protection-mg

Consortium for Electric Reliability Technology Solutions White Paper on Protection Issues of The MicroGrid Concept Prepared for Transmission Reliability Program Office of Power Technologies Assistant Secretary for Energy Efficiency and Renewable Energy U.S. Department of Energy Prepared by William E. Feero, Consultant, Reedsville, PA Douglas C. Dawson, Consultant, North Hollywood, CA John Stevens, Sandia National Laboratories, Albuquerque, NM March, 2002 The work described in this report was coordinated by the Consortium for Electric Reliability Technology Solutions, and funded by the Assistant Secretary of Energy Efficiency and Renewable Energy, Office of Power Technologies of the U.S. Department of Energy under Contract No. DE-AC03-76SF00098 MICROGRID PROTECTION ISSUES Introduction and Summary This report examines the protection problems that must be dealt with to successfully operate a microgrid when the utility is experiencing abnormal conditions. There are two distinct sets of problems to solve. The first is how to determine when an islanded microgrid should be formed in the face of the array of abnormal conditions that the utility can experience. The second is how to provide segments of the microgrid with sufficient coordinated fault protection while operating as an island separate from the utility. As used in this discussion, the term microgrid refers to conventional distribution systems with distributed resources (DR) added. This is not to imply that the simple addition of DR to a distribution system creates a microgrid. In a microgrid the DR(s) has sufficient capacity to carry all, or most, of the load connected to that portion of the distribution system that houses the DR. In addition, a microgrid can operate as an electrical island in times of disturbance to the main utility system. Thus, there will be a well-defined interconnection point where the microgrid can be disconnected from the bulk of the electric utility system if so desired. Figure 1 is an example of what a microgrid’s network might look like. Shedable loads Power/Heat Coordinator Power Flow Controller Circuit Breaker Heat Load Protection Coordinator Communication Fast protection Shedable loads Power/Heat Coordinator Power Flow Controller Circuit Breaker Heat Load Protection Coordinator Communication Fast protection Figure 1 Typical Microgrid 2 The discussion in this report generally presumes that basic elements of microgrids (such as fuses, circuit breakers and over-current detection devices) will perform in a manner consistent with those present in existing distribution systems. An element that is not well defined is the inverter interface with the power system. There is a concern that, not only will each inverter design have different constants, but that the basic characteristics of the unit presented to the system can change markedly depending on the design goals of a particular manufacturer and/or application. It seems certain that the inverter fault current capability will be less than twice the rated current of the inverter unless the unit is specifically designed to provide high fault current. This is a marked departure from the relatively high fault-current capability typical of synchronous generation capability. If a significant portion of the microgrid’s generation has inverter interfaces, the change from utility-connected operation to islanded microgrid operation may aggravate a concern for using current-based fault detection. The protection systems must respond based on a pre-set understanding of the boundary between normal and abnormal conditions. Many protection issues associated with microgrids will not be truly resolved until the millisecond-by-millisecond dynamics prior to, during, and following microgrid separation from the utility are well understood. In fact, during an impromptu meeting on this subject with approximately 30 utility protective relay engineers, the salient need they expressed was the ability to simulate the expected response of the microgrid when separating from the utility and in disconnected microgrid operation. Therefore the findings of this report are general observations based on the present understanding of expected microgrid dynamic response characteristics. 3 Making the decision of when to separate from the utility first requires a realistic expectation of what benefit the microgrid will gain from rapid separation. While equipment standards such as SEMI F47 suggest that many manufacturers would benefit from separation times of less than 50 milliseconds after an abnormal condition is started on the utility, such times are not possible with presently available protective devices (as noted in footnote 1, below). If very high speed separation is required and nuisance trips are to be avoided (see the Nuisance Separation section below for a discussion of whether nuisance separations should be tolerated in a trade for improved reliability), then transfer trip systems must be installed between the utility substation and the breaker at point of common coupling (PCC) with the microgrid. (Note that the PCC is the point where the Main Microgrid Separation Device is shown on figure 1.) Separating for non-fault conditions will also benefit from having high-speed communication channels between the utility and the microgrid. When the microgrid has separated from the utility, a host of new issues must be accommodated. A means to assure appropriate grounding must be provided. The equipment used for detection of faults internal to the microgrid must work with, or around, the protection that exists for fault detection while connected to the utility. A means of detecting faults that is not dependent on a large ratio between fault current and maximum load current must be provided. Any existing anti-islanding schemes will have to be examined, and perhaps modified, to prevent instability or loss of DR units with sensitive settings. Any load shedding schemes set up by the utility in the microgrid area will have to be closely coordinated. The following sections explore these areas to determine what might be done with present protective devices to resolve these issues. Modern protection techniques are highly evolved from nearly a century of experience. Thus caution should be used in any attempts to revolutionize approaches to system protection. Grid Separation In a preliminary evaluation for this report, the following issues were listed for consideration: x� Protection operation speed required to approach SEMI F47 specifications1 x� Nuisance separations: how to minimize x� Is non-fault separation (for low voltage, open phase, voltage imbalance, etc.) desirable? x� Separation protection limitations imposed by microgrids that export power to the utility 1 Note that it is not possible, at least with electromechanical switching devices, to meet the SEMI F47 specification for all fault types and locations. The SEMI specification does not permit voltages below 50%, even with durations as short as 0.05 sec. (3 cycles at 60 Hz.) If a utility feeder fault causes the voltage to drop below 50%, there is nothing that can be done except to separate as quickly as possible (see Figure 2). 