40.2.1 Steady-state Voltage Frequency and Magnitude

In most areas of North America, steady-state frequency regulation is not a significant issue due to the sufficient levels of generating capacity and the strong interconnections among generating companies and control areas. In other parts of the world, and North America under extreme conditions, frequency can deviate from 1/4 to 1/2 Hz during periods of insufficient generating capacity. Under transient conditions, frequency can deviate up to 1–2 Hz.

Frequency deviations can affect power electronic equipment that use controlled switching devices unless the control signals are derived from a signal that is phase-locked with the applied voltage. In most cases, phase locks are used, or the converters consist of uncontrolled rectifiers. In either case, frequency deviations are not a major cause of problems. In most cases, frequency deviations have more impacts on conventional equipment that does not use electronics or in very inexpensive electronic devices. Clocks can run fast (or slow), motor speeds can drop (or rise) by a few revolutions per minute, etc. In most cases, these effects have minimal economic impact and are not considered a real power quality problem.

Steady-state voltage regulation is a much more pronounced issue that can impact a wide range of end-use equipment. In most cases, utility supply companies do a very effective job of providing carefully regulated voltage within permissible ranges. In North America, ANSI Standard C84.1 suggests steady-state voltage ranges both at the utility service entrance and at the point of connection of end-use equipment. Furthermore, equipment manufacturers typically offer equipment that is tolerant of steady-state voltage deviations in the range of ± 10%. Table 40.1 shows the voltage ranges suggested by ANSI C84.1, with specific mention of normal (Range A) and contingency (Range B) allowable voltages, expressed in percent.

TABLE 40.1 ANSI C84.1 Voltage ranges

Range A is for normal conditions and Range B is for emergency or short-time conditions.

Virtually all equipment, especially sensitive electronic equipment, can be effected by deviating voltage outside the ± 10% range. In most cases, overvoltages above + 10% lead to loss of life, usually over time; excessive overvoltages can immediately fail equipment. Undervoltages below − 10% usually lead to excessive current demands, especially for equipment that has a controlled output like an adjustable speed drive controlling a motor to a constant speed/torque point. The impacts of these prolonged excessive currents can be greater voltage drop, temperature rise in conductors, etc. In the extreme, undervoltages of greater than 15–20% can cause equipment to immediately trip. In most cases, such extreme undervoltages are associated with system faults and the associated protection system. These extreme undervoltages are so important that they are classified in a power quality category of their own called voltage sags.

40.2.2 Voltage Sags

Other than improper grounding, voltage sags are probably the most problematic of all power quality problems. At this time, a number of standards-making bodies, including IEEE, ANSI, and IEC, are working on standards related to sags. In most cases, sags are generally agreed to be more severe and outside of the scope of ANSI C84.1 and they are temporary in nature due to the operation of system protection elements. Because the electrical system is a continuous electrical circuit, faults in any location will have some impact on voltages throughout the network. Of course, areas closer to the faulted area will see a greater voltage sag due to the fault than other, more (electrically) remote areas. Sags can originate anywhere in a system, but are more pronounced in utility distribution systems because of the greater exposure of low-voltage systems to the causes of short circuits.

Most utility companies implement distribution system protection in what is known as a “fuse saving” methodology. Figure 40.1 shows a typical overhead distribution system with two feeders being supplied from the same substation transformer. Each primary circuit has its own automatic circuit recloser (ACR) and shows one fused tap.

FIGURE 40.1 Overhead distribution system.

With the protection system set up based on fuse-saving methodology, any fault downstream of a fault will be cleared first by the substation recloser followed by a reclosing operation (re-energization of the circuit) 1/2–2 later. If the fault is still present, the closest fuse should blow to permanently isolate the fault. (Note that in some cases, multiple reclosing attempts are made prior to the clearing of the fuse.)

For a fault on the load side of fused tap # 2 in Fig. 40.1, customers on feeder # 1 will see a voltage sag determined by the system and transformer impedance at the substation. Because this impedance is typically on the same order (or larger) as the feeder circuit impedance, a sag in substation bus voltage of 50% is common. This sag will persist until feeder # 2 is cleared by the recloser opening. When the recloser re-energizes the circuit, a permanent fault will still be present and the substation bus will again experience a voltage sag. Of course, any sag in substation bus voltage will be delivered directly to all customers on feeder # 1, even though there is no electrical problem on that feeder. Figure 40.2 shows a possible rms voltage profile that might be supplied to the customers on feeder # 1 for a permanent fault on the load side of fused tap # 2. Only one recloser operation is shown prior to fuse clearing.

