Tensioning 3D Prints For Lightweight, Strong Parts

Desktop 3D printers have come a long way over the past decade. They’re now affordable for almost anyone, capable of printing in many diverse materials, and offer a level of rapid prototyping and development not feasible with other methods. That said, the fact that they are largely limited to printing different formulations of plastic means there are inherent physical limitations to what the machines are capable of, largely because they print almost exclusively in plastic. But augmenting prints with other building techniques, like this method for adding tensioning systems to 3D printed trusses can save weight and make otherwise unremarkable prints incredibly strong.

The build from [Jón Schone] of Proper Printing consists of printed modular sections of truss which can be connected together to make structural components of arbitrary length. To add strength to them without weight, a series of Kevlar threads are strung from one end of the truss to the other on the interior, and then tensioned by twisting the threads at one end. Similar to building with prestressed concrete, this method allows for stronger parts, longer spans, less building material, and lighter weight components. The latter of which is especially important here, because this method is planned for use to eventually build a 3D printer where the components need to be light and strong. In this build it’s being used to make a desk lamp with a hinged joint.

For other innovative 3D printer builds, [Jón] has plenty of interesting designs ranging from this dual extrusion system to this 3D printed wheel for a full-size passenger vehicle. There’s all kinds of interesting stuff going on at that channel and we’ll be on the edge of our seats waiting to see the 3D printer he builds using this tensioned truss system.

28 thoughts on “Tensioning 3D Prints For Lightweight, Strong Parts

  1. Pre-stressing doesn’t change the Young’s modulus of the material, so the beams become more bendy as you remove material. The compression doesn’t help with stiffness in any direction – the Young’s modulus of the material is the spring constant, which stays constant no matter the tension or compression within the proportional load limits. What also happens when a beam is compressed lengthwise, you bring the structure closer to the critical load for buckling.


    I’ve never heard of trusses being compressed lengthwise along the neutral axis, which does not help the stability of the structure at all – actually the opposite. It becomes non-stable against buckling and actually prefers the bent position. Pre-stressed concrete is always done on the tension side to balance the weight of the structure and prevent cracks from opening up.

    1. That said, there’s a point in compressing all those truss sections together to keep the seams under compression at all times, but the string should rather go through the corners, not through the middle of the whole truss, unsupported.

    2. Yes the prestressed concrete comment just confuses things. But if you think about it, these trusses are not being used in the standard way concrete columns are, which is compression. These are being used to prevent any deflection or Bending of the trusses by keeping the separate sections together. The Kevlar line does not really “increase” the strength, but really to create the strength. They wouldn’t work without something like them. But yes I agree with the guy below saying they should he anchored at the corners

      1. >These are being used to prevent any deflection or Bending of the trusses by keeping the separate sections together.

        But the tension string doesn’t do that because it’s an unstable configuration. First of all, it goes right through the neutral axis where there is no stress under deflection, so it contributes nothing to the stiffness of the beam. Secondly, when the beam does deflect, the string moves off the center and starts to pull the beam even more into the bend. There’s no centering force: the stressed beam _wants_ to bend and it’s only waiting for something to disturb the delicate balance of internal forces.

        That’s why end-to-end compressive loads on beams cause buckling when the force exceeds some characteristic critical load: the beam will go banana and collapse on itself out of any small disturbance.

        1. Considering how hard it is to make a perfectly symmetric object, the beam is already bending when the guy puts tension on the kevlar string.

          If you put the beam on a flat table, you can turn it and find which side is concave or convex, and then bend it to make it switch to the opposite shape. That’s what it’s going to do when put into a 3D printer frame where the forces change direction – it has different stable shapes and it switches between them when disturbed with enough force.

    3. Its not the same modulus as just the plastic pieces though, its now the plastic pieces plus a tiny section of kevlar. Like you say though its in the worse place for helping with bending, 3 strings at the extremities would at least mean you couldn’t bend it without stretching the kevlar and actually bring it into play. The copper tubes are probably doing as much for it as the string is.

      1. The string does nothing as long as it stays in the middle, because it does not stretch when the beam bends. The beam is the longest, measured from end to end, when it is straight. The string instead is trying to make the beam not straight by bringing the ends together by its tension, so it works WITH the bending load, not against it.

    4. When you have high tensile cables and weaker/’squishy’ beams, it might be better to think of it as a tensegrity or tensairity structure rather than a prestressed beam/slab. The goal then is to only have enough non-cable material needed to keep all your cables in pure tension.

  2. It’s an interesting exercise, but at the end of the day, it’s doomed to failure because of the added stress without strengthening anything. Maybe this is just a dry run for modeling in PLA but ultimately using another material like aluminum or other metal blend.

    1. Well, it does use brass rods at the corners for actual mechanical strength, and the truss is merely acting as a spacer between them.

      It’s the brass rods that are in the most stressed position in terms of external forces. The Young’s modulus of plastic is something like 1-2 GPa while brass is 110 GPa. The rods are a hundred times stiffer than the 3D printed material anyways and when placed like that, they carry 99.99…% of the load. Whether the plastic pieces are compressed together or not has really no significant contribution to the stiffness or strength of this structure.

