Design and Implementation of a 154 kV, ±50 MVAr Transmission STATCOM Based on 21-Level Cascaded Multilevel Converter

Burhan Gültekin*,¤, Cem Özgür Gerçek¤, Tevhid Atalık¤, Mustafa Deniz¤, Nazan Biçer¤,

Student Member Student Member Student Member Non- Member Non-Member Muammer Ermiş*, Nadir Köse¤, Cezmi Ermiş¤, Erkan Koç¤, Işık Çadırcı¤,†, Adnan Açık¤,

Member Non-Member Non-Member Non-Member Member Non-Member Yener Akkaya‡, Hikmet Toygar‡, Semih Bideci‡

Non-Member Non-Member Non-Member * Middle East Technical University, Ankara, TR ¤ TÜBİTAK-UZAY Institute, Ankara, TR †Hacettepe University, Ankara,TR ‡Turkish Electricity Transmission Co., Ankara,TR

{burhan.gultekin, cem.gercek, tevhid.atalik, mustafa.deniz, nazan.bicer, nadir.kose, erkan.koc, cezmi.ermis, adnan.acik, cadirci@uzay.tubitak.gov.tr}, ermis@metu.edu.tr, [yener.akkaya, hikmet.toygar, semih.bideci@teias.gov.tr]

Abstract -- In this research work, design and implementation of a 154 kV, ± 50 MVAr Transmission STATCOM (T-STATCOM) has been carried out primarily for the purposes of reactive power compensation and terminal voltage regulation, and secondarily for power system stability. The implemented T-STATCOM consists of five 10.5 kV, ±12 MVAr Cascaded Multilevel Converter (CMC) modules operating in parallel. The power stage of each CMC is composed of five series connected H-Bridges (HB) in each phase, thus resulting in 21-level line-to-line voltages. Due to modularity and flexibility of implemented HBs, a CMC module power density of 250kVAr/m3 is reached, thus making the mobility of the system implementable. DC link capacitor voltages of HBs are perfectly balanced by means of the Modified Selective Swapping Algorithm proposed. The field tests carried out at full load in the 154 kV transformer substation where T-STATCOM is installed and put into service have shown that the steady-state and transient responses of the system are quite satisfactory.

Index Terms-- T-STATCOM, Multilevel Converters, Terminal Voltage Regulation, Modified Selective Swapping

I. INTRODUCTION Flexible AC Transmission Systems (FACTS) are being

increasingly used in power systems, to enhance the system utilization and power transfer capacity as well as the stability, security, reliability and power quality of AC system interconnections. The STATCOM converters used in transmission system applications should control a few tens or hundreds of MVArs, operate at medium voltage levels and are connected to EHV or HV transmission system buses via specially designed coupling transformers. These are called Transmission STATCOMs (T-STATCOM) in the literature. Various Multilevel (MLC) topologies operating at switching frequencies at or near the This research and technology development work is carried out as a subproject of the National Power Quality Project of Turkey. Authors would like to thank to the Public Research Support Group (KAMAG) of the Scientific and Technological Research Council of Turkey (TUBITAK) for full financial support of the project.

supply frequency fit best to the needs of T-STATCOMs [1-6]. In T-STATCOM systems, the use of 6- pulse converters is not feasible owing to the need for series connected power semiconductor switches, and much higher switching frequencies for waveform synthesizing. Multilevel Converters for T-STATCOM applications are classified in three groups [1]: a) Diode Clamped Multilevel Converters (DCMC), b) Flying Capacitor Multilevel Converters (FCMC), c) Cascaded Multilevel Converters (CMC).

Equalization problem of the DC link capacitor voltages is being the common drawback of these MLC topologies [1-9]. Among these, CMC topology is based on H-bridge (HB) circuits connected in series and has the advantage of modularity, flexibility and lower component count in comparison with the others [1]. Commercial CMC-based T-STATCOMs employ either GTO or GCT/IGCT devices with inverse-parallel connected power diodes. To limit di/dt overvoltages on GTOs during turn-off operation, bulky snubber circuits were used [8-9]. To equalize individual capacitor voltages, high frequency IGBT based auxiliary circuits supplied from low voltage side were employed. Specially designed isolating transformers are to be used to isolate auxiliary circuits from the power stage of CMC. An important contribution to the solution of voltage equalization problem of DC link capacitors is known as the Selective Swapping Algorithm [7] which can be embedded in the control algorithm of CMC, thus eliminating the need for bulky auxiliary circuits. Although the effectiveness of voltage equalization algorithm decreases as the number of H-bridges connected in series and/or peak-to-peak ripple content of the capacitor voltage increase/s, it provides the lowest switching frequency for the power semiconductors.

In this research and technology development work, ±50 MVAr, 21-level CMC based T-STATCOM has been designed and implemented primarily for the purposes of reactive power compensation and terminal voltage regulation, and secondarily for power system stability. Since the operating voltage of CMC is chosen to be 10.5 kV l-to-l, it is connected to 154 kV l-to-l transmission bus

978-1-4244-5287-3/10/$26.00 ©2010 IEEE 3936

through a specially designed 50/62.5MVA YNyn coupling transformer. The power stage of the CMC is composed of five series connected H-bridges in each phase. The Selective Swapping Algorithm in [7] has been improved to balance DC link capacitor voltages perfectly at the expense of higher switching frequency, and hence switching losses. This implementation is the first industrial application of HV IGBT based CMC and the modified capacitor voltage balancing algorithm for T-STATCOM Technology. The CMC, its control system and the de-ionized water cooling system excluding the water-to-air heat exchanger are located on a specially designed trailer to equip T-STATCOM with the advantage of mobility in addition to modularity and flexibility.

