ASMES device is a dc current device that stores energy

Transcription

1 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 21, NO. 2, APRIL Detailed Modeling of Superconducting Magnetic Energy Storage (SMES) System IEEE Task Force on Benchmark Models for Digital Simulation of FACTS and Custom-Power Controllers, T&D Committee Abstract This paper presents a detailed model for simulation of a Superconducting Magnetic Energy Storage (SMES) system. SMES technology has the potential to bring real power storage characteristic to the utility transmission and distribution systems. The principle of SMES system operation is reviewed in this paper. To understand transient and dynamic performance of a SMES system, a detailed SMES system benchmark model is given with extensive simulation results. This system is demonstrated using an electromagnetic transient program PSCAD/EMTDC. Index Terms Custom power controllers, flexible ac transmission systems (FACTS), power electronics, power system transients, superconducting coils, superconducting magnetic energy storage (SMES), transient analysis. I. INTRODUCTION ASMES device is a dc current device that stores energy in the magnetic field. The dc current flowing through a superconducting wire in a large magnet creates the magnetic field. Generally it consists of the superconducting coil, the cryogenic system, and the Power Conversion/Conditioning System (PCS) with control and protection functions [1]. The total efficiency of a SMES system can be very high since it does not require energy conversion from electrical to mechanical or chemical energy. Depending on the control loop of its power conversion unit and switching characteristics, the SMES system can respond very rapidly (MWs/milliseconds). The ability of injecting/absorbing real or reactive power can increase the effectiveness of the control, and enhance system reliability and availability. Consequently, SMES has inherently high storage efficiency, about 90% or greater round trip efficiency. Comparing with other storage technologies, the SMES technology has a unique advantage in two types of applications: Power system Manuscript received February 25, 2003; revised November 20, The work was supported in part by the National Science Foundation, Department of Energy under Grant no. DE-FG36-94GO10011, and in part by BWX Technologies, Inc. Naval Nuclear Fuel Division. However, any opinions, findings, conclusions, or recommendations expressed herein are those of the authors and do not necessarily reflect the views of the sponsoring organizations. Paper no. TPWRD L. Chen and Y. Liu are with the Bradley Department of Electrical Engineering, Virginia Polytechnic Institute and State University, Blacksburg, VA USA ( yilu@vt.edu). A. B. Arsoy is with the Department of Electrical Engineering, Kocaeli University, Izmit-Kocaeli, Turkey. P. F. Ribeiro is with the Department of Engineering, Calvin College, Grand Rapids, MI USA. M. Steurer is with the Center for Advanced Power Systems (CAPS), Florida State University, Tallahassee, FL USA. M. R. Iravani, Chair of the IEEE Task Force on Benchmark Models for Digital Simulation of FACTS and Custom-Power Controllers, T&D Committee, is with the Department of Electrical and Computer Engineering, The University of Toronto, Toronto, ON M5S 3G4, Canada. Digital Object Identifier /TPWRD Fig. 1. Components of a typical SMES system. CSI: current source inverter. VSI: voltage source inverter. transmission control and stabilization, and power quality improvement. For instance, SMES can be configured to provide energy storage for FACTS controllers at the transmission level, or custom power devices at the distribution level. The efficiency and fast response capability of a SMES can be further exploited in different applications in all levels of electric power systems [2]. In order to achieve the best system configuration possible, the design of the SMES system needs to take into account many factors. The performance evaluation of a SMES system also requires extensive knowledge about the SMES and the associated power systems. Computer aided simulation is one of the cost effective ways to carry out. This paper intends to provide a detailed model of the SMES system for the SMES related power system computer simulation. The benchmark system will provide the basis for the comparison of the different simulation tools, control strategies and algorithms related to SMES systems. The proposed SMES system will utilize parameters from BWX Technologies, Inc. for the SMES coil modeling. A GTO based Voltage Source Converter/Inverter (VSC/VSI) will be used for modeling the PCS of the proposed SMES system. The principle of the SMES system operation is reviewed, and the detailed SMES benchmark system configuration is presented with extensive simulation results. II. SMES SYSTEM OVERVIEW As can be seen from Fig. 1, a SMES system consists of several sub-systems. A large superconducting coil is the heart of a SMES system, which is contained in a cryostat or dewar consisting of a vacuum vessel and a liquid vessel that cools the coil. A cryogenic system is also used to keep the temperature well below the critical temperature of the superconductor. An ac/dc /$ IEEE