      1. And, since the rods appear to be merely press-fit into channels in the truss pieces, there’s nothing preventing the rods from sliding lengthwise, and there’s nothing transmitting the forces from the truss pieces to the rods except a slight bit of friction.

        It means, if you apply torsion to the beam, it should appear as a shear force between the truss pieces and the rods, but since they slide against one another the whole thing is more or less free to twist into a screw, until the geometry changes enough that the pieces lock up or the rods bow and pop out of the channels.

  3. I hate to say it folks, but this is where we are headed; sympathetic media people will be presenting “designs” with implied improved functionality beyond visual appeal, and everyone will eat it up. I wouldn’t be upset about it, if it wasn’t for the infered expertise which is clearly lacking.

  4. like other people said, i think this isn’t very structurally sound.

    but the other thing is, it just doesn’t make sense. or anyways, it’s not how i use my 3d printer. for a truss / rod / column, the hardware store has a wide variety of dowels, pipes, extrusions. these things have been readily available since before i was born. the 3d printer revolution is fine control of precise shapes. brackets that mate a dowel rod to an aluminum extrusion, with screw hole mounting points for a servo, and holes for a bearing axle, all aligned at arbitrary positions and angles, with some decent (but wanting) degree of precision.

    the 3d printer is bad at trusses. they aren’t very strong, and they take forever to print, and you wind up constantly bumping against your build area limit. i use mine for what it’s good at.

  5. I fully agree with the other comments here that linear tensioning does nothing to add actual strength, and if anything it makes is worse if there is any bending at all. On top of that, many plastics suffer from “cold flow” under continuous stress. It might take a while depending on the plastic, and PLA might be better than most, but I’ll bet his lamp stand collapses before the bulb goes bad.

    Cute isn’t necessarily clever.

    1. heh from my experience, PLA is pretty bad about cold flow. i don’t know how other plastics are, but PLA is awful. from my experience with legos, i’m guessing ABS is *much* better.

      i print something with a little tension (like a friction-fit sleeve around a consumer product), and after a month at the most, it is loose. not just bent out to be conformal, but actively loose.

  6. Hold up – there’s also brass rods running through this which the article doesn’t mention, so what’s actually doing the work here?

    If the 3D printed stuff is just there to hold the rods then this is a massive waste of plastic besides all the other good points raised in the comments about the “engineering” of this.

    Seriously HaD you could at least run stuff through a basic sanity test before giving it the credence of publishing it. Or are we about to get a load of “free energy” videos?

  7. You’re missing the (physical) point of the tensioning which is twofold:

    1. It makes it so you can print small parts and turn them into something BIG using just one single fastener (the tensioner parts) instead of using fasteners between each part.
    2. It changes the “constant to worry about” from the flexural modulus (tendency for a material to resist bending) to Young’s modulus (modulus of the elasticity in tension).

    For a material like PLA you’re not gaining much by using tensioning since it’s already quite stiff. However, by using something like PETG you’d be turning a bendy-ish sort of material into a reasonably stiff/rigid material by keeping it under tension at all times.

    1. 1. You don’t need much tension for just holding the parts in place, and in fact it would be far better to glue them in place.

      2. The flexural modulus results from the Young’s modulus when you calculate the deflection of the beam.

      “For very small strains in isotropic materials like glass, metal or polymer, flexural or bending modulus of elasticity is equivalent to the tensile modulus (Young’s modulus) or compressive modulus of elasticity. However, in anisotropic materials, for example wood, these values may not be equivalent.”

      Note that for large strains (over 0.2% in common metals), you’re typically going into the plastic deformation region of the material where the strain is no longer proportional to stress and the Young’s modulus doesn’t describe material behavior. It’s also the region where your design for rigidity or stiffness has already failed because the part won’t return back to its original shape when the load is removed. Plastics generally don’t have a definite proportional limit because of creep under static loading, so the elastic modulus only applies for dynamic loading.

      1. The Young’s modulus, or the flexural modulus, and a bunch of others, are merely different names describing the same material property under different circumstances when dealing with isotropic materials.

  8. I built a 2M high Kossel using 2020 aluminum, and steel slider rails, with cast aluminum corners. I knew it was ill-advised. It works suprisingly well, not really cost/time effective DIY but great for learning. Anyways, some clever guy described how he strung picture wire around the corner and sides of the prism to stiffen the 2020 on his shakey Kossel.

    I have noticed as I get above 10mm/s, sharp corners/acceleration/high frequencies excite the prism, creating a wavey print. Of course, there’s a dozen Marlin & slicer setting involved too, but the stiffness of your trusses determine how much imbalance and jerk before it shows.

    Hey, perhaps a structure, lighter and even cheaper, there’s active vibration damping with solenoids and microphones. A great way to boost the price of cheap hardware with expensive DSP/GPU chips (its not working so well with cars now, I’ve heard). And more expensive software/firmware too, with feed-forward pre-compensation.

Leave a Reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.