Uninterruptible Power Supplies: Classification, Operation, Dynamics, and Control

Stoyan B. Bekiarov and Ali Emadi

Power Electronics and Motor Drives Laboratory Electrical and Computer Engineering Department Illinois Institute of Technology Chicago, IL 60616-3793 Phone: (312) 567-8940; Fax: (312) 567-8976 EML: emadi@iit.edu URL: http://www.ece.iit.edu/~emadi/
AbstractThis paper presents a comprehensive review of Uninterruptible Power Supply (UPS) systems in terms of topologies, operation, dynamics, and control. UPS systems are classified with emphasis on static systems. This paper also addresses fundamental problems faced in these systems in different distributed and centralized applications. In addition, a brief description of the conventional UPS systems, their disadvantages, scope for improvement, advanced architectures, role of power electronics, and current trends in UPS research are explained. Negative impedance dynamics in UPS systems with constant power loads are also introduced. Finally, this paper concludes with a brief outline of the advancement to be made in the future.

operation and related control strategies are addressed in Section V. In Section VI, concept of negative impedance instability in UPS systems with constant power loads is introduced. Section VII deals with the control techniques suitable for these systems. At last, Section VIII summarizes the results obtained in this paper. II. CLASSIFICATION UPS systems are classified to three general types of static, rotary, and hybrid static/rotary systems. In this section, we explain these three categories of UPS systems. A. Static UPS Static UPS systems are the most commonly used UPS systems. They have a broad variety of applications from low power personal computers and telecommunication systems, to medium power medical systems, to high power utility systems. Their main advantages are high efficiency, high reliability, and low THD. The inherent problems related to static UPS systems are poor performance with nonlinear and unbalanced loads and high cost for achieving very high reliability. On-line, off-line, and line-interactive configurations are the main types of static UPS systems. 1) On-line UPS: On-line UPS systems consist of a rectifier/charger, a battery set, an inverter, and a static switch (bypass). Other names for this configuration are inverterpreferred UPS and double-conversion UPS [1]-[3]. Fig. 1 shows the block diagram of a typical on-line UPS. The rectifier/charger continuously supplies the DC bus. Its power rating is required to meet 100% of the power demanded by the load as well as the power demanded for charging the battery bank. The batteries are rated in order to supply power during the back up time. The duration of this time varies in different applications. The inverter is rated at 100% of the load power since it must supplies the load during the normal mode of operation as well as during the back up time. It is connected in series with the load; hence, there is no transfer time associated with the transition from normal mode to stored energy mode. This is the main advantage of on-line UPS systems. The static switch provides redundancy of the power source in the case of UPS malfunction or overloading.

I. INTRODUCTION Uninterruptible Power Supply (UPS) systems provide uninterrupted, reliable, and high quality power for vital loads. They, in fact, protect sensitive loads against power outages as well as overvoltage and undervoltage conditions. UPS systems also suppress line transients and harmonic disturbances. Applications of UPS systems include medical facilities, life supporting systems, data storage and computer systems, emergency equipment, telecommunications, industrial processing, and on-line management systems [1][9]. Generally, an ideal UPS should be able to deliver uninterrupted power and, simultaneously, provide the necessary power conditioning for the particular power application. Therefore, an ideal UPS should have the following features: regulated sinusoidal output voltage with low Total Harmonic Distortion (THD) independent from the changes in the input voltage or in the load, on-line operation that means zero switching time from normal to back-up mode and vice versa, low THD sinusoidal input current and unity power factor, high reliability, high efficiency, low EMI and acoustic noise, electric isolation, low maintenance, and low cost, weight, and size. In this paper, classification, operation, dynamics, and control of UPS systems are reviewed. In Section II, UPS systems are classified to static, rotary, and hybrid static/rotary systems. Static UPS systems are explained in details. In Section III, distributed and centralized applications are presented. Performance evaluation of UPS systems is given in Section IV. Principles of parallel

