Hawaiian electric co. is using distributed generagion (DG) on the idland of Oahu as a strategy to meet growing loads and to satisfy reserve margin requirements until a new 110-MW central power plant becomes operational in 2009. HECO (Honolulu, Hawaii, U.S.) is installing 4.92 MW of DG at each of six substations, for a total of 29.52 MW. Five sites are already in operation, and the other site will become operational in 2007. Four of these DG sites are located at distribution substations and two are on primary feeders within 1 mile (1.6 km) of the substation. All the distribution systems are 12.47-kV or 11.5-kV, four-wire multigrounded neutral circuits.

The generators employed are Caterpillar trailer-packaged diesel units with separately excited synchronous generators (Type SR4B). The rating of each machine is 1.64 MW and 2.05 MVA. They were supplied as integrated packages by Hawaiian-based Hawthorne Power Systems. Each DG site consists of three of these units in parallel yielding a total capacity of 4.92 MW per substation.

The units are under the control of the HECO dispatch center and used primarily as peak shavers to support the central power system. They are typically dispatched on peak load days between 10 a.m. and 9 p.m. local time. The total annual operating hours for each plant is typically in the range of 300 hours to 500 hours, but the sites are air permitted for up to 1200 hours of annual operation per unit if needed. The central power system on the island of Oahu has a peak load of about 1200 MW, so these plants represent 2.5% of that peak load.

Each site includes three generator units, three 1500-kVA step-up transformers, protective relays and metering/remote terminal unit equipment as shown in Fig. 2. The step-up transformers interface the 480-V generator output to primary feeder levels of 11.5 kV or 12.47 kV. Protection is done primarily with a Beckwith 3410A relay for each generator using a wide range of functions. In addition, there is a direct-transfer trip (DTT) from the feeder circuit breaker to the DG for islanding protection.

When needed, some sites also include a DTT between the subtransmission system breaker and the DG. The sites are not designed for islanded operation and will trip off-line when subjected to a sufficiently severe voltage or frequency disturbance, or when the DTT trip signals arrive. Voltage-check relays are also installed, and reclosing dead times are adjusted on certain circuit breakers as a further safeguard. The grounding reactor on the DG was sized to limit line-ground fault levels and generator zero-sequence currents, while still maintaining effective grounding with respect to the distribution system. This limits the ground-fault overvoltage on the distribution system to acceptable values.

The large size of these generators, with respect to the substation capacity, means that they have a large influence on the distribution system and an influence on the local subtransmission system. Because of this concern, extra care was applied to address the system protection and power-quality interactions with various modeling studies by both HECO and Nova Energy Specialists (NES; Schenectady, New York, U.S.), the contractor selected for the independent system assessment.

A variety of tools were used for the analysis. NES used the EMTP-RV software for steady-state and dynamic-response simulations that included factors such as fault contributions, voltage regulation analysis, stability, ferroresonance, ground-fault overvoltages and others. These studies identified several important power-system changes and upgrades to ensure proper coordination of protective devices and that system power quality, reliability and safety were preserved.

One of the areas studied was the DG plant impact on steady-state voltage. It was immediately apparent that the DGs are so large they “mask” the feeder load currents normally seen by the substation transformer load-tap changer (LTC) controls. Therefore, on circuits where line-drop compensation is employed, the LTC would not boost the voltage properly. This will result in low voltage at the ends of the distribution circuit during peak system loads (Fig. 3).

To mitigate this problem, the studies showed that either extensive modification of the LTC controllers would be required or the line-drop compensation of the LTC would need to be disabled and a higher LTC set voltage applied to compensate for the lack of line-drop compensation. The latter approach requires no equipment modifications other than setting changes, making it a good solution for temporary DG sites, which is why HECO adopted it.

Another voltage regulation concern involved reverse power passing through the substation transformer during light load conditions. Reverse power can confuse some types of LTC voltage regulator controllers, resulting in lock up of the LTC, or runaway tap changers. Under these conditions, the voltage regulation could be quite unpredictable. To avoid problems, all LTC controllers with reverse-power detection functions were identified and the function was disabled.

The starting and stopping of generators can cause discernible voltage flicker or voltage change on the feeders. The magnitude of voltage change depends on how many units are started, the machine's power factor, the system impedance and how rapidly the units are ramped up to full power. Calculations show that the units would cause discernible voltage flicker per the IEEE 519-1992 flicker curve if they were all stopped or all started suddenly.

Table 1 presents some of the scenarios examined. The studies determined that the generator should ramp power output up and down slowly over a couple of minutes. This avoids a sudden change of voltage and makes it much less discernible to the human eye. This practice also gives the LTC a chance to respond, given its 30-second time delay. In addition, the effect can be reduced further if only one machine is started at a time, even in quick succession.

The operating mode of the generator exciter and governor is also important. To avoid “hunting” interactions between LTC and DG, the DG were set to operate in a fixed power-factor mode and not in a voltage-regulating mode. This allows the units to “voltage follow” and will prevent them from fighting against the LTC controller for a desired voltage set point. It also allows the operator to dispatch a known quantity of real and reactive power.

The EMTP-RV models were used to perform dynamic simulations of the machine response to various voltage-sag events on the power system. Stability of the machines during moderate events was important since it was desirable to maintain reliability of dispatch. It was shown that a large DG connected to a high-impedance utility source can experience rotor angle oscillations during minor disturbances that amplify the duration of the voltage disturbance. This may make minor disturbances more visible than they otherwise would be (Fig. 4).

Utilities considering the use of very large DG on lines need to consider these dynamic effects and the stability of the machines. In the case of the HECO sites, the units were at or close enough to the substation bus to help limit the voltage oscillations. But, if the DG is further out on the feeder (or near the end) or of a higher rating, the oscillations could become unacceptably large.

The generators contribute fault current to power system faults with a decaying envelope as illustrated in Fig. 5. The large size of the DG compared to the substation transformer rating means the DG will have a significant impact on power system fault levels. This fact is especially important for evaluating sympathetic tripping and lateral fuse coordination issues.

The fault levels at the substations when DG is operating were 15% to 40% larger than fault levels from the utility source alone. Surprisingly, even with such large fault contributions, they were not above thresholds that would cause interference with feeder circuit breakers or lateral fuses. This is because HECO does not use instantaneous tripping on the feeder circuits studied and does not practice “fuse saving.” If those practices are present, then the study conclusions are quite different. In that case, the DG could have caused a variety of issues including sympathetic trips during adjacent feeder faults and problems with lateral fuse coordination.

One area where the DG does impact fusing at HECO is distribution transformer fusing. The fault level is increased sufficiently that current limiting fuse (CLF) application points for distribution transformers on feeders may need to be reevaluated.