Escolar Documentos
Profissional Documentos
Cultura Documentos
I. INTRODUCTION
HERMAL constraints under intermittent loading conditions are becoming more significant with the expansion
of distributed generation (DG) and demand-side management.
Knowing the true capacity of the grid in real time is essential,
especially if expensive network assets are to be used efficiently and safely. Exceeding the steady-state ratings of power
transformers may be necessary in open electricity markets for
economic reasons or simply to ensure a continuous energy
supply [1]. Besides, short-time peak overloads, without significantly decreasing their life expectancy, are very often requested
from distribution transformers installed in prefabricated substations [2]. The true capacity varies with the power system
dynamics as well as with varying environmental conditions.
Dynamic thermal ratings utilize these factors to optimize and
better manage the grids power transfer capacity in real time.
Manuscript received March 20, 2013; revised June 06, 2013; accepted July
25, 2013. Date of publication August 23, 2013; date of current version March
20, 2014. This work was supported by Aalto energy efficiency program through
the SAGA project. Paper no. TPWRD-00329-2013.
The authors are with the Department of Electrical Engineering, Aalto
University, Aalto FI-00076, Finland (e-mail: merkebu.degefa@aalto.fi;
john.millar@aalto.fi; matti.lehtonen@aalto.fi; petri.hyvonen@aalto.fi).
Color versions of one or more of the figures in this paper are available online
at http://ieeexplore.ieee.org.
Digital Object Identifier 10.1109/TPWRD.2013.2276941
0885-8977 2013 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
Fig. 1.
RC thermal circuit.
source to the heat sink. Many heat-transfer problems can be reduced to an equivalent RC circuit, where the distributed parameters can be lumped to a suitable combination of thermal capacitances and resistances. In some cases, these elements must be
nonlinear, such as the thermal resistance shown in Fig. 1. Generally, most transformer top-oil or hot-spot thermal models go
through the following steps:
Step 1) Identify all heat-transfer components (i.e., convection, conduction, and radiation).
Step 2) Evaluate the thermal resistances and thermal capacitances of each medium between the heat source and
heat sink (Fig. 1).
Step 3) Produce a simplified RC network and relevant firstorder linear differential equation to solve the temperature at the required point.
(1)
, analogous to electric
The thermal resistance
resistance, obstructs the flow of heat
, which itself is
analogous to electric current in electric circuits
(2)
is the heat-transfer coefficient
where
and
is the area
.
The presence of thermal resistances and capacitances
in the network requires iterative numerical methods,
such as the Euler and Runge Kutta methods, to solve differential equations like (1). This approach is used in almost all
of the literature, for example, the IEEE guide for loading
mineral-oil-immersed transformers [3][5], [7], [8].
The heat-transfer coefficient in (2) depends on the physical
parameters of the medium which, in turn, might depend on temperature, hence giving nonlinearity to the thermal resistance.
The coefficient may include convection and radiation when the
medium is air.
For convection [9]
(3)
787
where
is the Nusselt number;
is the Prandtle number;
is the Grashof number; is the characteristic dimension, length,
width, or diameter
; is the gravitational constant
;
is the fluid thermal conductivity
;
is the fluid
density
; and are empirical constants; is the fluid
thermal expansion coefficient
;
is the specific heat
of the fluid
; is the dynamic fluid viscosity
;
is the fluid temperature gradient
; and
is the convection heat-transfer coefficient
.
The radiative heat transfer from a hot surface in air could
be surface_to_ambient and/or surface_to_surface radiation,
where the radiation heat-transfer coefficient
is calculated
as either:
a) Surface_To_Ambient: The ambient surrounding behaves
as a blackbody. This means that the emissivity and absorptivity are equal to 1, and zero reflectivity
(4)
is the surface emissivity and
is the
where
StefanBoltzmanns constant
.
b) Surface_To_Surface (from both ambient and surrounding
surfaces)
(5)
is the mutual irradiation, going to other boundwhere
aries;
is an ambient view factor whose value is equal
to the fraction of the field of view that is not covered by
other boundaries; and
is the assumed far-away temperature in the directions included in
The total heat-transfer coefficient for a heat source body in a
fluid medium
is given by
(6)
Although various combinations of empirical correlations and
modified solution mechanisms could be implemented, the general heat-transfer theories explained before are the basis of the
modeling in this paper.