II. SYSTEM DESCRIPTION Fig.1 shows the schematic diagram of a 154-kV, ±50-

MVAr T-STATCOM system. It is composed of five 10.5-kV, ±12-MVAr Cascaded Multilevel Converter (CMC) modules which are connected in parallel via air-core reactors. The combination of five STATCOM modules is then connected to 154kV bus via a 154/10.5kV,

50/62.5MVA,YNyn connected coupling transformer. It is a 21-level CMC based T-STATCOM and has been designed and implemented primarily for the purposes of reactive power compensation and terminal voltage regulation, and secondarily for power system stability. It is equipped with a digital control system except some analog interface and protection circuits. A master controller manages all control and most of the protection facilities by sending the necessary commands to the slave control units. All controllers are built up by using up-to-date Digital Signal Processing (DSP, TI-TMS320F28335) and Field Programmable Gate Array (FPGA, XILINX Spartan-3, 1,500,000-gate) technologies. 11-level line-to-neutral voltage and hence 21-level line-to-line voltage waveforms are created by using five series connected H-Bridges in each phase of the Y-connected CMC module. The developed ±50MVAr T-STATCOM has been installed in 380kV/154kV Sincan Transformer Substation, Ankara as shown in Fig.2. TABLE I summarizes the technical specifications of each CMC module and the resulting T-STATCOM system.

CT

VT

154/10.5 kV50/62.5 MVA

Uk= 16%

CouplingTransformer

10.5 kV Connection Bus

Pre-chargeResistor

Fiber Optics Communication Bus

CB-2 CB-3

154 kV Bus ( PCC )

Master DSPBoard

CT

O/H Lines(154kV Loads)

Istatcom

380kV Bus

IS

CB

380/154 kV250 MVAUk= 12%

380/154 kV150 MVAUk= 12%

O/H Lines

CouplingReactor-5

CT

+

FPGABoard-5

Slave DSPBoard-5Vpcc

CouplingReactor-4

CT

FPGABoard-4

Slave DSPBoard-4Vpcc

+

CouplingReactor-1

CT

+

FPGABoard-1

Slave DSPBoard-1Vpcc

CouplingReactor-3

CT+

FPGABoard-3

Slave DSPBoard-3Vpcc

CouplingReactor-2

CT

+

FPGABoard-2

Slave DSPBoard-2Vpcc

RCVT

Fig. 1. Simplified single line diagram of the T-STATCOM

3937

(a) Interior view of Trailer with Power Stages

(b) Control Cabinets

(c) Trailer for ±50 MVAr T-STATCOM on road

(WxLxH:3700mmx16800mmx4300mm)

(d) General view of T-STATCOM

Fig. 2. 154kV, ±50MVAr T-STATCOM based on 21-level Cascaded Multilevel Converters

The associated coupling transformer (ONAN/ONAF), high voltage (HV) and medium voltage (MV) switchgear equipment, input filter reactors and the water-to-air heat exchanger are installed outdoor (Fig.2d). The power stage, the control system and part of the de-ionized water cooling

system excluding water-to-air heat exchanger of the T-STATCOM are placed into a non-magnetic container mounted on a trailer (Fig. 2c). This makes possible easy relocation of the T-STATCOM to another problematic point of the transmission system whenever this is required.

3938

III. OPERATING PRINCIPLES

A. Reactive Power Control Simplified single-phase Y-equivalent model of the T-

STACOM is given in Fig.3. Equivalent series resistance of the coupling transformer and the input reactor of the CMC in Fig.3 is neglected to simplify the derivations for reactive power control. Phasor diagram for the resulting lossless system is drawn in Fig.4 according to power sink convention.

Fig. 3. Simplified single line diagram of T-STATCOM

where, Vs′ :RMS line-to-neutral AC grid fundamental voltage

referred to the T-STATCOM side Vc :RMS line-to-neutral T-STATCOM fundamental

voltage Ic : RMS T-STATCOM fundamental current Qs : PCC reactive power Qc : T-STATCOM reactive power δ : Phase-shift-angle between fundamental components

of STATCOM and AC grid voltages θ : Phase angle between fundamental components of

converter voltage and current R :Y-equivalent of total loss resistance including

coupling transformer losses, coupling reactor losses and converter losses

Xc :Y-equivalent of total reactance including leakage reactance of coupling transformer and reactance of input reactors

Since active power delivered by the supply at PCC, Ps, is equal to the active power absorbed by CMC, Pc, per phase active power flowing in the circuit, P, can be expressed in (1 ):

. .s c c cP P P V I cosθ= = = (1) Since Xc.Ic.cos θ =Vs

'.sin δ as marked in Fig .4, P can be expressed as in (2):

'.s c

c

V VP sinX

δ= (2)

where, δ is the power (load) angle.

