See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/260587623

Prediction of the oil flow distribution in oil-immersed transformer windings

by network modelling and computational fluid dynamics

Article in IET Electric Power Applications · February 2012

DOI: 10.1049/iet-epa.2011.0122

CITATIONS READS

36 805

4 authors, including:

Stefan Tenbohlen
Universität Stuttgart
377 PUBLICATIONS 3,540 CITATIONS

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Online Monitoring of Power Transformers View project

Electromagnetic Compatibility View project

All content following this page was uploaded by Stefan Tenbohlen on 12 February 2015.

The user has requested enhancement of the downloaded file.

www.ietdl.org
Published in IET Electric Power Applications
Received on 30th April 2011
Revised on 18th August 2011
doi: 10.1049/iet-epa.2011.0122

ISSN 1751-8660

Prediction of the oil flow distribution in oil-immersed

transformer windings by network modelling and
computational fluid dynamics
A. Weinläder1 W. Wu2 S. Tenbohlen1 Z. Wang2
1
Institute for Power Transmission and High Voltage Technology, University of Stuttgart, Stuttgart, Germany
2
School of Electrical and Electronic Engineering, University of Manchester, Manchester M13 9PL, UK
E-mail: zhongdong.wang@manchester.ac.uk

Abstract: In the context of thermal performance and thermal design, it is of significance to predict the magnitude and the
location of the ‘hot-spot’ temperature inside a power transformer. In an attempt to accurately predict this hot-spot in an
oil-immersed transformer, various numerical modelling approaches have been developed for calculating the cooling oil
flow distribution, which are generally categorised as either ‘network models’ or the methods that incorporate forms of
computational fluid dynamics (CFD). In network modelling, the complex pattern of oil ducts and passes in a winding is
approximated with a matrix of simple hydraulic channels, where analytical expressions are then applied to describe oil
flow and heat transfer phenomena. On the other hand, CFD models often adopt discretisations of much higher fineness,
which can be expected to offer a higher order of accuracy but also comes with a large increase in the required
computational resources. In order to compare both modelling approaches, the network model implementation TEFLOW
and a commercial CFD package, ANSYS-CFX, were applied on a typical ‘zigzag’ oil channel arrangement of a disc-type
winding to predict oil flow distribution and disc temperatures; experiments on hydraulic models have also been
performed to validate the models. The principle work of this study is then comparing the results and concluding
recommendations to industrial practices.

Nomenclature Subscripts

D equivalent hydraulic diameter of fluid duct cond value of heat conduction

dp thickness of insulating paper conv value of heat convection
f average dimensionless friction coefficient of fluid b value at bulk temperature
duct c value at winding conductor
k thermal conductivity of transformer oil w value at wall temperature
L length of fluid duct 1 value at the duct of the straight-through direction
Nu Nusselt number of a junction
Pr Prandtl number 2 value at the duct of the branch direction of a
q̇ heat flux from winding to fluid duct junction
Re Reynolds number m value at the common duct of a junction
tb bulk temperature of fluid duct 1m value for combining junction, from duct 1 to m
tc temperature of winding conductor 2m value for combining junction, from duct 2 to m
tw wall temperature of fluid duct m1 value for dividing junction, from duct m to 1
T absolute temperature m2 value for dividing junction, from duct m to 2
U average flow velocity of fluid duct Acronyms/shorthand
DP pressure drop between the inlet and outlet
of the duct CFD computational fluid dynamics
m dynamic viscosity of fluid HTC heat transfer coefficient
r density of fluid JPL junction pressure loss

