Proceedings of the ASME 2013 32nd International Conference on Ocean, Offshore and Arctic Engineering

OMAE2013
June 9-14, 2013, Nantes, France

OMAE2013-10135

NUMERICAL PREDICTION OF THE COOLING OF BURNT CABLE FOR THE

PREVENTION OF RE-IGNITION
Dmitry G. Lanin Boris V. Balakin
All-Russian Institute for Fire Safety University of Bergen
Moscow, Russian Federation Department of Physics and Technology
Bergen, Norway

Olga A. Sabodash Sergey G. Tsarichenko

Far Eastern Federal University All-Russian Institute for Fire Safety
School of Engineering Moscow, Russian Federation
Vladivostok, Russian Federation

ABSTRACT f drag force per unit volume (N/m3)

Fires in the cable structures are a serious danger for the off-
shore facilities. Inflammations of the cables are mostly caused M molar mass (kg/mol)
by cable overheating as a result of power loads in the net or by q heat flux per unit length (W/m)
the shortcut circuits. Fire extinguishing comes, in the first place,
to neutralization of the external flame on its surface. However, q1-2 interphase heat transfer per unit volume (W/m3)
even when the external flame is prevented, the re-ignition of the Rc radius of cable (m)
cable`s insulation may happen in case its metal core has not
t time (s)
been sufficiently cooled down to a safe temperature level. It is
therefore important to clarify the dynamics of the core cooling T temperature (K)
under the influence of the flow of extinguishing agents in order
u velocity (m/s)
to manage the temporal characteristics of the entire process.
The current paper reports a numerical model of a hot cable x,y Cartesian coordinates (m)
cooling under the influence of external flow. The model
Greek letters
resolves the temperature fields as in the cable as in the
surrounding media in connection with the flow profiles of the α heat transfer coefficient (W/m2K)
cooling agent. Several cooling scenarios are considered for the β heat expansion coefficient (1/K)
problem: natural convection of the cooling gas around the
cable, and forced heat utilization due to an induced gas-water λ thermal conductivity (W/mK)
flow directed towards the cable. The model is validated with the µ dynamic viscosity (Pa·s)
use of experimental data presenting a suitable agreement with
empirics. ρ density kg/m3
σ Stefan-Boltzmann constant W·m-2·K-
4
NOMENCLATURE
φ volume faction m3/m3
2
a thermal diffusivity (m /s)
Dimensionless criteria and coefficients
A area of droplets per unit volume (m2/m3)
CD drag coefficient
C specific heat (J/kgK)
Gr Grashof number
d diameter of droplet (m)
Nu Nusselt number
e internal energy per unit mass (J/kg)

1 Copyright © 2013 by ASME

Pr Prandlt number ∂T  ∂ 2T ∂ 2T  λ  ∂ 2T ∂ 2T  (1),
= a 2 + 2 =  2 + 2
Ru universal gas constant ∂t  ∂x ∂y  ρ C  ∂x ∂y 
Ra Rayleigh number where T – temperature; t –time; x,y – coordinates; a – thermal
Rer relative Reynolds number for a droplet diffusivity; λ – thermal conductivity; C - specific heat; ρ –
density of solid material.
Subscripts The conjugate heat transfer imposes on Eq. 1 a convective
a ambient boundary condition; hence the heat flux per the cable’s unit
length is given by:
q = −2π Rα (T [ Rc ] − Ta ) − 2π Rε rσ (T [ Rc ])
f fluid (air) 4 (2),
1,2 phase indeces
where q – heat flux per unit length; Ta – ambient temperature; α
1-2 interphase interaction
– heat transfer coefficient; Rc – outer cable’s radius; σ=5.67x10-
8
W·m-2·K-4 – Stefan-Boltzmann constant; εr – emissivity.
1. INTRODUCTION The process of cable’s cooling at natural convection in the
The formation of fire in the electrical networks is an air converges to the set of Euler equations for compressible
important problem for the offshore facilities. The main reason flow.
of the fire is an inflammation of the cable shell due to Continuity equation:
overheating of the internal conductor which can be caused
∂ρ
( )
(3),
+∇ ρf u = 0
f


either by a power overload or by the shortcut circuits. However
the ignition temperature for a shell is normally less than the ∂t
melting point of a conductor. where ρf – gas density; u – velocity.
From the fire extinguishment point of view it is therefore Momentum equation:
important to understand whether the temperature of the
∂ρ f u