Fig. 4. Phasor diagram for lossless system (exaggerated)

TABLE I T-STATCOM AND CMC SPECIFICATIONS

T-STATCOM Rated Power ±50 MVAr Rated Voltage 154 kV Coupling Transformer YNyn, 154/11,1 kV,

50/62.5 MVA, 18 %Uk Number of CMC Modules in Parallel

5

Cascaded Multilevel Converters (CMC) Rated Power ±12 MVAr Rated Voltage 10.5 kV Number of H-Bridges per phase 5 Number of Levels in Voltages 11 Levels Line-to-Neutral,

21 Levels Line-to-Line Power Semiconductor 3300V, 1200A HV IGBT with

parallel inverse diode Harmonics Eliminated by SHEM 5th, 7th, 11th, 13th Effective Switching Frequency 500Hz Coupling Reactors 2.5 mH, air core Rated DC Link Voltage 1900 Vdc, mean Cooling System De-ionized water cooling

During the operation of the CMC in the steady state, P is very low to supply only the CMC losses and hence δ takes a very small value (δ is around 0.017 rad ≡ 1 degree). For such small values of δ, sinδ ≈ δ holds and hence P can be approximated by (3):

'.s c

c

V VPX

δ= . (3)

The per phase value of the reactive power delivered by the supply at PCC, Qs, is as given by (4):

. . ( )s s cQ V I sin θ δ= + . (4) Since Xc.Ic.sin(θ+ δ)=Vs-Vc

'.cosδ, then '2 '.s s c

sc c

V V VQ cosX X

δ= − (5)

Since cosδ≡1 for very low values of δ, then Qs can be approximated to (6):

'' s c

s sc

V VQ VX

⎡ ⎤−= ⎢ ⎥⎣ ⎦

(6)

On the other hand, per phase value of the reactive power by CMC, Qc, which occurs for inductive operation mode can be expressed as in (7):

. .c c cQ V I sinθ= (7)

Since '. . .c c s cX I sin V cos Vθ δ= − , then Qc can be

approximated by'

s cc c

c

V VQ VX

⎡ ⎤−= ⎢ ⎥⎣ ⎦

. (8)

The difference between Qs and Qc is the reactive power absorbed by 2.c cI X which is always positive.

In summary, δ, P, Qs and Qc are positive and Qs>Qc, and θ is negative (≡ -π/2) for inductive mode of operation of T-STATCOM in the steady state. However, in capacitive mode of operation θ (≡+π/2), P is positive but δ, Qs and Qc are negative and Qs<Qc. The transition between capacitive and inductive mode of operations occurs when Qc=Ic

2.Xc which corresponds to unity pf operation. During the operation of the T-STATCOM, P is controlled by varying δ in (3) for full range hence keeping the DC link capacitor voltages constant at a pre-specified value over the entire

3939

operating range. On the other hand, Qs and Qc are controlled by varying Vc by PWM technique. If Vc is made smaller than Vs′, T-STATCOM operates in inductive operation mode as can be understood by (8). On the other hand, Vc is made sufficiently larger than Vs′, it starts to operate in capacitive operation mode and delivers reactive power to the supply. In practice, the situation is more complex than this because, the supply is not an infinite bus. That is the capacitive operation mode causes a rise in the supply voltage, Vs′, while the inductive operation, a drop.

B. Wave Shaping The input voltage waveform of voltage source converter

(VSC), vc(t), will be approximated to a pure sinusoidal voltage at supply frequency by using the CMC topology and Selective Harmonic Elimination Method (SHEM). It is well known that the number of voltage levels, n, in the staircase voltage waveform in Fig.5 produced by CMC is defined by (9)-(10):

n=2s+1 in the line-to-neutral voltage (9) n=4s+1 in the line-to-line voltage, (10)

where, s, is the number of HBs in one phase. As an example, in the T-STATCOM described in this paper, five HBs are connected in series in each phase to form a Y-connected multilevel converter topology. This CMC yields 11 voltage levels in the line-to-neutral voltage waveform and 21 voltage levels in the line-to-line voltage waveform. As illustrated in Fig.5, five pulses with different widths and the same magnitudes (Vdc) are to be superimposed in order to create a half-cycle of an 11-level line-to-neutral voltage waveform. This makes necessary assigning five different angles θ1,θ2,…,θ5 in such a way that the harmonic distortion of line-to-neutral voltage waveform will be minimum. 3rd harmonic and its integer multiples will not appear in the line-to-line voltage waveform. There remains only the elimination of dominant voltage harmonics which are the 5th, 7th, 11th and 13th. In a staircase waveform with odd-quarter symmetry, these are eliminated by using SHEM in the paper. If a higher harmonic content were allowed in line-to-line voltage waveform, a larger series filter reactor would be needed at the expense of voltage regulation problem. The expressions which are used in the calculation of optimum values of θ1,θ2,…,θ5 using a hybrid algorithm are given in Appendix. The hybrid algorithm is a combination of the genetic algorithm [10-11] and the gradient based method. First, the genetic algorithm is used for determination of proper initial conditions. Then, these initial conditions are applied to the gradient based method to reach the global minima much faster than the use of genetic algorithm only. The magnitude of the CMC fundamental output voltage can be controlled by adjusting modulation index, m, as given (11): m=(Vc

*/Vcmax), (11)

where 3 4max 2

V Vc dcπ= .

Vc*, is the rms value of fundamental line-to-line output voltage of CMC, and Vcmax denotes the maximum value of fundamental line-to-line rms voltage that can be produced by the CMC. Optimum values of θ1,θ2,...,θ5 are obtained for different modulation index values so that-50 MVAr to +50MVAr reactive power control range is divided into 150

Fig. 5. 11-level line-to-neutral voltage waveform [4]

steps and corresponding results are arranged as a 150x6 Look-up Table.

