Está en la página 1de 8

Proceedings of the ASME 2017 Power Conference Joint With ICOPE-17

POWER2017-ICOPE-17
June 26-30, 2017, Charlotte, North Carolina, USA

POWER-ICOPE2017-3319

STUDY ON CALCULATION METHOD OF COOLING OIL FLOW RATE ALLOCATION


FOR NATURAL OIL CIRCULATION TRANSFORMER

Rong Xie Na Wang


Key Laboratory of Ocean Energy Utilization and Key Laboratory of Ocean Energy Utilization and
Energy Conservation of Ministry of Education, Energy Conservation of Ministry of Education,
School of Energy and Power Engineering, School of Energy and Power Engineering,
Dalian University of Technology, Dalian University of Technology,
Dalian, Liaoning, China,116024 Dalian, Liaoning , China,116024

Zifu Lu Zhonglin Cheng Baopeng Xu


Hangzhou Steam turbine Co., Key Laboratory of Ocean Energy Key Laboratory of Ocean Energy
Ltd, Utilization and Energy Utilization and Energy
Hangzhou, Zhejiang, China, Conservation of Conservation of
310022 Ministry of Education, Ministry of Education,
School of Energy and Power School of Energy and Power
Engineering, Engineering,
Dalian University of Technology, Dalian University of Technology,
Dalian, Liaoning, China,116024 Dalian, Liaoning, China,116024

ABSTRACT which was based on the part of experimental data, the numerical
For the natural oil circulation power transformer, the hot calculation results are in agreement with these data. And the
spot temperature and winding temperature should be within error between them is within 6 percent. Therefore, the
prescribed limits so as to ensure its life and reliability. feasibility and accuracy of calculation method, which was used
Temperature rise of inner windings not only depends on the to calculate oil flow of natural oil circulation power transformer
transformer loss, but relates to oil flow closely. Therefore, windings by the porous media model, has been verified as well.
whether the oil flow of transformer and oil flow distribution There are two kinds of conventional method to calculate oil
rules among the three windings can be predicted or not, relates flow, including an empirical method and an experimental
to success or failure of the winding temperature rise research. method. On the one hand, although the former kind of method
According to the structural characteristics of transformer, a can be applied conveniently, it cannot consider all the factors so
three-dimensional computational model including windings and that the accuracy cannot be guaranteed. On the other hand,
external radiators was built by a three-dimensional modeling experimental method to calculate the flow can present high
software. Initially, from the view of flow, the flow resistance of accuracy, but the cost of carrying out experimental study is high.
natural oil circulation power transformer windings has been Therefore, this paper presents another calculation method aimed
calculated and analyzed using the CFD methodology, and the to the flow resistance and oil flow by CFD. And the calculation
resistance characteristic curves about high voltage windings, method presented has higher calculation accuracy compared to
middle voltage windings, low voltage windings, radiators and above. Meanwhile, the new method provides a theoretical
pipelines were fitted respectively by the least square method. reference about oil flow distribution for the design of the power
And then, on the basis of above, the porous media model was transformer.
applied to simplify the integral model of transformer so as to
Key words: transformer, flow resistance, inlet flow rate
build a three-dimensional coupled porous media-pure fluid
allocation, CFD
computational model. Meanwhile, the author calculated
porosities and linear resistance coefficients of high voltage, INTRODUCTION
middle voltage and low voltage windings. At the end of this
paper, to obtain the relationship of oil flow distribution among Transformer is one of the core equipment in the power
three windings, the pressure loss was calculated by numerical system, and its reliability and stability under operation directly
simulation. Compared with the theoretical calculation results concerns the security of the entire grid. With the growing

1 Copyright © 2017 ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 09/14/2017 Terms of Use: http://www.asme.org/about-asme/terms-of-use