FIGURE 40.2 The rms voltage supplied to feeder # 1 customers.

Just based on the voltage information shown in Fig. 40.2, it is impossible to tell if the end-use loads on feeder # 1 will experience a problem. Equipment tolerance curves are required to assess the vulnerability of equipment to voltage deviations, including sags, and all equipment is different. Figure 40.3 shows the lower portions of two equipment tolerance curves, the (older) CBEMA and the (newer) ITIC curves for computer equipment. Most, but not all, power electronic-based equipment has a similar shape. Voltage sags with a duration that correspond to a point that is “below and to the right” of the tolerance curve will result in loss of equipment function, while sags of duration that plot “above and to the left” of the tolerance curve will not effect equipment performance. Note that only the lower portion of the curve has been shown; an upper tolerance curve also exists that is often used in transient (overvoltage) studies.

FIGURE 40.3 CBEMA (curved line) and ITIC (square shape) tolerance curves.

Voltage sags are probably the most common power quality problem that is “given” to the end-user by the supplying utility. However, improper equipment grounding is responsible for the vast majority of power quality problems on the customer's side of the meter.

40.2.3 Grounding

Grounding of equipment was originally conceived as a personnel safety issue. However, the presence of an electrical conductor that is at zero potential has been widely used in many power electronic and microprocessor-controlled loads. In the United States, electrical systems in residential, commercial, and industrial facilities fall under the purview of the National Electric Code (NEC) which establishes specific criteria for grounding of equipment. While it was once thought that proper grounding according to the NEC was detrimental to power quality concerns, these opinions have gradually faded over time.

From a power quality perspective, improper grounding can be considered in three broad categories

The ground loop problem is a significant issue when power, communications, and control signals all originate in different locations, but come together at a common electrical point. Transients induced in one location can travel through the created ground loop, damaging equipment along the way. Improper neutral-to-ground connections will create a “noisy” ground reference that may interfere with low-voltage communications and control devices. Excessive neutral-to-ground voltage may damage equipment that is not properly insulated or that has an inexpensive power supply.

Figure 40.4 shows a common wye-connected service (assumed at the terminals of a transformer) that supplies power to equipment that is also remotely monitored and controlled from another location with a separate ground reference.

FIGURE 40.4 Powering and control ground loop.

For any shift in ground potential for the power circuit, often caused by lightning as shown in Fig. 40.4, potentially large currents can flow through the grounding circuits and through the sensitive electronic equipment. Such currents can easily lead to equipment damage. Situations like these are common in

Figure 40.5 shows an example of an improper neutral-to-ground connection, and how this connection can create power quality problems.

FIGURE 40.5 Improper neutral-to-ground connections.

Load current returning in the neutral conductor will, at the point of improper connection to ground, divide between neutral and ground. This current flow in the ground conductor will produce a voltage at the load equipment, which can easily disrupt equipment operation.

Figure 40.6 shows an example of the possibility for excessive neutral-to-ground voltage and how this can lead to power quality problems.

FIGURE 40.6 Excessive neutral-to-ground voltage.

For load equipment that produces significant voltage drop in the neutral, such as laser printers and copying machines when the thermal heating elements are on, the voltage from the neutral-to-ground reference inside the equipment can exceed several volts. In many cases, this voltage is sufficient to damage printed circuit boards, disrupt control logic, and fail components.

40.2.4 Harmonics

In most cases, power electronic equipment is considered to be the “cause” of harmonics. While switching converters of all types produce harmonics because of the non-linear relationship between the voltage and current across the switching device, harmonics are also produced by a large variety of “conventional” equipment including

Any ac current flow through any circuit at any frequency will produce a voltage drop at that same frequency. Harmonic currents, which are produced by power electronic loads, will produce voltage drops in the power supply impedance at those same harmonic frequencies. Because of this inter-relationship between current flow and voltage drop, harmonic currents created at any location will distort the voltage in the entire supply circuit.

In most cases, equipment is not overly sensitive to the direct impacts of harmonic current flow. Note, however, that equipment heating is a function of the rms value of the current, which can significantly exceed the fundamental frequency value when large harmonic components are present. It is because harmonic currents produce harmonic voltages that there is a real power quality concern.

Most equipment can operate satisfactorily as long as the voltage distortion at the equipment terminals does not exceed around 5%. Exceptions to this general rule include ripple-control systems for converters (which are impacted by small even-order harmonics) and small harmonics at sufficiently high frequency to produce multiple zero crossing in a waveform. (Note that voltage notching due to simultaneous commutation of switching devices can also create multiple zero crossings.) Such a multiple crossing scenario is shown in Fig. 40.7 and represents a 60 Hz waveform plus a 1% voltage harmonic at 3000 Hz.

