I want to start firing ceramics at home, but the mini-kilns I’m finding on MercadoLibre suck shit. They’re electric and top out at low temperatures like a bit over 1200°. (Maybe this is a consequence of their heating elements?) Although earthenware can fire at as low as 1000°, temperatures in the 1350° range are needed for many other ceramics.
For example, in July, http://articulo.mercadolibre.com.ar/MLA-618696798-horno-para-vidrio-ceramica-gres-box-1-_JM was a 1200-watt kiln with a 160 mm × 160 mm × 115 mm inside space (2.8 ℓ) that cost US$500. And it only reached 1200°, but weighed 6 kg.
I was thinking that maybe a gas-fired kiln would work better, maybe fueled by a portable LP gas cylinder and with regenerators to keep noxious exhaust to a minimum.
I want to beat that shitty electric kiln handily on all axes except weight: energy usage, volume, temperature, and cost. I’m okay if it weighs a lot more, though. Let’s shoot for a 320 mm cube inside capacity, 1400° max, cost under US$200, weight under 100 kg, and use under 600 W.
(See files Millikiln and An electric furnace the size of a sake cup for versions of this with further reduced scope.)
The gas heater we use to heat the living room is 8000 BTU/hour, which is 2300 watts. The little electric space heater I’m using to heat my office is 2000 watts. Suppose 2000 watts is a reasonable power envelope and we need to be able to heat at 300° per hour (0.083 K / second) once the ceramic is dry. Then the maximum thermal mass in the kiln is 24 kJ/K. The specific heat of water is 4.2 kJ/K per kilogram, so this is about 6 kilograms of water; other substances like quartz have lower specific heats like 0.83 kJ/K/kg, so it’s in the neighborhood of 30 kg. This is pretty close to the amount of stuff you’d like to have on the inside of a kiln’s insulation.
(Engineeringtoolbox lists specific heats of .70, .73, and .83 kJ/K/kg for different forms of quartz.)
This implies that you need pretty high efficiency; we can’t afford to just spew out the vast majority of your heat in an exhaust and expect to get the kind of temperatures we need.
How much energy do you need at a minimum? The minimum would be when you heat your kiln up inadvisably fast. At 300° per hour (83 mK/s) you can reach 1200° in four hours, which would be 2000 W · 4 hours = 29 MJ. LP gas is 46.4 MJ/kg or 26 MJ/ℓ, so this is 630 g of gas, occupying 1.12 ℓ. A regular 10 kg gas cylinder would power 15 such firings before refilling, a cost of about US$6 per firing.
This doesn’t take into account the heat lost while the kiln is hot, which depends on insulation and regenerator losses, or the energy lost in boiling off the water, which is about 2.3 kJ/kg and thus relatively small in this context.
Using regenerators in this context involves feeding the flame with air preheated from the regenerators. LP gas burns at 1970° in cool air, which is hot enough, and presumably this temperature increases as the inlet temperature increases. This means, though, that the chamber where the flame is happening will need to be able to withstand such temperatures. Some amount of exhaust gas recirculation may reduce the load on the regenerators and lower the flame temperature.
You probably don’t need to run the flame the whole time; the autoignition temperature of LP gas is around 400° or 500°, so you should be able to turn the gas back on and ignite it just from the walls of the burner chamber.
You need two regenerators because you need to pull the intake air through one while running the exhaust air through the other, periodically alternating direction. The use of a recuperator instead of regenerators would avoid this necessity, conventional (non-fractal) recuperator designs need to be fabricated from high-thermal- conductivity materials such as copper or aluminum, which can’t withstand such high temperatures. (Copper melts at 1085°.)
What regenerator material can withstand 1350°, though? Lots. Alumina melts at 2072° and even quartz doesn’t melt until 1670°, even though quartz-based felsic lavas and well-fluxed quartz-based glasses melt at much lower temperatures. Regular fireclay is good to 1515°. So a pebble bed of ordinary fireclay should work fine. Limestone would calcine to quicklime, which would also work, and would be a useful product in its own right, though it might be an unnecessary hazard. Quicklime could also perhaps eliminate some acid contaminants from the exhaust gases, such as HCl from salt firing.
Olivines are not suitable for the regenerator itself; although you might expect them to be much more refractory due to their role in raising the melting temperatures of mafic rocks, not to mention their foundry use as refractory sands, not only do some of them melt as low as 1200° (fayalites, though forsterites melt at higher temperatures up to 1900°), but they oxidize exothermically to quartz and magnesite when exposed to hot CO₂. This might be useful to get a zero-carbon-emission gas-fired kiln if the reaction is fast enough; the exhaust air could contain less CO₂ than the input air.
If the alternation time on the regenerators is 1 minute and the advection reaches 2000 watts (that must be retained), each regenerator must hold 120 kJ. If the temperature swing in the regenerator averages 600° (more for the innermost chunks, less for the outermost), then it would be sufficient to have 250 g of quartz per regenerator at 0.83 kJ/K/kg. This is very promising!
Let’s suppose that we want the kiln to stay at 1350° for 12 hours without losing more than, say, 10% of the heat energy through its insulation, so that we run the burner at about 10% power to keep it warm. Earlier I said the inside should be a 320-mm cube, which gives it 0.61 m² surface area. R-values are m²·K/W, and in this case we want to lose under 200 W through 0.61 m² and a 1330-K difference, which means our R-value needs to be at least 4.1. Fiberglass is about 22 m·K/W, so that’s about 190 mm of fiberglass. Loose-fill perlite is about 19 m·K/W, loose-fill vermiculite can be as low as 15 m·K/W, so you might need as much as 275 mm of vermiculite. This is eminently feasible, although it doesn’t leave a lot of extra room.
NIST’s data http://ws680.nist.gov/bees/ProductListFiles/Generic%20Fiberglass.pdf says R-13 fiberglass batts (i.e. 13 m²·K/W, which is 3.2 times what we need) are 89 mm thick and weigh 12.1 kg/m³ (i.e. 12.1 g/ℓ). This works out to be 146 m·K/W, which is 6.6× as high as the value I calculated above and twice as high as the values I find online for silica aerogel, so it cannot possibly be correct.
Ah, this random CertainTeed™ brochure http://www.certainteed.com/resources/IGProductKnowledge200608.pdf explains that “R-13” is actually 2.3 m²·K/W, giving a thermal resistivity for fiberglass of 25.8 m·K/W, which is correct. So we can probably take the density as correct.
Using such batts (minus their facing layer, I suppose) to achieve the insulation level needed would require two layers of them, providing 4.6 m²·K/W of insulation and increasing the dimensions of the kiln by 178 mm on each side, for a total of (178+320+178) mm = 676 mm. The total mass of the fiberglass then would be 3.3 kg, entirely reasonable.
But you can’t actually use just those batts, because the innermost layers of insulation need to withstand 1400°, and glass won’t. Glass will only handle up to 500° or so, so only the outer reaches can be made of glass. You need inner layers of kiln bricks or vermiculite or something, and that’s going to drive up the weight considerably. (Or you could use ceramic fiber batts, which look like glass fiber batts but are made of alumina to withstand high temperatures.)
To observe the inside of the kiln without losing a ridiculous amount of heat, although peepholes are the traditional solution, you might want some kind of optics. I have no idea how this could work; wouldn’t a mica window melt? Even if it doesn’t melt, could you see through it while it’s glowing?