Leaving fences ungrounded is sometimes done at power plants where the fence is not close to equipment. Sometimes an ungrounded section is insulated from a grounded section at the power plant substation. IEEE Std 80 has a section that discusses the pros and cons of fence grounding.
I'm from the UK. Here we have Engineering Tech Std 41-24. Substation fences here are always grounded. The issue is whether they are connected to the substation main grounding system, or independently grounded. If they are more than 2m outside the main substation grounding system they are independently grounded using rods 3-4m long at each corner and at 50m spacings, plus rods at points where overhead conductors pass over. Gate posts are bonded below ground level. All equipment connected to the main grounding grid has to be at least 2m inside the fence.
Where the fence is encompassed by the main grounding system or within 2m, it is connected to the main grounding system. If the calculated touch potential exceeds the limit in the standard, we would install a bare grounding conductor 1m outside the fence at 0.5m depth.
If an independently grounded fence abuts a substation grounded fence then an insulated panel at least 2m long has to be put in. This can be metal, but is insulated at either end with at least a 5cm gap so it "floats" electrically.
Regards
Marmite
Thanks everyone for your responses on the fence grounding. We are using WinIGS to evaluate our grid design and at the fence the GPR drops off dramatically. (The fence is 3' inside the edge of the grounding grid.) We thought a way to eliminate the fence as a hazard during a ground fault event is to procure a plastic coated fence and have it ungrounded. This addresses the touch potential only and not the step potential. The step potentials seem to pass OK even at the GPR drop-off point. Any comments for you guys on the coated fence scenario?
If the surface potential (not GPR) drops off dramatically outside the edge, it's no problem as long as the edge is outside the reach of the fence. GPR is the voltage of the grid wire with respect to remote earth, not the surface potential.
See thread238-192760 for a discussion of coated fences.
He said it drops off rapidly at the fence, not the edge of the grid. This is a problem if the difference between the GPR at the feet and that at the hands on the fence is different enough.
Coatings can become damaged, and NESC requires grounding of any metallic fence. Perhaps there is a non-metallic fence that would suit.
In response to the fence OK if a few feet inside the grid, you logic is there and thanks to the others that addressed the fence/edge GPR drop-off concern. There are some sides of our older substations where the fence is built right on the property line. We cannot go outside by 3' with a horizontal copper conductor to prevent the GPR from being dangerous at the fence since that is not our property. We can on some cases move the fence 3' back away from the property line and then add the horizontal grid conductor, but that is only some cases. Anyway, using a plastic coated fence although there is a concern that the coating may go away with time, may be our best option. Using a non-metallic fence doesn't appeal to our company. If there are other comments, please do so. We welcome them!
