Optimizing CAD designs for strength and filament consumption?

Hi, I was wondering about the logic behind a 3D print. Common sense and the world as we know it would suggest that the less volume a part has, the less material it needs. Also strategically placing ribs and orienting walls perpendicular to the force the part is going to be subjected to and the like make sense to me.

I was first surprised to be told that truss structures in 3D printed parts, counter intuitively, don't work as expected. Also basic shapes like square profile is better than intricate ribbed profiles. In 3D printing one also has to consider support material and total length of the nozzle path when designing anything more complex.

Now to my issue. I am trying to reduce material consumption and also design for strength the following part. Does it make any sense to try the approach of the bottom part of the image?



PS: i have no engineering experience nor can do a CAM analysis, all you see above is my own logic. Nothing very mathematical. I am less interested in the correctness of the design from a structural analysis pov and more interested in how should I think a design from a 3D printing pov. I know that rather than intricate designs one can play with infill, number of shells, etc but I am new to 3D printing world so my best bet is ask the experienced guys.

Edited 3 time(s). Last edit at 06/12/2015 06:25PM by realthor.

Sorry to say, but it doesn't work that way. Printing the lower design will reduce material usage, but it will not straightened the part and will extend printing time.

The way you can straighten the top design will be putting through holes from top to bottom and using a sutil arc on the sides.

Surely you can't talk about strengthening the part without specifying the loads that it will be subjected to?

Let me rephrase so I am sure I understand: holes from top to bottom will increase the number of walls the slicer will build around the hole diameter while reducing somewhat the plastic usage and having a tear shape beam (this is what comes to mind when you say subtle arc from top to bottom) achieves again some plastic economy while resembling a triangle truss right?

Subtle lateral arc like this?


@JamesK: I wouldn't know where to begin. Talking about loads meas talking dimensions, material, etc. I am trying to understand the difference in how a 3D printed structure should be designed differently from a milled/etc similar structure. I am thinking of an I-beam when the forces are vertically distributed along the shape and so on.

Edited 3 time(s). Last edit at 06/12/2015 07:51PM by realthor.

I meant something like the attached sample.

ok, thank you for the effort; i'm ashamed now ... I keep thinking of I-beams and trusses when sketching and try to come up with resembling shapes.

Is this design specific to a 3D printed part or would it be the same if it would be built the conventional subtractive manufacturing way?

Edited 1 time(s). Last edit at 06/13/2015 04:22AM by realthor.

Where lightness (and therefore minimal material) is required in engineering, monocoque designs are often preferred. Think of how light aircraft are constructed. Most or all of the strength is in the outer shell. So for your part, I suggest you try using a very low infill (so that the main function of the infill is to provide support while printing the top), and adjust the number of perimeters and top/bottom layers to get the right balance between strength and economy.

---

Large delta printer [miscsolutions.wordpress.com], E3D tool changer, Robotdigg SCARA printer, Crane Quad and Ormerod

Disclosure: I design Duet electronics and work on RepRapFirmware, [duet3d.com].

Ok, here are the forces that this part will be subjected to:
1) rotation side to side 180 deg, acceleration forces
2) supporting a pretty low weight at the tip.

This part is 150mm in length, 25x25mm square profile as of now, the rotational accelerations end are quite high but the weight they are moving around isn't that much (0.5kg maybe). The piece that actuates the rotation is the small part that goes into the slot on the larger part. I hope I make sense



I have to make sure the thin wall of the slot is wide enough to take the forces generated by accelerating the weight at the tip (inertia of the weight).

This is what a google search came up with. The upper sketch seems what you are talking about:


Edited 1 time(s). Last edit at 06/13/2015 06:36AM by realthor.

Ah, excellent, now we have something to get our teeth into. 0.5 Kg of mass at the tip sounds like quite a lot to move around for the size of the part. (Sounds like a fun project!)