FIGURE 40.7 Multiple zero crossings.

Converters that have a time-limited firing signal can directly suffer from excessive voltage distortion. For a six-pulse converter, a maximum time of 1/(6× 60) seconds is available to turn on a switching device. Similarly, for a 12-pulse converter, a maximum of 1/(12× 60) is available to turn on a switching device. Considering that all switching devices have a short (but non-zero) turn-on time, manufacturers tend to design drive circuits that bring up the firing pulse for a limited amount of time. If, for example, a firing pulse is maintained for 100 µs, the device must begin conduction in that time. In situations where voltage distortion is excessive, the device to be switched could be reverse biased during the first several milliseconds of the time available for device firing during which time conduction cannot begin. If the firing signal is removed before the certain classes of switching devices are correctly biased, conduction will not begin at all. This situation, commonly called a “misfire,” can lead to equipment mis-operation and failure.

Because some switching devices can conduct in both directions when the firing signal is applied (but only one direction is intended to carry appreciable current), applying the firing pulse at a time when the voltage is of the wrong polarity can destroy the device. Excessive voltage distortion can certainly lead to such a situation, and manufacturers typically design products to function only under limited-distortion conditions.

Because of the numerous potential problems with harmonic currents, standards exist for their control. The IEC goes as far as to limit the harmonic currents produced by certain individual pieces of equipment, while the IEEE takes more “system-level” point of view and prescribes limits for harmonic currents for a facility as a whole, including one of more harmonic producing loads. Harmonic standards will be further discussed in Section 40.4.

40.2.5 Voltage Fluctuations and Flicker

Voltage flicker is not directly caused by electronic loads except in the largest of applications. Voltage fluctuations, and the corresponding light flicker due to them, are usually created by large power fluctuations at frequencies less than about 30 Hz. In most applications, only

are potentially problematic. Each of these types of end-use devices can create large, low-frequency (about 30 Hz or less) variations in the system voltage, and can therefore lead to voltage flicker complaints. At this time, the IEEE prescribes a “flicker curve” based originally on research conduction by General Electric. The IEC, however, has adopted a different methodology that can consider voltage fluctuations and flicker that are more complex than those considered by the IEEE flicker curve.

Most equipment is not sensitive to the voltage fluctuations that cause flicker complaints. The change in output of incandescent lamps as viewed by human observers becomes objectionable at levels of change around 0.3%, but electronic equipment will not be affected at all. Because most utility supply companies limit voltage fluctuations, regardless of the frequency of repetition, to less than a few percent, equipment malfunction or damage due to flicker is very rare. Figures 40.8 and 40.9 show plots of single-cycle rms voltage fluctuations due to large dc welders and arc furnaces, respectively; it is clear that the magnitude of the fluctuations are well above the level that could impact equipment. The waveform in Fig. 40.8 probably would generate numerous light-flicker complaints, whereas the waveform in Fig. 40.9 probably would not. Neither would disrupt equipment.

FIGURE 40.8 Single-cycle rms voltage fluctuations due to a large dc welder.

FIGURE 40.9 Single-cycle rms voltage fluctuations due to a large dc arc furnace.

Due to the advances in power electronics that have offered devices with higher power ratings, reactive compensation systems have been developed to compensate for voltage fluctuations by adding or removing reactive power from the supply circuit. These devices have allowed large flicker-producing loads like arc furnaces to be served from utility circuits that, without the compensator, could not serve the load. However, because the compensators can so directly impact system voltage, they can create flicker problems if they are not properly applied and controlled.

40.2.6 Transients

Transients, especially in the voltage supply, can create numerous power quality problems. The major sources of transients are

Lightning events can create the most severe overvoltages, but these transients decay rapidly. A typical lightning transient has decayed to zero in a few hundred microseconds, but it can reach a peak magnitude of several hundred percent if not controlled with surge suppression devices. Other categories of transients associated with power system switching are much smaller in magnitude (typically less than 200%), but last in the order of several hundred milliseconds. Considering the energy available in a transient, therefore, there is a considerable overlap in the range of severity of lightning and switching transients. It is the available energy that typically determines whether or not an equipment will be affected or damaged. Figure 40.10 shows a capacitor switching transient on a low-voltage (480 V) circuit. The magnitude and duration of the event are quite clear.

FIGURE 40.10 Capacitor switching transient.

