The 2002 edition of IEEE 1584 incident energy equations consider the effect of system grounding. The calculations for a solidly grounded system (i.e. wye connected system) result in a slightly lower calculated three phase incident energy than an Un-grounded or impedance grounded system

IEEE 1584 only considers the three phase case assuming either the arc flash originates with all three phases or will escalate into a three phase event. For un-grounded or impedance grounded systems, a line-to-ground arc flash can not occur so there is nothing to escalate.

Here is this week's question:

What type of grounding does your/client's system have?

Solidly Grounded
Un-grounded
Impedance grounded

Select all that apply

I was not aware the Hazard went up, but would also expect the Risk to go down; with resistance grounded systems. The single phase events would be limited to low amperage with low probability this escalates to 3 phase arcing fault, theory corroborated by field experience.

Are folks applying Risk factors to their calculated results that somehow reduce the level of clothing required, to match the table results?

I work in a facility with a solidly grounded system. Although I would not vouch for how perfect that grounding is. Prior to this, I worked in a facility with a high resistance grounding system.

Solidly grounded systems are not much fun. For those that are using phase overcurrent protection as "double duty" for ground faults, try this some time. Look at the trip curves and pick a reasonable trip time for a ground fault. Look at the required fault current from the time-current-curve for the overcurrent protection. Now calculate the required maximum resistance for the ground fault path using the line-to-ground voltage. The maximum allowable total circuit resistance is often scary how low it is.

Take for example a 100 HP motor with 124 A as FLA. Multiplying by 6 to get to starting current, we have 744 A as an absolute minimum trip setting, and chances are it is much higher than this. Taking 480/sqrt(3) to get the line-to-ground voltage of 277 V, 277/744 = 0.35 ohms.. This extremely small number barely insures that the overload relay will trip pretty quickly and a breaker/fuse is probably never going to trip. But achieving a circuit resistance of only 0.35 ohms is by no means easy.

If on the other hand you have a ground fault relay even in a solidly grounded system, it is going to react fairly quickly. NEC has changed over the years but basically it is only required above about 1000 A (1200 A frame breakers). I would suggest however that this cutoff is way too high based on simple calculations of how much current it takes to trip even such a large breaker.

What about these systems where a known issue exists (phase grounded) and the location will not shut down to perform ground fault locating and repair? I have worked at locations that let this go for 10 years! And that was with me consistently raising the issue. If you base calcs on resistive grounding please ensure the location maintains GF warning systems and subsequently performs troubleshooting. Just my observation.

Although not authoritative, many companies use Factory Mutual as an insurance carrier. FM Global is pretty clear on what is required for ungrounded or resistance grounded systems. It must either have an alarm or trip. If it is alarm only, then it must alarm in a continuously supervised area (e.g. control room). Otherwise, it needs to trip. If it does alarm, it needs to be handled and responded to exactly like a fire alarm. Otherwise, it needs to trip. In other words just like a fire alarm immediate action is taken to track down the source of the alarm and make repairs to eliminate it or fix the alarm if it is a false alarm.

The logic behind this is that if the ground fault escalates into a line-ground-line fault, it can and does start a fire through a process known as "burn down". FM also omre or less hints at or suggests converting and all ungrounded systems into resistance grounding systems.

In a resistance grounded system you have options. You can install ground fault relays which use time delays so that coordination becomes possible and practical. This allows for the efficient removal of the most common culprits (motor failures) off the system while preserving power everywhere else. With ungrounded systems it is not possible to narrow down the source most of the time except using the trial-and-error method (open disconnects and breakers until the ground fault disappears).

Interesting that you mention burn down because that same facility experienced just that issue 2 years ago. I left there about 6 years ago but a current EE told me about it. A 480/277 v lighting panel literally melted to the ground in the process. This indicating system had shown a phase A ground for years and was ignored...a prime example of your point and need of immediate response if a resistive grounding approach is used.

As per FM Global standard 5-10:

"2.3.3.1 Provide ground fault detection for all high resistance systems. In areas not frequently inspected by electrical personnel (once per shift), an alarm should be provided to a constantly attended location in the plant. Promptly locate and remove all ground faults that develop. The high resistance grounded system is similar to an ungrounded system in that it limits the available ground fault current and allows for an orderly shutdown if a ground fault occurs. Unlike the ungrounded system the high resistance grounded system prevents transient overvoltages which can cause potential failure of insulation and misoperation of micro processor based equipment."

The same wording is used in many of their standards. I just did a quick search and took the first one I found. They have a lot of these that are industry specific but in terms of system grounding, there is no variation in their standards. The standard for ungrounded systems is identical except that they also recommend converting them to resistance grounded systems.

