet fan
pair #4
x
x
x
x

Jet fan
pair #5
x
x
x

Jet fan
pairs #6 - 10
x
x
-

Settings of the ventilation system for the four fire scenarios considered

The calculations have been performed by using direct coupling with a 400 m long CFD model of the near
field. The 1D representation of the rest of the tunnel included the 10 pairs of jet fans. A sketch of the
coupling is presented in Fig. 5.

K+2

K+4

North portal

340

38
0

54
0

K+5

500
460

K+3

5866
050
06042600
70

K+1

0 6 380
60

0
42

South portal

t+1

i+1
t

t+3
t+2

t+5
t+4

L3D

Far field

Figure 5

Near field

Far field

Multiscale coupling procedure to model the fire region and the rest of the tunnel

Table 2 presents a comparison of the bulk flow rate computed in the multiscale, full CFD and full 1D
predictions for the 7 scenarios under investigation. The multiscale results show a good agreement when
compared to full CFD predictions. The deviations range between 0.01% and 7%. On the other hand, it can
be seen that simple 1D model does not provide highly accurate bulk flow results since the deviations from
full CFD data range from around 12 % to 67% worsening for higher HRR.
Full Scale CFD
mass flow
rate [kg/s]

Scenario 1
Scenario 2
Scenario 3
Scenario 4
Scenario 5
Scenario 6
Scenario 7

Table 2

Multiscale direct
mass flow
rate [kg/s]
deviation

1D model
mass flow
rate [kg/s]
deviation
216
221
2.1%
277
28.2%
301
301
0.01%
356
18.1%
435
435
0.1%
490
12.6%
299
296
1.1%
351
17.3%
205
204
3.1%
236
15.4%
227
234
0.02%
310
36.5%
194
209
7.4%
325
67.3%
Accuracy of the multiscale simulation for the four fire scenarios considered. Comparison of full
CFD, multiscale (direct and indireact coupling) and 1D model results

The results are also presented in Fig. 6 as temperature and horizontal velocity fields on the tunnel
longitudinal plane. For sake of simplicity, only the results for the first scenario are presented. The rest of
scenarios can be found in [17].
The multiscale predictions are shown to be in excellent agreement with the full CFD predictions. In
particular, no important differences are observed the temperature fields. Some local deviations are observed
in the horizontal velocity fields. These are due to flow perturbations from the operating jet fans.

433

Fourth International Symposium on Tunnel Safety and Security, Frankfurt am Main, Germany, March 17-19, 2010

CFD: Tunnel Longitudinal section:

CFD: Tunnel Longitudinal section:

x-velocity: -3 -1 1 3 5 7 [m/s]
3
4

temperature: 325 375 425 500 600 [K]

325
580

590

600

610

620

Longitudinal coordinate [m]

425
37
5 4

00

630

640

580

590

temperature: 325 375 425 500 600 [K]

425
40 425
0
37
5
325

200

210

220

Longitudinal coordinate [m]

230

5
620

Longitudinal coordinate [m]

630

640

x-velocity: -3 -1 1 3 5 7 [m/s]
3
4

Figure 5

190

610

35

180

600

MULTISCALE: Tunnel Longitudinal section:

MULTISCALE: Tunnel Longitudinal section:

325

35
0

37

240 180

190

200

210

5
220

Longitudinal coordinate [m]

230

240

Comparison of results near the fire for the multiscale and the full CFD simulations for a fire
of 30 MW and ventilation scenario 1. Velocity and temperature values are expressed in m/s
and K respectively. The longitudinal coordinates start at the upstream boundary of the
corresponding CFD domain

