Arc flash hazard in outdoor 230kV substation

[ters]

What would be a practical approach for assessing actual arc flash hazard risk level when one is just walking inside a 230kV switchyard (N. American case)? It does not seem to me that empirical formulas in NFPA 70E and IEEE 1584 would be adequate to use since, given the typical fault level of say 40kA, being well over 10,000MVA, it will result in such HRC level of hundreds of kcal/cm2 for somebody waking just 4-5m below the bus (at his head level), which would make entering the substation prohibitive irrespective of the PPE level. It seems to be that NFPA 70E and IEEE 1584 formulas are accurate enough for panel type of indoor switchgear up to about 25kV, but applying the same rules for a 230kV facility with high SC level seems to suggests that even drive by on a public road passing by the sub can be classified extremely danger… Thank you.

 

[wbd]

This is a common mis-conception of arc flash hazards. There have been many discussions on this including a janitor walking by an MCC panel.

Unless you are actively interacting with the equipment (ie working on), AR clothing is not required. That being said many utilities will require anyone entering a substation to have on AR clothing. Note that NESC C2-2012 states that an assessment is performed to determine potential exposure to an arc for employees who work on or near energized equipment. This assessment considers the employees assigned tasks and work activities. So a worker entering a substation to read meters will have different PPE requirements than a worker entering to perform switching operations.

 

[dpc]

In the US, the National Electric Safety Code (NESC) covers electric safety for electric utilities including arc-flash protection. This standard has a much different method of assessing arc-flash hazards in outdoor substations. IEEE 1584 is not used.(BTW, the IEEE 1584 equations only cover up to 15 kV). Due to the relatively large bus spacings, arc-flash energy is based on an assumption of a line-to-ground fault only. Three-phase faults are not considered. Also, in high voltage substations, fault clearing is generally extremely fast, the working distances are greater and the fault is in open air. I've done the calculations and generally arc-flash hazards are much lower than you might expect. There is certainly a hazard, but orders of magnitude less than the numbers you state.

NFPA 70E specifically excludes electric utility facilities.

 

[ters]

Thanks you very much gents. Looks like there is significant confusion about arc flash out there and some more room for separating myths from the reality, although I’m obviously not an expect for arc flash studies and making comments based on common sense and general electrical knowledge only.

For this particular questions though, the NFPA 70E does actually mention that a different method should be used for open switchyards, called heat flux rate, but I overlooked that one before asking the question. And it refers to the same method you described. Actually, I don’t look into NFPA 70E directly, but rather to the Canadian version of it called CSA Z462, but I understand it is more or less copy/paste.

Speaking of other arc flash fuzziness, here is one example. The IEEE method seems to be based on the assumption that no enclosure provides any protection against arc exposures unless it is an arc resistant design. This is still a reasonable assumption for most enclosures which are subject to faults in the range of hundreds or at least tens of MVA, even though some enclosure do actually provide a degree of protection and may not necessarily explode at lower fault powers. But since that level of protection is not quantifiable in any way, and is different for each enclosure, looks like the assumption has to be it is zero for all non-arc resistant rated boxes.

However, what happens if inside an enclosure the fault level is very low, say less then 1MVA, but clearing time is infinite? The standard says that if clearing time is infinite, consider 2 seconds when calculating incident energy. 2 sec being enough time for somebody standing next to the panel to somehow realize that something is going on and move away from the line of fire.

And here is an example for the above case from real life. On a solar installation with a total installed DC capacity of about 700kW and with a bipolar collector having positive and negative arrays of 600V each (1200V total), and given non-linear characteristics of PV panels, we came up with what appear to be a reasonable estimate that in case of a short circuit on main DC bus (across 1200V), it may result in an arc of 900Amp. With the assumtion that arc voltage will still be high, say over 900V, that is fault power of < 1MW. But as the fault current is barely higher then nominal, the DC breakers which connect individual combiner boxes to the main DC bus will not trip for 2 sec (actually they will never trip, but only first 2 sec of fault durations is used for calculations).

So, as the clearing (or get away) time in this case is very long, the calculated incident energy is still high and requires high level of PPE.

But is a fault of 1MVA and 2 sec really equally danger and one occuring say at the same voltage but 20kA, being 20MVA, and lasting only 0.1 sec? The same amount of energy but released in 20 times shorter period.

The latter case is a very good candidate to instantly blow off the enclosure door and those standing around might be exposed. But in case of former, will the current of only 1kA produce such sudden pressure raise that the enclosure will also explode, just with a delay after 2 sec? Or it is much more likely that, since energy is being released rather slowly, that everything will burn inside the panel before the enclosure explodes or it deforms in such a way that door opens, which would probably take minutes, not mili seconds?

 

[Skythian]

But is a fault of 1MVA and 2 sec really equally danger and one occuring say at the same voltage but 20kA, being 20MVA, and lasting only 0.1 sec? The same amount of energy but released in 20 times shorter period.