(4),
( )
+ u∇ ρ f u = −∇p + g ρ f




conductor has become below the ignition temperature of the
shell after the fire is prevented. If this condition is not satisfied ∂t
then the shell is re-ingnited, i.e. the secondary flame is formed where p – gas pressure and g – gravitational constant.
[1]. Energy equation:
From this point of view any extinguishment system could (5),
∂   u 2 
  u 2 
ρ f +   + u∇  ρ f + = ρ − ∇


be used for an additional cooling of the hot cable after the fire  e  e  q ` u p
∂t 
f
has been stopped. The CFD-prediction (computational fluid  2    2 
dynamics) of the cooling process would help to make a decision where e=CfTf – internal gas energy per unit mass.
on the duration of post-extinguishment utilization of such The gas density is defined by the ideal gas state equation:
system.
ρ (6),
The current paper illustrates a computational fluid dynamic p= f
RuT f
model developed for the prediction of the dynamics of a hot Mf
cable cooling. The cooling considered here is of two basic where Ru=8.31 – J·mol-1·K-1 – universal gas constant; Mf –
types: the natural convection of air around the cable and the molar mass of gas.
forced heat exchange by the falling flow of water drops in air. The heat transfer coefficient from Eq.(2) is given by [3] for
The model is validated with experimental data and clearly the natural convection of gas around a horizontal cylinder:
demonstrates the difference in between two selected cooling
λ λ (7),
techniques. Additional sensitivity studies were done for the α= f
Nu = 0.25 Ra 0.25
model in terms of the influence of the water droplet size and the 2 Rc Rc
thermal radiation mechanism used there. Moreover, the cooling
times given by the model were predicted for the standard Ra = Gr Pr
electrical cables used at the most industrial sites in Russian g β f Rc 3 (T [ Rc ] − T f )
Federation and could therefore be used for reference. Gr = 8
ν 2
f
2. METHODOLOGY νf
Pr =
2.1 Numerical model af
The temperature distribution in the infinite cable is defined where β is the heat expansion coefficient.
by the heat equation with spatial derivatives with respect to the For the cooling process in the multiphase flow (air – water
cable axis lateral coordinates [2] (two-dimensional spray) Eulerian-Eulerian approach was applied. This concept
representation of process is suggested): includes the phase balance equations in a computational cell:

2 Copyright © 2013 by ASME

 2 (8), 1. a circular cable of solid material. The diameter of the cable
∑ϕ p = 1
p =1
is set according to validation experiment and standards [5]
while the thermal parameters of solid material are
where φp is the volume fraction of the pth component. calculated as to be defined by the mass fractions of the
Continuity equation is written for each phase (phase 1 – air, conductor and the shell assuming the cable to be
phase 2 – water): homogeneous. The temperature field in the cable is given
∂ϕ ρ
( )
(9). by Eq. (1). The only boundary condition is located at the
+ ∇ ϕ1 ρ1 u1 = 0


1 1
air-cable interface. It is defined by the heat flux as in Eq.
∂t (2).
∂ϕ 2 ρ 2
+ ∇ (ϕ ρ u ) = 0 2. cooling (fluid) media. The fluid domain is rectangular with