82 IET Electr. Power Appl., 2012, Vol. 6, Iss. 2, pp. 82– 90

& The Institution of Engineering and Technology 2012 doi: 10.1049/iet-epa.2011.0122
 www.ietdl.org
LV low voltage critical temperatures such as hot-spot are required and
NM network modelling detailed local flow/temperature information is not.
The objective of this paper is using a same winding pass
design as an example to compare the two different numerical
1 Introduction approaches with the experimental results; a pass is defined as
the section of a winding between two adjacent oil-block
Power transformers are key and expensive components in washers. The differences between the results from them
electric system networks. To avoid failure and ensure would provide recommendations for those who are choosing
continual power supply, the thermal management of power thermal analysis tools for oil-immersed transformers.
transformers is critical in controlling their insulation ageing
due to high hot-spot temperatures, finally causing electrical
failure. Thereby, accurate thermal assessment is of 2 Dimensions of the investigated geometry,
significance for both design procedure in manufacturers and material properties and excitation conditions
asset management policy-making in utilities [1]. In
particular, large power transformers are generally cooled by The pass example is from a disc-type winding and the studied
natural or forced oil flow, and as such for these oil- section is between two neighbouring spacers, as shown in
immersed transformers, improved understanding of the oil Fig. 1a, followed by three sub-pictures depicting the front,
flow distribution across the oil ducts inside transformer side and top views of the model structure, respectively.
windings is meaningful to avoid localised oil starvation and There are eight discs in the pass; they are cooled with oil
hot-spot temperatures. Commonly, the cooling oil flows up which flows in from the bottom inlet, through horizontal
from the bottom to the top of a winding; however, the hot- channels between the rows of ‘heat-generating’ discs, and
spot is not always found on the topmost winding disc, due joins up with a single vertical channel at the opposite side
to, in part, the effect of a non-uniform flow distribution [2]. that carries the oil upwards and through a gap to next pass.
Numerical modelling has been used to predict the oil flow The next pass starts from the oil-block washer equipped just
and hot-spot for over 40 years [3]. Two categories of below the ninth disc; all washers are assumed as fully tight.
numerical approaches were developed, which are ‘network A winding can then be periodically composed by a series of
modelling’ such as TEFLOW (developed in the UK in the this type of passes, resulting in a zigzag-like oil flow, and
late 1980s) [2, 4 – 7], and the methods that incorporate due to the periodicity, only a single pass is investigated in
computational fluid dynamics (CFD) [1, 8 – 11]. Generally, this paper. In an ideal cooling design, first sufficient oil
with much higher fineness of the discretisation, CFD should be supplied into the pass by buoyancy or oil pumps,
simulations can provide more detailed results but and secondly the oil can be distributed uniformly across the
meanwhile the required computational resources largely horizontal ducts for avoiding any localised oil starvation.
increase. In comparison, network modelling provides fast The geometric parameters, the physical properties of the
solutions which are often more convenient for industry to material and the investigated parameter ranges are
use, and a large range of parameters can be tested with this summarised in Tables 1 – 3, respectively. The dimensions
tool for relatively low computational effort, when only were used to construct the geometry of the numerical

Fig. 1 Structure of the studied disc-type winding pass

a Studied section in winding
b Front view
c Side view
d Top view

IET Electr. Power Appl., 2012, Vol. 6, Iss. 2, pp. 82 –90 83

doi: 10.1049/iet-epa.2011.0122 & The Institution of Engineering and Technology 2012
www.ietdl.org
Table 1 Geometric parameters of the winding pass example

Parameter name Value

width of vertical channels, mm 5.0

width of horizontal channels, mm 4.22
height of discs, mm 9.4
clear distance between the spacers 129.7
(circumferentially uncoiled), mm
radial length of the discs, mm 90.1
bevel corner radius the disc, mm 0.92
thickness of insulating paper, mm 0.5

