Aluminum Casting Parts

Hello Everyone, I just want to share some of the shenanigans I’ve been up to on the off season.

I got obsessed with making a one piece gearbox housing that could bolt to a section of 2x1 box section. Not because we really really need it but more or less to go through the process:

First I started by figuring out a furnace:
I Did some super crude tests with my students regarding how hard it would be to melt aluminum: Result even with the shadiest furnace ever aluminum melted very easy with propane and way faster than we all expected:

I was pretty grossed out by some of the furnaces on Amazon with Kaowool insulation and wasn’t too interested in having the thin fibers of it going everywhere. I found an old empty R134a tank and some castable refractory from Skyline Components in Tucson, AZ. We cast a small furnace. It was pretty easy to work with and has held up really well to multiple firings, no degredation and minimal cracking:

For my first test I made a really simple bracket with curves and shapes that would even be a pain to make with a CNC mill. I cut parts out of foam on my router in 2 halves. (cutting these parts took the longest out of this whole process:

This first bracket was cast using green sand. It was alright. the weather was pretty dry so i was struggling with the sand drying out and things kind of falling apart. This first test I went pretty fast and didn’t sweat so many details. I totally expected it to be a failure. But to my astonishment it worked… even with such small passages (though barely):

I tried the lost foam method in green sand. It was a lot of work with few good things:

  • The foam isn’t exactly my favorite thing to machine as it is very flimsy
  • Surface finish is garbage
  • Tons of nasty fumes probably took a few years off my life
  • The positive is consumed
  • The aluminum didn’t penetrate very far

I tried cutting the positive out of a chunk of HDPE. Throughout this whole time i’m constantly tweaking the design here and there trying to make it more reasonable for casting. I initially had an idea of having a center positive then spacers that I could put on either side to make it a right or left gear box. I learned the hard way about ALWAYS having draft on the edges I wasn’t able to get a cast from a positive with straight sides. I thought that it was so thin i could probably get away with it… nope.

At this point ive spent a lot of time routing out positives and way more time than i would like trying to cut out the spacer sections that lock onto the aluminum tube. All i was getting was garbage then something happened…

Let me stop here for a second…

Our team first got a 3D printer in 2015 and for the next 4 years I’d come to loathe 3d printers. I had a really hard time justifying putting so much time into a machine that wasn’t cranking out durable parts and that someone would be kind enough to “re-calibrate” every week or so. I hate them… I mean i still kinda hate them but i’m warming up… to them… maybe with an enclosure i’ll warm up much fa… (stop) :blush:

ok back on track…
One of my students had recently pulled the moth balls out of one of our 3D printers and offered to 3d print the part I was struggling to cut cleanly with a router (Anyone else out there do front side and back side 3d contours in 2 ops on an x-carve? yes it does work you just have to plan steps ahead for your Z0) I was schooled the 3d printed part worked perfectly.

One trick I came up with along the way is to make a thin wall where you want holes. In this first successful casting i made my life a lot easier by not packing Petrobond through the entire frame. the sand just formed a 0.030" void between the parts that the aluminum just didn’t flow into:

Holes were then chased through on a mini mill but you could probably use a drill press. I intentionally gave the task of drilling it out to a super new member just to see if it would still work. Fit was alright. A smidge tight for the lower motor but bringing the bolt holes up from a snug fit to a free fit fixed it:

Yes there are 4 holes that have to be in just the right spot on the box section to make this work. We plan on making a template that we can just clamp to the metal and chase the holes in with. we’ll start with an 1/8" pilot then just chase it with a rotabroach hole saw.

So long story short a 3D printer and casting old aluminum off cuts is pretty promising:

I made these pretty key observations:

  • Scale up your model by 1.017 before making the positive
  • Stay away from aluminum cans there is a lot of paint and plastic in them
  • There is a lot to casting I did not cover in this post intentionally, do you homework and find someone with experience
  • Oil sand (petrobod) is great for very dry regions
  • Don’t even dream of holes just put in spot to drill through
  • If you are planning on drilling it out make sure some feature will fit in a vise.
  • It’s actually pretty easy

This post is kind of a mess I hope it’s still okay.
Oh also cad files here:

Next… Soduim Silicate and Generative Design… is anyone else running down this rabbit hole?


This sounds really interesting.
A few benefits I can see right off the top:

  • BOM cost either low or non-existent if casting done by team members - Not sure how the aluminum scrap would count for this though
  • Allows teams to actually use the ridiculous amounts of scrap cncs generate
  • Allows teams without access to high end 3D printers to make parts that are near impossible to CNC

Super cool - I love these projects that lead into whle series of discoveries because you learn a ton.

A couple of years ago, I made silver bullets loaded cartridges for some of my friends for Christmas - you know, Lone Ranger, werewolves, vampires stuff.

I started in July and worked on it on the side all the way through Christmas; learned a lot during the process (including what is takes to melt 100% silver for starters). I’ve thought about some casting opportunities and was considering using plaster and with PLA as a lost wax sort of process.

Casting is harder than it seems - kudos for such cool results.

One tip from machining similar cast and forged parts - pick a good reference point for finish machining. Something that can be definitively located and resolved well enough to be able to reference the rest of the machining. Getting that first hole drilled or the first surface cut sets the reference for everything that follows so your advice about making something you can fixture in vise (or whatever) is a big deal towards being successful and repeatable.

Super cool project - I get sucked into the same sorts of things - it’s not about whether you should but about whether you can :wink:

What sort of strength do you expect to get from the cast material?

We’ve been playing with similar projects in our shop. Our high school still has a foundry and we have a lot of leftover aluminum scraps that make for some interesting experiments. Our biggest issue has been determining good draft angles with 3d printed patterns for a clean release. That and that our sand is used by high school classes so it is pretty well worn out for bonding to itself properly.


