Sie sind auf Seite 1von 4

IEEE MELECON 2004, May 12-15,2004, Dubrovnik, Croatia

Thermal Analysis for Determination of Current Carrying Capacity of PE and XLPE Insulated Power Cables Using Finite Element Method
I. Kocar*, A. Ertas** * Aselsan Electronics Industries, Inc., Ankara, Turkey I* Middle East Technical University/Department of Electrical and Electronics Engineering, Ankara, Turkey
Abstract-Transient heat transfer analysis of medium voltage, PE and XLPE insulated underground power cables are investigated by applying finite element modeling developed under a finite element software utilizing principle of virtual temperatures which is the quadratic functional form of the variational formulation of the transient heat transfer equation. Obtained results are compared with and verified by experimental heat transfer data for PE insulated cable. In order to minimize the truncation error arising from the boundary condition choice, two different boundary conditions for the selected boundaries namely Dirichlet and Neumann conditions are applied for different domain dimensions, and, it is shown that appropriate solution domain can be obtained. Then, PE and XLPE insulated cables are compared from the heat transfer point of view. Cable temperature dependency to variations in thermal conductivity of the surrounding medium and loading is investigated. Thus, correction factors given by cable manufacturers regarding cable current carrying capacity are obtained by using the finite element model developed in this study.

I. INTRODUCTION Determination of thermal behavior of underground power cables plays an important role in the design and manufacture processes of the cables as the heat generated inside the cable may lead the cable breakdown thermally. Maximum allowable power that can be carried through electrical cables is specified by two factors: withstanding level of the insulator to excessive temperature and the rated voltage level of the cable. Heat generated inside the cable is the result of the current passing through the cable (Ohmic losses, 12R) and insulation losses respectively. As insulation of the cable and soil are not good thermal conductors, the heat generated inside the cable may not be transferred away effectively resulting thermal breakdown of the cable eventually. Therefore, transfer of the heat generated inside the cable should be well investigated in steady state and transient conditions. As it can be understood, a strong dependence exists between the current carrying capacity of the cable and heat transfer from the cable to its surrounding. Thus, current ratings of cables has been established analytically or numerically based on the thermal conductivity of the mediums by power utilities so far. The analytical techniques are usually based on IEC publication 287. Numerical techniques are based on either finite difference or finite element methods. Finite element methods can be employed for complicated boundary conditions and cable 0-7803-8271-4/04/$20.00 02004 IEEE

geometries [l, 2 and 31 where finite difference methods can not easily handle complicated geometries. In this study, finite element modeling developed under Adina-T software utilizing principle of virtual temperatures, which is the quadratic functional form of the variational formulation of the transient heat transfer equation, is used. Firstly, experimental setup established by A.E. for 95 mm2, 35 kV rated, PE insulated power cable is modeled in order to verify the finite element modeling. Then, environmental conditions are modeled for both XLPE and PE insulated power cables. Usually, the semi-infinite region of the model is truncated by placing an artificial Dirichlet boundary far from the cable. The temperature on this boundary is accepted to be equal to the ambient temperature. If the distance of the boundary is selected very far away from the cable, truncation error could be minimized, however this time finite element modeling errors may increase due to finite elements dimensional aspect ratios; moreover the computational effort will also increase. This paper proposes a method to specify optimum boundary distance based on comparison of two boundary conditions for different boundary distance selections. After having obtained appropriate geometry and dimensions of the model, cable temperature dependency to variations in thermal properties of the surrounding medium and loading is investigated.
11.
HEAT CONDUCTION IN POWER CABLES

A . Governing Differential Equation The governing equation solved for heat distribution is, [41: dT V . ( k V T ) + q = p C> (1)
at

in which q is the internal heat generation, k is the thermal conductivity, p is the density of the material, c is the specific heat capacity of the material, T is the temperature, t is time, and pc (C) is the heat capacity coefficient. B. Boundary Conditions Whether heat transfer solution is implemented using finite element modeling or any other tool, a solvable heat conduction problem should also have boundary conditions (b.c.) well defined [5]. The b.c.s are very often of three common types: I . Dirichlet conditions, or b.c.s of thefirst kind T is specified on the boundary fort > 0.