Transients such as those shown in Fig. 40.10 are generally sufficient to cause nuisance trips of electronic loads like adjustable speed drives. For these types of loads, the protection system settings are usually very tight due to the use of sensitive switching devices. Overcurrent and overvoltage settings of 120% are not uncommon. For a transient similar to that shown in Fig. 40.10, there is sufficient overvoltage for very large currents to flow through any conducting switching device to the drive's dc bus. The device's protection system see these overcurrents as a fault, and trip the drive. Similarly, the overvoltage at the terminals can be passed through to the dc bus and accumulate, where the drive may trip due to overvoltge on the dc bus.

40.2.7 Monitoring and Measurement

To consider or be able to diagnose power quality related problems, it is imperative to be able to measure various power quality parameters. Several different categories of monitoring and measurement equipment exist for these purposes, with costs ranging from a few hundred dollars to $ 10,000–20,000 for fully equipped disturbance analyzer.

The most basic category of power quality measurement tool is the handheld voltmeter. It is important that the voltmeter be a true-rms meter, or erroneous readings will be obtained that incorrectly suggest low or high voltage when harmonics are present in the signal. It is especially important to have true rms capability when measuring currents; voltage distortion is not typically severe enough to create large errors in the readings of non-true rms meters. Virtually all major measurement equipment vendors offer true rms meters, with the costs starting around $ 100.

The next step up from the basic voltmeter is a class of instruments that have come to be called “power quality analyzers.” These instruments are handheld and battery powered. These instruments can measure and display various power quality indices, especially those that relate to harmonics like THD, etc. and can also display the input waveform. Newer models feature 20 MHz (and higher) bandwidth oscilloscopes, inrush measurements, time trending, and other useful features. Manufacturers such as Fluke, Dranetz, BMI, and Tektronix offer these types of instruments for around $ 2000.

In most power quality investigations, it is not possible to use handheld equipment to collect sufficient data to solve the problem. Most power quality problems are intermittent in nature, so some type of long-term monitoring is usually required. Various recorders are available that can measure and record voltage, current, and power over user-defined time period. Such recorders typically cost in the order of $ 3000-10,000. More advanced long-term monitors can record numerous power quality events and indices, including transients, harmonics, sags, flicker, etc. These devices, often called “line disturbance analyzers”, typically cost between $ 10,000 and $ 20,000.

It is important to use the right instrument to measure the phenomenon that is suspected of causing the problem. Some meters record specific parameters, while others are more flexible. With this flexibility comes an increased learning curve for the user, so it is important to spend time on them before going out to monitor, to make sure all aspects and features of the equipment are understood.

It is equally important to measure in the correct location. The best place to measure power quality events is at the equipment terminals that is experiencing problems. With experience, an engineer can evaluate the waveforms recorded at the equipment terminals and correlate them to events and causes elsewhere in the power system. In general, the farther away from the equipment location the monitoring takes place, the more difficult is to diagnose a problem.

40.3.1 Typical Harmonics Produced by Equipment

In theory, most harmonic currents follow the “ 1/n” rule where n is the harmonic order (180 Hz = 3 × 60; n = 3). Also in theory, most harmonic currents in three-phase systems are not integer multiples of three. Finally, in theory, harmonic currents are not usually even-order integer multiples of the fundamental. In practice, none of these statements are completely true and using any of them “exactly” could lead to either over- or under-conservatism depending on many factors. Consider the following examples:

Table 40.2 gives the magnitudes, in percent of fundamental, of the first 25 (integer) harmonics for a single-phase switched-mode power supply, a single-phase fluorescent lamp, a three-phase (six-pulse) dc drive, and a three-phase (six-pulse, no input choke) ac drive. Together, these load types represent the range of harmonic sources in power systems. Note that seemingly minor changes in parameter values and control methods can have significant impacts on harmonic current generation; the values given here are on the conservative side of “typical.”

TABLE 40.2 Typical harmonic spectra of load equipment

40.3.2 Resonance

Considering only the harmonic current spectra given in Table 40.2, it would appear that a large number of harmonic-related power quality problems are on the verge of appearing. In reality, most current drawn by many residential, commercial, and industrial customers is of the fundamental frequency; the amplitudes of the individual harmonic currents, in percent of the total fundamental current, are often much less than that shown in Table 40.2. For this reason, end-use locations employing non-linear loads often do not lead directly to significant voltage distortion problems. A parallel resonance in the power supply system, however, changes the picture entirely.

