You are on page 1of 6

Suitability of Excitation Systems for PSS Application Roger E. Beaulieu and Murray E. Coultes Goldfinch Power Engineering Inc.

The Early Days Power system stabilizers (PSS) have now been with us for several decades. Papers by Larsen & Swann [1], De Mello, Hannett and Undrill [2], Bayne, Watson & Lee [3], and several others [4] to [9], have covered the problems associated with the application of stabilizers on hydraulic and thermal units from the late 1960s to the late 1970s. The problems discussed in those papers covering the early years of this technology, had to do with the actual design of the stabilizer hardware, the choice of signal, the implementation problems, the interaction between the stabilizer signal and shaft torsionals, effects of loading and unloading on terminal voltage, and of course, the acceptability of the excitation systems themselves, as perceived at that time. Gradually, the theoretical concepts advanced by people such as de Mello overcame most of the application problems, leading to the present situation where the Integral-of-Accelerating-Power stabilizer has become the de facto standard for power system stabilizers, at least in North America. This type of PSS was installed on hundreds of small and large units of all types, hydraulic, thermal, nuclear and gas turbines. The period from the early 1980s to the present saw several excitation system manufacturers incorporate the latest design of PSS into their digital excitation systems. There were no secrets about the structure; it had been published in IEEE papers and in the 1992 IEEE 421.5 Standard. Ontario Hydro, an active player in this technical area for decades, made the design available to anyone that wanted it and in fact worked with several manufacturers to commercialize the design, either as a built-in feature of digital exciters or as a stand alone device. By and large, this engineering phase met with success. The problems associated with the design and implementations of PSS were fairly well documented in the literature or in internal utility reports. For example, the interaction between the stabilizer speed signal derivation and the presence of shaft torsionals in the speed signal merited a lot of attention [9]. The suitability of the excitation systems for use with PSS was, however, left undocumented in large part. A number of authors, [1], [2], etc., discussed the need for compensating for the phase lag through the system but very little was said about how much compensation you could apply before running into problems. Some found out the hard way that there is limit to the amount of phase lead that can be used. The WSCC, in 1992, published an excellent report on Criteria to Determine Excitation System Suitability for PSS in WSCC System[10 ]. This report prepared by Baj L. Agrawal of Arizona Public Service Company, did an excellent job of identifying a major factor in determining the suitability of an excitation system, namely the speed of response of the overall system. The WSCC 1992 Report The WSCC 1992 report provided a benchmark transfer function against which the excitation system/power system combination could be compared. If the system was slower than the benchmark transfer function, it was deemed not suitable for installation of a PSS. The work presented in this paper is still valid, for the conditions for which it is presented. At the time, the author considered the frequency range of interest to be 0.1 to 1 Hz. This range is suitable for inter-area modes of oscillations, a prime concern of the WSCC. For general application, particularly where one is dealing with low-inertia machines, it is necessary to deal with a wider bandwidth, up to at least 3 Hz. The WSCC report dealt with the speed of response of the exciter/power system and did not go into the many other problems associated with the installation of PSS. It was believed, by many utility people, that the hardware implementation problems were the responsibility of the exciter/PSS manufacturers and they would solve those problems. That may be one of the reasons why, out of the many early PSS that were installed, a large number ended up not commissioned, or else commissioned but quickly turned off. Manufacturers were not always in a position to resolve all the interfacing, mounting, noise and performance problems as they did not necessarily have control over all the factors. Gradually, as new exciters came on the market in the late 1970s to the early 1990s things improved and PSS could be applied, where required, without worrying too much about suitability or performance. Then came the deregulation of the electricity market. The Situation Today The break up of large utilities, the layoffs or early retirements of experienced manpower, the highly competitive gas turbine market, and the arrival of manufacturers not familiar with large power system requirements, all in a short time span in the early to mid 1990s, all contributed to a sudden rash of application