2 700 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 21, NO. 2, APRIL 2006 PCS is used for two purposes: One is to convert electrical energy from dc to ac, and the other is to charge and discharge the coil. Finally, a transformer provides the connection to the power system and reduces the operating voltage to acceptable levels for the PCS. For a SMES system, the inductively stored energy (E in Joule) and the rated power (P in Watt) are commonly the given specifications for SMES devices, and can be expressed as follows: where is the inductance of the coil, is the dc current flowing through the coil, and is the voltage across the coil. During SMES operation, the magnet coils have to remain in the superconducting status. A refrigerator in the cryogenic system maintains the required temperature for proper superconducting operation. Since the refrigeration load can affect the overall efficiency and cost of a SMES system, the refrigeration load that has loss components (such as, cold to warm current leads, ac current, conduction and radiation, etc) should be minimized to achieve a higher efficient and less costly SMES system. A PCS provides a power electronic interface between ac power system and superconducting coil. It allows the SMES system to respond within tens of milliseconds to power demands that could include a change from maximum charge rate to maximum discharge rate of power. This rapid response allows a diurnal storage unit to provide spinning reserve and improve system stability. Converters may produce harmonics on the ac bus and in the terminal voltage of the coil. Using higher pulse converters can reduce these harmonics. A PCS could be either a current source inverter or a voltage source inverter with a dc dc chopper interface. A bypass switch is used to reduce energy losses when the coil is on standby. And it also serves other purposes such as bypassing dc coil current if utility tie is lost, removing converter from service, or protecting the coil if cooling is lost. The superconducting coil is charged or discharged by adjusting the average (i.e., dc) voltage across the coil to be positive or negative values by means of a dc dc chopper. When the unit is on standby, the coil current is kept constant, independent of the storage level, by adjusting the chopper duty cycle to 50%, resulting in the net voltage across the superconducting winding to be zero. III. SMES SYSTEM ELEMENTS MODELING A. SMES Coil Modeling 1) Coil Modeling Considerations: A number of methods are available to determine coil characteristics such as direct measurement on the actual winding, direct measurement on an electromagnetic model of the coil, or building a mathematical model and determining the voltage distribution and frequency response by means of the computer-aid analysis. As explained in [3], [4], the first two methods, with no doubt, give more reliable and accurate results than the mathematical analysis. But setting up (1) Fig. 2. Schematic representation of a winding. C = shunt capacitance. C = series capacitance. L = self inductance. M = mutual inductance. a mathematical model is the most convenient method with the lowest cost. At the same time, a greater variety of design alternatives and detailed analysis can be achieved with the mathematical model of a winding. The most detailed model would require a representation of single turns, which take the magnetic mutual couplings to all other turns into account. However, such a model is difficult to obtain and to handle as well as impractical in most cases. Unless the time delay of traveling wave phenomena is really of interest, e.g., for very fast transients such as a lightning surge, a lumped parameter network model proofs sufficient [9]. A lumped parameter network model contains magnetic and dielectric circuits, which have the following sets of elements: The magnetic circuit is represented by self and mutual inductance ( and ) of each turn. The dielectric circuit is represented by capacitances between adjacent turns, axially separated turns and turn to the outside surface. Due to the high memory and computing costs, various degrees of simplification are necessary. The simplification methods are given in [3]. According to [5], a relatively small number of dominant resonance frequencies are sufficient for the analysis. Therefore, a distributed winding can be represented by an equivalent circuit of a finite number of lumped elements. When a superconducting coil is simulated for a purpose of dynamic operation, it is a common practice to represent the coil as an inductor. On the other hand, for transient analyses, the more detailed coil model representing disks, even turns with associated mutual inductance and capacitance yields more accurate results. 2) Calculation of Electrical Parameters: An electrical lumped parameter model, illustrated in Fig. 2 is constructed for a superconducting coil to determine voltage distribution and frequency response of the coil. It is assumed that the coil consists of a number of disks (pancakes) comprised of a number of turns. Given the geometrical dimension of a coil, the following parameters need to be calculated for each turn of the coil including,,,, [6]. In order to avoid computing cost, a lumped double pancake parameter model is developed using the parameters computed for turns. In transient analysis simulations, representing the first and last few double pancakes with turn-to-turn representation may satisfy the requirement for the detailed modeling. Physical dimensions of the coil as illustrated in Fig. 3 should be given in order to calculate the inductance and capacitance parameters for a particular turn in the coil. Fig. 3(a) shows the cross section of the entire coil having a number of disks, Fig. 3(b)