Table 1 and 2 also show that the number of operating jet fans required to achieve critical ventilation
velocity in the fire region varies with the fire size. In particular 2, 3, 4 and 5 jet fan pairs must be activated
to provide super-critical ventilation velocity for a 10 MW, 30 MW, 50 MW and 100 MW fire respectively.
These results show that throttling effect of the fire is large as it was already proved experimentally in 1979
[20]. In [17] it has been showed that a 100MW fire in a 1.2 km long tunnel can produce a decrease in the
tunnel bulk velocity as high as 30%.
Time dependent simulation of fire emergency scenarios
Two ventilation scenarios have been considered in the time dependent simulations corresponding to
scenario 2 and 3 of Table 1. The fire growth curve has been designed following the recommendations
presented in [21]. It is constituted by an incipient phase (4 minutes long) characterized by a slow growth
rate (0.5 MW/min) followed by a second phase characterized by a higher fire growth rate (15 MW/min).
The maximum HRR is reached after around 350 s (see Fig. 6.Left). The detection time is assumed to be 2
minutes. The ventilation system is supposed to be activated once the fire has been detected. The temporal
duration of the simulated fire scenarios was 10 min when approximately steady state conditions were
achieved.
The multiscale calculations have been performed by using direct coupling with a 300 m long CFD model
of the near field. The 1D representation of the rest of the tunnel included the 10 pairs of jet fans (see Fig.
5). No time dependent full CFD calculation could be performed given the unreasonable computing time
required to conduct such analysis.

434

Fourth International Symposium on Tunnel Safety and Security, Frankfurt am Main, Germany, March 17-19, 2010

500
450

Time to detection

HRR [MW]

35
30
25
20
15

10

Mass flow rate [kg/s]

40

15
MW/min

Scenario 3

400
350
300
250
200
150
100

Scenario 2

0
0

100

200

300

400

500

600

3
2
1

100

200

time [s]

Figure 6

5
4

50
0

Velocity[m/s]

Incipient phase

45

Time to detection

50

300

400

500

0
600

time [s]

Left). Fire growth curve

Right). Time dependent evolution of the mass flow rate through the tunnel for scenario 2 and
3. The time to detection is 2 minute. Supercritical conditions (vair> 3m/s) are reached after
190 s and 160 s for scenario 2 and 3 respectively.

temperature: 300 320 340 360 380 400

330

320

310

4
2
0

50

100
x-velocity:

-0 .
6

200

150

250

-1 -0.8-0.6-0.4-0.2 0 0.2 0.4 0.6 0.8 1

0.4

-0.4

0.6

2
0

Figure 7

50

100

150

200

Longitudinal coordinate [m]

elevation [m]

elevation [m]

Figure 6.Right shows the temporal evolution of the mass flow rate through the tunnel as computed by the
multiscale model. Supercritical ventilation conditions are reached after 190 s and 160 s for scenario 3 and 2
respectively when the fire is still in its incipient phase and its HRR is lower than 2MW.

250

Multiscale results in the vicinity of the fire computed 2 min after the fire outbreak for
scenario 2 and 3. The ventilation system is about to be started. Velocity and temperature
values are expressed in m/s and K respectively. The longitudinal coordinates start at the
upstream boundary of the corresponding CFD domain

Figure 7 shows the conditions within the tunnel 2 min after the fire outbreak. As it can be seen velocity and
temperature profiles are still symmetric as the ventilation system has not been activated yet. The smoke
fronts are located around 110 m far away from the fire source.

435

elevation [m] elevation [m]

Fourth International Symposium on Tunnel Safety and Security, Frankfurt am Main, Germany, March 17-19, 2010

temperature: 300

Figure 8

360

380

400

100 p

330

0
32

310

200

150
310

5
0
50

340

310

5
0
50

320

3 20

310

310

100

250

150
200
Longitudinal coordinate [m]

250

Multiscale results in the vicinity of the fire computed 3 min after the fire outbreak for
scenario 2(top) and 3(bottom). Temperature values are expressed in K. The longitudinal
coordinates start at the upstream boundary of the corresponding CFD domain

temperature: 300

320

380

400

320

360

39

340

380

33
0

0
50

310

100

0
50

Figure 9

200

100

250
32

150
33

elevation [m] elevation [m]

Fig. 8 shows the multiscale results calculated for the fire near field 3 min after the fire outbreak. As in can
be seen the ventilation velocity is high enough to counteract against the smoke spread in both the cases.
The back layering distance has been reduced from 110 m to 70 m and ~0 m for scenario 2 and 3
respectively. Further analysis of the results have demonstrated that the upstream part of the tunnel is
completely cleared from the smoke after around 220 s and 180 s for scenario 2 and 3 respectively.
Furthermore it can be seen that, given the relatively low ventilation velocity, smoke stratification is
maintained both in the upstream and downstream regions.

