Power quality monitoring systems are getting more powerful every day. Permanent monitoring systems are used to track ongoing system performance and watch for potential power quality problems. They also provide a wealth of information for utility and customer personnel to review when there is a disturbance. But one of the most exciting developments in this field is the implementation of intelligent systems that can automatically evaluate disturbances and system conditions, draw conclusions about their cause, and even predict problems before they occur.
These "expert" systems are the wave of the future. In this column, we'll address one example of an automated evaluation system for identification and characterization of capacitor-switching transients.
Capacitor-Switching Issues We've looked at power quality problems associated with capacitor switching in previous issues. But how can a monitoring system help you identify these problems and find solutions before they occur?
Is it Capacitor Switching? The most basic task for an intelligent disturbance identification system is to determine the cause of the disturbance. Capacitor-energizing transients (Fig. 1) are usually characterized by a step change in the voltage, followed by an oscillation as the voltage across the capacitor equalizes with the system voltage.
The oscillation occurs at the natural frequency of the capacitor with the inductance of the power system, which is usually in the range of 250 Hz to 800 Hz. You can use Fourier and Wavelet analysis to help identify the signature of a capacitor-switching transient and distinguish it from other transient disturbances superscript [1].
Locating the Capacitor Bank It also would be nice to know the location of the capacitor bank that is being switched. One important characteristic that can be determined is whether the capacitor is upstream or downstream from the monitoring location. If the monitor is at the service entrance to a facility, this would tell if the capacitor bank was located in the plant or if it was on the utility supply system. If the monitor is located at a substation, you could tell if the capacitor bank was a transmission capacitor bank or if it was located on the distribution system. If you have multiple monitors on the system and know their actual locations, you can determine the exact capacitor that is being switched. The Dranetz-BMI Signature System capacitor switching answer module is one system on the market that offers this capability superscript [2].
Magnification of Capacitor-Switching Transients In our previous discussions of capacitor-switching concerns, we discussed the phenomenon of capacitor-switching transient magnification. This occurs when a large utility-owned capacitor bank is switched on the supply system, and a smaller customer-owned bank picks up the switching transient.
The series resonance formed by the step-down transformer and the customer's capacitor bank can magnify the switching transient, resulting in high-transient voltages within the facility. Many customers have reported equipment misoperation (e.g., adjustable-frequency drives tripping) and even equipment failures (e.g., SCR circuit board failures in DC drives) caused by these magnified capacitor-switching transients.
Capacitor-switching transients usually have a peak value in the range of 1.1 to 1.7 times the normal peak voltage. If the switching transient magnitude is higher than this, it could be a magnified capacitor-switching transient (Fig. 2, on page 39). The expert system should identify these events and provide an alarm so the situation can be investigated. Solutions include changing the configuration or size of the capacitors within the facility or adding switching control to the capacitor being switched on the supply system.
Is the Capacitor Switch Operating Properly? Capacitor-switching devices can misoperate, resulting in serious capacitor switching problems. Two types of misoperation are possible:
The switch or breaker can misoperate during closing. Normally, the individual phases of the switching devices should close at nearly the same time. A significant separation between the phases could be an indication of a problem with the device itself. In a grounded capacitor bank, this can be detected by looking at the instantaneous transient voltage on each phase. With an ungrounded capacitor bank, there is no transient until the second pole of the switching device closes. Therefore, it is harder to tell when the first pole closed. Pole separation during closing operations is not usually a major problem. However, it can be an indication of a breaker problem, which could show up as a restrike problem during opening.
A more important problem occurs when the switch misoperates on opening. Normally, there is no transient when a capacitor bank is deenergized. If the monitoring system detects a transient voltage during an opening operation, an alarm should sound because it is an indication of a restrike of the capacitor-switching device. This means that the switch was not able to withstand the recovery voltage across its contacts as it was opening. During this activity, an arc forms across the contacts which energizes the capacitor again. This energizing action can be much more severe than normal energizing because there is a trapped charge on the capacitor when it occurs (Fig. 3). Restrike transients can result in arrester failures, blown capacitor fuses, switch failures, or other equipment failures.
Capacitor switch restrikes can often occur without actual breaker failures, and monitoring may be the only way to detect a problem. If restrike events are identified the first time they occur, breaker maintenance or replacement may avoid additional restrike events and equipment failures.