3 CHEN et al.: DETAILED MODELING OF SMES SYSTEM 701 Fig. 4. Coil representation by Lyle s method for b>c. where is determined by the ratio of, which determines the thickness of the coil as shown in the equation at the bottom of the page. Mutual inductance between two circular filaments is calculated by using the formula developed by Maxwell [7], [9] (4) Fig. 3. Illustration of the physical dimensions for a disc coil. (a) Disc winding. (b) Two disk coils. (c) Disk coil with rectangular cross section. illustrates two disks on top of each other, and Fig. 3(c) displays the required dimensions in parameters computation for a turn/disk coil. It should be noted that the MKS unit system is observed in all of the parameter formulas given below. a) Self and mutual inductance calculation: Inductance calculations are based on geometric entities. Formulas developed in [7] to compute the self-inductance (in Henry) are where and are the radii of the circular filaments 1 and 2, is the distance between circular filaments, is the permeability of free space, and and are the complete elliptic integrals of the first and second kind. Equation (4) is valid for circular filaments of a negligible cross section. If the cross sectional dimensions are not small enough when comparing with the distance between the coils, then Lyle s method is applied to calculate the mutual inductance between turns or coils. Lyle s method states that each coil (Coils 1 and 2) can be represented by two equivalent filaments as shown in Fig. 4. Mutual inductance between each equivalent filament is calculated by using (4), where and are replaced with an equivalent radius of and spacing between coils vary between and in (5) (2) where is the number of turns in a pair of disk coils (double pancake), other dimensions are shown in Fig. 3(c). If the inductance of a turn is to be calculated, then. Another formula is developed by using tables in [6], [8] (3) The average of each calculated mutual inductance gives the mutual inductance between the two coils. b) Capacitance calculation: Capacitance calculation is also based on the geometric entities. In a double pancake, the series capacitance between turns, shunt capacitances between axially separated turns, and turn to ground can be easily and (5) then then is found from the tables given in [6]

4 702 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 21, NO. 2, APRIL 2006 accurately calculated by using the parallel plate formula given in (6) (6) where, and is the constant of the dielectric between two plates, is the area between two plates, and is the distance between plates. The capacitances of turns in a double pancake are combined to represent capacitances between double pancakes and to ground. Equations for series capacitance of the total winding developed by several authors have been summarized in [6]. 3) Developing a SMES Coil Model With Lumped Parameter: a) Modeling Assumptions: It is assumed that the SMES coil can be accurately represented by a lumped parameter model, shown in Fig. 3(a). In this model, each double pancake is represented by self and mutual inductors, and series and ground capacitors. The inductance and capacitance values of a double pancake are obtained by lumping those of turns forming the pancake. The following additional assumptions are made in [10]: The dielectric constant of the insulating material does not vary with frequency. The thermal enclosure and the tank do not carry current, and they are represented as ground plane A small value of resistor represents skin effect and eddy current losses. Parallel plate model is employed to calculate ground and series capacitances of each turn. b) Modeling Steps: As stated before, the most detailed mathematical model for a coil can be obtained if each turn in the coil is represented by its associated,,,, and. But this detailed model requires much memory and computing time if the coil consists of excessive numbers of turns. number of turns in a number of double pancakes in a coil can be lumped to model the coil in the level of double pancakes. For better understanding, an example coil is considered with single pancake ( double pancake) where each Single Pancake (SP) consists of turns. Forming inductance matrix Calculate self-inductances for each turn in a SP by applying the Miki s formula [7], and mutual inductances between each turn in a SP by applying the Lyle s method. Construct an matrix block (A in (7)) Fig. 5. Derivation of equivalent series capacitance for a double pancake a1 = (2N 0 1)Cax, b1 =(2N 0 1)Cax, a2 =(2N 0 3)Cax, b2 =(2N 0 3)Cax -, a3 = (2N 0 5)Cax, b3 = (2N 0 5)Cax -, an 0 1 = 3Cax -, bn 0 1 = 3Cax, an = 1Cax, bn = 1Cax. where N is the number of turns in a SP. Diagonal elements correspond to the self inductances and off-diagonal elements correspond to mutual inductances between each turns in a SP. See (7) at the bottom of the page. We applied Lyle s method to calculate mutual inductances between turns in the first SP and turns in the other SPs. A series of matrix blocks ( to in (7)) are constructed where each represents mutual inductances between the first SP and other SPs. These matrix blocks and the one constructed above builds a column with a size of. Once the first column of turn is formed, a lumped inductance matrix representing the double pancakes of the coil is computed as given in (8) (9), shown at the bottom of the next page. Calculating capacitances for a double pancake Calculate capacitances between adjacent turns, axially separated turns and turns to ground using (6). Capacitances between adjacent turns and axially separated turns are combined in such a way shown in Fig. 5 to compute the equivalent series capacitance for a double pancake. Ground capacitances calculated for each (7)