problems with PSS. Many of these new problems can be attributed to the installation of low-cost, limitedfeatures exciters, originally intended for small isolated industrial power systems. Since the mid 1990s engineers charged with tuning excitation systems and PSS have been struggling with some or all of the following issues: Response time or pilot-exciter/rotating exciter systems Low-cost pilot exciters with lack of negative field forcing Incorrect internal AVR structure that does not allow for a proper PSS connection Crude digitization process that results in noisy performance or wrap-around numerical problems Unnecessary filtering stages added for unknown reasons Unsuitability of internal filtering that causes large phase shift over a narrow range of frequency Interaction between the underexcitation limiter and PSS signal Loss of PSS when UEL comes into play Lack of access to internal variables to verify suitability of settings (H-constant in Delta-POmega) Range and resolution of settings available Difficulties of deriving accurate computer models What Led to This Situation? In the past, the North American power industry consisted of large utilities with a number of expert staff who knew something about exciters, stability and performance requirements. Excitation systems were obtained from large established manufacturers, who had knowledgeable staff that could provide customers with what they needed, even if the customers did not always know what they needed. Now, particularly in the area of gas turbine generators, many are turnkey projects with little or no design involvement from the owner-operator. Several of us have been repeatedly called in at the last minute to commission PSS and model excitation systems, because of regulatory requirements. This is when we find out that the PSS and/or exciters are really not suitable. No one is directly at fault. It is a matter of not having accurate specifications for the exciters and PSS and expected performance goals.

What follows is discussion of the newly created problems, and possible solutions. Response Time In modern excitation systems, obtaining the desired response time it is mostly a matter of setting the PID controllers in the automatic voltage regulator. Rotating exciters are generally designed with much shorter time constants than those seen several decades ago. Nevertheless, systems made up of a pilot exciter, or pilot AVR, and a rotating exciter, require special considerations. Attempting to overcome too much lag in the exciter, by adding phase lead to the PSS, can result in a very noisy PSS signal. In many instances, it is necessary to resort to three stages of phase lead. Though, on paper, this approach works correctly, applying this amount of phase lead, in practice, can lead to a noisy field voltage, noisy reactive power out of the unit, objections by the operators and often disconnection of the PSS once the commissioning staff leaves the station. It becomes essential that the PSS make use of all information available, namely using a minimum of two phase currents to derive the electrical power signal and the frequency signal. All unnecessary delays must be eliminated for the correct application of PSS on rotatingtype excitation systems. The challenge is particularly difficult on smaller machines located near noisy load centers. Smaller machines are generally equipped with the slowest exciters, necessitating the most lead compensation, but due to noisy loads, the PSS is almost ineffective due to the need to use low gains. In such instances, it may be better to use a significant amount of transient gain reduction on the excitation system and avoid the use of PSS altogether. Lack of Bi-Directional Field Forcing The industry has seen, over the last decade, the installation of hundreds of gas turbines in the range of 30 to 100 MW, equipped with low-cost analog or digital pilot exciters feeding rotating brushless main exciters. These pilot exciters are often equipped with half-controlled full-wave bridges. These bridges can only produce positive voltages, so no negative field forcing is possible. Unfortunately, due to regulatory requirements, most of these systems must have PSS installed. During large disturbances, in which the PSS signal can exceed 0.05 pu of the terminal voltage reference, the result is a generator overvoltage. Figure 1 is a simulation of an external fault. The first part of the record is for a full-controlled bridge; the second part is a half-controlled bridge. The half-controlled bridge produces a generator terminal voltage increase of over 10%. The settings for PSS on these systems must consider this effect. Typically, the output limits of the PSS are set at +/-0.05 pu, instead of the recommended +/-0.1 pu.

80 Exciter Field Voltage, VR 60 40 20 0 -20 -40 -60 7.0 Generator Field Voltage 6.0 5.0 4.0 3.0 2.0 1.0 0.0 1.2 Generator Term inal Voltage 1.1 1.0 0.9 0.8 0.7 0.6 1.3 1.2 1.1 1.0 Pow er 0.9 0.8 0.7 0.6 0.5 0.4 0.15 0.10 0.05 PSS 0.00 -0.05 -0.10 -0.15 0 1 2 3 4 Time (s) 5 6 7 8

