NFPA 660 and Combustible Dust: What Changed, and What Your Plant Has To Do About It

What actually changed in December 2024

NFPA published NFPA 660, Standard for Combustible Dusts and Particulate Solids, 2025 edition, in December 2024. It consolidated six standards that had grown up separately over decades:

• NFPA 652, the fundamentals of combustible dust
• NFPA 654, combustible particulate solids in chemical, dye and pharmaceutical operations
• NFPA 664, wood processing and woodworking
• NFPA 61, agricultural and food processing
• NFPA 484, combustible metals
• NFPA 655, sulfur

If you have been working to one of those six, the obligations underneath have not been rewritten from scratch. What changed is the structure: a common core of fundamentals that applies to everyone, with industry-specific chapters layered on top, rather than six documents that each restated the same principles slightly differently and occasionally contradicted each other.

For a plant that handles more than one kind of dust, that consolidation is a genuine improvement. A food plant that also runs a woodshop used to sit under two standards with two vocabularies. Now there is one.

The other notable addition is a documented emergency planning and response requirement, which is new as an explicit chapter rather than something scattered through the previous documents.

What makes a dust combustible, and how you find out

A combustible dust is a finely divided solid that will burn or explode when it is suspended in air at the right concentration and meets an ignition source. OSHA’s working definition says much the same thing and adds the part people miss: regardless of particle size or shape. The old rule that only particles below about 420 microns counted has been dropped from the standards, because fibrous and irregular material that will not pass that sieve has been shown to deflagrate anyway.

Judgment does not settle whether your dust qualifies. A laboratory test does, on a sample taken from where the dust actually accumulates, not from the bag the raw material arrived in. The screening test is a go/no-go explosibility test in a 20-liter sphere. If the sample propagates a flame, it is combustible and the follow-on tests describe how bad it is:

• Kst and Pmax (ASTM E1226): how fast and how hard it burns. Kst puts the dust in a class, St 1 to St 3, and that class drives the size of the explosion vents and the design of the isolation on your collectors.
• Minimum ignition energy (ASTM E2019): how small a spark will set it off. Dusts with a low MIE are the ones where static from an ungrounded hose or a plastic vacuum is enough.
• Minimum explosible concentration (ASTM E1515): how much has to be in the air before it can propagate at all.
• Layer and cloud ignition temperatures: whether a hot surface or a hot bearing will do it without any spark.

The first number is the one most plants need and do not have. A plant that cannot say whether its dust has been tested cannot say whether NFPA 660 applies to it, and an inspector reads that gap the same way an insurer does.

NFPA 652: the fundamentals standard, and where it went

NFPA 652, the Standard on the Fundamentals of Combustible Dust, was first published in 2015 and revised in 2019. It was written to sit underneath the industry-specific standards and say the things all of them had in common: that the owner is responsible for knowing whether the dust is combustible, that a Dust Hazard Analysis is required, that housekeeping, ignition source control, training and management of change are part of the program, and that none of this is optional for an existing plant.

Its single most consequential clause was the DHA deadline. Existing processes had to have a completed Dust Hazard Analysis by September 7, 2020. That date passed with a large share of plants still without one, and it is still the first question an OSHA inspector or an insurance engineer asks.

In NFPA 660, the content of 652 became the general chapters that apply to every plant in scope, before any industry chapter is read. Nothing in it was withdrawn. If your program was built on 652, it maps onto the front of 660 almost section for section, and the DHA you did under 652 is the DHA 660 expects you to maintain and revalidate.

NFPA 654: the standard most general plants were working to

NFPA 654, the Standard for the Prevention of Fire and Dust Explosions from the Manufacturing, Processing, and Handling of Combustible Particulate Solids, was the catch-all. If your dust was not food, wood, metal or sulfur, you were under 654: paper and corrugated, plastics and resins, chemicals, pharmaceuticals, rubber, textiles and most general manufacturing.