Table 2 Physical properties of oil and solid materials

Parameter name Value

Fig. 2 Scheme of equipments for pressure acquisition
(a) Oil properties of Shell Diala DX at absolute temperature T, K
dynamic viscosity, mPa s 0.0757 × exp[605.8/(T – 178.3)]
Table 4 Properties of the hydrocarbon used for hydraulic
density, kg/m3 874 – 0.65 × (T – 293.15)
measurements at ambient temperature
heat conductivity, W/(K m) 0.124 – 6.25 × 1025 × (T – 293.15)
heat capacity, J/(K kg) 2020 + 4375 × (T – 293.15) Parameter name Value
(b) Oil properties of Shell Diala DX at absolute temperature 608C
dynamic viscosity, mPa s 3.8 dynamic viscosity, mPa s 2.4
density, kg/m3 848 density, kg/m3 803
heat conductivity, W/(K m) 2195
heat capacity, J/(K kg) 0.1215
(c) Properties of the solid materials usually small and it is possible to keep its dimensions
thermal conductivity of 410 according to a real transformer; therefore there was no need
conductors, W/(K m) to apply laws of similarity to the measured data afterwards.
thermal conductivity of 0.15 Since for the first step only hydraulic data were of interest,
insulating paper, W/(K m) the discs were made of transformer board according to the
thermal conductivity of 0.15 outer form of real discs.
spacers, W/(K m) The scheme of equipments for acquiring the pressure drop
is outlined in Fig. 2. From the picture, it can be seen that each
pressure transducer is switched between two channels. The
Table 3 Investigation parameter ranges
measured data are constantly recorded by a computer until a
steady state is reached. Once finished, a new flow rate can
Parameter name Value be imposed and studied. To have realistic properties of the
fluid according to them of transformer oil at a typical
flow rate at the pass inlet, l/min 2, 5, 10, 15, 20
operating temperature, a special hydrocarbon was used
oil temperature at the pass inlet (8C) 60
instead of regular transformer oil. This hydrocarbon is
(only for the hydraulic –thermal models)
similar to kerosene but has a higher flame point, an eligible
loss power per disc (W) (only for the 15, 30
viscosity and density at room temperature, listed in Table 4.
hydraulic– thermal models)
Owing to this in the measurements there is no heating
required and the oil temperature remains constant.

models, and the inlet oil flow rate and temperature, in Table 3,
were applied as boundary conditions. 4 Network modelling
In brief, network modelling reduces the complex pattern of
3 Experimental measurements the oil flow inside a transformer winding down to a matrix
of simple hydraulic channel approximations, interconnected
Hydraulic measurements were especially done to validate the by junction points or ‘nodes’ [12]. Fig. 3 shows the
simulations. The procedure of the hydraulic measurements geometry approximated from the experimental setup in
was taking a physical model of the winding pass example, Fig. 1 for 2D network model; lumped elements such as
illustrated in Fig. 1, and inputting the specified oil flow rate discs, ducts and nodes are indicated. Bulk averaged
from the pass inlet. The supplied flow rate causes a pressure parameters are assumed to represent the variation in
drop along the flow path, and this pressure drop can be physical quantities at each duct and node, based on which a
measured at some locations that are reachable without set of ‘lumped parameter’ equations are applied to construct
disturbing the flow considerably. Only the pressure drop both the so-called thermal and hydraulic networks across
was measured because it is often difficult to measure the the entire winding.
in-duct pressure profiles accurately; on the other hand, Additionally, the following physical assumptions are made
numerical simulation can yield detailed results in the entire in network modelling: oil is modelled as laminar flow
domain with much less practical effort [8]. between a pair of infinite parallel flat plates [12, 13]; oil
The model that was applied in this experiment represents a temperature is assumed to rise linearly along horizontal
section of a real transformer winding. Such a section is channels due to the uniform heat flux at disc surfaces [14];

84 IET Electr. Power Appl., 2012, Vol. 6, Iss. 2, pp. 82– 90

& The Institution of Engineering and Technology 2012 doi: 10.1049/iet-epa.2011.0122
 www.ietdl.org
    2 
U U 1000
K1m = 0.580 − 1.729 + 1.419 1 ×
Um Um Re1
72
K2m =
Re2
    2 
U1 U 1000
Km1 = 1.079 − 3.337 + 2.735 1 ×
Um Um Re1
276
Km2 =
Re2
(6)