demand for electricity, the transformer voltage level and and ignored the influence of transformer temperature on
capacity also continue to be improved. Because of the limitation winding loss. Therefore, compared with three-dimensional
of transportation and installation, the increasing space of weight model, 2D model has certain errors. Ziyu Jia[6] put forward oil
and size for transformer gets less and less, so that the problem flow distribution engineering calculation method for the forced
of winding temperature rise is getting serious. oil circulation guide cooling transformer on the basis of a large
For oil immersed transformer, the winding temperature rise number of experiments. Yongteng Jing[7] established three-
not only depends on the losses of transformer, but also closely dimensional multi-physics coupling mathematical model and
relates to the oil flow rate. Therefore, whether the oil flow rate researched the transformer winding heat problem based on the
distribution law among different windings can be accurately nonlinear influence of material. M.A. Taghikhani[8] used finite
predicted or not, closely relates to the success or failure of the element method to simulate a transformer with non-directed oil-
winding temperature rise research. All kinds of problems make forced cooling and directed oil-forced cooling. He also
multiple physical fields research for transformers deserving of compared the results with those obtained from finite integral
more attention. The researchers can reveal the physical model. The experimental test checks the validity and accuracy
properties in electromagnetic, mechanical aspects for of the proposed method. Smolka[9] presented a state-of-the-art of
transformers. However, as for the transformer cooling problem, the numerical modeling of the coupled problems involving heat,
the research is relatively weak. Currently, three main methods fluid flow and electromagnetic phenomena in electrical
are often used to calculate the spot temperature rise, namely transformers by establishing three-dimensional model. Chao
empirical formula method, hot road model method and Liu[10] studied temperature rise of windings by introducing a
numerical simulation method. The first one has been applied in quasi-3D coupled-field method, and applied it on a ventilated
certain practical engineering, offering different parameters to dry-type transformer.
different types of transformers, but it ignores too many factors. In view of mentioned above, a three-dimensional coupled
The second one mainly uses the thermoelectricity analogy to porous media-pure fluid computational model was put forward
build thermal model of transformers in order to solve the hot in this paper to analyze the pressure loss of the different
temperature. But the result is closely related to the parameters windings and distribution of oil mass flow among the windings.
set in the method. To a certain extent, the two methods above The flow resistance characteristics and the cooling oil flow
ignore the transformer structure and the cooling medium on the distribution rule for oil-immersed transformer were focused on
influence of the hot spots temperature. Therefore, in order to from the flow view by the commercial CFD software.
more systematically research the temperature distribution inside
PHYSICAL MODEL
the transformer, many researchers proposed the numerical
simulation method based on fluid mechanics and heat transfer An ODFS - 334000/500 type transformer was studied in this
theory, and established the physical model to analyze the paper—The capacity is 334000KVA, and the high voltage is
temperature field. However, due to the complex internal 500KV. The geometric model of windings is composed of 48
structure of the transformer, the model had to be simplified in groups of blocks and stays along the circumferential direction,
the calculation. In addition, the pressure loss was mainly the windings was thus divided into 48 equal parts. According to
calculated by empirical formula to, so the errors are inevitable. the characteristic of axial symmetry, only 1/48 module between
To some extent, multiple physical fields coupling analysis two sets of blocks and stays was taken for three-dimensional
method is put forward to enhance the accuracy of the modeling. The external radiators are symmetrically arranged on
temperature field solving. Yuan Gao[1] built a two-dimensional both sides of the oil tank. Each side consists of the same 14
model for windings, and analyzed the temperature field and independent radiator pieces, except four radiator pieces located
flow field using fluid-solid interaction method. The method in the lateral are slightly different from the others, the rest of the
provides a reference for improving of winding temperature rise. radiator pieces has the similar structure and thermal properties.
Haibin Fang[2] adopted finite volume method to study the flow Therefore, the only one of the radiator pieces is taken to
rate distribution and temperature rise of oil-immersed construct model according to its symmetrical characteristic. The
transformer by two-dimensional model. Mufuta[3]adopted a two- model diagram of high-voltage winding, middle-voltage
dimensional model according to symmetrical characteristics to winding, low-voltage winding and radiator are respectively
simulate temperature field and flow field under different mass shown in figure 1.
flow. Meanwhile, the author researched the effect of Reynolds
number and Gr number on heat dissipation effect. Skillen[4]
established integral two-dimensional model to simulate the
effect of heat flux on oil temperature and flow about low
voltage windings. Zhang et al.[5] developed two-dimensional
model to investigate the temperature distribution in a disc-type
transformer and formulated empirical correlations for
calculating local heat transfer coefficients. Most of the current
analysis was based on two-dimensional axisymmetric model,

2 Copyright © 2017 ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 09/14/2017 Terms of Use: http://www.asme.org/about-asme/terms-of-use