The strength is definitely lower due to a lot of voids and imperfections in the material, that and the fact that we mix alloys without much for discretion…

In a low load part like a hub this seems to work fine and they hold up just as well as a purchased part would. The only downside now is that AM lowered their prices to where this process costs more in time than it could ever be worth!

1 Like

Super cool project, thanks for sharing. Love seeing ideas that most people would immediately dismiss come to life.

How did you attach the 3D printed parts to the HDPE chunk to prepare for casting?

Looking at your problem with Lost Foam Casting I think that could have been solved with an entrance and exit port.
For example this video:
But I would love to try casting part and it is really cool that you tried it.

I’m super happy you asked!

From Machinery’s handbook:

(these would be best case, it may be possible that some of the alloying is lost or the metal picks up other garbage along the way, or pits, grains and inclusions.)

6061-O properties which is in it’s softest annealed state:
Ultimate tensile stress of 18ksi Yield stress of 8 ksi. if we manage to get good enough to put a T6 temper on it with an oven then ultimate jumps to 45ksi Yield to 40ksi (so probably worth it)

If we use 6063:
as cast 13ksi ultimate, 7ksi yield
with a T6 temper 35ksi ultimate and a 31ksi yield

This is the part i didn’t know until you asked and i looked it up:
if we use old andy mark chassis:
the Chassis is 5052 aluminum I don’t know what temper but in all cases the 5052 has a lower strength than 6061-T6 However in an annealed state it is actually stronger than 6061 with an ultimate of 28 ksi a yield of 13 ksi and both in hardness and shear strength it beat annealed 6061 and 6063.


The 3d printed parts had location pins pressed into them. the hole in the HDPE is drilled out a few thou over so its a nice easy fit. I actually kind of switched up my strategy and ended up 3d printing the entire half of the mounting side with the spacers in it as one part. (See the Grab Cad Link) I think it’ll be easier to keep the spacers in the correct orientation when packing. The original way i did it i felt that the sand was pushing them in or out as i was packing them together. If they are integrated as one part i think it’ll be easier to keep them in a more precise position.

I kinda didnt go to crazy talking about how i went to a fully 3d printed positive because I havent actually tried casting with it yet.

  • I think if we use this method we’ll count the cost as remnant cost which is about $2.79/lb which would make this part less than $5
  • Making use of a huge bin of scrap aluminum bits is how this whole thing got started. i could see a future where we are asking other teams for their scrap
  • i think even an entry level 3D printer is going to work fine. you are really not asking the part to do much and even if the print volume is too small and you have to glue/pin several together it is still going to cast the same. My team has an Ultimaker and a lulzbot. i bought a Prusa i3 Mk3 and it is lightyears easier to use than the other two and much cheaper!

Please post the results once you’ve completed a cast with them. I’d love to look at the different results. I have a few more questions if you don’t mind.

What thermoplastic for the side positives did you use? Basic PLA or ABS or something else?

How much did it cost to make the setup for the casting process?

Any plans to use something like this in the main season? (I know it’s not very practical probably but for the cool factor)

The white sides were made in ABS but for no other reason than we have a bunch of ABS filament. I’ve since made some more positives in PLA. I think all that really matters it that it is easy for you to print.

The whole casting set up i would estimate to be around $800 bucks. There are a lot of things i learned along the way that could have made it cheaper or easier. I’d like to refine a few more things then if there is interest i could make a big post just on the build and operation of a furnace.

I would argue that there are moments when the casting process is actually more practical than other manufacturing methods. For the most part a 3d print will show up overnight and there is not time spent developing CAM tool paths or creating new set ups on a machine, watching the first and only time you run through all the tools or even any time spend finding or buying the correct size stock if it is even available. The packing and casting part of it takes an hour at most (disappointingly fast). so for a complicated 3d part a team with a 3D printer and a drill press could have a part before the team with the Haas. It kind of hurts to say that because i loathe 3D printers and I love flipping parts on a Haas. (granted the tolerance just won’t be there but probably close enough)… so anyway yes I will be keeping an eye out for an opportunity to eliminate part count through casting this year… and hope to prove out lost PLA casting over the x-mas break. If the lost PLA works i will freak out.

We had success with lost PLA, but you will need a lot of venting and also you will need to pour hot and from multiple sprues quickly. It actually looks like our lost PLA went a little smoother than your lost foam pour. Burning out the pla with a furnace would yield much better results than the direct pour we tried.


Alloy 5052 is not hardenable by heat treatment. It can be significantly hardened by cold work (eg by cold rolling) …

Cool! What was your gating system like? Seems like there are portions in your lost foam where having additional gates and sprues could have been helpful in getting a more complete cast. If you can, maybe do something a little more traditionally lost-wax and dip the pattern into a fine sand slurry so you can get a more desirable surface finish.


5052 Can’t be heat-treated, only cold-worked. Typically sheet metal products are H32 condition, which is reasonably strong.

Looks sketchy, but fun!

I’ve been working in an aluminum billet casthouse for the past year and gaining a greater respect for how much work goes into the precise alloying, filtering, and heat treating of aluminum to achieve the material properties we see in some of the higher strength alloys we use in our airplanes, spacecraft, and robots. It’s not something that can be done easily on a small scale.

Seeing the gearbox casting you made fills me with wonder and paranoia! Love the concept of casting gearbox cases, but wouldn’t recommend the cheesing method that we usually see on teams’ gearbox plates. If you can make any shape you want, make a shell with ribs near high stress areas instead. You can make up for a drawback in material strength with your ability to make stronger shapes with your mold design.


That is the REAL thing! making something from nothing. AWESOME. I’m impressed and inspired. well done