THEORETICAL AND APPLIED MECHANICS
Volume 45 (2018) Issue 2, 151–165
DOI: https://doi.org/10.2298/TAM171201010S
EXPERIMENTAL AND NUMERICAL ANALYSIS
OF FLOW FIELD AND VENTILATION
PERFORMANCE IN A TRAFFIC TUNNEL
VENTILATED BY AXIAL FANS
Milan Šekularac and Novica Janković
Abstract. To investigate air ѕow in longitudinally ventilated traffic tunnels,
a scaled model of a typical road-traffic tunnel with an appropriate ventilation
system based on axial ducted fans, is designed and built in the Lab. The focus
of this paper is the airѕow in a bi-directional traffic, two-lane tunnel. At the
scale ratio of approx. 1:20, at 20.52 m length it represents ≈ 400 m of a realscale tunnel. The model consists of two parallel tunnel tubes, where the main
tunnel (with a hydraulic diameter of Dh1 ≈ 0.4 m) has the geometry of a scaled
road traffic-tunnel. The second tunnel (Dh2 ≈ 0.16 m) has a smaller size and
is circular in cross-section, used only to simulate airѕow towards an evacuation
tunnel tube. Thus the two tunnels are connected by the evacuation passages,
equipped with adjustable escape doors. By a combination of experimental and
numerical work, the air ѕow-ҥeld and the performance of the ventilation system are investigated. The velocity ҥeld and its turbulence properties exiting
the fans were determined experimentally using hot-wire anemometry. These
data were further processed to be used in the tunnel ѕow computations by
CFD. The efficiency of momentum transfer (ηi , Kempf factor) of the longitudinal tunnel ventilation is determined. The effect that the imposed boundary
conditions and the level of their detail, have within a CFD computation of
tunnel airѕow, with respect to accuracy, velocity distribution and computed
ηi . Finally a traffic-loaded (traffic ҡjamә) case of ѕow is studied through experiment and CFD. The difficulty in assessing the required thrust of the plant
in traffic-jam tunnel conditions is discussed, and the ventilation efficiency is
estimated. Based on later results, the two limiting shapes of axial velocity
distribution with respect to height above the road, in this type of tunnel and
traffic, are estimated. The last result can be used as a realistic boundary condition (as inlet b.c. and/or initial condition) for numerical studies of ѕow and
ҥre scenarios in such tunnels with the traffic load critical for design.
2010 Mathematics Subject Classification: 76-05; 76D99; 76F99.
Key words and phrases: ventilation ѕows, tunnels, axial ducted fans, turbulence, numerical
boundary conditions, CFD.
The paper is presented at the 6th Serbian Congress in Theoretical and Applied Mechanics
(19—21 June 2017), Minisimposia Turbulence. The paper is a part of the collection of papers from
the Congress published in the issue 1 of the same volume.
151
152
ŠEKULARAC AND JANKOVIĆ
1. Introduction
A review of different design concepts for traffic tunnel ventilation can be found
in [1]. Recent advances in the ҥeld related to the design of more energy efficient
ventilation systems can be found in [2]. Consider here a longitudinal ventilation
system with axial ducted fans, attached to the tunnel ceiling, which can exert a
static thrust T , and the effective thrust exerted by these fans to the traffic tunnel’s
airstream is Tef :
(︁
vt )︁
· ηi = N · T · k 1 · ηi ,
(1.1)
T = ρ · Af · w2 , Tef = N · T · 1 −
w
where: w is the mean axial velocity of the fan’s jet, vt the tunnel air-stream mean
velocity, Af is the fan exit’s cross sectional area, while the Tef is the effective
thrust exerted on the tunnel’s air ѕow by the system equipped with N fans, [1, 3].