0-7803-7404-5/02/$17.00 (c) 2002 IEEE

597

The AC line and load voltages must be in phase in order to use the static switch. This can be achieved easily by a locked-phase control loop. During normal mode of operation, the power to the load is continuously supplied via the rectifier/charger and inverter. In fact, a double conversion, i.e., AC/DC and DC/AC, takes place. It allows a good line conditioning. The AC/DC converter charges the battery set and also supplies power to the load via the inverter. Therefore, it has the highest power rating in this topology increasing the cost. When the AC input voltage is outside the preset tolerance, the inverter and battery maintain continuity of power to the load. The duration of this mode is the duration of preset UPS back-up time or till the AC line returns within the preset tolerance. When the AC line returns, a phaselocked loop makes the load voltage in phase with the input voltage and after that the UPS system returns to normal operating mode. The main advantages of on-line UPS are very wide tolerance to the input voltage variation and very precise regulations of output voltage. In addition, there is no transfer time during the transition from normal to stored energy modes. It is also possible to regulate or change the output frequency [3]. The main disadvantages of this topology are low power factor, high THD at the input, and low efficiency. The input current is destroyed by the rectifier unless an extra Power Factor Correction (PFC) circuit is added; but, this adds to the cost of the UPS system [4].
S t a tic S w itc h (B y p a s s )

available at all. During this mode of operation the power to the load is supplied by the battery set via the inverter for the duration of the preset back up time or till the AC line is back again. The duration of the switching time depends on the starting time of the inverter. The transfer time is usually about of the line cycle, which is enough for most of the applications such as personal computers. The main advantages of this topology are simple design, low cost, and small size. The line conditioning, when there is such, is passive and, hence, the technique is very robust. On the other hand, lack of real isolation of the load from the AC line, no output voltage regulation, long switching time, and poor performance with nonlinear loads, are the main disadvantages.
AC Line Static Switch (Normaly on) Load

AC/DC Rectifier/ Charger

DC/AC Inverter

Battery Bank

Fig. 2. Block diagram of a typical off-line UPS.

AC L in e

A C /D C R e c tif ie r

D C /A C In v e rte r

Load

B a tte ry B a n k

Fig. 1. Block diagram of a typical on-line UPS.

3) Line-interactive UPS: Line-interactive UPS systems, as is shown in Fig. 3, consist of a static switch, a series inductor, a bi-directional converter, and a battery bank. An optional output filter can be added at the output of the bidirectional converter or at the input side of the load. A lineinteractive UPS can operate either as an on-line UPS or as an off-line UPS. For an off-line line-interactive UPS, the series inductor is not required. However, most of the lineinteractive UPS systems operate on-line in order to either improve the power factor of the load or regulate the output voltage for the load.
AC L in e
S e rie s In d u c to r

2) Off-line UPS: Off-line UPS configuration is also known as standby UPS or line-preferred UPS [1], [2]. As is shown in Fig. 2, it consists of an AC/DC converter, a battery bank, a DC/AC inverter, and a static switch. A filter may be used at the output of the UPS or inverter to improve the quality of the output voltage. The static switch is on during the normal mode of operation. Therefore, load is supplied from the AC line directly without any power conditioning. The AC/DC converter charges the battery set. It is rated at a much lower power rating than the rectifier/charger in an on-line UPS since it is not required to meet the power demand of the load. This, in turn, allows the off-line UPS systems to be cheaper than on-line UPS systems. The inverter is rated at 100% of the load demand. It is connected in parallel to the load and stays standby during the normal mode of operation. It is turned on only when the primary power is out of a given preset tolerance or is not

S ta tic S w itc h

Load

B i-d ire c tio n a l A C /D C C o n v e rte r

B a tte ry B a n k

Fig. 3. Block diagram of a typical line-interactive UPS.

When the AC line is within the preset tolerance, it feeds the load directly. The AC/DC converter is connected in parallel with the load and charges the battery. It may also supply the reactive power required to keep the power factor close to unity or to regulate the output voltage [1], [5]. In stored-energy mode of operation, the bi-directional converter operates as an inverter and supplies the load from the battery

598

set. The static switch disconnects the AC line in order to prevent back feed from the inverter. The main advantages of the line-interactive UPS systems are simple design and, as a result, high reliability and lower cost compared to on-line UPS systems. They also have good harmonic suppression for the input current. Since this is a single stage conversion topology, the efficiency is higher than that of the double-conversion UPS. The main disadvantage is the lack of effective isolation of the load from the AC line. Employing a transformer in the output can eliminate this; but, it will add to the cost, size, and weight of the UPS system. Furthermore the output voltage conditioning is not good because the inverter is not connected in series with the load. In addition, since the AC line supplies the load directly during the normal mode of operation, there is no possibility for regulation of the output frequency. The new series-parallel line-interactive topology, called delta-conversion UPS, can achieve simultaneously both unity power factor and precise regulation of the output voltage, which is not possible with a conventional lineinteractive UPS [6]. Its configuration is shown in Fig. 4. It consists of two bi-directional converters connected to a common battery set, static switch, and a series transformer. The series bi-directional converter is rated at about 20% of the output power of the UPS and it is connected via a transformer in series with the AC line. The second bidirectional converter is the usual inverter for a lineinteractive UPS connected in parallel to the load and rated at 100% of the output power.
SeriesTransformer AC Line Static Switch Load