To support the static load at the tip, the vertical dimension of the beam is going to be key, so for that your original model of a thinned beam makes sense - I and T shape beams being classical for cantilevered support. The rotational accelerations put sideways force on the beam, and if the center of gravity of the load is off the beam axis there will also be a twisting force on the beam. You could balance the vertical, lateral and torsional strengths to match the forces by varying the relative width and height of an I beam, but DCs suggestion of sticking with the original shape and treating it as a monocoque seems like the more natural approach given the way the 3d printing software is setup. It's easy to adjust the thickness of the vertical walls separately from the top & bottom walls, and you still have infill % to play with as another variable.

Depending on how fast you want to accelerate the part, I suspect you may need to widen the first half of the part to provide more material around the actuator. Or maybe add a vertical pin & hole engagement between the actuator and the beam so that there's more material to transfer the forces than just the section of overlapping outer wall.

Ok, another pin between the actuator and the beam makes perfect sense and is easy to add.

Now about the monocoque, a beam with 15-20% infill with the original simple design and with a number of shells on the vertical walls and a smaller number of shells on the horizontal front: is this what the general suggestion inclines to?

The sparse infill will somehow resemble the monocoque in itself, would I need to add ribs myself within the structure? I would also like to get the design a little slimmer because there will be a couple more of these in the grand scheme and it does look a bit bulky to me.

Quote
realthor
Now about the monocoque, a beam with 15-20% infill with the original simple design and with a number of shells on the vertical walls and a smaller number of shells on the horizontal front: is this what the general suggestion inclines to?

The sparse infill will somehow resemble the monocoque in itself, would I need to add ribs myself within the structure? I would also like to get the design a little slimmer because there will be a couple more of these in the grand scheme and it does look a bit bulky to me.

By "horizontal front" do you mean the top & bottom surfaces? The easy-to-change parameters in most slicers are the thickness of the top and bottom, and the thickness of the vertical walls. If you wanted to have a different thickness between a side wall and a front wall you'd have to get into modelling the part as a non-solid object.

The infill vs monocoque is interesting. In a true monocoque the forces are all handled by the skin of the object. However, often monocoques are constructed by forming them around a lightweight shape made out of foam or balsa. Obviously the form adds something to the strength of the piece even if for foam vs carbon fiber (for example) it isn't a big percentage. With a low percentage infill it won't add much to the strength and is basically just there as a support for printing the top. As the percentage increases you're going to get an increasing load bearing function. I wonder which is stronger - 100% infill or perimeters thick enough to fill the width?

I think you will need to print a few and see how strong they are in practice. If you can easily support the weight/accelerations you need then you can look at slimming it down like ggherbaz's sexy curved version. You can't beat destruction testing for finding out what a part can do!

Yes I was referring to the top and bottom layers by "horizontal front".

I don't have much experience yet with 3D printing, i'll make the parts at a friend's and I was expecting that multi-shell means that the software is somehow smart enough to create a structure where the out most layer is followed by an infill layer then again another shell and so on. Just how the aluminum honeycomb panels are created incredibly strong. They actually have two skins rendered very rigid due to the core( infill/honeycomb). I now read that many shells means just another skin glued to the first one, making the wall thicker.

At least the slicer should know to offset the next skin by half the layer thickness so that each new layer fits in between two of the previous "skin"'s layers. That would solve some delamination issues or structural failure points.

I think that the design suggestion of ggherbaz would work well with a low infill. If I could only have variable infill density options somewhere in any slicer so I don't design the internal structure myself in CAD.

Slic3r will do it. If you export multiple bodies in the same coordinate system, you can right-click on the first body you load, go to Settings, then load additional Parts or Modifiers, with different settings as you see fit.

Yes I was just reading about this on this page about variable density infill on insoles. . Steeve from GyRobot explain his technique on his G+ account:"It is semi-automatic. I use Gimp and Inkscape to seperate out the regions, I then export .dxf files which can then be imported in to a cad programm (e.g. OpenSCAD). Once the curves have been turned into stls, then its over to Slic3r and using its modifier mesh option, different infill densities can be applied to different regions.".

It is interesting and could totally work for mechanical parts as the needed density variations are predictable and aren't organic in nature, thus requiring few densities that can be saved as patterns.