Factor k1 denotes the inѕuence of the mean axial velocity of the tunnel ѕow which
offloads the fan, while ηi factor is usually referred to as installation efficiency, Kempf
factor [1, 2], and denotes the efficiency of momentum transfer from the jet fans to
the air stream in the tunnel. Factor k1 is of kinematic character, it can be increased
only by increasing the velocity of the jet at the fan’s exit, since the tunnel air-stream
velocity is predetermined by ventilation design requirements. Some authors split
the ηi into two factors: k2 and k3 . The former should account for the inѕuence of
the relative separation between the tunnel ceiling and the fan’s axis, while the later
should account for the possible downward tilting of the fan’s jet axis with respect to
the tunnel longitudinal axis, which can increase the ηi . In this paper the distance
of the fan axis to the ceiling had a common value of approximately equal to fan
diameter and the fan axis was parallel to the tunnel axis. Aside the aforementioned
aspects, the overall energy efficiency of the ventilation system is inѕuenced by the
hydraulic efficiency of the fan which is in the ∼ 0.40ҫ0.80 range and the motor
efficiency ∼ 0.90ҫ0.95. These factors greatly decrease the overall energy efficiency
of the system, whereas the space for improvement is mainly related to the increase
of the fan hydraulic efficiency, the fan’s jet velocity and the momentum transfer
efficiency ηi [2]. The effective thrust required to push the air-ѕow in the tunnel
can, in principle, be determined by using the following formula [1, eq. 3]:
)︁
(︁
∑︁
Lt
2
T = ζin + ξ
+ 1 At ρvtun
/2 +
(1.2)
ζ+ Ai ρ(vv + vtun )2 /2
Dh
veh+
∑︁
ζ− Ai ρ(vv − vtun )2 /2,
+
veh−
where vv is the absolute value of vehicle velocity, the ς+ , ς− denote the hydraulic
resistance coefficients of the vehicles in their streamlined and their opposite orientations, respectively (for a bi-directional traffic situation studied here), whereas the
dynamic pressure term is computed using the absolute value of the air’s relative
velocity with respect to the vehicle.
The aforementioned efficiency factor ηi is dependent on the properties of the
turbulent ѕow generated by the fans, in particular the secondary ѕow components
of the exiting jet swirl [3ҫ5]. However, in the published studies with numerical
EXPERIMENTAL AND NUMERICAL ANALYSIS OF FLOW FIELD...
153
simulation of tunnel airѕow, the fans are usually represented as momentum sources
or jets of a constant axial (mean-ѕow) velocity w. The w value is predetermined
by the available fan’s static thrust measurements [2, 6, 7].
On the other hand, the results of the available experimental studies on ҥre
and smoke movement in the tunnel are based on experiments obtained using scaled
laboratory installations where the airѕow in the model-tunnel is induced by an
external source (fan or pressure tank) through a ѕow-straightening honeycomb, so
the velocity proҥles in the tunnel are ѕat when they approach the ҥre source [8ҫ12].
The properties of ѕow generated at the fan’s exit will indirectly inѕuence the
smoke movement and control. The fan design, possible fan jet tilting, will affect
the axial velocity distribution [13]. It is noted [12] that a more efficient design will
reduce the friction losses on the tunnel ceiling and redistribute the axial velocity
by lowering its value near the ceiling, which may promote smoke back-layering.
One can expect the changes in velocity distribution to have an effect of vertical
stratiҥcation of temperature and pollutant ҥelds too. Some recent papers address
this topic with respect to fan location and other inputs, in an empty tunnel ѕow
case [13]. These effects should be further investigated. For the smoke control
problem, an overview of the available literature results for estimating the critical
velocity and smoke back-layering length is given in [14].
This paper attempts to give a better insight into the details of the ѕow ҥeld
and the results accuracy attainable through CFD, with respect to used boundary
conditions. An experiment in the Lab tunnel model is used for veriҥcation.
2. Experimental results
The sketch of the built experimental installation [3] and the CAD model of the
utilized axial fans are shown in Figure 1.
The installation consists of two parallel tunnel tubes: the main tunnel (at a
scale ratio ≈ 1 : 20, shaped as a road-traffic tunnel) with a hydraulic diameter of
Dh1 ≈ 0.40 m) and a secondary tunnel tube circular in cross section (Dh2 = 0.16 m),
which can simulate airѕow from/to an evacuation tunnel tube. Tunnels are connected to the evacuation tunnel by three evacuation passages (EP), and are approx.
100 fan diameters DF apart, equipped with shut-off valves. The characteristic distance L between fan batteries (F1F2, F3F4, F5F6) is approx. ∼ 5 m or ∼ 100 fan
diameters, or 12.5 Dh1 . At Lt = 4L = 20 520 mm tunnel length the installation
corresponds to approx. 400 m of a real-scale traffic tunnel.
2.1. Experimental results. The utilized fans are model-aircraft vehicle
(UAV) propulsion axial fans by Great Planes ҡHyperѕowә with a 55 mm rotor diameter, the rotor spinning rate in the results reported in this paper was 27 500 RPM.
The fan exerts approx. 2.28 N of static thrust and generates a jet with a mean
axial exit velocity w = 34.5 m/s. The detailed results of experimental study on
the turbulent ѕow generated with such fans can be found in [16] where a hot-wire
anemometry measurement-system with an appropriate signal processing procedure
was utilized to determine the ѕow. The details on hot-wire probe design and the
measurement setup can be found in [17]. The mean value of axial velocity at fan
154
ŠEKULARAC AND JANKOVIĆ
exit obtained by static thrust measurements on the test-table and use of eq. (1.1)
agree very well with the value obtained by numerical integration of the measured
turbulent ѕow ҥeld. A comparison of these results with larger-scale fans reveals
that the dimensionless proҥles of turbulence kinetic energy, turbulence intensity,
and the dimensionless Reynolds stresses, show very similar proҥles and maximum
values for this fan compared to larger-scale fans (larger by a factor of ∼ 10ҫ20,
at approx. correspondingly lower rotor speeds, typically ∼ 1500 RPM), [3]. The
time-averaged axial, radial and tangential components of fan’s ѕow ҥeld, given in
detail in [3, 16] are presented in Figure 2.