C. Voltage Balancing of DC Link Capacitors The major drawback of multilevel converters is the

voltage balancing problem of DC link capacitors [1-9]. An important contribution to the solution of voltage equalization problem of DC link capacitors is known as the Selective Swapping Algorithm [7] which can be embedded in the control algorithm of CMC. It is called the swapping algorithm because at each level change (θ1 to θ5 in Fig.5), the previous set of H-Bridge/s which are in operation is going to be interchanged with a new set in each phase of each CMC module. It is called selective swapping algorithm because the swapping operation is done according to the values of DC link capacitors. If the current and the voltage are both positive or negative, the input power (instantaneous) to the HB converter is positive and hence the associated DC link capacitor is going to be charged. On the other hand, if one of these quantities is positive while the other is negative, the input power to HB converter is negative and hence the associated DC link capacitor is going to be discharged. Therefore, in order to determine which HB converter/s are going to be operated at each level change, the values of individual DC link capacitor voltages, polarity of the voltage and direction of the current should be measured. The polarity of the input voltage to HBs is determined indirectly by using a PLL circuit which is locked to line-to-neutral voltage at PCC. The digital signal produced by PLL is phase advanced or delayed by power (load) angle, δ in Fig.4 to determine the zero crossing point of the CMC modules fundamental line-to-neutral voltage. After determining the new status of DC link capacitors (charging or discharging) by this way, HBs having the lowest DC link voltages will be put into operation for charging and HBs having the highest DC link voltages, for discharging mode of operations. This method is called the Conventional Selective Swapping (CSS) in the paper. The CSS gives minimum switching frequency for the power semiconductors. For the T-STATCOM described in this paper, the effective switching frequency (the number of turn-on in 1sec) is measured to be 200Hz and 250Hz in

3940

the field, respectively for capacitive and inductive operation modes. The major drawback of the CSS is the higher variations observed in the DC link capacitor voltages as the number of HBs in each CMC is reduced. In order to eliminate this drawback, the CSS is modified and then applied to T-STATCOM described in the paper. In the modified selective swapping algorithm, selective swappings are applied not only at level changes but continuously at a pre-specified frequency during the operation of the CMC. Fig.6 illustrates the application of the proposed method in comparison with the CSS. It is called the Modified Selective Swapping (MSS) in the paper. In the field tests of the T-STATCOM, the MSS has been applied at a pre-specified period of ∆ts=400usec and also of ∆ts=200usec, at expense of higher switching frequency and hence switching losses. The effects of MSS on voltage balancing problem and switching frequency will be discussed in the results section in comparison with those of CSS.

IV. DESIGN AND IMPLEMENTATION

A. Power Stage Fig. 8 shows the power stage of the ±12 MVAr, 10.5 kV,

21-level CMC module. The compact, and withdrawable, ±840 kVAr HB unit has a laminated L-shape busbar system with minimized stray inductance of Lstray=70nH and two pairs of 3300V, 1200A, wire-bond HV IGBTs (Mitsubishi, CM1200HB-66H) mounted on two cold plates (R-Theta). Fiber reinforced composites (FRC) were used as the construction materials for HBs to keep clearances at a minimum, and to have a strong mechanical structure. Further flexibility is achieved by quick couplings for water connection to cold plates and by using some special connectors for connection of fiber optics and the power supply cables entering to the front panel. HB units in each CMC module have been placed on the shelves which are built up of a specially designed Glass-fiber Reinforced Polyester (GRP) 3x5 matrix structure (Fig.8.b). To minimize the volume occupied by the ±60 MVAr converter system, five CMC modules have been located in a non-magnetic container (build up of PVC panels and joinery) permanently mounted on the trailer (Fig. 2c). By making these choices in the design, the power density is approximated to 250kVAr/m3, thus making mobility of the system implementable. A 700kW de-ionized water based cooling system (SwedeWater+Flexcoil) has been used for the power stage of the T-STATCOM.

B. Control System Simplified block diagram of the T-STATCOM control system is given in Fig.9. The T-STATCOM can be operated in either reactive power compensation mode (Q-mode) or terminal voltage regulation mode (V-mode) by using a selector switch. Operation at the set value of the bus voltage at PCC is achieved by close loop control which employs the digital implementation of a simple PI controller. The operation in Q-mode tends to reduce the reactive power flow in between 154kV bus and autotransformers in Fig.1 to zero. Elimination of 5th, 7th, 11th and 13th voltage harmonics in the line-to-neutral voltage waveforms according to SHEM is digitally

Fig. 6. The illustration of CSS and MSS

implemented by using a off-line prepared Look-up Table. The look-up Table contains information about angles θ1,θ2,...,θ5 (Fig.5) as a function of 150 discrete values of modulation index, m, to cover the entire operating range. Voltage balancing problem of DC link capacitors has been solved by applying the selective swapping algorithm. The operator can select the execution of either the CSS or the MSS with ∆ts=400usec (Fig.6). ∆ts of the MSS is programmable. Furthermore, each DC link capacitor is equipped with a parallel connected, chopper controlled discharge resistor in order to prevent the DC link from dangerous overvoltages. When the T-STATCOM is turned off, the chopper controlled discharge resistors are automatically activated to discharge the capacitors for safety. Active power delivered by the 154kV bus to the T-STATCOM supplies the losses of the coupling transformer, series reactors and CMCs in order to keep DC link capacitor voltages at their rated values (1900Vdc,mean).This is achieved by employing a digitally implemented close loop power (load) angle (δ in Fig.9) control. The T-STATCOM is also equipped with an automatic pre-charge circuit at 10.5kV side for cold start.

