You are on page 1of 11

4

IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 21, NO. 1, JANUARY 2006

Nonlinear-Coupled Electric-Thermal Modeling of Underground Cable Systems


Nik sa Kova c, Ivan Saraj cev, and Dragan Poljak, Member, IEEE
AbstractAn original nonlinear-coupled electric-thermal model of underground cables with the solid sheaths is proposed. The model deals with the numerical evaluation of losses, heating, and ampacity. The computation of the current dependent losses is undertaken by means of the lament method, where conductors and sheaths are represented by a number of smaller subconductors or laments. Furthermore, heat-transfer phenomena through an innite domain beneath the soil surface are modeled combining the nite and the mapped innite elements, respectively. The corresponding nite-element meshes are generated by the advancing front method. The numerical results presented throughout this work suggest that the International Electrotechnical Commission relation concerning the external thermal resistance for touching cables, placed in at formation, having appreciable sheath losses, should be re-examined. Index TermsAmpacity, numerical modeling, underground cables.

I. INTRODUCTION

OSSES, heating, and ampacity are unavoidable parameters in underground cable design depending on cable materials, laying condition of the cable system, thermal properties of the media, bonding arrangement, etc. Generally, numerical methods provide more accurate modeling of underground cable systems than purely analytical or analytical/numerical techniques. However, numerical methods are sometimes too complex to be handled by engineers. Some of the papers dealing with a numerical approach to the analysis of underground cable losses [1][4], external or internal thermal resistances [5][10], as well as heating and/or ampacity calculations [11][30] are highlighted in this work. On the other hand, rare are the papers dealing with the complete numerical approach for the evaluation of all quantities of interest (i.e., losses, heating, and ampacity). Thus, losses have often been assumed to be known values, or computed by means of International Electrotechnical Commission (IEC) or NeherMcGrath relations, while heat-transfer phenomena have been handled via numerical methods using losses as input data [11][17], [19], [21][23], [27][30].
Manuscript received April 7, 2004; revised September 4, 2004. Paper no. TPWRD-00173-2004. N. Kova c and I. Saraj cev are with the Department of Electrical Engineering, Faculty of Electrical Engineering, Mechanical Engineering and Naval Architecture, Split University, Split 21000, Croatia (e-mail: nkovac@fesb.hr; isarajcev@fesb.hr). D. Poljak is with the Department of Electronics, Faculty of Electrical Engineering, Mechanical Engineering and Naval Architecture, Split University, Split 21000, Croatia (e-mail: dpoljak@fesb.hr). Digital Object Identier 10.1109/TPWRD.2005.852272

Hwang [18] has proposed the nite-element modeling of both losses and heating for underground cable systems. The numerical computation of these parameters always requires considerably high computational effort. In [18], the cable cross-section has been approximated by an octagon. Also, the shield and dielectric losses have been incorporated into the model as heat sources distributed at the cable surface. However, this simplied modeling of the cable cross-section is not quite satisfactory, since the temperature drop between conductor and cable surface could be signicant. Moreover, by posing the Dirichlets boundary conditions at a certain distance from the cable structure to solve the heat conduction equation, an innite domain beneath the soil surface has been truncated. The truncation of the domain size represents a compromise between opposite requirements: the higher accuracy of numerical results and the lower computational cost. It has been shown in [27] that the cable domain temperatures have been inuenced by the truncation, even in the case when the truncated domain boundaries have been placed quite far away from the cables. The more accurate treatment of the innite domain containing underground cables can be carried out by the integral or integro-differential equation formulation. Thus, the application of the boundary-element method (BEM) and the coupled nite/boundary-element approach (FEM/BEM) have been reported in [28], [29], and [30], respectively. The present work deals with an original nonlinear coupled electric-thermal model of underground cables with the solid sheaths. The proposed model provides numerical evaluation of losses, heating, and ampacity. The computation of current dependent losses is performed by means of the lament method [4], [31], [32], where conductors and sheaths are replaced by a number of smaller subconductors or laments. Furthermore, heat-transfer phenomena through an innite domain beneath the soil surface are modeled combining the nite elements and the mapped innite elements [33], [34]. In particular, the nite-element meshes are generated by the advancing front method [35]. The accurate thermal modeling of both cable cross-sections and surroundings is performed by the second-order isoparametric elements. Some illustrative computational results, presented throughout this work, clearly demonstrate the efciency of the proposed approach.

II. NONLINEAR-COUPLED ELECTRIC-THERMAL MODEL The electric-thermal model is outlined in a few steps: 1) losses evaluation; 2) heating evaluation; 3) electric-thermal coupling; and 4) ampacity evaluation.

0885-8977/$20.00 2006 IEEE

et al.: NONLINEAR-COUPLED ELECTRIC-THERMAL MODELING OF UNDERGROUND CABLE SYSTEMS KOVAC

where

. . .

. . .

. . .

Fig. 1.