905

2. Neumann conditions, or b.c.s of second kind: The derivative of T normal to the boundary is specified on the boundary for t > 0. Such a condition arises when the heat flux is specified on a boundary or when, with the help of insulation, we set heat flux or equally rate of temperature change with respect to the direction equal to zero. 3. Convection or b.c.s of the third kind A derivative of T in a direction normal to the boundary is proportional to the temperature difference with respect to the ambient on that boundary. Such a condition arises when convection occurs at a boundary, and it is expressed as :

For the particular problem considered in this study k is assumed to be temperature independent. The principle of virtual temperatures expresses the heat flow equilibrium at all times of interest. It is aimed to develop incremental equilibrium equations. First it is assumed that the conditions at time t have been calculated and that the temperatures are to be determined for time t + At, where At is time increment.

Iv.

EXPERIMENTAL STUDIES

In (2), h denotes the convective heat transfer coefjcient where T, denotes ambient temperature.
11 1.
METHOD OF FINITE ELEMENTS

The finite element method, applied in this paper to calculate detailed and accurate schemes of the cable temperatures under various circumstances, is a powerful computational technique that is widely used at present for the solution of differential and integral equations arising in virtually every field of engineering and applied science. As far as the solution of continuous system mathematical models are considered the finite element method is a generalization of the classical variational and weighted residual methods which are based on the idea that the solution of a differential equation can be represented as a linear combination of unknown parameters and appropriately selected functions in the entire domain of the problem [SI. Finite element methods divide the physical region into sub regions or fmite elements (Most commonly triangle). Temperature variable is approximated in functional form over each element and over whole the domain. The parameters of this approximation as a result become the unknowns of the problem. Substitution of the temperature approximations into the parabolic formulation of the governing differential equation and imposing the relevant boundary conditions yields a set of algebraic equations.

A . Experimental Setup According to the experimental set up established by A.E., 35 kV, 95 mm2, PE insulated power cable was placed into a specific wooden box full of dry soil having dimensions of 50x40~100 (WxHxD) in the laboratory. cm Then cable is subjected to various loads to obtain the thermal transients inside the cable and at the surrounding of the cable. Sensitive thermocouples were used to record the temperatures of specific points of the cable. This geometry forms the basis geometry that will be used initially to implement finite element solution and, of course, to correlate the application results obtained in this study using the finite element solution method explained above.

A . Formulation of Governing Differential Equation


Governing differential equation of heat transfer is formulated by using principles of virtual temperatures which is a form of variational formulation. Functional of (1) can be written as follows, [6]:

B. Modeling of Experimental Setup Experimental set up illustrated above is divided into 780 elements by defining 557 nodes using a rule based mesh algorithm to obtain a viable mesh which represents the computational domain and boundary conditions accurately. It is important that the mesh should not contain elements with very large aspect ratios (i.e. ratio of the largest side to the smallest side of an element), especially in regions of large gradients. Then, material models are attained to the geometry as illustrated in Table I. Following, boundary conditions are determined for the experimental setup. As three boundaries of four are adjacent to the ambient environment of the domain given, these boundaries can be of kind 3, convective boundary. For the 4th boundary, adjacent to the base, boundary condition of first, constant temperature is taken. After completing the model of the experimental set up, the finite element program is executed. It is observed that temperature distribution results at different points such as center of conductor and cable surface (Points A and B in Fig. 1) almost exactly matched the experimental results as shown in Fig. 2. When Fig. 2 is inspected, it can be understood that temperature values are far away from saturation points for a duration of 4 hours. Due to the
Air

- Iq,Tds+ I h i f T-

(3)
40

cm

In (3) qn is a surface heat flow input over a very small area, q is the internal heat generation per unit volume inside the cable and, qc, the rate of heat absorbed by the elements is: -dT q- =c-. dt 906

Air

I
Base

-,-->PointB
(Shield Ins. Surface)

-- --

Air

-5Point A
(Cable Core)

Main Ins.

Figure 1. Experimental Setup

50

Center
40

Shce

I
00 .
0.5
1,O

1.5

2,0

2,s

3,0

3.5

40 .