C. Further Design Objectives A smooth transition between different Q or V settings is aimed at the design of the control system. This will minimize the oscillations in current, voltage and reactive power in transition periods and hence reduces the settling time. The extreme operating condition is the transition from full capacitive to full inductive operation mode, or vice versa. For this purpose, first, the changes in the modulation index have been applied only at zero crossing points of the CMC line current. Second, the control system in Fig.9 is equipped with a feedforward controller block. The modification shown in Fig.10 is made on FPGA Board-3 in Fig.9. The gain of the feed-forward controller is determined as described below: Fig.11 shows the variations in power angle, δ, against modulation index, m, in the steady state for the control system having no feedforward block. These variations are obtained by a field test. The m values in Fig11 include all delays (phase lag) which amounts to nearly -2 degrees arising from conventional voltage transformers, PLL block, and etc.

3941

N

Vdc

1V

dc5

Vdc

4Vd

c3Vd

c2

Va

R

R

R

R

R

Vdc

1V

dc5

Vdc

4Vd

c3Vd

c2

Vb

R

R

R

R

R

Vdc

1V

dc5

Vdc

4Vd

c3Vd

c2

Vc

R

R

R

R

R

Fig. 8.a. Single line diagram of 21-level CMC

Fig. 8.b. Power stage of a 21-level CMC at MV Laboratory

1)L-shape Laminated Buses 2) HV IGBTs with drive circuits 3)Front panel with optical connectors 4) FRC-based structure of HB 5)Two

parallel connected DC link capacitors 6)Water pipe pairs with quick couplings 7) GRP-based structure of Power Stage

Fig. 9. Simplified block diagram of T-STATCOM control

system

The non-linear characteristic of δ is approximated by a straight line as marked on in Fig.11. The gain of the feedforward controller is chosen to be proportional to the slope of the approximated straight line in Fig.11. Fig.12 illustrates the benefits of modulation changes made at zero crossing points of line currents and those of the feedforward controller. These voltage and current records are made at the AC input of one of the CMCs. The Qset is varied from 40 MVAr capacitive to 40MVAr inductive. In both set of records, modulation changes are made at zero crossing points of line current waveforms. Fig.12.a is obtained for the control system having no feedforwad block, while Fig.12.b is the one of with feedforwrd controller. It is seen from Fig.12 that i)modulation index changes made at current zeros provide a smooth transition between two operating points, and ii) addition of feedforward controller reduces the settling time from 140ms to 100ms.

Fig. 10. Control system with feedforward block

3942

Fig. 11. The variations of power angle with modulation index

(a)

(b)

Fig. 12. The transient-free transitions of T-STATCOM (CMC side field Data)

V. FIELD TEST RESULTS

A. Voltage and Current Harmonics at PCC (154kV) Harmonic contents of the line-to-ground voltages at PCC with and without T-STATCOM are as given in TABLE II. These values are deduced from the measurements of resistive-capacitive voltage transducers (RCVT by TRENCH). The RCVT has a frequency bandwidth of

1MHz and can measure low order harmonic components with unity gain. It is seen from TABLE II that the operation of T-STATCOM at its rated capacity in both capacitive and inductive operation modes does not affect the harmonic content of the PCC voltage. Harmonic contents of the coupling transformer line current waveforms on the HV side (154 kV) are as given in TABLE III. Both individual voltage and current harmonic values, and THD and TDD values comply with IEEE Std. 519-1992.

B. Voltage Waveforms at the Input of CMC In this subsection, line-to-line and line-to-neutral voltage waveforms at the AC input terminals of a CMC will be given when it is producing rated MVAr in both the inductive and capacitive regions. Fig.13 shows the voltage waveforms when the CMC is operated with CSS and modified selective swapping (∆ts=400us and ∆ts=200usec).These waveforms show that the CMC successfully create 11-level line-to-neutral and 21-level line-to-line voltage waveforms regardless of the selective swapping technique applied. Harmonic contents of the line-to-neutral and the line-to-line voltage waveforms just at the input of the CMC are given in TABLE IV and TABLE V for rated inductive and capacitive operation modes, respectively. Following conclusions can be drawn from TABLE IV and TABLE V:

1. The SHEM applied in this research work successfully minimizes 5th, 7th, 11th and 13th harmonics in the line-to-neutral voltage waveforms.

2. Although the 3rd harmonic component is significant in the line-to-neutral waveforms, it becomes vanishingly small in line-to-line voltage waveform.

C. Effects of Series Reactors on Voltage Harmonics Harmonic contents of the line-to-ground and line-to-line voltage waveforms on the medium voltage side of the coupling transformer (after series connected input filter reactor) are as given in TABLE VI. These results show that voltage harmonics are filtered out by the input filter reactor to a certain extent. Harmonic contents of the line current on the medium voltage side of the coupling transformer are given in TABLE VII. Their magnitudes are relatively low, as expected.

D. Performance of Selective Swapping Methods The performance of Modified Selective Swapping method in balancing the DC link voltages is investigated by using field test results. The MSS is also compared with CSS in view of the instantaneous voltage peak-to-peak ripple, δVdc, and peak-to-peak mean dc link voltage ripple, ∆Vdc, of DC link capacitor voltage and effective switching frequency, fsw(eff) (number of turn-ons in 1 second per switch). Fig.14 shows the variations in instantaneous values of DC link capacitor voltage against time for conventional and modified selective swapping methods (sampling rate=1MHz). The variations in mean dc voltages (Vdc,mean) are also marked on the same waveforms (20ms averaged data).These results are summarized in TABLE VIII. The modified selective swapping algorithm reduces considerably both the instantaneous dc link voltage ripple and mean dc link voltage ripple at the expense of higher switching frequency losses for power semiconductors. It is not beneficial to reduce swapping period, ∆ts, considerably.