The average values of yaw and pitch angles are: 3∘ and 2.3∘ respectively. By
utilizing the deҥnition for total turbulent ѕuctuations [3], graphs of time-averaged
turbulent intensity T u and dimensionless turbulent kinetic energy TKE were computed [3], used further in this paper, see Figure 4.
The maximum and mean-value of dimensionless TKE over the ѕow-exit area
are approx. 0.33 % and 0.10 %. The turbulence intensity mean-value is 4.41 %
whereas the maxima are at 13.1 %. Fan exit zones are the areas of the highest
turbulence intensities in the tunnel.
The hydraulic properties of the installation were measured in the Lab: the
effective relative roughness of the tunnel tubes (0.00035), the entry losses coefficient
a)
b)
c)
Figure 1. (a) Installation photo, (b) 3D CAD model of the fan
(shroud is removed), (c) Scheme of the instalation.
EXPERIMENTAL AND NUMERICAL ANALYSIS OF FLOW FIELD...
a)
b)
155
c)
Figure 2. Time-averaged velocity components of fan exit turbulent ҥeld: (a) Axial vz , (b) Radial vr , (c) Tangential vz .
a)
b)
Figure 3. Fan ѕow turbulence properties: a) intensity T u% over
the whole fan exit cross section area, b) dimensionless TKE (1/6
of ѕow’s exit plane is shown).
(0.564) and the hydraulic resistance coefficients of the used vehicle models: see
Table 2, and [3]. Results are given for the empty tunnel ѕow in Table 1, and with
stationary vehicle models in Table 3.
Table 1. Measured and derived values for ѕow in traffic-free
model tunnel.
T [N ]
≈ 7.2
≈ 13
v[m/s]
4.175
6.27
w[m/s]
25
33.5
1 − v/w[/]
0.83
0.81
ηi [/]
≈ 0.70
≈ 0.85
Two traffic-load densities where adopted corresponding to two load regimes:
normal (corresponding to a real tunnel with 80 km/h vehicle speed) and congested
(at 10 km/h), [3]. The number of vehicles is estimated using RVS formulae [3].
For the scaled conditions of model tunnel and traffic share of 25 % of large vehicles
(buses) and 75 % passenger cars, the proper number of vehicles is: a) ∼ 43 passenger
vehicles and 11 buses in the congested-traffic case; b) ∼ 11 passenger vehicles
and 3 busses, at the 80 km/h regime case. To determine the hydraulic resistance
coefficients of the used model-vehicles an experiment was conducted. The obtained
values are reported in Table 2.
156
ŠEKULARAC AND JANKOVIĆ
Table 2. Measured values of hidraulic resistances: streamlined
and opposite orientations.
Vehicle
Car
Bus
ζ+
≈ 0.49
≈ 1.1
ζ−
≈ 0.71
≈ 1.1
From the conducted experiments, the following values of k1 and ηi are determined, see Table 3:
Table 3. Measured and derived values for traffic loaded tunnel ѕow.
Reg.
a)
b)
T [N ]
≈ 12.8
≈ 12.8
v [m/s]
≈ 4.12
≈ 4.90
w [m/s]
33.5
33.5
k1
0.88
0.86
ηi
≈ 0.80
≈ 0.70
The ηi value is in the 0.70ҫ0.85 range for fan-jet speeds above 30 m/s, in the
empty tunnel ѕow case. In the traffic-loaded tunnel the results are in the range
0.70ҫ0.80. To analyse the aforementioned issue and the relative inѕuence of fan
velocity ҥeld on the derived results, the numerical simulations are used further.