Representation of an electrical section of a cable transmission line.

A. Losses Evaluation In order to determine heating of underground cable system, the system losses have to be known. The current dependent losses of cables with the solid sheaths are computed by means of the lament method [4], [31], [32]. The calculation of the dielectric losses is considered as a rather straightforward task, and details can be found elsewhere [31]. Sheaths are predominantly made of aluminum, lead, or lead alloy. The lament method provides conductors and sheaths to be represented by a number of smaller subconductors or laments, sufciently small to assume the uniform current density. Moreover, the governing equations are formulated using the additional assumptions: 1) cables are arranged in parallel, and 2) cable line is longitudinally homogenous. The skin and proximity effect are taken into account by the lament method, as well. An electrical section of a transmission line composed of three single-core cables is considered, as shown in Fig. 1. The line is assumed to be a part of an earthed circuit. The sheaths are solidly bonded and earthed at both ends. The conand laments, reductors and sheaths are divided into spectively. Hence, the total number of laments is . The earthed circuit can be replaced by a balanced oper, as well as a balanced load ating voltage system impedance system (Fig. 1). This simplied circuit modeling is satisfactory for the losses evaluation, since the systems are used only to set up the corresponding owing through the laments. currents The lament currents are calculated by the mesh-current method. Each mesh is represented by a loop consisting of: 1) the associated conductor or sheath lament; 2) both the associated voltage and the load impedance phase, if the loop contains the conductor lament; and 3) the ground return path. The corresponding matrix equation can be written as follows: (1) where vector of the loop voltages; matrix of self and mutual impedances of the loops; vector of the lament currents. Equation (1) can be written in a more convenient matrix form

are vectors of phase voltages concerning the loops with stand for the conductor laments; self and mutual impedance submatrices of the loops with the laments of the conductors , and , respectively; denote mutual impedance submatrices between the loops with the laments of the different is mutual impedance submatrix conductors; between the loops with the conductor laments and those stands for containing the sheath laments; self and mutual impedance submatrix of the loops with the sheath laments. are given by Elements of the submatrices (3) where mutual impedance between th and th lament in the presence of the ground return path; self impedance of the lament in the presence of the ground return path. The mathematical details regarding the assessment of the imand , as well as their testing procedure, can be pedances found in the Appendix. The elements of the other impedance submatrices are (4) The vector of the lament currents can be obtained by simply inverting the matrix in (2). If the cables are a part of a circuit with the insulated neutrals, between these points, providing there is a voltage difference the following condition equation is to be posed: (5) The resulting matrix equation then can be written as follows: . . . . . .

. . .

(6) The matrix equation (6) contains equations with unknowns (i.e., lament currents and the voltage difference ). Solving the matrix (6), the unknown currents and voltage difference are obtained.

(2)

IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 21, NO. 1, JANUARY 2006

as well as the load vector The global conductivity matrix are assembled from each element

Fig. 2. Boundary conditions.

(13)

Knowing the lament currents, the losses of the conductors and sheaths are simply computed from the relations (7) (8) where laments of conductors; laments of sheaths; resistances of conductor and sheath laments, respectively. The resistance of each lament is computed on the basis of of its associated metallic partmp the average temperature (i.e., the conductor or sheath). B. Heating Evaluation Two-dimensional steady-state temperature distribution in the innite domain beneath the soil surface, generated by underground cables (assuming negligible moisture migration in the cables vicinity) is governed by the heat conduction equation (9) where is the thermal conductivity coefcient and is the heat generated per-unit time and volume. The following boundary conditions [40], associated with the heat conduction equation, are: on on where temperature at innity; portion of the domain boundary at innity; heat ux on the ground surface; portion of the domain boundary with the heat ux ; convective heat-transfer coefcient by which the average radiation is included [14], [18]; air temperature, as shown in Fig. 2. Using the weighted residual approach and applying the GalerkinBubnov scheme of the nite-element method, the following matrix equation is obtained: (12) where global conductivity matrix; vector of the unknown values of the nodal temperatures; global load vector. where total losses of a particular conductor or sheath con; taining element cross-section of a particular conductor or sheath con. taining element (10) (11) where (14) where total number of elements; area of an element; number of nodes assigned to any element; vector of the shape functions of an element; heat generated within an element area per unit time and volume; portion of the boundary of which lies on . Heat sources associated with the current dependent losses are conductors and sheaths (i.e., the metallic partsmp). As they are not placed alongside a domain boundary , the last term in (14) concerning the load vector of the metallic parts equals zero. The heat generated per-unit time and volume within an element of the cable metallic part is determined by (15) denotes the losses generated within element , is the area of element , and stands for the length of a cable section. Due to the rather high thermal conductivity values of the metallic parts, the uniform losses distribution can be assumed over a particular conductor or sheath through the heating evaluation. On the other hand, the nonuniform distribution is taken into account through the losses evaluation by means of the lament method, since the nonuniformity can affect the total value of a particular sheath losses and, consequently, the cable heating. Therefore, it follows: (16)