of this UPS is relatively high. Therefore, the deltaconversion UPS is used in high power rating applications. However, the complicated control of this topology limits its applications. B. Rotary UPS A typical rotary UPS is shown in Fig. 5. It consists of an AC motor, a DC machine, an AC generator, and a battery bank. Electric machines are mechanically coupled. There are two operating modes: normal and stored energy. During the normal mode of operation, the AC line supplies the AC motor, which drives the DC machine. The DC machine drives the AC generator, which supplies the load. During the stored energy mode of operation, the battery bank supplies the DC machine, which, in turn, drives the AC generator. The AC generator supplies the load. The rotary UPS systems are much more reliable than the static UPS systems. Yet, they require more maintenance and have much bigger size and weight.
Static Switch (Bypass)

AC Motor

DC Machine

AC Generator

AC Line

M/G

Load

Battery Bank

Fig. 5. Block diagram of a typical rotary UPS.

C. Hybrid Static/Rotary UPS

Series Bi-directional Converter DC Link Parallel Bi-directional converter

battery set

Fig. 4. Block diagram of a typical series-parallel line-interactive UPS.

Hybrid static/rotary UPS systems combine the main features of both static and rotary UPS systems. They have low output impedance, high reliability, excellent frequency stability, and low maintenance requirements [7], [8]. In Fig. 6, a typical hybrid static/rotary UPS is depicted. It consists of a bi-directional AC/DC converter, an AC motor, an AC generator, a battery bank, and a static switch. Hybrid UPS systems are usually used in high power applications.

Parallel converter keeps output voltage stable and precisely regulated by PWM control. Series converter compensates any differences between output and input voltages. It also controls the input power factor to unity and, at the same time, controls the charging of the battery. When the AC line is within the preset tolerance, most of the power is supplied directly from the AC line to the load. Only a small portion of the total power, usually up to 15%, flows through the series and parallel converters. This power is needed to compensate for any differences between the input and the output voltages and to make the input power factor unity. Since an important portion of the power flows without any conversion from the AC line to the load, the efficiency

Static Switch (Bypass)

AC Motor
AC Line
Switch

AC Generator

Load

Bi-directional AC/DC Converter

Battery Bank

Fig. 6. Block diagram of a typical hybrid static/rotary UPS.

599

III. APPLICATIONS UPS systems not only supply power during the blackouts, but also, in most cases, provide power conditioning. They are now not only highly desirable, but also a required standard in many cases. A. Distributed Approach In a distributed approach, many separate UPS units operate in parallel to supply critical loads. UPS units are placed flexibly in the system to form a critical load network. A typical distributed UPS system is shown in Fig. 7.

UPS

Load

AC Line

Static Switch

Secure Network

UPS

UPS

Load

Load

trained group of staff will be in charge of maintenance and service reducing the risk of misuse of the system and increasing its reliability. Employing line-interactive topology in centralized UPS systems can give some additional useful functions apart from ensuring the continuity of the power supply. By using a parallelconnected UPS to control the real and reactive power drawn from the AC line, one can use it for load leveling, voltage regulation, harmonic suppression, or reactive compensation. Capability of line-interactive UPS systems to control real and reactive power flow from the AC line is very valuable from power system operation and control point of view. Utilities use parallel UPS systems as Battery Energy Stored Systems (BESS) for load leveling, voltage stabilizing, frequency control, and active filtering [10]. As a load leveling device, BESS helps to cover the peak power demanded by the loads. BESS can also work as a frequency control device if its capacity is large enough compared to the overall power of the system. It can supply additional real power to the system in the case of generator outage in order to stabilize the frequency. IV. PERFORMANCE EVALUATION

UPS

UPS

Load

Fig. 7. Block diagram of a typical distributed UPS system.

