The NEC Article 230.95 states that ground fault protection is required for solidly grounded wye electrical service rated 1000 Amp or more. This is very understandable and we provide protection device with ground fault protection (Ground Fault Pickup & Ground Fault Delay) for any service that equal or exceeds 1000 Amp. But recently, many consultant colleagues are pushing for ground fault protection for services as low as 400 Amps. Their reason was "It is not the matter of code, it is a safety for Arc Flash." Well, the bible for arc flash safety is NFPA 70E & IEEE 1584 and I haven't read such statement there. In arc flash study, we do not have anything like there is a high incident energy (ei) & we mitigate with ground fault protection. Regardless of the setting of the ground fault protection, the incident energy will remain same. I too do not know why we do not see the effect of ground fault setting on incident energy in arc flash study reports and I would appreciate if you can shed some light on that. What I do know is, theoretically ground fault protection may minimize incident energy as its TCC is roughly in L-shape and could pickup the fault current before the fault current hits the breaker short time pickup, short time delay, long time pickup & long time delay. But this is theoretical and may only satiate the consultant engineer, whereas the customer has to not only pay extra for ground fault protection (below 1000 Amp service), but also have to wear the same PPE as without ground fault protection because mitigation if any by ground fault protection does not show in study report. Please share your thoughts on this.

I have had concerns that the Arc Flash Analysis does not take into account ground fault protection. Everything I read assumes the ground fault will develop into a three phase fault and it is that incident energy that must be protected against. However, my customer is installing 10 amp grounding systems on all 480V systems. With initial ground fault current this low I have a problem believing the plasma ball created by such a fault will develop into a phase-to-phase or three phase fault very rapidly, if at all. I realize the spacing on 480V is pretty close, but I would think the arc resistance of a 10 amp arc would be quite high, and would likely be more like a 120V contact.

PaulEngr, if you are reading this please weigh in on the subject. You seem to be the most knowledgeable source in the forum.

The probability of a three phase arc flash would be reduced because most arc flash incidents begin with a single phase to ground and then erupt into a three phase arc flash. I read that somewhere. . .

Great questions.

I think there are several issues here. Using the electronic trip breaker example that was given, usually the minimum ground fault delay is 0.1 seconds / 6 cycles (60Hz.) If a phase-to-ground arc flash occurred below the instantaneous and the hope was the GF function would trip, would it? It might but I believe the problem is more about uncertainty and the fact that the language in IEEE 1584 is always about three phase.

However, the 2011 edition of the NEC addressed this issue differently with Article 240.87. The requirement began with electronic trip breakers that did not have an instantaneous - similar concern as the one raised here. An electronic trip breaker can have a 30 cycle short time delay if no IP exists. The requirement was to have some form of high speed tripping such as a maintenance switch, zone selective interlocking, differential protection etc. to protect a worker from an arc flash with the breaker having potentially long time delay/arc duration.

The 2014 Edition of the NEC 240.87 made a significant change again, to address the concern of an arc flash with an arcing current below the instantaneous trip.

Arc Energy Reduction. Where the highest continuous current trip setting for which the actual overcurrent device installed in a circuit breaker is rated or can be adjusted is 1200 A - one of the following or approved equal shall be provided: Zone selective interlocking, differential relaying, Energy reducing maintenance switch, energy reducing arc flash mitigation system, approved equal means.

This requirement is to address the area of concern here, an arc flash with an arcing current below the IP of larger devices where a longer clearing time can exist.

Regarding the Grounding Resistor issue, the resister greatly reduces the likelihood of a phase-to-ground arc flash from ever occurring. In this case, there isn't anything to escalate. However, there is still an arc flash issue - if the arc flash begins as phase-to-phase or three phase.

I look forward to the view of others too!

I'd agree about the statement of lowering the limit to 400 A, for two reasons.

First, do the math. Let's say we have a 2.77 ohm resistance in the ground return connections. This is pretty easily typical with a few 10's of feet of conduit and "typical" ground conductors going through an MCC, several connections, etc. It is pretty typical in my experience especially on larger (50+) HP motors that a total resistance of 2-3 ohms is not unusual at all unless you have a very "small" system consisting of a transformer directly feeding a combination disconnect and a motor within a few feet from there. The other major reason for choosing this number is you can do the math in your head.

So on a 480 V system, the solidly grounded connection is 277 V to ground. That gives us 100 A of total ground fault current. This won't cause instantaneous or near instantaneous tripping until we get down to 10-15 A breakers considering standard off the shelf non-adjustable thermal magnetic circuit breakers are set to 6-10 times their long term rating, and except for very small motors (under 10 HP), the breaker never trips and it is up to the overload relay to trip (eventually....). As a general rule even if one follows Code to the letter, the breakers won't trip at all or at least take an excessively long period of time to trip on ground fault with motor wiring starting at around 100 HP and it is questionable down to around 50 HP.

In the mean time, the equipment is energized and there is a shock potential hazard of considerable proportions present. Considering that shock is responsible for twice the number of arc flash injuries (as per ESFI results and Cawley), going with the "cheapest" grounding system (solidly grounded) without ground fault relaying opens up a very real safety concern. This is over and above the arc flash concern which is drastically limited with resistance grounding since on a resistance grounded system (with tripping), ground fault-initiated arc flash no longer occurs, eliminating between 70 and 98% of all arc flash incidents, depending on whose statistics you want to use. The shock hazard extends all the way to 4160 V for a 25 A resistor for indefinite periods of time, and long enough for high resistance grounding with tripping that the benefits extend to 7200 V systems as well.