5 CHEN et al.: DETAILED MODELING OF SMES SYSTEM 703 turn within a double pancake are summed to obtain an equivalent ground capacitance for a double pancake. c) Credibility of SMES Coil Modeling: The inductance calculation method developed here is applied for the example coils given in [11], [12]. A capacitance measurement experiment is conducted on a sample coil to compare calculated and measured values, and the results are well matched. B. Model of Power Electronics Conversion and Control Unit The power electronics interface between a superconducting coil and the ac power system is called SMES PCS. A PCS is expected to transfer energy into or out of the SMES on command to control real and reactive power, and to be able to bypass the coil when there is no need for energy into or out of the coil [13] [16]. Certain factors such as semiconductor device types, switching technologies, system configuration and reactive power requirement have been considered/evaluated for a PCS design. Two basic types of inverters, Current-Sourced Inverters (CSI) or Voltage-Sourced Converters (VSC) are commonly used for the power conversion unit between SMES and ac power system. The SMES coil studied in this work is being built for FACTS applications, so VSC along with a dc dc chopper interface is used. Attaching SMES devices to the VSC based FACTS controllers used in power systems can improve the effectiveness and capability of overall system. VSC is composed of a turn-off device based converter, a dc link capacitor, and inductance on the ac side. The large capacitance ensures that the voltage is uni-polar. It also handles sustained charge/discharge current. An inductive interface is needed to ensure that the dc link capacitor is not short-circuited. It should be noted that the VSC must have bi-directional valves to allow current flow in either direction. Natural commutated devices initially used in power conversion of SMES systems are replaced by high power forced (self) commutated semiconductor devices, which offer more controllability and flexibility. And since GTOs are well established and employed devices, the simulation work presented in this benchmark work uses GTO devices for the power electronics interface. Varying the width of the voltage pulses, and/or the amplitude of the dc bus voltage can control the ac output voltage. Due to the nature of converters, harmonics are present. To reduce harmonic magnitude, either a multi-pulse VSC with 180-degree conduction or a three-phase PWM scheme is utilized. PWM scheme has Fig. 6. VSI/dc dc chopper configuration for SMES. TABLE I PARAMETERS OF THE COIL STUDIED not been justified for high power converters due to the switching losses [17]. The use of VSC for SMES applications has been proposed for the Engineering Test Model (ETM) project [14], [15], [18]. A 24-pulse VSC and a two-quadrant multi-phase dc dc chopper for SMES have been introduced. The VSC and the chopper are linked by a dc link capacitor that behaves as a stiff but controllable dc voltage source providing the desired characteristics. A three-phase VSC and single-phase chopper connection is illustrated in Fig. 6. VSC can provide continuous rated capacity VAR support even at low or no coil current whereas CSI is dependent of coil in providing VAR support. IV. DETAILED SMES MODEL SYSTEM A. SMES Coil The SMES Coil modeling is based on a 50 MW, 100 MJ SMES coil built by BWX Technologies, Inc. for a FACTS/energy storage application. The entire SMES coil has a width/height ratio of 3.66 m (144 in)/1.53 m (60 in) made of 48 Double Pancakes (DP). Each double pancake has 40 turns. The calculated series and shunt capacitances for each turn have been lumped to represent capacitances for a double pancake, and lumped further to obtain the equivalent capacitances for the entire coil for simulation of the dynamic operation as showed in Table I. (8) (9)

6 704 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 21, NO. 2, APRIL 2006 B. The Power Conversion System a GTO-Based Chopper A multiphase GTO-based chopper is modeled as shown in Fig. 7. Components in dashed rectangles are for bypass switching and transient suppression purposes. A constant 24 kv dc voltage source represents the dc side of the chopper. In FACTS/SMES applications, the dc dc chopper is connected to a FACTS device through a dc link capacitor maintaining a constant dc voltage. Operation principles of the multiphase dc chopper can be explained with the help of a single-phase dc chopper. In a singlephase dc chopper, the GTO firing signal is a square wave with a specified duty cycle. The average voltage and current of the SMES coil are related to the dc source voltage and average current by the duty cycle applied [18]. These relationships can be expressed as (10) Fig. 7. Structure of a GTO-based chopper. The self and mutual inductances for each turn also have been lumped to obtain the equivalent self and mutual inductances for each double pancake. The total inductance is 12.5 H. In order to reduce the computational burden, an equivalent circuit of the coil is represented by a 6-segment model comprised of self inductances, mutual couplings, ac loss resistances, and series and shunt capacitances, as shown in Fig. 2, including the mutual couplings between segments to obtain more accurate frequency and voltage response [5]. The inductance and capacitance values for segments are based on the previous design of the SMES coil provided by BWX Technologies. When calculating the parameters of the 6-segment model, the following steps are used: 1). Turn to turn values are lumped to obtain the parameters of a double pancake, 2). These lumped values are used to calculate the parameters of the entire coil, 3). The entire coil parameters are evenly distributed into 6 segments. It is assumed that inductance and capacitance values are equal for each segment, and PI model is adopted. The parameters for 6 segments are used throughout the voltage distribution and transient analysis studies. The inductance matrix used in the simulation is as follows: where - is the average voltage