5 CFD modelling and visual localised results

The CFD computation was done with commercial CFD
Fig. 3 Geometry for 2D network modelling of the experimental software ANSYS-CFX, which is a finite-volume-based
setup CFD-solver, while the mesh generation was done with
ICEM-CFD. Two-dimensional modelling is sufficient [12],
oil mixing at nodes is complete hydraulically and thermally. A but CFX does not have the explicit capability to treat 2D
group of mathematic expressions, that is (a–e) as follows, are problems (due to the underlying finite-volume algorithm),
then employed to constitute the hydraulic and thermal and the approach was to model the geometry in an ordinary
networks, respectively. Owing to the temperature dependence 2D way and then extruding only one cell into the
of the physical properties of oil, such as viscosity and circumferential direction for constructing 3D elements. For
density, the hydraulic and thermal networks are coupled and the simulation, this model was assumed as infinitely
as such an iterative approach is required for a solution. extruded along this direction. It implies that the small wall
effect of the spacers, which bounds the horizontal ducts in
1. Mass conservation at nodes; the circumferential direction, was neglected. On the other
2. Pressure drop equation, namely Darcy– Weisbach hand, the radial boundaries, wherein reality pressboard
equation, (1), applied onto ducts [12] cylinders are bounding the winding, were applied as
isothermal wall. Finally for the hydraulic –thermal
4fL rU 2 modelling, constant loss density was impressed into the
DP = (1) conductor volumes.
D 2 At the inlet, oil flow with homogenous velocity and
3. Thermal energy conservation at nodes; temperature was impressed, respectively, while the outlet
4. Conductive heat transfer equation, (2), to express the heat was closed by a zero static-pressure condition, as shown in
conduction across insulating paper Fig. 4. There are three passes involved in this model and
only the middle pass was intended to deliver the results; the
upper and lower passes were facilitated to deliver proper
k
q̇cond = (t − tw ) (2) boundary conditions for the middle pass. When modelling
dp c the material, Newtonian fluid model was applied, where
5. Convective heat transfer equation, (3), along horizontal viscosity depends on temperature in an exponential manner;
ducts to express the heat convection from the duct walls to density, specific heat capacity and heat conductivity are all
the flow bulk assumed to be temperature dependent in a linear manner.
No turbulence model was employed because of low
Nu × k Reynolds number.
q̇conv = (tw − tb ) (3) The discretisation was done with around 900 000 elements,
D
shown in Fig. 4. For the shape functions the second-order
upwind scheme was used and the single-precision solver
Moreover, empirical equations are incorporated for estimating
was tested to be sufficient for this problem. The
the Nusselt number, Nu, friction coefficient, f, at oil ducts and
convergence criterion was set as a RMS-residual of 1025,
junction pressure losses (JPL). These equations were
and a global balance of each conservation quantity of 1023
previously from general fluid dynamics and heat transfer
has been required and reached.
handbooks, but have recently been calibrated by large sets
In Fig. 5 the streamlines resulting from hydraulic-only
of CFD simulations for a wide range of transformer
models were plotted for both flow rates of 2 and 20 l/min.
designs, with overall minimised deviation from CFD
It gets obvious that separation eddies are blocking a portion
predictions [15, 16]. The calibrated (4) – (6) were applied in
of the entrance for horizontal channels, especially at the
this paper.
lower region, because the oil washer equipped at the top
    prevents oil flowing up, effectively forcing them to turn to
L/D −0.38 mw −0.16 the horizontal channels, and thus the eddies at these
Nu = 1.29 + 3.08 (4)
Re × Pr mb channels are largely suppressed. The entrance separation
  eddies actually account for the junction pressure losses
 −0.55  0.90 described by (6) in network modelling. By the comparison
24 0.15 L mw
f = 0.17 Re 0.37
Pr + 0.61 (5) between Figs. 5a and b, the eddies are strengthened with
Re D mb
high inlet flow rates, and because of this the flow

