Transfer Switches: Open or Closed?

Editor’s note: Following is a white paper prepared by James Daley on specifying open vs. closed transfer switches. Daley is a participant in CSE’s upcoming Feb. 2003 M/E Roundtable discussion on switchgear and transfer switches and is a consultant to ASCO Power Technologies, Florham Park, N.J.

What factors should engineers take into consideration when determining whether to specify a closed- or open-transition switch, or three-pole vs. four-pole transfer switch? How about insulating buswork?

Let’s answer the easy question first, insulating bus. Switchgear and switchboard bus is insulated. If the question is asking about encapsulated bus, then there are some issues that need to be addressed. In medium voltage switchgear the bus is commonly encapsulated. However, the continuous bus rating is 3 kA and below.

The issue is a bit more complex for low-voltage systems. Some manufacturers have bus encapsulation features for continuous bus ratings of 3 kA and below. The most likely incentive for this accommodation is the demand for a smaller system footprint thus, reducing the available volume for larger through air and over surface clearances.

For grid service switchgear, bus encapsulation presents little or no problem. For alternate power service derived from multiple paralleled engine generators on a common bus, system bus ratings are frequently above 3 kA and reach 10 kA — maybe even higher current ratings in the future. There is a multitude of scientific and engineering considerations that determine the final bus design. To mention just two, consider skin effect and proximity. Skin effect is the propensity for current to flow in the outer perimeter of a bus bar in alternating current power systems. The tendency for increased skin resistance increases with the square of the current. Economics drive bus design towards the use of spaced flat bar lamination. This allows for air circulation providing convection cooling. Alternate power source systems should not depend on forced cooling. Failures in the cooling provisions will curtail operation of the alternate power source at its full rating.

Proximity is the result of through air spacing of parallel phase conductors of opposite polarity. Given two parallel bus conductors separated by some distance, as that distance is reduced, the magnetic fields of the currents — due to polarity attraction and repulsion — will cause more current to flow either on the inner or outer cross sections of the bus bars. This will negatively affect economic use of bus material in cross section.

Consequently, as the continuous rating of the bus increases, one can expect more heat generation for a given design geometry. Thus, one can expect a need for more efficient cooling of the bus. Bus encapsulation defeats this requirement. Larger spacing between bus burs of opposite polarity reduces the proximity effect and adds to the insulating air gap. Economic application of bus bar lamination and geometry reduce skin effect. Taken together, these considerations drive the bus design to larger through air and over surface clearances. Such designs mitigate the need for and disallow the use of bus encapsulation.

Open vs. closed
Should load transfer be open or closed transition? The load transfer strategy should be the most reliable strategy for the application. All alternate power systems should be tested and exercised under the load they will serve on loss of the preferred power source. The highest reliability transfer strategy is open transition. This strategy will show a blink in lights on test transfer and retransfer after an outage or test. For motor loads, in-phase transfer effectively mitigates the back emf phenomena for the open-transition transfer strategy. For transfer of transformer primaries between power sources, the magnetizing inrush can be avoided by closed-transition transfer only. Consider this. On loss of power, the transformer is transferred in the open-transition mode, thus the system protective functions must be set to ignore this phenomena. Where that is the case, then open transition is suitable.

Closed-transition transfer accommodates two system constraints. The first concerns restoration of the host facility to normal operation upon the return of the preferred power system to acceptable conditions. If closed-transition transfer is used, the critical loads are retransferred without any interruption. Since the critical load composition is typically less than half of the facility loss, restarting the non-critical loads can be time consuming. Closed-transition transfer of the critical loads leaves the facility staff free to restart the non-critical loads. As an added thought, high intensity discharge lighting having long restrike times is best served by closed-transition transfer strategies. The second constraint considers the rating of the critical load transfer switches in comparison to the rating of the engine generator(s). When a critical load segment comprises 25% or less of the alternate power source bus rating, then short duration (&100 ms) closed-transition transfer is a strategy that can make system testing practically transparent to the operation of the facility.

Alternate power source generators are typically driven by turbocharged high speed diesel engines (1800 rpm). These engines become more responsive when these turbochargers are spooled. That occurs above about 40% of the engine power rating. So, if the alternate power system is comprised of four loads, each equal to about 20% of the engine power rating, a staggered short duration closed-transition load transfer strategy may be appropriate.

Where a critical load segment is greater than 25% of the bus power rating at load transfer, soft load closed-transition transfer may be required to avoid unacceptable voltage and frequency transients on the alternate power source. The soft load closed-transition transfer strategy maintains the power sources in parallel for a longer period of time, although 20 to 30 seconds is usually adequate. During this period, the transfer switch control system synchronizes the alternate power source to the preferred source, parallels it and then increases the engine and generating loading using a ramping function. On retransfer to the preferred source, the ramp function slowly transits the load to the preferred source. One advantage of soft load closed-transition transfer is that it permits use of the load in peak demand reduction operating strategies for distributed generation applications since it already includes the appropriate protective relaying.