Hi, I am trying to learn about CAE and reading about open source software to get a CAD model (STL?) analyzed got me nowhere. I have no experience with any CAE at all but I am thinking that a generated heatmap of stress can be turned into a grayscale image which then can pe postprocessed into a Voronoi pattern that would have a denser structure where the map is darker.

Does anyone have any idea how this can be implemented? I know it might be overkill for this particular application but developing an understanding of doing it might prove lucrative in the long run.

My idea is that it should be as easy as save STL, generate heat map, import both in slicer, slicer does the slicing with an infill pattern based on the heatmap (variable density).

What do you think?

ok, just learned that STL is not the best format for anything more than 3D printing or CNC cutting. The steps above would be the same with the difference that instead of a mesh format I would use a solid geometry descriptive format like STEP or IGES. Too bad my currend CAD tool, the freeware DesignSpark Mechanical, can't save in anything but mesh formats besides its rsdoc own format .

I was referring in the previous post at a 2D stress analysis heatmap that would be used to generate a 2D Voronoi pattern and this one would be extruded to fill the inside of the shelled object vertically or horizontally, depending on the direction the designed item will take the most stress in real life, basically replacing the infill. I am not sure that a 3D volumetric heat map can be saved from any CAE package and I wouldn't know what format would that be so that a 3D voronoi pattern can be generated. If that would be possible than one should get a 3D volume of Voronoi cells the same size with the designed item.

Edited 1 time(s). Last edit at 06/17/2015 05:50AM by realthor.

It's a great idea and I'd love to see some sort of opensource tooling in this area. I'm not sure Voronoi is the ideal choice for FDM printers - the resulting shapes tend to be hard to print and don't maximise the benefits of continuous XY extrusion, but some form of stress analysis that could identify areas where material could be removed for minimum loss of strength would be neat.

Yeah, a separate tool would be great. What would be even greater is having one of the open source slicers pick up another open source FEA/FEM software and integrate it, at least the way Repetier host integrates the slicer. Then, based on the FEA, the slicer could optimize the infill.

While the previous would be an ideal situation, I was talking about some way to do it based on what we already have around. We have CAE Linux that could potentially be used to generate a file (heatmap?). Then this file could be used to generate an infill pattern in Slic3r for example via the Modifier IMBoring25 has mentioned.

The way I see it is a cloud solution with slicer and FEA included and an option to either download the sliced model or send it to a 3d printing service. This way the company running it would benefit from clients that pay for the 3d print or commissions from 3d printing services and the casual user with a home 3d printer could get the service for free. Also it is a case of optimal usage of computational resources as opposed to running the analysis on your own computer.

That is unless a collaborative CAE Linux or combo CAE Linux/3dprinterOS that uses a Distributed Computing paradigm for user-aggregated "cloud".

Edit: hmmm, CAE Linux seems to do pretty much what I am talking here, just not as straight forward as I'd like it to be ... have to take a deeper look into that.

Edited 2 time(s). Last edit at 06/17/2015 08:22AM by realthor.

Nothing magic that I can see in CAE . . . any of that stuff should build on any relatively recent Linux distribution . . .

- Tim

It's still kind of magic to me as I have never tried it but this is not the point. People that want to 3D print don't really care about Linux, CAD, CAE, they want an easy interface and perhaps all this FEA should be included in the slicer.

Now back to the subject, is such a method (stress map -> variable density infill pattern) a realistic one to increase the strength of a part?

Yes and no. Yes you increase the strength of the part but only in the plane you are printing. So to make a part strong enough in different planes, your part needs to be composed by more than one print assembled together.

ok, so if I'm understanding correctly you advise for designing several parts (if possible) per assembly, each printed in a plane that you get the most strength out of when 3d printing.

Assuming I got that right, having each such part optimized with a variable density infill would only increase the strength. And if we add strategically placed holes and surface curvatures that naturally increase the strength like @ggherbaz suggested, we might have a "method" for designing better parts !?