In this system, the voltage regulator, and the limiters, cannot differentiate between the terminal voltage signal and the PSS signal, since they are summed together and appear as a combined signal. If the stabilizer now fails, with its output going against the limit values (if one is lucky and the output does not reach the power supply level), the initial effect is that the voltage regulator senses a sudden drop or a sudden increase in terminal voltage. The regulator action will be to bring the sensed signal back to its proper value. This, in turn, can result in a large increase or decrease in generator terminal voltage. Say the net effect is a large increase in terminal voltage. In the type of excitation systems we are used to, the limiters, such as the overvoltage or V/Hz limiter will be useless since the signal they sense is within the proper range of values. In the case of a large decrease in terminal voltage, the underexcitation limiter will come into play since it senses reactive loading. It will, however be fooled by the sensed terminal voltage signal since it will contain the offset value from the PSS. Consequently, the normal recalibration of the limit line will not occur and the limiting level will not be where it is intended to be.
Terminal Voltage PSS Signal Extra Input
Input Signals.DWG

+ve OP. AMP. -ve

Convert to Digital Values

To Regulator and Limiters

Figure 1: Effect of Lack of Bi-directional Forcing PSS Interfacing Problems The marketing of low-end excitation systems, not designed for use with stabilizers, has resulted in some installations with serious shortcomings. For decades now, the PSS signal has effectively been brought into the summing junction with the generator terminal voltage, the reference signal and possible other signals such as the V/Hz limiter, the Underexcitation limiter and the reactive current compensation. This of course requires that there be a spare or dedicated input for the stabilizer signal. In some systems, this spare input does not exist and it became necessary to sum the PSS signal with the generator terminal voltage signal. By inverting the polarity of the PSS signal, one achieves a workable system, at least with respect to the PSS operation. The above approach, however, leads to some unexpected problems with limiters and output limits on the stabilizer. Consider the simplified representation shown in Figure 2.

Figure 2: Inappropriate Summation of PSS Signal Crude Digitization Process that Results in Noisy or Faulty Performance If we were to compare an analog excitation system to a digital version, one could say that the analog system uses an infinite number of bits in the digitization process and an extremely high sampling rate. Small or large signals can therefore be accommodated without problems. If the signals are too large, then well-defined limits are reached, either created by zener diodes or the power supply levels. In a digital system, it is necessary to make compromises on the number of bits to use, the sampling rate and the use of integer or floating point arithmetic. Sometimes, the impact of these factors is not well understood by the programmers with the results that internal limits are unknown, large or small signals are not properly handled and unexpected results, such as wrap-around in the computation results ensues, with serious impacts on the performance of the exciter or the PSS. If the sampling rate is low, this is equivalent to delays that must then be overcome by additional lead in the PSS signal path. Also, crude filters, such as those implemented

by using a sliding window average suffer from additional delays. If the position of the firing pulse is computed digitally via a look-up table, a lack of resolution results in a field voltage that is noisy and could contain undesirable frequency components. This problem is particularly noticeable in systems with high field voltage forcing levels. The impact of the above limitations has been erratic derived signals, such as power, at low levels because of lack of resolution or erratic/noisy behaviour of the controlled variable. Unnecessary Filtering Adds Delay The presence of significant filtering in the PSS signal path, at the input to the summing junction leads to further difficulties in obtaining the right amount of phase lead in the PSS signal. This is particularly annoying when this feature is not documented and one has to find it out the hard way. Digital systems are slow enough without having to cope with this additional delay, which in one instance has been equivalent to a low-pass filter with a time constant of approximately 0.1 second. It would be of some help if such a feature could be defeated at site. The input connection, for an externally applied PSS, should be clean, with no intentional delay except for the digitizing process. Unsuitability of Internal Filtering The popularity of PID (Proportional-IntegralDerivative) stages in Automatic Voltage Regulators has led some manufacturers to add too much filtering in the Derivative path, to correct a perceived noise problem. Heavy filtering causes a rapid change in phase angle, with only small changes in frequency, Figures 3 and 4. It therefore becomes impossible to tune a PSS for the local modes and the inter-area modes of oscillations, a span of 0.1 to 3 Hz. Interaction Between the Underexcitation Limiter and PSS Signal Interactions between the PSS and the Underexcitation limiter have been a problem on some excitation systems, over the last decades. Well-designed systems have conquered this problem by providing fully independent tuning of the voltage regulating loop and the limiter parameters. The limiter signal is summed into the main AVR summing junction, resulting in
D ifferentiator O utput (pu)

