CFD analysis of directed oil flow in power transformers.

Pupaza, Cristina ; Stancu, Octavian ; Parpala, Radu Constantin 等

1. INTRODUCTION

Large power transformers as generator step-up (GSU) are operating life time under full load. An efficient cooling system is a crucial factor determining the operational safety. Actually, the best solution is the Oil Directed (OD) cooling system. It improves cooling, lowers the hot-spot temperature and reduces the thermal aging of the insulation. The design task is to assure sufficient oil flow in all the parts of the system, while keeping the pressure drop as small as possible. Recommendations for the design of power transformers are found in loading guides (IEC 60076-7, 2005), but this standard does not address the local distribution of the coolant flow inside the windings. It is thus up to each transformer manufacturer to make use of appropriate tools such as Computational Fluid Dynamics (CFD) in the design process. Detailed studies were undertaken with regular CFD solvers on two dimensional models. However, CFD is considered unfeasible for the daily design work due to the required computer time (Kranenborg et al., 2008). One complication for CFD models of transformer cooling is that the flow in the tank is three dimensional and unsteady. Although OD cooling has already been calculated in the past (Karsai et al., 1987), (Zhang & Li, 2004) and experiments were described in the literature (Felber et al., 1984), (Weinlaser & Tenbohlen, 2008), we promote a method to rapidly calculate the cooling system using 3D modeling of the windings in an early stage of the design. The simulation takes advantage of the axisymmetry, the cylindrical shape and the repetitive pattern of the winding sections in the flow path. Different oil patterns were performed in order to obtain a balanced pressure drop on each winding. The results were compared with analytical calculations and a good accuracy was found. The method proved to be efficient in a simulation driven design approach.

2. CFD PROCEDURE

The fluid flow problem is defined by the laws of conservation of mass (1) and momentum (2) as follows:

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (1)

where [v.sub.x], [v.sub.x], [v.sub.x] are components of the velocity vector, [rho] density and t--time. The rate of change of density is replaced by the rate of change of pressure. In a Newtonian fluid flow the relationship between the stress and the rate of deformation of the fluid is:

[[tau].sub.ij] - [rho][[delta].sub.ij] + [mu] ([[partial derivative]u.sub.i] [[partial derivative]x.sub.j] + [[partial derivative]u.sub.j] [[partial derivative]x.sub.i] (2)

where [[tau].sub.ij] is the stress tensor and [u.sub.i], [u.sub.j]--the orthogonal velocities and [mu]--the dynamic viscosity. These laws are expressed in terms of partial differential equations which are discretized with a finite element based technique. The solution method used in CFD codes is known as the Finite Volume technique (ANSYS CFX, 2005).

3. MODELING AND SIMULATION

The CFD analysis aimed to determine the pressure loss for several oil patterns in the Low Voltage (LV) and High Voltage (HV) windings of a 440 MVA GSU transformer. The 3D models were processed in CATIA V.5.18 and imported in ANSYS CFX using a neutral file.

[FIGURE 1 OMITTED]

Because the symmetry and the vertical pattern of the models only 11 and respectively 8 disc sectors of the HV and LV windings were processed, as shown in Fig. 1.

3.1 Mesh definition

The models were meshed using advanced front and inflation 3D methods combined with controlled body, edge and face spacing. A proper angular resolution option was set in order to obtain a sufficiently refined mesh (Fig. 2).

[FIGURE 2 OMITTED]

Some faces and edges were declared in a virtual topology environment to eliminate narrow or short edges.

Finally the HV and LV cooling system models had 281652 nodes and 1306187 elements and 135312 nodes corresponding to 616485 elements, respectively. Mesh quality control checks were performed and no inconsistencies were found.

3.2 Simulation conditions

The oil flow capacity at the inlet was imposed as shown in Table 1, in order to eliminate the copper losses using the temperature drop in each winding. The simulation conditions were laminar flow of pure and nearly incompressible substance. The wall influence on the flow was set no slip. The uninhibited oil transformer met the required specifications (IEC 60296, 2003). Oil properties were assumed to be at the average temperature of the windings. The convergence criterion was chosen the RMS residuals, with a suitable residual target. For adequate convergence the physical timescale was calculated taking into account the average residence time. The residuals were plotted each time step in order to monitor the solution convergence. All the computations terminated with a normal convergence, after 70-80 iterations. No parallel computing was needed and the solver enabled the calculation in a practical amount of time.