et al.: NONLINEAR-COUPLED ELECTRIC-THERMAL MODELING OF UNDERGROUND CABLE SYSTEMS KOVAC

C. Electric-Thermal Coupling The nonlinear behavior of the electric-thermal model is caused by an electrical conductivity of a material, which depends on the corresponding temperature value. The iterative procedure of the temperature matching starts with the evaluation of losses according to Section A, by using the arbitrarily chosen conductor and sheath temperatures as well . Assembling the global equation as the load impedance system (12) yields the rst iteration of nodal temperatures , where denotes the total number of nodes. Subsequently, the average temperature of a particular conductor and sheath is calculated by the following formula:

Fig. 3.

Two-dimensional mapping.

Hence, the load vector of the metallic part element can be written in the form (17)

(22) The innite domain beneath the ground surface is treated via the mapped innite elements [33], [34]. Applying the method of images, the heat transfer of underground cables can be considered as the dipole source-type problem. Two-dimensional (2-D) dipole-source-type problems have been solved very accurately using the mapped elements [33]. Consequently, the application of the mapped innite elements is expected to be well suited for the treatment of underground cables. Furthermore, if the nite/innite approach is applied, one can retain the differential equation formulation (9). Since the detailed theoretical background of the mapped innite elements can be found in [33], for the sake of brevity, only the basic concepts are given in this work. The algorithm is based on a simple mapping of the global innite element into the local nite element. The mapping of a 2-D quadratic innite element, Fig. 3, can be written as where

. . . and is the total number of elements associated with a particstands for the total number ular conductor or sheath, while of conductors and sheaths. Using thus obtained average temperatures, the losses are computed again. The procedure goes on repeatedly as long as the prescribed permissible temperature discrepancy through successive iterations is achieved. It is to be mentioned that the noncoupled model incorporates the computation of the conductor and sheath resistances on the basis of the arbitrarily assumed temperatures, thus causing the non-negligible current and losses calculation errors. For example, the discrepancy of 15 C between the assumed and actual temperature of aluminum conductor results in the corresponding resistance error of around 5%. D. Ampacity Evaluation If the ampacity evaluation is of interest, as well, a new external iterative procedure of the load impedance matching is required. As a matter of fact, it is necessary to determine the and, hence, the corresponding currents, which impedance give the temperature rise to cable insulation up to the permissible value . Cable losses are not affected by the load power , for the constant absolute value of . Thus, the factor problem can be reduced to the assessment of the load resistance only. The load resistance in the th iteration is obtained using the linear interpolation of the maximum temperature reand sults obtained for cable insulation within the iterations (Fig. 4). The external iterative procedure ends up when the prescribed and is achieved. permissible discrepancy between The complete internal iterative procedure of the temperature matching is performed through each external iteration. The

(18)

(19) where global coordinates; local coordinates; global nodes coordinates of an innite element; mapping functions; standard Lagrange shape functions. The mapping functions are given in the form (20) (21) The mapped elements retain the nite-element integration weights and abscissae and shape functions as well.

IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 21, NO. 1, JANUARY 2006

Fig. 4. Assessment of the load resistance.

Fig. 7. Fig. 5. Cable system domain.

Mesh of the cable cross-section.

Fig. 6. Boundary between the nite and innite elements.

ampacity equals the conductor currents owing in accordance evaluated in the last iteration. to the load resistance value III. MESH GENERATION Losses, heating, and ampacity are evaluated for the underground cable system in a at arrangement. Three 35-kV, direct-buried 400 mm copper conductor cables having impregnated paper insulation and aluminum sheath are shown in Fig. 5. An automatic mesh generation, as a part of the nite-element approach, is an important step for the accuracy of solution and reduction of computational time. The advancing front method [35] has been widely used for the meshing of arbitrarily shape domains. This meshing procedure starts with the polygonal [polyhedral in three-dimensional (3-D)] discretization of material boundaries. The only input data required for the mesh generation are the boundary nodes arising from the discretization. The complete mesh generation algorithm has been promoted in [35]. The soil meshing is performed using a software package based on the advancing front method. Additional software is developed to mesh the cable cross-section, consisting of the specic shape materials, such as cylindrical conductor and tubular insulation, sheath, and covering. The boundary between the nite and innite elements is chosen in accordance to Fig. 6 to avoid the high-temperature gradients inside the innite elements. The nite-element mesh of the cable cross-section is shown in Fig. 7. The curved edges of the six-noded elements are somewhere drawn by two lengths connecting the corresponding nodes due to the graph determination with less difculty. The

(a)

(b) Fig. 8. (a) Mesh of the surrounding soil, cable spacing s = 0 mm. (b) An enlarged section in the cable vicinity, cable spacing s = 0 mm.