310

150
200
Longitudinal coordinate [m]

340
330

250

Multiscale results in the vicinity of the fire computed 10 min after the fire outbreak for
scenario 2(top) and 3(bottom). Temperature values are expressed in K. The longitudinal
coordinates start at the upstream boundary of the corresponding CFD domain

For sake of simplicity only the conditions established in the tunnel 10 min after the fire outbreak are
presented (see Fig. 9). As it can be seen the ventilation velocity is high enough to avoid back-layering in
both the scenarios. Average higher temperatures are recorded for scenario 2 given the lower bulk flow
velocity achieved. Smoke stratification is completely destroyed in the fire downstream region in both the
scenarios.
It has to be asserted that the assumption of 1D flow at the upstream boundary must be maintained during
436

Fourth International Symposium on Tunnel Safety and Security, Frankfurt am Main, Germany, March 17-19, 2010

the whole calculation in order to achieve highly accurate results. This condition is always guaranteed for
supercritical ventilation scenarios. For subcritical ventilation conditions, the location of the upstream
boundary must be properly chosen to ensure that all the back-layering is captured within the CFD domain.
As it can be seen from time dependent calculations, the fire effluents formed a long back-layering (~ 110
m) at the early stages but it was largely contained in the CFD domain. A rough estimation of the backlayering distance to properly design the dimension of the CFD zone for sub-critical ventilation scenarios
can be obtained using the experimental correlation presented in [22]. For time dependent calculation a
rough estimation of the smoke front velocity and the consequent travelled distance can be based on the
correlation presented in [23]. However, a posteriori post-processing of the CFD results must always be
conducted to clarify this matter.
FINAL REMARKS
The simulation of tunnel ventilation flows and fires by using multiscale techniques is a relatively new
computational approach. The multiscale model, built by coupling 1D and CFD representations of the flow
phenomena, allows a significant reduction in the computational time, as the more time consuming tool is
only used to simulate a small part of the domain (near field). This region is characterized by high velocity
and temperature gradients. The regions within the domain where the flow behaves as a fully-developed
channel flow are modelled using a 1D model which is computationally effective and still able to provide
sufficiently accurate results for this type of flow. The numerical coupling of the model takes place at the
1D-CFD interfaces where the two models provide the required boundary conditions to each-other.
The reduction of the computational complexity is significant. In the presented examples for a modern 1.2
km long tunnel, the computational time was reduced by two orders of magnitude. Direct coupling
calculation required less than 2 hr and around 150 hr for steady state and time dependent simulations
respectively. However, the full CFD steady state simulations required between 48 and 72 hr to converge.
No time dependent full CFD calculation could be performed due to the untreatable amount of computing
time required.
The full coupling achieved between ventilation system and fire allowed the evaluation of the fire throttling
effect. It is pointed out that it can be as high as 30% for a 100MW fire in 1.2 km long tunnel.
The adoption of this novel approach requires care on the location of the 1D-CFD interfaces. Previous
studies [16,17] have showed that their position have a significant impact on the accuracy of the solution.
Better predictions are achieved when the interfaces are located in regions where the flow behaves in a 1D
fashion. The attainment of this flow pattern depends on the specific tunnel layout, jet fan characteristics, jet
fan installation details and the presence of any other obstacles.
Simulations of tunnel ventilation flow have some degree of uncertainty due to the real flow conditions at
the portals (atmospheric pressure and wind), wall roughness, fire load, fire geometry, throttling effects of
vehicles, and various other perturbing factors. For these reasons, the error introduced by using the
multiscale model is well within the uncertainty range of the calculations and is acceptable for engineering
purposes. On the other hand, the significantly lower computational time required by the multiscale models
offers the advantage that many ventilation scenarios can be considered.
Furthermore, it is asserted that the multiscale modelling approach represents the most practical way to
perform accurate simulations of coupled ventilation flow and fire in tunnels longer than a few kilometres,
when the limitation of the computational times becomes seriously restrictive
REFERENCE LIST
1. Ferro V, Borchiellini R, Giaretto V, Description and application of tunnel simulation model,
Proceedings of Aerodynamics and Ventilation of vehicle tunnels conference, 487-512, 1991.
2. Jacques E, Numerical simulation of complex road tunnels, Proceedings of Aerodynamics and
Ventilation of vehicle tunnels conference, 467-486, 1991.
3. Riess I, Bettelini M, Brandt R, SPRINT - a design tool for fire ventilation, Proceedings of
Aerodynamics and Ventilation of vehicle tunnels conference, 2000.
4. Cheng LH, Ueng TH, Liu CW, Simulation of ventilation and fire in the underground facilities,
Fire safety Journal, 36(6), 597 619, 2001.
437