across the SMES coil, - is the average current through the SMES coil, - is the average dc source voltage, - is the average dc source current, and d is the duty cycle of the chopper ( conduction time/period of one switching cycle). There is no energy transferring at a duty cycle of 0.5, where the average voltage of the coil is zero and the average coil current is constant. In the cases of duty, when cycle being larger than 0.5 or less than 0.5, the coil is either charging or discharging respectively. Adjusting the duty cycle of the GTO firing signals controls the rate of charging/discharging. For a n-phase dc chopper, the duty cycle of firing signals of each phase is of the total duty cycle. In the 3-phase chopper used, the duty cycle of each phase changes from 0 to 1/3, and the frequencies of the GTO firing signals are 100 Hz. In Fig. 7, small inductors are placed at the output of each chopper phase for the purpose of current sharing, and resistors represent the resistances of these inductors and the leads. V. SMES SYSTEM SIMULATION A. Resonance Analysis The SMES coil has 48 double pancakes. Since currently available version of EMTDC can only model a limited number of mutual couplings, turn-to-turn values of both self and mutual inductances are lumped together and then divided equally into 6-segments. In EMTDC model of the parameters for the 6-segment, mutual couplings between any two segments have been considered (total of 30 mutual inductances) to obtain frequency and voltage response. Frequency scan analysis is performed to predict resonant frequencies. The model for the frequency scan is the same equivalent circuit of the SMES coil represented by a 6-segment model comprised of self inductances, mutual couplings, ac loss resistances, and series and shunt capacitances. The coil s resonance frequencies are studied by using this electrical lumped parameter model to determine the frequency response of the coil by using the built in functions in EMTDC. Fig. 8 shows the result of the magnitude of the coil terminal voltage vs. frequency. As can be seen, the coil has several resonance frequencies, parallel resonance at frequencies around 60 Hz, 400 Hz, 890 Hz, and series resonance at 280 Hz, 830 Hz, which can lead to possible magnification of transients. Since the SMES coil has a rather high inductance of 12.5 H, the resonance frequencies of the coil are relatively low. While similar results have been reported in [21] on the same coil structure any discrepencies of specific resonance frequencies may be due to differences in model input parameters (e.g., coil dimensions, dielectric constant of insulation, ground plane influence).

7 CHEN et al.: DETAILED MODELING OF SMES SYSTEM 705 Fig. 8. Frequency response of the six-segment SMES coil model. Fig. 10. Transients under normal operation condition. Fig. 9. Initial voltage distribution of the coil. The frequency scan results show that when an accidental operation of the coil at its natural mode frequencies occurs, an internal over voltage is observed in the coil, which could break down the coil insulation. B. Voltage Distribution Analysis Determining voltage distribution along the coil is essential in the coil insulation system design. The voltage distribution is obtained by recording the internal voltages at a snapshot time after a surge impulse is applied to the coil. EMTDC/PSCAD is used as a platform to compute the voltage distribution of the SMES coil. An amplification ratio is defined as the square root of the ratio of ground capacitance to series capacitance of the coil. This factor plays an important role in the voltage distribution of the coil. The amplification ratio of each double pancake in the SMES coil is found to be approximately The per-unitized initial voltage distribution of the coil at this amplification ratio can be obtained when an impulse surge voltage is applied to the SMES coil. This is depicted in Fig. 9. The same scenario is considered for the 6-segment coil model that is derived by lumping the double pancake parameters further. It is observed that similar voltage distribution curves are obtained. If better accuracy is desired from the coil model, it can be achieved by using more sections and finer parameters in the model. C. Transient Analysis Transients seen by the SMES coil can originate from ac or dc system faults/switching. The normal switching operation of the devices generates periodic pulse sequences that may be continuously applied to the SMES coil. In order to minimize the impact of the transients caused by the chopper switching, a bypass switch is used to short the coil when there is no power/energy exchange required. Operation of the bypass switch also introduces transients that may affect the coil. 1) Transients Generated by GTO Switching Under Normal Chopper Operation Condition: Under normal chopper operation condition, transient characteristics depend on the duty cycle and the average coil current. For example, Fig. 10 gives the simulation results when the duty cycle is 0.5 and average coil current is 0.4 ka. The coil is charged at full speed (duty cycle equals to 1) from to s. After 0.22 s, the duty cycle is kept at 0.5 and the SMES system is in standby mode. Fig. 10(a) shows the duty cycle for the time interval s. The SMES coil terminal voltage,, is shown in Fig. 10(b) for the same time interval. An expanded inset of is also given in Fig. 10(b-1) to show the high frequency components in detail. Fig. 10(d) shows the maximum voltage (to ground) at each node along the entire coil. As can be seen, voltage transients appear after the duty cycle is changed to 0.5. The terminal voltage can easily go up to over 40 kv. It should be noted that the transitions of terminal voltages from negative to positive or vice versa have two transient processes. These are due to the delay of the two firing signals for the two GTO branches in the

