The batch is three hours in, the reactor is at temperature, and the desiccant dryer flips over to regeneration. Twenty seconds later the header sags fifteen pounds, half the positioners in the unit start hunting, and the board operator is watching two control valves drift while he figures out whether this is a process problem or an air problem.
It is an air problem. It is almost always an air problem.
Compressed air for chemical plants is not the same job as compressed air for a machine shop. In a machine shop, low air means a slow impact wrench. In a chemical plant, low air means valve actuators losing position, dampers drifting, and a fail-safe cascade that shuts a unit down and costs you a day of production. The air system is not a convenience utility here. It is part of the control system, and it gets designed like one.
Plant air and instrument air are two different systems
The first thing that separates a chemical plant from a general manufacturing shop is that you will almost certainly run two air services, and they have different specs.
Plant air is the utility side. Hose stations, pneumatic tools, drum pumps, blow-off, diaphragm pumps moving solvent or slurry, the occasional air motor on an agitator in a classified area. Plant air can be dirtier and wetter because nothing downstream of it holds a setpoint.
Instrument air feeds control valve positioners, I/P transducers, on-off actuators, solenoid pilots, analyzer purges, and any instrument enclosure you are purging to keep it out of an area classification. This is the air that has to be clean and dry every single hour of every single day, including the coldest night in January.
The practical rule is simple. Feed instrument air from the plant air header, never the other way around, and put a check valve and a dedicated receiver between them so a hose station somebody left open cannot pull down your control air. One compressor set feeding both is fine. A single header with no separation is not, because the day a crew drops a 3/4 inch hose on the ground you find out what your actuators do at 60 psi.
What ISA-7.0.01 actually asks for
The instrument air spec most US plants write to is ANSI/ISA-7.0.01-1996, Quality Standard for Instrument Air. It is short and it is specific, which is why people keep using a standard from 1996.
| Parameter | ISA-7.0.01 requirement | What that means in practice |
|---|---|---|
| Pressure dew point | At least 18°F (10°C) below the lowest ambient temperature any part of the system sees, and never above 39°F (4°C) at line pressure | Indoor-only headers can live on a good refrigerated dryer. Any outdoor pipe run in a freezing climate pushes you to desiccant. |
| Particle size | 40 micron maximum | A generous ceiling. Most plants filter to 1 micron or better anyway, because positioner orifices are small. |
| Oil content | 1 ppm by weight maximum, total liquid and vapor | At normal air density that is roughly 1.2 mg per cubic meter, which is looser than ISO 8573-1 oil Class 3. Most modern specs go tighter. |
| Contaminants | No corrosive or hazardous gases, no flammable content | This is really an intake location requirement. See below. |
Read the dew point line twice, because that is the one that trips people. The standard does not say minus 40. It says 18 degrees below your lowest ambient, with a hard ceiling of 39°F. If your instrument air tubing runs across a pipe rack in Ohio and the design low is minus 10°F, you need a pressure dew point of minus 28°F or better, and that is desiccant territory. If every inch of your instrument air is inside a heated building that never drops below 55°F, a refrigerated dryer holding 35 to 38°F pressure dew point satisfies the standard. Our breakdown of ISO 8573-1 air quality classes lines these up if your corporate spec is written in ISO classes instead.
Sizing it without guessing
Chemical plant air demand is lumpy, and the average tells you very little. Here is the order that works.
- Count the continuous bleeds first. Older I/P transducers and positioners bleed constantly. Analyzer purges bleed constantly. Enclosure purges under NFPA 496 bleed constantly. These add up faster than anyone expects, and unlike actuators they never stop.
- Add the stroking load. Pull the air consumption per stroke off the actuator data sheets and figure out how many valves move in your worst minute, which is usually a unit trip or a batch changeover, not steady state.
- Add plant air with a diversity factor. Hose stations and tools never all run at once. Count them, then apply something like 25 to 40 percent depending on how many crews you have.