0 16-Sample Filter 4-Sample Filter Pure Derivative

-1

4 6 Time (seconds)

10

Figure 3: Differentiators Performance at 0.2 Hz


20

Differentiator O utput (pu)

10

-10

Pure Derivative 4-Sample Filter 16-Sample Filter

-20

0.2

0.4 0.6 Time (seconds)

0.8

1.0

Figure 4: Differentiators Performance at 3 Hz excellent dynamic performance. In some recent systems, however, particularly where the limiter acts in a take-over manner, and yet still has the PSS signal feeding through the limiter parameters because the PSS signal is part of the terminal voltage signal, problems have surfaced. If the PSS is tuned for the PID parameters of the main voltage control loop, it will be necessary to keep the limiter PID stage identical to the AVR PID; otherwise the PSS may not perform correctly. This constraint may lead to an unstable limiter. Loss of PSS When UEL Comes Into Play Manufacturers offer quite a mixture of approaches when it comes time to operate against limiters, particularly the Underexcitation Limiters (UEL or MEL). If a generator is operating in the underexcited region, this is the time when the PSS is most required as the generator can be highly unstable in that region. Several manufacturers have chosen to have the UEL limiter take over control of the terminal voltage via an auctioneering circuit. This

effectively removes the PSS signal from duty. A better approach is the use of summation blocks, leaving the PSS in service at all times. Of course, there is then a need to ensure that with the PSS and the Limiter working in parallel, the dynamic performance will remain stable. This is not an insurmountable task: it has been achieved by several manufacturers. Lack of Access to Internal Variables Analog exciters and PSS allowed access to all internal signals. This often proved to be of great advantage when tuning a system in a difficult/unstable location. In particular, when tuning an Integral-of-AcceleratingPower PSS, when H or Xq are not known precisely, it is advantageous to have access to the internal signals to confirm that appropriate values have been set. Some recent digital implementations of excitation systems and PSS allow very limited access to internal variables. It is impossible to verify internal limits and possible wrap-around problems, or to obtain data to accurately model the excitation system. In one instance, dealing with an overseas manufacturer, we encountered a serious problem accessing the digital exciter, at any level. The manufacturer of the hardware indicated that unless we could locate the settings files left on a floppy disc at the station, we could not talk to the hardware with our computers. The file was never found and even a company representative sent to assist us was unable to gain a high enough access level to change something as minor as the manual setpoint lower limit. Old exciters, of the analog type, could be accessed by anyone with a screwdriver and no one worried about the possibility of someone walking around changing settings randomly. Why is it so different with digital systems? If a customer wishes to have someone look at his system, he accepts the risks. Hardware manufacturers should recognize that these security measures are counterproductive with regard to the reliability of the system. Luckily we have encountered only one such manufacturer so far. Most manufacturers have indeed been quite cooperative. Range of Settings and Resolution Available The range of settings and the resolution with which the settings can be applied should never be a restriction or limiting factor in achieving the best possible dynamic response, from the PSS and the excitation system. In digital systems, the hardware should never present a limitation. In analog system, the use of fixed switch settings can sometimes prevent obtain the desired value. It would be an easy matter to mount some key