Distributed UPS systems have many advantages. They are highly reliable because of the redundancy. In fact, it is an advantage to achieve the so called N+1 redundancy in distributed UPS systems, where N UPS units supply the load and one additional unit stays in reserve. They are also highly flexible to increase the capacity of the system, when the load grows, by simply adding additional units. The problems associated with distributed UPS systems are the following. First, the load sharing between separate units has to be very fast and precise because of the fast dynamic response of each module and the low overload limit of a typical static UPS. In practice, this load sharing is difficult to achieve and require very fast and complicated digital controllers. The second disadvantage is that the monitoring of the whole system is difficult and requires specially trained staff [9]. The distributed approach is expected to be more attractive for highly proliferated loads such as medical equipment, data processing, and telecommunications. B. Centralized Approach In centralized approach, one large UPS unit supplies all the critical loads. Main goal is to ensure the continuous operation of the whole system rather than the operation of specific critical loads. This approach is more desirable for industrial and utility applications. The problems here are associated with the high relative cost for achieving redundancy and increasing of capacity with load expansion. Centralized approach offers a clear advantage in respect to maintenance and service of UPS systems. A specially

Performance of a UPS system is evaluated in terms of several factors. First and the most important factor is the quality of the output voltage. The output voltage should be sinusoidal with a low THD, usually less than 5% even when feeding non-linear or unbalanced loads. In addition, UPS systems should have good transient response towards sudden changes in the load. They should also be capable of maintaining the regulation of the output voltage. Considering the power electronic topologies, on-line UPS configuration gives superior performance followed by line-interactive UPS and finally by off-line UPS. In on-line UPS system, the inverter is connected in series with the load in both normal and stored-energy operating modes and continuously provides power conditioning to the load. In line-interactive UPS systems, the inverter is connected in parallel with the load and during the normal mode provides only limited power conditioning to the load. Only during the stored-energy mode of operation, the performance of lineinteractive UPS can be the same as that of the on-line UPS. The second important factor in terms of which performance of UPS systems is evaluated is the power factor of the input current. For higher power applications, active Power Factor Correction (PFC) techniques are used. In these techniques either a PWM rectifier is controlled in PFC technique [11] or an additional DC/DC PFC circuit is used [12]. These active techniques are applicable only for on-line and off-line UPS topologies. For line-interactive UPS systems, PFC is achieved by controlling the power angle between the input and output voltages [13]. Another important factor for evaluating performance of UPS systems is the transfer time. In this aspect, the on-line UPS topology is superior. Its configuration naturally ensures

600

that the load is continuously supplied with power without any transfer time. The transfer time in line-interactive UPS systems depends on the time necessary for converting the power flow from the battery bank through the inverter to the load. Improved performance is achieved by choosing the DC bus capacitor voltage at the battery side to be slightly higher than the floating voltage of the batteries. Therefore, when the AC line fails, it is not necessary to sense the failure because the DC bus voltage will immediately fells under the floating voltage of the batteries and the power flow will naturally turn to the load. For off-line UPS systems, the transfer time is the longest. It depends upon the speed of sensing the failure of the AC line and starting the inverter. V. PARALLEL OPERATION A typical on-line UPS consists of an AC/DC rectifier, a battery bank, and a DC/AC inverter. The concept of parallel operation can be applied to rectifiers alone, to inverters alone, or to the whole UPS. As shown in Fig. 8, two or more AC/DC rectifiers can operate in parallel to supply the DC link with rectified voltage/current. On the other side of the DC link, one or more inverters supply the critical loads.
Static Switch (Bypass)

from a common DC link, supply loads connected to a critical bus [15]. In addition, in a system with several UPS units operating in parallel, UPS modules are close to each other and parallel operation is achieved by communication between separate units [16]-[19].
AC/DC Rectifier AC Line 1 AC/DC Rectifier DC/AC Inverter DC/AC Inverter AC/DC Rectifier AC/DC Rectifier AC Line 2 AC/DC Rectifier DC/AC Inverter Secure DC Ring Bus DC/AC Inverter Load Load

DC/AC Inverter

Load

AC/DC Rectifier Battery Bank

Fig. 9. Block diagram of a typical UPS with parallel configurations forming a ring DC bus.
VDC Vnom VDC Vnom Vdrop

AC/DC Rectifier

DC/AC Inverter

Vdrop

AC Line

AC/DC Rectifier

DC/AC Inverter

Load

I1

I 1'

IDC1

I2

I 2'

IDC2

AC/DC Rectifier

Battery Bank

Fig. 10. Voltage drop characteristics of the rectifier groups in Fig. 9.

