Field Mounted Switch Enclosure and Grounding/Bonding

[Ground_Loop]

I have a project in a large industrial facility that requires several field installed Ethernet switch enclosures dedicated to my equipment.  The power source begins at a 480 3p distribution panel dedicated to my equipment, through a 480-208 transformer to a 208/120 panel board through to the load.  Proper grounding and bonding is maintained from the source to the load and its enclosure.  The only connections from our switch cabinets to the main server room are non-conductive fiber runs through air gapped conduit to cable tray.  All conduit and cable tray is properly bonded, but there is no conductive path between our switch cabinet and the server room, just single mode fiber without armor or shield.

Our client is insistent that we install a ground conductor from our switch cabinet to the server room central equipotential ground bus.  I am arguing that installing a conductive path where none currently exists has no technical merit.  Further, tying all my cabinets to the server room ground would connect all the power supply commons to the server room common potentially creating ground currents.

The client is quoting ANSI/TIA 607 which clearly implies the scope of equipment is that which is already co-located in the server room or has an existing electrically conductive path.

And finally to my question: Have any of you experienced a situation like this and is there justification to ground the switch enclosures to the server room central common?
This is literally a $1M question.

[bdunham7]

How is the server room connected relative to the 480V/3PH panel that you said was the beginning of the distribution chain powering your stuff?  If you trace back the grounding of your cabinets and the server room, do they go back to a common point or are they otherwise bonded now?  Or are they on completely different distribution systems?

If the former, obviously you are creating a....ahem.....Ground_Loop.  If the latter, I do have some experience with this and it's all bad, but I can't provide an expert advice on regulations or anything like that.  What I can suggest is that if you are compelled to install something like that, you look at how it is routed and whether you can use (or parallel) the existing grounds back to some point where they are closer together and just bridge or bond them there.  Just running a random ground wire between two currently unbonded, unrelated points seems just as idiotic to me as it apparently is to you.

[Ground_Loop]

How is the server room connected relative to the 480V/3PH panel that you said was the beginning of the distribution chain powering your stuff?  If you trace back the grounding of your cabinets and the server room, do they go back to a common point or are they otherwise bonded now?  Or are they on completely different distribution systems?

If the former, obviously you are creating a....ahem.....Ground_Loop.  If the latter, I do have some experience with this and it's all bad, but I can't provide an expert advice on regulations or anything like that.  What I can suggest is that if you are compelled to install something like that, you look at how it is routed and whether you can use (or parallel) the existing grounds back to some point where they are closer together and just bridge or bond them there.  Just running a random ground wire between two currently unbonded, unrelated points seems just as idiotic to me as it apparently is to you.

My 480 source is solely dedicated to my equipment.   The power sources for the server room and my system merge at the substation 4160.

[bdunham7]

Is the "server room central equipotential ground bus" itself bonded directly to the grounding system of its  own power service entrance?

It seems that situations like this come down to winning an argument.  And they effectively want to bridge the air-gaps you deliberately put in the conduit for the fiber.   

You may have already thought of this, but perhaps argue like this:  If the service panel for the servers and the one that you get your power from are fed from separate transformers that get their 4160V 3PH--but not ground-- from the substation, and the initial grounding point for each is separate and they are not bonded together, then you might propose simply adding whatever grounding is needed to complete the connection between your service panel and wherever the "server room central equipotential ground bus" point is.  If that seems unacceptable, then point out that the ground wire they are proposing will do the exact same thing.

Now in reality, those points may already be connected.  I don't know anything about this facility, but a lot of times connections end up being made whether they are intended or not...so maybe it would be interesting to test that.

[Brumby]

Firstly - I have no experience in this type of topography, but I can offer you a suggestion as to how you might argue the point....

What happens when something goes wrong?

The objective is to define fault scenarios and demonstrate the risks involved in what they have asked for against the risks in what you have delivered.

Anything goes here, including lightning strikes, somebody driving a truck through a wall, flooding or any other hazardous event.  Maybe look up an agency that reports on electrical incidents for ideas.  There will be scenarios where such a grounding conductor would become live - something with very obvious risk.


Just my thoughts.


bdunham7 seems to have some experience on the subject.   

[Ground_Loop]

Unfortunately, this is completely a letter-of-the-law compliance issue.  Technical arguments will not win at this level. I'll be headed up the chain for this one. Thanks for the support of my confirmation bias.