IET Electr. Power Appl., 2012, Vol. 6, Iss. 2, pp. 82 –90 85

doi: 10.1049/iet-epa.2011.0122 & The Institution of Engineering and Technology 2012
www.ietdl.org

Fig. 4 Principal model and details with mesh

Fig. 5 Streamlines for two oil flow rates

a 2 l/min
b 20 l/min

distribution becomes more unequal; for example, in Fig. 5b of thermal boundaries of the middle ducts and the cooling for
20 l/min, the three upmost channels obtain almost the whole the bottom and top ducts is better. In particular, due to high
amount of flow rate which has been supplied into the pass. Prandtl number (typically 200 for oil [2]), it can be seen
Owing to this reason the five lower channels get small that a strong cold streak from the former pass is entering the
proportion of oil and they would get relatively higher oil bottom duct and moving from there lasts until the outer
temperature if constant heating power were imposed. vertical duct; however, the cold streak does not reduce the
With a hydro-thermal model, Fig. 6 displays the temperature thermal boundary temperature, because it flows almost along
distribution within the fluid domain for the case with loss the centreline rather than contacting a channel wall. On the
power of 15 W per disc and flow rate of 2 l/min and the case other hand, at the outer side vertical duct, there are hot-
with 30 W per disc and 20 l/min. For the case with 2 l/min, streaks formed and lasting till the pass outlet and they could
the oil flowing through the middle region is less and local affect the cooling efficiency at the entrance of next pass [9].
temperature at these ducts therefore becomes higher; in For the case with 20 l/min, most extra supplied oil flows
Fig. 6a the highest temperature is observed to occur at the through the upper half of the pass and the duct wall

86 IET Electr. Power Appl., 2012, Vol. 6, Iss. 2, pp. 82– 90

& The Institution of Engineering and Technology 2012 doi: 10.1049/iet-epa.2011.0122
 www.ietdl.org

Fig. 6 Contours of fluid temperature

a 2 l/min and 15 W per disc
b 20 l/min and 30 W per disc

temperatures at the lower half remain higher. In Fig. 6b the rises linearly from the upstream to the downstream of a
worst temperature is observed at the lower right corner; at channel, so the HTC profile follows a linear reduction. With
the dead corner oil is almost stagnated. Secondly, due to the the help of CFD, the assumption can be examined in a
high flow rate at some horizontal ducts, there are second more detailed way.
eddy circulations generated at the entrance regimes; Fig. 7 shows the local HTC values for the cases with 2 l/min
fortunately it was not found that these second eddies would and 15 W per disc and 20 l/min and 30 W per disc. Fig. 7a
reduce the flow rate at the ducts. Furthermore, at the outer shows the values along the bottom side of the disc, which
vertical duct there are also hot-streaks discovered, but these actually bounds on the last horizontal duct of the previous
hot-streaks have lower temperatures than those in Fig. 6a pass. Since the portion of flow in this horizontal duct is the
(due to high flow velocities), and therefore, their influence highest of all passes, the HTC values are also high. The kink
upon the next pass is smaller than that in the case of 2 l/min. distribution patterns at the upstream end, that is, the left-hand
To emphasise the effect from the flow rate on the end, are due to the entry eddy circulations, then the HTC
convective heat transfer, the heat transfer coefficient (HTC) values gradually reduce, typically following hyperbolic
distribution around the bottom disc of the studied pass was trends, and are finally involved into the downstream flow
plotted in Fig. 7, for both the case of 2 l/min and 15 W per combination. The difference between the two cases can be
disc and the case of 20 l/min and 30 W per disc. These explained by the difference in the oil flow rates; high flow
HTC values were evaluated from the heat flux and the wall rate brings high HTC values.
temperature difference, on the oil side of the insulating Fig. 7b displays the HTC values at the upside of the disc.
paper, from a reference temperature; the heat flux and wall Beginning with the entry eddy caused kink patterns at the
temperature could be extracted from the CFD results, and upstream end, that is, the right-hand end, the HTC values
the inlet temperature of the pass, 608C, was taken as the decrease hyperbolically along the duct; this is known from
reference temperature. As network modelling assumed, the the literatures for heated infinite parallel channels. At the
heat flux is uniform and the wall temperature difference downstream combination profiles are then observed. The