Two things from 654 are still quoted on plant floors every day. The first is the layer-depth criterion, the one thirty-second of an inch over five percent of the floor area or 1,000 square feet, whichever is smaller. Later editions added a mass-based method and a risk evaluation method as alternatives, which is where the idea that the threshold depends on your dust and your building came from. The second is its treatment of housekeeping methods, which is the source of the sequence most contractors still follow: vacuuming where the dust can be captured where it sits, sweeping where it cannot, and blowdown only with ignition sources controlled and the area prepared.

In NFPA 660 the 654 material became the industry chapter for chemical, pharmaceutical, plastics and general particulate operations, sitting on top of the common fundamentals. For a corrugated plant or a plastics compounder, the practical change is small: the same housekeeping expectations, the same methods, and a threshold now routed through the DHA rather than read off a page.

NFPA 664: wood dust, and why its threshold was different

NFPA 664, the Standard for the Prevention of Fires and Explosions in Wood Processing and Woodworking Facilities, covered sawmills, furniture and cabinet plants, engineered wood, pellet mills and any operation that sands, saws or shapes wood. It introduced the term deflagrable wood dust for the fine fraction that will propagate a flame, as distinct from chips and shavings that will not.

Its accumulation threshold was deeper than 654’s, with one eighth of an inch the figure most often cited from it, because wood dust is light: an eighth of an inch of wood flour weighs about what a thirty-second of an inch of a denser dust does, and it is the mass in the room that makes a cloud, not the depth on the beam. That difference is the clearest illustration of why NFPA 660 moved the threshold question into the DHA. The same depth of two different dusts is not the same hazard.

The 664 content is now the wood processing chapter of NFPA 660. The equipment requirements that woodworking plants know, on dust collection, spark detection and extinguishing in the duct, and on where collectors sit relative to the building, carried across.

NFPA 61, 484 and 655

The other three standards covered narrower territory and moved the same way:

• NFPA 61, agricultural and food processing: grain, flour, sugar, starch, feed and food ingredients. OSHA’s grain handling standard, 29 CFR 1910.272, sits alongside it and is enforced directly.
• NFPA 484, combustible metals: aluminum, magnesium, titanium, zirconium and others, where the hazards include reactions with water and the extinguishing agents that work on everything else.
• NFPA 655, sulfur.

If a document, a quote or an insurance letter still cites one of the six by number, it is not wrong, it is just citing the chapter by its old name. The obligation is the same one.

The question most plants get wrong first

The most common reason a plant has no combustible dust program is not negligence. It is that nobody believed the standard applied to them.

The instinct is to picture a grain elevator or a sugar refinery, decide the plant is not that, and move on. But the test is not the industry. It is whether the process handles, generates or accumulates a combustible particulate solid, and the list of materials that qualify is longer than most people expect:

• Wood, paper, cardboard and anything cellulose
• Flour, sugar, starch, milk powder, spice, grain and most food ingredients in dry form
• Most plastics and resins, including regrind and buffing dust
• Rubber, coal, sulfur, and many pharmaceutical actives
• Aluminum, magnesium, titanium, zinc, iron and other metal fines

A material that is inert in a sack can be energetic as a suspended cloud. That is the whole subject in one sentence, and it is why a corrugated plant, a bakery, a machine shop and a furniture factory all end up in the same standard despite having nothing else in common.

If you are not sure whether your dust qualifies, that question is answered by testing a sample, not by judgment. It is inexpensive relative to the consequence of being wrong.

The five things an explosion needs

Fire needs three things: fuel, oxygen and an ignition source. A dust explosion needs five, and knowing which two are extra is the most useful piece of technical knowledge on this subject, because those two are the ones housekeeping controls.

1. Combustible dust, the fuel
2. Oxygen, which is simply the air in the room
3. An ignition source, which can be a spark, a hot surface, friction, static discharge or a bearing running hot
4. Dispersion, the dust suspended in air at a concentration that will carry flame
5. Confinement, a space that lets pressure build rather than dissipate

Take away any one of the five and you do not get an explosion. The first three are present in most plants most of the time and are hard to eliminate. The last two are the ones under your control, and they are the reason housekeeping is a fire safety measure rather than a tidiness measure.