surrounding soil meshes concerning the different cable spacmm; double brick thickness ings (touching 140 mm) are shown in Figs. 8 and 9, respectively. IV. NUMERICAL RESULTS The cable system, Fig. 5, having 1.5-mm-thick aluminum sheaths bonded at both ends of an electrical section, is assumed to be a part of a 35-kV circuit with the insulated neutrals. The stranded 400 mm copper conductor consists of ve layers replaments. The layers are composed of resenting wires with the diameter mm. laments, as it is visible The sheath is divided into from Fig. 10. Hence, the total number of laments related to the . In general, one deals with a three-phase system is

et al.: NONLINEAR-COUPLED ELECTRIC-THERMAL MODELING OF UNDERGROUND CABLE SYSTEMS KOVAC

TABLE I AMPACITY COMPARISON

(a)

(b) Fig. 9. (a) Mesh of the surrounding soil, cable spacing s = 140 mm. (b) An enlarged section in the cable vicinity, cable spacing s = 140 mm.

Fig. 10.

Sheath laments.

signicantly lower number of unknowns than is required for the losses evaluation via the nite-element method [1]. The operating voltage system can be written as follows: kV, kV, kV. Furthermore, the elements , (3) and (4), of the submatrices associated with (2), are computed and of the using the mutual and self impedances laments with the ground return path. The impedances are determined by (A.2) and (A.7), presented in the Appendix. The geometric mean distances, incorporated in impedance expressions, are calculated via (A.8) and (A.9) concerning the conductor laments, while the distances of the sheath laments are computed using (A.10)(A.12). computed using the coupled nonThe ampacity results linear electric-thermal model are compared with the results obtained via the IEC relations [41]. The relations are derived

in accordance with the assumption of the isothermal temperature value of the actually convective and radiative soil surface. Hence, the results calculated via the coupled model are obtained using the same assumption. Otherwise, the discrepancy between the results would be affected by the different boundary condition. The temperature values at both the soil surface and innity, respectively, are chosen to be 20 C. In general, the convective and radiative heat transfer through the soil surface can be taken into account by the coupled model via (11). The dielecW/m are calculated in a straightforward tric losses manner and included in the thermal calculations. It is worth mentioning that the effective cross-section area mm (i.e., slightly of the conductor considered is differs from the value of 400 mm covered by IEC 228, where the maximum dc resistances of stranded conductors at 20 C are given). Therefore, the following equation [31] is used:

where 1.02 is the empirical factor taking the strands into account, while is the electrical resistivity of conductors. The further IEC procedure for the ampacity evaluation follows the standard one. The ampacity values related to the various cable spacings and soil conductivities are shown in Table I. The negligible , and , obtained differences between the phase currents by the model, arise from the unequal temperatures of the phase conductors as well as the mutual electromagnetic (EM) coupling. The average phase current is assigned to as the ampacity . value

10

IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 21, NO. 1, JANUARY 2006

TABLE II LOSSES COMPARISON

Fig. 11.

Sheath losses distribution s = 0 mm, k = 1 W/m C.

When single-core cables with appreciable sheath losses are installed in at formation, the losses nonlinearly increase together with the cable spacing , providing the greater generated heat per-unit time and volume. The whole procedure of the sheath losses evaluation can be found in [31], [42]. On the other hand, the external thermal resistance decreases for the greater spacing [31]. Hence, the optimal spacing, for which the maximum ampacity is obtained, stands in a balance between these effects. Since the actual cable has a rather thick aluminum sheath and, thus, the low value of the associated resistance, a very rapid increase in the sheath losses is achieved for mm and mm, compared with the spacing of mm. Consequently, the greatest ampacity value, taking into account the spacings shown in Table I, refers to the cable touching. If the cables are touching each other, the discrepancy between the ampacity results evaluated via different approaches %. On the other hand, if the cables are is approximately spaced, very close agreement is achieved. The analysis of the ampacity discrepancy for the touching cables starts up with the comparison of the system losses obtained via the coupled model and IEC relations (Table II). The , IEC losses are calculated for the coupled model currents shown in Table I, since the losses comparison should be based on the same conductor current through both of the approaches. The phase and sheath losses computed by the coupled model are and . The conductor and sheath losses obtained via the IEC relaand . Obviously, tions are the very close agreement is achieved for the spaced cables. The discrepancy of the sheath losses for the touching cables (particularly for the middle onephase ) arises from the nonuniform distribution of a particular sheath losses (Fig. 11), which is not taken into account by the IEC relations. Such a distribution originates from the proximity effect. Particular sheath losses for the spaced cables do not possess such a nonuniformity (Figs. 12 and 13). Also, the discrepancy arises from the linkage simplications undertaken in the evaluation of the inductances via the IEC relations, according to the approach presented by Arnold [42]. In order to eliminate the inuence of the losses discrepancy to the value of the ampacity discrepancy , the losses for the touching cables, obtained by the proposed model, are used

Fig. 12. Sheath losses distribution s = 70 mm, k = 1 W/m C.