∂t
2 2 2
the height equal to 6 and the width to 4 cable diameters.
where ρ1 – air density defined by Eq. (6), ρ2 = ρwater = 1000 The flow of the natural convection of air is described by
kg·m-3. Eqs.(3)-(7) while the multiphase air-water flow is given by
The momentum equation for the first phase takes into Eqs.(8)-(13). The heat flux given by Eq.(2) is also applied
account interphase interaction f1-2 as a drag force of the water to the cooling media at the interface. The periphery of the
droplets which impacts on the gas flow: model is the pressure boundary with the fixed temperature
and the atmospheric pressure. When the multiphase flow
∂ϕ1ρ1 u1 
(10).
( )


+ u1∇ ϕ1 ρ1 u1 = −ϕ1∇p + ϕ1 g ρ1 + f1− 2




was simulated, a constant volume fraction of air and water
∂t was prescribed at the upper edge of the model while the
3 ρ
( )
zero-gradient volume fraction approach [6] was used for
f 1− 2 = − f 2−1 = ϕ1 1 CD u1 − u2 u1 − u2







the rest of pressure boundaries.
4 d
CD =
24
Re r
(1 + 0.15 Re0.687
r )
u1 − u2 d



Rer =
νf
Water is a viscous fluid, so shear stress is to be added to the
momentum equation for the second phase :
∂ϕ2 ρ 2 u2
( )


(11),
+ ϕ 2 u2∇ ρ 2 u2 = −ϕ2∇p + ϕ2 µ2∇ 2 u2 +




∂t
+ϕ2 g ρ 2 + f 2−1



where µ2 is the dynamic viscosity of water.

Energy equation is also modified with volume fractions of
the phases and interphase heat transfer q1-2=-q2-1:
∂   u12   
  u12   (12),
ϕ1 ρ1  e1 +   + ϕ1 u1∇  ρ1  e1 +   = ϕ1 ρ1q1 `−ϕ1 u1∇p + q1− 2


∂t   2    2 
∂   u2  
2
  u22  
ϕ2 ρ 2  e2 +   + ϕ 2 u2∇  ρ 2  e2 +   = ϕ 2 ρ 2 q2 `−ϕ2 u2∇p + q2−1




∂t   2    2 
q2−1 = A1−2α1−2 (T1 − T2 )
where A1-2 – surface of the interphase interaction per unit
volume, m2·m-3 , α1-2 – interphase heat transfer coefficient,
given according to Ranz and Marshall [4] as:
α = 2 + 0.6 Re0.5 Pr 0.3
1− 2
(13).
r Figure 1. Computational domain and boundary conditions
The initial condition for Eq. (1) was defined by either the The model was further discretized with 0.5-mm regular
experimental initial temperature or it was around the melting computational cells and Eqs.(1)-(13) were solved with the use
temperature of a conductor from the standard cable [5] (copper of built-in STAR-CCM+ v.7.06 [6] solver based on the
in most cases, T=1084°C). The initial condition for Eqs. (3)- SIMPLE technique [7] with a time step of 10 ms.
(12) was: zero velocity and water phase volume fraction fields, The mesh-independence check was performed for the
normal conditions. model refined twice relatively to the referenced value. The
maximum difference in between the results produced on 0.5-
2.2 Process geometry mm and 0.25-mm grids was not more than 4 %. The coarse grid
The 2D model (presented in Fig.1) of the process is was then selected in order to minimize the computational costs.
subdivided by two domains:

3 Copyright © 2013 by ASME

 Figure 3. Model sensitivity to thermal radiation for 2.5-
mm copper conductor
The results of model validation have demonstrated the
2.3 Experiment
importance of the thermal radiation. The simpliest model of the
The experimental part of the current study was aimed to
thin (2.5-mm) uninsulated conductor was selected for the
record the dynamics of cooling of a cable due to the natural air
current sensitivity test in order to minimize the computational
convection at normal conditions. A 13.0-mm cable with the
costs required for the simulation of the case with pure
conducting core made of stainless steel (12N18H10T according
convection.
to Russian sortament) and 2-mm thick PVC-shell was heated up
Fig.3 illustrates the sensitivity of the model to the existence
to 450° C by an electrical current passing through the core.
of radiation, i.e. the thermal histories presented in figure are for
After the conductor was warmed up to the desired temperature,
the cases when: the second member on the right-hand-side of
the electrical current was stopped there and the cooling of the
Eq.(2) is neglected and when it is included into the model. The
test cable was controlled by a thermocouple.
results in Fig.3 are for the natural convection of air around a
copper 2.5-mm naked conductor thus the maximum temperature
3. RESULTS was 1084 °C according to the methodological section. It
follows from the figure that the temperature reduction from
3.1 Model validation and sensitivity test
1084°C to 200°C is almost twice longer in case the radiation is
Fig.2 illustrates the process of 13-mm insulated cable
not accounted for.
cooling due to the natural convection in air predicted by model
and recorded experimentally. The temperature of the cable was
3.2 Cooling by natural convection
reduced from 450 to 150 °C during the experiment so the initial
Fig.4 demonstrates vectors of the air velocity around 2.5-
condition for model was set to the same maximum value.
mm copper conductor at the early development of the cooling
It follows from the plot that the initial cooling rate of the
process (10 s). It is seen from the figure that the upward
cable was higher than the terminal phase of the process. This is
convectional current has been established just above the cable.
explained by more intensive thermal radiation at higher
The mean convection velocity is about 0.05 m/s. The air
temperatures. The model demonstrates slight underprediction of
velocity in the vicinity of the cable is lower than at the
the temperature history at the initial stage of the process so the
periphery of the model due to the formation of a boundary layer
amount of heat irradiated by the model was lower due to the use
on the cable surface. Two slow vortices are formed in the top
of the thermal parameters homogenization mentioned in section
corners of the model as the swirling takes place by the sides of
2.2. The rest of the process dynamics was in the suitable
the conective stream. Slight asymmetry of the flow profile is
agreement with the experiment quantified in terms of
observed at the bottom of the model due to the low-magnitude
discrepancy as to be not more than 11%.
convectional instability formed there.

Figure 2. Cable average temperature as a function of time

for the cooling of 13-mm cable due to the natural
convection of air. Experimental results compared with
numerical data

Figure 4. Air velocity vectors around 2.5-mm copper

conductor. 10 s of the process
Figs.5A-B illustrate the temperature profiles of the model
for 10 and 100 s of the convection around 2.5-mm copper
conductor.
The only driving force for the flow development is the
convection of air due to the hot cable. At the beginning of the
process the boundary layer in the vicinity of the cable becomes

4 Copyright © 2013 by ASME

warmed by the same temperature as the cable. The convective cooling of cable due to injection of 50 µm water droplets done
upward stream mentioned above forms an upward temperature with the flow rate of 0.04 kg/s·m2.
pattern which is see in Fig.5A.
A

Figure 6. Flow velocity vectors around 150-mm copper

conductors. ASBL 3x150 cable. 100 s of the process. 50
B
µm water droplets. Water injection rate 0.04 kg/s·m2
It follows from the figure that there are two main flow
patterns formed in the system: the downward current of the
cooling media which is formed under the influence of water
drops injection is competing with the upward current of the
rising air heated in the vicinity of cable. Two vortex structures
are therefore formed by the sides of the cable.
Figs.7A-B demonstrate contours of temperature in the
system at 100 s (Fig.7A) and 300 s (Fig.7B) of the colling of
ASBL 3x150 with copper conductors. As compared with
natural convection presented in Figs.5A-B, there is no an
upward temperature stream governed by the rising air. Two
wing-like temperature patterns (Fig.7A) are located by the sides
of the cable. They are associated with the vortices seen in Fig.6.
It follows from Fig.7B that these warm regions decrease in line
with the cooling of the heat source (cable).
Figure 5. Temperature contours for 10 s (A) and 100 s (B) of A
the process. Cooling of 2.5-mm copper conductor in air
However the cable releases its heat due to the radiation and
convective heat exchange with air. Fig.5C clearly demonstrates
the stagnation of the convective stream after 100 s of the
process. It follows from the figure that the intensity of the
convective heat transfer is reduced and the relatively warm zone
mostly occupies the boundary layer near the warm wall.

