Paper accepted for presentation at 2003 IEEE Bologna Power Tech Conference, June 23th-26th, Bologna, Italy
Simulation by MATLAB/Simulink of active filters for reducing THD created by industrial systems I. Zamora, A. J. Mazon. P. Eguia, I. Albizu, K. J. Sagastabeitia, E. Fernández
Abstract—Nowadays, power electronics are widely used in industry for supplying loads with an amplitude and frequency controlled voltage. These systems comprises mainly rectifiers and inverters which, as non-linear loads, produce currents with high harmonic content. In order to fulfil the legislation concerning voltage harmonic distortion it is necessary to put in place corrective actions. Among these corrective actions active filters are one of the most effective. For the design of these filters simulation has been proved to be a very useful tool. In this paper, the simulation by MATLAB/Simulink of an active filter for the reduction of the harmonic distortion is analysed. Two examples are presented: a steel plant and an underground traction system. Index Terms— Active filter, computer simulation, induction furnace, steel plant, metro traction system, THD.
T
I. INTRODUCTION
he main requirement of any electric power system is the supply of electricity with a determined power quality and reliability to the minimum possible cost. Due to the increased quality of life, it has taken place a spectacular increase of the number and installed power of non-linear loads, especially of electronic devices used mainly in the control of systems and power hardware. Depending on the nature of these loads, the distortion created can be very high and affect the voltage supplied by the distribution network. In this case, it is highly possible that other loads served from the same network will be affected. For regulating this situation a complete legislation exists at national and supranational levels concerning voltage harmonic distortion. To comply with legislation, corrective actions have to be taken to reduce harmonic distortion to established
I. Zamora is with Department of Electrical Engineering-ESI of Bilbao, University of the Basque Country, Spain (e-mail:
[email protected]). A. J. Mazón is with Department of Electrical Engineering-ESI of Bilbao, University of the Basque Country, Spain (e-mail:
[email protected]). P. Eguia is with Department of Electrical Engineering-ESI of Bilbao, University of the Basque Country, Spain (e-mail:
[email protected]). I. Albizu is with Department of Electrical Engineering-ESI of Bilbao, University of the Basque Country, Spain (e-mail:
[email protected]). J.K. Sagastabeitia is with Dep. of Electrical Engineering-EUITI of Bilbao, University of the Basque Country, Spain (e-mail:
[email protected]). E. Fernández is with Department of Electrical Engineering-ESI of Bilbao, University of the Basque Country, Spain (e-mail:
[email protected]).
0-7803-7967-5/03/$17.00 ©2003 IEEE
regulated levels. Among the possible corrective actions active filters are one of the most effective. Before taking any corrective action, it is necessary to evaluate the distortion introduced by the installation into the distribution network and the expected reduction when the active filter is in use. In this stage, simulation has been proved to be a useful tool. It allows to quantify the harmonic distortion created by a system and, when a corrective action is introduced, simulation will show the reduction in the distortion. Besides, simulation can be used as a tool for the design of the active filter. The work presented in this paper is based on the simulation of active filters for reducing harmonic distortion created by industrial loads. Simulations have been carried out under the MATLAB/Simulink environment and two examples of industrial harmonic polluting loads are presented: a steel plant and an underground traction system. In order to perform the harmonic analysis of the voltage and current signals present in the industrial systems, a Simulink block has been developed. This block calls an M-file that makes the required calculations and shows graphically the harmonic spectrum of the analysed signal (Fig. 1). The 1st harmonic is out of scale so that the rest of the harmonics can be visualized properly. The calculated values correspond to peak values.
Fig. 1. Harmonic analyser result
A real time harmonic analyser, that shows the peak values of the first twenty harmonic components as the circuit is being simulated, has been developed too. This block has been developed using Simulink blocks. The results format is shown in Fig. 2.