3943

TABLE II PCC VOLTAGE LINE-TO-GROUND HARMONICS

MEASURED BY RCVT SENSORS (MAXV=170KV)

Harmonic Number

No Operation Capacitive Mode

Inductive Mode

IEEEStd.519-

1992 (%)

Vb,g (%)

Vc,g (%)

Vb,g (%)

Vc,g (%)

Vb,g (%)

Vc,g(%)

1 100 100 100 100 100 100 1003 0,6 0,5 0,6 0,5 0,6 0,5 1,05 0,5 0,4 0,5 0,4 0,4 0,2 1,07 0,4 0,3 0,4 0,3 0,3 0,2 1,0

11 0,1 0,1 0,1 0,1 0,1 0,1 1,013 0,1 0,1 0,1 0,1 0,1 0,1 1,017 0,1 0,1 0,1 0,1 0,1 0,1 1,019 0,1 0,1 0,1 0,1 0,1 0,1 1,021 0,1 0,1 0,1 0,1 0,1 0,1 1,0

THD 0,9 0,9 0,9 1,5

TABLE III 154KV SIDE CURRENT HARMONICS

AT RATED POWER (IL=188A, ISC=20KA)

Harmonic Number

Capacitive Mode Inductive Mode IEEE Std.519-1992

(%) Ia

(%) Ib

(%) Ic

(%) Ia

(%) Ib

(%) Ic

(%) 1 100 100 100 100 100 100 1003 0,2 0,2 0,2 0,2 0,2 0,2 3,05 0,4 0,4 0,4 0,4 0,4 0,4 3,07 0,2 0,2 0,2 0,2 0,2 0,2 3,0

11 0,1 0,1 0,1 0,1 0,1 0,1 3,013 0,1 0,1 0,1 0,1 0,1 0,1 1,517 0,5 0,5 0,5 0,5 0,5 0,5 1,1519 0,3 0,3 0,3 0,3 0,3 0,3 1,1521 0,1 0,1 0,1 0,1 0,1 0,1 1,15

TDD 0,77 0,77 3,75

TABLE IV CMC VOLTAGE HARMONICS AT +10MVAR

Harmonic Number Va,g(%) Vb,g(%) Vc,g(%) Vab(%) Vbc(%) Vca(%)

1 100 100 100 100 100 1003 12,6 12,2 12,0 0,1 0,1 0,25 0,5 0,4 0,5 0,4 0,4 0,5 0,3 0,2 0,3 0,2 0,2 0,3

11 0,1 0,2 0,1 0,2 0,2 0,113 0,5 0,6 0,6 0,5 0,6 0,517 3,0 3,0 3,0 3,0 3,0 3,019 2,8 2,8 2,8 2,8 2,8 2,821 2,0 2,1 2,1 0,1 0,1 0,1

TABLE V CMC VOLTAGE HARMONICS AT -10MVAR

HarmonicNumber Va,g(%) Vb,g(%) Vc,g(%) Vab(%) Vbc(%) Vca(%)

1 100 100 100 100 100 1003 10,7 10,4 10,4 0,1 0,1 0,15 0,7 0,5 0,5 0,6 0,5 0,67 0,2 0,1 0,2 0,2 0,2 0,1

11 0,1 0,1 0,2 0,1 0,1 0,113 0,4 0,4 0,5 0,4 0,5 0,417 3,3 3,3 3,3 3,3 3,3 3,319 3,1 3,1 3,1 3,1 3,1 3,121 0,1 0,1 0,1 0,1 0,1 0,1

Fig. 13. Voltage waveforms of a CMC under (a) conventional selective swapping (CSS) and (b) modified selective swapping (MSS) methods

(a) CSS method (Field Data)

(b) MSS method with ∆ts =400 μsec (Field Data)

3944

TABLE VI 10.5KV SIDE VOLTAGE HARMONICS MEASURED BY RCVT SENSORS

Harmonic Number

Capacitive Mode Inductive ModeVa,g(%) Vb,g(%) Vc,g(%) Va,g(%) Vb,g(%) Vc,g(%)

1 100 100 100 100 100 1003 0,3 0,2 0,2 0,3 0,1 0,25 0,2 0,1 0,1 0,4 0,3 0,57 0,1 0,1 0,2 0,1 0,1 0,1

11 0,1 0,1 0,2 0,1 0,1 0,113 0,1 0,1 0,1 0,1 0,1 0,117 2,0 2,0 2,1 1,7 1,7 1,719 2,3 2,2 2,2 2,0 1,9 1,921 0,1 0,1 0,1 0,1 0,1 0,1

TABLE VII CMC CURRENT HARMONICS AT RATED POWER

HarmonicNumber

Capacitive Mode Inductive ModeIa(%) Ib(%) Ic(%) Ia(%) Ib(%) Ic(%)

1 100 100 100 100 100 1003 1,2 1,2 1,2 2,0 2,0 2,05 0,5 0,5 0,5 0,4 0,4 0,47 0,6 0,6 0,6 0,2 0,2 0,2