Should the transfer switch be three- or four-pole?
The possible permutations of power system configuration produces a number of different ways to treat the neutral conductor of AC systems. Confusion can creep in when trying to make the best design decision for any given power system that incorporates both a preferred and alternate power source.

The National Electrical Code, NFPA 70, provides specific requirements for the treatment of the neutral conductor and provisions for its grounding. The purpose of grounding the neutral conductor in a defined manner is to assure protection against inadvertent excitation of conductive surfaces of equipment, enclosures, cable conduits and raceways. The obvious motivation is concern for the safety of people and protection of property from electric shock and fire hazards.

In the early 1970’s, the NEC was revised to include the requirement for ground fault protection on certain power system configurations. Article 230-95 requires ground fault protection for solidly grounded wye electrical services of more than 150 volts to ground, but not exceeding 600 volts phase-to-phase for service disconnects rated 1000 amps or more. The most popular power system encountered in commercial, industrial and institutional facilities is 480/277 Y, three-phase four-wire. As a result of advances in the design of electrical lighting equipment, it became common place to accomplish bulk lighting in large buildings with 277 VAC fixtures. As this practice became widespread, the occurrence of electrical equipment burn down and fire began to increase. Upon investigation, it was found that at 277 VAC, the environment for arcing faults to ground was enhanced. Since arcing faults have impedance, quite frequently these faults would not be detected by the phase overcurrent protection until significant damage had occurred. It therefore became necessary to include provisions in the code that would provide for protection against arcing ground faults.

The expected current in an arcing fault is considerably less than that expected in a line-to-line or line-to-neutral fault. Consequently, phase overcurrent protection devices would take a considerable time to recognize and interrupt such faults. It was therefore necessary to develop a means of detecting and interrupting arcing faults. Several manufacturers responded by making various ground fault arcing current detection schemes available. These were dependent on the presence of a known return path for the arcing current to the power system neutral point. Consequently, deliberate treatment of neutral conductor grounding became even more important. If arcing ground fault current is to be reliably detected, the return path for that current must be known so that, when necessary, corrective actions can be taken.

The issue of grounding or not grounding a power source neutral impacts the cost of electrical distribution equipment. Where two or more power sources feed a load, and the neutrals of those sources are separately grounded, it may become necessary to switch the neutral conductor with the phase conductors when the load is transferred from one source to the other. This increases the cost of transfer switches. Consequently, as a cost control measure, the neutral should not be switched unless it absolutely must be. As an additional consideration, neutral conductor switching should be accomplished in a manner that assures that its switching contact does not interrupt current. Avoiding current interruption on this contact maintains assured low resistance in the neutral path.

Current in an arcing ground fault is limited by the voltage drop across the conductive path. The voltage across the arc is relatively constant. Thus, the current in the arcing path is (Esource — Earc )/Zpath. This arcing current can be detected by measuring current in the bonding jumper that connects the neutral conductor to the system ground. It can also be detected by summing the three phase and neutral currents at any point along the conductor path. In single- and three-phase currents, the algebraic sum on the instantaneous currents at any point in the path should be zero. When it is not, then a current is flowing outside the design path and is therefore objectionable. In reality, due to distributed capacitance between parallel conductors and ground (i.e. between the phase conductors and the wall of a conduit through which the conductors travel), there will always be some current flow in the ground return path. However, when that current becomes excessive, considerable damage can result if it remains uninterrupted. It is established, then, that a means to detect ground current is required and a means to differentiate between acceptable and objectionable currents is required.

The NEC fully defines required grounding practices including when power systems are required to be grounded and when they are not. Article 250-20(b) establishes when the power system shall be grounded. Article 250-20(d) states that separately derived systems shall be grounded. The fine print note goes on to state that when the neutral conductor of an alternate power source is solidly connected to the service supplied system, that alternate power source is not considered a separately derived system. What does this mean?

If separately derived sources meeting the required criteria of Article 250-20(b) include an alternate power source whose neutral conductor is solidly connected to that of the preferred source, then the alternate source neutral is considered to be grounded through the ground at the preferred source service disconnect. This means that there are times when the neutral of a generator power source will be grounded at the generator neutral and times when it won’t. Thus, there is environment for confusion over treatment of the generator neutral conductor.

