Does Your Feedstock Even Let You Choose a Pyrolysis Reactor Type?

Which of the pyrolysis reactor types gives the best oil yield? That's the first question almost everyone asks me, and it's the wrong one to start with. Yield is something you tune over a commissioning season. What you can't tune is the particle size your front end is able to deliver, and that single number eliminates most of your reactor shortlist before you've costed anything. A bubbling fluidized bed wants biomass particles of less than 2-3 mm to hit the heating rates it's designed around, according to IEA Bioenergy's Task 34 review of pyrolysis reactor designs. If your material is shredded tire or dewatered sludge cake, you just bought a mill, a dryer, and the parasitic load that comes with both, before a single kilogram reaches the reactor.

So a useful comparison runs on four criteria: what the front end has to deliver, how heat gets into the solid, what the seal architecture does to your air permit, and what the feedstock does to the reactor by year five. I'll take them in that order, because that's the order they bite you in. Twenty years around thermal plant have taught me that the reactor is rarely the thing that fails. It's whatever had to be built around it.

Particle size is the first gate, and you pay for it upstream

Rotary kilns are forgiving on size. Metso's indirectly fired rotary kiln for tires runs washed chips at roughly two inches square [equipment spec, per Metso's published process description], and a single shredder stage gets you there. A fluidized bed pyrolysis reactor is not forgiving at all. Grinding rubber crumb or fibrous biomass down to a few millimetres is a different machine class from shredding it, which means a Vecoplan or Doppstadt stage feeding a mill rather than replacing one, plus the screening and recirculation that any mill circuit drags along (and a dust explosion assessment, once you're milling dry organics).

Actually, that 2-3 mm figure is a fast-pyrolysis number. Run slow pyrolysis for char and the tolerance widens considerably, which is part of why rotary kilns and screw reactors dominate the biochar end of the market, which is why rotary kiln pyrolysis shows up wherever char is the product and fluidized beds dominate the liquid-fuels end. Same physics, different objective.

A screw pyrolysis reactor sits in between. They don't need fluidizing gas at all, and the Task 34 survey notes that product recovery and gas cleaning get simpler as a result, since you aren't diluting your vapour stream with a carrier. What they need is abrasion tolerance. Feed a screw a construction-and-demolition fines stream with silica in it and you'll learn how fast a flight wears. On our Gulfport CDW processing line in 2023 we spent most of a quarter blaming chloride for ash agglomeration that turned out to be silica from a stream we were accepting without any inbound QA. Different unit operation, same lesson: the abrasive fraction nobody specified is the one that writes your maintenance budget.

Reactor	What the front end must deliver	What limits it first
Rotary kiln	Coarse chips from one shredder pass; tolerant of mixed sizing	Heat flux through the shell per unit length; seal and refractory wear
Fluidized bed, bubbling or circulating	Milling to a few millimetres, plus a sand inventory and its handling	Heat transfer at large throughput; char attrition carrying solids into the oil
Screw or auger	Tolerant on size, intolerant of abrasives and sticky phases	Indirect heating through the wall; flight and liner wear

How the heat gets in sets the ceiling on one train

Vendors gloss this one. Pyrolysis is endothermic, so every design is a heat-delivery argument dressed up as a reactor. A fluidized bed delivers heat through a circulating solid, and the rotating-cone variant runs sand-to-biomass ratios of 10 to 30 per Task 34, which tells you how much inert mass you're heating and moving for every ton of feed. A screw reactor usually delivers heat through the wall, and Campuzano, Brown and Martinez concluded in their 2019 review of auger pyrolysis in Renewable and Sustainable Energy Reviews that wall heat transfer is exactly what limits scale-up when energy is supplied indirectly.

Wall-heated geometry scales with surface area while throughput scales with volume, so doubling the tube doesn't double the output. That's why you see screw installations multiply trains instead of growing them. The Task 34 figures put a representative auger pilot at 0.5 t/h on wheat straw; getting to industrial tonnage means repeating that unit, with every repeat bringing its own drive, seals and instrumentation.

