Deploying generation along utility distribution systems naturally creates some conflicts because the design of the system assumes only one source of power. A certain amount of generation can be accommodated without making any changes. At some point, the conflicts will be too great and changes must be made.
In this article, several of the operating conflicts that can result in power quality problems are described. They are:
1. Utility Fault-Clearing Requirements:
Figure 6.7 shows the key components of the overcurrent protection system of a radial feeder. The lowest-level component is the lateral fuse, and the other devices (reclosers and breakers) are designed to conform to the fuse characteristic. There will frequently be two to four feeders off the same substation bus. This design is based mostly on economic concerns. This is the least costly protection scheme that is able to achieve acceptable reliability for distributing the power.
One essential characteristic is that only one device has to operate to clear and isolate a short circuit, and local intelligence can accomplish the task satisfactorily. In contrast, faults on the transmission system, which easily handles generation, usually require at least two breakers to operate and local intelligence is insufficient in some cases.
In essence, this design is the source of most of the conflicts for interconnecting Distributed Generation with the utility distribution system. Because there is too much infrastructure in place to consider a totally different distribution system design to better accommodate Distributed Generation, the Distributed Generation must adapt to the way the utility system works. With only one utility device operating to clear a fault, all other Distributed Generation devices must independently detect the fault and separate to allow the utility protection system to complete the clearing and isolation process. This is not always simple to do from the information that can be sensed at the generator.
Reclosing utility breakers after a fault is a very common practice. Most of the distribution lines are overhead, and it is common to have temporary faults. Once the current is interrupted and the arc dispersed, the line insulation is restored. Reclosing enables the power to be restored to most of the customers within seconds.
Reclosing presents two special problems with respect to DG:
a. Distributed Generation must disconnect early in the reclose interval to allow time for the arc to dissipate so that the reclose will be successful.
b. Reclosing on Distributed Generation DG, particularly those systems using rotating machine technologies, can cause damage to the generator or prime mover.
The Distributed Generation relaying must be able to detect the presence of the fault followed by the opening of the utility fault interrupter so that it can disconnect early in the reclose interval as shown.
Normally, this detection and disconnection process should be straightforward. However, some transformer connections make it difficult to detect certain faults, which could delay disconnection.
A greater complicating factor is the use of instantaneous reclosing by many utilities. This is used for the first reclose interval for the purpose of improving power quality to sensitive customers. The blinking clock problem can be largely averted, and many other types of loads can ride through this brief dead time.
The interval for instantaneous reclose is nominally 0.5 s, but can be as fast as 0.2 s. This is in the range of relaying and opening times for some Distributed Generation breakers. Thus, instantaneous reclose is very likely to be incompatible with Distributed Generation. It greatly increases the probability that some Distributed Generation will still be connected when the reclose occurs or that the fault did not have enough time to clear, resulting in an unsuccessful reclose.
A reclose interval of at least 1.0 s is safer when there is Distributed Generation on the feeder. Many utilities use 2.0 or 5.0 s for the first reclose interval when Distributed Generation is installed. This minimizes the risk that the Distributed Generation will not disconnect in time. If it is deemed necessary to maintain the instantaneous reclose, it is generally necessary to employ direct transfer trip so that the Distributed Generation breaker is tripped simultaneously with the utility breaker.
This can be a very expensive proposition for smaller Distributed Generation installations. Thus, for some distribution systems it will be necessary to compromise one aspect of power quality to better accommodate significant amounts of Distributed Generation.
3. Interference with Relaying:
Three of the more common cases where Distributed Generation can interfere with the overcurrent protection relaying on distribution feeders will be examined here:
i. Reduction of reach
ii. Sympathetic tripping of feeder breakers
iii. Defeat of fuse saving
Each overcurrent relay device has an assigned zone of protection that is determined by its minimum pickup value. Some refer to this generically as the “reach” of the relay. Distributed Generation infeed can reduce the current that the relay sees, thereby shortening its reach. When the total Distributed Generation capacity increases to a certain amount, the infeed into faults can desensitize the relays and leave remote sections of the feeder unprotected.
A low-current (high-impedance) fault near the end of the feeder is more likely to go undetected until it does sufficient damage to develop into a major fault. The power quality consequences of this are that voltage sags will be prolonged for some customers and the additional fault damage will eventually lead to more sustained interruptions.