DC/AC Inverter

A. Control Strategies in Parallel Operated UPS Systems

Fig. 8. Block diagram of a typical UPS with several rectifiers and inverters in parallel.

The concept of parallel operation can be taken further, as shown in Fig. 9, where two separate rectifiers, formed a DC ring bus, fed from separate AC lines [14]. The load sharing is achieved by built-in voltage drop. In fact, changes in DC voltage are used as communication signals. Principle of operation of the UPS system in Fig. 9 can be explained by using Fig. 10. A change in load causes a corresponding change in DC voltage. For example, a load increase causes a decrease in DC voltage. Rectifiers work in voltage control mode to increase their currents in order to restore the proper voltage level. The overall increase in current (I=I1-I2) is distributed between the separate rectifiers by the build-in voltage drops. It can be seen in Fig. 10 that the steeper the voltage drop, less load will be taken from the rectifier. This configuration is used mostly in industrial applications where many different critical loads have to be supplied. The concept of parallel operation can also be applied to inverters. In this configuration, two or more inverters, fed

1) Concentrated Control: In this control method, as shown in Fig. 11, a Phase Locked Loop (PLL) in each inverter is used to synchronize its output voltage in frequency and phase with the voltage of the critical bus ensuring proper real power load sharing between separate inverters. A parallel control unit detects the total load current I, divides it by number of inverters n in the system, and passes the signal I/n to each inverter. This signal is used as a reference for the minor current control loop, which takes as a feedback signal the current through the output filter inductor. Both signals are compared and an error is obtained, which is used to compensate any inequality in voltage amplitude of different inverters. Therefore, a proper reactive power load sharing is achieved [19]. 2) Master-Slave Control: Basically, the master-slave control is the same as the concentrated control. The difference is that only the master unit has enabled the PLL which is in charge of providing constant sinusoidal output voltage synchronized in frequency and phase with the

601

critical bus voltage. The other inverters work as slaves, which are controlled to track the reference current provided by the parallel control unit. The parallel control unit has the same function as in a concentrated control method. Hence, the reference current derives from load current; so, it can serves as a frequency and phase reference as well. As a result, the slave units do not need to have separate PLL circuits enabled. If the master unit fails, one of the slave units replaces it by simply enabling its PLL circuit [15], [17]-[19].
Critical Bus AC Line
Voltage Regulator fo UPS #1 + V* PWM Inverter f* PLL

VI. NEGATIVE IMPEDANCE DYNAMICS In general, there are two kinds of loads for UPS systems. One group is conventional positive incremental impedance loads that require a regulated voltage for their operation. The other group is constant power loads sinking constant power from their input power supply. Power electronic converters, when tightly regulated, behave as constant power loads. In fact, there is a trend in the power electronic loads of UPS systems to be constant power. On the other hand, several loads such as electric motors, actuators, and power electronic converters have to be controlled such that constant output power is maintained for them. If we neglect the losses, input power is constant. As a result, these loads present constant power characteristics to the UPS systems. An example of constant power loads is a DC/AC inverter which drives an electric motor and tightly regulates the speed when the rotating load has one-to-one torquespeed characteristic. Usually, a controller tightly regulates the speed; therefore, the speed is almost constant. Since the rotating load has one-to-one torque-speed characteristic, for every speed there is one and only one torque. As a result, for a constant speed, torque is constant and power, which is the multiplication of speed and torque, is constant. If we assume a constant efficiency for the drive system, considering the constant power of the rotating load, the input power of the DC/AC inverter will be constant. Therefore, the DC/AC inverter presents a constant power load characteristic to its input power supply. Another example is a DC/DC converter which feeds an electric load and tightly regulates the voltage when the load has one-to-one voltage-current characteristic. In constant power loads, although the instantaneous value of impedance is positive; but, the incremental impedance is always negative. In fact constant power loads have negative impedance characteristics, which might impact the power quality and stability of the UPS systems. Current through a constant power load decreases/increases when the voltage across it increases/decreases. This is a destabilizing effect for the power supply system and known as negative impedance instability. Fig. 12 depicts the negative impedance behavior of constant power loads. Because of the non-linearity and time-dependency of power converters and because of the negative impedance destabilizing characteristics of constant power loads, classical linear control methods, which are often used to design controllers for UPS systems, have stability limitations around the operating points. In [20]-[22], we have presented an assessment of the negative impedance instability concept of the constant power loads in the multi-converter power electronic systems. Furthermore, recommendations on the design of power supply systems to avoid negative impedance instability have been given. Different stabilizing control methods have been proposed in [20]-[22] to ensure large-signal stability.