4. RESULTS

A balanced pressure drop on each winding was obtained. The windings were calculated for both flow directions, when the oil enters from the inner and outer section, because the flow conditions were different.

The oil velocity in each cylindrical sector was processed in order to avoid stagnant regions and to check the risk of oil electrisation at cold start. Figure 3 shows the velocity distribution for the HV and LV calculated segments, when the oil enters from the inner section. The velocities were also displayed in a horizontal plane (P1) located in the neighborhood of the inlet (Fig. 4). Results were included in Table 2

[FIGURE 3 OMITTED]

[FIGURE 4 OMITTED]

When calculating the total pressure loss not only the number of calculated sections, but also the relative position of the flow path were considered. The numerical results of the pressure drop were compared with the analytical values and this verification confirmed that the prediction of the pressure loss was accurate.

5. CONCLUSION

The proposed CFD procedure enforces traditional capital intensive design processes of power transformers with a simulation driven design method. It is performed on a three dimensional model, which is the most reliable simulation for the power transformer cooling system. It avoids late design changes and takes advantage of the parametrical modeling of the transformer windings, enabling the optimization of the cooling system.

At this stage of the research we neglected the dependence of the oil properties with the temperature. The next step of the simulation will be a multiphysics approach, coupling the fluid dynamic analysis with the thermal-electromagnetic solution.

This approach provides a proper understanding of the cooling system, reduces the need for expensive prototypes and gives comprehensive data that is not easily obtainable from experimental tests. When optimizing new power transformer designs, many "what if' scenarios can be analyzed in a short time. The result is an improved performance of the power transformer, as well as a reliable and a durable design.

6. REFERENCES

ANSYS CFX. (2005). Fluids Analysis Guide, ANSYS Inc., Southpointe, 275 Technology Drive, Canonsburg, PA15317

Felber, W.; Damm, B.; Loderer, K. & Preininger, G. (1984). Evaluation of Thermal Conditions of Large Transformers, CIGRE International Conference on Large High Voltage Electric Systems, Imprimerie Louis-Jean, Paris

IEC 60076-7. (2005). Power transformers. Part 7: Loading guide for oil-immersed power transformers, International Electrotechnical Commission, December 1, 2005

IEC 60296. (2003). Fluids for electrotechnical applications. Unused mineral insulating oils for transformers and switchgear, Int. Electrotechnical Commission, Nov. 1, 2005

Karsai, K.; Kerenyi, D. & Kiss, L. (1987). Large Power Transformers, Akademiai Kiado, ISBN 963-05-4112-2, Budapest

Kranenborg, E.J.; Olsson, C.O.; Samuelsson, B.R.; Lundin, L.A. & Missing, R.M. (2008). Numerical Study on Mixed Convection and Thermal Streaking in Power Transformer Windings, Available from: www.eurotherm2008.tue.nl/ Proceedings_Eurotherm2008/papers/Mixed_Convection/M CV_3.pdf, Accessed: 2009-03-15

Weinlaser, A. & Tenbohlen, S. (2008). Thermohydraulische Untersuchung von Transformatorwicklungen durch Messung und Simulation. Thermal-hydraulic Analysis of Transformer Development, Available from: www.unistuttgart.de/ieh/forschung/veroeffentlichungen/30_Weinlae der%20Andreas.pdf Accessed on: 2009-02-20

Zhang, J. & Li, X. (2004). Coolant Flow Distribution and Pressure Loss in ONAN Transformer Windings, IEEE Trans. on Power Delivery, pp. 186-199, Vol. 19, No. 1, Jan. 2004

Tab. 1. Inlet input data

Model Flow capacity Surface [m2] Speed [m/s]
 [[m.sup.3]/s]

 HV 2.576 x [10.sub.-4] 6.46 x [10.sub.-4] 0.39878
 LV 1.554 x [10.sub.-4] 4.36 x [10.sub.-4] 0.35655

Tab. 2. Simulation results

 HV LV

Flow Pressure Min. Pressure Min.
case loss Velocity [Pa] Velocity
 [Pa] [m/s] [m/s]

1 247.853 9.942-[10.sup.-5] 163.392 1.002-[10.sup.-5]
2 489.836 4.079-[10.sup.-5] 335.630 1.913-[10.sup.-5]