3. Numerical results
3.1. The mathematical model and the numerical grid. A CAD model
of the tunnel installation was made to be identical to the real laboratory tunnel
model. The fans were modeled accordingly, without particular approximations
in their geometry, with an air-domain of the fan in the form of annuli between
the fan’s shroud and fan’s hub. The experimentally determined properties of ѕow
ҥeld were imposed there as boundary conditions. A hybrid-type of grid with 8.7
million elements in the ѕuid-domain was used. The elements consisted of prismatic
and tetrahedral ҥnite-volumes with boundary layers around fan surfaces, tunnel
walls, vehicles-models were covered with layers of prismatic cells, in the form of
prisms with triangular basis. The bulk mesh tetrahedral cell size spans from ≈
1 mm (which is ≈ 50 % of the mean-value of integral length scales of the turbulent
ѕow at the fan’s exit) up to 25 mm (6 % of tunnel’s hydraulic diameter) far from
the fans. The wall-normal size of prismatic cells at the walls was ҥne enough to
provide an average y + ≈ 1. Flow was computed by numerically solving a system of
incompressible RANS (Reynolds-averaged NavierҫStokes) equations in the Fluent
ѕow solver [3]:
(3.1)
∂v̄i
∂(ρv̄i v̄j )
∂ p̄
∂ 2 v̄j
∂Rij
+
=−
+µ 2 +
+ fi
∂xj
∂xj
∂xj
∂xj
∂xj
(3.2)
∂ρ ∂(ρv̄i )
+
=0
∂t
∂xi
EXPERIMENTAL AND NUMERICAL ANALYSIS OF FLOW FIELD...
157
where the Rij = −ρvi′ vj′ is the Reynolds turbulence stress tensor which has to
be modeled. Turbulence was modeled using standard k − ε and the k − ω SST
turbulence models. Enhanced wall function approach was used with the k−ε model.
All the model constants had their default values (Cµ = 0.09, C1ε = 1.44, C2ε = 1.92,
*
P rk = 1, for k − ε; and for the k − ω SST model the constats were: α∞
= 1,
*
*
α∞ = 0.52, β∞ = 0.09, ζ = 1.5). Appropriate boundary conditions were used and
are discussed bellow. The iterative procedure was carried until all the residuals
were reduced under 10−5 , and integral quantities showed no further change.
3.2. CFD cases and boundary conditions. Boundary conditions for presented CFD cases are listed in Table 4. Two layouts of the tunnel installation were
considered: 1) A tunnel with no vehicles with the ventilation system exerting the
thrust (cases S1, S2, S3, S4); 2) Tunnel with a congested traffic case occurring at
a 10 km/h vehicle speeds in a two-lane traffic tunnel with a bi-directional traffic,
in the two-opposite traffic lanes. The adopted vehicle share (in CFD-cases S5, S6,
S7, S8) is approx. 30 % of large vehicles (model buses) vs model passenger cars.
The number of vehicles in the analysed CFD tunnel was 38 passenger cars and 12
buses. A segment of 1.75Dh1 near the exit tunnel-portal was clear from vehicles,
while the ҥrst vehicles are 0.6Dh1 downstream of the entry-portal.
Table 4. Overview of numerical cases and boundary conditions
on the fans (and tunnel inlet).
Case
S1
S2
S3
S4
S5
S6
S7
S8
Model
k−ε
k − ω SST
k − ω SST
k − ω SST
k−ω
k−ω
k−ω
k−ω
SST
SST
SST
SST
Boundary conditions on fans
v x , vy , vz
vx , vy , vz ,TKE
vz = w = const, v̄x = 0, v̄y = 0
vr ≡ 0 → vx* , vy* , vz ,TKE
with traffic, fans ON: vx , vy , vz
with traffic, fans OFF: (△ptot )in
S5
with traffic, fans OFF: vin = vtun
with traffic, fans ON, Re = Reob
For S1 case a full 3-component time-averaged turbulent ѕow ҥeld was prescribed
on all fans, whereas in S2 an added Dirichlet condition of the measured TKE ҥeld
at the fan exit was used. In S3 case a simpliҥed fan representation as a constant
axial-velocity ѕow jet was used. In S4 case the ҥelds of vx , vy , vz ,TKE where used,
but secondary ѕow was modiҥed to represent an idealised nozzle with a zero radial
component. Since the vehicle’s surface have a velocity of 10 km/h = 2.78 m/s
the obtained results represent an averaged (quasi-steady) picture, i.e., an averaged
time-snapshot of a ѕow ҥeld, since the overall and averaged velocity results will be
independent of the vehicle instantaneous position in this case.
To account for the hydraulic resistances induced by vehicles in the tunnel and
isolate the ventilation system’s efficiency in such speciҥc ѕow conditions, two separate additional CFD cases were computed: S6 and S7. There the identical layout
158
ŠEKULARAC AND JANKOVIĆ
a)
b)
Figure 4. Features of the computed ѕow ҥelds: S1-case.
was used but the ѕow was induced not by operating the ventilation fans inside the
analysed tunnel segment but by an external thrust source, outside of the considered CFD model-tunnel space. In S6 case a speciҥed gauge total pressure δptot was
prescribed, and in S7 case an inѕow-velocity vt at the tunnel entry portal. The air
was able to freely pass through the fan annulus space. The air physical properties
for all computed cases were: µ = 1.8 · 10 × 10−5 Pa · s, ρ = 1.238 kg/m3 , and they
correspond to the conditions of conducted laboratory experiments.
