Escolar Documentos
Profissional Documentos
Cultura Documentos
=
[W m-2] (3)
Where; q is heat flux, r is the radius at which heat-flux is
evaluated and R
1
and R
2
are inner and outer radius of the
dielectric respectively, bounded by the conductor and the
earthed sheath.
D. Modeling of Forced Cooling
The 2-D models include solution of the heat-transfer at the
pipe boundary due to forced-convection water coolant. This
problem is complex in nature due to the dependence of the
heat transfer coefficient on viscosity, specific heat capacity
and their dependence on coolant temperature. For the 2-D
FEA model a solution has been implemented by dividing the
cable route into sections, the heat transfer for the FEA model
2-D section is then solved and a heat transfer coefficient at the
pipe inner diameter is then calculated and used for the
subsequent section calculation.
Calculating for the average coolant temperature along the
route, here estimated to be 20C since the maximum allowable
temperature of the pipe outlet is specified to be 30C, and inlet
temperature is 10C; the Reynolds number, Re, is 92839; and
the Prandtl number, Pr, 7.0. Thus the water in the pipes is a
developed turbulent flow. In the 2-D FEA model the heat
transfer coefficient is calculated using Equation 4.
D
k Nu
T T
q
h
b s
.
) (
=
=
[W.m-2.K-1] (4)
Where; h is heat transfer coefficient, T
s
is the surface
temperature of the pipe, T
b
is the bulk temperature of the
coolant, k is the thermal conductivity of the coolant (W.m
-1
)
and D the characteristic dimension (m), in this case the tube
inner diameter. Nu is the dimensionless Nusselt number.
Assuming a smooth tube wall; the Nusselt number is
491
calculated by the Gnielinski relation [5], Equation 5, using the
Blasius resistance formula [6] to calculate the friction factor, f,
Equation 6. Both these equations are suitable for use with this
range of Re and Pr numbers [7].
{ }
{ } ) 1 (Pr ) 8 ( 7 . 12 1
Pr 1000 Re ) 8 (
3 2 2 1
+
=
f
f
Nu
D
D
(5)
(6)
25 . 0
Re 3164 . 0
=
D
f
Where; Nu
D
is the Nusselt number for the characteristic
length for the case of flow in a closed conduit, such as in this
example, the pipe inner diameter, D. Furthermore assuming
incompressible flow and considering the case for a length of
pipe with an applied heat flux, q, the increase in bulk
temperature can be found using Equation 7.
p
in b out b
C m
DL q
T T
+ =
, ,
(7)
Where; L is the pipe length and is the mass transfer rate
and C
m
p
the specific heat capacity of the coolant. Thus with a
heat transfer coefficient at the pipe wall calculated using
Equation 4, a 2-D FEA model can be solved to provide the
heat flux at the pipe wall and Equation 7 used to calculate a
new bulk temperature for the subsequent route section.
Due to the interaction between the inner and outer pipe and
cable heats multiple iterations along the total route length are
required to achieve convergence of the heat fluxes, coolant
temperatures and cable temperatures along the route. The
convergence criteria used in this study was less than 0.1C for
all cable temperatures along the total route length in
subsequent iterations; in practice this always provided
convergence better than 0.2C for the pipe temperatures and
convergence was achieved within 3 iterations. With a solution
found the load can be adjusted until the maximum operating
temperature limit of the cable is found.
III. RESULTS
Table II presents results obtained using both methods, the
FEA model has been used to investigate the influence of four
different ground surface conditions. Firstly an isothermal case
at 10C, and also convective heat transfer for still air, 1 m.s
-1
and 10 m.s
-1
wind speeds. To model this the heat transfer
coefficient has been approximated using an experimental
relationship, Equation 8 [8].
(8) w h 6 . 4 0 . 6 + =
Where; w is wind speed.
If the current loads for all cables, and water flow in coolant
pipes is equal for all pipes, the symmetry about the centerline
means that the FEA model only need solve for one half the
geometry using a no heat flux boundary along this line. Fig 2
and Fig 3 show the convergence of solutions for a 2-D FEA
model, the case for an isothermal top surface of 10C. It can
be seen that it is not trivial to know where the hottest point
along the cable will be, indeed whilst the cables nearest the
centerline are hottest nearest the inlet, the center cables of the
two groups are hottest at the turnaround point. As may be
anticipated the outermost cables are coolest.
1 2 3 4 5 6 7 8 9 10
357
357.5
358
358.5
359
Cable 1
section number
T
e
m
p
e
r
a
t
u
r
e
( K
)
1 2 3 4 5 6 7 8 9 10
354
356
358
360
Cable 2
section number
T
e
m
p
e
r
a
t
u
r
e
( K
)
1 2 3 4 5 6 7 8 9 10
352.5
353
353.5
354
Cable 3
section number
T
e
m
p
e
r
a
t
u
r
e
( K
)
1st run
2nd run
3rd run
Fig. 2. Cable temperature profile for 300m section length solution with
isothermal ground surface, 2-D FEA method (right hand group of cables,
numbered left to right from centerline in Fig 1.).