This also explains the pattern nearly every serious dust incident follows. A small event happens somewhere confined, often inside a dust collector or a duct. The pressure wave from it shakes the building and lifts the accumulation that has been sitting on overhead steel, beams and cable trays for years. That suspended dust finds the flame front from the first event, and the second explosion is the one that takes the building down.

The first event is usually an equipment problem. The second one is a housekeeping problem, and it is almost always the one that causes the deaths. That is the entire case for cleaning the overhead.

Where the first event usually starts

Since the secondary explosion needs a primary, it is worth knowing where primaries come from. In most plants it is not the open floor.

Dust collectors are the single most common location. They concentrate fine material in an enclosed volume with moving air, and they frequently sit outside where nobody visits them. A collector that has not been opened in years, or one that is undersized for what the plant now produces, is the highest-risk object on many sites.

Ducting and conveying systems carry fine material through confined spaces by design. Scrap and trim systems in converting plants are a recurring example.

Silos, hoppers, bins and bucket elevators, where material is concentrated and mechanical equipment can generate heat or sparks.

Mills, grinders, dryers and blenders, where the process itself creates fines and puts energy into them.

Kelly cleans these systems and reports condition as well as accumulation, because on a collector or a duct what is found is frequently worth more than the cleaning: worn ducting, a collector running past its capacity, or a pit with no practical access.

The Dust Hazard Analysis, in plain terms

The Dust Hazard Analysis, or DHA, is the document everything else hangs from. It is worth being clear about what it is, because plants frequently think they have one when what they actually have is a cleaning schedule.

A DHA is a systematic assessment of where combustible dust exists in your process, what could ignite it, what would happen if it did, and what safeguards are in place or needed. It covers the process equipment and the building. It looks at the dust itself, at the places it accumulates, at ignition sources, and at what would propagate if something started.

Three things about it that plants routinely misunderstand:

It applies to existing plants. The requirement covers operations already running, not only new or modified processes. A plant built in 1978 is not grandfathered out of it.

It expires. Revalidation is expected at least every five years, and sooner if equipment, materials, process or building configuration change enough to alter how the dust behaves. Adding a line, changing a supplier’s grind, or enclosing an area that used to be open can all matter.

It is engineering work. It is performed by someone qualified to do it, and Kelly does not perform DHAs. We clean to what one says, and where a plant does not have one we will say plainly that the analysis comes first, because cleaning without it is work with no defined endpoint.

The threshold question, answered honestly

Ask ten people in this industry what the accumulation limit is and most will say one thirty-second of an inch, about the thickness of a paperclip, over five percent of the floor area.

That figure comes from older layer-depth guidance, it is a genuinely useful rule of thumb, and it is the wrong thing to treat as your legal threshold.

The honest version: the threshold that matters for your building depends on your dust’s bulk density, the volume of the room, the surface area the dust can cover and how readily that dust suspends and ignites. A light paper or wood dust and a dense metal fine do not present the same hazard at the same depth. That is exactly why the standard routes this through a facility-specific analysis rather than publishing one number for everyone.

What this means practically is that a contractor who quotes you the paperclip rule as though it settles the matter is working from a memory of the old standard rather than from your plant. The number you should be managing to is the one your DHA established, and if you do not have that number, that is the gap to close first.

One thing the rule of thumb is genuinely good for: if you can write your name in it, or if the color of a surface is a matter of opinion, you are past the point where any threshold debate is interesting. Deal with the accumulation and argue about the measurement afterwards.

Cleaning without creating the hazard you are removing

This is the part where well-intentioned plants cause incidents.

Settled dust is a fire risk. Suspended dust is an explosion risk. Anything that lifts an accumulation into the air converts the first problem into the second, which is why the method matters as much as the frequency.

