Current Affairs: Closing the Loop on Automatic Transfer Switches
A reader took issue with my position on the use of three-pole switches on four-wire systems, stating: "Your article suggests that three-pole switches with solid neutral bars [should] never be used due to '...electromagnetic interference from currents flowing in the ground, and can interfere with the proper operation of ground fault protection systems...' Manufacturers have been selling three-pole automatic transfer switches without switched neutrals for 120/208 and 277/480-V, three-phase, four-wire systems for years without incident. Personally I've been selling them for over 20 years. Please provide me with your insight as to why we should discontinue using these switches in applications without ground fault protection."
I did not say that a three-pole ATS should never be used; it is perfectly acceptable for three-wire systems where there are no neutral loads. I also said it can be used with four-wire loads if provided with a solid neutral bus, but require that the neutral not be bonded to ground at the alternate source, which can present a safety hazard and should be carefully considered by the designer before being specified. There are two ways this can occur:
- The normal source neutral conductor's entire length from the ATS back to the source becomes part of the ground return path for fault currents supplied from the alternate source. This can increase the ground return impedance, limiting the fault current, and extending the time for a circuit breaker or fuse to operate.
- If the reason the normal source failed is physical damage to either the service or the normal source feeder to the ATS, the normal neutral path or the normal service neutral-to-ground bond may be broken, leaving the alternate source operating ungrounded.
If both sources are located in close proximity, the first item may be eliminated as a concern. However, the risk of the second item must still be weighed against the cost advantage of the three-pole switch. Also, because you are depending on the normal service equipment to establish the ground path for the alternate source, it is not possible to support the emergency system from the alternate source while doing maintenance on the normal service that could interrupt the neutral or bonding path.
Ground Fault CoordinationAnother reader wrote: "I have a lingering question about the use of four-pole closed transition switches. During the overlap period of the transfer, what impacts are seen with regard to the ground fault scheme? Given that the generator is a separately derived system in this configuration and is bonded accordingly and the service is bonded, two paths are created to the source. This also brings into question how the intent of the code is met that requires that the neutral be bonded at a point only on the supply side of the first disconnecting means. Is there an exception that allows this dual neutral bond condition even though it is for a limited amount of time?"
The neutral overlap that occurs with closed-transition transfer is a potential problem for the ground fault system when there is significant unbalanced load. The possibility exists that neutral currents from one source will be diverted at the transfer switch through the other source neutral-to-ground bond during the overlap period.
To prevent this from causing a ground fault trip, the ground fault function on any breakers supplying the ATS must have a time delay setting that exceeds the maximum parallel time of the ATS. I believe most switches use somewhere between 100 and 200 milliseconds as a limit for having both sources tied together, so a ground fault delay setting of 0.3 seconds will probably coordinate. However, this is something that should be specifically addressed when the protective device coordination study is performed for the system.
Obviously, if the parallel interval of an ATS were extended to allow a soft load/unload of the generator, or when switchgear with three-pole breakers is used instead of an ATS, you can't coordinate the ground fault function in this manner, and you must modify the ground fault system to recognize the dual-source arrangement in a manner similar to what is done for a double-ended unit substation. These systems can become very complicated, and are difficult to commission and maintain, so if I am designing such a system, I prefer to make it three-wire with no neutral. This requires providing additional isolation transformers to serve 277-V lighting.
As to the NEC issue, I interpret FPN No. 3 to Paragraph 230-95(c) to indicate that this method of transfer is acknowledged by NFPA and not prohibited. If it were prohibited, there would be no need for this note. ES