Circulating fluidized beds are the standard answer for very large throughputs, and Task 34 is blunt that heat transfer at large scale remains one of the unproven areas. Higher gas velocities also mean more char attrition, and that char ends up in the oil. If you've ever filtered a bio-oil with high solids you know what that does to your downstream equipment and your offtake spec.

Rotary kilns sit between those extremes. Indirect firing through the shell gives you the same surface-area problem as a screw, but a kiln's length and the tumbling action buy you a lot more contact area per unit of volume. The electrically heated kilns VOW ASA supplies for sludge are sized at around 160 tons per day of dewatered cake across two parallel lines on the Rhode Island project described below [per the facility's own process description to EPA], which is roughly where kiln geometry lands at industrial scale.

Air ingress is a reactor design choice with a permit attached

Air ingress is the criterion nobody puts in the comparison table, and it moves more money than any yield point on the brochure. On 13 May 2025, EPA Region 1 issued an applicability determination for the QSS Biosolids sewage sludge pyrolysis facility proposed in North Kingstown, Rhode Island. The determination found the facility is not subject to 40 CFR Part 60 subparts LLLL, O or CCCC, on the reasoning that its pyrolysis reactors do not combust anything.

What made that argument stick was hardware. The supplier estimated total oxygen entering the process at 30 kg/hour, a stoichiometric ratio of 0.005, and got there with infeed airlocks, outfeed airlocks, equipment seals, under-pressure control and inert gas purging, with the equipment leak-tested before start-up. The reactor runs at 600 to 750 degrees C under indirect heat. Every one of those items is a reactor-architecture decision made years before the permit application.

Think about what that does to the reactor comparison. A screw with a well-sealed infeed and a nitrogen purge has a straightforward air-ingress story. A rotary kiln has rotating seals at both ends, which is a harder seal problem but a very well understood one, and VOW's case shows it closes. An open-topped or loosely sealed design, or any arrangement where you're pulling combustion air anywhere near the reaction zone, gives a reviewer something to pull on. And a reviewer who decides you are combusting solid waste hands you a whole subpart you did not budget for. Fix the seal and airlock architecture before you file, not after a reviewer asks about it.

None of that is a general exemption, by the way. EPA relied on the applicant's own numbers, said so in the letter, and noted the determination may no longer hold after a process change. But it does tell you which reactor characteristics an air reviewer in the United States will ask about, and the answer is mechanical, not thermal.

The one study that costed all three on the same basis

Public, apples-to-apples capital comparisons of pyrolysis reactor types barely exist. The best one I know is old: Westerhout and colleagues at Twente designed a 50 kton/year mixed plastic waste plant three ways, as a bubbling fluidized bed, a circulating fluidized bed and a rotating cone reactor, and costed all three for 1996 construction in the Netherlands.

Per that paper, fixed capital ran from 14 million guilders for the cone to 20 million for the circulating bed. Return on investment, in the same study, came out at 29.5 percent for the cone, 21.7 percent for the bubbling bed and 14.2 percent for the circulating bed.

The absolute numbers are a museum piece. The ranking isn't, and the reason it isn't is that the spread came from the same places it comes from today: lower reaction temperature means lower energy demand, and a design needing little or no additional utilities carries less of everything around it. The same study found something else worth carrying into any plastics project: low-temperature pyrolysis converted 98 percent of the chlorine to HCl gas, and only 50 to 60 percent of the polymer ended up as product gas. That second figure is the one most plastics-to-fuel business cases quietly round up.

Chlorine and ash decide what the reactor looks like in year five

Chloride is the slowest and most expensive failure mode in thermal plant, and reactor geometry changes where it lands. On the Pinellas County retrofit in 2021 we added a thermal scrubber stage, and the chloride attack showed up anyway on a year-five superheater, downstream of where anyone was looking for it. Pyrolysis moves that problem rather than removing it. Strip 98 percent of chlorine as HCl in the reactor and you've handed it to your gas cleaning, your condenser metallurgy, or your thermal oxidizer.