Fig. 7 HTC around first disc

a Bottom side
b Top side

IET Electr. Power Appl., 2012, Vol. 6, Iss. 2, pp. 82 –90 87

doi: 10.1049/iet-epa.2011.0122 & The Institution of Engineering and Technology 2012
www.ietdl.org
HTC values after 40 mm from the entry is almost overlapped the hydraulic impedance the supplied oil flow should resist
because the flow velocities within the duct are in a quite to flow upwards through the winding.
similar range; for the high flow rate case, this first duct is Since only pressure drop of unheated cases was available
actually blocked by the separation eddy at its entry. from measurements, hydro-only calculations with the
material data, Table 4, were performed for comparison. The
6 Results comparison obtained results are shown in Fig. 8a; the calculation results
closely match the measurement values with the maximum
As the first step, the pressure drop along the complete pass has deviation less than 15 Pa, which equates to 1.8 mm
been compared. The pressure drop over a winding describes hydraulic head. In Fig. 8b, the pressure drop for the case of

Fig. 8 Pressure drop over the pass

a Case without losses
b Case with 30 W per disc

Fig. 9 Flow distributions on the horizontal oil ducts

a From network model for 15 W per disc
b From CFD for 15 W per disc
c From network model for 30 W per disc
d From CFD for 30 W per disc

88 IET Electr. Power Appl., 2012, Vol. 6, Iss. 2, pp. 82– 90

& The Institution of Engineering and Technology 2012 doi: 10.1049/iet-epa.2011.0122
 www.ietdl.org
30 W per disc losses is shown and the deviation between the cases of 2 l/min, visible in Figs. 9b and d. At such a low
network model and the CFD results is still small, less than flow rate, the buoyancy forces are able to drive the oil flow
30 Pa. The difference of the pressure drop between the from the hotter outlet vertical channel to the colder inlet
cases with 15 and 30 W per disc losses was below 2%, so vertical channel, and this tends to reverse the flow in the
that the results for 15 W are relinquished to display here. upper horizontal ducts into the opposite direction and the
From the comparison results, the network modelling was flow in the lower horizontal ducts into its original direction.
proved to give well fit pressure drop correlations. Finally, the flow distribution with higher losses becomes
In Fig. 9 the flow distribution among the horizontal more symmetrical.
channels is shown; the percentage refers to the whole oil In Fig. 10 the maximum temperature of the individual discs
flow, which enters the inlet of the pass and distributes is displayed. In the network modelling assumptions, this
among the individual horizontal channels. As seen in maximum temperature locates at the downstream end of
Figs. 9a and c, the flow distribution calculated by the each disc; however, in CFD results the whole volume of
network model is quasi-parabolic and nearly symmetrical to each conductor bears almost the same temperature due to
the axial middle of the pass, even though slightly more oil the high thermal conductivity of copper, and this
tends to flow through the upper half of the pass for the temperature is taken as the maximum temperature. The
cases with higher flow rates. In contrary, the distribution temperature profiles in Fig. 10 correspond to the flow
calculated by CFD is distinctly and visibly asymmetrical distributions in Fig. 9. As observed the temperatures
even for low flow rates. It gets increasingly imbalanced for calculated by the network model are scaled by the factor of
higher flow rates; for example, in the case of 10 l/min most loss density. For the CFD results the situation is similar, but
of the flow passes through the top two ducts. This was the only exception is in the case of 20 l/min, where a slight
indicated by the streamline plots in Fig. 5, in which it got temperature peak is recognisable at the second disc from
visible that separation eddies are blocking the lower ducts the bottom, probably due to the channel entry eddy
in the case with such a high flow rate. circulations. Another difference between the two modelling
The difference in the flow distribution between the cases approaches is that, for the low flow rates, particularly
with losses of 15 W per disc and 30 W per disc is 2 l/min, the network model predicted higher temperatures
obviously small, which implies that buoyancy forces are not than the CFD for both losses. This is because the network
very dominant, especially for high flow rates. In particular model predicted less uniform flow profiles than the CFD at
the network model did not show any difference between 15 the low flow rates. For high flow rates, especially 20 l/min,
and 30 W and tiny difference is only shown within the CFD the network model temperature prediction is relatively