Blowdown has to be planned. It is fast, it is how Kelly cleans most plants, and it lifts dust into the air by design. That makes the conditions around it the whole job: ignition sources dealt with before the work starts, the area below prepared, and the plan consistent with the Dust Hazard Analysis.

Vacuum where the surface or the material calls for it, with equipment that matches the hazard: bonded and grounded to prevent static discharge, rated for the material, HEPA filtered where the dust is fine enough to matter. A standard shop vacuum in a combustible dust area is an ignition source with a hose attached.

Metal dusts change the rules again. Several combustible metals react with water to produce hydrogen, which means wetting down an accumulation, the instinctive response, can make it considerably worse. Aluminum, magnesium and titanium fines need equipment and procedures chosen for them specifically.

Work top down. Overhead steel, cable trays, ductwork, lighting and roof structure hold the accumulation nobody sees, and it becomes floor dust the moment anything disturbs it. Cleaning floors first is money spent twice.

No ignition sources introduced by the work. Non-sparking tools, bonded equipment, and attention to what else is happening in the area while the cleaning is underway.

What OSHA actually cites

There is no single comprehensive OSHA standard for combustible dust. That surprises people, and it leads to the assumption that there is therefore nothing to be cited for, which is wrong.

Enforcement generally arrives through three routes. Housekeeping and walking and working surfaces, 29 CFR 1910.22, which requires that workplaces be kept clean, orderly and free of recognized hazards. Specific standards where they apply, such as grain handling facilities under 1910.272. And the General Duty Clause, Section 5(a)(1) of the OSH Act, which covers recognized hazards likely to cause death or serious harm where no specific standard applies. NFPA 660 is frequently what establishes that the hazard was recognized.

OSHA also runs a Combustible Dust National Emphasis Program, directive CPL 03-00-008, effective July 2023, which directs inspections at facilities that generate or handle combustible dusts. A plant in a targeted industry can be inspected without a complaint or an incident preceding it.

What your insurer is looking at

In practice, more Kelly work is triggered by an insurer than by OSHA.

Property insurers inspect plants that carry combustible dust exposure, and what they photograph is predictable: horizontal surfaces above the production floor, dust collection equipment, ducting, and the scrap or conveying systems where fine material concentrates in enclosed volumes with moving air and mechanical equipment. In a converting plant, the baler and the scrap system are where the attention goes, because that is where these fires historically start.

What changes a renewal conversation is not a single deep clean performed three weeks before the walk. It is a documented program: a stated interval, records of the last several cycles, and photographs. An insurer can tell the difference, and so can an auditor.

Closing a finding: what “done” actually looks like

Most of this work is commissioned because something has to be closed out. A DHA raised a housekeeping gap. An insurer asked a question. An inspection produced a finding with a date attached.

Cleaning that cannot be evidenced closes none of those. What closes them is a record:

• What was cleaned, by area, and to what condition
• Dated photographs before and after
• The method and equipment used, which is what demonstrates the cleaning did not create a hazard
• Entry permits and atmospheric logs for any confined space
• Disposal documentation and waste characterization for what was removed
• Training and insurance records for the people who did it

Where the scope was written against specific findings, the report should be structured against those findings, so the person who raised them can close them line by line rather than reading a general account of cleaning.

Where to start, in order

If your plant handles a combustible dust and you are not confident in your position, the sequence is not complicated.

One. Establish whether your dust is combustible, by testing rather than assumption, if that has never been documented.

Two. Get a Dust Hazard Analysis, or revalidate the one you have if it is more than five years old or the process has changed. This is the document that sets your thresholds and your frequencies, and everything downstream depends on it.

Three. Clean to what it says, with methods that match the material, working top down, and document it.

Four. Set an interval from what the first cleaning cycle reveals about how fast the building reloads, rather than from a calendar convention. Then keep to it, because the interval and the records are what turn a recurring finding into a closed one.

Kelly performs step three and step four. Steps one and two are somebody else’s work, and we will tell you so rather than sell around it.