Grounding the neutral
When should this neutral be grounded? The remainder of this discussion will attempt to offer some considerations in making that decision. There are some simple differentiation factors that indicate when it is unnecessary to separately ground a generator neutral. The first is if ground fault sensing is not required by the code. Generally, solid connection of the generator neutral to the preferred service neutral will preclude separately grounding the generator neutral. Any place where the power system profile does not come under the jurisdiction of NEC Article 250-20(b), connecting the generator source neutral to the preferred source service neutral serves to effectively ground it. Therefore, for 480/277Y three-phase, four-wire power systems rated less than 1,000 Amps (833 kVA), the generator neutral conductor can be directly connected to the preferred service neutral.

Additionally, the generator neutral conductor can be directly connected to the preferred service neutral for all 208/120Y three-phase, four-wire power systems. Where the facility service is 480/277 Y, three-phase, four-wire, and the generator is permanently installed, the neutral conductor can be treated in such a manner so as to preclude the need for neutral switching. In the following figure, such a circuit is shown. Note that this service is limited to 833 kVA maximum so as not to invoke Article 230-95 of the NEC. If ground fault protection is not required, the generator neutral can be solidly connected to the service neutral. The service neutral is grounded by the bonding jumper between the main service switchboard neutral and ground bus.

When the service falls under the auspices of NEC Article 230-95, the neutral should be grounded at each source and switched where ground fault detection coordination is required.

The circuit shown above represents a power system that is impacted by NEC Article 230-95. When the service rating exceeds 1,000 Amps (833 kVA), ground fault protection is required. The code requires that protection on the service disconnect. The presence of automatic transfer switches suggests that some of the load in this facility is of such importance that it justifies the installation of an alternate power source to avoid outages. That being the case, it may be advisable to expand the ground fault protection scheme to second level branch circuit protection as well. (NEC Art. 230-95 FPN No. 2)

Where ground fault protection is triggered by the code and an alternate power supply is included, switching of the neutral becomes necessary. The circuit shown above illustrates such a circumstance. The service is larger than 1,000 Amps. Consequently, ground fault protection is required at the main service disconnect at a minimum. If it is installed there and the generator neutral grounding was through a solid connection to the main service neutral, a ground fault when the generator is feeding the load will cause the main service disconnect to open. This will not disconnect the arcing fault from the generator. Coordination is thus lost. In the power system where the neutrals of the two sources are separately grounded, switching of the load neutral conductor to the source feeding the load is required. (NEC Art. 230-95 FPN No. 3) It can be seen that ground fault current will return only to the source from which it originates. This provides for coordination of the ground fault protection scheme.

In a power system where there are several preferred source services and multiple generators, ground fault protection can be readily coordinated. The multiple source circuit shown has a double-ended unit substation service from the preferred source and multiple generators paralleled onto a common bus. The transfer switches in this scheme have neutral transfer contacts where four-wire loads are involved. The ground fault protection scheme shown is only one of many alternatives.

The ground fault protection scheme begins with the preferred sources. It is noted that the neutrals of the two preferred services are connected to a common neutral bus. The bonding jumper connects this bus to the ground bus at only one location. There are two current relays monitoring that connection. The first one is set to trip the tie breaker 12 cycles after the fault starts. This provides for six cycle trip on the feeder ground fault sensors. The second current relay will trip one or both main breakers in 18 cycles if the fault is not cleared by either the tie or appropriate feeder breaker. The ground fault protection scheme on the alternate source provides selective coordination as well. However, depending on the nature of the affected load, the detection system may either trip or alarm the condition. Note that the generator neutrals are connected to the switchgear neutral bus. This neutral bus is bonded to the switchgear ground bus at one location with a current relay monitoring that connection. The ground fault sensors on the load and generator feeders would be set to six cycles. While the load feeder sensors may not trip their respective breakers, the generator feeder sensors will trip their respective breaker. Their operation detects arcing faults in the generator and its feeder up to the breaker at the switchgear bus. The current relay sensing the ground to neutral connection operates in 12 cycles and will most likely be alarm rather than trip.

It is not always necessary to separately ground the generator neutral conductor. However, when it is separately grounded, it may become necessary to switch a load neutral along with its phase conductors when transferring loads between power sources. This becomes necessary when ground fault protection is used. Ground fault protection is code mandated for 480/277Y 3q, four-wire services rated 1,000 Amps or more. For other power systems, it is left to the designer to decide. As a generality, if ground fault protection is not used, neutral conductor transfer is probably not required. However, where a branch circuit neutral conductor is transferred between sources, the switching means should assure that the neutral conductor switching contact does not interrupt current.

To sum up, open- vs. closed-transition transfer, three-pole vs. four- pole, insulated (encapsulated) bus? They are all driven by application requirements and sound engineering judgement.

Do you have experience and expertise with the topics mentioned in this content? You should consider contributing to our CFE Media editorial team and getting the recognition you and your company deserve. Click here to start this process.