I do blacksmithing with my friends. We want to try casting aluminum. I have some questions about this.
A coal forge is not too hot. I’ve never heard of anyone setting aluminum on fire by accident, though I guess you could burn through your crucible.
A steel pot is fine, like if you weld a plate over the end of a short section of pipe. Thicker (like around 1/4") walls are better for longevity. Thinner crucibles might work a few times, at the risk that they would fail sometime unexpected and dump a lot of molten metal somewhere bad.
Soda cans work, but the paint generates a lot of dross and probably isn’t too good for your health. There’s also not a lot of aluminum in a can. I’d go to a junkyard and buy some scrap, like castings from a car. You can break up big pieces (to make them fit in your crucible) by heating them in the forge and then hitting them with a hammer, since aluminum gets brittle (“hot short”) when it’s hot.
You might like “The Charcoal Foundry” by David Gingery. He has a series of books that starts with a home foundry, and makes a succession of machine tools, each one using the previous ones and not much else.
The King of Random has a video on casting molten aluminium. He uses charcoal, not coal. Cans work just fine: the paint on the cans just burns off.
Aluminium burns (= forms Al2O3) even below the melting point. You will have to keep the molten metal pretty much isolated from air.
I would suggest using a cast iron pot with a heavy lid for crucible. And if you do it outdoors, paint, plastic coating and other stuff in cans wouldn’t really be a problem. Just compact them really good before melting!
“Burns” means “forms the oxide in a self-sustaining exothermic reaction”, not “forms a protective layer of oxide and then stops”. There is no need to take special precautions to exclude air, though I guess it couldn’t hurt.
A cast iron crucible is probably okay, but more brittle than steel. If you use cans, then take extreme care not to add water (like a few remaining drops of soda in the can) to molten metal.
I never suggested the reaction were self-sustaining. In fact, it’s most likely not: At ATP, with a normal form of metal (not dust!) you have to add a lot of heat to keep it melting. Whether oxidation stops at a certain point or not, is, I guess, a matter of fluid dynamics and whatnot.
I’m not sure of the mechanism, but I can empirically confirm that nothing bad happens if you melt aluminum in an open crucible. I guess the thin layer of dross/slag that floats on top, and protects the aluminum from air? I could see how theoretically an oxide layer might form at the top, then get broken up by the motion of the fluid, then reform on the newly exposed aluminum, etc., and end oxidizing lots of the metal, but that doesn’t seem to happen noticeably for real.
I’ve played around with this. In addition to some of the advice above, I’d strongly support the idea of using old cast aluminium car parts as feedstock. Aluminium cans or building products like window extrusions don’t cast as well. I assume different alloying components are to blame,and the excess dross from cans.
Empirical knowledge certainly trumps theoretical. You are in charge, TommySeven!
As a teenager, I’d always wanted to make magnalium, a magnesium-aluminum alloy. That one absolutely does need to be melted under some kind of inert gas, or else it burns by any definition. In certain ratios, it’s very brittle, so that you can ball-mill it into fine powder for fireworks with very simple equipment (which is also rather dangerous, since milling in a closed container can produce a powder without any protective oxide or other coating, which becomes pyrophoric…). I could never get a big enough piece of magnesium, though.
If the OP doesn’t want to do old-school sand casting like from the Gingery book, then lost foam is fun: you can make the pattern from normal expanded polystyrene foam (like the white stuff sold for home insulation), and sculpt it with hot knives and other tools. Then embed it in the sand, pour in the molten metal, and watch flames briefly shoot out while the pattern disappears… The finishes and tolerances aren’t quite as good, but less skill is required, and you can make undercuts and stuff.
Google “casting aluminum ant mound youtube” for some practical tips on melting aluminum.
I tried this in about 8th grade. My fire was charcoal stoked with a blower from a vacuum cleaner. my crucible was some junk cast iron cover about the size of large coffee cup and I used various scrap, maybe pieces of lawn mower engines. Melting the aluminum is the easy part. Chunks are better feed stock; thin, airy scrap like pop cans may oxidize into ash and slag before they melt.