8 706 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 21, NO. 2, APRIL 2006 Fig. 11. SMES transients under bypass switching. same phase. It is a common practice to keep this delay large enough so that the transitions do not cause high transients. The values of the current sharing inductors could play a major role in the simulation results and the transients could be very different for different valves. The plot of the coil current at the terminal,, and its expanded inset are given in Fig. 10(c) and (c-1), respectively. 2) Transients Generated by Bypass Switch: The bypass switch is used to isolate the coil, and it also has the function of minimizing the chopper switching losses. In standby mode operation, the bypass switch is turned on (closed) in advance before the chopper is turned off so that the coil is properly disconnected from the chopper. On the other hand, if the chopper needs to be reconnected to the coil, the chopper must be turned on before the bypass switch can be turned off (opened). The firing signals which control bypass and chopper operations are shown in Fig. 11(a), where 1 corresponds to ON and 0 corresponds to OFF for either bypass or chopper. Failure of this sequence and improper selected time delay between chopper and bypass switching may result in extremely high transient over voltages. If these two aspects are properly considered, high transient voltages can be avoided as can be seen in Fig. 11(b). 3) Transients Generated by GTO Faults: Fig. 12 gives the transients caused by GTO faults. The faults are simulated by opening or short-circuiting one GTO branch at a specific time. The left column corresponds to the open-circuit case, and the right column corresponds to the short-circuit case. The open-circuit fault is initiated when GTO is conducting under normal chopper operation condition. Whereas, the short-circuit fault is created when GTO is not conducting under normal chopper operation conditions. Fig. 12. SMES transients caused by GTO faults. As expected, the open-circuit case results in higher transient voltages, while the short-circuit case causes increasing GTO current. As a consequence, the transient suppression schemes should be optimized to reduce voltage/current transients to acceptable levels. 4) Transient Suppression Schemes: Transients experienced in previous cases can be suppressed by the following schemes: adding surge capacitors along with ground resistors adding Metal Oxide Varistors (MOV) to limit over voltages tuning the current sharing inductances Studies showed that having those schemes applied, the terminal voltage and the current of the coil are reduced to the acceptable level [19]. VI. STACOM SMES COMBINATION SYSTEM A. STACOM SMES Experiment System Setup To utilize and verify the SMES model, an integrated system of a Static Synchronous Compensator (StatCom) with a SMES system is simulated for testing. A 100 MJ 96 MW (peak) SMES coil is attached to the voltage source inverter front end of a StatCom via the dc dc chopper. The real and reactive power responses of the integrated system-to-system oscillations are studied using the electromagnetic transient program (PSCAD/EMTDC). A StatCom is a second-generation flexible ac transmission system controller based on a self-commutated solid-state

9 CHEN et al.: DETAILED MODELING OF SMES SYSTEM 707 Fig. 13. Fig. 14. AC system equivalent. Detailed representation of statcom, dc dc chopper and SMES coil. voltage source inverter. However, it can only absorb/inject reactive power, and consequently is limited in the degree of freedom and sustained action. The addition of energy storage will allow the StatCom to inject and/or absorb active as well as reactive power simultaneously, therefore provides additional benefits and improvements in the system. The voltage source inverter front-end of a StatCom can be easily interconnected with an energy storage source such as a SMES coil via a dc dc chopper. The characteristics of a SMES system such as rapid response (Milliseconds), high power (Multi-MW), high efficiency, and four-quadrant control can help to meet the power industry s demands for more flexible, reliable and fast active power compensation devices. The simulation circuit representing the integrated ac system is shown in Fig. 13. The detailed representation of the StatCom, dc dc chopper, and SMES coil are depicted in Fig. 14. In this figure, the units of resistance, inductance, and capacitance values are Ohm, Henry, and Microfarad, respectively. 1) AC Power System: The ac system equivalent used corresponds to a two-machine system where one machine is dynamically modeled (including generator, exciter and governor) to demonstrate the dynamic oscillations. Dynamic oscillations are simulated by creating a three-phase fault in the middle of one of the parallel lines at Bus D in Fig. 13. A bus that connects the StatCom-SMES to the ac power system is named a StatCom terminal bus. The location of this bus is selected to be either Bus A or Bus B. 2) StatCom SMES System: As can be seen from Fig. 14, two-gto-based six-pulse voltage source inverters represent the StatCom used. The voltage source inverters are connected to the ac system through two 80 MW coupling transformers, and linked to a dc capacitor in the dc side. The value of the dc link capacitor has been selected as 10 mf in order to obtain smooth voltage at the StatCom terminal bus. The primary function of StatCom is to control the reactive power/voltage at the point of connection to the ac system. Basically, there