4,s

Tim 0

Figure 2. Cable surface and center of the conductor temperature versus time under 300 A load, dry soil conditions. TABLE I. MATERIAL MODELS FOR EXPERIMENTAL WORK Air Core Ins.
Dry Soil

I I 1

Material Models
Constant h, h = 5.234 W/m2.K, 20 'C PE, c: 2.302 kJ/kg.K, k: 0.335 W/m.K, Isotropic c: 0.837 kJ/kg.K, k: 0.268 W/m.K, Isotropic, 20 ' C PVC. c: 2.302 kJ/kg.K.

I
I

Shield

experimental facilities available, experiments were only carried out for 4 hours (Difficulties such as to keep environment temperature constant etc.). So, time step is increased in the program to reach the saturation point of the temperatures. It is observed that it may take several days for the cable to reach a stable temperature value. For instance, the temperature value of the center of conductor after 10 days of operation is found to be 67.86 "C, approximately 18 "C higher than the value reached after 4 hours of operation under 300 A loading. Modeling and experimental results under different loadings are shown in Fig. 3. However, if cyclic daily variations of ambient temperature are taken into account saturation temperature figures will obviously change.

180 160 140 120 100


80 60 40 20

C. Solution o PE Insulated Cable Buried in Ground f In the analysis carried above, the cable was considered to be in a box to resemble the experimental conditions. In other words boundaries located at right and lei? hand side of the domain were convective boundaries as they were surrounded by air as shown in Fig. 1. What kind of an analysis should be carried out if these boundaries were surrounded by soil as the boundary located at the bottom of the domain? There are two possibilities for these three boundaries: First, b.c. of the first kind (Dirichlet), T is specified on the boundary. Depending on the boundary distance selection, an error may arise as it is expected that heat generated by the cable does not change the temperature of the soil after a specific distance. However, with the selection of correct boundary distances, the temperature distributions will be closer to the actual distribution. Second, b.c.'s of second kind, (Neumann) the derivative of T normal to the boundary (heat flux) is specified on the boundary of the domain. However, heat flux value can not be easily specified on the boundaries. On the other hand, with the selection of correct boundary distance, zero flux can be assigned to the boundaries (Boundary is assumed to be insulated.), and, temperature distribution will reflect the actual distribution. As it can be understood from the above explanations, both boundary conditions are not actually contradicting each other when proper boundary distances are selected. So, to minimize the error arising from the selection of boundary conditions we should choose such a distance that temperature distribution found by applying both of the boundary conditions will be closer. To do so, analysis were executed with different boundary distances. Table 2, illustrates the saturation values of conductor temperatures for different boundary distances, and 160 cm seems as an appropriate distance choice. When the domain is enlarged, meshing should be modified accordingly. In other words, instead of enlarging element dimensions of the mesh, number of nodes should be increased. So, aspect ratios between the elements inside the cable and soil will be preserved. However, increasing the number of nodes increases the computational effort. It should be also noted that soil thermal conductivity is taken higher to be in accord with the experimental studies compared to rated thermal properties given by cable manufacturers. Cable manufacturers give correction factors to correlate rated current for different soil thermal conductivities which will be considered in the following section.

D. Comparison ofXLPE and PE Insulations


XLPE insulated, single phase 1x95 mm2, 35 kV rated cable is modeled following the same steps in section IV.B,
TABLE 11. TEMPERATURE VERSUS BOUNDARY DISTANCE
250

0
200

300

350

400

450

500

Load (A)
~

XLPE 10D -PE

10D

- t -

PE 4HR

Figure 3. Maximum center of conductor temperature distributions after 10 days and 4 hours of operation for XLPE and PE insulated cables