Fig. 13.

Sheath losses distribution s = 140 mm, k = 1 W/m C.

within the IEC procedure of the ampacity evaluation. The novel , as well as the discrepancy value are ampacity value is around %. shown in Table III. It can be noticed that Consequently, the use of the same losses evaluation procedure through both of the approaches decreases the ampacity discrepancy of approximately 3%. The rest of the ampacity discrepancy originates from the IEC formula of the external thermal resistance for touching cables in at formation, based on Symms paper [43], since the computation of the internal resistances for the single-core cables is rather straightforward. The integral equation method assuming the isothermal cable surfaces has been used in [43]. The numerical results presented in Table III show that the IEC relation should be re-examined for touching cables with unequal appreciable sheath losses. The temperature distribution of the cable

et al.: NONLINEAR-COUPLED ELECTRIC-THERMAL MODELING OF UNDERGROUND CABLE SYSTEMS KOVAC

11

TABLE III AMPACITY COMPARISON FOR THE TOUCHING CABLES (s = 0 mm)THE LOSSES COMPUTED BY THE MODEL ARE USED WITHIN THE IEC EVALUATION PROCEDURE OF THE AMPACITY I

Fig. 15. Temperature values through the cable cross section, phase B, s = 0 mm, k = 0:5 W/m C, # = 0 .

Fig. 16. Temperature values through the cable cross section, phase B, s = 0 mm, k = 1:5 W/m C, # = 0 .

Fig. 14. Temperature distribution of the cable surfaces s = 0 mm, k = 1 W/m C. TABLE IV AMPACITY COMPARISON FOR THE TOUCHING CABLES (s = 0 mm)BOTH THE LOSSES COMPUTED BY THE MODEL AND VAN GEERTRUYDENS FORMULA EVALUATION ARE USED WITHIN THE IEC PROCEDURE OF THE AMPACITY I

of external thermal resistance than the relation accepted in IEC standards, as can be noticed in Table IV. The ampacity results , obtained using both the losses computed by the model and Van Geertruydens formula within the IEC procedure of the ampacity evaluation, are compared with the coupled model results . The absolute value of the discrepancy is below 3%. Upon the whole, it is worth underlining how important the accurate modeling of the cable cross-section is, as can be seen in Figs. 15 and 16. The temperature drops between conductor and cable surface for the touching cables are around 6 C for W/m C as far as 14 C for W/m C. V. CONCLUDING REMARKS This paper proposes an original nonlinear coupled electricthermal model of underground cables with the solid sheaths, thus providing the numerical evaluation of losses, heating, and ampacity. The model is based on the lament method as well as the FEM, including the innite domain modeling carried out by the mapped innite elements. The corresponding niteelement meshes are generated by the advancing front method. The principal contributions of the approach presented so far are as follows. 1) The cable systems, which cannot be handled via simplied analytical or empirical equations, can be treated by the lament/FE model. Moreover, the proposed model shows a certain advantage compared to the fully FE-based model, since the use of the lament method for the losses evaluation requires rather lower computational cost than the FEM. 2) The nite/innite element approach provides the differential equation formulation of the thermal problem, contrary to, up to now used, more complex integral or in-

surfaces cannot be assumed as an isothermal one, as can be noticed in Fig. 14. For example, the temperature variation for the outer cables (phases and ) exceeds 5 C. Van Geertruyden [44] has also developed a relation for the external thermal resistance, based on the nite-element analysis by which the innite domain beneath the soil surface has been truncated. However, the cable temperatures are inuenced by the domain truncation, even in the case when the domain boundaries have been placed quite far away from the cables [27]. Hence, the appropriate relation taking into account the innite domain inuence should be developed. Nevertheless, the expression given in [44] is better suited for the computation

12

IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 21, NO. 1, JANUARY 2006

tegro-differential formulations featuring the boundary-element method (BEM) and coupled FEM/BEM approach, respectively. However, the thermal part of the model should be modied in accordance to the equations given in [24] and [25] to solve the problem of moisture migration in the cable vicinity. The numerical results presented throughout this work suggest that the IEC formulation concerning the external thermal resistance for touching cables, laid in at formation, having appreciable sheath losses, should be re-examined. Future work will deal with an extensive numerical analysis to develop a new equation related to the external thermal resistance for touching cables, by which the innite domain inuence will be taken into account. APPENDIX and are derived from The mutual and self impedances Carsons innite integral solution, where the ground inuence is taken into account [36]. Well-known simple closed-form approximations for the overhead wires in the low-frequency range are used in the present work. This range is relevant for the evaluation of underground cable losses. The mutual impedance per-unit length of two solid circular conductors in the presence of the ground return path is determined by [36] (A.1) where operating frequency; angular frequency; absolute permeability; distance between the conductor centers; electrical resistivity of the soil. If conductors are arbitrarily shaped, the mean geometric distance should be used. The mutual impedance of the laments with the ground return path, related to the underground cable system in Fig. 1, is written by (A.2) where length of an electrical section; geometric mean distance between the th and th lament. The self ground return impedance per-unit length is given by [36] (A.3) where is the conductor radius. The self impedance of the circular conductor with the ground return path is obtained by to (A.3) adding the internal impedance (A.4)

Fig. 17.

ith

lament associated with the conductor.

