The decision is based on whether it is a 3 phase 3 wire, 3 phase 4 wire, and whether you will provide ground fault detection at the generator. If it is 3 phase 4 wire and you provide GF protection/detection, then you will want a neutral to ground bond to be placed at the generator and provide a 4 pole xfer switch, making it a "separately derived source".
If not, then you can use a 3 pole switch and ensure that the neutral to ground bond does not exist at the generator.
If the generator neutral is grounded for any reason and you have ground fault protection at the normal source, you should typically use a 4-pole ATS. Otherwise the generator's ground will bypass some of the ground current from the normal source GF relay, reducing it's sensitivity.
There are ways around this with differential ground fault schemes.
I've got a similar problem with a transfer scheme with two 'utility' feeds and a standby genset. The utility feeds are from our own internal 11kV network and are solid-grounded to the site earthing grid. All three feeds are 4-pole ACB's equipped with a mid-range protection unit using CTs on all four conductors to provide a residual connection for ground fault detection, plus the usual functions for IDMT and instanaeous trip. Line-neutral loads are present, so the neutrals must be solidly connected to earth to meet British regs.
The genset has the ability to synch to either utility for a closed transfer, or to synch and run in parallel with one of the utility feeds to provide an on-load test facility. In this mode our ground fault protection is compromised because the neutrals of utility feeds and genset are both connected to the site earthing grid at points remote from each other, thus creating a low-impedance path through the earth grid in parallel with the neutral conductor.
I'm sure there is solution to this, but I'm scratching my head at the moment. I don't want to desensitise the earth fault protection because it seems a very crude solution and will screw up coordination elsewhere. Has anyone an off-the-shelf solution to this which they have used before?
TIA for any assistance.
----------------------------------
If we learn from our mistakes,
I'm getting a great education!
Assuming the generator feeds only one bus at a time (in parallel with the utility), I would think a 4-wire differential system on each util/generator pair of breakers would work. Sensing could be residual (easy on 4-pole breakers) or core-balance CT's.
Thanks for the suggestion. My problem is that I'm fairly sure that there is current through the parallel path, but I can't remove either of the neutral-earth bonds as this would leave me with a floating neutral in one or other state where the utility supplies the load or the genset supplies the load and the other is out of service.
British regs preclude any non-ganged switching device in the earth conductor (unless anyone knows of a specific exception?), so I'm between a rock and a hard place!
----------------------------------
If we learn from our mistakes,
I'm getting a great education!
I can think of a few possible solutions, but neither none may work in your case.
1. Disconnect one of the neutral grounds or don't connect the generator neutral when operating in parallel. I'm not up on British code but it sounds as if this may not be possible.
2. Add a small reactance in the generator neutral ground. If there is circulating current, it's a good bet that it's 3rd harmonic. A fairly small reactance may be enough to make the circulating current tolerable. It would have to be sized so as not to upset the protection scheme.
3. Remove the neutral grounds at the transformers and generators and put a grounding transformer on each bus.
1. Disconnecting the generator neutral when in parallel is what I'm considering at present if I can prove that it is acceptable to our regs. I need to be careful in case the generator G59 protection operates and disconnects the set from the utility, leaving me with a floating neutral with line-neutral loads connected.
2. Agree that circulating current is likely to be triplen harmonics plus any unbalance current. This may be possible.
3. British regs require solidly grounded neutral for LV systems
----------------------------------
If we learn from our mistakes,
I'm getting a great education!
