Adjusting Prestress; High Prestress

New materials and methods are needed to create prototypes with high prestress and easily-adjustable prestress. Tom designed 3D printed end caps that attach to the end of a strut and allow tension to be adjusted.

Context: Tom Flemons Archive

New materials and methods are needed to create prototypes with high prestress and easily-adjustable prestress

(Sept 1, 2015)  My main problem right now (besides the form finding which is proceeding step by step to reproduce the various components needed i.e. gripper, appendage, torso and links to each) is to find materials and build components which allow me to increase the prestress to a useful degree. Traditionally turnbuckles and cables attached to fixed length metal struts were used to good effect but this is a tedious method with lots of moving parts. I’m looking at the possibility of 3-D printed end caps which could capture tension members and allow for tensioning to be incrementally increased through some sort of cam system but this is a hard design problem (especially miniaturization). Alternatively it may be the best approach is to build structures that are more akin to baskets using more flexible materials and constrain their compliancy through stiffeners of one sort or another.

(June 28, 2017) It would be a wonderful advancement to be able to adjust tensegrities on the fly as they are being built or afterwards when they were being fine-tuned. One problem, as Kevin found out, is they can be very difficult to keep from exploding or falling apart while being assembled.  If the model is large enough and you don’t mind if it’s a little sloppy when finished then various kinds of tensioned knots will work but it’s not an elegant solution…

The attachment method must be secure enough that you can assemble a Tensegrity without it coming apart but adjustable so you can make changes afterwards. This usually necessitates guessing fairly accurately the lengths of the tension members beforehand- even so it’s almost impossible to get it right the first time so it requires going over the entire structure and tediously re-tensioning every line. Depending on the method used this can take hours and be very frustrating. Add to this that any one tensegrity is usually just one of a long string of prototypes… necessitating building dozens if not hundreds before something starts to work. When I was designing my leg foot combination I lost track of how many versions there were. If I hadn’t developed a system using elastic cord caught in slots in the ends of dowels I wouldn’t have been able to make any progress over the years. At this point I have a literally built thousands of tensegrities using this method. Granted most of them were toys, but even so there were hundreds of iterations of biotensegrity models. (I have large boxes stored in my workshop filled to the brim with well worn out models). My system works so well because I don’t have to measure the exact length of the tension members. I just have to get it tight enough and the elasticity takes care of any inaccuracies. This is great for prototyping but is pretty much useless in testing out actual purpose driven designs. For example, I’ve designed a number of possible prosthetic legs including a knee joint and a foot. They look plausible but they are nowhere near strong enough to bear weight. I could thicken the elastic cords but then it starts to get congested at the nodes and I have to beef up the strength of the struts which adds weight and usually means thickening the dowels which creates more congestion…

Limits of materials

(Feb 9, 2016) I’ve had to negotiate these boundaries in my models because of the limits to the materials I was working with. Too loose elastics meant floppy models that didn’t demonstrate the properties of transference of force – they just got dampened in the elastic material.

Want rapid form finding, with more durable and functional models

(March 19, 2018) I put myself to sleep every night trying to solve the problem of building and tensioning tensegrities. As you know I think it is important to devise improved methods for rapid prototyping. My method of using elastic cord and slotted dowels has allowed me to make thousands of tensegrities over the years (and be responsible for millions more – 4 million plus Skwish and counting). It really has made it possible to do all this form finding – stuff that may also be done on a computer but I am convinced that having a tactile experience with a real model gives you information you cannot get from a simulation. But I also think my method only goes so far. If I or others wanted to continue with rapid form finding but ending up with more durable and functional models other means have to be found. This is where my exploration of end cap design comes in. There are a number of ways that a tensegrity model could be built out of more durable materials – clearly all the attempts all over the world demonstrate this – my point is – I know how much work goes into making them one at a time. If you get something wrong or want to change one variable you have to start again. This slows down the design and exploration process incredibly. So while you may want to end up with something as robust as the Superball Bot with spooling cables or the duct robot with 3D printed components I think there is a need for a new way to rapid prototype models.

So for example at present I have a spool of dyneema (= spectra) which looks like kite string but has a breaking strength of 150 lbs and I play with ideas to incorporate it into adjustable tensegrity design. So as I said every night I visualize ways to connect it to strut nodes such that they can be tensioned or slackened as the structure is being built. So far it’s all in my head – I haven’t been able to get out to my workshop for a while… This for me is a difficult problem to solve. For a good engineer who understands the problem thoroughly maybe there is an obvious approach.