This issue can be a particular problem for peaking generation located near the end of the feeder. This generation is on at peak load level where the overcurrent relaying would normally be very sensitive to a high-impedance fault. The Distributed Generation infeed has the potential to mask many faults that would otherwise be detected.
i. Add a line recloser to create another protection zone that extends far past the end of the feeder.
ii. Decrease the relay minimum pickup current to increase the zone. This may not be practical for ground relays that are already set to a very sensitive level.
iii. Use a transformer connection that minimizes DG contribution to ground faults, since high-impedance faults are likely to be ground faults.
Sympathetic tripping describes a condition where a breaker that does not see fault current trips “in sympathy” with the breaker that did. The most common circuit condition on utility distribution feeders is backfeed into a ground fault. For the situation shown in Fig. 6.10, the source of the backfeed current is the Distributed Generation. Most utility feeder breakers do not have directional sensing. Therefore, the ground relay sees the Distributed Generation contribution as a fault and trips the breaker needlessly. This situation is exacerbated if the service transformer for the Distributed Fig. 6.10 Sympathetic tripping of feeder Generation has a grounded wye-delta connection.
The main solution to this problem is to use directional overcurrent relaying. If appropriate potential transformers are not already present, this could end up being an expensive alteration. Since the Distributed Generation contribution in breaker B is likely to be much lower than the fault current through breaker A, it may be possible to achieve coordination with the appropriate time-delay characteristic or by raising the instantaneous (or fast) trip pickup past the amount of Distributed Generation infeed.
The power quality impact of the sympathetic tripping is that many customers are interrupted needlessly. The Distributed Generation is also forced off-line, which could be a problem for the Distributed Generation owner. There could be impacts from the solutions as well. By slowing the ground trip, there will be more arcing damage to lines and through-fault duty on transformers.
This could eventually lead to increased failures. Fuse saving is commonly practiced in utility overcurrent protection schemes, particularly in more rural regions.
The desired sequence for the situation depicted in Fig. 6.11 is for the recloser R to operate before the lateral fuse has a chance to blow. If the fault is temporary, the arc will extinguish and service will be restored upon the subsequent reclose, which normally takes place within 1 or 2s. This saves the cost of sending a line crew to change the fuse and improves the reliability of customers served on the fused lateral.
Fuse-saving action is a “horse race” in the best of circumstances. It is a challenge for the mechanical recloser to detect the fault and operate fast enough to prevent damage to the fuse element. Distributed Generation infeed adds to the current in the fuse and makes this race even tighter. At some amount of Distributed Generation capacity that is capable of feeding the fault, it will no longer be possible to save the fuse.
This phenomenon limits the amount of synchronous machine Distributed Generation that can be accommodated without making changes to the system. Fuse- saving coordination fails for about the same level of generation that causes voltage regulation problems.
Solutions includes are:
i. Increase the size of the lateral fuses. All fused cutouts in the zone would have to be changed, which could be quite expensive.
ii. Choose to simply abandon fuse saving, particularly if the Distributed Generation is only connected intermittently.
iii. Require Distributed Generation to have transformer connections that do not feed single-line-to-ground faults.
The power quality impacts of this are mixed. While the utility generally views fuse saving as an improvement in power quality, customers tend to view the short blink as poor service. Therefore, many utilities have already abandoned fuse saving in many areas.
4. Voltage Regulation Issues:
Voltage regulation issues are more likely to occur and cause interconnection problems. Figure 6.12 illustrates one voltage regulation problem that can arise when the total Distributed Generation capacity on a feeder becomes significant. This problem is a consequence of the requirement to disconnect all Distributed Generation when a fault occurs. Figure 6.12(a) shows the voltage profile along the feeder prior to the fault occurring. The intent of the voltage regulation scheme is to keep the voltage magnitude between the two limits shown. In this case, the Distributed Generation helps keep the voltage above the minimum and, in fact, is large enough to give a slight voltage rise toward the end of the feeder.
When the fault occurs, the Distributed Generation disconnects and may remain disconnected for up to 5 min. The breaker recloses within a few seconds, resulting in the condition shown in Fig. 6.12(6). The load is now too great for the feeder and the present settings of the voltage regulation devices.
Therefore, the voltage at the end of the feeder sags below the minimum and will remain low until voltage regulation equipment can react. This can be the better part of a minute or longer, which increases the risk of damage to load equipment due to excessively low voltages. Of course, this assumes that the voltage regulation devices are not already at the maximum tap position. Utility planners will often point out that this is one of the dangers of relying on Distributed Generation to meet capacity. It masks the true load growth on the system, and there is insufficient base capacity in the wires to deliver the power.