If the internal impedance is derived under dc conditions, it follows: (A.5) where is the resistance of the circular nonmagnetic conductor per-unit length. Hence, the self impedance can be written as (A.6) where is the self geometric mean distance of the solid circular conductor. Consequently, the self impedance of the th lament of an electrical section of the cable system with the ground return path (Fig. 1) is (A.7) where resistance of the th lament; self geometric mean distance of the th lament. It is to be mentioned that , concerning the conductor laments, is multiplied by the empirical factor 1.02, by which the strands are taken into account. A. Conductor Each layer of the stranded conductor is considered to be a associated with lament. The ith lament with the radius conductor is shown in Fig. 17. It consists of wires with the radius . According to [37], the self mean geometric distance can be approximated by (A.8) The mutual mean geometric distance between the th and th conductor lament, where the second is placed inside the th lament, is determined by (A.9) If the th lament is located outside the th one, the mutual mean geometric distance equals the distance between their centers. B. Sheath The nonuniform losses distribution of the solid sheath, arising from the proximity effect, can be taken into account using the sheath partitioning. Thus, the sheath is replaced by a number

et al.: NONLINEAR-COUPLED ELECTRIC-THERMAL MODELING OF UNDERGROUND CABLE SYSTEMS KOVAC

13

Fig. 22. Fig. 18. Sheath partioning.

Relative discrepancy in the resistive part of Z

;f

= 950 Hz.

Fig. 19.

Conguration with two underground conductors.

Fig. 23.

Relative discrepancy in the inductive part of Z

;f

= 950 Hz.

The mutual mean geometric distance between the sheath laments is approximated by (A.12) Although Carsons method is widely applied to the overhead lines, the underground cable analysis in this work is undertaken and due to their simby the approximate relations for plicity. In order to test the accuracy of the approximations for buried cables in the low-frequency range, the comparison to Saads closed-form solution is performed [39]. Saads approach to the impedances of underground cables seems to be the very accurate for frequencies up to 1 MHz. For the comparison purposes, the conguration with two underground conductors is considered as it is shown in Fig. 19. The relative discrepancies from the Saads solution in the resistive and inductive part of , for the fundamental frequency Hz the impedance are shown in Figs. 20 and 21, respectively. The cable distance varies from the 0.02 m (touching) to 1 m. The same data for the Hz) are presented in Figs. 22 and 23. 19th harmonic ( Moreover, the relative discrepancies in the resistive and inducHz are % and %, respective part of , for tively. The values corresponding to the 19th harmonic frequency % and %. are Therefore, the satisfactory agreement between the different approaches is achieved. It can be also concluded that the simple closed-form relations arising from Carsons integral solution can be utilized for the computation of underground cable losses. REFERENCES
[1] D. Labridis and P. Dokopoulos, Finite element computation of eld, losses and forces in a three-phase gas cable with nonsymmetrical conductor arrangement, IEEE Trans. Power Del., vol. 3, no. 4, pp. 13261333, Oct. 1988. [2] J. Kuang and S. Boggs, Pype-type cable losses for balanced and unbalanced currents, IEEE Trans. Power Del., vol. 17, no. 2, pp. 313317, Apr. 2002.

Fig. 20.

Relative discrepancy in the resistive part of Z

;f

= 50 Hz.

Fig. 21. Relative discrepancy in the inductive part of Z

;f

= 50 Hz.

of laments having cross-section shape similar to a rectangle (Fig. A2). The self geometric mean distance of the th sheath lament according to [38] can be written as (A.10) where and are assigned in Fig. 18. is given by (A.11) where distance between the th sheath lament and cable center; central angle of the th lament.