III. INCLUSION OF CABIN ENVIRONMENT
A typical prefabricated substation (cabin) is comprised of an
MV/LV transformer, LV panel and MV switchboard, connections, and auxiliary equipment in an enclosure to supply the LV
energy from the MV system [2] (see Fig. 2). Due to the thermal
limits imposed by the enclosure, there is only a slight possibility
of exceeding the nominal rating of such transformers. Natural
convection plays the dominant role in the heat transfer, and is
mainly influenced by factors, such as the entry and exit grill
shapes, size, and position. The grills should be designed in such
a way as to reduce the slowing down of air circulation within
the substation or loss of air pressure differential [9]. Ventilation
openings are also arranged to prevent any undesired condensation on electrical equipment and inner wall surfaces.
Short-term peak loads can be handled by distribution transformers inside kiosk substations for economic reasons without
significantly decreasing their life expectancy [1], [2]. Hot-spot
temperatures considerably higher than 98 C can be carried for
788
789
(10)
In Susas model, besides the nameplate-rating information,
results of the heat-run test of the transformer conducted during
commissioning are among the input parameters.
IV. STACK EFFECT
Inside a cabin, after the air is heated at the surface of the transformer by natural convection, the hot air will rise to the roof. The
accumulation of hot air at the top creates positive pressure while
the abandoned floor area of the cabin experiences negative pressure (see Fig. 4). With the developed pressure difference and
the presence of inlet and outlet openings at the bottom and top
areas of the cabin, respectively, a sustained circulation of air will
occur. This phenomena is called the stack effect and continues
to act as long as the inside air is hotter than the ambient environment. This form of natural ventilation is so effective that, for
instance, without the need to introduce forced ventilation, computational fluid dynamics (CFD) techniques applied to partially
ventilated cable troughs have shown an increase in the continuous rating of up to 31% [15].
The air streaming out of the outlet is quantified by the volumetric flow rate
and is given in (11) [6]
(11)
where
is the ventilation rate (m /s);
is a discharge coefficient;
is the free area of inlet opening
(m ), which is equal to the area of the outlet opening;
(m/s ) is acceleration due to gravity; is the vertical distance
between inlet and outlet midpoints (in meters);
is the temperature of indoor air (in Kelvin); and
is the temperature of
outdoor air (in Kelvin).
790
The heat loss due to the ventilating moving air is given [6]
(12)
is the specific heat
where is the heat loss (in kilowatts),
capacity air1.005 (kJ/kg C), and is the density of air1.2
(kg/m ).
Equation (11) consists of a discharge coefficient which directly incorporates the effect of the grill type and a covering
panel just behind the vent openings. The discharge coefficient
is a value describing the aerodynamic channeling of the airflow
on discharge. It represents the ratio between the actual air flows
compared with the theoretical airflow. The coefficient, ranging
from 0 to 0.65, can easily be attained experimentally and is
given in the manufacturers specification of the specific grill or
louver. The thermal resistance associated with ventilation openings will be as shown in (13), which is based on (11) and (12)
(13)
The temperature dependence of the air density and air viscosity could easily be incorporated in (11) to accurately account
for the pressure-driven ventilation and air humidity effects.
V. SCALED MODEL OF THE PREFABRICATED CABIN
Before crafting the overall thermal circuit, we closely studied
a scaled down model of the cabins as well as a 3-D FEM simulation. The model transformer has steel walls and wall-attached
ventilating fins, and was filled with mineral oil. The heating load
was supplied with 14 ceramic-walled hollow resistive rods with
a connected resistance of about 31 (see Fig. 5). Thermocouples were placed at different locations in the cabin, including
on the surfaces of the resistive rods and inside the model transformer top oil. The placement of the thermocouples was guided
by results from a 3-D FEM simulation shown in Fig. 3 and an
analytical heat-loss calculation of the parallel-connected resistive loads.
