You are on page 1of 8
HIS - 01 - Heating by Internal Sources C-l COMPUTER SIMULATION OF INDUCTION HEATING SYSTEM WITH TRANSISTOR INVERTER Jerzy Zgraja and Jerzy Bereza Department of Electroheat, Technical University of Lodz. Al. Politechniki 11, 90-924 Lodz, Poland ABSTRACT. The paper presents the results of computer simulation of the process of induction heating with series transistor inverter. In the time harmonic simulation, the coupled electromagnetic and thermal fields of induction heating system were combined with the simulated real energy source. As a source, the transistor bridge inverter with series resonant circuit was applied. The source parameters (voltage and frequency) depend on the load parameters obtained by numerical calculations of the coupled fields. As the examples the induction heating before hardening of not moving cylindrical workpiece and the moving flat workpiece was considered. The simulation results were compared with these obtained for constant voltage and frequency supply. INTRODUCTION ‘An induction heating system comprising the energy source, inductor and charge is usually for computer simulation simplified to inductor and charge only. As a result the energy source is only a device supplying the constant voltage and frequency or the constant current and frequency, ‘Such simplification can lead to serious simulation errors especially when electronic sources supply are used in which output parameters are depended on inductor-charge system, In the paper the simulation of the coupled electromagnetic and thermal fields of inductor- charge system were combined with the real series transistor bridge inverter as the energy source supplying the inductor, as shown in Figurel. The load of the bridge inverter is represented by its equivalent resistance R, inductance L and series compensating capacitance C of constant value. The workpiece temperature changes have in consequence the variations of L and R so that inductor voltage and supplying frequency also change. The problem of the load and source parameter interdependence was solved using the commercial FLUX2D program combined with the elaborated GEN program simulating the inverter operation. ‘As resulting of big differences between heating time and period of inductor voltage supply the time harmonic simulation is the most ‘efficient one for simulation induction heating systems although its brings some problems with charge and power supply non-linearity. The inverter simulation presented in the paper takes into account the dead-time ta necessary for proper on-off switching of IGBT transistors of the bridge inverter and assumes the existence of only first harmonics of inductor voltage and current. This last assumption is. of no importance when dead-time is small enough in comparison with the output frequency period, in other cases it need further discussion. 131 ey y Figure 1. The inductor-charge system with power supply Besides, it was assumed that inverter load is inductive in the frequency range of operation Tt means that the damped angular frequency of RLC circuit a is smaller than the output inverter angular frequency @ (x-o/oo is bigger thanl), The sequence of calculations of energy source parameters and its coupling with electromagnetic and thermal fields is shown in Figure 2, where’ V-inverter voltage, Q=o0LAR, Ve-de voltage supplying the inverter. hue frteae | - Tt 7 uQtNp Figure 2. Simplified flow-chart of the simulation program The computer simulation presented in this paper is based on the examples of induction heating for surface hardening process of stationary cylindrical and moving flat workpiece. For moving charge the starting and end part of load are especially considered MATHEMATICAL MODEL AND SIMPLIFYING ASSUMPTIONS The simulation of induction heating of non-linear magnetic charge especially with inverter supplying system should be considered as transient magnetic problem Unfortunately computation of such a problem coupled with transient thermal problem causes a really high computing cost For this reason time harmonic magnetic problem is considered in the paper 132 ‘The mathematical analysis of coupled electromagnetic and temperature fields was carried out by solving Maxwell and Fourier-Kirchhoff equations. For considered problem the electromagnetic field was assumed to be harmonic, which allowed employing a complex vector potential A. Hence, the analysis was reduced to solving mutually coupled equations Joy A+-curlf(\/ wcurtAKs, a) where © - angular frequency, y -electrical conductivity, jt - magnetic permeability, J, excitation current density, and 29 ot div((-k) grado }+ pe @ where k - thermal conductivity, p - density, c - specific heat, py — volumetric heat source power density, with the existing boundary conditions taken into consideration To consider voltage supplied multi-tur inductor with non-uniform current density distribution in the turns , it is necessary to couple Maxwell equations with electrical network equations. Such strong coupling of magnetic and electric equations is possible in commercial FLUX2D program in which simulated conductors can be connected in an external circuit with current or voltage source The inductor-charge system supplied by series inverter can be considered, from electrical network point of view, as an equivalent resistance R, equivalent inductance Land constant series compensating capacitance C connected with non-linear voltage source (Figure 3) Source voltage and frequency depend on R, L, C values, on dead-time ts, dc supply voltage V, and drive angle. LV. £9) VIR,L,Cp¥. 0) AIR.L, C, ty) Fig.3 Equivalent electrical circuit of inductor-charge system supplied by inverter. The inverter simulation takes into account the dead-time ty necessary for proper on-off switching of transistors of the bridge inverter and assumes existence of only first harmonics of inductor voltage and current. For such assumptions the analytical analysis of the inverter from Figure! allows to describe its output current by equation: kot WD @) Feet w(ka,t-0,,) - de voltage supplying the inverter. where 133 and: = 81/, , (4) i Ye (5) oxlo,t,en)-¢ ® ealke-0,t,-n) © n=arc «( 4o) . aM (8) The current i (3)has a shape of exponential damped sinusoid. This function has property of skewed symmetry and therefore, the Fourier series has only odd harmonics. The amplitudes of the harmonics depend on k parameter and indirectly on dead time ty and Q factor. Taking into account that time ts is small in comparison of current (3) period and assuming that Q changes during heating of ferromagnetic material are in the range of 4-20, the third harmonic of current can be neglected. Therefore, the real current i has been substituted for first harmonic only. It permits to simplify the calculation process. The k parameter appearing in above relations, for assumed ts value and calculated in successive time steps factor Q, can be calculated from the relation: kan ta(o.t4)= 0) © +coskr Parameter _k permits to calculate the transient value of inverter angular frequency o, according to equation. (8) Basing on the Fourier analysis the first harmonic of the current i can be expressed as follows: i,(@gf)=B, sin aot + C, COS at (10) where B= afi (cot) sin Wot d(at) : 2 fio cos @,t d(a,t) a) Inserting the equation (3) into equations (11) the coefficients B1 and C1 can be written as sett {0s ots -e* Gysin 2kq +sin 2ka) +40 (I-e)] + aera?) . (12) +4sin otal e *°(b sin 2ka +2cos 2kx)~2 ] } cos ot, e*5 (dh-sin 2k +2.c0s 2k) ~ 2 + (13) +sin agt, [ °° (— tcos” ka +sin 2k) + 40(1 - 134 The first harmonic of the current (3) can be expressed as iy = sin (wt+9,) aay where g, =arcig$ as) The error resulting from considering only the first harmonic is less than 2% of rms current value for Q=4.8 and td=3ys, When Q=10 the errors falls to 0,8% of rms. current value The simulation of inductor~charge system supplied by a transistor inverter have been made by coupling the commercial FLUX2D program used for computation time harmonic magnetic field coupled with transient thermal field with the elaborated GEN program simulating the inverter operation, as has been shown in Figure 2 Z FEM presser, caleulates mynratin felt fe f-var), is shorter than in simulation with assumed constant frequency and voltage values. As a consequence the calculated hardening layer thickness Ay is about 25% smaller then value calculated in simplified simulation For the same example the power P, frequency f and inductor voltage Vou as functions of time, for calculation for real energy source, as well as power P, for calculation with constant V and f values, are shown in Figure 6. 8 rc) 1200 1020 200 ooo 400 200 simm) ot > a ee Figure 5. Temperature distribution in workpiece cross-section. 