are two fundamental control strategies for the dc bus voltage. The first one is basically to keep the constant dc bus voltage, therefore operating with a variable modulation of the inverter. The second will have the variable dc bus voltage and always operating with a modulation of near one [17]. The second control is basically a stabilizer control that regulates the SMES power according to the changes that may happen in the ac real power. Different control schemes will certainly have different effect on the performance of the system, which is not the focus of this paper. The SMES coil is connected to the VSI through a dc dc chopper. It controls dc current and voltage levels by converting the inverter dc output voltage to the adjustable voltage required across the SMES coil terminal. A two-level three-phase dc dc chopper used in the simulation has been modeled and controlled according to [19], [21]. The phase delay is kept at 180 degrees to reduce the transient over voltages. The average voltage of the SMES coil is related to the StatCom output dc voltage with the relationship displayed by (10). Where - is the average voltage across the SMES coil, - is the average StatCom output dc voltage, and d is duty cycle of the chopper (GTO conduction time/period of one switching cycle) [18]. Three measurements used in this Chopper-SMES control are: SMES coil current, ac real power measured at the StatCom terminal bus, and dc voltage measured across the dc link capacitor. B. Case Studies Several cases are simulated in order to demonstrate the effectiveness of the StatCom-SMES combination. A three-phase fault is created at Bus D in Fig. 13 to generate dynamic oscillations in each case. 1) StatCom-Only for Damping AC System Oscillations: A two-machine ac system is simulated as illustrated in Fig. 13. The inertia of the Machine-I is adjusted to obtain approximately 3 Hz oscillations from a three-phase fault created at time and cleared at time. When there is no StatCom-SMES connected to the ac power system, the system response is depicted above in the first column of Fig. 15 in the interval of 3 to 5 s. The first and second rows correspond to the speed of Machine-I and ac voltage at Bus B respectively. When a StatCom is connected to the system, the response is given in the second column of Fig. 15. Since the StatCom is used for voltage support, it may not be effective in damping the oscillations. 2) StatCom-SMES at Bus-B and Bus-A for Damping AC System Oscillations: In this case, 100 MJ 96 MW SMES coil is attached to a 160 MVAR StatCom through a dc dc chopper first at Bus B. The SMES coil is charged by making the voltage across its terminal positive until the coil current becomes 3.6 ka. Once it reaches this charging level, it is set at the standby mode. In order to see the effectiveness of the StatCom-SMES combination, the SMES activates right after the three-phase fault is cleared at 3.25 s.

10 708 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 21, NO. 2, APRIL 2006 Fig. 15. Dynamic response to ac system oscillations. Fig. 17. Real and reactive power responses of StatCom-SMES at different locations. Fig. 16. Dynamic response of StatCom-SMES to AC system oscillations. The dynamic response of the combined device to ac system oscillation is shown in the first column of Fig. 16. The first row shows the machine speed. The second row gives voltage at Bus B. When comparing with No StatCom-SMES and StatCom- Only cases in Fig. 15, we find out both frequency and voltage oscillations are damped out faster. Secondly, the StatCom-SMES combination case is re-simulated by connecting StatCom-SMES system to ac power system at Bus A near the generator bus. The same scenario above also applies to this simulation case. The results are shown in the second column of Fig. 16. Comparing with other two cases, the StatCom-SMES connected to a bus near generator shows effective results in damping electromechanical transient oscillations caused by a three-phase fault at Bus D. Then, the real and reactive power responses of the compensator to oscillations are also compared for different locations Bus A and Bus B. Fig. 17 shows the comparison results of the StatCom real and reactive power responses between these two cases (StatCom-SMES at Bus B, and StatCom-SMES at Bus A). When the StatCom-SMES is located at Bus B, it provides a voltage support by injecting approximately 50 MVA, and damps the oscillations. When the combined compensator is located at Bus A, no reactive power injection is necessary since Bus A voltage is fixed by the exciter of Machine-I. The differences between these two cases are clearly showed in Fig ) Comparison the Responses to the System Oscillation of the StatCom-SMES Combination at Different Locations: To study the response to system oscillations, the operation of SMES is also compared for different locations. The results are shown in Fig. 18. From top to bottom, SMES current, SMES terminal voltage, dc capacitor voltage and SMES energy are plotted in each row for different locations. SMES current/energy does not change abruptly, which is expected. Otherwise, transient over-voltages are observed at the terminal of SMES. SMES terminal voltage changes its polarity as the coil charges or discharges. The positive SMES voltage charges the SMES coil, which absorbs power from the ac system. Any variation on the StatCom ac terminal voltage is reflected to the dc capacitor voltage or the input voltage to the dc dc chopper as shown in the 3rd row of Fig. 18. From the above simulations, we find the disturbances caused by low frequency oscillations in the ac system can be damped