The first measurement was conducted by loading the resistors
with 150 W and measuring the incoming and outgoing air temperature. The measurements, shown in Fig. 6, reveal that the hot
air leaves the cabin at the very top of the upper windows, indicating that the remaining openings are less relevant. The measurements also show that after a certain limit of opening area,
further increasing the size of the ventilation opening does not
have an impact on the heat transfer. It is primarily the height
of the placement of the outlet opening with respect to the inlet
opening that will affect the top-oil temperature of the distribution transformer. This observation supports the logic behind the
stack effect.
The FEM simulation in Fig. 4 supports the fact that the hot air
requires only a small area at the very top of the outlet opening to
stream out, as explained before. Besides, the simulations show
locally circulating air at the nonventilated walls of the cabin.
The heat transfer from the transformer surface to the nonventilated cabin wall side, therefore, depends on surface-to-surface
radiation and convection at the inside cabin wall surface.
The thermal network shown in Fig. 7 is our proposal. The network considers that the surface-to-surface radiation between the
Fig. 6. Air temperature at different levels of the scaled down cabin model
loaded with a 150-W heat supply. (a) Right-side ventilation openings. (b) Frontside ventilation openings (see Fig. 5).
(14)
where is the characteristic dimension, length, width, or diameter (in meters); is the gravitational constant (m/s ); is the oil
thermal conductivity (W/mK);
is the oil density (kg/m );
is the oil thermal expansion coefficient (1/K);
is the specific
heat of oil (J/kg K); is the oil viscosity
;
and
is the oil temperature gradient.
The empirical values for
and are given in Table III.
The thermal resistances
and
are computed
based on the natural convection Nusselt numbers for vertical
and horizontal planes facing either upwards or downwards and
either inside or outside, as shown in the Appendix. The thermal
resistance
is given in (13). The thermal ladder network
is implemented with an extended Matlab Simulink tool called
791
Fig. 8. Step response of outgoing air temperature through the outlet ventilation
of a prefabricated distribution transformer cabin supplied with 3200 W followed
by 4780 W.
Fig. 7. Overall thermal ladder network for the cabin installation. (a) The
detailed network consisting of the major thermal resistances and capacitance.
(b) The simplified network.
Fig. 9. Top-oil and transformer room air temperature comparisons for a fully
loaded 1000-kVA distribution transformer inside the cabin.
TABLE I
RAYLEIGH AND NUSSELT NUMBERS FOR HORIZONTAL PLATES [9]
TABLE II
RAYLEIGH AND NUSSELT NUMBERS FOR VERTICAL PLATES [9]
792
2) Vertical Plates:
Characteristic length
3) Vertical Wall With Ventilating Fins: The Rayleigh number
is expressed as
TABLE III
EMPIRICAL VALUES FOR CONSTANTS
AND
[3]
ACKNOWLEDGMENT
The authors would like to thank Dejan Susa (Ph.D.) for explaining the transformer thermal model and to N. Tatu for assistance in the experimental setup.
REFERENCES
Characteristic length
793
M. Lehtonen (M11) received the M.S. and Licentiate degrees in electrical engineering from
Aalto University (formerly Helsinki University of
Technology), Espoo, Finland, in 1984 and 1989,
respectively, and the D.Sc. degree from the Tampere
University of Technology, Tampere, Finland, in
1992.
Since 1987, he has been with VTT Energy, Espoo,
and since 1999, he has been with the Department of
Electrical Engineering, Aalto University, where he is
a Professor of IT applications in power systems. His
main activities include earth fault problems, and harmonic-related issues and
applications of information technology in distribution automation and distribution energy management.