3.3. Numerical results for the empty tunnel ѕows. An overview of the
computed velocity ҥeld is given in Figure 5. The velocity proҥle with respect
to height above road-surface is presented in Figure 6. The axial time-averaged
a)
b)
Figure 5. Axial velocity with respect to height above road surface: a) in S1 case taken at different cross-sections along the tunnel:
13Dh ҫ51.3Dh from the entry portal, b) Dimensionless axial velocity for S1ҫS4 cases, and the experiment, taken at the midpoint of
the last tunnel section (L-long).
EXPERIMENTAL AND NUMERICAL ANALYSIS OF FLOW FIELD...
159
velocity proҥles with respect to y/ymax = (0 − 1) were plotted for 7 different
equidistant cross-sections downstream along the tunnel (at the exit planes of the
fan batteries, the outlet portal and at the mid-sections between these): v * = v̄z /vt
The average value of y + over the tunnel wall with the used numerical grid was
fairly low: 1.075. The axial velocity proҥle starts from a ѕat turbulent proҥle at
the tunnel entry region, deforms along the tunnel into an asymmetrical proҥle with
2ҫ3 times higher velocities near the tunnel ceiling. This occurs intermittently - from
the F1F2 battery exit cross-section, it deforms from ѕat to asymmetric and later
ѕattens again as the ѕow approaches the next fan battery. Some results are given
in Table 5, including a skewness-coefficient of the axial velocity proҥle taken at the
midpoint of the last section (L-long) of the tunnel; the point C is the midpoint
of tunnel height at the symmetry-plane of the tunnel. Experimental result was
obtained by a small Pitot tube, on the Lab. tunnel-model in the same ventilation
operating conditions. The computed under-pressure value in the cross-section of
the ҥrst fan battery F1F2 exit corresponds to the measured under-pressure in the
Lab. model tunnel. It arises from the sum of all losses (entry loss, friction losses
and the achieved ѕuid dynamic pressure).
Table 5. Comparison of results of S1ҫS4 cases and the experiment.
Case
Exp
S1
S2
S3
S4
vt
6.41
6.44
6.23
6.64
6.38
vC
7.06
7.19
7.50
7.60
7.30
k1
0.81
0.813
0.82
0.807
0.815
ηi
0.82
0.825
0.81
0.891
0.806
Skewness coeff.
0.263
0.204
0.048
0.072
0.057
Figure 6. Average static pressure over a tunnel’s cross-section along
the tunnel at z/Dh downstream of the tunnel inlet (z is the distance
from inlet). The values are taken at the each battery exit positions, and
at midpoints between them.
160
ŠEKULARAC AND JANKOVIĆ
Figure 7. Velocity ҥeld in a traffic loaded funnel ѕow of S5 case
(contours of absolute velocity).
3.4. Numerical results for ѕow in the tunnel with traffic. Flow features
in the computed case S5, are given in Figure 8. In the presence of vehicles, the ventilation system exerted thrust is consumed to counteract the hydraulic resistances
of the tunnel and the moving vehicles in the two traffic lanes with opposing traffic
directions.
In order to evaluate the ηi , the required cumulative pressure drop could be, in
principle, calculated using equation (1.2), but this requires an accurate information
of speciҥc vehicle’s resistance coefficients, in the tunnel ѕow layout. To avoid this
problem, two additional CFD cases are computed in identical geometrical layout
as the previous S5-case, but the ventilation fans are switched ҡoffә while the thrust
is exerted by some source outside of the tunnel (upstream of the tunnel entry).
Such CFD-case can reveal the velocity distribution which will occur in the tunnel
much further downstream of the last operating fan-battery. Or, if the thrust is
exerted by outside atmospheric sources (like wind or existing pressure differences
between portals).
Thus, the boundary condition at the tunnel entry portal in S6 and S7 were:
a) In the S6-case a user-prescribed gauge total pressure of 90 Pa (which will
produce tunnel velocity of similar value as in S5-case with working ventilation) and
a supplied entry-loss coefficient;
b) In the S7-case the user-supplied constant-proҥle inlet velocity vt , with a
value prescribed as identical to the mean-value of ѕow velocity vt computed in the
S5-case where the same layout was studied but with the fans operating. In this
way, it was possible to numerically compute the required pressure drop (and thus
the effective thrust force Tef ) in the tunnel and using equation (1.1) rather than
equation (1.2), isolate the ηi value in the S5 case, for such ѕow conditions with
higher adopted traffic load (at 10 km/h layout).
The obtained results are given in Table 6.