11 0,3 0,3 0,3 0,1 0,1 0,113 0,2 0,2 0,2 0,1 0,1 0,117 0,8 0,8 0,8 0,5 0,5 0,519 0,9 0,9 0,9 0,2 0,2 0,221 0,1 0,1 0,1 0,1 0,1 0,1

Fig. 14. Variations in the DC link voltage of an H-Bridge under (a) conventional selective swapping (CSS) and, (b) and (c) modified selective swapping (MSS) methods

TABLE VIII PERFORMANCE OF SELECTIVE SWAPPING ALGORITHMS

Applied Method

Inductive Capacitive CSS MSS with 400us MSS with 200us CSS MSS with 400us MSS with 200us

Instantaneous Voltage Ripple (peak-to-peak) 190V 145V 130V 205V 170V 190V Mean Voltage Ripple (peak-to-peak) 30V 14V 16V 50V 23V 27V

Effective Switching Frequency 250Hz 500Hz 650Hz 200Hz 500Hz 700Hz

(a) CSS method (Field Data)

(b) MSS method with ∆ts =400 μsec (Field Data)

(c) MSS method with ∆ts =200 μsec (Field Data)

3945

(a)

(b)

Fig. 15. Variations in the DC link mean voltages of a CMC ( Field Data) The swapping period, ∆ts, should not exceed 400usec. The variations in DC link capacitor voltages in one phase of anyone of the five CMC modules are also recorded. A typical record (120usec averaged data) is as shown in Fig15. The instantaneous variations in the first H-bridge

capacitors in all phases of the same CMC module are also recorded as shown in Fig15. These results show that the modified selective swapping algorithm and its digital implementation balance the DC link capacitor voltages perfectly.

E. Terminal Voltage Regulation The 154kV bus-bar to which the T-STATCOM is

connected is a strong bus (5300MVA,SC). Therefore, the effect of T-STATCOM on bus-bar voltage in V-mode does not exceed ±1% depending upon the power demand of the loads supplied from the same bus. The performance of the T- STATCOM in V-mode is illustrated in Fig.16.

F. Reactive Power Compensation The records in Fig.17 show the performance of the T-

STATCOM in Q-mode. In the first 5-minute part of the record, the T-STATCOM brings the power factor to unity as viewed from the 154kV side of the autotransformers. For this purpose, T-STATCOM produces nearly 45MVAr- inductive. When T-STATCOM is operated in standby

(a) (a)

(b) (b)

(c) Fig. 16. The performance of T-STATCOM in V-mode (Field Data)

(c) Fig. 17. The performance of T-STATCOM in Q-mode (Field Data)

3946

mode to produce 0MVAr, the autotransformers start to deliver nearly 27 MVAr-capacitive to the PCC. This reactive power flow is not equal to the 45MVAr inductive reactive power previously generated by the T-STATCOM. This is attributed to the fact that since the bus voltage at PCC increases nearly 2kV line-to-line after operating of T-STATCOM is standby mode, the overall response of overhead lines, power plants and loads connected to the same bus is non-linear.

G. Transitions between Full Capacitive and Full Inductive Modes

The reactive power variations on the 154kV side and 10.5kV side of the coupling transformer are given when Qset of T-STATCOM is varied from +50 MVAR (inductive) to -50MVAr (capacitive) in Fig.18 and Fig.19, respectively. There is a 10 MVAr (inductive) difference between the reactive power values of 154kV and 10.5kV sides. This is the reactive power consumption of the coupling transformer. As can be seen from these waveforms in Fig.18 and 19, the settling time of the closed loop control system is only 80ms. On the other hand, when Qset is varied from -50 MVAr (capacitive) to +50MVAr (inductive), the response of the closed loop control system will be as seen in Fig.21 and 22. For this case, the settling time is measured to be 100ms. It is worth to note that if the T-STATCOM were operated in open loop control loop, the

settling time would not exceed 45ms. The line-to-neutral voltage waveform and the associated line current waveform on the 10.5kV side of the coupling transformer during transition from Qset =+50 Mvar to -50MVAr with the closed loop control system are given in Fig.20. The associated waveforms from Qset=-50MVAr to +50MVAR are shown in Fig.23. The responses in Fig.20 and 23 show the smooth and fast response of the control system against changes in Qset from full inductive to full capacitive, or vice versa.

VI. CONCLUSIONS This is a research and technology development project

devoted to the design and implementation a 154 kV, ±50 MVAr Transmission STATCOM based on 21-level cascaded multilevel converters, and is primarily dedicated to reactive power compensation and terminal voltage regulation. It is demonstrated by field tests that the developed system can contribute to terminal voltage regulation of a 5300 MVAsc, 154 kV bus by an amount of ±1%. This is the first industry application of HV IGBT based CMC in T-STATCOM applications. Five ±12 MVAr, 10.5 kV CMC modules are operated in parallel by the specially designed fully digital control system composed of DSP-FPGAs.

