Every material recovery facility we have ever walked into has the same tell. Stand next to an optical sorter for thirty seconds and listen to the ejector bank fire. That machine-gun rattle is your compressed air system's entire workload, and if you have more than one sorter, the compressor room is the second largest electrical load in the building after the motors on the belts.
Compressed air for recycling facilities is a sorting problem disguised as a utility problem. The sorters set the spec. Everything else on your air system is a rounding error next to them, and almost every air complaint in a MRF traces back to either not enough volume or air that was too dirty for the valves.
What Optical Sorters Actually Pull
An optical sorter sees a target on the belt and fires a short blast from a bank of solenoid ejector valves to knock it off the stream. An 80 inch single-eject unit typically carries around 80 valves. Wider machines and multi-eject setups run well past 180.
Those valves want roughly 87 to 116 PSI, which is 6 to 8 bar, at the manifold. Not at the compressor. At the manifold, after everything upstream has taken its cut.
The volume is where people get surprised. Consumption is not a fixed number, and anyone who quotes you one without asking questions is guessing. It scales with:
- Reject rate. A sorter pulling 8 percent of the stream fires far less than one pulling 40 percent.
- Throughput. More tons per hour is more targets per second.
- Sensitivity and recipe. A tight recipe fires more often, including on marginal targets.
- Material. Fiber, container line and PET all behave differently.
Published figures for an individual sorter often land in the 35 to 70 CFM range at nominal settings, but a multi-sorter container line running hard is a different animal, and large facilities routinely spec systems in the hundreds of CFM, with the biggest running past 1,000 SCFM at 90 PSI. The only honest way to size is to get consumption data from your sorter manufacturer for your recipe and your throughput, then add the rest of the plant on top.
Size for the Plant You Run on a Bad Day
The mistake is sizing for the average shift. Sorter air demand is spiky and it climbs exactly when you least want it to, which is when the inbound stream gets dirty and reject rates go up.
A workable method:
- Get per-sorter consumption from the OEM at your actual reject percentage, not the brochure number.
- Add the balance of plant: baler air, cylinder actuation on diverters and chutes, blow-off stations, shop air in the maintenance bay. In most MRFs this totals 20 to 60 CFM.
- Add 25 to 30 percent for leaks. A MRF runs long hours in a vibrating, dusty building, and leak rates of 25 percent are common rather than exceptional.
- Add headroom for the sorter you will add in two years. Everybody adds one.
For the compressor itself, a variable speed rotary screw fits this load well, because sorter demand genuinely varies with the material coming down the belt and a fixed-speed machine will spend a lot of its day unloaded burning power. Two smaller machines in lead lag also beats one large one here, both for turndown and because a single-machine failure does not stop the line.
Storage Is Not Optional on a Sorter Line
Ejector firing is the textbook case for receiver capacity. Thousands of short, hard pulls per minute, with the demand curve jumping the instant a contaminated bale of inbound hits the line.
Put real storage close to the sorters, not just at the compressor. A dedicated receiver tank near the sorter deck does two things: it absorbs the burst so manifold pressure stays in band, and it decouples the sorters from pressure events elsewhere in the plant. If your sorters lose ejection accuracy when the baler cycles, that is a storage and piping problem and no amount of extra compressor will fix it cleanly.
Air Quality Is Why Your Valves Fail
This is the part that costs MRFs the most money and gets blamed on the wrong thing.
Ejector solenoids are precision parts firing millions of cycles. Feed them wet or oily air and they stick, either open or closed. A valve stuck open bleeds air continuously and drags your whole system pressure down. A valve stuck closed puts a dead spot in the ejector bank, and you will not see it in a gauge, you will see it as a quality problem in the outbound bale weeks later.
If sticking valves are your specific complaint, we went deeper on the dryer side of this in air dryers for recycling facilities, including how to choose between refrigerated and desiccant when part of your pipe run crosses an unheated tip floor.
What the train needs to look like:
| Stage | Why it is there |
|---|---|
| Aftercooler and water separator | Drops the bulk liquid before anything else sees it |
| Particulate filter | Catches pipe scale and rust from the header |
| Refrigerated dryer, 38 to 50 F PDP | Stops vapor from condensing in the manifold. This is the baseline for an indoor MRF |
| Coalescing filter, 0.01 micron | Last defense on oil aerosol before the ejector bank |
| Point of use filter at each sorter | Cheap, and it catches whatever the main train missed |
Go to desiccant only if you have outdoor pipe runs that freeze or a sorter OEM who specifies a lower dew point. Indoors, a refrigerated dryer is the right tool and costs a fraction as much to own. Our breakdown of ISO 8573-1 classes is worth ten minutes before you sign off on a spec, because sorter manuals often quote a class rather than a dew point.