the radiator and air was chosen the third boundary condition.
Because the fan is used to strengthen heat transfer by air side,
according to the experience, a given surface heat transfer
2
coefficient was set to 24W⋅ m ∙ K 1 . The ambient temperature
was set to 20℃.
NUMERICAL SOLUTION
In low-voltage winding, for example, the pressure contour
was illustrated in Fig.2 under given 0.0375kg/s inlet flow rate. It
is evident that the maximum value is located at the oil inlet, and
the oil pressure decreases gradually from bottom to top. But the
pressure difference is not obvious along horizontal direction.
The main reason is the oil viscidity. On the one hand, the
frictional resistance loss was caused by the friction effect
between the oil and the flow channel in the flow process. On the
other hand, the local resistance loss was produced by the oil
a. HV winding b. MV winding convectional effect and vortex dissipation in the horizontal and
vertical flow channels intersections. The comprehensive effect
of above two kinds of losses leads to fluid pressure decreases
along the vertical direction. However, owing to slower speed
and smaller flow loss of the oil in the horizontal direction, the
pressure difference is not obvious.
The pressure contour of the radiator was shown in Fig.3. It
can be seen that, the flow channel in the radiator is relatively
simple and the flow direction does not change. So the local
resistance loss in the flow process is low, leading to the total
pressure loss also less than that in windings. At the same height
of inlet side, the pressure in the center channel is higher than
that in both side channels. While the pressure change is exactly
opposite at the exit. The main reason is that the middle oil flow
channel is near the upper collection tube, the pressure loss is
small. While the pressure is lower on either side because the oil
flow channel is far away from the upper tube. Similarly, at the
exit, compared with the oil flow channel on both sides, the
c. LV winding d. radiator middle oil flow channel is close to the lower collection tube, the
pressure loss of the oil is lower before entering the lower
Fig.1 Physical model collection tube, so the pressure is higher. Based on the above,
the oil flow rate in middle flow channel is higher than that in
MESHING AND BOUNDARY CONDITIONS both sides of the channels. But because both sides of the
The winding model was divided into solid domain and channels directly contact with the atmosphere, the cooling effect
fluid domain. The two domain structured grids were divided by is better than that in middle passage. It indicates that the radiator
ICEM soft. And the grid was finely meshed near the wall. After cooling effect does not achieve the optimal status.
grid independence test, the grid number of high voltage
winding, middle voltage winding and low voltage winding is
respectively 5 million, 5.7 million and 5.4 million.
As the oil flow velocity is very slow for natural oil
circulation transformer, and Reynolds Number is below critical
degree by several orders, the flow in the transformer was
modeled laminar.
Circumferential boundaries was set as symmetric
boundary. The inlet condition was set to the velocity boundary
and the outlet was set to the pressure boundary. The inlet
temperature of fluid domain for a given oil was set to 29℃ and
the outlet pressure was set to 0 Pa. The heat transfer between

3 Copyright © 2017 ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 09/14/2017 Terms of Use: http://www.asme.org/about-asme/terms-of-use


Bottom Top
Fig.2 Pressure contour in LV winding Fig.3 Pressure contour of radiator

under 0.0375kg/s inlet flow rate On the basis of above computing results, the relation curves
between the pressure loss and the mass flow were fitted by
In order to obtain pressure loss-flow rate curves of windings using the least square method, shown in Fig.4, respectively for
under different mass flow, every winding was simulated under high-voltage winding, middle-voltage winding, low-voltage
eight flow rates. Big distinction between the selected mass flow winding, radiators and pipelines (including upper and lower
and the real mass flow may lead to a deviation in the fitting collection tube). From the figures, it can be seen that the
results. For this reason, the flow rate was firstly evaluated by an pressure loss increases with the mass flow increasing. When the
empirical equation (1), and then the eight inlet mass flow were inlet mass flow increases, the oil average temperature will
selected from 0.5Qm to 2Qm (in the area [0.5Qm, 2Qm]). reduce, resulting in increased oil viscosity. The higher oil
viscosity is, the more obviously the flow is hindered.
Qm  1.1PL (1)
Where Qm is the inlet mass flow
PL is the loss in the windings.
According to energy conservation equation, the transformer
loss can be obtained by equation (2). The inlet oil temperature
in the radiator is approximately equal to the outlet temperature
of the windings,
PT  c pQm T (2)

Where PT is the transformer loss


Qm is inlet mass flow
c p is the average specific heat coefficient of oil
∆T is the temperature rise between inlet and outlet of the
radiator.

4 Copyright © 2017 ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 09/14/2017 Terms of Use: http://www.asme.org/about-asme/terms-of-use