It is reasonable to expect that for any given tunnel of sufficient length, the
axial-velocity proҥle for an arbitrary cross-section along the tunnel should have the
shape between these two estimated extremes, shown in Figure 8.
The S8 case was computed as equal to the S5-case, only the Reynolds number
of ѕow was adjusted to correspond to a 20 times larger size tunnel, to evaluate
EXPERIMENTAL AND NUMERICAL ANALYSIS OF FLOW FIELD...
161
Table 6. Results for traffic-loaded tunnel ѕow cases
Case
S5
S6
S7
S8
User input b.c.
(vx , vy , vz , k), ζin
△ptot at inlet, ζin
vtun at inlet
(vx , vy , vz , k), ζin
vt
4.77
4.975
4.77
4.95
k1
0.863
0.857
0.863
0.858
ηi
≈0.86
≈0.92
a possible inѕuence of Re number on the results in larger-scale object ѕow. The
computed mean-value of axial velocity is approx. 3.9 % higher than in the S5-case
with model-tunnel Re-number ѕow. It can be concluded that the Re number has
a small inѕuence on the results, and the obtained large-scale tunnel ѕow values
of mean-ѕow velocity and ventilation efficiency are a few percent higher than for
model-scale cases, whereas the average axial-velocity proҥle has a fairly similar
shape, Figure 8.
Figure 8. The envelope of possible dimensionless axial-velocity
distribution in a traffic-loaded tunnel ѕow case, with respect to
dimensionless height above road surface. Obtained by averaging
the velocity proҥles in the last segment, of characteristic length L
(100 fan diameters).
The available static thrust of the ventilation system in the model-tunnel S1-S7
cases is approx. 13.6 N or 91.4 Pa. Given the same hydraulic resistance conditions
in the tunnel in the three considered traffic-jam cases (S5ҫS7) it can be calculated
that the efficiency of momentum transfer for the S5 case (with working ventilation)
is 0.86, see Table 6.
An additional effect is observed by comparing the required thrust to a value
which could be computed by using the experimentally determined resistance coefficients for these model-vehicles and equation (1.2), as is usually done in ventilation
design. Using computed CFD results, it can be concluded that the appropriate
effective values of hydraulic resistance coefficients of the here used model-vehicles
162
ŠEKULARAC AND JANKOVIĆ
in the traffic layout analysed in traffic-cases S5, S6, S7 are lower than the values
in Table 3, i.e., 0.30 vs 0.49 for passenger cars. Although values in Table 3 are
subject to experimental error (which was an additional motivation to follow the
approach presented in this chapter when computing traffic-loaded tunnel ѕow cases
S5ҫS7) the appearance of lower effective values of vehicle-resistance coefficients in
tunnel ѕows can be attributed to the drafting or slipstreaming phenomenon, given
the congested traffic layout at vehicle speed 10 km/h.
4. Conclusions
The following results are obtained through the conducted experimental and
numerical work:
∙ From the obtained results of ѕow measurements (on fans and on the scaled
Lab. tunnel model) a value of efficiency for momentum transfer ηi for the tunnel
with no traffic load is: ≈ 0.70ҫ0.80, with higher values corresponding to higher
fan-jet exit velocities (usually up to 35 m/s). For the ѕow in a Lab. model tunnel
occupied with stationary vehicles the estimated ηi value is approx. 0.70.
∙ The analysis of the relative inѕuence of the boundary conditions for a numerical simulation of tunnel ѕow and their effect on the achieved results, show that
the fan’s ѕow properties do have a non-negligible effect on the computed tunnel
ѕow-ҥeld, in terms of the derived conclusions, like ηi values. The determined ηi
value is approx. 0.80 with the used fans. The negative effect of the secondary ѕow
components is assessed. Thus depending on the adopted boundary conditions in the
numerical CFD case setup the ηi value varies from 0.81 to 0.89, i.e., by about 9 %;
∙ In a road-tunnel of a bi-directional traffic type, congested with vehicle-traffic
in two traffic lanes occurring at the adopted standard critical regime of 10 km/h,
the ηi value was assessed using CFD: the results show there isn’t any noticeable
deterioration of the ventilation efficiency ηi value in this speciҥc case, compared
to the traffic-free tunnel ѕow case. The ηi value of 0.86 is obtained, vs 0.85 value
in the empty tunnel ѕows. The experiment showed lower values but its result
here should be taken with caution given the difficulties in estimating the effective
vehicle’s resistance forces in the model-tunnel, and the mean-value of axial velocity
in such disturbed ѕow;
∙ The effect of adopted boundary conditions on mean-value of ѕow velocity
in S1ҫS4 cases is small (approx. 3.3 %), but more important on velocity proҥle
asymmetry and the secondary ѕows in the tunnel. The maxima of axial velocity
proҥle values under the tunnel ceiling, in the empty-tunnel ѕow cases, fall approx.
