Anda di halaman 1dari 7

IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 23, NO.

1, JANUARY 2008

221

Heating Effects of Short-Circuit Current Impulses on Contacts and ConductorsPart I


Lszl Koller, Balzs Novk, and Gyrgy Tevan
AbstractAn analytical description combined with nite-element calculations of the current displacement caused by short-circuit current impulses in 1-D and 2-D conductor models is presented. It is shown that in many cases, the Joule integral used in the technical practice gives a false representation of the thermal stress in the conductors and contacts. A simplied method is given for determining the degree of current displacement caused by different current impulses, hence, the nonuniformity of the losses and the heating within the conductors. Index TermsConductors, contacts, current distribution, shortcircuit currents.
Fig. 1. Opening-closing nger contact arrangement.

I. INTRODUCTION HE increasing demand for electric power and for its economical use entails the use of energy converters with higher power. This also means that the serial network elements (mainly the conductors and the contact units)used at high and low voltage, at operating frequency, and at substantially higher frequenciesmust withstand the thermal and dynamic effects of several hundred kiloamps operating and short-circuit currents. In switchgears, the conductors and contacts must have appropriate geometry and sufciently large cross sectional areas. On the other hand the economical installation and power transmission requires the possible smallest size that can be achieved. To minimize size different switchgear combinations, e.g., the metal enclosed gas insulated switchgears, and gas insulated bus bars have been developed. As a result, the size of the conductors and contacts has risen and the distance between them has become smaller and smaller. In energy converterslike induction furnaces or electric motorsthe high currents thermal effects and their dynamic forces are harnessed to change the properties of materials or to accomplish mechanical work. Through crucible-type induction furnaces the distribution of the eddy currents within the workpiece is highly relevant, and by such applications the primary goalalong with the technological processis the economical energy conversion [7][11]. The current distribution within the serial network elements of electrical equipment is nonuniform no matter if the current is harmonic or an impulse [3], [5]. In solid conductors the current distribution is formed by the interaction of the electromagnetic eld and the thermal eld. This interaction occurs because

Manuscript received October 9, 2006; revised December 12, 2006. Paper no. TPWRD-00624-2006. The authors are with the Department of Electric Power Engineering, Budapest University of Technology and Economics, Budapest 1111, Hungary (e-mail: koller@ntb.bme.hu; nbalazs@eik.bme.hu). Digital Object Identier 10.1109/TPWRD.2007.905806

the physical parameters of the conducting material and the heat transfer to the environment are temperature dependent. The electromagnetic effects of a harmonic current or a current impulse produce current displacement, which means nonuniform current-, thermal power-, and temperature-distribution. Because of these effects, not only the distribution within the conductors connected parallel (e.g. bus bars or fuse elements), but also the amount of current carried by each of these conductors will be uneven. The nonuniform temperature distribution modies the physical parameters of the conductors, which reacts on the current distribution. This nonuniformity can cause the overheatingpossibly the breaking downof the conductors and contacts: the conductors can lose their temper and they cannot withstand the mechanical stress (mainly the electrodynamical forces). The contacts, where the current is constrained to ow through a narrow path, are exposed to even higher thermal stress. An example of an opening-closing nger contact arrangement is shown in Fig. 1. If there are parallel connected contacts, the temperature of some parts can reach the melting-point of the contact material, and the contacts can bond together or evaporate. The risk of evaporation is the highest in case of lamella contacts. The main cause of the increased thermal stress is the current displacement. Its degree at a given current with given timefunction and at given conductor material is higher if the cross sectional area of the conductors is higher and the distance between them is smaller. It has been already mentioned that in switchgears there are large conductor sizes with the possible smallest distance between them, and therefore a current displacement of higher degree with its undesirable effects can be expected. These considerations are conrmed by the results of 2-D and 3-D nite-element calculations for harmonic currents in SF6 gas insulated circuit breakers [12]. The analyses proved that the nonuniform current distribution of the parallel connected contacts is affected mainly by the geometrical arrangement of the current conductors and thus the proximity effects of the electromagnetic eld.