Fig.4 Relation curves between pressure loss and mass flow

INTEGRAL MODEL ESTABLISHMENT


The structure size of windings is 6m×4m×5m,while the
gap between two layers of windings is 2mm, so the number of
grids meeting computing requirement is considerable. The grid
quantity of high-voltage winding, middle-voltage winding, low-
voltage winding, and radiators is respectively 5 million, 5.7
million, 5.4 million and 136, 000. If integral model is directly
built, too large grid quantity may not be able to run in a
workstation and will take a very long computing time.
Obviously, the method of integrally modeling cannot conform to
the engineering practice. Therefore, it is convenient to simulate
by dividing the integral model into several blocks.
In previous studies, the middle-voltage winding was
simulated in isolation. The inlet of oil domain for a given oil
temperature was set to 29 ℃ , and given mass flow was
0.052kg/s. Pressure of outlet was set to 0 Pa. The top oil
temperature rises was presented in Table 1. There is a big error
between the simulative result and the experimental one. One
reason is the inlet mass flow of different windings is achieved
from empirical equation (1). Another one is the computing
model is isolated.
In order to obtain more accurate computing results and
make the grid quantity meeting the workstation requirements, a
porous media model was established to simplify the whole
structure, including high-voltage winding, middle-voltage
winding, low-voltage winding, pipeline system and radiators.
And the inlet mass flow in different windings is allocated
according to above pressure loss-flow rate curves. The
simplified model effectively reduced the grid quantity, took all
kinds of critical parts into consideration synthetically. As seen in
Fig.4, the pressure loss of oil in the pipeline between the
windings and the radiator accounts for small percentage
compared to total loss. So the loss in the pipeline has little
impact on oil flow. Above pipeline system was simplified to
single pipe with the same diameter as pipeline installed in

5 Copyright © 2017 ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 09/14/2017 Terms of Use: http://www.asme.org/about-asme/terms-of-use


radiators. The temperature field of integral model was presented Table 2. Simulation results of integral model
in figure 5.
linear
Table 1. Comparison of top oil temperature rise for the Top oil pressure resistance mass
middle-voltage winding Model temperature loss coefficient flow
type (K) (Pa) 3 1 (kg/s)
Model type Top oil Top oil (kg∙ m ∙ s )
(middle-voltage temperature temperature
HV 54.85 183.1 31550 0.139
winding) (℃) rise(℃)
Isolated model 77 48 MV 59.05 201.8 33818 0.096
Integral model 59.5 30.5 LV 56.35 197.8 18024 0.060
experimental data --- 30.4

For a porous media model, the continuity equation and


energy equation are the same as the standard governing
equations, except for the momentum equation, obtained by
adding a resistance term on the standard N-S equation. And for
the porous media model, the resistance term can be written as,
p (3)
  C1vi  C2 v j v j
xi
Where p is the pressure
C1 is the first resistance coefficient
C 2 is the second resistance coefficient
vi and v j are the velocity components on i and j
direction.
Because of low oil velocity in the winding, the second
resistance term donates only a little to the total resistance, Fig.5 Contours of temperature of integral model.
which can be omitted, so eq. (3) is equivalent to, THEORETICAL CALCULATION
p
  C1vi (4) For furthermore verifying the accuracy of porous media
xi model, theoretical calculation was conducted according to the
Eq. (4) is known as Darcy’s Law. It shows that, under low balance principle of driving force and resistance under steady
oil velocity, flow resistance is proportional to the velocity. flow. The curves of buoyancy and resistance are presented in
Furthermore, there is little temperature variation of oil in the Fig.6. The flow balance lies in the intersection of the two curves.
pipeline, so it was treated as isentropic flow.
The oil temperature, linear resistance coefficient and mass
flow of oil are shown in Table 2 by simulating the simplified
model. Because only one single piece of the radiator was
simulated, the real mass flow is 28 times of the simulation
solution, thus the real mass flow in high-voltage winding,
middle-voltage winding and low-voltage winding is respectively
3.89kg/s, 2.69kg/s and 8.27kg/s.
The top oil temperature of middle-voltage obtained from
integral model was compared with the experimental result,
shown in Table 1. Obviously, the two are in good agreement,
verifying the porous media model is effective and credible.

Fig.6 Curves of buoyancy and resistance

6 Copyright © 2017 ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 09/14/2017 Terms of Use: http://www.asme.org/about-asme/terms-of-use


(2) to obtain the blow formulations:
13.69
1  6  (12)
Q1
23.89
2  6  (13)
Q2
29.29
3  6  (14)
Q3
146.6
4  6  (15)
QT
73.3
5   6  (16)
QT
Where Q1 , Q2 , Q3 , QT is respectively the mass flow in high-
Fig.7 Simplified network model voltage windings, middle-voltage windings, low-voltage
windings, radiators.
In the porous media model, the windings and the radiator Refer to the resistance characteristic of pipeline, the inlet
respectively locate on two sides, and the cooling oil circularly and outlet pipelines of radiator was simplified. The pressure loss
flows, so the influence of gravity was ignored. and the flow rate respectively achieved from the theoretical
Buoyancy appears due to the density difference between calculation and the simulation of the porous media integral
different location oil, the computation formula is as follows: model are presented in table 3.
h
F     R  x   S x gdx (5) Table 3. Comparison of theoretical results
0
and simulation results
Where F is buoyancy for oil, Pressure loss Mass flow
 R is density of oil around windings, Model
type Pa kg/s
 S is density of oil around radiator.
theoretical simulation error theoretical simulation error
According to the simplified network model of the results results results results
transformer, as shown in figure 7, the buoyancy expression of HV 193.3 197.8 2.3% 1.73 1.69 2.3%
winding circuit is as follows: MV 218.8 201.8 7.8% 2.85 2.69 5.6%
FLV   5 gh5  6 gh6    1gh1  4 gh4  (6)
LV 205.4 183.1 10% 4.04 3.89 3.7%
FMV   5 gh5  6 gh6    2 gh2  4 gh4  (7) radiator 61.7 58.4 5.3% 8.62 8.27 4.1%