Fig. 2. Real time spectrum analyser
II. ACTIVE FILTER Passive filters consisting of tuned LC filters have traditionally been used to improve power factor and to absorb harmonics in power systems because of their simplicity, low cost and high efficiency. However, these devices may fall in series and parallel resonance with the source impedance. Besides, the source impedance may vary, and influence
in the performance of the passive filter. In order to overcome these problems active power filters have been developed. Active power filters were first proposed for harmonic compensation in the early 1970's, but they could not be used in real power systems because high-power high-speed switching devices were unavailable. Since then, and because of the high development of power electronics technology, much research has been done on active filters and their practical applications. Currently, a great number of these devices has been installed, for different purposes as: harmonic compensation of non-linear loads, harmonics isolation between utilities and customers, harmonic damping throughout power distribution systems, reactive power/negative-sequence, flicker compensation, etc. The operation of an active filter is based on a continuous monitoring and conditioning of the distorted current created by the non-linear load. The same harmonic currents, but with a 180º phase shift are generated by the filter, so that harmonic components are cancelled and only fundamental component flows from the point of common coupling of the load. The filter used in the simulations is based on an inverter connected in parallel with the load. The inverter has a voltage source configuration and its control is based on the p-q control theory [1-2]. The SIMULINK model for the active filter can be seen in Fig. 3. In these figure, the filter is connected to the low voltage winding of a power transformer.
Fig. 3. Active filter model
III. CASES STUDIES A. Underground Traction System The main components of the traction substation modelled are the following ones: - Four rectifier groups with 3R+1 configuration. That is to say, three rectifiers are connected at the same time and one is always in reserve. Each rectifier group is composed of a 30/1,295 kV transformer (TGR) with 2.250 kVA rated power and a threephase rectifier. The rectifiers convert the voltage at the output winding of the transformers group into DC voltage of 1750 V when they work without load and into 1650 V when they work with 2000 kW rated power. - Two 30/13,8 kV power transformers with 2.500 kVA rated power each one, for station services. - One 30/0,4 kV transformer with 1.000 kVA rated power, for auxiliary services. Each metro traction unit or convoy has a traction power of 2.880 kW and nominal voltage of 1.500 Vdc and consists of four motor wagons. Each wagon leans over two bogies with two motors each, this is, each convoy has 16 induction motors of 180 kW supplied by a PWM inverter connected directly to DC line voltage.
The traction system is supplied by a source with 1230 MVA of short-circuit power and feeds a linear and another non-linear load. The linear load has been considered as a pure resistance and represents the station and auxiliary services. For modelling considerations, the two underground station service transformers and their load have been grouped into a single constant load of 5.000 kVA supplied at 30 kV and the auxiliary services transformer and its load as a constant load of 1.000 kVA supplied at 30 kV. The non-linear load is shown in Fig. 4. It includes three traction groups that feed two meter units. The rated power of the non-linear load is 6 MW and it is adjustable through the convoys torque. In the simulation, the torque measurements are real. To evaluate the rate of the total harmonic distortion in the substation, two measurement points have been located. The "point 1" is located in the connection with the distribution system of the electrical utility and the "point 2" is located in the feeding to the non-linear load (traction unit). In Fig. 5, the whole system, including the active filter, can be seen. As the frequency of underground trains varies along the day and so, the number of convoys that at the same time are supplied by the same substation, the simulation have been carried out at three different traction load levels (developing a 100%, 50% and 10% of the nominal torque) to study the variations produced in THD along the day.
Fig. 4. Traction system model
Fig. 5. Complete substation and traction system model
As example, we show the results of the case that both convoys are working in steady-state and developing a 100% of the nominal torque each one. Besides, in the “point 1” of measurement (connection with the distribution system) the graphs of voltage harmonic spectrum are shown, while in the “point 2” (feeding to the no-linear load) the graphs of current harmonic spectrum are shown. The results obtained in the harmonic distortion analysis of voltage at the connection “point 1” without the active filter are shown in Fig. 6. The THD value corresponding to the current is 4.85% while the voltage THD is 0.43 %. It can be seen that the harmonic voltage, generated by the substation, hardly have weight. This is due to the nonlinear load consumption which is about 6 MVA in front of the 1234 MVA short circuit power of the source. The highest harmonics are the 5th (250 Hz) and the 7th (350 Hz) and they are repeated at the 11th (550 Hz) and the 13th (650 Hz). These harmonics are characteristic of nonlinear loads that include three-phase rectifiers of six pulses.
If the harmonic analysis is performed including the active filter, the results change as can be seen in Fig. 7. The THD of the current is reduced from 4.85% to 0.10% and the voltage THD from 0.43% to 0.31%. The values of the highest harmonics have been reduced considerably. But some new components of low value (harmonics due to the injection of the filter) can be observed.