4 x� Re-synchronization to utility: enable automatically or manually, match frequency and voltage automatically or manually? Implicit in this list is the acceptance of many concurrent factors. Microgrid means very small power capacity relative to the utility, e.g., less than 10 MVA. Any system that is designated as a microgrid will have sufficient generation sources to carry a significant portion of the microgrid’s load. As such, under improper separation from the utility, i.e., not at the PCC, it will have the possibility of carrying part of the utility. IEEE standards activity will soon be setting minimum interconnection protection criteria, which any microgrid may have to meet while operating interconnected. Economics is important. Finally, the reality that even excellent protection packages cannot compensate for all of the special needs to operate a microgrid must be recognized. Rapid Separation from a Faulted Feeder One of the goals of forming a microgrid is to maintain uninterrupted power to critical loads. As indicated above, if the loads of the microgrid are so sensitive to voltage dips that they require the system to meet SEMI F47 specifications, then existing protective equipment cannot act fast enough to separate from the fault under all conditions. Secure relay time to detect an under (or over) voltage can be up to two cycles. Most medium voltage breaker times, after receiving a trip signal, require from three to five cycles to interrupt the circuit. Therefore, unless the microgrid interface device at the point of common coupling is a solid state circuit breaker, other system design considerations will have to be employed to prevent the voltage from going below 50% for three cycles or longer. SEMI F47 Sag Immunity 0 10 20 30 40 50 60 70 80 90 100 0.01 0.1 1 10 100 S a g Dur a t i on ( se c ) Sag Ride-Through Region Figure 2, SEMI F47 Voltage verses Time Limits 5 To combine design and protection improvements, first let’s look at the case where separation is not required, i.e., the fault is not located between the PCC and the substation breaker. For example, a fault causing a sag on a substation bus may be on an adjacent feeder fed from the same substation. An example of design considerations could be to install electronic sag correctors that have recently become available. Or, if the transformer at the point of common coupling is the widely used wye-wye connection, replacing it with delta-wye transformer and adding a high side breaker may be another option. (For single-line-to-ground, SLG, faults in the utility, this transformer connection will ensure that the phase-to-ground voltage in the microgrid does not drop below 58%). These two possible options are useful to demonstrate how protection considerations and design options must be considered together in evolving economic microgrids. Electronic sag protectors are available that do not use energy storage. These devices are typically effective for only a couple of cycles. For longer protection periods, extending into the realm of many minutes, electronic sag protectors that incorporate an energy storage device, such as a battery, are necessary. Microgrid owner/operators might be reluctant to pay the increase in cost to install electronic sag protectors that utilize any significant amount of energy storage. However, emerging devices do not require significant storage if the under voltage does not last too long or is not less than 50%. For abnormal conditions that cause zero voltage, most electronic sag protectors can hold up the voltage for three cycles. If instantaneous relays and three cycle breakers are applied to all feeders adjacent to the one supplying the microgrid, then, at least for adjacent feeder faults, the combination of electronic sag correctors and high speed relaying may meet the SEMI F47 requirements. Of course, the issue of faults within the microgrid still remains. The solution will depend on utility practices such as fuse saving. If modern power quality requirements govern the utility protection practice, they will be using quick blow fuses in which case the electronic sag correctors should still make it possible to meet SEMI F47. If the utility uses instantaneous overcurrent tripping of the feeder (for fuse saving purposes), then separation of the microgrid will be required. Using a delta-wye interconnecting transformer may be a low cost, albeit less effective solution. Since SEMI F47 only allows the voltage to be depressed below 70% for 0.2 seconds, the utility protective action must still be rapid. Also the delta-wye interconnection is only effective for single-line-to-ground faults on the delta side and loads connected line-to-neutral on the wye side. Now consider the case where the fault is on the feeder between the microgrid and the substation (that is, “upstream” of the Main Microgrid Separation Device indicated on figure 1). For this case the microgrid must separate from the utility. Emerging standard requirements, utility protection requirements, and any hope of meeting SEMI F47 requirements all converge on the need to separate rapidly. For this condition, there is no maintaining even a low under voltage tie to the utility. The present speed limitations of the protective devices now become insurmountable without electronic sag correctors with storage if SEMI F47 is the target criteria. The amount of storage contained in the electronic sag corrector can be kept reasonably small if high speed tripping is employed. This presumes the microgrid has adequate controls to rapidly recover the voltage and maintain near system frequency when islanded by clearing a tie line fault. 