This issue can be one of the more limiting with respect to how much Distributed Generation can be accommodated on a feeder. It is particularly an issue for lengthy feeders on which the Distributed Generation is located some considerable distance from the substation. This may be an attractive application of Distributed Generation because it defers the construction of major wire facilities to serve the remote area. However, it can come at the cost of having to modify long-established operating practices and sacrificing some reliability of the system.
It also suggests one test an analyst can perform to determine if a proposed DG application will likely require changes on the utility system.
One useful analysis is to determine Distance from how much DG capacity can be accommodated (without change) at various distances along the feeder. For example, if one were to establish a 5 percent change criterion for the limit, there would be a curve of generation limit versus distance similar to that shown in Fig. 6.13.
This simple analysis will work for one DG site per feeder. This may be adequate if penetrations of DG are low, but the problem can get complicated quickly as more sites are added. One approach is to study many random distributions of small DG at peak load. This will result in a more conservative screening curve that is shifted downward and to the left.
When something must be done, solutions include:
i. Requiring customer load to disconnect with the Distributed Generation. This may not be practical for widespread residential and small commercial loads. Also, it is difficult to make this transition seamlessly and the load may suffer downtime anyway, negating positive reliability benefits of Distributed Generation.
ii. Installing more voltage regulators, each with the ability to bypass the normal time delay of 30 to 45 s and begin changing taps immediately. This will minimize the inconvenience to other customers.
iii. Allow Distributed Generation to reconnect more quickly than the standard 5-min disconnect time. This would be done more safely by using direct communications between the Distributed Generation and utility system control.
iv. Limit the amount of Distributed Generation on the feeder.
Utility voltage regulators commonly come with a reverse-power feature that allows the regulators to be used when a feeder is supplied from its alternate source. The logic is that when the net power through the regulator is in the reverse direction, the regulator control switches direction and regulates the original source terminal so that the regulator can work properly. Otherwise, the control will attempt to regulate the alternate source side, which would not be possible.
The tap position would generally move to one extreme or the other and stay there. Assume, for example, that several cogeneration sites have been added to a feeder and there is excess generation when the load is low.
The regulator now senses reverse power and attempts to regulate the utility source. However, the Distributed Generation is not nearly as strong as the utility source and the regulator will not succeed. Similar to the case where the controls fail to switch direction on the alternate feed, the tap will be run to an extreme position, often in the worst possible direction.
To prevent this, regulator vendors have come up with cogeneration features on the controls that can detect this condition. The desired result is to keep the regulator looking in the forward direction. The line-drop compensator R and X settings may also be changed while the reverse-power condition exists.
Generation technologies whose output varies rapidly can be difficult to handle on a distribution feeder. Wind-turbine generation is the most difficult because there is seldom a substation near the proposed site. The generation is typically sited several miles from the nearest substation on a feeder that already may have several switched capacitors and a voltage regulator.
One example based on a proposed wind farm at a ski resort is shown in Fig. 6.15. The line is a typical un-transposed, horizontal cross-arm geometry that leads to special issues. As the power output of the generator varies, one outside phase will tend to rise in voltage while the other tends to drop. Not only is there a magnitude issue but a balance issue.
When the fluctuation is too great, the main recourse is to build a separate feeder for the wind generation. Also, some wind generators employ doubly fed wound- rotor induction machines that can control reactive power very quickly. With proper control, this can help tame the voltage fluctuations.
Harmonics from Distributed Generation come from inverters and some synchronous machines. The PWM switching inverters produce a much lower harmonic current content than earlier line-commutated, thyristor-based inverters.
In IEEE Standard 519-1992, generators are limited to the most restrictive values in the tables on the allowable amount of harmonic current injection. While generator inverters are not necessarily any worse than power converters used in loads, the developers of the IEEE standard allocated all the capacity in the system to loads, leaving very little for generators. Fortunately, the shift to PWM switching technology has made it relatively easy for inverters to meet the standard.
One new distortion problem that arises with the modern inverters is that the switching frequencies will occasionally excite resonances in the primary distribution system. This creates non-harmonic frequency signals typically at the 35th harmonic and higher riding on the voltage waveform. This has an impact on clocks and other circuitry that depend on a clean voltage zero crossing.
A typical situation in which this might occur is an industrial park fed by its own substation and containing a few thousand feet of cable. A quick fix is to add more capacitance in the form of power factor correction capacitors, being careful not to cause additional harmful resonances. Characteristics of synchronous machines, there can be harmonics problems related to zero-sequence triplen harmonics.