0885-8977/$25.00 2007 IEEE

222

IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 23, NO. 1, JANUARY 2008

In [13] and [14], analytical calculations determining the transient current distribution in conductors were carried out. These methods can be used only for particular cases namely for single cylindrical conductors, and thus cannot be used to evaluate the effects on other conductor arrangements, especially on contacts. In [15] the heating of contacts under short-circuit currents were calculated but the current displacement and its effect on the contact resistance was not reckoned with. Our primary goal in this two part paper is to give a general approach to this problem by comparing the effects of different current impulses in simple models reecting the physical background of the phenomena. Based on our results it can be estimated when to use accurate analytical or nite-element calculations. To accomplish this task: degree of current displacement in the conductors (the space- and time-functions of current distribution); losses; space-function of dissipated energy density will be determined in circuits with inductive characteristic and current impulses with short duration. These results reect the increased and nonuniform thermal stresses within the contacts. An answer can also be obtained to the question, that whether the

Fig. 2. Time functions of the current impulses. The dashed line shows the prospective current, the continuous line shows the current impulse with t width: (a) sinusoidal impulse; (b) sinusoidal impulse with dc component; (c) impulse with current limitation; and (d) dc impulse.

(1) the short-circuit occurred at no-load condition. Three different type of ac and one type of dc impulses were analyzed (Fig. 2): solely sinusoidal impulse; sinusoidal impulse with a dc component generating the highest possible current peak; ac impulse occurring at current limitation; impulse occurring at a dc short-circuit, increasing with an exponential function. The length of the impulses is , which in the ac case can be given with an angle or in the dc case ratio. The impulses were determined for such with the electrical circuits, where the components (inductance, effective resistance) are constant. With this assumptionamong other thingsthe effect of the current displacement on the currents time-function is neglected. The real short-circuit inductance and resistance are much higher than the analyzed circuits. The impulse models reect two practical cases. 1) Regarding the main contacts of the circuit breakers: By the (a), (b) and (d) impulse models means the time of contact opening. This is the end of the thermal strain of the main contacts caused by the nonuniform current distribution among the asperities on the mating contact surfaces. the current begins to ow through the arcing conAt tacts, thus from this time on the current has no effect on the main contacts. 2) Regarding the conductors and all other contacts (especially the sliding contacts): When the arcing contacts open the time function of the current will be changed by the arc voltage. This change can be neglected in case of a high voltage ac short-circuit as the arc voltage is signicantly lower than the mains voltage. Practically the same current continues to ow after opening of the contacts, which breaks in a current zero (or near a current zero) by the (a)

Joule integral determining the thermal effect of a current imtime function and length can be used in the pulse with technical practice in the traditional way, which means that the value of the integral is given independently from the shape of the impulse, the geometrical parameters and the arrangement of the serial circuit elements. Thus the effect of the current displacement is neglected by considering only the dc resistance of the conductors. The effect of resistivity change is considered also with uniform current- and an approximately uniform temperature distribution. The results were obtained from analytical and nite-element calculations (and their combination) of the electromagnetic eld. In the calculations the interaction with the thermal eld was neglected, because the heat transfer to the environment and the temperature rise of the conductors with large cross sectional areas are negligible during the time of the short-circuit current impulses. In the 1-D and 2-D mathematical models a homogenous copper conductor with constant temperature and ) is assumed. The 2-D ( model calculations and the conclusions are presented in Part II of this paper. II. TIME-FUNCTIONS AND SPECTRAL AMPLITUDE DENSITIES OF THE SHORT-CIRCUIT CURRENT IMPULSES The heating effects of the short-circuit current impulses on conductors and contacts were examined for both ac and dc shortcircuits. It is assumed that the current starts to ow from 0 A in an inductive circuit with time constant. In the ac case the current is induced by a sinusoidal voltage with frequency. This means the switching on of a short-circuit, or that