- Add the dryer purge. A heatless desiccant dryer gives back roughly 15 to 18 percent of its rated flow as purge air. That is real compressor capacity you do not get to use. Heated blower-purge units cut it down to single digits at the cost of more complexity.
- Add a margin, then size storage separately. Do not solve a transient problem by buying a bigger compressor. Solve it with a receiver.
On that last point, the storage math is worth knowing:
V = (T × C × 14.7) / (P1 - P2)
Where V is receiver volume in cubic feet, T is the ride-through time you want in minutes, C is the demand in SCFM, P1 is your starting pressure and P2 is the lowest pressure your instruments will tolerate. Run that with a two minute ride-through at 100 SCFM from 110 psi down to 80 psi and you get about 98 cubic feet, or roughly a 750 gallon tank. That is a large receiver, and that is the point. Instrument air ride-through is bought in tanks, not horsepower. We keep ASME air receivers in a wide range of sizes for exactly this reason.
The dryer decides whether this system works
You can get the compressor wrong and limp along. Get the dryer wrong and you will be chasing sticky positioners and frozen tubing for years.
For a plant with any outdoor instrument air, that means a twin tower desiccant dryer holding minus 40°F pressure dew point. Two things matter more than the brand on the cabinet.
First, protect the desiccant. Liquid water destroys activated alumina, and oil carryover poisons it permanently. That means a properly sized wet receiver, an aftercooler, a water separator, and a coalescing prefilter ahead of the towers, plus a particulate afterfilter to catch desiccant dust on the way out. Skipping the prefilter is the single most expensive shortcut in this entire system. Our desiccant air dryers and matched compressed air filters are meant to be specified together for that reason.
Second, instrument the dryer. A dew point transmitter on the outlet with an alarm back to the DCS is cheap insurance. Without one, the first indication that your dryer has quit drying is a control valve sticking on a cold morning, and by then the desiccant may already be soaked.
If you are weighing dryer types, the comparison in refrigerated vs desiccant air dryers walks through the tradeoff in more detail.
Where the compressor goes, and where it breathes
This is the part that is specific to chemical plants and gets handled badly more often than it should.
Standard air compressors are not rated for classified areas. A packaged rotary screw has a motor, a starter, a controller and a cooling fan, and none of that is built for a Class I Division 2 location under NFPA 70. The normal answer is to put the compressor in a dedicated, unclassified compressor building or an electrical room, well away from the process, and run pipe to the units. That is the cheap answer and it is usually the right one.
The harder question is the intake. ISA-7.0.01 says the air must be free of corrosive and hazardous contaminants, and a compressor is very good at taking whatever is in the yard and concentrating it into your control system. If your intake sits downwind of a vent stack, a tank farm breather, or a scrubber discharge, you are pumping that into every positioner in the plant. Run the intake up and away from process vents, check the prevailing wind, and put it on the side of the building that stays clean. An intake filter change is easy. Replacing corroded actuator internals across a unit is not.
Cooling deserves a look too. Most air-cooled rotary screws are rated to about 104°F (40°C) ambient, and they typically discharge 100 to 110 degrees above whatever the room is. Chemical plant compressor houses tend to be small, closed and hot. Size the ventilation for the heat, not the floor area.
Backup air, and the nitrogen shortcut
Every plant eventually asks the same question: what happens when the compressor trips?
The honest first answer is redundancy. Two compressors with a proper lead lag control scheme and enough storage to cover the start of the standby machine handles the overwhelming majority of failures, and it handles them without changing the gas in your header.
The second answer, common in plants that already have bulk nitrogen on site, is a nitrogen backup that regulates into the instrument air header on low pressure. It works, and it is a real design used across the industry. It also has consequences people forget to write down. Nitrogen is an asphyxiant, and an instrument air header that can be filled with nitrogen must be treated as a nitrogen system from then on: no breathing air taken from it, ever, no using an instrument air hose to blow off a pump seal in a confined space, and clear labeling at every station. If you put in a nitrogen backup, that has to land in your procedures and your training, not just on the P&ID.