At my plants we use resistance grounding from 480 to 23 kV. It is originally designed and intended for 1 kV to 10 kV. We use the coal mining "medium voltage standard" of 25 A resistors for 1 kV through 10 kV. Above 10 kV charging current gets so high that a higher current is necessary. The typical standard for this is 400 A but my predecessor designed it with 2000 A resistors. Below 1 kV the wire size gets so small that we drop down to 15 A to accomodate smaller (#14-16) wire. In newer installations we use tripping with time-based coordination. In the older ones there is still a lot of alarm-only equipment prevalent.

I have seen all three grounding types. I would in general agree that solidly grounded systems would result in lower IE since there higher current/faster trip times. Ungrounded could result in higher since everything is phase-phase.

Using impedance grounding could (probably does) result in higher calculated values BUT the incident rate for ground fault drops dramatically. There is no calculation to account for ground fault tripping due to impedance grounded systems.

There is a "calculation". Although 70E-2012 gave some vague wording in the arc flash hazard definition alluding to the idea that a risk assessment should be done, 70E-2015 edition clearly states that this is the case. A risk assessment has to consider both the severity of the hazard, and the likelihood that it occurs. For instance we do not require drivers to wear FR jump suits (except for race car drivers) as protection against the possibility of a fire during a car accident even though the hazard exists because the likelihood is not very large.

IEEE 493 contains failure mode data for both types of arcing faults. This provides a source of data for estimating the amount of line-to-line arcing faults vs. line-to-ground arcing faults. The nature of arc flash is that line-to-line faults and line-ground-line faults will rapidly convert to 3 phase arcing faults so other than an estimate of line-to-ground faults, the other types can be treated as a single case. Examination of the equipment damage cannot break it down further. Only waveform level measurements or high speed photography can break down the fault types that result in a 3-phase arcing fault any further because they all escalate into a 3-phase arcing fault. The IEEE 493 data only breaks the faults down into two categories.

SKM PTW determines a system is grounded based on a user defined parameter involving the fault current ratio SLG/3P calculated at each bus. I have been setting the parameter so that if SLG/3P >= 15%, then the system is grounded. Does 15% seem too conservative to you? Thank you.

Both solidly grounded systems and impedance-grounded systems have their place, depending on the application.

Coefficient of Grounding
I have seen voltage used as well to determine whether a system is "effectively grounded" You calculate what is known as the "Coefficient of Grounding"

EXAMPLE:
Take the case of an ungrounded 480 V. delta.

The normal phase-phase voltage is 480V between A-B, B-C, C-A.

Under a line to ground fault condition, the voltage on the faulted phase goes to zero and the voltage on the un faulted phase remains as full phase-phase voltage except it is now between an ungrounded phase and a grounded phase i.e. phase to ground connection.

Let's look at an example where the B phase that is faulted:

B Phase voltage = 0
A-B = 480V (but B is grounded so it is 480 Phase-Ground)
C-B = 480V (but B is grounded so it is 480 Phase-Ground)

The Coefficient of Grounding is used to determine how effectively grounded a system is. It is calculated by taking:

Coefficient of Grounding = Vline-ground / Vline-line

Where:

Vline-ground = voltage during a fault on unfaulted phases
Vline-line = Normal unfaulted line-line voltage

Effectively grounded Coefficient of grounding < 80%

Ungrounded / noneffectively grounded Coefficient of grounding > 80%

So using the example above, Vline-ground = 480 Volts (phase A to B and phase C to B where B is grounded)
Vline-line = 480 Volts

Coefficient of Grounding = 480 Volts/ 480 Volts = 1.0 - Not effectively grounded.

My understanding of effectively grounded is if the impedance to ground exceeds the impedance of the system shunt capacitance X sub C0 by at least a factor of 3. This is somewhat arbitrary but the whole idea is to provide a low enough impedance so that the system capacitance is effectively shorted out. Otherwise during arcing fault conditions in particular the system acts as a charge coupled circuit and the line to ground voltage is limited only by the insulation coordination...the motors get zapped by transients. As a local example, a GE resistor+ground fault monitor had a 2 A resistor but charging current was about 1.5 A. The resistor was small but it was not an effective ground because transients were not controlled. It is being upgraded to 25 A. Grqnted anytying above 4.5 A but less than the maximum current on the wire during fault conditions is acceptable but I just standardized all the resistors on 25 A for medium voltage based on charging current and Dalziel's formula for ventriculation.