Fig. 18. Transition from full inductive to full capacitive (10.5kV side, field data)

Fig. 21. Transition from full capacitive to full inductive (10.5kV side, field data)

Fig. 19. Transition from full inductive to full capacitive (154kV side, field data)

0.0Time (sec)

Rea

ctiv

e Po

wer

(MVA

r)

0.5 1.0 1.5 2.0 2.5-50-40-30-20-1001020304050

Time (ms)20 120-50

0

50

Ts=100ms

-50MVAr

+50MVAr

Fig. 22. Transition from full capacitive to full inductive

(154kV side, field data) 10 1.0

Line

-to-N

eutr

al V

olta

ge (k

V)

Line

Cur

rent

(kA

)

Time (ms)0 20 40 60 80 100 120 140 160

-10-8-6-4-202468 0.8

0.60.40.20.0-0.2-0.4-0.6-0.8-1.0

Ts=80ms

Fig. 20. Variations in line-to-ground voltage and current during the transition from full inductive to full capacitive (10.5kV side, field data)

Time (ms)0 20 40 60 80 100 120 140 160

-10-8-6-4-202468

10 1.00.80.60.40.20.0-0.2-0.4-0.6-0.8-1.0

Ts=100ms

Fig. 23. Variations in line-to-ground voltage and current during the transition from full capacitive to full inductive (10.5kV side, field data)

3947

A Modified Selective Swapping (MSS) algorithm has

been applied by the control system to reduce the instantaneous dc link voltage ripple of H-bridges by an amount of 25%, and the mean dc link voltage ripple by an amount of 50%, as compared to the Conventional Selective Swapping (CSS) algorithm, thus resulting in perfect balancing of the H-bridge dc link voltages, as the expense of higher semiconductor switching losses. Furthermore, owing to the application of the Selective Harmonic Elimination Method (SHEM), the 5th, 7th, 11th, and 13th voltage harmonics are successfully minimized. It has been shown by field tests that the total harmonic distortion (THD) of the line-to-neutral voltage waveform at the point of common coupling is as small as 0.9% with respect to the fundamental component. The total demand distortion (TDD) has been measured as 0.77%, in the 154 kV bus. The transient response of the developed system is also found to be quite satisfactory. The settling time of the T-STATCOM at full-load from inductive mode to capacitive mode transition, and vice versa has been recorded as 80 ms, and 100 ms, respectively in the field, with a transient free transition between modes. The modularity of H-bridges for easy replacement in the case of a failure, and the mobility and flexibility of the overall T-STATCOM for easy relocation when necessary are being further advantages of the developed system.

APPENDIX The SHEM expressions which are used in the calculation of optimum values of θ1,θ2,…,θ5 using a hybrid algorithm are given as follows: cos( ) cos( ) cos( ) cos( ) cos( )51 2 3 4 mθ θ θ θ θ+ + + + = , (A.1)

cos(5 ) cos(5 ) cos(5 ) cos(5 ) cos(5 ) 051 2 3 4θ θ θ θ θ+ + + + = , (A.2)

cos(7 ) cos(7 ) cos(7 ) cos(7 ) cos(7 ) 051 2 3 4θ θ θ θ θ+ + + + = , (A.3)

cos(11 ) cos(11 ) cos(11 ) cos(11 ) cos(11 ) 051 2 3 4θ θ θ θ θ+ + + + = , (A.4)

cos(13 ) cos(13 ) cos(13 ) cos(13 ) cos(13 ) 051 2 3 4θ θ θ θ θ+ + + + = (A.5)

REFERENCES [1] Lai J.-S., Peng F.Z., “Multilevel Converters-A New Breed of Power

Converters”, IEEE Trans. Ind. App., vol.32, No.3, pp.509-517, May/June 1996.

[2] Peng F.Z., McKeever J.W., Adams D.J., “Cascade Multilevel Inverters for Utility Applications”, IECON 97, vol.2, pp.437-442

[3] Patil K.V., Mathur R.M., Jiang J., Hosseini S.H., “Distribution System Compensation using a new Binary Multilevel Voltage Source Inverter”, IEEE Trans. on Power Delivery, vol.14, No.2, pp.459-464,

[4] Tolbert L.M. , Peng F.Z., Habetler T.G., “Multilevel Converters for Large Electric Drives”, IEEE Trans.on Industrial Apps., vol.35, No.1, pp.36-44, Jan./Feb.1999

[5] Peng F.Z., Lai J.-S., McKeever J.W.,VanCoevering J., “A Multilevel Voltage-Source Inverter with Separate DC Sources for Static Var Converters”, IEEE Trans. Ind. App., vol.32, No.5, pp.1130-1138, Sept/Oct 1996

[6] Sirisukprasert S., Lai J.-S., Huang A.Q., “Modeling, Analysis and Control of Cascaded Multilevel Converter Based STATCOM”, Power Engineering Society General Meeting, 2003, IEEE , vol.4, pp.2561-2568

[7] Sirisukprasert S., “Modeling and Control of a Cascaded-Multilevel Converter-Based STATCOM”, PhD Dissertation, February 2004

[8] Ainsworth J.D., Davies M., Fitz P.J., Owen K.E.,Trainer D.E., “Static VAr Compensator (STATCOM) Based on Single-phase

Chain Circuit Converters”, IEE Proc. on Generation Transmission Distribution, Vol.145, No.4, July 1998

[9] Hanson D.J., Horwill C., Gemmel B.D.Monkhouse D.R., “A STATCOM Based Relocatable SVC Project in the UK for National Grid”, IEEE Power Engineering Society, Winter Meeting, 2002

[10] Ozpineci,B., Tolbert L.M., Chiasson, J.N., “ Harmonic Optimization of Multilevel Converters Using Genetic Algorithms”, IEEE Power Electronics Letter, vol.3, No.5, pp.92-95, Sept.2005

[11] Tolbert L.M. et al., “Modulation Index Regulation of a Multilevel Inverter for Static Var Compensation”, IEEE Power Engineering Society General Meeting,vol.1, pp.194-199, July 2003

3948