0 2 4 6 8 10
292
294
296
298
300
Pipe 1 - return
section number
T
e
m
p
e
r
a
t
u
r
e
( K
)
0 2 4 6 8 10
291
292
293
294
295
296
297
298
section number
T
e
m
p
e
r
a
t
u
r
e
( K
)
Pipe 4 - return
1st run
2nd run
3rd run
Fig. 3. Pipe temperature profile for the return pipes 1 and 4 (right hand group
of pipes, numbered left to right from centerline in Fig 1.).
492
TABLE II
CABLE RATING RESULTS
Model
Surface
boundary
Section
length (m)
Current
rating (A)
Max. coolant
temperature
(C)
isothermal
regions
matrix
10C
isotherm
300 2084 22.39
FEA
10C
isotherm
a
300 1647 25.25
FEA
10C
isotherm
100 1650 25.08
FEA
10C
isotherm
50 1650 25.03
FEA
10C
isotherm
10 1650 24.99
FEA
Still air
convection
100 1638 25.49
FEA
wind speed
1 ms
-1
100 1642 25.34
FEA
wind speed
10 ms
-1
100 1648 25.14
a
No change in the rating or pipe exit temperature was found for a
constant Nu matching that used by the isothermal region method.
A. Influence of Section Length
The FEA model makes the simplifying assumption of a
constant heat flux over each section length and thus results
will be influenced by this parameter, which must be chosen
correctly. The results of Table II demonstrate that for the case
of the isothermal ground surface the resultant current rating
does not significantly alter, less than 1 Amp, for a refinement
of less than 100 m section length. Thus 100 m results have
been presented for discussion.
The isothermal region matrix model does not make such an
assumption, but fits a curve of heat flux to coolant through all
sections in the longitudinal direction. There is therefore
negligible influence below 300 m.
B. Rating Results
Results clearly show that the isothermal ground surface
produces the most optimistic rating. This is due to an
effective additional cooling term at the surface being held at a
specified temperature. In practice with air above the surface
and heat transfer by convection the surface temperature may
be raised considerably considering the relatively shallow
burial depth of the cable circuit. The FEA model shows
however that this effect is less pronounced for the case with
forced cooling pipes positioned between the surface and the
cables. The simulation results show a reduction of only 0.7%
between isothermal and still air boundary conditions. As the
wind speed and heat transfer coefficient at the boundary
increases the resultant rating approaches the isothermal case.
C. Comparison of Methods
Of significant interest is the discrepancy of the results from
the two different rating methods. This must be due to greater
simplifying assumptions used by the isothermal regions
method. Solving the FEA model for the constant Nusselt
number used by the isothermal region matrix method showed
no change in the rating. The construction of the isothermal
region matrix and use of superposition is therefore suspected
to be at fault, the precise source of discrepancy will be sought
in future investigations.
IV. CONCLUSIONS
As was found in a previous model for directly buried cables
[2] the assumption of an isothermal ground surface boundary
produces an optimistic cable rating. This effect is less
pronounced in the case for forced cooled circuits being offset
by the dominance of the cooling action of the water pipes on
the thermal environment and their position between the cables
and the ground surface.
The results show that the isothermal regions method using
superposition and method of images produces a result that is
considerably different from the 2-D FEA model. It was found
that the discrepancy was not due to the assumption of a fixed
Nusselt number and heat transfer coefficient; an assumption
not used in the FEA model, which varies values with bulk
water temperature. Future investigations will seek to clarify
the cause of the discrepancy in results more precisely.
ACKNOWLEDGMENT
The financial support of National Grid Transco for this work
is gratefully acknowledged.
REFERENCES
[1] Electrotechnical-commitee, "Electric Cables - Calculation of the
current rating IEC 60287 -1 -1:1994, BS 7769-1.1:1997," British
Standards Board.
[2] D. J. Swaffield, P. L. Lewin, and M. LeBlanc, "Investigation into
the conservatism of high voltage cable rating methods: A
comparison between IEC60287 and Finite Element Analysis,"
presented at International Symposium on High Voltage
Engineering 2005, Beijing, 2005.
[3] Cigre_committee-21, "Computer method for the calculation of the
response of single core cables to a step function thermal transient,"
Electra, vol. 87, pp. 41-64, 1983.
[4] Cigre_committee-21-08, "The Calculation of Continuous Rating
for Forced Cooled Cables," Electra, vol. 66, pp. 59-84, 1979.
[5] V. Gnielinski, "New equations for heat and mass transfer in
turbulent pipe and channel flow," Int. Chem. Eng., vol. 16, pp.
359-68, 1976.
[6] H. Blasius, "Das Ahnlichkeitsgesetz bei Reibungvorgangen in
Flussigkeiten.," Forsch. Arb. IngWes, vol. 131, 1913.
[7] C. A. Long, Essential Heat Transfer: Longman, 1999.
[8] L. A. Ramdas, "Some new instruments and experimental
techniques developed in the Agricultural Metreology Section at
Poona," J. Sci. Ind. Res., vol. 7, pp. 16-29, 1948.
493