Towards Self-Supporting Design for Additive Manufacturing: Part 1 (Standard Guidelines)

1. Background:

When it comes to Additive Manufacturing (AM), there is a lot to consider before hitting the print button. One of the biggest constraints in most AM processes is the need for supports for overhangs, which are aspects of the design that will not print properly without supports either due to the force of gravity acting on the material (natural free-falling state of the material with no support forcing it into position), or the thermomechanical effects associated with printing with no underlying thermally conductive and warpage-constraining material.

The solution is to either redesign any of the problem areas or reorient the whole piece to avoid any overhangs that need these supports. During my internship at PADT Inc., I will be focusing on strategies to minimize the need for supports, towards the ideal goal of manufacturing only self-supporting structures, because it’s never a bad idea to decrease waste, both in terms of additional material used and the labor involved in removing the support materials after the print. This post (part 1) of this blog series is going to be about evaluating the most basic guidelines of printing a self-supporting structure to extract some insight.

2. Methodology:

dfy5 e1499451751414

Using inspiration from some machine accuracy tests found online, I designed my own prints to evaluate the Makerbot Replicator 5th generation’s ability to print overhangs using angles, upright holes, bridges, arched bridges, and 90 degree overhangs—and I present each one of these standard guidelines below. My process parameters for almost all of the tests with, of course, supports OFF were as follows:

  • Extruder Temp: 212 C
  • Travel Speed: 70 mm/s
  • Infill Density: 10%
  • Layer Height: 0.20 mm
  • Number of Shells: 2

3. Observations:

3.1 Angles

For testing overhangs with angles, I printed out two different sets of trapezoids. The first was a set of six ranging from 25-75 degrees (or 65-15 degrees from the leveled plane).

 19250930 683601601828737 529593455 n  19415712 683601531828744 1416379942 n

  19427421 683601585162072 162636770 n  19398577 683601595162071 270826606 n

As shown by the photos above, the prints were of good quality and only started to show visibly poor quality on the 65 and 75 degree samples. The thinnest edge on the 65 degree sample curled up due to the heat of the extruder. The same issues were present on the 75 degree piece, but this is more exaggerated because of how harsh the angle is.

 19415832 683601608495403 1190936473 n  19427562 683601621828735 2033462656 n

My hope of printing self-supporting pieces was shattered when I printed out an 85 degree trapezoid. To save material, I only printed out a section of the trapezoid, but the angled edge did not print smoothly at all. Not only that, but it did not print at a true 85 degree angle. With these tests, it is safe to say that a machine can handle up to a 65 degree angle with light finishing needed, but further experimentation can be done to see if these angles can be improved.

3.2 Upright Holes

  19416067 683601628495401 218953062 n  19427642 683601631828734 1791084263 n

For these, I did 2 quick tests. The first was printed with the settings listed above, and the second was printed with only one shell (contour). The numbers next to the circles (1, 2, 4, 6, 8, 10) represent the radii in millimeters. The double-shelled print came out a lot better than the single-shell replica on the edges of the piece, but the single-shelled piece had slightly cleaner holes due to less weight on the overhang. However, both pieces had defects that can easily be sanded down.

3.3 “H” Overhangs/Bridges

 19250970 683601671828730 791432500 n  19415658 683601675162063 1587093843 n

Bridges are sometimes referred to as an “H” overhang due to the overhang having two sides to support it. When testing bridges with 90 degree overhangs of 0.25, 0.75, 1.25, 1.75, and 2.25 inches, the results showed increasing stringing with length for all but the 0.25 inch sample.

3.4 Arched Bridges

 19398917 683601645162066 1118307867 n  19433668 683601665162064 1466245648 n

The inspiration for these came from the shape of an egg. That’s because I learned during an egg drop lab that an egg is stronger when weight is being put on it length-wise than if the sides are pinched. As expected, the pieces where the curves are less steep (like an egg laying so the shorter distance is perpendicular to the ground) have more defects, and the steepest curve (as if the top of an egg was the mold for this piece) was almost perfect. The wider the curve becomes, the less it can support itself and the more the piece is unrecoverable.

3.5 “T” Overhangs/Cantilevers

 19433545 683601678495396 1791483513 n  19427867 683601688495395 1883780102 n

The final test for this section is the “T” overhang, which only has a support on one side. This happened to be the only test that completely failed, as none of these pieces are usable – it’s safe to say that pieces should not be made without supports on both side of the overhang.

4. Insight

A rule-of-thumb “overhang rule” used in the industry is that a piece can be self-supporting as long as the overhang does not exceed the angle to the horizontal by more than 45 degrees. A back-of-the-envelope (literally) calculation shows that if we approximate an angular edge with stair-steps of thickness t, the overhang length l equals t/tan(Θ). According to this equation, this means that to increase the allowable angle, the layer thickness can be increased or the unsupported length should be reduced.

mathpic scaled

This observation is confirmed by a previous investigation into the angles of self-support for ULTEM-9085 on Stratasys Fortus systems showed how the maximum angle that can be self-supported is indeed a function of layer thickness, but also a function of the contour width (see graph below). In the graph, the lower the angle, the lesser the support needed, since everything above that angle will need to be supported. Thus, thicker layers result in lesser support. Due to the nature of contouring in the FDM processes, a thin contour that forms the edge of the overhang is likely to droop off. But as it gets thicker, it maintains greater contact with the supported portion.

angle width3

The fact that thicker layers and contour widths may yield larger support angles is counter intuitive since we generally assume thinner layers improve print quality – and this is in general true. But if the aim is to design parts without supports, both these variables can push the limits of the process.

5. Conclusions

Basic design guidelines for overhangs can be, to a first order, simplified to one design rule: the angle below which material needs to be supported. This angle in turn, for the Fused Deposition Modeling process on a given machine and material, can be optimized by manipulating layer thickness and contour width.

In my next post, I will look for inspiration for self-supporting strategies from other disciplines. Stay tuned.

Share this post:

Upcoming Events

Jan 07
- Jan 10
, 2025
CES 2025
Jan 22
- Jan 25
, 2025
Arizona Photonics Days 2025
Feb 04
- Feb 06
, 2025
MD&M West 2025
Feb 04
- Feb 06
, 2025
2025 Transformative Vertical Flight
Apr 07
- Apr 10
, 2025
40th Space Symposium

Search the PADT Site

Contact Us

Most of our customers receive their support over the phone or via email. Customers who are close by can also set up a face-to-face appointment with one of our engineers.

For most locations, simply contact us: