Arc Flash Certification
NFPA 70E & OSHA Compliance Consulting
Direct line
(813) 725-0999
Office hours
Mon–Fri 8am–6pm ET
Audits Programs Procedures DC Systems Training Compliance About Contact
Nationwide — onsite & live virtual
Service — DC Electrical Safety
The 2027 edition of NFPA 70E added an article on direct current. Your arc flash study almost certainly stops at AC.
We assess your dc systems — solar, battery storage, EV charging and data centre dc plant — against the 2027 thresholds, establish whether the study you already hold covers them, and write the procedures and qualified-person training that follow from the answer.
Request a dc hazard assessment
or call us direct on (813) 725-0999
The change, at a glance
New article
310 — Direct Current (dc) Electrical Hazards
Status
Not present in the previous edition
Travels with
360 Batteries · 380 Photovoltaic Systems
Systems in scope
Solar PV · BESS & UPS · data centre dc · EV charging
Commercial terms
Fixed scope, quoted up front
Article 310
Three numbers decide whether your dc equipment is in scope
Article 310 sets out dc hazard thresholds and applies them to the systems covered by Chapter 3. Where any one of the three is exceeded, risk control has to be selected and applied from the hierarchy of risk control. Values below are cited as facts; no standard text is reproduced.
Hazard
2027 threshold
What it means on your site
Contact thermal
1000 W
power at or above
Power is the trigger, not voltage. A modest-voltage dc source with enough available current behind it puts a contact thermal hazard on a terminal no ac rule of thumb would flag. Battery terminal and dc bus work are where this lands first.
Electric shock
100 V dc
& 40 mA
at or above 100 V dc, with current greater than 40 mA
Both conditions have to be present, so available current matters as much as nameplate voltage — and somebody has to have worked it out for the equipment in question. The long-running 100 V versus 50 V argument in the dc literature is settled for this edition.
Arc flash
150 V dc
& 1.2 cal/cm²
above 150 V dc, with incident energy above 1.2 cal/cm²
Above 150 V dc the arc flash question turns entirely on incident energy — which is exactly the number no consensus method calculates for dc. That is the subject of the next section, and it is why a facility can be compliant on paper and uncovered in fact.
Chapter 3, 2027 edition
300
Introduction
310
Direct Current (dc) Electrical Hazards
New in this edition
320
Capacitor Electrical Hazards
330
Electrical Hazards 1 Hz to 110 MHz
340
Mixed Frequencies (Reserved)
350
Electrolytic Cells
360
Batteries
370
Electrical Double Layer Capacitors
380
Photovoltaic Systems
390
R&D Laboratories
Chapter 3 has been renumbered. Any program, procedure or training deck citing prior-edition Chapter 3 article numbers now points at the wrong articles — a concrete, checkable audit finding.
Which systems in a real facility cross those thresholds
PV strings and combiners
String open-circuit voltage is on the array drawings, not inferable from the ac side. Read it, compare it with 150 V dc, and remember the array generates whenever there is light on it — and keeps generating after grid power to the area is shut off (National Wildfire Coordinating Group, Solar Farm Safety).
Battery strings in a UPS or BESS
String voltage and available fault current live on the battery system’s own documentation, not on the switchgear study. There is no isolation point, so the question is never whether it is on. It is always on, day and night.
dc bus in a data centre
dc distribution and the battery plant behind the UPS routinely fall outside the study commissioned for the ac switchgear. If the dc side is not in the equipment schedule it was not assessed — and the people working on it weekly are working to an ac program.
EV fast charging
A fast-charging cabinet holds a dc output stage and, in some designs, an internal battery buffer. Technical Safety BC investigated an arc flash that injured a vehicle owner at a fast charger in Hope, British Columbia on 16 August 2024; its final finding, reported 22 October 2025, attributed it to a short circuit inside the charger’s own battery module.
The survey gap
The standard method most arc flash studies are built on does not cover dc at all.
IEEE 1584-2018, the guide almost every incident energy analysis in the United States is performed to, covers three-phase ac systems from 208 V to 15 kV. It explicitly excludes single-phase ac, direct current systems, coordination studies and PPE recommendations.
Source: IEEE Standards Association, IEEE 1584 scope statement; standard published 30 November 2018.
There is no dc equivalent
NREL states the gap plainly: because PV has proliferated so fast and there is no formal calculation guideline equivalent to IEEE 1584 for dc, practitioners have had to rely on a scatter of equations and models published by different researchers.
NREL, Methods for Evaluating DC Arc Incident Energy in PV Systems, NREL/CP-5K00-78331, August 2021.
The models disagree with measurement
EPRI tested published dc arc-flash models against real arcs on real PV equipment. Two industry “maximum power” methods predicted more than five times the incident energies actually measured. A third was reasonable at a half-inch electrode gap and over-predicted by a factor of two at two inches.
EPRI, DC Arc Flash on Photovoltaic Equipment, Technical Report 3002014124, June 2018.
And it cuts both ways
In the same programme every incident energy measured at 18 in. from the electrodes came in below 3.6 cal/cm², with a maximum adjusted value of 5.2 cal/cm² on a combiner box. Over-conservative dc numbers drive over-specified PPE and unnecessary work restrictions. Under-scoped ones leave people uncovered. Neither is a safe default.
EPRI 3002014124, June 2018. The NFPA Fire Protection Research Foundation currently has an open project to develop a dc arc flash hazard model for NFPA 70E.
The operational consequence, stated plainly
A facility can hold a current, professionally executed arc flash study, produced by a competent engineering firm, and have no dc coverage whatsoever — and nothing in the study will say so. The exclusion lives in the method, not in the report. Nobody misled anybody. The gap is simply invisible unless you go looking for it.
That is the survey gap. It is also, in our experience of reading these reports, the single most common finding on a site that has added solar, storage or charging since the study was commissioned.
Run this against your own study
Six checks
1 · Find the methodology statement.
If it names IEEE 1584-2018, that method covers three-phase ac only. Everything dc on your site sits outside it by definition.
2 · Read the equipment schedule and count the dc items.
Combiners, the dc side of inverters, battery strings, UPS dc buses, dc distribution, charger cabinets. If they are not listed, they were not studied.
3 · Read the exclusions and limitations section.
A report that is silent on dc is not a report that found no dc hazard. Silence and clearance look identical on the page and are not the same thing.
4 · If dc was studied, find out which model was used.
And whether the report justifies applying it to that equipment. A maximum-power method used without comment deserves a question, given EPRI’s measured over-prediction of more than five times (EPRI 3002014124, June 2018).
5 · Compare the study date with your dc installation dates.
Rooftop solar, a storage container or a bank of chargers added after the study was issued is, by definition, not in it.
6 · Walk the dc equipment and look at the labels.
Then ask who on your qualified-person roster is qualified for that equipment, and which record proves it. That answer is usually the fastest route to the whole problem.
Article 380 — Photovoltaic Systems
Solar now carries its own training requirement. A general 70E class does not evidence it.
Article 380 has its own scope, its own exposure levels — the same three dc thresholds as Article 310 — and, unusually, its own article-level electrical safety training requirement.
What Article 380 contains
380.1
Scope. Electrical safety for employees working on PV arrays that present an electrical hazard. An informational note directs you to other Chapter 3 articles for other PV components such as combiner boxes and inverters — so a PV procedure written only around “the panels” is incomplete on its face.
380.2
Exposure levels. The same three thresholds: 1000 W contact thermal; 100 V dc with 40 mA shock; 150 V dc with 1.2 cal/cm² arc flash.
380.3
Electrical safety training. Training requirements are applied to employees exposed to electrical hazards working on or near PV arrays, interconnected PV equipment and collection systems — with a defined syllabus for qualified persons, set out opposite.
380.4–.6
Qualified person, PV risk assessment, specific work procedures. Three further requirements, each of which produces a document somebody has to be able to hand an auditor.
What qualified-person PV training has to cover
The operating characteristics of the PV modules and equipment in front of them
Recognising the continuously energised nature of PV panels
Performing a risk assessment for energised work
Developing and then actually following procedures that minimise exposure
Using control measures, including personal protective equipment
Why the second one matters
It is the only topic on the list that has no ac analogue at all. Everything in a 70E class assumes something can be turned off.
What this does to your records
A general NFPA 70E class does not evidence PV qualification, and an attendance sheet from one will not survive the question. Three separate documents now have to say something they probably do not say today: the training record needs line items that map to the topics above; the qualified-person roster needs to name who is qualified for PV work specifically, rather than qualified in general; and there have to be written PV work procedures and a PV risk assessment to point at.
If you use contractors for array work, the same evidence is now something you should be asking them to produce before they go on the roof — and something a host employer will be expected to have verified.
Supporting the point
OSHA’s solar guidance advises employers to cover the panels while installing or servicing them, in addition to protecting workers from the circuits — an acknowledgement that the usual isolation step does not exist.
OSHA, Green Job Hazards — Solar Energy: Lockout/Tagout.
NREL reports that in testing, module short-circuit current under artificial light sources ranged from 1.5 mA to 212 mA, and notes that tens of milliamps can kill a person — which is the whole argument for the 40 mA figure in the shock threshold.
NREL, Solar Photovoltaic DC Systems: Basics and Safety, NREL/CP-5B00-68696, March 2018.
Article 360 — Batteries
A battery gets a lower risk work condition, not an electrically safe work condition
The wording is the whole point. Article 360 covers scope, safety procedures, and establishing a lower risk work condition in batteries — because a battery cannot be de-energised.
Everything in a conventional energy control program assumes an isolation point and a verified zero-energy state. OSHA’s 29 CFR 1910.147 is built on it; so is every lockout procedure written against it. On a stationary battery string neither exists. Waiting for the energy to run down is not a control measure — on a large system it is not even a plausible one.
So the LOTO logic does not transfer, and procedures have to be written specifically for battery work. That means sectionalising the string, insulated tooling and single-point-of-contact rules, chemical and thermal hazards carried alongside shock and arc flash, and a defined “lower risk” end state that a named person signs. Sandia National Laboratories’ published work on energised battery work describes exactly this approach — reducing risk by sectionalising batteries into lower-voltage, lower-energy strings.
D. M. Rosewater, Sandia National Laboratories, “Reducing Risk When Performing Energized Work on Batteries,” IEEE Transactions on Industry Applications, Vol. 60, No. 2, March–April 2024.
An honest note on battery arc flash
The standards community does not agree on whether a stationary battery string can sustain an arc. A 2013 committee presentation to the NFPA Conference & Expo by B. Gray (HBI) and S. McCluer (Schneider Electric) argued that in most cases it cannot, because a battery’s stored energy depletes rapidly and the literature shows millisecond sparks rather than sustained blasts. Presenting battery arc flash as uniformly catastrophic would be indefensible. Presenting it as trivial would be equally so. The procedure has to be built on your own string configuration, not on a slogan.
Documented incident
McMicken Energy Storage, Surprise, Arizona — 19 April 2019
APS McMicken facility, a 2 MW / 2.16 MWh lithium-ion (NMC) battery energy storage system in service since March 2017.
16:55
Smoke detector activation. Clean agent suppressant deployed.
18:28
Hazmat team on scene, detecting hazardous levels of hydrogen cyanide and carbon monoxide.
20:01
After developing an incident action plan, the team opens the enclosure door.
20:04
Deflagration. Four career firefighters seriously injured.
Sequence and injury count from the line-of-duty-injury report by the Fire Safety Research Institute (UL Research Institutes), which documents four seriously injured career firefighters. Contemporaneous trade reporting stated that eight firefighters and one police officer were transported to hospital. Both figures are on the public record; the difference between “seriously injured” and “transported for evaluation” is not resolved in the published material, so we cite the basis rather than pick a number.
The root cause finding — and the dispute
APS commissioned DNV GL, whose final report of 27 July 2020 determined the root cause to be failure of a single cell on one rack from abnormal lithium metal deposition and dendritic growth, cascading into thermal runaway across the rack. Contributing factors identified included the absence of thermal barriers to limit cell-to-cell and module-to-module cascading, a suppression system unable to stop thermal runaway, accumulation of flammable off-gas in a sealed enclosure, and inadequate emergency response planning and coordination.
That finding was contested. LG Chem, through its investigator Exponent, rejected the internal-cell-failure conclusion — arguing that metallic lithium plating did not cause an internal cell failure and that voltage profiles from testing did not match the incident data — and proposed instead an external cause: intense heating from electrical arcing on the same rack. The disagreement has not been publicly resolved.
DNV GL findings and the LG Chem response as reported by IEEE Spectrum and Utility Dive, 29 July 2020, and by APS. FSRI notes the facility was commissioned before the current consensus standard for stationary energy storage installations existed, and complied with the codes in force at the time.
What a dc practice should take from this: the mechanism that injured people was not shock and not arc flash. It was a flammable-gas deflagration released when a door was opened on an enclosure that could not be de-energised, could not be safely ventilated, and had no site-specific response plan. dc consulting that stops at incident energy and labels misses the thing that actually hurt people.
The dc service lines
What we do about it
Five pieces of work, each producing something written that a manager, an insurer or an inspector can read. Fixed scope, quoted up front.
01
dc hazard assessment against the 2027 thresholds
We walk the dc plant, establish where 1000 W contact thermal, 100 V dc with 40 mA, and 150 V dc with 1.2 cal/cm² are crossed, and record which systems are consequently in scope of the hierarchy of risk control. The output is a written assessment naming equipment, not a colour-coded summary.
02
Review of an existing arc flash study for dc coverage
We run the six checks above against your own report and hand back a written statement of what is covered, what is not, and what to require if a dc study is commissioned — in language your engineering firm and your insurer can both act on.
03
dc-specific written procedures
Program language, energised work permits and risk assessment forms rewritten for a source with no isolation point — including the parts of your existing program that silently assume one, which is usually most of them.
04
PV and battery work procedures
Array, combiner and inverter dc-side work on the PV side. On the battery side: sectionalising, entry, gas detection and response procedures written around a lower risk work condition rather than a zero-energy state that cannot be achieved — and coordinated with whoever responds to an alarm at your site at eight in the evening.
05
Qualified-person training for dc systems
Training built to the PV article’s topic list and to battery work, delivered onsite or live virtual, with records that name the equipment and evidence the qualification rather than logging attendance. We run 55+ classes a year and have trained 500+ electrical workers in the past six months, at an average student rating of 9.6 out of 10.
The scope boundary, stated plainly
We do not perform incident energy analysis engineering studies and we do not produce arc flash labels. That is true of our ac work and it is true here. On dc it is arguably a feature rather than a limitation: no consensus calculation method exists, so the defensible work is establishing what is and is not covered, writing the procedures, and qualifying the people who do the work.
Where a dc study is genuinely needed, we tell you what to ask an engineering firm for and what to require in the deliverable — including that the model used be named and its applicability justified. We take no part of that fee.
Context
Why the exposed workforce is growing
Every figure below is quoted from the named publisher with its date. We do not repeat industry statistics whose origin cannot be traced — including the widely quoted daily arc flash incident figure, which NFPA’s own research foundation says has unclear origins.
Solar
43.4 GW
of utility-scale solar planned for 2026 — 51% of all planned US capacity additions.
EIA, Preliminary Monthly Electric Generator Inventory, 20 February 2026.
262 GWdc
installed nationwide; solar’s share of US electricity generation is now over 8%.
SEIA, Solar and Storage Industry Research Data, updated 1 June 2026.
280,119
solar energy jobs in the United States, 178,713 of them in installation and project development.
IREC, National Solar Jobs Census, published November 2025 (2024 data).
Battery storage
24 GW
of battery storage planned for 2026 — 28% of all planned US capacity additions.
EIA, Preliminary Monthly Electric Generator Inventory, 20 February 2026.
70% a year
average annual growth in operating utility-scale battery storage over three years, reaching nearly 52 GW nameplate after the first half of 2026.
EIA, Battery storage capacity averaged 70% growth over the last three years, 7 August 2026.
57.6 GWh
of new energy storage installed in 2025 — the largest single year of new battery capacity additions on record.
SEIA, 2025 year-in-review release.
EV charging
75,011
dc fast charging ports in the United States, among 256,487 total charging ports at 82,091 station locations.
US DOE Alternative Fuels Data Center, station counts by state; data last updated 22 August 2026.
Each of those cabinets is dc equipment with a maintenance population around it, sited in car parks and forecourts rather than in electrical rooms — and in many cases maintained by people whose electrical safety training was written entirely around ac.
Data centres
176 TWh
of US data centre electricity use in 2023, about 4.4% of total US electricity — projected to reach 325 to 580 TWh by 2028.
Lawrence Berkeley National Laboratory, 2024 United States Data Center Energy Usage Report, December 2024, released by the US Department of Energy.
Every one of those halls has a battery plant behind the UPS and dc distribution in front of it — the part of the installation least likely to appear in the arc flash study commissioned for the switchgear.
Further reading
Two pieces from the journal that go further into the ground covered here.
Electrical Safety for Data Centers
Where the dc bus, the UPS battery plant and continuous-uptime pressure meet a written electrical safety program.
Read the article
2027 NFPA 70E: What Actually Changed
The edition-level changes, including the restructured Chapter 3 and the arrival of Article 310.
Read the article
Next step
Find out whether your study, your program and your roster cover the dc plant you already own.
Send us the arc flash study you hold and a list of the dc equipment on site. We will tell you what is covered and what is not before any engagement is scoped. Fixed scope, quoted up front.
Speak with us directly
(813) 725-0999
safety@arcflashcertification.com
Mon–Fri 8am–6pm ET
Request a dc hazard assessment
Arc Flash
Certification
An NFPA 70E and OSHA electrical safety compliance practice. Procedural and auditing work only — we do not perform incident energy analysis studies or produce arc flash labels.
Services
Compliance Audits Electrical Safety Programs Energized Work Procedures
PPE Program
Training — onsite & virtual
Practice
About the practice
Scope of practice
DC electrical safety
Standards we work to
Contact the practice
Contact
(813) 725-0999
safety@arcflashcertification.com
Mon–Fri 8am–6pm ET
Onsite and live virtual, nationwide
© 2026 Arc Flash Certification · All rights reserved.
NFPA 70E 2027 Edition OSHA 29 CFR 1910 OSHA 29 CFR 1926