Series and parallel resonance exist in any ac power supply network that contains inductance(s) and capacitance(s). The following simple (but usable) definitions apply:

For a given harmonic-producing load that generates harmonics at frequencies that correspond to parallel resonance in the supply system, even small currents at the resonant frequencies can produce excessive voltages at these same frequencies. The principle of series resonance, however, is usually exploited to reduce harmonics in power systems by providing intentionally low impedance paths to ground.

In many cases, end-users will install power factor correction capacitors in order to minimize reactive power charges by the supply utility. In most cases, these capacitors are located on the customer's low-voltage supply buses and are therefore in parallel with the service transformer. In most cases where the power factor correction capacitors are sized to provide a net power factor (at the service entrance) to 0.85(lag)–1.0, the parallel resonance occurs somewhere between the fifth and nineth harmonic. Considering Table 40.2, it is apparent that a large number of loads produce harmonics at these frequencies, and the amplitudes can be significant. Even a small increase in impedance at these frequencies due to resonance can lead to unacceptable voltages being produced at these same frequencies. Figure 40.11 shows and example plot of impedance looking into a utility supply system when a typically-sized capacitor bank has been installed to improve overall plant power factor.

FIGURE 40.11 Driving point impedance.

From Fig. 40.11, it is clear that 1.0A at the 9th harmonic will produce at least 150 times more voltage drop than would be produced by the same 1.0 A if it were at the fundamental frequency. A look back at the Table 40.2 shows the clear potential for problems. Fortunately, the series resonance principle can be used to provide a low-impedance path to ground for the harmonic currents and thus reduce the potential for problematic voltage distortion.

Since capacitors are required to produce parallel resonance, it is often a “cheap fix” to slightly modify the capacitor to include a properly sized series reactor and create filter. This filter approach, designed based on the series resonance concept, is usually the most cost-effective means to control harmonic voltage distortion.

40.3.3 Harmonic Filters

Harmonic filters come in many “shapes and sizes.” In general, harmonic filters are “shunt” filters because they are connected in parallel with the power system and provide low impedance paths to ground for currents at one or more harmonic frequencies. For power applications, shunt filters are almost always more economical than series filters (like those found in many communications applications) for the following reasons:

Shunt filters are designed (or can be purchased) in three basic categories as follows:

The single- and double-tuned filters are usually used to filter specific frequencies, while the damped filters are used to filter a wide range of frequencies. In applications involving small harmonic producing loads, it is often possible to use one single-tuned filter (usually tuned near the fifth harmonic) to eliminate problematic harmonic currents. In large applications, like those associated with arc furnaces, multiple tuned filters and a damped filter are often used. Equivalent circuits for single- and double-tuned filters are shown in Fig. 40.12. Equivalent circuits for first, second, third, and “c-type” damped filters are shown in Fig. 40.13.

FIGURE 40.12 Harmonic filters: (a) single-tuned and (b) double-tuned.

FIGURE 40.13 Damped filters: (a) first-order; (b) second-order; (c) third-order; and (d) C-type.

A plot of the impedance as a function of frequency for a single-tuned filter is shown in Fig. 40.14. The filter is based on a 480 V, 300 kvar (three-phase) capacitor bank and is tuned to the 4.7th harmonic with a quality factor, Q, of 150. Note that the quality factor is a measure of the “sharpness” of the tuning and is defined as X/R where X is the inductive reactance for the filter inductor at the (undamped) resonant frequency; typically 50<Q< 150 for tuned filters.

FIGURE 40.14 Single-tuned filter frequency response.

A plot of impedance as a function of frequency for a second-order damped filter is shown in Fig. 40.15. This filter is based on a 480 V, 300 kvar capacitor bank and is tuned to the 12th harmonic. The quality factor is chosen to be 1.5. Note that the quality factor for damped filters is the inverse of the definition for tuned filters; Q=R/X where X is the inductive reactance at the (undamped) resonant frequency. Typically, 0.5<Q< 1.5 for damped filters.

FIGURE 40.15 Second-order damped filter frequency response.

In most cases, it is common to tune single-tuned filter banks to slightly below (typically around 5%) the frequency of the harmonic to be removed. The reasons for this practice are as follows:

Damped filters are typically used to control higher-order harmonics as a group. In general, damped filters are tuned in between the corresponding pairs of harmonics (11th and 13th, 17th and 19th, etc.) to provide the maximum harmonic reduction at those frequencies while continuing to serve as a (not quite as effective) filter bank for frequencies higher than the tuned frequency. Because damped filters have significantly higher resistance than single- or double-tuned filters, they are usually not used to filter harmonics near the power frequency so that filter losses can be maintained at low values.