Fourth International Symposium on Tunnel Safety and Security, Frankfurt am Main, Germany, March 17-19, 2010

5. Jang H, Chen F, On the determination of the aerodynamic coefficients of highway tunnels,

Journal of wind engineering and industrial aerodynamics, 89 (8), 869 896, 2002.
6. Woodburn PJ, Britter RE, CFD simulation of tunnel fire part I, Fire Safety Journal, 26, 35-62,
1996.
7. Woodburn PJ, Britter RE, CFD simulation of tunnel fire part II, Fire Safety Journal, 26, 6390, 1996.
8. Wu Y, Bakar MZA, Control of smoke flow in tunnel fires using longitudinal ventilation systems a study of the critical velocity, Fire Safety Journal, 35, 363-390, 2000.
9. Vauquelin O, Wu Y, Influence of tunnel width on longitudinal smoke control, Fire Safety
Journal, 41, 420426, 2006.
10. Van Maele K, Merci B., Application of RANS and LES field simulations to predict the critical
ventilation velocity in longitudinally ventilated horizontal tunnels, Fire Safety Journal, 43, 598
609, 2008.
11. Armstrong J, The ventilation of vehicle tunnels by jet fans the axisymmetric case, Proceedings
of the seminar on installation effects in fan systems, Mechanical Engineering Press, London, 1993.
12. Tabarra M, Optimizing jet fan performance in longitudinally ventilated rectangular tunnels,
Separated and complex flows, ASME FED 217, 3542, 1995.
13. Karki KC, Patankar SV, CFD model for jet fan ventilation systems, Proceedings of
Aerodynamics and Ventilation of vehicle tunnels conference, 2000
14. Massachusetts Highway Department, Memorial tunnel fire ventilation test program
comprehensive test report, 1995.
15. Galdo Vega M, Arguelles Diaz KM, Fernandez Oro JM, Ballesteros Tajadura R, Santolaria
Morros C, Numerical 3D simulation of a longitudinal ventilation system: memorial tunnel case,
Tunnelling and Underground Space Technology, 23(5), 53951, 2008.
16. Colella F, Rein G, Borchiellini R, Carvel R, Torero JL, V. Verda V, Calculation and Design of
Tunnel Ventilation Systems using a Two-scale Modelling Approach, Building and Environment,
44, 2357-2367, 2009.
17. Colella F., Rein G., Borchiellini R., Torero J.L., A Novel Multiscale Methodology for Simulating
Tunnel Ventilation Flows during Fires, Fire Technology, (in press).
18. Kunsch JP, Simple model for control of fire gases in a ventilated tunnel, Fire Safety Journa,l
37(1), 6781,2002.
19. Oka Y, Atkinson GT, Control of smoke flow in tunnel fires, Fire Safety Journal, 25(4), 30522,
1995.
20. Lee CK, Chaiken RF, J.M. Singer JM, Interaction between duct fires and ventilation flow: an
experimental study, Combustion Science & Technology, 20, 59-72, 1979.
21. Carvel, R., Design fires for tunnel water mist suppression systems, Proceedings of 3rd
International Symposium on Tunnel Safety and Security, Stockholm, 12-14 March, 2008
22. Ingason, H., Fire Dynamics in Tunnels. In The Handbook of Tunnel Fire Safety (R. O. Carvel
and A. N. Beard, Eds.), Thomas Telford Publishing, 231-266, London, 2005.
23. Heselden, A.J.M., Studies of Fire and Smoke Behavior Relevant to Tunnels, Current Paper
CP66/78, Building Research Establishment, 1978.

438