The second advantage that arises is that the fault current in a high resistance grounded system is less than the normal operating current even on the smallest wiring in the whole system so none of the unfaulted equipment will be exposed to excessively high currents. The result is no damage to unfaulted equipment. This benefit does not extend to phase-to-phase faults which still generate enormous bolted fault currents and cause a lot of damage, but for the vast majority of incidents, damage is very minimal both at low and medium voltage.

Finally, high resistance grounding installations can be less expensive than even solidly grounded systems. This is true for the simple reason that the size of the grounding conductor can be almost arbitrarily small. It is limited essentially by the "bare minimum" limits in Code such as use of a #8 AWG cable going to the ground rod. For that matter, trying to achieve crazy low (<1 ohm) Earth ground resistance is no longer necessary in any application. Although Earth grounding is no longer "critical", I still advocate getting it down reasonable to say under 10 ohms for grounding purposes other than system grounding where system grounding shares a ground rod with other grounding systems.

On a 480 V system the ground resistors can typically run less than $2K depending on amperage. I recommend 15 A but others want to go lower. Generally 480 V systems don't get over 5 A of system capacitive charging so 300% of that (to ensure it doesn't start acting like an ungrounded system with large transients, etc.) is 15 A. At 2400-10 kV, I've been using 25 A resistors which run around $4K for 2400 V and around $10K for 7200 V. Above this point the charging current becomes so high that you need to go to a low resistance grounding system using either a 100 A or 400 A resistor. A continuous duty resistor is no longer practical. At 23 kV a recent quote for a 400 A resistor was around $25K.

I do not recommend using the "pulser" things. They don't work all that well in reality, and leave electricians fumbling around trying to find the fault with the new fancy tool instead of relying on their tried-and-true skills. I also don't recommend buying a prepackaged system from the big names (iGard or Post Glover). Their prepackaged systems are crazy expensive for what you get. All you need is the resistor in a box and sensitive ground fault relaying for your breakers. Allen Bradley resells and most importantly, STOCKS ABB molded case breakers that fit the old Cutler Hammer Series C footprint and are available with relatively inexpensive ground fault modules. Since ABB is poor about maintaining inventory (at least in the U.S.) for 480 V systems this offers a great way to go.

I also don't recommend going with non-tripping systems. FM Global which carries a lot of weight for a lot of companies also does not recommend alarm-only. If you go alarm only, it needs 24/7 monitoring and needs to be treated like a fire alarm. It requires an ORDERLY shutdown, but still a shut down, to find the fault and remove it. Most companies think that the alarm going off at the substation for WEEKS is adequate.

Typically you see around 1-3 A of charging current on 480 V. The highest I've seen is 5 A. If your resistor is too small (resistance is too high), then what happens is that charging current is no longer a minor issue and the system begins to take on the characteristics of an ungrounded system...basically you get very high transients during an arcing fault, typically an unlimited one. That's unhealthy for your insulation so you will see an increase in failures rather than a decrease. The rule of thumb is to set the resistor to at least 3 times the charging current. So at 5 A for 480 V, this means the resistor should be 15 A. That's a good general purpose number to use. Larger has no positive benefits and neither does smaller. Set tripping generally to about 60% of this or 10 A although in practice I use a very mild curve with a 5 A tap and vary the time dial to give grading (time-based coordination), setting the time dial so that the 10 A trip is at the desired setting. I also prefer to install a transformer or resistive divider across the power resistor set to trip at 60% of full voltage and a very long time delay (out around 10 seconds). This is backup protection in case the resistor itself fails as well as backup protection in case the primary (current) relays fail.

And no, not even at 4160 V phase-to-ground with a 25 A resistor (7200 V phase-to-phase) have I seen any evidence of a seriou arcing fault. In fact about 6 weeks ago someone I know very well more or less "tested" a 7200 V enclosure by accidentally dropping a ground cluster into it, realized it was falling and attempted to grab it. The rest of the details are fuzzy but suffice to say that he received relatively minor burns. I've seen some pretty extraordinary things but no evidence that if a high resistance ground is set up correctly that there's enough energy there to get to a self-sustaining arcing condition on a phase-to-ground fault. On phase-to-phase of course nothing changes and we still get a full on arcing fault which rapidly turns into a 3 phase arcing fault. An interesting effect though that was reported in some recent testing on 480 V MCC's is that it appears that when the center phase crosses zero current and extinguishes the A-B and B-C arcs, the A and C phases will initiate arcs with the cabinet (L-G-L fault) until such time as the B phase voltage is high enough to restrike the arc. This multi-arc behavior is kind of interesting, but it should be expected since from simple geometry otherwise we'd see a big difference in arcing fault currents where the middle and outer phases behave differently. It is also interesting that the same study compared what happened in a high resistance grounded system vs. solidly grounded. The arcing pattern was the same...it seems that at least for sustained 3 phase arcing faults, having a grounded path is not necessary and does not affect the arcing in any appreciable way, although clearly from the equations IEEE 1584 predicts that ungrounded and high resistance grounded systems have slightly elevated incident energy during a 3 phase arcing fault.

From what I've read, it seems solidly grounded systems have their advantages, but also have some serious disadvantages. Why then, do we still (mostly) install and use solidly grounded systems if low resistance grounded systems can reduce the possibility of arc flash?

Not LRG, but HRG for low voltage systems. I think the main reason is that extra training is needed, e.g. fault location. Not good to allow a fault to persist or to go unrepaired. Also, can't have L-N loads.

Mike