14

IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 21, NO. 1, JANUARY 2006

[3] A. Konrad, Integro-differential nite element formulation of two-dimensional steady-state skin effect problems, IEEE Trans. Magn., vol. MAG-18, no. 1, pp. 284292, Jan. 1982. , and I. Medic , Calculation of losses in power ev, M. Majstrovic [4] I. Sarajc cables as the base for cable temperature analysis, in Advanced Computational Methods in Heat Transfer VI. Southampton, U.K.: Wessex Inst. Technol., 2000. [5] G. J. Anders, A. K. T. Napieralski, and W. Zamojski, Calculation of internal thermal resistance and ampacity of 3-core unscreened cables with llers, IEEE Trans. Power Del., vol. 13, no. 3, pp. 699705, Jul. 1998. [6] G. J. Anders, A. Napieralski, and Z. Kulesza, Calculation of internal thermal resistance and ampacity of 3-core screened cables with llers, IEEE Trans. Power Del., vol. 14, no. 3, pp. 699705, Jul. 1999. [7] M. A. El-Kady, J. Motlis, G. A. Anders, and D. J. Horrocks, Modied values for geometric factor of external thermal resistance of cables in duct banks, IEEE Trans. Power Del., vol. 3, no. 4, pp. 13031309, Oct. 1988. [8] K. E. Saleeby, W. Z. Black, and J. G. Hartley, Effective thermal resistivity for power cables buried in thermal backll, IEEE Trans. Power App. Syst., vol. PAS-98, no. 6, pp. 22012214, 1979. [9] E. Tarasiewicz, M. A. El-Kady, and G. J. Anders, Generalized coefcients of external thermal resistance for ampacity evaluation of underground multiple cable systems, IEEE Trans. Power Del., vol. PWRD-2, no. 1, pp. 1520, Jan. 1987. [10] M. A. El-Kady and D. J. Horrocks, Extended values for geometric factor of external thermal resistance of cables in duct banks, IEEE Trans. Power App. Syst., vol. PAS-104, no. 1, pp. 19581962, Aug. 1985. [11] N. Flatabo, Transient heat conduction problems in power cables solved by nite element method, IEEE Trans. Power App. Syst., vol. PAS-92, no. 1, pp. 5663, Jan./Feb. 1973. [12] J. K. Mitchell and O. N. Abdel-Hadi, Temperature distribution around buried cable, IEEE Trans. Power App. Syst., vol. 98, no. 4, pp. 11581166, Jul/Aug. 1979. [13] M. A. Hanna, A. Y. Chikhani, and M. M. A. Salama, Thermal analysis of power cables in multilayered soil: Part 1: Theoretical model, IEEE Trans. Power Del., vol. 8, no. 3, pp. 761771, Jul. 1993. [14] , Thermal analysis of power cables in multilayered soil: Part 2: Practical considerations, IEEE Trans. Power Del., vol. 8, no. 3, pp. 772778, Jul. 1993. , Thermal analysis of power cables in multilayered soil: Part 3: [15] Case of two cables in a trench, IEEE Trans. Power Del., vol. 9, no. 1, pp. 572579, Jan. 1994. [16] , Thermal analysis of power cables in a trench in multilayered soil, IEEE Trans. Power Del., vol. 13, no. 2, pp. 304309, Apr. 1998. [17] W. Z. Black and S.-I. Park, Emergency ampacities of direct buried three phase underground cable systems, IEEE Trans. Power App. Syst., vol. 102, no. 7, pp. 21242132, Jul. 1983. [18] C. C. Hwang, Calculation of thermal elds of underground cable systems with consideration of structural steels constructed in a duct bank, Proc. Inst. Elect. Eng., Gen. Transm. Distrib., vol. 144, no. 6, pp. 541545, Nov. 1997. [19] D. Mushamalirwa, N. Germay, and J. C. Steffens, A 2-D nite element mesh generator for thermal analysis of underground power cables, IEEE Trans. Power Del., vol. 3, no. 1, pp. 6268, Jan. 1988. [20] G. J. Anders, M. Chaaban, N. Bedard, and R. W. D. Ganton, New approach to ampacity evaluation of cables in ducts using nite element technique, IEEE Trans. Power Del., vol. PWRD-2, no. 4, pp. 969975, Oct. 1987. [21] M. A. Kellow, A numerical procedure for the calculation of the temperature rise and ampacity of underground cables, IEEE Trans. Power App. Syst., vol. 100, no. 7, pp. 33223330, Jul. 1981. [22] M. A. El-Kady, Calculation of the sensitivity of power cable ampacity to variations of design and environmental parameters, IEEE Trans. Power App. Syst., vol. 103, no. 8, pp. 20432050, Aug. 1984. [23] M. Liang, An assessment of conductor temperature rises of cables, caused by a sudden application of short pulse in a daily cyclic load, IEEE Trans. Power Del., vol. 14, no. 2, pp. 307313, Apr. 1999. [24] G. J. Anders and H. S. Radhakrishna, Computation of temperature eld and moisture content in the vicinity of current carrying underground power cables, Proc. Inst. Elect. Eng. C, vol. 135, no. 1, pp. 5162, Jan. 1988. [25] , Power cable thermal analysis with consideration of heat and moisture transfer in the soil, IEEE Trans. Power Del., vol. 3, no. 4, pp. 12801288, Oct. 1988. , I. Sarajc ev, D. Poljak, and B. Jajac, Electromagnetic-thermal [26] N. Kovac model for power cables analysis, in Elect. Eng. Electromagn. VI. Southampton, U.K.: Wessex Inst. Technol., 2003, pp. 225234.