6 Nuisance Separations The above discussion should have made it obvious that maintaining the tie between the utility and the microgrid is highly desirable except when the fault is on the tie between the substation and the microgrid. It also should be obvious that when separation is required it should be rapid as possible. Unfortunately inexpensive protective schemes are not secure, that is, they are prone to false trips. The emerging standards specify mandatory voltage and frequency trip settings for measurements made at the PCC. However, voltage and frequency are poor discriminators for determining if the fault is on the feeder to the microgrid, or within the microgrid itself. Currently the only reliable and secure means for rapidly tripping the microgrid breaker is a transfer trip from the substation breaker. This security from false or nuisance trips is not just an electromechanical relay and breaker problem. Even the most sophisticated microprocessor package acting solely on information available at the PCC cannot always determine the fault location given the extreme difference in energy capability between the utility and the microgrid. The difference between a distributed resource and a microgrid is most clear when considering the importance of nuisance trips. For a simple DR interconnection, the cost of a nuisance trip is normally just the lost kWh sales for a brief period and the cost of restart and re-synchronizing. For a microgrid, a nuisance trip is significant exposure to unacceptable power quality problems. Therefore the cost of interconnection protection should be examined as insurance against potential manufacturing production lost, not merely kWhrs lost. However, if the microgrid has been designed as a valid backup source of power for the loads of the microgrid or a “power park,” then nuisance separations are not all that serious and can be tolerated as preferable to remaining connected to a utility feeder that is undergoing a disturbance and may experience an outage. Separating from the utility removes the disturbance from the microgrid system and allows it to continue operating unmolested. A nuisance separation has little impact on either the microgrid or the utility as long as the microgrid has been adequately designed. The big advantage of tolerating nuisance tripping is that the relaying threshold conditions for separation can be defined by voltage and frequency deviations and time durations, without regard to whether these are good indicators of the location of a utility system fault. That is, if the voltage is out of range for a time exceeding the allowable, separation should occur, regardless of whether the underlying disturbance will ultimately result in an outage to the utility feeder. This is a simpler relaying problem than trying to estimate, from voltage and current measurements at the PCC, whether the utility fault is between the utility substation and the PCC, or elsewhere. Microgrid designs that might argue against such an approach are: Microgrids which are intended to shed non-critical load upon separation. (That is, the DR capacity within the microgrid cannot dependably support the loads of the 7 microgrid.) In such cases, the outages to non-critical loads might become a real nuisance. Microgrids which export power to the utility under normal operating conditions. In these conditions, nuisance separations would entail a loss of revenue and a period of overfrequency operation while the frequency on the microgrid stabilizes. Also, the utility might feel that such frequently interrupted generation was not worth as much as generation not so subject to interruptions. Non-fault Separation Low voltages can exist for non-fault conditions. The determination of whether a low- voltage condition is an indication of a fault in the system between the PCC and the substation can be difficult without high-speed communication between the PCC and the substation. In general, the utility and the microgrid would benefit from remaining connected while the utility works to resolve the cause of the low voltage if it is not the result of a fault that requires tripping at the PCC. It might be desirable for the microgrid and the utility to negotiate a trip control to coordinate with the SEMI F47 voltage limits for balanced voltage conditions. Such control could either be achieved by communication with the utility or by balanced voltage blocking of single phase under- voltage relays at the PCC when the desired voltage trip levels are lower than the delayed trip settings that may be required by other standards such as are being developed within the IEEE P1547 project. However, since the levels being considered in P1547 are set to determine an unintended island condition, there may not be a conflict because P1547 as presently written does not cover “intentional islanding”. Relaying systems could be designed from presently available devices to provide the above trip restraint for balanced under voltage conditions. The larger question is can the microgrid recover from operation at such depressed voltage levels if tripping is eventually required? These microgrid operational restraints are more likely to determine the under-voltage trip point. Voltage unbalance normally exists to some degree on most distribution feeders under normal operating conditions. The desirability for the microgrid to separate at some level of voltage unbalance will be a function of such factors as the transformer connections and grounding points within the microgrid. The distribution apparatus, distributed generation and loads being served all need to be considered for their sensitivity to voltage unbalance to determine the criteria for establishing the microgrid design. For voltage unbalance conditions, the protective function dilemma is how to determine whether the cause for the unbalance is internal to the microgrid or external to it. Such a determination may be complicated by the ratio of the power supplied internal to the microgrid to the load demand internal to the microgrid. Some form of intelligent controller at the PCC could be needed to make the appropriate decision to separate or not based on voltage unbalance. 8 Open phase occurrences are generally associated with systems where fuses are located between the substation and the PCC. While non-fault initiated open phases are rare, they do occur. Detection of the event is complicated, depending on the number and type of transformers between the open phase and the PCC where a three-phase switching device must exist. Since an open phase will permit phase-to-phase voltages to remain at or above 50%, it might be tempting for the microgrid not to want to separate for this condition. However,