Vr +

fcr + I1

Load
I/n + In

Vr + fo UPS #n

Voltage Regulator +

V*

PWM Inverter f*

PLL

- fcr

Fig. 11. Block diagram of a typical concentrated control strategy.

In the above two described methods, a failure in parallel control circuit can cause a whole system failure. This is prevented in the distributed control method. Each inverter sends signals to the others. Compensating voltage and frequency signals are sent to each inverter [19]. 3) Wireless Independent Control: In this method, there are no inner-connective load sharing wires between inverters. Control is based on information available locally at the inverter terminals: voltage, current, and frequency. Each inverter has a build-up frequency drop characteristic for real power load sharing and a build-up voltage drop characteristic for reactive power load sharing [9], [16], [19]. It is well known that when there is an unbalance in generated and consumed real power in the system, there is a change in the system frequency. Hence, the frequency can be used as a communication signal between two inverter connected in parallel. For example, if the real power demanded by the two inverters in parallel is increased by P=P1+P2, than the frequency will initially drop by f. The increase of power P will be taken by separate inverters accordingly to the slope of their build-up frequency drop. New steady state equilibrium will be achieved at a lower frequency. In order to restore the initial frequency the whole drop characteristic has to be shifted up vertically. A proper active power load sharing control is then implemented without communication needs between inverters.

602

V=

P = Cons tan t I V <0 I

Fig. 12. The negative impedance behavior of constant power loads.

loops: an outer and an inner. The outer control loop uses the output voltage as a feedback signal, which is compared with a reference signal. The error is compensated by a PI integrator to achieve stable output voltage under steady-state operation. This error is also used as a reference signal for the inner current regulator loop, which uses the inductor [28] or the capacitor [29] output filter current as the feedback signal. The minor current loop ensures fast dynamic responses enabling good performances with nonlinear or step-changing loads.

VII. CONTROL TECHNIQUES THD of the output voltage and dynamic responses of a UPS system are among the most important performance features of the UPS. They depend mostly on the control strategy applied to the UPS inverter. UPS control strategy as well as UPS power electronic topology determine the overall performance of the UPS systems. A. Single Voltage Control Loop Strategy This strategy, as shown in Fig. 13, uses a single feedback loop to provide well-regulated output voltage with low THD. The feedback control can be continuous [23] or discontinuous [24], [25]. Analogue techniques are used in continues approach. The most usual continuous feedback control is Sinusoidal PWM (SPWM), which can be of natural sampling type, average type, or instantaneous type [26], [27].
Vr + K(s) PWM Modulator DC/AC Inverter L V o C Load
V* + PI + Regulator IL* Io Current Regulator IL DC/AC Inverter

Vo Load

Fig. 14. Block diagram of a typical control system with multiple feedback loops.

Fig. 13. Block diagram of a single feedback loop.

In natural sampling type, the peak value of the output voltage is detected and compared with a reference voltage in order to obtain the error, which is used to control the reference to the modulator. The average approach is basically the same; but, the sensed voltage is converted to an average value and after that is compared with a reference signal. These approaches control only the amplitude of the output voltage and are good only at high frequencies. In instantiations voltage feedback SPWM control, the output voltage is continuously compared with the reference signal improving dynamic performances of the UPS inverter [27]. B. Multiple Control Loop Strategy Better performance even with non-linear and stepchanging loads can be achieved by multiple control loop strategies [28]. As shown in Fig. 14, there are two control

The basic current regulators employed as minor current loop are: hysteresis regulators, sinusoidal PWM regulator, and predictive regulators. In a typical hysteresis regulator, the reference signal is compared with the feedback signal. The sign and predetermined amplitude of the error determine the output of the modulator, which has two possible levels +/-Vout. The duration between two successive levels is determined by the slope of the reference signal. The output voltage tracks the reference signal within the upper and lower boundary levels [30], [31]. This hysteresis control has fast transient response; but, the switching frequency varies widely. In SPWM control technique, the output voltage feedback is compared with a sine reference signal and the error voltage is compensated by a PI-regulator to produce the current reference. The current through the inductor or the capacitor is sensed and compared with the reference signal. After compensated by a P-regulator, the error signal is compared with a triangular waveform to generate SPWM signal for switching control. The SPWM current control has constant switching frequency and also provides fast dynamic responses [29], [32]. In predictive current control method, the switching instants are determined by suitable error boundaries. When the current vector touches the boundary line, the next switching state vector is determined by prediction and optimization in order to minimize the error [33]. Predictive current control requires a good knowledge of the load parameters. All these current regulators are typically used as an inner loop to regulate the current in the filter inductor. The current reference for the current regulator is obtained by summing together the error in an outer voltage loop with the actual load current to yield the rated output voltage.