Worth saying plainly: we sell the compressors, dryers, filtration, receivers and nitrogen generators that make up this system. We do not engineer your instrument air backup philosophy or stamp your PSM documentation. That belongs to your process safety group.
Where instrument air shows up in your PSM program
OSHA's process safety management standard, 29 CFR 1910.119, has no line item that says "instrument air." What it has is a process hazard analysis element and a mechanical integrity element, and in most covered plants instrument air lands squarely in both. If the PHA concludes that loss of instrument air drives a unit to a fail-safe state that protects people, then the air system is part of that safeguard, and it has to be maintained and tested like one: the dryer, the receivers, the check valves between plant and instrument air, and the backup, all on a documented interval. Plants that do this well treat the air system as process equipment. Plants that do not treat it as a facilities item and find out during an incident investigation.
Five mistakes we see repeatedly
- One header for everything. No separation between plant air and instrument air, so every hose station is a potential control system upset.
- A refrigerated dryer feeding outdoor tubing. Works beautifully until the first hard freeze, then you are thawing impulse lines with a heat gun.
- No prefilter ahead of the desiccant towers. Cheapest way to destroy an expensive dryer. Our guide to coalescing air filters covers what has to sit ahead of the towers.
- Storage sized as an afterthought. Then a bigger compressor gets bought to fix a problem that was never about capacity.
- Intake in the wrong place. Nobody notices for two years, and then the actuator rebuilds start.
Frequently Asked Questions
Do I need oil-free compressors for instrument air in a chemical plant?
Usually not. ISA-7.0.01 allows up to 1 ppm by weight of oil, and a lubricated rotary screw with good coalescing filtration and a carbon adsorber comes in well under that. Oil-free becomes the answer when your corporate spec calls for ISO 8573-1 oil Class 0 or Class 1, when the air touches product, or when the plant has already had an oil carryover event that nobody wants to repeat. It costs more to buy and it is not automatically more reliable, so make it a decision rather than a default.
What pressure dew point should I specify?
Start with the coldest temperature any part of the instrument air system will see, subtract 18°F, and that is your ceiling. Then check it against the 39°F hard limit in the standard. In practice, if any instrument air pipe or tubing goes outdoors in a climate that freezes, specify minus 40°F and buy a desiccant dryer. If it is all indoors and heated, a good refrigerated dryer is honest engineering and it costs a lot less to run.
Can I put the air compressor in the process area?
Not a standard packaged unit. A typical air-cooled rotary screw has electrical gear that is not rated for a Class I Division 1 or Division 2 location. The usual solution is a compressor building outside the classified boundary with piping run to the units. If a compressor genuinely has to sit in a classified area, that becomes a custom package with rated electrical components, and the cost goes up sharply.
How much storage should an instrument air system have?
Enough to ride through the start of your standby compressor plus a margin, which for most plants means one to three minutes of full instrument air demand. Run the formula above with your actual numbers rather than using a gallons-per-horsepower rule of thumb, because instrument air demand and compressor size are not tied together the way they are on a shop air system.
Is nitrogen backup on the instrument air header a good idea?
It can be, and plenty of plants run it. The engineering is straightforward. The part that gets underestimated is the administrative side: once that header can contain nitrogen, it is an asphyxiation hazard, and every hose station, every procedure and every new hire orientation has to reflect that. If your site cannot commit to maintaining that discipline, a second compressor is the safer answer.
The short version
Getting compressed air for chemical plants right comes down to four decisions: separate instrument air from plant air, hit the ISA-7.0.01 dew point based on your real minimum ambient rather than a habit, protect the dryer with proper filtration and storage, and put the compressor and its intake somewhere that is both unclassified and clean. Do those four and the air system stops showing up in your incident reports.
If you want help matching a dryer, filter train and receiver to an existing plant, give us the compressor model, the header pressure, the lowest ambient your tubing sees and the instrument count, and we can work it out with you.