Fig. 7. Voltage harmonic spectrum in the point 1, with filter
For the “point 2” of measurement, the results obtained in the harmonic distortion analysis of current without the active filter are shown in Fig. 8. It is observed a higher deformation of the waves than in the case measured in the point 1. This distortion is due to the characteristic consumption of the non-linear load that contains electronic power devices. The THD value corresponding to the current is 9.16% while the voltage THD is 0.31%. Fig. 6. Voltage harmonic spectrum in the point 1, without filter
-
Fig. 8. Current harmonic spectrum in the point 2, without filter
If the harmonic analysis is performed including the active filter (Fig. 9), the THD of the current is reduced from 9.16% to 0.35% and the voltage THD from 0.43% to 0.31%.
Fig. 9. Current harmonic spectrum in the point 2, with filter
From the whole simulated cases results without and with the active filter, it can make a comparative analysis: - The THDI decreases more than 86% in anyone of the three cases, which implies a high effectiveness of the active filter (TABLE I and TABLE II). TABLE I. – Results for measurement point 1 (substation interconnection) Reduction of Current Current current distortion with Traction load distortion distortion active filter level (THDI %) (%) (THDI %) 100 % 4,85 0,09 98,14 50 % 4,07 0,09 97,79 10 % 1,72 0,08 95,35 TABLE II. – Results for measurement point 2 (traction load) Reduction of Current Current current distortion with Traction load distortion distortion active filter level (THDI %) (%) (THDI %) 100 % 9,16 0,35 96,18 50 % 11,51 0,63 94,53 10 % 15,08 2,01 86,67
The harmonics are decreased almost completely and the highest value becomes the 11th and 13th order, instead of the 5th and 7th order in the case without filtering. Even harmonics are also observed (included the 0 order harmonic which is the DC component). These harmonics are introduced by the active filter, but they take worthless and highly variable values.
B. Steel Plant The steel plant substation considered is fed alternatively from two 30 kV lines switched by means of a disconnector. The electric load of the plant is composed of six induction furnaces, temper furnaces and general services. There are two different types of induction furnaces. Four furnaces are of type 1 and the other two of type 2. Each furnace of type 1 is fed from a 4 MVA threewinding transformer which reduces the voltage from 30 kV to 770 V. The secondary winding feeds a six pulse rectifier and the tertiary feeds another identical rectifier. The rectification has a 12-pulse configuration. Both rectifiers are connected in parallel including filtering coils that improve the direct current obtained. A capacitor bank is connected in parallel with the induction furnace coil to achieve a controllable resonance of the coil. The voltage in the coils that melt the steel is 2200 V with a frequency of 500 Hz and an approximate consumption of 3300 kW. Each furnace of type 2 is fed from a 5 MVA threewinding transformer. The voltage is reduced to 945 V. Similarly to the type 1 furnace configuration, the secondary winding feeds a six pulse rectifier and the tertiary feeds another identical rectifier. The rectification has a 12-pulse configuration. Both rectifiers are connected in series including filtering coils. A capacitor bank is connected in parallel with the induction furnace coil. The approximate consumption of the coil is 4 MW. The rectifiers used to get the DC voltage are the cause of the injection of current harmonics in the system and consequently the cause of the voltage distortion. The characteristic harmonics injected by a 12-pulse rectifier are harmonics of order 11th and 13th. The distribution transformers of the general services and the temper furnaces consume 1800 kW with a cosφ of 0,85. For simulation purposes they have been modelled as a linear load of these characteristics. This is acceptable as their consumption is only a little portion of the total power consumed in the plant and they do not produce any distortion. All these elements have been modelled using existing SIMULINK blocks contained in the SymPowerSystems blockset. As there is no induction furnace electrical model, new blocks have been created for the two types of induction furnaces (Fig. 10 and 11).
Fig. 10. Type 1 induction furnace model
Fig. 11. Type 2 induction furnace model
The simulation model of the steel plant including the active filter inserted in parallel with the furnaces can be seen in Fig. 12. The steel plant has been simulated considering different levels of load (one, three and six furnaces working). Firstly,
the voltage harmonic distortion created by the steel plant has been measured. Secondly, after the active filter has been added, the distortion has been measured again.