603

VIII. CONCLUSIONS A general classification of UPS systems has been explained. A brief description of the topology, operation, and applications has been presented with emphasis on static UPS systems. Basic principles of parallel operation have also been explained. In addition, different configurations as well as basic control strategies for UPS systems operated in parallel have been described. Effects of constant power loads on the stability and dynamics of UPS systems have been introduced as well. Furthermore, UPS control systems have been reviewed. A brief description of the two main control techniques of single voltage control loop and multiple control loops has been presented. The basic current regulators employed as a minor current loop in multiple control strategies, such as hysteresis regulators, sinusoidal PWM regulator, and predictive regulators, have been described. REFERENCES
[1] [2] [3] [4] F. Kamran and T. G. Habetler, A novel on-line UPS with universal filtering capabilities, IEEE Trans. on Power Electronics, vol. 13, no. 2, pp. 366371, March 1998. R. Krishnan, and S. Srinivasan, Topologies for uninterruptible power supplies, in Proc. IEEE International Symposium on Industrial Electronics, Hungary, June 1993, pp. 122127. S. Karve, Three of a kind, IEE Review, vol. 46, no. 2, pp. 2731, March 2000. W. J. Ho, J. B. Lio, and W. S. Feng, Economic UPS structure with phase-controlled battery charger and input-power-factor improvement, IEE Proc. Electric Power Applications, vol. 144, no. 4, pp. 221226, July 1997. J. C. Wu and H. L. Jou, A new UPS scheme provides harmonic suppression and input power factor correction, IEEE Trans. on Industrial Electronics, vol. 42, no. 6, pp. 2216-226, Dec. 1995. S. Rathmann and H. A. Warner, New generation UPS technology, the delta conversion principle, in Proc. 31st IEEE Industry Applications Society Annual Meeting, Oct. 1996, pp. 23892395. A. Windhorn, A hybrid static/rotary UPS system, IEEE Trans. on Industry Applications, vol. 28, no. 3, pp. 541545, May/June 1992. W. W. Hung and G. W. A. McDowell, Hybrid UPS for standby power systems, Power Engineering Journal, vol. 4, no. 6, pp. 281291, Nov. 1990. M. C. Chandorkar, D. M. Divan, Y. Hu, and B. Banerjee, Novel architecture and control for distributed UPS systems, in Proc. 9th IEEE Applied Power Electronics Conference, Feb. 1994, pp. 683689. M. T. Tsai, C. E. Lin, W. I. Tsai, and C. L. Huang, Design and implementation of a demand-side multifunction battery energy storage system, IEEE Trans. on Industrial Electronics, vol. 42, no. 6, pp. 642-652, Dec. 1995. T. Shimizu, T. Fujita, G. Kimura, and J. Hirose, A unity power factor PWM rectifier with DC ripple compensation, IEEE Trans. on Industrial Electronics, vol. 44, no. 4, pp. 447-455, Aug. 1997. C. Qiao and K. M. Smedley, A topology survey of single-stage power factor corrector with a boost type input-current-shaper, in Proc. 15th IEEE Applied Power Electronics Conference, Feb. 2000, pp. 460-467. W. J. Ho, J. B. Lio, and W. S. Feng, Economic UPS structure with phase-controlled battery charger and input-power-factor improvement, IEE Proc. Electric Power Applications, vol. 144, no. 4, pp. 221226, July 1997. B. K. Johnson and R. Lasseter, An industrial power distribution system featuring UPS properties, in Proc. 24th IEEE Power Electronics Specialists Conference, June 1993, pp. 759-765.