components on standoffs so they could be field-replaced to achieve the desired settings. Modelling Requirements High-end digital systems have shown the way to accurate modelling by providing the necessary internal structure and scaling factors. A detailed model for these systems can be constructed and verified at site without too much effort. The model parameters can be linked to the settings directly and accurately without the need for fudge factors. Determining the suitability of an excitation system for use with a PSS, in the final analysis, must be based on simulations conducted with accurate models of the exciter, the PSS, the generator, and the power system, in this order of importance High-end analogue and digital excitation systems have always been well documented, with enough details about their circuit diagrams or internal structure to allow knowledgeable staff to construct good computer models, representing all of the internal transfer functions, gains and limits. Such is not the case for low-cost system and, to some extent, this is to be expected. It costs almost as much time and money to write a manual for a low-cost system as it does for the high-end system. If an exciter sells for $20K, there is not a lot of money left over for the documentation, unless the number of units sold is very large. It has become a challenge to derive accurate models of some of the low-cost systems installed on gas turbines and elsewhere. These systems, mostly digital, have often been programmed by staff that is long gone. The exact algorithms used are not documented properly. We are then left with having to go through several iterations with manufacturers to obtain a model. It becomes difficult to relate model parameters to actual settings. It is not a matter of scaling but rather of not knowing what else is there? How is the derivative block actually implemented? How is the filtering implemented? What are the internal limits? Simple Simulations to Establish Suitability of Exciter/PSS Hardware In 1992 it was appropriate to establish a benchmark transfer function to determine exciter suitability. However, it is also important to construct a detailed model of the local generator, excitation system and PSS to check that the hardware is suitable. Range of settings, response to large and small signals, and interactions between the PSS and various limiters in the excitation, are some of the items to be verified. It is essential, to perform this task properly, to have all the limits, gains, and time constants included in the model.

Conclusions The purpose of this paper is to highlight some of the problems that have arisen in the application of PSS to new types of generation that is appearing. It is not intended to point fingers at any particular segment of the industry. Things can be easier in the future if: 1. System operators make their requirements clear to connecting generators. 2. Companies contracting for new generation ensure that whoever is doing the engineering understands and meets the system operators standards. 3. Manufacturers take an active role in understanding the use for which their equipment is being purchased to ensure that it is suitable. 4. Technical bodies such as the IEEE ensure that problems that have been encountered are made known to the industry. References 1. S. V. Larsen, D. A. Swann, 'Applying Power System Stabilizer., Part I General Concepts'.IEEE Trans. on PAS Vol. 100, No. 6, June 1981, pp. 3017-3024. "Criteria and Definitions for Excitation Systems for Synchronous Machines'. IEEE Standards 42172, December, 1972. 'IEEE Guide for Identification, Testing, and Evaluation of the Dynamic Performance of Excitation Control System,' IEEE Standards 42.1.2 - 1990. C. Concordia, F. P. De Mello, 'Concepts of Synchronous Machine Stability As Affected by Excitation Control', IEEE Trans PAS Vol. 88, Apr. 1969.' pp 316-329. J.P. Bayne, D.C. Lee and W. Watson. A Power System Stabilizer for Thermal Units Based on Derivation of Accelerating Power. IEEE Transactions Vol. 96 (PAS) No 6:1777-1783, November/December 1977. W. Watson and G. Manchur. Experience with Supplementary Damping Signals for Generator Static Excitation Systems. IEEE Transactions Vol. 93 (PAS) No 1:199-203, January/February 1973. J.P. Bayne, P. Kundur and W. Watson. Static Exciter Control to Improve Transient Stability.

IEEE Transactions Vol. 94 (PAS) No 4:1141-1146, July/August 1975 8. D.C. Lee, R.E. Beaulieu, J.R.R. Service. A Power System Stabilizer Using Speed and Electrical Power Inputs-Design and Field Experience. IEEE Transactions Vol. 100 (PAS), No 9, September 1981. W. Watson and M.E. Coultes. Static Exciter Stabilizing Signals on Large Generators - Mechanical Problems. IEEE Transactions Vol. 92 (PAS):204-211, January/February 1973. 10. WSCC Modelling Work Group, Criteria to Determine Excitation System Suitability for PSS in WSCC System, December 17, 1992. Roger E. Beaulieu graduated from the University Of Waterloo, Ontario, in 1967 as an Electrical Engineer, B.A.Sc. From February 1969 to November 1, 1993, he worked for Ontario Hydro's Research Division on the development of protection systems, conducted research, development and modelling of excitation systems and power system stability controls, governors and overall system performance. He retired in November 1993 and now works as an advisor to power utilities in his role as Senior Engineer at Goldfinch Power Engineering Inc. His main fields of interest remain power system operation and protection. He is a Registered Professional Engineer in the Province of Ontario. beaulieu@goldfinchpower.com

9.

2.

3.

4.

5.

6.

7.

You might also like