Fig. 12. Steel plant model including the active filter
As example, we show the results obtained with the six furnaces working. This is the most common working situation of the steel plant in order to get the maximum profitability. This is the worst case from the point of view of the harmonic distortion and the active filter should be able to correct it properly. We can see that the current THD value is 7.41% while the voltage THD is 6.38%. The voltage distortion is too high as it is above the limit imposed by the regulations (5% for 30 kV). The highest harmonics are the 11th and the 13th. The 9th harmonic also increases its value from the previous cases. The results obtained in the harmonic distortion analysis are shown in Fig. 13 and 14. Fig. 14. Current harmonics (6 furnaces)
The results obtained including the active filter are shown in Fig. 15 and 16.
Fig. 13. Voltage (6 furnaces)
Fig. 15. Voltage harmonics (6 furnaces + active filter)
IV. CONCLUSIONS The use of simulation tools as MATLAB/ Simulink, allows to reproduce the behaviour of the power systems in different situations, analyse how the system answers in these situations and choose the solution that better fit with the particular problem without additional costs. Besides, active filters with different rated values can be simulated in order to analyse different reductions of the harmonic distortion. By means of the simulation carried out, the voltage and current harmonic distortions created by an underground traction system and a steel plant have been obtained. Moreover, the reduction of the distortion by an active filter has been simulated for both systems. Fig. 16. Current harmonics (6 furnaces + active filter)
The THD of the current is reduced from 7.41% to 3.33% and the voltage THD from 6.38% to 3.61%. The new value of the voltage distortion is below the limit of 5%. Therefore the active filter is adequate for the steel plant to fulfil the voltage quality requirements. The 11th and 13th harmonics are still the highest harmonics and the 9th is nearly removed. But there are many new components introduced by the filter spread over the spectrum, but they take worthless and highly variable values. Additional results obtained whit one or three furnaces working are shown in TABLES V and VI. No. of furnaces working 1 3 6
TABLE V. - Current distortion Reduction of THDI with Current THDI active filter THDI (%) (%) (%) 4.47 1.87 58.1 6.91 3.85 44.2 7.41 3.33 55
No. of furnaces working 1 3 6
TABLE VI. – Voltage distortion Reduction of THD with Voltage THD active filter THD (%) (%) (%) 1.04 0.66 36.5 3.23 2.06 36.2 6.38 3.61 43.4
V. REFERENCES [1]
H. Akagi, Y. Kanazawa and A. Nabae, “Generalized theory of instantaneous reactive power in three-phase circuits”, Int. Power Electronics Conf., pp. 1375-1386, Tokyo, 1983. [2] H. Akagi, “New trends in active filters for power conditioning” , IEEE Trans. Ind. Applications, Vol. 32, No. 6, 1996. [3] J. Shen, N. Butterworth, "Analysis and design of a three level PWM converter system for railway-traction applications", IEE Procedings Electric Power Applications, Vol. 144, No. 5, 1999. [4] K. Al-Haddad, B. Singh and A. Chandra, “A review of active filters for power quality improvement”, IEEE Transactions on Industrial Electronics, Vol. 46, No. 5, pp. 960-971, 1999. [5] G. Casaravilla, A. Salvia, C. Briozzo and E. Watanabe, “Selective active filter applied to an arc furnace adjusted to harmonic emission limitations”, Latin America T&D IEEE Conference, San Pablo - Brasil, 2002. [6] P.T. Cheng, . Bhattacharya, D.M. Divan, "Application of dominant harmonic active filter system with 12 pulse nonlinear loads", IEEE Transactions on Power Delivery, Vol. 14, No. 2, 1999. [7] Chih-Ju Chou, Chih-Wen Liu, June-Yown Lee, Kune-Da Lee, “Optimal Planning of Large Passive- Harmonic-Filters Set at High Voltage Level”, IEEE Transactions on Power Systems, Vol.15, No. 1, 2000 [8] Bhim Singh, Kamal Al-Haddad, Ambrish Chandra, “A New Control Approach to Three-Phase Active Filter for Harmonics and Reactive Power Compensation”, IEEE Transactiones on Power Systems, Vol.13, No.1, 1998. [9] Carlos V. Nunez-Noriega, George G.Karady, “Five Step- Low Frequency Switching Active Power Filter for Network Harmonic Compensation in Substations”, IEEE Transactions on Power Delivery, Vol. 14, No.4, 1999. [10] Thierry Thomas, Kévork Haddad, Géza Joós, Alain Jaafari, “Design and Performance of Active Power Filters”, IEEE Industry Applications Magazine, 1998.