Experiment 3 - top-burn reactor (C)

Following a presentation of the TLUD stove concept by Dr. Paul Anderson in June 2023, I determined to explore biochar production in a more traditional way. I constructed reactor C as follows:

one-gallon paint can - drill several 1/4-inch (6mm) holes in the bottom to admit some air. From galvanized sheet metal and 5/16-inch aluminum pop-rivets, fabricate a cap and chimney. The cap is made by cutting a disc (actually about 250 degrees of a complete circle) of sheetmetal with diameter of approximately 25cm and center hole diameter of approximately 8cm. This disc is formed into a cone with about 2cm overlap and riveted in shape. A 28cm x 91cm piece of sheetmetal (sized by the Method of First Approximations - experimenter had a piece that size and said to self “that looks approximately right for a first go!”) is formed into a 91cm length tube (the chimney) and riveted into shape. Tabs are cut in the cone center area to facilitate riveting the chimney onto the cap. This work should be performed with gloves and in solitude (due to exclamations in foul language that may be necessary).

Paint can loaded with wood, bottom has various ¼-inch (6mm) holes for air. Cap and chimney of galvanized sheet steel. Three tabs cut in cap keep it aligned over the paint can and admit some excess air.

mass of empty can = 199g* (final empty mass was 295g, so I have used that value instead for all calculations)
mass of full can = 1296g
The biomass is pieces of dry milled lumber (possibly spruce?) salvaged from a discarded box-spring. The wood has been cut into approximately 15cm lengths and split into approximately 1.5cm x 1.5cm pieces of “kindling”, and packed as tightly as will fit,
into the reactor.
ignited at 15:15; rain began at 15:40. Elapsed time 25 minutes.
Smothered flame by sandwiching the paint can between two heavy concrete tiles (smooth flat surface), so no oxygen could enter. Biomass conversion was not complete. There was no smoke evident during burning, but only an occasional spark would reach the top of the chimney. Some zinc melted on the cap. IR thermometer indicated T=380C (melting point of zinc is 419.6C).

final mass of full can = 850g

removed friable char that was readily separable from unburned or torrefied wood.

final mass of full can with char removed = 782g

I will supplement this material with additional wood and continue the burn.

reloaded can mass = 1250g

08:00 - ignition
08:25 - roar of flames is reduced
08:32 yellow flame is sometimes not visible, but flame is evident with cap removed
08:45 low flame, but still a little. Smothered can between two pre-wetted tiles
08:50 no smoke in exposed char at first, then faint tendrils appear; re-sealed
08:55 char is cool

final mass = 490g
final mass of empty can = 295
there were a few incompletely charred bits; total mass 26g - left this fraction combined with char.

cut a piece of wood to measure density. Sample size: 43g, dimensions 6.75cm x 6.1cm x 1.6 cm

Analysis spreadsheet —> “Top-fire burn” tab

The char yield from this experiment was 18% of biomass. The void fraction in the can averaged 50%. If the wood is assumed to be dry pure cellulose, the fraction of theoretical yield is 40%. In discussion with a large scale biochar operator (Rodriguez, BRI), biomass carbon content may actually be only about 30% on average.


More developed top-burn reactors of this ilk have the refinement that an outer shell is fitted around the burn chamber so air that will enter the bottom of the burn chamber is pre-heated by passing downward beside the outer surface of that chamber before entering the holes at the bottom. This pre-heats the biomass and makes the process more efficient. If further experiments are to be performed with this reactor, that is a good addition.

The chimney is tall and slender, and poorly supported on the paint can. It becomes very hot (greater than 200C at the top - the exiting gas will ignite paper). The cap and chimney could be better supported by a wider base (such as if an outer shell were added as noted above), and if the chimney were shorter and thus had a lower center of mass.

Bonus Test!

I performed another experiment with this reactor, but packing it with waste cardboard instead of wood. I was able to pack in 335g of cardboard. The cardboard gave some fly-ash from the chimney during the early stage of burning. Burning appeared to be nearly complete within 30 minutes. But I discovered hot coals or embers banked under the accumulated char after ten minutes of sealed cool-down. The coals produced copious smoke once exposed to air, so I replaced the chimney and reignited the reactor. It burned vigorously (flames reached the top of the chimney), for about another 10 minutes with the exposed remnants of cardboard on top of the accumulated char. The mass of char was 110g and there was a further 39g of unburned cardboard. There were again some fugitive embers which I extinguished by breaking them apart. The biomass conversion was 33%, greater than for the wood - but this could be due to inert additives in the cardboard and incomplete combustion of the char. Cardboard alone in the reactor appears to self-smother. It requires greater care to recover char without water-quenching.

Water quenching is commonly used in flame-pyrolysis for this reason of extinguishing fugitive embers. But that practice makes accurate mass-measurements impossible.

I reloaded reactor C with wood from the same inventory, but shortened the chimney to 60cm and placed the reactor on a cement block inside a 5-gallon plastic pail. This will force some of the inlet air to flow downward past the sides of the can, perhaps and thus become preheated. However this test was aborted because the wood failed to ignite. I retried with several milliliters of kerosene accelerant, and a little bit of the wood in the center burned briefly, but although this support an initial burn for a minute or so, it ceased when the kerosene was consumed and then there was much smoke. I had to quench with water hose. Will try again once the wood dries.

This is an Experiment 3 repeat with the shorter chimney, with the reactor inside a plastic pail, and using the dried wood. The mass of wood is 899g. This wood, less dense that the earlier test due to having been slightly termite-damaged, was difficult to ignite and I finally resorted to using about 20mL of kerosene accelerant. In the interest of using the entirety of this wood, once the paint can was ignited and burning well, I added 218g of additional wood gradually, piecewise, by dropping pieces down the chimney. These added pieces caused much popping and showers of ash escaped the chimney (I believe there were no actual termites remaining in the wood). Campfire lore attributes these phenomena to moisture trapped in rapidly heated wood, and this wood was recently extinguished with water and at least the surface had dried.

Combustion with open flame was complete in fifty minutes and the reactor was quenched between tiles for ten minutes. The yield was 143g of char - 12.8% of the added biomass.

The HDPE plastic pail melted in one area - melting point of HDPE is approximately130C. The outside of the reactor during pyrolysis was up to 330C and the outside of the HDPE pail was consistently near 130C, but in most areas it did not deform. Maybe it was a little hotter in the deformed area which was on the downwind side of the reactor. A metal heat shield/air preheating duct will be necessary.

The char yield in this test was lower than before. This could be due to several factors.

  1. the preheating may have caused more carbon to oxidize
  2. the gradual addition of 20% of the biomass into the upper flame zone may have promoted more carbon to oxidize
  3. the more complete combustion in less time suggests hotter flame which may have favored carbon combustion.
  4. this wood was less dense, and burning was accompanied with more fly-ash - some char may have been lost up the chimney. The shorter chimney may also have contributed to the loss of fly-ash.

As I inspected the char and crushed it to make sure no embers remained and to prepare it for soil sequestration, I noted that the resulting char powder is extremely fine dust. This is a good reason besides quick fire-extinguishing for adding water to the char - to suppress dust formation during crushing.