11 CHEN et al.: DETAILED MODELING OF SMES SYSTEM 709 The paper considered a transient modeling of a SMES coil and the chopper. The SMES coil is modeled as sections where each section is represented with its series capacitance, shunt capacitance, self and mutual inductances to other sections. The computation of models is developed to represent the entire coil to reduce computational effort. The voltage distribution and transient analysis of the SMES coil find the transient over voltages to SMES coil are generated during normal chopper operation, any open or short circuit fault on the chopper GTOs, and the bypass switch operation. This paper also presents the modeling and control of the integration of a StatCom with SMES system, and its dynamic response to system oscillations. It has been shown that the StatCom-SMES combination can be very effective in damping power system oscillations. More effective damping and faster stabilization of the system can be obtained if StatCom-SMES is located near a generation area rather than a load area. Adding energy storage enhances the performance of a StatCom and possibly reduces the MVA ratings requirements of the StatCom operating alone. This is important for cost/benefit analysis of installing FACTS controllers on utility systems. It should be noted that, the StatCom provides a real power flow path for SMES, but the SMES controller is independent of the StatCom Controller. While the StatCom is ordered to absorb or inject reactive power, the SMES is ordered to absorb/inject real power. REFERENCES Fig. 18. SMES operation as a response to ac system oscillations at different locations. by either injecting or absorbing the reactive/real power in the system. However, the reactive power injected to the system is dependent on the StatCom terminal voltage. On the other hand, the SMES operates according to the variation of the real power flow in the system. Damping power oscillations with injection of the real power may be more effective than reactive power since it does not affect the voltage quality of the system. VII. CONCLUSIONS This paper provides a detailed model for the simulation of the SMES system. The model is intended to provide guidelines for a detailed SMES device simulation in the power system, as well as to provide a basis for comparison of various simulation tools, control strategies, algorithms and realization approaches. [1] W. V. Hassenzahl, Superconducting magnetic energy storage, IEEE Trans. Magn., vol. 25, no. 2, pp , Mar [2] P. F. Ribeiro, SMES for enhanced flexibility and performance of FACTS devices, in Proc. IEEE Summer Meeting, Edmonton, AB, Canada, Jul [3] W. J. McNutt, T. J. Blalock, and R. A. Hinton, Response of transformer windings to system transient voltage, IEEE Trans. Power App. Syst., vol. PAS-93, no. 2, pp , Mar./Apr [4] R. C. Degeneff, A general method for determining resonances in transformer windings, IEEE Trans. Power App. Syst., vol. 96, no. 2, pp , Mar./Apr [5] P. A. Abetti and F. J. Maginniss, Natural frequencies of coils and windings determined by equivalent circuit, AIEE Trans., pt. III, vol. 72, pp , Jun [6] P. Chowdhury, Calculation of series capacitance for transient analysis of windings, IEEE Trans. Power Del., vol. PWRD-2, no. 1, pp , Jan [7] A. M. Miri, C. Sihler, M. Droll, and A. Ulbricht, Modeling the transient behavior of a large superconducting coil subjected to high voltage pulses, in Proc. Int. Conf. Power System Transients, Sep. 1995, pp [8] F. W. Grover, Inductance Calculations: Working Formulas and Tables. New York: Instrum. Soc. Amer., [9] A. Greenwood, Electrical Transients in Power Systems. New York: Wiley, [10] R. C. Degeneff, Calculating Voltage Versus Time and Impedance Versus Frequency for the SMES Coil: Babcock & Wilcox by Utility Systems Technologies, Inc, [11] B. Osman, Verification, validation, and testing, in The Handbook of Simulation, J. Banks, Ed. New York: Wiley, 1998, ch. 10, pp [12] K. A. Wirgau, Inductance calculation of an air-core disk winding, IEEE Trans. Power App. Syst., vol. PAS-95, no. 1, pp , Jan./Feb [13] R. L. Kustom, J. J. Skiles, J. Wang, K. Klontz, T. Ise, K. Ko, and F. Vong, Research on power conditioning systems for superconductive magnetic energy storage (SMES), IEEE Trans. Magn., pt. 4, vol. 27, no. 2, pp , Mar [14] R. H. Lasseter and S. G. Jalali, Power conditioning systems for superconductive magnetic energy storage, IEEE Trans. Energy Convers., vol. 6, no. 3, pp , Sep

12 710 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 21, NO. 2, APRIL 2006 [15] R. H. Lasseter and S. J. Jalali, Dynamic response of power conditioning systems for superconductive magnetic energy storage, IEEE Trans. Energy Convers., vol. 6, no. 3, pp , Sep [16] B. M. Han and G. G. Karady, A new power conditioning system for superconducting magnetic energy storage, IEEE Trans. Energy Convers., vol. 8, no. 2, pp , Jun [17] N. G. Hingorani and L. Gyugyi, Understanding FACTS Concepts and Technology of Flexible AC Transmission Systems. New York: IEEE Press, [18] I. D. Hassan, R. M. Bucci, and K. T. Swe, 400 MW SMES power conditioning system development and simulation, IEEE Trans. Power Electron., vol. 8, no. 3, pp , Jul [19] A. B. Arsoy, Z. Wang, Y. Liu, and P. F. Ribeiro, Electromagnetic transient interaction of a SMES coil and the power electronics interface, in Proc. 16th Annual VPEC Seminar, vol. 16, Sep , 1998, pp [20] A. B. Arsoy, Z. Wang, Y. Liu, and P. F. Ribeiro, Transient modeling and simulation of a SMES coil and the power electronics interface, IEEE Trans. Appl. Superconduct., vol. 9, no. 4, pp , Dec [21] M. Steurer and W. Hribernik, Frequency response characteristics of a 100 MJ SMES coil Measurements and model refinement, IEEE Trans. Appl. Superconduct., pt. 2, vol. 15, no. 2, pp , Jun