, B. Jajac, and D. Poljak, Domain optimization in FEM mod[27] N. Kovac eling of power cable heat transfer, in Advanced Computational Methods in Heat Transfer VI. Southampton, U.K.: Wessex Inst. Technol., 2000. [28] G. Gela and J. J. Dai, Calculation of thermal elds of underground cables using the boundary element method, IEEE Trans. Power Del., vol. 3, no. 4, pp. 13411347, Oct. 1988. [29] E. Tarasiewicz and J. Poltz, Mutually constrained partial differential and integral equations for an exterior eld problem, IEEE Trans. Magn., vol. MAG-19, no. 6, pp. 23072310, Nov. 1983. [30] E. Tarasiewicz, E. Kuffel, and S. Grzybowski, Calculation of temperature distributions within cable trench backll and the surrounding soil, IEEE Trans. Power App. Syst., vol. 104, no. 8, pp. 19731978, Sep. 1985. [31] G. J. Anders, Rating of Electric Power Cables. New York: IEEE Press, 1997. [32] P. de Arizon and H. W. Dommel, Computation of cable impedances based on subdivision of conductors, IEEE Trans. Power Del., vol. PWRD-2, no. 1, pp. 2127, Jan. 1987. [33] O. C. Zienkiewicz, C. Emson, and P. Bettess, A novel boundary innite element, Int. J. Numer. Meth. Eng., vol. 19, pp. 393404, 1983. and D. R. J. Owen, Mapped innite elements in transient [34] F. Damjanic thermal analysis, Comput. Structures, vol. 19, no. 4, pp. 673687, 1984. , S. Gotovac, and D. Poljak, A new front updating solution [35] N. Kovac applied to some engineering problems, Archiv. Comput. Meth. Eng., vol. 9, no. 1, pp. 4375, 2002. [36] J. R. Carson, Wave propagation in overhead wires, with ground rreturn, Bell Syst. Tech. J., vol. 5, pp. 539554, 1926. [37] B. Stefanini, Power Transmission (in (in Croatian)). Zagreb, Croatia: Zagreb Univ., 1966. [38] D. Oeding and K. Fesser, Geometric mean distances of rectangular conductors (in (in German)), ETZ-A, vol. 86, no. 16, pp. 525533, 1965. [39] O. Saad, G. Gaba, and M. Giroux, A closed-form approximation for ground return impedance of underground cables, IEEE Trans. Power Del., vol. 11, no. 3, pp. 15361545, Jul. 1996. [40] H. C. Huang and A. S. Usmani, Finite Element Analysis for Heat Transfer. Berlin, Germany: Springer-Verlag, 1994. [41] Electric CablesCalculation of the Current Rating, 1993/94. IEC, Publication 287. [42] A. H. M. Arnold, Theory of sheath losses in single-conductor leadcovered cables, J. Inst. Elect. Eng., no. 67, pp. 6989, 1929. [43] G. T. Symm, External thermal resistance of buried cables and troughs, in Proc. Inst. Elect. Eng., vol. 166, 1969, pp. 16961698. [44] A. Van Geertruyden, External Thermal Resistance of Three Buried Single-Core Cables in Flat and in Trefoil Formation, Laborelec Rep., DMO-RD-92-003/AVG, 1992.

a Kovac was born in Split, Croatia, on December 28, 1968. He received Niks the Ph.D. degree from the University of Split, Split, Croatia. Currently, he is an Assistant Professor of Faculty of Electrical Engineering, Mechanical Engineering, and Naval Architecture, University of Split. His research interests are numerical analysis related to the underground cables and thermal effects of human exposure to electromagnetic elds as well.

ev was born in Split, Croatia, on October 28, 1947. He received the Ivan Sarajc Ph.D. degree from the University of Zagreb, Zagreb, Croatia. Currently, he is Associate Professor of Faculty of Electrical Engineering, Mechanical Engineering, and Naval Architecture, University of Split. His primary interest is in the elds of power cables, overvoltage protection, and electromagnetic compatibility (EMC).

Dragan Poljak (M96) was born in Split, Croatia, on October 10, 1965. He received the Ph.D. degree from University of Split, Split, Croatia. Currently, he is Associate Professor of the Faculty of Electrical Engineering, Mechanical Engineering, and Naval Architecture, University of Split. His research interest is in computational methods in electromagnetics, particularly in the numerical modeling of wire antennas and related electromagnetic-compatibility (EMC) problems using both frequency- and time-domain techniques. He also deals with the numerical modeling applied to the environmental aspects of electromagnetic (EM) elds. He is a Series Editor of Advances in Electrical and Electronic Engineering, Wessex Institute of Technology (WIT) Press, Southampton, U.K. and was a Guest Editor of the International Journal of Engineering Analysis with Boundary Elements (EABE) Special Issue on Electromagnetics. He is author of ve books, published by WIT Press. He is a reviewer for IEEE TRANSCATIONS ON ELECTROMAGNETIC COMPATIBILITY. He has published ve papers in several IEEE TRANSACTIONS.

You might also like