907

modeling of experimental conditions. Differences of XLPE insulated cable from the PE one having the same core cross sectional area are its overall diameter which is less then PE cable and its current carrying capacity (363 A) which is higher then PE insulated cables (335 A). The reason of lower diameter of XLPE cable is due to its thinner insulation thickness achieved by means of a different manufacturing technology. XLPE cable is loaded with 300 A as did PE cable previously, while center conductor of PE cable reaches to 67.86 T , XLPE cable reaches to 62.87 C only which indicates that XLPE insulated cables can have more current ratings although XLPE and PE has the same thermal properties. E. Solution ofXLPE Cable for Unity Correction Up to this point of the study, thermal analyses of the cables were carried out under the conditions specified in the experimental set up and reasonable solutions were achieved by the methods in this study. Further analysis can be made to consider other conditions. First of all normal operating conditions accepted by the cable manufacturers is supplied and analysis is done accordingly. Normal conditions mean the conditions for which correction factor of rated current carrying capacity is equal to unity. Normal Conditions for XLPE cable given by the manufacturer Pirelli [7] is as follows: Soil Temperature: 20 T , Air Temperature: 30 C, Thermal Conductivity of Soil, k: 1 W/mK. For these conditions the manufacturer is proposing a rated current of 363 A which corresponds to a unity correction factor. Also note that, unity correction factor corresponds to unity thermal conductivity in W/mK. Considering the above parameters, the model is regenerated using the boundary conditions and domain proposed in section 1V.C and center conductor temperature is found to be 56.09 C against 62.87 C found for the experimental conditions in section 1V.D. This is imposing a large safety margin compared to the permissible operating temperature, 90 C. F. Soil Effect on Correction Factor Cable manufacturers give correction factors for different soil thermal conductivity values. Correction factors of Pirelli for XLPE Cables with respect to thermal conductivity of soil are given in [7]. Keeping these correction factors in mind, XLPE cable is analyzed applying normal conditions except varying soil thermal conductivity and corrected current load. Table 3 lists maximum center conductor temperatures for different soil thermal conductivities under fixed current load (Rated current) while Table 4 gives temperature values under correlated current load for the concerning thermal conductivity. Application of correction factors results temperature distributions similar to the normal operating condition distribution. Correction factors given by manufacturers depend on either experimental studies or IEC Standard 287 (Calculation of the Continuous Current Rating of Cables). Correction factors are also specified for other environmental parameters such as air and soil temperature which can be also investigated by the method proposed in this study.

Thermal Conductivity of Soil. W/m.K

Thermal Resistivity of

Center Conductor Temperature (C)

Thermal Conductivity of Soil, W/m.K 0.4 0.67


1.o

Corrected Loading

1.43

0.81 x 363 A 0.9 x 363 A 1.0 x 363 A 1.12 363 A

Center Conductor Temperature (C) Results 57.14 59.70 56.09 60.08

v.

CONCLUSIONS

In this study, thermal behavior of underground 35 kV power cable is investigated both experimentally and numerically by using finite element method and it is understood that rated current for the cable defined by the manufacturers are given in a very safe margin. It is clear that the use of Dirichlet or Neumann boundary conditions in soil during modeling of the underground power cable will obviously cause a truncation error as the solution area is limited and approximated; however the temperature distribution is shown to become closer to actual distribution when boundary distances are properly chosen. The effect of critical parameters such as soil conductivity and loading are also investigated. It is observed that, obtained results are matching with the manufacturers values by the use of current correction factors. Although Finite Element Method can never replace the experimental studies, it may well reduce the experimental work as depicted in this study.

REFERENCES
G. J. Anders, and H.S. Radhakrishna, Power cable thermal analysis with consideration of heat and moisture transfer in the soil, IEEE Transactions on Power Delivery, vol. 3 , pp. 12801285, October 1988. N. Flatabo, Transient heat conduction problem in power cables solved by the finite element method, IEEE Transactions on Power Apparatus and Systems, vol. 92, 1973, pp. 161-168. G. Gela, and J. J. Dai, Calculation of thermal fields of underground cables using the boundary element method, IEEE Transactions on Power Delivery, vol. 3, pp. 1341-1347, October 1988. J. H. Lienhard, Heat Transfer Textbook. Cambridge, MA: Phlogiston Press, 2002. J. N. Reddy, FEM in Heat Transfer and Fluid Dynamics. New York: McGraw Hill, 1994. K. J. Bathe, Finite Element Procedures. Englewood Cliffs, NJ: Prentice Hall, 1996. Turk Pirelli Cable & Systems, Insulated Wires and Power Cables, TPKSUUO-2. Bursa: Pirelli, 2000.

908

Das könnte Ihnen auch gefallen