KOLLER et al.: HEATING EFFECTS OF SHORT-CIRCUIT CURRENT IMPULSES ON CONTACTS AND CONDUCTORSPART I

223

and (b) impulse models . This can be the case also at low voltages if there is no current limitation and the arc voltage is low. The currents shape is completely different at low voltage in dc case or in ac case with current limitation, where the arc voltage is higher than the instantaneous value of the mains voltage. Fig. 2(c) shows an example of the latter case, where the sinusoidal current is modied by the arc appearing at the opening of the contacts at . The nal breaking of the current (the current zero) occurs at . The time-functions of the four impulse models are if otherwise if otherwise (2) (3)

is the shown in (4) and (5) at the bottom of the page, where peak value of the ac case in the steady-state range, and is the steady-state value of the dc case. The effect of the arc voltage in the low voltage ac case with current limitation (4) is represented by two sine curves, the frequencies of which depend on the time of contact opening. The peak value of the impulse occurs at . The parameter is the ratio of the impulses peak to the value . During occurring at the time of contact opening was assumed, and the maximum value the calculations of was considered where current limitation occurs, namely . where Both in the ac and dc cases the mains voltage was assumed to be the same, thus the effective value of the ac voltage was as. The time sumed to be the same as the dc voltage . This time constant of the circuits was equal too which is constant represents a power factor of an average of the typical values in high- and low-voltage ac circuits. Because of the same voltage and time constant (6) therefore the steady-state value of the dc current is , if the sinusoidal peak current is . It shall be noted here, that in case of the 1-D conductor models the extent of which in two or three directions is innite, in the shall be replaced with formulas (2)(6) and in Fig. 2. and with , where .

Fig. 3. Normalized spectral amplitude densities at different t times. (a) Sinusoidal impulse; (b) sinusoidal impulse with dc component; (c) impulse with current limitation; and (d) dc impulse.

The spectral amplitude densities of the current impulses can be calculated from the time-functions with the following formula [1], [2]: (7) The calculated spectral densities for some characteristic times are shown in Fig. 3. As it can be seen, at short current breaking times, namely when the impulse width is small, the amplitude of the high-frequency components is signicant. Below 5 ms impulse width there is no major difference between the solely sinusoidal (a) and the dc (d) case. Within this time the spectrum of the sinusoidal impulse with dc component (b) is much wider. The proportion of the high-frequency components is the lowest in the case of current limitation (c), as there is no such instant change in the time-function as by the other impulses. The steps of determining the spectral amplitude density of the solely sinusoidal impulse, and the result in a closed form can be seen in Part II, Appendix A1. For the other three signals, because of their complexity, the equations are not published here.

if if if otherwise if otherwise (5) (4)

224

IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 23, NO. 1, JANUARY 2008

It can be seen from the denition of the Fourier-spectrum that , namely with the conthe inner integral is equal with , that is . Furthermore jugate of

which is an even function. Therefore

Since
Fig. 4. One-dimensional models: (a) innite half-space and (b) innite plate conductor.

III. 1-D CONDUCTOR MODELS In the following section, the distribution of current-densities and energy-densities are calculated for the most general and simple 1-D model: the innite half-space conductor [Fig. 4(a)]. The current density distribution at different times and the distribution of the losses inside the conductor will be shown. A slightly more difcult but better representation of the problem is the innite conductor plate [Fig. 4(b)]. By this model, the losses can be calculated to include only the dc resistance, which cannot be dened for innite half-space. By both conductor models the total energy loss will be compared with the losses on mains frequency resistance, and by the plate model with the losses also on the dc resistance. A. Energy Losses by Current Impulses in Innite Half-Space and Plate Conductor Models The calculation is accomplished with the use of the frequency-dependent impedance of the space. The phasor of the , where is the phasor electric eld strength is of the magnetic eld strength and (8) is the space impedance. Therefore, the spectrum of is and according to (8) the conjugate of is

It can be seen that is an even and is an uneven falls out and is function. In the previous integral doubled in the integral from 0 to . So the loss per surface unit is (10) The frequency-dependent resistance of the innite half space [Fig. 4(a)] is (11) where is the skin depth at angular frequency. The energy loss on the mains frequency resistance with (12)

and the time-function of mula:

can be given with the following forIn case of innite conductor plate with nite width [Fig. 4(b)] the frequency-dependent resistance is [3], [4] (9) (13)