within a 16 % range scatter;
∙ The real-scale Re-number numerical results (S8-case) are very similar to
model-tunnel numerical results (S5-case), but with a computed value of the meanѕow velocity 4 % higher (using the same boundary conditions);
∙ In addition, the CFD results of tunnel ѕow with vehicle-traffic load reveal a
possible slipstreaming effect of vehicle resistance in the traffic-jam case, i.e., possibly
lower effective values of hydraulic resistance coefficients of the model-vehicles in
the tunnel, compared to the values estimated experimentally for the same vehicle
models in a single-vehicle tunnel ѕow;
EXPERIMENTAL AND NUMERICAL ANALYSIS OF FLOW FIELD...
163
∙ The asymmetrical shape of the axial velocity proҥle in the traffic-loaded
tunnel-ѕow (at 10 km/h traffic density regime) is estimated in the form of an envelope of 2 limiting extremes. The shape of the velocity distribution will thus depend
only on the distance of the analyzed cross-section with respect to the upstreamnearest operating fans, or if the fans do not operate but the ѕow is induced by
atmospheric overpressure at the inlet, on the distance to the tunnel inlet. For any
such tunnel ѕow situation, the distribution of velocity will have a shape within this
envelope. Thus this result can be used as an appropriate inlet boundary condition, and/or initial condition in setting up a numerical simulation of ѕow and ҥre
scenario, in a bi-directional traffic tunnel situation, with critical traffic load.
Acknowledgments. This work was funded by the grant from the Ministry of
Education, Science and Technological Development (MESTD), Republic of Serbia,
Project No. TR 35046 and Bilateral cooperation between MESTD, Serbia and
Ministry of Science, Montenegro (2016ҫ2018), coordinated between University of
Belgrade, Faculty of Mechanical Engineering and University of Montenegro, Faculty
of Mechanical Engineering: Investigation of the inѕuence of turbulent swirl ѕow on
the energy parameters of the axial fans by use of the contemporary measurement
techniques, what is gratefully acknowledged.
References
1. F. Tarada, R. Brandt, Impulse ventilation for tunnels - a state of the art review, 13th International Symposium on Aerodynamics and Ventilation of Vehicle Tunnels, New Brunswick,
New Jersey, 2009.
2. F. Tarada, Design, testing and application of an energy-efficient longitudinal ventilation system, 14th International Symposium on Aerodynamics and Ventilation of Tunnels, Dundee,
Scotland, 2002.
3. M. Šekularac, Analiza strujnih polja slozenih sistema ventilacije saobracajnih tunela, PhD
thesis,University of Montenegro, Faculty of Mechanical Engineering, 2015.
4. E. Jacques, P. Wauters, Improving the Ventilation Efficiency of Jet Fans in Longitudinally
Ventilated Rectangular Ducts, Universite catholique de Lovain-la-Neuve, 2009.
5. A. D. Marteggani, G. Pavesi, C. Barbetta, Uno studio sperimentale sulla ventilazione longitudinale nelle gallerie, Gallerie e grandi opere sotterranee 48 (1996).
6. P. Levoni, D. Angeli, E. Stalio, E. Agnani, G. S. Barozzi, M. Cipollone, Fluid-dynamic characterisation of the Mont Blanc tunnel by multi-point airflow measurements, Tunneling and
Underground Space Technology 48 (2015), 110ҫ122.
7. E. Eftekharian, D. Alireza, O. Abouali, J. Meigolinedjad, G. Ahmadi, A numerical investigation into the performance of two types of jet fans in ventilation of an urban tunnel under
traffic jam conditions, Tunneling and Underground Space Technology 44 (2014), 56ҫ67.
8. Y. Wu, M. Z. A. Bakar, Control of smoke flow in tunnel fires using longitudinal ventilation
systems - a study of the critical velocity, Fire Safety Journal 35 (2000), 363ҫ390.
9. H. Ingason, Y. Z. Li, Model scale fire tests with longitudinal ventilation, Fire Safety Journal
45 (2010), 371ҫ384.
10. L. Barbato, F. Cascetta, M. Musto, G. Rotondo, Study of critical velocity and backlayering
length in longiudinally ventilated tunnel fires, Fire Safety Journal 45 (2010), 361ҫ370.
11. ] J. S. Roh, S. S. Yang, H. S. Ryou, M. O. Yoon, Y. T. Jeong, An experimental study on the
efect of ventilation velocity on burning rate in tunnel fires - heptane pool fire case, Building
and Environment 45 (2008), 1225ҫ1231.
12. F. Tarada, New prospectives on the critical velocity for smoke Control, International Symposium on Tunnel Safety and Security 3, 2009.