[bdunham7]

bdunham7 seems to have some experience on the subject.   

Unfortunately just enough to understand the difficulty of the problem, not to offer much of a solution.  My ears perk up when I hear the term 'equipotential' because it seems to acquire some mystical meaning and the actual practicalities are sometimes lost.  At some level everyone seems to understand that 'equipotential' grounds serve a different purpose than 'fault protection' grounds, but some people seem to think that just because you call it one or the other it magically acquires some metaphysical property that distinguishes it from the other.  First, "equipotential" is a local concept and trying to extend it over any distance runs into some serious issues.  Then there is the practical issue that even though some might imagine that the two are somehow separate, they always end up connected and not always in a way that prevents the possibility of fault currents in the nominally equipotential branch.  But I'm pretty sure Ground_Loop already knows all of that and is just...

Sometimes if you have level-headed, common-sense people having a good faith argument you can point out that the regulation isn't completely clear in some way (subject to interpretation) and that it would be really bad to interpret it in one of those ways.  Other times you run into folks that only seem to know one thing and they have that pretty much wrong as well, but they won't let go. 

[Ground_Loop]

Is the "server room central equipotential ground bus" itself bonded directly to the grounding system of its  own power service entrance?

It seems that situations like this come down to winning an argument.  And they effectively want to bridge the air-gaps you deliberately put in the conduit for the fiber.   

You may have already thought of this, but perhaps argue like this:  If the service panel for the servers and the one that you get your power from are fed from separate transformers that get their 4160V 3PH--but not ground-- from the substation, and the initial grounding point for each is separate and they are not bonded together, then you might propose simply adding whatever grounding is needed to complete the connection between your service panel and wherever the "server room central equipotential ground bus" point is.  If that seems unacceptable, then point out that the ground wire they are proposing will do the exact same thing.

Now in reality, those points may already be connected.  I don't know anything about this facility, but a lot of times connections end up being made whether they are intended or not...so maybe it would be interesting to test that.

Bridging the air gap aint gonna get it. They are insisting on me running a green #6 ground wire in conduit from my cabinet bus bar to their common ground point.  In some cases this is nearly 2000 feet.  Completely Efing insane.  And no I haven't done anything to piss them off.  Most of those involved share my opinion at least to some extent, but they also need to support their 'inspector'.  I'm sure many of you have run into intransigent inspectors. They don't have a grasp of the intent, but by God they know that code inside and out.  Unfortunately for me this is buried in second level doc references in my contract.  The most vexing part is that this is not a code issue, it's a local facility standard.

[bdunham7]

Bridging the air gap aint gonna get it. They are insisting on me running a green #6 ground wire in conduit from my cabinet bus bar to their common ground point.

I don't have the whole standard to look at, but you may find that they cite the phrase "is or may become" or something similar.  The remote panel may not be connected, but it may become so.

I'm sure many of you have run into intransigent inspectors. They don't have a grasp of the intent, but by God they know that code inside and out.  Unfortunately for me this is buried in second level doc references in my contract.  The most vexing part is that this is not a code issue, it's a local facility standard.

Yes, but in most cases the cost of compliance was small enough that it was the easiest way out.  Only once did a moron zoning inspector threaten to cause me some serious harm, and that only resulted in a lot paperwork and sleepless nights.  In your case it is worse--since it is their standard, they interpret it.  Unless you get lawyers involved, which I can't recommend.  You have my sympathy. 

[Ground_Loop]

Yea, no lawyers.  Been there.  I once spent more than 7 months arguing the order of precedence of conjunctions, stated as (1 and 2 or 3) interpreted as ((1 and 2) or 3) or (1 and (2 or 3)), with legal teams of both sides.

[Ground_Loop]

I am pleased to report that reasonableness under persistent pressure has prevailed. I have been informed that I can simply tie my enclosure grounds to local building steel.

[Doctorandus_P]

Since the other "wires" are fiber optics, I suggest to also lay down a fiber optic ground wire.

[Ground_Loop]

Since the other "wires" are fiber optics, I suggest to also lay down a fiber optic ground wire.

That was going to be one of our proposals, but cad welding was going to be to be problematic.

[f4eru]

Lol yeah, a fiber with no photons in it becomes a "grounded" fiber if run through 20m of compacted soil, every code inspector knows that. 

it's the relativistic effect of the absence of photons that bond it to the optical ground...