100 Vi : 80 kw) ‘ 60 | Ves Po i) | - ey nee ———— Pea ° tis) 0 02 04 08 08 1 12 14 Figure 6. Power, voltage and frequency as function of time. ‘The power increasing at the end of heating process is mainly caused by increased inverter output voltage above the charge magnetic transformation point (Curie Point) because in the series resonant circuit LRC, inductor voltage (across the R, L elements) increases with R decreasing, As can be seen in Figure 6 the power value calculated considering energy source parameters is at the beginning of heating lower than that calculated for V=const and f=const. During the heating time the frequency values change, but its value at the beginning and at the end of heating time is nearly the same. In Figure 7 the time variation of charge surface temperature in its middle part in inductor for both types of calculation is shown ai°c} 1200 - \Veconst, fsconst 1000 | > 800 + 600 | 400 200 ° 002 04 06 08 1 12 14 16 Figure 7. Time dependence of the surface temperature of the charge 136 In magnetic state of the charge the speed of temperature increasing is higher for calculation with simplified energy source, but after passing the Curie Point the output voltage of inverter increases and temperature of the charge (in such type of calculation) also rapidly increases. Finally, in considered example, the calculated heating time to reach the hardening temperature of the charge, is lower than for calculations taking into account the non-linear simulation of the source parameters. The influence of considering the non-linearity of energy source simulation of the inductor- charge system results in strong dependence on inductor-charge dimensions and on charge parameters. For considered example of inductor-charge system it has been shown in Figure 8 and Figure 9 how the frequency and inverter output power change in time for inductor-charge gap values A=1.5mm, 3mm and 9mm. For all these cases the starting inverter output voltage value has been chosen to have the same charge surface temperature value of about 1200°C [kHz] oo plete) \ A= 1.5:mm ts) °o 02 O46 06 (OB 1 12 14 Figure 8. The frequency vs. time for different inductor-charge gap A values. PiKW) 9mm 8 8 8 88 A=3mm Smm » tis} 0 02 of 06 o8 1 2 14 ° Figure 9. The inverter output power vs. time for different inductor-charge gap A values As can be seen in Figure 8 for relatively big gap value, A-9 mm, the frequency value during heating time is nearly constant. COMPUTER SIMULATION OF INDUCTION HEATING MOVING FLAT CHARGE The computer simulation of induction heating before hardening of moving flat charge has been made. The induction ~charge system presented in Figure 10 with inductor supplied by series transistor inverter has been considered 137 TT Y BLM, W. ti Hh ae. mage sur Figure 10. The inductor-moving flat charge system For moving charge the computer simulation of induction heating of the starting and end part of load is especially interesting. In the initial period of heating process, the electrical inductor-charge parameters rapidly change, thus having strong influence on supply source. For series resonance inverter leads to strong increase of the output voltage, if it is not reduced by voltage or current protection. Not taken into consideration so strong source-load interaction can lead to serious of simulation inaccuracy. In Figure 11 the inverter output power P(t), output voltage Vou(t) and frequency f(t) dependencies for initial period of induction heating of moving flat charge is presented. For comparison the dependence P(t) calculated assuming V=const and f=const, which are equal to the values calculated in simulation with inverter together at the half-length of the charge situated in the inductor, are also presented. P Kw) Veal =~ S Piveconst const) ° 2 0 0 0 10 120 Figure 11. Power, voltage and frequency as function of time As can be noticed in Figure 11 the inverter voltage protection does not allow to increase the output voltage over 50 V thus protecting the beginning part of charge against melting, CONCLUSIONS Computer simulation of induction heating system should be realised as coupled with supply source simulation, In rather simple and with low computer cost way, it can be reached by coupled numerical simulation of magneto-thermal field with analytical source description Such an approach in which only the first harmonic of output source voltage is considered is especially acceptable for series inverter with full drive Computer simulation of induction heating system without taking into consideration supply source simulation can lead to serious calculation errors, especially when induction-charge parameters rapidly change during the heating process. REFERENCES {1] Bottari, S..,”High-Frequency 200 kHz inverter for induction heating applications”, IEEE Ind. Applications. Soc. Conf. Rec, 308-316, 1985 138

You might also like