Wall thickness rules

The rules for WSF say that the thickness of walls should be greater than 0.7mm. I take it that this distance is generally measured from a polygon, through the model in the direction normal to the polygon, until it hits a polygon on the other side. This is how netfabb measures it, and how the measurements appear in the images Shapeways service send back from a rejected model.

However, I think there are some edge cases when this measurement doesn't work. Here's an example:



This is a triangular prism, and although the prism is quite large, you can find a "wall thickness" measurement that is under 0.7mm.

Here are my questions:

1) Is this triangular prism printable?

2) If it is printable, how big does the triangular cross-section of a triangular prism need to be to be printable?

I may be dense, but would not this "problem" occur whenever two faces meet at an acute angle (an "edge case" indeed,
but not specific to your prism) ?

Yes, you're right, it would. The triangular prism is just a simple case. As it happens the model that I'm trying to get printed has lots of wires that are in the form of triangular prisms.

I have printed literally thousands of 5 to 6 mm wide tetrahedra and octahedra now. What could be an issue
is how/where your prismatic wires meet. (Sharing an edge or vertex is obviously not going to work)

The images with netfabb measurements I'm getting back from Shapeways are measured in the middle of the prismatic wire, nowhere near the joins. I've got an image here with a measurement of 0.38mm "wall thickness", on a triangular prism where the triangular cross section has side lengths of 2.625mm, 2.625mm and 3.464mm.

Hmm. Quite a bit smaller than in your first image. I am too lazy to do the math myself: if you inscribe a cylinder, would its
diameter be safely above the "minimum free wire" limit ?

Diameter of inscribed circle is 1.57mm.

Hmm. You could use that information to argue your case. On the other hand, how big is your model overall ?
Maybe they would not like it even if the wires were cylindrical.
Unfortunately the excerpt from the EOS-sponsored diploma thesis as reproduced on the materials page for WSF does not cover this case. Its full text does not appear to be available online, but searching brought up a link to a similar study (pdf file) performed at the University of Texas . (It has some additional data on gears etc, but still nothing like your model).

They regularly allow airplane wings, but they prefer that the depth of the "edge" of the wing be less than two times the minimum wall thickness.
Note: this is not an official design rule.. it's just what I've found that works to yield the least rejections.

Ask for clarification as to why they have rejected. The cross section sounds more than big enough, and there are any number of models on SW where details are smaller than this.

For example in my model Potentiometer Scale on the '2' and the '7' which are Boolean Differenced from the plate there is a situation exactly like yours, and trust me, it prints well in WSF polished and Alumide.

Henryseg, can you share the image from the rejection email, along with the reason for rejection? This way we can better tell you what needs to be done to solve the issue.

I use the same rule Stonysmith illustrated for airplane wings - for printing in WSF (0.7mm minimum wall thickness) the points on your prism need to be 0.7mm thick (W) at a distance of 0.7mm or less from the point.

Or in this case, the angle between the two faces must be greater than 2*arctan(0.5), or about 53 degrees. For the model I submitted, I now see that the angle is a little smaller, at 48.7 degrees.

Here is the relevant part of the image:



The measurements are both on the triangular prism with triangular cross-section having side lengths of 2.625mm, 2.625mm and 3.464mm.

This was from an email to service@shapeways.com rather than an attempted printing of an uploaded model, because I wanted to check that I had fixed a previous problem. Joost wrote back: "I have checked the attached models but unfortunately still some thin walls (see capture).
Please make the walls at least 0.7 mm thick otherwise the model(s) will get rejected."

As I said above, this reason doesn't make sense to me for a triangular prism. But I see that the airplane wing condition (or angle condition) makes sense. If this really is the condition then the only problem is that I didn't get that information from the email.

The triangular cross-section isn't a necessary part of my design, so I will be bevelling off all of the edges, which should remove the angle problem.

Thanks all!

Chamfers are the better part of valor.