Week 8 - 3D machining, moulding and casting

Designing the mould

This week I wanted to try and make a mould to cast a octohedron. I've been working a lot with this shape at NASA where we have been using them to create cellular composite structures with incredibly high specific stiffness. Initially I tried designing a 12 piece mould consisting of two halves for the exterior shape and 8 internal slots that would be removed after moulding.The following images show how I developed the mould design using Rhino.

After talking with Sam about the design we came up with the idea of a mould that you could fold up to cast the piece. The following images show how I came up with the negative form for the final mould which I aim to make from oo-moo and then how I created the positive form.

Machining the mould for the mould

I had to use foam to machine my design from as it was too big for the wax pieces provided. I use the Shopbot to carry out two tool paths, the first was a roughing cut using a 1/4" ball nosed end mill. The second was a finishing cut using a 1/8" ball nosed end mill. I didn't realise at the time but the 1/8" end mill was unable to machine the majority of the pyramid plugs. I decided to continue anyway as the edges that were machined would suffice in blocking off material from getting into the central void.

Finishing the mould surface

I removed excess dust and pieces using a file and a chisle. Then I decided to try out using Gesso to seal the foam and improve the surface finish. I applied it using my fingers, given the output from the Shopbot I was pretty happy with the finishing techniques that I used. I then sprayed it with a release (more info below).

Pouring the mould and making the part

I sprayed the mould with a release and used oo-moo, a silicone rubber produced by Smooth On as the mould material. I later casted a second mould without release and it was far harder to remove, therefore I would recommend using release on foam moulds. After removing the mould from the foam, I used scissors to get rid of the flashing around the edges and then used rubber bands to hold it together. Initally I tried using hot glue but it did not adhere to the silcone sufficiently well to hold it together. I did however use hot glue to seal the seams ahead of pouring.

I poured Hydrostone for the first test of the mould. Some of the hot glue didn't seal completely so I had a few leaks. The Hydrostone only took about 30 minutes to set. Upon demoulding I discovered that some of the material had made it into the internal cavity, luckily this piece was small enough to be removed.


The surface finish is very accurate in terms of the foam surface that it came from, however it would good to remake the mould from wax in order to get a smoother finish. When redesigning the mould I would add square plugs to create flattened corners that would allow multiple voxels to be assembled into a larger structure. Also, if I machine the mould again I would smaller tool to properly remove the material for the triangular plugs.

Plus a bit of failure

I tried to cast a resin voxel, the pour seemed to go well however upon return I discovered that the exothermic reaction of the left over resin setting caused the cup to melt. Luckily it didn't get through the surface below. In addition, the cast itself bubbled at the surface of the pour so the voxel didn't come out very well.