164
ŠEKULARAC AND JANKOVIĆ
13. V. Betta, F. Cascetta, M. Musto, G. Rotondo, Fluid dinamic performance of traditional and
alternative jet fans in tunnel longitudinal ventilation systems, Tunneling and Underground
Space Technology, 25 (2010), 415ҫ422.
14. L. Barbato, F. Cascetta, M. Musto, G. Rotondo, Fire safety investigation for road tunnel
ventilation systems - a overview, Tunneling and Underground Space Technology, 43 (2014),
253ҫ265.
15. M. K. Camby, E. W. M Lee, A. C. K. Lai, Impact of location of jet fan on airflow structure in
tunnel Fire, Tunneling and Undeground Space Technology 27 (2012), 30ҫ40.
16. M. Šekularac, Experimental determination of tunnel ventilation ducted fan performance,
Thermal Science 53(4) (2014), 1237ҫ1242.
17. P. Vukoslavčeević, J. Wallace, A 12-sensor hot-wire probe to measure the velocity and vorticity
vectors in turbulent flow, Springer Ser. Meas. Sci. Technol. 7(7) (1996), 1451ҫ1461.
18. Ansys, Ansys, Fluent solver theory guide, ANSYS, 2013.
EXPERIMENTAL AND NUMERICAL ANALYSIS OF FLOW FIELD...
165
ЕКСПЕРИМЕНТАЛНА И НУМЕРИЧКА АНАЛИЗА
СТРУJНОГ ПОЉА И ПЕРФОРМАНСИ ВЕНТИЛАЦИJЕ
У САОБРАЋАJНОМ ТУНЕЛУ ВЕНТИЛИСАНОМ
ПУТЕМ АКСИJАЛНИХ ВЕНТИЛАТОРА
Резиме. Ради испитивања струjања ваздуха у вентилисаним саобраћаjним тунелима, умањени модел типског друмског саобраћаjног тунела са одговараjућим системо вентилациjе на бази аксиjалних вентилатора, jе конструисан
и направљен у лабораториjи. Циљ овог рада jе струjно поље у тунелу двосмjерног саобраћаjа, са двиjе коловозне траке. Са размjером од око 1:20, а дужином
20.52 m модел представља око 400 m дуг стварни тунел обjекат. Модел се састоjи од двиjе паралелне тунелске циjеви, гдjе jе главни тунел (хидрауличког
пречника ≈ 400 m има облик попречног пресjека умањеног друмског тунела
обjекта. Други тунел (Dh2 ≈ 0.16 m) jе мањих димензиjа и кружног jе попречног пресjека, а користи се само као модел евакуационог тунела симулираjући
струjање ваздуха из главног саобраћаjног у евакуациони тунел. Дакле два тунела су повезани евакуационим путевима, опремљеним подесивим преградним
вратима. Комбиновањем експементалног и рада на рачунару, струjно поље и
перформансе вентилационог система су истражене. Поље брзине и турбуленциjе на излазу из вентилатора вентилационог система jе дефинисано експериментално користећи термалну анемометриjу. Ови подаци су даље обрађени да
би се употриjебили при нумеричком симулирању струjног поља у тунелу путем
CFD. Ефикасност преноса импулса (ηi , Кемпф фактор) подужне вентилациjе
jе одређена, те ефекат коjи задати гранични услови и ниво њихове детаљности
имаjу, у оквиру CFD симулациjе струjног поља у тунелу, са аспекта тачности
резултата, поља брзине и срачунатог ηi . На краjу, случаj ҡзагушеногә тунела
(саобраћаjем) jе проучен путем експерфимента и CFD. Тешкоће у дефинисању
потребне силе потиска вентилациjе у ситуациjи оптерећеног саобраћаjа су дискутоване, и срачуната jе ефикасност преноса импулсе вентилациjе. На бази
ових резултата, два краjња случаjа облика распођеле аксиjалне компоненте
брзине ваздуха у тунелу у функциjи висине од површине друма, у овоj врсти
тунела и саобраћаjа, jе дефинисана путем CFD. Последњи резултат се може
користити као реалистичан гранични услов (као улазни гранични услов и/или
почетни услов) за нумеричке анализе струjног поља и преноса топлоте у сличним тунелима у условима саобраћаjног загушења, што jе критична ситуациjа
за проjектовање.
Laboratory for Fluid Mechanics and Energy Processes
Faculty of Mechanical Engineering
University of Montenegro
Podgorica, Montenegro
milans@ac.me
Hydraulic Machinery and Energy Systems Department
Faculty of Mechanical Engineering
Univrsity of Belgrade
Belgrade, Serbia
nolebb@gmail.com
(Received 01.12.2017)
(Available online 05.09.2018)