The instantaneous power per surface unit is energy loss per surface unit is

, so the where is the width of the plate. The energy loss per unit surface on mains frequency resistance can be obtained by with in formula (10). Since and replacing , the energy loss per unit surface on dc resistance (14)

KOLLER et al.: HEATING EFFECTS OF SHORT-CIRCUIT CURRENT IMPULSES ON CONTACTS AND CONDUCTORSPART I

225

Fig. 5. Ratio of real energy loss to the energy loss on dc and mains frequency resistance. (a) Sinusoidal impulse; (b) sinusoidal impulse with dc component; (c) impulse with current limitation; and (d) dc impulse.

Fig. 6. Current density distribution in innite half-space conductor at t = 2:5 ms. (a) Sinusoidal impulse; (b) sinusoidal impulse with dc component; (c) impulse with current limitation; and (d) dc impulse.

A more detailed deduction for solely sinusoidal impulse can be observed in Part II, Appendix A2. Fig. 5(a)(d) shows the ratio of the energy loss considering the current displacement to the energy loss considering only and dc resistance the mains frequency resistance of the conductor in the function of . The is the is the skin width of the conductor plate and depth at mains frequency. Since the resistance of the innite half-space is zero in steady-state dc condition [Fig. 5(d)], the real losses are compared also in this case with the losses on mains frequency resistance. It can be seen that when the time is short, and thus there are considerable high-frequency components in the spectrum (Fig. 3), the real losses are signicantly higher in all the four cases than when considering just the dc or mains frequency resistance generally used in the engineering practice, e.g., in case of solely sinusoidal impulse (Fig. 5(a)) at plate widths substantially larger than , and at the ratio . If the is close to the current zero, this ratio to the mains frequency losses is less than for sinusoidal impulse), 1 (at which means that if the current breaking occurs near the current zero the energy loss is smaller than when calculating in the traditional way. Increasing the impulse width this ratio converges to 1, which is true also for the sinusoidal impulse with dc component (b), but in the dc case (d) the limit of is 0, and the limit of is 1. If the plates width is much smaller than there is no signicant difference, and therefore the use of the dc resistance for determining the energy loss is acceptable. If the plates width is large enough the ratio equals the ratio calculated for innite half-space. The convergence of this ratio to the 1 limit is the quickest in the case of current limitation modelling the electric arc (c).

Fig. 7. Energy-density distribution in innite half-space. (a) Sinusoidal impulse; (b) sinusoidal impulse with dc component; (c) impulse with current limitation; and (d) dc impulse.

Here the impulse always ends in current zero, hence its shape is just marginally different from a half sine wave with width [Fig. 2(c)]. Thus, the proportion of the high frequency components in the spectrum is determined mostly by the impulse width.

226

IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 23, NO. 1, JANUARY 2008

Fig. 8. f =f in function of t (f = 50 Hz). (a) Sinusoidal impulse; (b) sinusoidal impulse with dc component; (c) impulse with current limitation; and (d) dc impulse.

B. Current Density Distribution in Innite Half-Space The Fourier-spectrum of the real function is (15) therefore (16) In case of sinusoidal excitation the current density phasor is [3], [4] (17) where is the phasor of the magnetic eld strength at the boundary surface of the half-space, is the coordinate perpenis the inverse dicular to the boundary surface, value of the skin depth in nonferromagnetic material, is the angular frequency of the sinusoidal excitation, and is the resistivity. If the spectrum of the magnetic eld strength at the , then from (17) the spectrum of boundary surface is the current density is (18) and according to (16) its instantaneous value at the boundary is distance from

which, similar to the deduction in Section III-A, can be expressed with the spectrum of the current density (21)