Ash behaviour splits the reactor types harder than oil yield does. A fluidized bed has a bed to defluidize, and a feedstock with the wrong alkali or silica content will agglomerate it. A rotary kiln will carry ash through and let you pull it cold (assuming your discharge seal holds, which is its own maintenance item). A screw will grind it into the flights. Most of the operational problems I see get written up as process issues are instrumentation drift instead - a thermocouple reading shell temperature and getting called bed temperature, a flow transmitter nobody has rezeroed since start-up. Two shifts chasing a drifting signal. That's the common case, and it's worth more attention than the reactor-selection argument gets.

If you're working through this for a sludge line specifically, the drying train in front of the reactor usually dominates both capex and parasitic load, which is why I've argued that sludge pyrolysis is a drying business with a reactor on the end. Get the dryer wrong and your choice between pyrolysis reactor types barely registers in the cost stack.

A shortlist by feedstock

Tires, mixed bulky plastics, anything you can only get to chip size: rotary kiln. The size tolerance is worth more than the yield difference, and the recovery route is proven. Metso notes its tire oil resembles No. 2 fuel oil in viscosity and heating value but carries higher sulphur and a lower flashpoint, which is an offtake conversation you should have before you order steel. The same holds across most of the waste conversion facilities I've walked: the product spec gets argued late and costs more than the reactor decision did.

Clean, dry, uniform biomass at scale where you want liquid: fluidized bed, accepting the mill and the sand loop. But what does that sand inventory cost you in availability at a hundred tons a day rather than a thousand? I've not seen a convincing answer from anyone selling one.

Sludge, digestate, single-stream feedstock where char is the product and the permit path is the risk: electrically heated kiln or screw, with the seal and airlock package specified from day one. The pyrolysis systems being built for biosolids today are almost all in this class, and the reason is that permit argument, which no thermal spec sheet addresses.

And the anchored warning for all three. Brightmark's Ashley, Indiana plastics pyrolysis subsidiaries missed a scheduled debt payment on 1 March 2025 and filed Chapter 11 sixteen days later, running at 5 percent of nameplate capacity against 178.3 million dollars of debt, according to court filings reported by Waste Dive. Years of parent-company equity and a state bond package had already gone in by then, per the same reporting. No reactor-type comparison would have predicted that. Front-end preparation, uptime and offtake did it, which is the same list as above.

Where does this stop holding? Below about 50 tons a day the whole analysis changes shape, because fixed costs swamp the reactor differences and capex per ton stops closing no matter which geometry you pick. Wet feedstock does the same thing from a different direction: anything above roughly 60 percent moisture turns the comparison into a dryer-selection exercise. And the air-ingress argument is specific to the United States. In the EU, whether your unit falls inside the waste-incineration provisions of Directive 2010/75/EU is a different test with different evidence, so a seal package that buys you an exemption in Rhode Island buys you a permit negotiation in Rotterdam. If you're sizing a line for landfill diversion outside North America, settle the classification question with your regulator before you shortlist a reactor at all, because in some jurisdictions the answer narrows the list to one before any engineer gets a say.

Sources & Notes

• Particle-size limits, sand-to-biomass ratios, the 0.5 t/h auger pilot and the unproven-at-scale heat-transfer caveat all come from IEA Bioenergy Task 34's reactor survey.
• The permit reasoning, the 30 kg/hour oxygen figure and the 600-750 degrees C operating window are quoted from EPA Region 1's 13 May 2025 applicability determination for QSS Biosolids. Read the whole letter before relying on it; the reasoning is specific to that process description.
• Capital and ROI rankings across the three reactor designs: Westerhout et al., "Techno-economic evaluation of high temperature pyrolysis processes for mixed plastic waste", Chem. Eng. Res. Des. 76(A), 1998, 427-439.
• Scale-up limits on indirectly heated screws: Campuzano, Brown and Martinez, "Auger reactors for pyrolysis of biomass and wastes", Renew. Sustain. Energy Rev. 102, 2019, 372-409.
• Kiln feed sizing and tire product split: Metso's tire pyrolysis process description. The Brightmark figures are from Waste Dive's reporting on the Chapter 11 filing and auction.
• Pinellas County and Gulfport details are from my own commissioning and operating notes on those lines, anonymized where the client asked.

Disclosure: I'm a senior plant engineer at Renewable Waste Energy and have worked on retrofit and commissioning teams for the thermal systems discussed here.