Fig. 10 Temperature distributions on the winding conductors

a From network model for 15 W per disc
b From CFD for 15 W per disc
c From network model for 30 W per disc
d From CFD for 30 W per disc

IET Electr. Power Appl., 2012, Vol. 6, Iss. 2, pp. 82 –90 89

doi: 10.1049/iet-epa.2011.0122 & The Institution of Engineering and Technology 2012
www.ietdl.org
lower than the CFD, since CFD showed that the bottom oil given to the colleagues of CIGRE WG A2.38 for inspiritive
ducts are blocked by entry eddy circulations and thereby discussions. Financial support is also gracefully received
localised temperature peaks are formed. from the Academic Research Collaboration (ARC)
Programme between the British Council and Deutscher
7 Conclusions Akademischer Austausch Dienst (DAAD) for facilitating
this collaborated paper.
As transformer thermal modelling tools, network models and
CFD both require the same input parameters such as model
geometry, oil properties and boundary conditions including 9 References
loading and bottom oil flow rate and temperature. 1 Tanguy, A., Patelli, J.P., Devaux, F., Taisne, J.P., Ngnegueu, T.:
Fundamentally, network models employ the mathematic ‘Thermal performance of power transformers: thermal calculation
equations for the average quantities, such as oil velocity tools focused on new operating requirements’. Session 2004, CIGRE,
and temperature and so on, on each lumped element, that is, rue d’Artois, Paris, 2004
2 Oliver, A.J.: ‘Estimation of transformer winding temperatures and
oil duct or junction node, and some of the equations are coolant flows using a general network method’, Proc. Inst. Electr.
empirical. On the other hand, CFD uses much higher spatial Eng. C, 1980, 127, pp. 395– 405
discretisation and gets rid of empirical equations especially 3 Allen, P.H.G., Finn, A.H.: ‘Transformer winding thermal design by
when the flow is laminar. In this paper, both modelling computer’. IEE Conf. Publications, 1969, vol. 51, pp. 589–599
4 Simonson, E.A., Lapworth, J.A.: ‘Thermal capability assessment for
approaches were compared based on the same winding pass transformers’. Second Int. Conf. on Reliability of Transmission and
model in order to assess the influence of the different Distribution Equipment, 1995, pp. 103– 108
discretisation of both approaches upon the modelling results. 5 Del Vecchio, R.M., Poulin, B., Feghali, P.T., Shah, D.M., Ahuja, R.:
From the comparison conducted, it was concluded that the ‘Transformer design principles: with applications to core-form power
deviation between the pressure drop calculated by CFD and transformers’ (Gordon and Breach Science Publishers, 2001)
6 Zhang, J., Li, X.: ‘Coolant flow distribution and pressure loss in ONAN
the measured values is quite low and that, once adapted, the transformer windings. Part I: Theory and model development’, IEEE
network modelling method delivered reliable results for the Trans. Power Deliv., 2004, 19, pp. 186– 193
pressure drop as well. It is a proof to show that network 7 Radakovic, Z., Sorgic, M.: ‘Basics of detailed thermal-hydraulic model
modelling would be able to provide a quick solution for for thermal design of oil power transformers’, IEEE Trans. Power
Deliv., 2010, 25, pp. 790– 802