[5] [6] [7] [8] [9] [10]

[11] [12]

[13]

[15] J. Holtz, W. Lotzkat, and K. H. Werner, A high-power multitransistor-inverter uninterruptible power supply system, IEEE Trans. on Power Electronics, vol. 3, no. 3, pp. 278-285, July 1988. [16] M. C. Chandorkar, D. M. Divan, and R. Adapa, Control of parallel connected inverters in standalone AC supply systems, IEEE Trans. on Industry Applications, vol. 29, no. 1, pp. 136-143, Jan./Feb. 1993. [17] J. Holtz, W. Lotzkat, and K.-H. Werner, Multi-inverter UPS system with redundant load sharing control, IEEE Trans. on Industrial Electronics, vol. 37, no. 6, pp. 506 -513, Dec. 1990. [18] J. F. Chen and C. L. Chu, Combination voltage-controlled and current-controlled PWM inverters for UPS parallel operation, IEEE Trans. on Power Electronics, vol. 10, no. 5, pp. 547-558, Sep. 1995. [19] D. Shanxu, M. Yu, X. Jian, K. Yong, and C. Jian, Parallel operation control technique of voltage source inverters in UPS, in Proc. IEEE 1999 International Conference on Power Electronics and Drive Systems, July 1999, pp. 883-887. [20] A. Emadi and M. Ehsani, Negative Impedance Stabilizing Controls for PWM DC/DC Converters using Feedback Linearization Techniques, in Proc. 35th Intersociety Energy Conversion Engineering Conference, Las Vegas, Nevada, July 2000. [21] A. Emadi and M. Ehsani, Dynamics and control of multi-converter DC power electronic systems, in Proc. IEEE 2001 Power Electronics Specialist Conference, Vancouver, British Columbia, Canada, June 2001. [22] A. Emadi, A new negative impedance stabilizing control technique for switching power supplies with constant power loads, in Proc. 2001 International Conference on Power Electronics, Seoul, Korea, Oct. 2001. [23] N. M. Abdel-Rahim and J. E. Quaicoe, Analysis and design of a multiple feedback loop control strategy for single-phase voltagesource UPS inverters, IEEE Trans. on Power Electronics, vol. 11, no. 4, pp. 532-541, July 1996. [24] A. Kawamura, R. Chuarayapratip, and T. Haneyoshi, Deadbeat control of PWM inverter with modified pulse patterns for uninterruptible power supply, IEEE Trans. on Industrial Electronics, vol. 35, no. 2, pp. 295-300, May 1988. [25] O. Kukrer, Deadbeat control of a three-phase inverter with an output LC filter, IEEE Trans. on Power Electronics, vol. 11, no. 1, pp. 1623, Jan. 1996. [26] D. M. Divan, Inverter topologies and control techniques for sinusoidal output power supplies, in Proc. 6th IEEE Applied Power Electronics Conference, March 1991, pp. 81-87. [27] T. Kagotani, K. Kuroki, J. Shinohara, and A. Misaizu, A novel UPS using high-frequency switch-mode rectifier and high-frequency PWM inverter, in Proc. 20th IEEE Power Electronics Specialists Conference, June 1989, pp. 53-57. [28] C. D. Manning, Control of UPS inverters, in IEE Colloquium on Uninterruptible Power Supplies, 1994, pp. 3/1-3/5. [29] N. M. Abdel-Rahim and J. E. Quaicoe, Multiple feedback loop control strategy for single-phase voltage-source UPS inverter, in Proc. 25th IEEE Power Electronics Specialists Conference, June 1994, pp. 958-964. [30] S. Sae-Sue, V. Kinnares, C. Tangsiriworakul, and S. Potivejkul, Comparative performance evaluation of fixed and adaptive hysteresis band delta modulation techniques for UPS, in Proc. IEEE 1999 International Conference on Power Electronics and Drive Systems, July 1999, pp. 956960. [31] N. M. Abdel-Rahim and J. E. Quaicoe, A single-phase deltamodulated inverter for UPS applications, IEEE Trans. on Industrial Electronics, vol. 40, no. 3, pp. 347-354, June 1993. [32] H. Wu, D. Lin, D. Zhang, K. Yao, and J. Zhang, A current-mode control technique with instantaneous inductor-current feedback for UPS inverters, in Proc. 14th IEEE Applied Power Electronics Conference, March 1999, pp. 951-957. [33] J. Holtz, Pulsewidth modulation for electronic power conversion, Proceedings of the IEEE, vol. 82, no. 8, pp. 1194-1214, Aug. 1994.

[14]

604