FHV   5 gh5  6 gh6    3 gh3  4 gh4  (8)


As seen in Tab.3, in addition to the pressure loss of the
Considering the balance principle of resistance and
low-voltage winding, most of the errors are within the range
buoyancy, equal (9)-(11) are available:
of engineering precision.
FLV  P1  P4  P5  P6 (9)
FMV  P2  P4  P5  P6 (10) CONCLUSION
(1) The resistance characteristic of transformer was focused on
FHV  P3  P4  P5  P6 (11)
by separate simulation for every components. The flow
Where ∆ P1 , ∆ P2 , ∆ P3 , ∆ P4 , ∆ P5 , ∆ P6 are respectively the resistance of different components increases with the increasing
resistance of cooling oil in low-voltage winding, middle-voltage of flow rate. The faster the flow rate increases, the bigger the
winding, high-voltage winding, cross-upper pipe, radiator, cross- flow resistance is. The determination of flow resistance lays the
foundation for the inlet flow rate calculation in different
lower pipe; 1 ,  2 ,  3 ,  4 ,  5 ,  6 are respectively the
windings.
density of cooling oil in low-voltage winding, middle-voltage (2) The porous media integral model was built to simplify the
winding, high-voltage winding, cross-upper pipe, radiator, cross- transformer. Compared with isolated model, it can obtain results
lower pipe. in good agreement with the experimental ones.
In addition, owing to the oil density changing with the
temperature, the above expressions were combined with equal

7 Copyright © 2017 ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 09/14/2017 Terms of Use: http://www.asme.org/about-asme/terms-of-use


REFERENCES [6] M.A. Taghikhani, Gholami. Heat transfer in power
[1] Yuan Gao. The Analysis on Thermal Field in Winding of transformer windings with oil-forced cooling [J]. IET Electric
Large Oil-immersed Forced-air Cooling Power Transformer [D]. Power Applications, 2009, 3(1):59-66.
Harbin: Harbin University of Science and Technology, 2012. [7] Ziyu Jia, Riqi Yao, Zhenxin Ding, etc. Engineering
[2] Haibin Fang, Xiaoming Liu, Zenghui Lai, etc. Influence calculation methods for the forced oil circulation guide cooling
factor of winding oil flow distribution and temperature rise [J]. transformer [J]. TRANSFORMER, 1998, 32(2):21-24.
TRANSFORMER, 2013, 50(1):18-22. [8] Yongteng Jing, Yan Li, Longnv Li,etc. Comprehensive
[3] J.M. Mufuta. Thermal hydrodynamic model of oil-immersed Calculation of Winding Loss, Fluid and Temperature Field in
power transformers [J]. ASME Fluids Engineering Division, Large Power Transformer, Trans Tech Publications, 2014.
2001, 256: 319-326. [9] J. Smolka, O. Bíró, A.J. Nowak. Numerical simulation and
[4] A. Skillen, A. Revell, H. Iacovides, W. Wu. Numerical experimental validation of coupled flow, heat transfer and
prediction of local hotspot phenomena in transformer windings electromagnetic problems in electrical transformers [J].Archives
[J].Applied Thermal Engineering, 2012, 36: 96-105. of Computational Methods in Engineering, 2009, 16: 319-355.
[5] J. Zhang, X. Li, M. Vance, Experiments and modeling of [10] Chao Liu, Jiangjun Ruan, Wu Wen, etc. Temperature rise of
heat transfer in oil transformer winding with zigzag cooling a dry-type transformer with quasi-3D coupled-field method [J].
ducts. Applied Thermal Engineering28 (2008): 36-48. The Institution of Engineering and Technology, 2015, 10 (7):
598-603.

8 Copyright © 2017 ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 09/14/2017 Terms of Use: http://www.asme.org/about-asme/terms-of-use

También podría gustarte