predicting winding pressure drops and thereby assist to 8 Takami, K.M., Gholnejad, H., Mahmoudi, J.: ‘Thermal and hot spot
choose capable oil pumps for forced oil cooling at thermal evaluations on oil immersed power transformers by FEMLAB and
design stage. It is to remark that only the model and MATLAB software’s’. Proc. Int. Conf. on Thermal, Mechanical
material parameters were inputted into the CFD and and Multi-Physics Simulation Experiments in Microelectronics and
network model programs and no any calibration or adaption Microsystems, EuroSime 2007, 2007, pp. 1–6
9 Kranenborg, E.J., Olsson, C.O., Samuelsson, B.R., Lundin, L.-A.,
from the measurement results was applied. Missing, R.M.: ‘Numerical study on mixed convection and thermal
Secondly, with high total oil flow rates such as 20 l/min, streaking in power transformer windings’. Fifth European Thermal-
the flow distribution delivered by the network modelling Sciences Conf., The Netherlands, 2008
deviates from the one by CFD. The main difference was 10 Weinläder, A., Tenbohlen, S.: ‘Thermal– hydraulic investigation of
transformer windings by CFD-modelling and measurements’. Proc.
particularly due to the separation eddies at the bottom ducts 16th Int. Symp. on High Voltage Engineering, South Africa, 2009
as observed in the CFD results; network model cannot fully 11 Torriano, F., Chaaban, M., Picher, P.: ‘Numerical study of parameters
capture the impact of complex phenomena like flow affecting the temperature distribution in a disc-type transformer
separation. Moreover, the resulted disc temperatures from winding’, Appl. Thermal Eng., 2010, 30, pp. 2034–2044
network model are lower than those from CFD for high oil 12 Wu, W., Wang, Z.D., Revell, A.: ‘Natural convection cooling ducts in
transformer network modelling’. Proc. 16th Int. Symp. on High
flow rates; however, the comparative relation acts Voltage Engineering, South Africa, 2009
oppositely for low flow rates. 13 Zhang, J., Li, X.: ‘Coolant flow distribution and pressure loss in ONAN
transformer windings. Part I: theory and model development’, IEEE
Trans. Power Deliv., 2004, 19, pp. 186– 193
8 Acknowledgments 14 Wu, W., Revell, A., Wang, Z.D.: ‘Heat transfer in transformer winding
conductors and surrounding insulating paper’. Proc. Int. Conf.
Stefan Tenbohlen and Andreas Weinläder would like to thank on Electrical Engineering, Shenyang, China, 2009
the Deutsche Forschungsgemeinschaft (DFG) for sponsoring 15 Wu, W., Wang, Z.D., Revell, A., Iacovides, H., Jarman, P.: ‘CFD
this research project. Zhongdong Wang and Wei Wu would calibration for network modelling of transformer cooling oil flows –
like to thank the Engineering and Physical Sciences Part I: heat transfer in oil ducts’, IET Electr. Power Appl., 2011, 6,
(1), pp. 19–27
Research Council (EPSRC) – National Grid Dorothy 16 Wu, W., Wang, Z.D., Revell, A., Jarman, P.: ‘CFD calibration for
Hodgkin Postgraduate Award (DHPA) and National Grid network modelling of transformer cooling oil flows – Part II: pressure
for their financial sponsorship. Due appreciation should be loss at junction nodes’, IET Electr. Power Appl., 2011, 6, (1), pp. 28–34

90 IET Electr. Power Appl., 2012, Vol. 6, Iss. 2, pp. 82– 90

& The Institution of Engineering and Technology 2012 doi: 10.1049/iet-epa.2011.0122

View publication stats