The Dust Problem Nobody Designs For
A MRF is one of the dustiest industrial environments there is. Fiber dust, glass fines, paper lint, and it is everywhere, including in the compressor room if you let it be.
Three things that matter more here than in a normal plant:
Intake location. Do not pull intake air off the tip floor or anywhere near the screens. Duct the intake from a clean side of the building or outdoors with a hood. A plugged intake filter costs you capacity long before it throws an alarm, and in a MRF they plug fast.
Cooler cleaning. Air-cooled aftercoolers and oil coolers blind over with fiber dust. Put cooler cleaning on the weekly list, not the annual one. A blinded cooler is the single most common cause of high temperature trips in this industry.
Room ventilation. A compressor rejects almost all of its input energy as heat. Seal a 75 HP machine in a dusty room with no airflow and you get high discharge temperatures, more condensate, shorter oil life, and eventually a summer shutdown.
Energy, Because This Is a Big Number
A MRF running two shifts with multiple sorters is spending real money on air. Two things pay back fast:
Fix the leaks. A plant with years of vibration, quick fittings and hose has a leak rate you would not believe until you measure it. Walk it during a shutdown with the compressor running and listen, or put an audit on the calendar. This is the cheapest CFM available to you.
Do not over-pressurize. If your sorters want 90 PSI at the manifold, running the header at 125 to cover a piping problem costs roughly 1 percent in energy for every 2 PSI. Fix the piping restriction instead. Every 2 PSI you carry unnecessarily is money leaving through the meter all day.
Also worth measuring rather than guessing: put a flow meter on the sorter header. Knowing what the sorters actually draw at different reject rates turns your next compressor purchase from a guess into a number.
What MRFs Get Wrong
Sizing off the sorter brochure. Those numbers assume a nominal recipe and a clean stream. Your stream is not clean. Ask for consumption at your reject rate.
Treating the dryer as phase two. The dryer costs less than one round of ejector valve replacement across a sorter bank. Buy it with the compressor.
Running a single large fixed-speed machine. Sorter demand varies. A fixed-speed screw unloaded still burns a meaningful fraction of full power, and one machine means one failure stops production.
Undersized header to the sorter deck. The compressor gauge reads 110 and the manifold reads 82. That is 28 PSI eaten by pipe, and it is the most common version of this complaint. Pipe sizing is cheaper than horsepower.
No condensate plan. You will make a lot of water, and it is oily. Find out what your local authority requires before the inspector does. We covered the rules in condensate disposal.
Where to Start
Call your sorter OEM and get real consumption numbers at your reject rate and throughput. Add the balance of plant, add 25 percent for leaks, and put storage near the sorter deck rather than only at the compressor. Then build the treatment train for the valves, not for the compressor. Browse rotary screw compressors when you have that number.
Get compressed air for recycling facilities right and the sorters do what they were sold to do. Get it wrong and you spend the next three years blaming the optics for a compressor room problem.
Frequently Asked Questions
How much compressed air does an optical sorter use?
It depends heavily on reject rate, throughput and recipe, so there is no single answer. Individual sorters often land in the 35 to 70 CFM range at nominal settings, while large multi-sorter facilities can run into the hundreds of CFM and the biggest past 1,000 SCFM at 90 PSI. Get the figure from your OEM at your actual reject percentage.
What pressure do sorter ejector valves need?
Typically 87 to 116 PSI, or 6 to 8 bar, measured at the manifold rather than at the compressor. If your compressor gauge looks fine but the manifold is 25 PSI lower, you have a piping or storage problem, not a compressor problem.
Why do my ejector valves keep sticking?
Air quality, nearly always. Wet or oily air makes solenoids stick open or closed. A valve stuck open bleeds system pressure continuously; one stuck closed leaves a dead spot in the bank that shows up as an outbound quality problem rather than an alarm. A refrigerated dryer plus 0.01 micron coalescing filtration is the fix.
Refrigerated or desiccant dryer for a MRF?
Refrigerated for an indoor facility, holding a 38 to 50 F pressure dew point. Go desiccant only if you have outdoor pipe runs that freeze or a sorter manufacturer specifying a lower dew point, since desiccant costs more to buy and burns purge air.
One big compressor or two smaller ones?
Two smaller machines in lead lag usually wins in a MRF. Sorter demand varies with the material on the belt, so turndown matters, and a single machine means a single failure stops the line. Variable speed drive on at least one of them handles the swing efficiently.