Figure 6.16 shows a typical situation where this occurs. The facility where the generator is located is served at 480 V by a common delta-wye transformer. When the generator is paralleled to the utility system through this transformer, the operator is frequently surprised to find a large amount of current circulating in the neutral.
In the example shown, the current is 26 percent of the machine’s rated current and is entirely third-harmonic current. This can adversely affect the operation and efficiency of the machine and may result in the failure of some circuit element. In this case, the problem is confined to the generator side of the transformer and does not affect the primary distribution system because the triplen harmonics are trapped by the delta winding. The same thing can happen with a grounded wye-wye transformer, except that the harmonic currents do reach the primary distribution system.
This problem is well known among vendors of standby generation equipment. If known beforehand, most will recommend a machine with a 2/3 winding pitch that can be paralleled without this difficulty. If it is necessary to parallel a design that does produce significant triplen harmonics, a reactor can be added in the neutral to limit the current flow.
Distributed Generation protective relays will generally perform their function independently of any outside knowledge of the system to which they are connected. Perhaps the greatest fear of the utility protection engineer is that Distributed Generation relaying will fail to detect the fact that the utility breaker has opened and will continue to energize a portion of the feeder.
Therefore, much attention has been paid to detecting islands or forcing islands to become unstable so they can be detected. The reliability concern is that other customers will be subjected to such poor-quality voltage that damage will be sustained. The utility is fearful it will be held liable for the damage. There is also the safety concern of a generator accidentally energizing the line resulting in injuries to the public and utility personnel.
Another concern is the Distributed Generation itself. Since reclosing is common, it is essential that the Distributed Generation detect the island promptly and disconnect. If it is still connected when the utility breaker recloses, damage can occur to prime movers, shafts, and components of machines due to the shock from out-of-phase reclosing. This highlights one area of potential conflict with utility practices: Those utilities using instantaneous reclose may have to extend the reclose intervals to ensure that there is sufficient time for Distributed Generation to detect the island and disconnect.
Relaying, is one way to address the issue. The main keys are to detect the deviations in voltage and frequency that are outside the values normally expected while interconnected.
Another approach to anti-islanding is to make requirements for the operating mode for the Distributed Generation while interconnected that significantly reduce the chances that the generation will match the load when an inadvertent island forms:
i. Inverters operating in parallel are less likely to form an island if they are acting as current sources and have a destabilizing signal that is constantly trying to shift the frequency reference out of band. Islanding would require another source to provide a voltage for the inverter to follow. Of course, this source could be provided by any synchronous machine Distributed Generation that remains on the island.
ii. Interconnected Distributed Generation should operate in a mode that does not attempt to regulate voltage. This usually means a constant power factor or constant reactive power mode. For many inverter-based devices, this will be unity power factor, producing watts only. Automatic voltage control should be avoided while Distributed Generation is interconnected to the distribution system unless the generator is directly connected to a control center to receive dispatching and transfer trip signals.
Ferroresonance is a special kind of resonance in which the inductive element is the nonlinear characteristic of an iron-core device. Most commonly, ferroresonance occurs when the magnetizing reactance of a transformer inadvertently is in series with cable or power factor capacitance.
One interesting case occurs for Distributed Generation served by cable-fed transformers. It is common practice for the larger Distributed Generation installations to have their own transformer. Also, it is nearly universal to require Distributed Generation to disconnect at the first sign of trouble on the utility system. This combination of requirements can lead to a common ferroresonant condition.
The circuit is shown in Fig. 6.17. Underground cable runs are normally fused at the point where they are tapped off the overhead feeder line. This is variously called the riser pole or dip pole. Should something happen that causes one or two fuses to blow, the relaying on the DG will detect an unbalanced condition and trip the generator breaker.
This leaves the transformer isolated on the cable with one or two open phases and no load. Either condition is conducive to ferroresonance because the cable capacitance in an open phase, or phases, now appears in series with the transformer’s magnetizing impedance.
There are several reasons why the riser-pole fuse may blow or become open. Normally, they are designed to blow for faults in the cable, but there are other reasons. Squirrels or snakes may climb the pole and get in contact with the line. Fuse elements may also fatigue due to frequent inrush currents or lightning surge currents. Fused cutouts may open due to corrosion or improper installation. Finally, they may be operated by a line crew for maintenance purposes.
Whether a blown fuse results in damage is dependent on many variables and the specific design of the equipment. A number of ferroresonance modes are possible, depending on the connection of the transformer, its size, and the length of cable. The most susceptible transformer connections are the ungrounded ones. The delta configuration with one phase open is shown in Fig. 6.17. The over-voltages for this condition that can occur can easily reach a value of 3 to 4 pu unless limited by arresters.