3.3 Cooling in multiphase flow

The model validated in the numerical experiments with the
natural convection was further used for the prediction of
cooling of the standard 150-mm cable ASBL 3x150 with 3
copper conductors covered by a multiplayer protective shell.
The cable in this case was immersed into the flow of water
droplets that were falling from the top of the model with the
prescribed mass flow rate (per unit area).
Fig.6 reports the developement of the flow velocity profile
for the case of forced convection in terms of the air-water
mixture velocity vectors. The results are presented for the

5 Copyright © 2013 by ASME

 B 4. CONCLUSION
The numerical model for the prediction of the hot cable
cooling after the fire extinguishment has been developed and
validated with experimental data for the natural convection of
air around the cable. The model was further extended for the
prediction of cooling process in the multiphase air-water
mixture.
The simulation results have demonstrated that the model is
sensitive in terms of the thermal radiation modelling, i.e. the
colling rate could be significantly (more then 100%)
overpredicted when the the radiation is not accounted for.
The difference between natural convection and the forced
convection around the cable has been demonstrated in terms of
the flow velocity and temperature profiles. The model was
capable to illustrate the enhancement of the cable-aerosol heat
exchange with the increase of the size of water droplet.
Figure 7. Temperature contours for 100 s (A) and 300 s (B)
of the process. Cooling of ASBL 3x150 cable with copper
conductors in air-water mixture. 50 µm water droplets. ACKNOWLEDGMENTS
Water injection rate 0.04 kg/s·m2 Far Eastern Federal University is acknowledged for the
Additional study was done for the determination of the financial support.
influence of the water droplet size on the cooling rate. It was
studied in terms of the time which is necessary for the ASBL REFERENCES
3x150 cable with copper conductors to decrease its temperature [1] L.I. Belousov, V.A. Bylinkin, D.G. Lanin, A.V. Pervykh,
from 1084°C to a safe temperature which ensured that the re- D.V. Polyakov, S.G. Tsarichenko. Fire extinguishment systems
ingnition was not possible (150 °C). The droplet size was in cable structures. Proceedings of the 19th conference on Fire
varried in the interval from 50 to 250 µm while the amount of Safety of Multifunctional Constructions, Moscow, p.26-27,
droplets injected into the system was kep constant. The 2005. In Russian.
simulation results are presented in Fig.8 where it is seen that the [2] J. D. Anderson Jr. Computational fluid dynamics, the basics
increase of the droplet size causes significant reduction of the with applications, McGraw-Hill Inc., 1995
cooling time due to the enhanced heat exchange of the cable [3] P.L. Kirillov, Y.S. Yuriev, V.P. Bobkov. Handbook on
with mixture. hydrothermal calculations (nuclear reactors, heat exchangers,
steam generators). Moscow, Energoatomizdat, 1990. In
Russian.
540 [4] W.E. Ranz and W.R. Marshall, Evaporation from drops.
time [s]

Parts I and II, Chemical Engineering Progress, 48(3), p. 141.,

490 1952.
[5] GOST 24183-80. Power cables. General technical
conditions. State standard. In Russian.
440 [6] CD-Adapco, 200 Shepherds Bush Rd. London.
Methodology, STAR-CD Version 4.02, 2006
50 150 250
[7] S. V. Partankar, Numerical heat transfer and fluid flow.
drop size [µm] ISBN 9780891165224.

Figure 8. Cooling time as a function of droplet size. ASBL

3x150 cable with copper conductors

4.

6 Copyright © 2013 by ASME