Identifying Harmonic Resonance Conditions Equipment powered by electronic power supplies (e.g., computers) or electronic power converters (e.g., adjustable- speed drives) continues to grow in terms of the total percentage of the electrical system load. These devices draw currents that include harmonic components. The interaction of these currents with the system impedance causes system voltage distortion. Normally, this is not a significant problem because the system impedance is low enough that the voltage distortion levels are not severe. However, capacitor banks can result in resonance conditions that cause the system impedance to be high at particular frequencies. These high-impedance resonances can result in voltage distortion problems in customer facilities or significant portions of the power system, depending on the capacitor location and system characteristics.
When capacitor banks result in resonance conditions, the high harmonic distortion levels can blow capacitor bank fuses, overheat transformers and motors, and cause other customer equipment problems.
It is important to identify these conditions and solve the problem before it results in equipment failures and misoperation. The problem can usually be solved by filtering the harmonics near the equipment causing the harmonics or by changing the characteristics of the resonance by changing the size of capacitors or converting capacitors into filters.
An expert system can watch for resonant conditions by looking for changes in harmonic distortion levels when a capacitor is switched (Fig. 4, on page 40). The resonance can be further characterized by determining the dominant frequency in the voltage and current distortion.
Looking for Blown Fuses and Failed Cans An expert system can evaluate the health of capacitor banks every time they are switched in or out. The system can evaluate the kvar change on each phase and determine whether all three phases switched in properly or whether there is an imbalance, which could indicate a blown fuse or failed can in the bank. This type of evaluation can be further enhanced with information about the capacitor bank sizes, their locations, and the switching procedures (are they switched based on a current control, kvar control, timer, etc.). It may even be possible to pinpoint the individual capacitor banks being switched and identify problems automatically.
Nuisance Tripping of Electronic Loads Monitoring systems placed within end user facilities can correlate capacitor-switching transients on the facility equipment, like nuisance tripping of adjustable-speed drives. This still can be a problem with smaller drives that do not include chokes as part of their design. Identifying the transient that caused the tripping can help you locate the source of the problem.
Implementing an Expert System The best place to process information about power quality events and characteristics is close to the location where the measurements are being taken. This frees the central system and the engineers to focus on the most important information, rather than being bombarded with huge amounts of low-level voltage and current information. An alarm from the local location can notify you whenever there is something important that needs to be investigated. And the detailed information is always available for more in-depth investigations if you need it.
A development project was initiated by EPRI, Electrotek Concepts, the Tennessee Valley Authority, and Dranetz-BMI a number of years ago to turn this concept into a reality. The research was designed to provide users with an autonomous expert system that would turn raw measurements into analysis results or answers. This resulted in the AnswerModule, a software program embedded in a power monitoring system, which is designed to provide answers, analysis results, or knowledge to specific questions. The AnswerModule works within the Signature System to process measurement results as they are obtained.
The AnswerModule system is a hybrid that pulls together data from many sources. It involves numerous signal processing techniques, knowledge-based techniques, and knowledge discovery techniques commonly known as data mining. The AnswerModule is part of a larger power monitoring system, which consists of data acquisition devices, a data aggregation system, a communications system, an Internet-based visualization tool, and enterprise management components. The AnswerModule is located in a device called an InfoNode, which processes the information from data acquisition devices and provides access to the processed information via a standard Web browser.
The general procedure for implementing an expert system is shown in Fig. 5, on page 40. The process involves data selection and preparation, information extraction, information assimilation, and report presentation. These steps are known as knowledge discovery or data mining.
This is just one example of useful expert systems for processing power quality data. If you have ideas or examples of other expert system applications, drop me an e-mail ([email protected]) so we can consider it for a future column.
References S. Santoso, W. M. Grady, E.J. Powers, J. Lamoree, S.C. Bhatt, "Characterization of Distribution Power Quality Events with Fourier and Wavelet Transforms," IEEE Trans. Power Delivery, Vol. 5, No. 1, Jan. 2000, pp. 247-254.
S. Santoso, J. Lamoree, R. Bingham, "AnswerModule: Autonomous Expert Systems for Turning Raw PQ Measurements into Answers," ICHQP 2000, Orlando, Fla., October, 2000.