Figure 6.19 shows the voltages computed for a 300-kVA delta-connected transformer fed by cable that has 30 nF of capacitance. This models a case in which there is no load on the transformer and no arresters. Arresters would clamp the voltage to a lower value unless they had thermally failed from prolonged exposure to this waveform.
The high voltages and the chaotic wave-shape are due to the transformer slamming in and out of saturation. The magnetic forces associated with this change cause the core to emit very loud noises that are sometimes described as the sound of a large bucket of bolts being shaken or a chorus of hammers on anvils.
This can cause failures of both the arresters protecting the transformers and the transformers themselves. Arresters fail thermally, leaving the transformers unprotected. Then the transformers may fail either from thermal effects or from dielectric failure. It is common for low-voltage arresters and transient voltage surge suppressors to suffer failures during this type of ferroresonance.
At one time, it was believed that grounded-wye connections were impervious to ferroresonance. However, this theory was shown to be false in a landmark paper. While grounded-wye transformers made up of three single-phase transformers, or a three-phase shell core design, are immune to this type of ferroresonance, the majority of pad-mounted transformers used in commercial installations are of three-legged or five-legged core design. Both are susceptible to ferroresonance due to phase coupling through the magnetic core.
Although not immune, the over-voltages are lower than with ungrounded connections, typically ranging from 120 to 200 percent. Sometimes, the voltages are not high enough to cause failure of the transformer. The utility line crew responding to a trouble call encounters the transformer making a lot of noise, but it is still functional with no detectable damage.
In other cases, there could be a burned spot on the paint on the top of the tank where the high fluxes in the core have caused heating in the tank. Primary arresters should be tested and the secondary system inspected for failed equipment before reenergizing.
This situation is not necessarily unique to Distributed Generation installations. Many modern commercial facilities are supplied with cable-fed service transformers that are disconnected from the mains when there is a problem on the utility system. The purpose would be to switch to UPS systems or backup generation. Unfortunately, it leaves the transformer isolated with little or no load.
As a general rule there should be no line fuses or single-phase reclosers between the generator and the utility substation. This is to prevent single-phasing the generator, which could not only result in Ferro resonance but could thermally damage rotating machines. This rule is particularly appropriate for Distributed Generation with cable- fed service transformers. The riser-pole fuses may be replaced with solid blades (no fuses) or three-phase switchgear such as a recloser or sectionalizer.
Replacing the fuses with solid blades will reduce the reliability of the feeder section somewhat. Each time there is a fault on the cable system, the entire feeder or feeder section will be out of service. If the cable is short and dig-ins unlikely, this may very well be the lowest-cost option. If protection is required, the three-phase switchgear option is preferred.
The type of ferroresonance shown in Fig. 6.19 is very sensitive to the amount of load. If the system can be arranged so that there is always a resistive load attached to the secondary bus, the resonance can be damped out. The load need not always be large, but must be significant.
In the example cited, a 2 percent load (6 kW) was sufficient. However, in other cases, more than a 10 percent load may be required.
8. Shunt Capacitor Interaction:
Utilities use switched capacitors to help support the voltage during high-load periods. These banks are mostly controlled by local intelligence, switching at predetermined times or at loading levels as measured by either voltage, current, or kvar. Some types of Distributed Generation can also produce reactive power (vars), and this can create control hunting and other difficulties.
There can be several capacitor banks on the feeder. The capacitors switch independently of the generator control unless special communications and control have been added to coordinate dispatch. A 2 to 3 percent increase in the voltage is common when a typical capacitor bank is energized. Generators in parallel operation are generally maintaining a constant power and power factor. The reactive power of the machine is controlled by the exciter field, which will have certain minimum and maximum voltage or field excitation limits as indicated on the diagram.
The generator control attempts to maintain a constant reactive power output until it bumps up against one of these limits. There can easily be conditions in which the total reactive power output of the generators and capacitors is too great, resulting in high voltages. This is particularly likely when capacitors are switched by time clock or by current magnitude (without voltage override).
There are at least three things that can happen at this point to trip the generator:
i. The generator control senses overvoltage at its terminals and attempts to back down the field to compensate. However, the utility system overpowers the generator and the field reduces to a level deemed to be too low for safe operation of the machine.
ii. When the generator reaches its voltage limit, reactive power flows back into the machine. When it reaches a certain level, the generator protection interprets this as a malfunction.
iii. Distributed Generation that does not produce reactive power simply trips on overvoltage.