Fig. 6 shows the instantaneous value of the current density and the within the conductor at different times, if . It is noticeable that magnetic eld strength is current still ows in the conductor as an eddy current after the excitation is terminated. The distribution of the energy-density produced by the current impulse in the conductor is shown for different times in Fig. 7. This can be compared with a harmonic energy-density distribution at mains frequency, which can be expressed at distance from the boundary as (22) It is shown in Fig. 7 with a dashed line. The energy loss, is and hence the temperature of the conductor, is higher if higher, therefore, as the current displacement is represented by the steepness of the curves, it is better to compare the normalized . Because of the current energy-densities to the value at displacement the energy-density curves deviate from the harmonic case and most of the energy loss is concentrated close to the surface. In ac cases the higher the is the smaller is the deviation. By dc impulse as increases the distribution becomes more and more uniform. Each curve can be substituted with an exponential one in the form of (22) where the skin depth must be replaced with a equivalent skin depth. It means an distance from the surface, fold. An where the curve decreases from its initial value to equivalent frequency can be assigned to this , which, in harmonic case, generates approximately the same energy-density distribution as the given type of impulse with a given width. in the function of is shown in Fig. 8. The ratio of This ratio can reach a signicant value if the impulses are short. E.g. by a solely sinusoidal impulse (Fig. 8(a)) with

(19) The expression of this formula for solely sinusoidal impulse can be found in Part II, Appendix A3. The cumulated energy-density at distance from the boundary surface is (20)

KOLLER et al.: HEATING EFFECTS OF SHORT-CIRCUIT CURRENT IMPULSES ON CONTACTS AND CONDUCTORSPART I

227

it is 17.1251, which corresponds to a harmonic distribu. By a dc impulse [Fig. 8(d)] with the tion with . Even if same it is 16.9934 corresponding to , namely at a quarter sine period, can be 4 [Fig. 8(a)], 5 [Fig. 8(d)] or 8 [Fig. 8(b)]. In case of current limit is itation [Fig. 8(c)] it decreases much quicker, at . 1.8715, which means an

REFERENCES
[1] K. Simonyi, Foundations of Electrical Engineering. Oxford, U.K.: Pergamon, 1963. [2] A. Papoulis, The Fourier Integral and Its Applications. New York: McGraw-Hill, 1962. [3] W. Rogowski, ber zustzliche Kupferverluste, ber die kritische Kupferhhe eines Nut und ber das kritische Wiederstandverhltnis einer Wechselstrommaschine, Archiv Elektrotech., p. 81, 1913. [4] G. Tevan, Models of Current Displacement in High Current Elec sram elektrotechtrotechnics. (ramkiszortsi modellek az ero nikban.) Budapest Univ. Technol., Inst. Continuing Eng. Educ., 1985, p. 184. [5] Copper for Busbars Copper Development Assoc., 1996 [Online]. Available: http://www.cda.org.uk/megab2/elecapps/pub22/index.htm. l.), Elek[6] L. Koller, About the bus bars. (Az ramvezeto snekro trotech., vol. 93, no. 3, pp. 9196, 2000. [7] L. Koller, Examination of the efciency of crucible type induction furnaces (Untersuchung des elektrischen Wirkungsgrades von Induktionstiegelfen (Wirkung der Spulenparameter)), Elektrowrme Int., pp. 158160, Apr. 2001. [8] L. Koller, Inuence of the shielding on the efciency of crucibele type induction furnaces 1. (Wirkung der Abschirmungselemente (Fluleiter) auf den elektrischen Wirkungsgrad von Induktionstiegelfen.), Elektrowrme Int., pp. 7678, Feb. 2002. [9] L. Koller, Inuence of the shielding on the efciency of crucibele type induction furnaces 2. (Wirkung der Abschirmungselemente (Kurzschluring und Mantel) auf den elektrischen Wirkungsgrad von Induktionstiegelfen.), Elektrowrme Int., pp. 110111, Mar. 2002. [10] L. Koller, New ribbed type of crucible for induction furnaces (Neuartiger Tiegel fr Induktionsfen mit gerippter Wandche.), Elektrowrme Int., pp. 7579, Feb. 2003. [11] L. Koller, G. Tevan, and I. Kiss, Optimisation of electrical parameters of induction melting furnaces with graphite crucible, Elektrowrme Int., pp. 122125, Mar. 2004. [12] G. Dvnyi, S. Gcsa, L. Koller, and S. Vincze, Analyzing the current distribution within the contacts of a three phase high voltage SF circuit breaker. (Hromfzis tokozs SF6-gzos nagyfeszltsgu rendszere rameloszlsnak vizsglata.), Elekmegszakt rintkezo trotech., vol. 98, no. 9, pp. 245248, 2005. [13] L. Hannakam, Berechnung der transienten Stromverteilung im zylindrischen Massivleiter ETZ-A Bd.91, 1970. [14] E. Bolte, Transient current distribution in a cylindrical solid conductor placed in a semiclosed slot, Elect. Eng., vol. 85, pp. 19, 2003, 2002. [15] M. P. Filippakou, C. G. Karagiannopoulos, D. P. Agoris, and P. D. Bourkas, Electrical contact overheating under short-circuit currents, Elect. Power Syst. Res., vol. 57, pp. 141147, 2001.