According to IEEE 1584 and NFPA 70E line of reasoning, indeed there is no difference between the 1MVA 2 sec fault and the 20MVA fault lasting 0.1 sec only. The issue of using incident energy as a measure of arc flash danger without taking into consideration the rate the energy was delivered (aka "heat flux") has been addressed in "Evaluation of Onset to Second Degree Burn Energy in Arc Flash" paper by M.Furtak and L.Silecky (the IAEI magazine. July 2012). Also, a quote from "Heat Transfer in Biotechnology" paper by A.Stoll (Advances in Heat Transfer, v.4. Academic Press. 1967) summarizes the issue of using incident energy in IEEE 1584 and NFPA 70E standard. The quote reads:

"Serious misconceptions have crept into this field of research through adoption of rule-of-thumb terminology which has lost its identity as such and become accepted as fact. A glaring example of this process is the “critical thermal load.” This quantity is defined as the total energy delivered in any given exposure required to produce some given endpoint such as a blister. Mathematically it is the product of the flux and exposure time for a shaped pulse. Implicit in this treatment is the assumption that thermal injury is a function of dosage as in ionizing radiation, so that the process obeys the “law of reciprocity,” i.e., that equal injury is produced by equal doses. On the contrary, a very large amount of energy delivered over a greatly extended time produces no injury at all while the same “dose” delivered instantaneously may totally destroy the skin. Conversely, measurements of doses which produce the same damage over even a narrow range of intensities of radiation show that the “law of reciprocity” fails, for the doses are not equal."

 

[wbd]

For substations and electric utility applications here in the USA, ArcPro is the predominate software used to analyze arc flash hazards in substations and overhead wires. This is the software used to develop the table in NESC C2-2012

 

[racookpe1978]

At some pint, the real-world results and design configurations interfere with a theoretical analysis also.

In door, inside the high motor controller room in the mezzanine deck (below the generator floor): A row of switchgear at very high voltages packed in right next to each other, with only a 4 ft corridor between opposite rows of switchgear panels, all of which are at high (top) and high volt (bottom). All energized. Racking out one panel completely and irrevocably exposes the interior of the panel, and pulls the gears out from the contacts. A slow, laborious movement which is done by the worker immediately near the opening panel in a very confined space, right?

You've got to have full PPE because you can't escape the fireball, a fireball is not expected of course, not probable, but is certainly possible (and has happened in the past!), and will kill people in the immediate area with "splashing" debris and arcs.

Compare that with the outdoor switchyard: Surrounded by the barbed wire, separation measured in terms of yards (not inches) and the wires high overhead in the open. Access is limited to a "need-to-be-there" basis, is granted only by approval from the control room, is rare, and requires a 10 foot margin (minimum!) between the high voltages and the CPA. just walking through (inspections or measurements for example, as I have done) does not open contacts nor change configuration nor "start" currents. (Hitting a HV conductor with a metal tape measure is not recommended though.)

Driving a crane through a switchyard or making a lift for repairs inside a switchyard is a highly controlled, briefed evolution with safety observers at all points and the shift supervisors of the rigging company, the power plant, utility (power distribution company - in addition to the power plant itself) and contractors all present as escorts and safety monitors in front, behind, and beside the crane at all times.

 

[tinfoil]

This thread covered some interesting discussion, but strayed a bit and did not answer the OP's question, and I too need the answer:

How DO you calculate the Arc Flash exposure risk in an outdoor substation yard to meet Standards (for Ontario, Canada in my case) and more importantly truly protect your workers?

Clearly IEEE-1584 and CSA-Z462 to not apply to a 115kV open-bus switchyard, but I am also leery of the results of Arc-Pro, as they seem 'too high' for my subjective 'gut feel'.
What is the industry actually using?

 

[wbd]

How DO you calculate the Arc Flash exposure risk in an outdoor substation yard to meet Standards (for Ontario, Canada in my case) and more importantly truly protect your workers?

Clearly IEEE-1584 and CSA-Z462 to not apply to a 115kV open-bus switchyard, but I am also leery of the results of Arc-Pro, as they seem 'too high' for my subjective 'gut feel'.
What is the industry actually using?

Since OSHA released a new 1910.269 regulations which deals with utilities, Appendix E references which models they feel is acceptable to determine a reasonable arc flash incident energy level. Basically for up to 15kV IEEE 1584 methodology is used and above 15kV, ArcPro is used.

 

[stevenal]

"Racking out one panel completely and irrevocably exposes the interior of the panel, and pulls the gears out from the contacts. A slow, laborious movement which is done by the worker immediately near the opening panel in a very confined space, right?"

Right. Except remote racking systems have been available for over twenty years, and can be retrofitted as well. There is no reason to accept this type of hazard.

 

[dpc]

How DO you calculate the Arc Flash exposure risk in an outdoor substation yard to meet Standards (for Ontario, Canada in my case) and more importantly truly protect your workers? ...but I am also leery of the results of Arc-Pro, as they seem 'too high' for my subjective 'gut feel'.

ArcPro is the basis for the NESC arc-flash levels. So that is what is used by nearly all utilties in the US - in many states, that's a legal requirement. ArcPro was developed in Canada. It uses line-to-ground faults instead of three-phase faults (as used in IEEE 1584). So the incident energy calculated in ArcPro will be much lower than that calculated using IEEE 1584 or the Lee equations.

"Gut feels" can let you down. Carl Sagan was once asked what he thought in his "gut" about some topic. He said "I try really hard to not think with my gut".

 

[wbd]

ArcPro is the basis for the NESC arc-flash levels. So that is what is used by nearly all utilties in the US - in many states, that's a legal requirement. ArcPro was developed in Canada. It uses line-to-ground faults instead of three-phase faults (as used in IEEE 1584). So the incident energy calculated in ArcPro will be much lower than that calculated using IEEE 1584 or the Lee equations.

IEEE 1584 is valid for 3 phase in a box type faults up to 15kv. It can be used with other type of faults but will be conservative. ArcPro is for single phase open air faults for all voltages (as long as the voltage will sustain an arc) and for 3 phase open air and other faults above 15kV as long as the multipliers supplied by ArcPro are utilized.