My failure (mostly) was not considering moisture in my molds which turns to steam and can ruin your part. Plaster of Paris does NOT work! And I didn’t know how to formulate casting sand at the time. Some people use iron or steel molds.
I’ll try it this summer!
I remelted aluminum cans as part of an “Engineering Day” demonstration in college. It went OK. Aluminum melts at around 1200F, so you don’t need a really hot fire. At 1200F, iron will just begin to glow a dull red if it is dark. In bright daylight, it will be black. So, if you do this outside and your iron/steel container gets red hot, you are probably way too hot.
We decided to use beverage cans so we would know the composition. Aluminum cans are made from grade 3004, which is fairly pure; with scrap from auto parts, you just don’t know.
We used a ceramic crucible and an electric furnace (we had those available). There is no reason you couldn’t use steel or cast iron (although you may get some iron pickup in the melt). Any rust present may be reduced to iron and contaminate the melt. A large amount of rust could be explosive, but if everything is clean, you should be OK.
The cans are covered with a lacquer on the inside (to prevent the contents from attacking the metal). This burns off around 500, but the fumes are not only very objectionable, but poisonous as well. Make sure you have plenty of ventilation. By the smell, you can tell its bad so you really don’t need any special knowledge; you just need to be able to smell. We burned the lacquer off in a separate operation, but I don’t know if that is really necessary.
You will get a bit of dross or slag on top, particularly if you are using cans (because of the large surface area). Aluminum is very reactive to air (regardless of temperature), but the oxide layer is effective at preventing a continuous reaction.
We used plaster of Paris molds, but we cured them in an oven at around 300F prior to use. You need to make sure anything you pour liquid metal into is dry, or it may come back out faster than it went in, and spraying liquid metal is a bad thing to be around.
Of course, wear heavy clothes. A leather welding apron and leather welding gloves are not a bad idea. A plastic face shield and goggles may also be called for. Sure, molten aluminum will go through any plastic as if it wasn’t there, but you are mostly trying to protect yourself from spatter and a two layer defense would probably do that.
One final word of caution. Hot aluminum does not look hot. It looks the same (shiny silver) hot or cold. Always assume it’s hot. Have extra buckets of cool water around and brush up on first aid for burns.
Would PC heatsinks melt better than cans? I have a bunch of old computer parts, and some of my heatsinks are solid aluminum.
That’s interesting. My PP molds steamed, lost definition, and were falling apart when I took my aluminum parts out. I later did some reading and I thought that water was part of the molecular structure of Plaster of Paris which makes it unsuitable. I also read that dental plaster and maybe ever household wall plaster are a different chemistry and resistant to high temperatures.
I once designed machinery for a cast aluminum automotive wheel factory and learned that they couldn’t recycle their own aluminum machining chips because of the entrapped cutting fluid and the air space. Apparently there are recyclers that have an inert gas furnace feed system that keeps thin chips from simply burning up.
I think that once you have a puddle of molten metal you could stir small pieces or thin scrap into it without too much deterioration.
Their larger mass will have to absorb a lot more heat to start melting - you might want to preheat them in direct flame before putting them in a crucible. Once the first one melts you’ll have a nice puddle of molten metal to add more pieces into. Molten metal provides excellent heat transfer. Look up “molten sodium nuclear reactor”. 
Well, our molds were very simple, rectangular shapes intended to represent ingots. I know you can drive the water from PP grind it up and reuse it. I have lost all of the ones I made for my self, but I made one with her HS graduation date cast into it with raised numerals and one I made for my father. I don’t know if those still exist, I’ll have to ask my sister. The one I made for my father, I am not sure about. He passed away a few years ago so I am not sure if I will ever see it again.
And, you are right, after a few uses, they were not usable any more. Honestly, I was not in charge of preheating the molds, but I remember we were advised by our professors to make sure they were dry by preheating them in a 300F oven.