Lszl Koller was born in Budapest, Hungary, in 1943. He received the Dipl.Eng. and the Ph.D. degrees in electrical engineering from the Budapest University of Technology and Economics in 1967 and 1994, respectively, and the Candidate of Engineering Sciences degree from the Hungarian Academy of Sciences, Budapest, in 1994. Between 1967 and 1974, he worked with different companies as a Technical Supervisor; and from 1970 to 1971, he was a Senior Lecturer at the Ministry of Public Education. From 1974 to 1995, he was an Assistant Lecturer at the Technical University of Budapest. Currently, he is an Associate Professor and Deputy Head in the Department of Power Engineering, Department of Electric Power Engineering, Budapest University of Technology and Economics, Budapest, Hungary. His major elds of research are high-current engineering (electrical switchgears and equipment, electrical heat engineering) and electrical safety engineering. He has been published in many publications. He also has 14 patents and inventions among them as well as 43 scientic articles. Dr. Koller is a member of the Public Corporation of the Hungarian Academy of Sciences from 2000; Hungarian Electrotechnical Association from 1976; and the Hungarian Engineering Chamber from 1996. He received the Gold Degree of Award of the Eminent Inventor, the Prize of Zipernowski, and the Prize of Academic Patent of Quality from the Hungarian Patent Ofce in 1988, 1995, and 2002, respectively.

Balzs Novk was born in Szeged, Hungary, in 1976. He received the M.Sc. degree in electrical engineering from the Budapest University of Technology and Economics, Budapest, Hungary, in 1999; the Postgraduate Engineer-Economist degree from the University of Economic Sciences and Public Administration, Budapest, in 2002, and is currently pursuing the Ph.D. degree in electric power engineering from the Budapest University of Technology and Economics. Currently, he is an Assistant Lecturer in the Department of Electric Power Engineering, Budapest University of Technology and Economics, Budapest, Hungary. His research interests include high-current engineering and electromagnetic compatibility. His publications include four conference papers and one paper. Mr. Novak is a member of the Hungarian Electrotechnical Association.

Gyrgy Tevan was born in Bkscsaba, Hungary, in 1927. He received the Dipl.M.Eng. degree from the Technical University of Budapest, Budapest, Hungary, in 1949 and the Candidate of Technical Sciences and Doctor of Technical Sciences degrees from the Scientic Qualication Board, Budapest, in 1967 and 1979, respectively. His major elds of research are the calculation of networks with Krons method; the voltage distortion of power systems; the investigation of active and reactive power in nonsinusoidal periodic systems; investigation of electric drives and induction heating systems; and linear algebra. He was with the Technical University of Miskolc, Hungary, and with the Technical University of Budapest. He is now retired. He wrote two textbooks, more than 50 papers, and has two patents. Dr. Tevan received the Level Award from the Hungarian Electrotechnical Association in 1970 and 1989, and the Excellent Teacher Award in 1974.

Anda mungkin juga menyukai