Need to be able to individually tighten each tension member

(April 12, 2015) I am missing the ability to 3D print a mechanism that will allow various tensioning cables to be captured at the end of a strut and tensioned on the fly. My rapid prototyping system is not adequate for building real testable prototypes. What is needed is some kind of miniature clutch that can accept multiple cables entering the strut end at various angles and holding them in loose tension until the structure is complete and then allowing the entire structure to be tuned.

(May 19, 2016) I’ve built thousands of tensegrities over the years because I didn’t have to measure every tension length precisely. All I needed to do was cut an elastic cord so that it would have some amount of stretch to it when the form was built. I knew approximately how long a tension member would have to be but didn’t calculate it precisely. The times where I have built more complex, larger forms I used steel cable or wire for the tension system and then I did have to have more precise lengths. Even so tensegrities are never built exactly balanced – they must be ‘tuned’ like a violin. So a means has to be found to individually tighten each tension member to put the structure into a certain amount of tautness referred to as prestress. A tensegrity is like an inflated exercise ball. You can keep adding pressure until it gets very hard and then finally bursts but adding just enough pressure to keep its shape as a sphere (and not have a flat bottom) is the basic amount of prestress the system requires to be ‘integral’.

This is key to the building the next generation of tensegrity prototypes

(June 28, 2017) However the problem is solved, it is key to building the next generation of tensegrities prototypes. It may be solved in different ways depending on the type of tensegrity, the size, the purpose, and of course the budget. Regardless there are a few key properties that must be kept in mind:

  • The system must be modular and ideally scalable.
  • It must be attached securely to the ends of compression struts as I think it would be prohibitive to design struts with highly purpose fashioned ends.   The forces of tension is oblique to the axis of the strut and would tend to drive the end cap down onto the strut so a pressure fit is probably sufficient.
  • It must be easy to produce, relatively cheap, strong (made of tough material),  and easy to use and adjust. Anything else I considered including ratchet designs found for example on Weedwhacker’s or Cam cleats found on boats,  seemed prohibitively expensive and hard to miniaturize. Tuning pegs systems on instruments are similarly expensive to make and hard to miniaturize.
  • Also because of strut and line congestion at the nodes, whatever is used must be compact.  There could be a special instrument designed to facilitate tightening and loosening of individual lines.
  • Generally a discrete tensegrity is going to have a global tension that is the same in every  tension component. But in some cases individual lines may want to be looser or tighter, especially in asymmetrical Tensegrities such as the foot model.

It’s not an easy problem to solve. Before the advent of 3-D printers it was pretty much impossible. As I don’t have a 3-D printer yet, I’m excited by the possibility that one of you may be able to take on the challenge. I’m available at any time to consult with and open to all ideas and suggestions.

This may be beyond the scope of your studies – it’s really a design and engineering problem but let me know if anyone wants to take a run at it.

Designs for new methods of assembly

Requirements for the tensioning mechanism: small ratcheting device that can accept multiple cables entering at the strut end

(June 21, 2015) What I’m really aiming towards these days is a 3-D printed end cap that can be inserted into hollow compression tubes. These designed end caps would accept tension cables at various angles and temporarily lock them in place while the structure is being assembled. Once the structure is built it can be tuned by individually tensioning each cable. I’ve designed such an end Cap that may work [discussion below].

(Dec 1, 2015) Some 3D printed end cap that allows for individual tightening and loosening of many individual tension lines is needed – vis a vis how a violin or guitar is tightened but considerably miniaturized. I have a few ideas regarding this [discussion below].

Collet lock

(June 28, 2017) I have two designs on Google Drive. The first one is what I call a Collet lock.  Basically it’s a way of capturing tension by compressing the line as it passes through an adjustable Collet.  It probably can’t be built to allow tension to be released but it’s likely to work if the proper materials can be found. It requires that the hole passing through the Strut end is conical  and matches the conical shape of the collet.  This is probably too complicated a system to start with.


3D-printed end cap that accepts up to four cables; use spectra line

(June 28, 2017) So my second idea was to build an endcap that fit over a dowel and could accept up to four tension lines which could be individually loosened or tightened  as needed. It works on a similar principal as cleats on a sailboat. There are a number of different kinds of cleats on a boat.  This is the simplest type. By wrapping lines in a figure 8 pattern over the horns of a cleat you can secure it.  The line enters the endcap through a hole and then proceeds to be kinked  and then slotted  back-and-forth  until it is secure.  Passing it back through the hole would compltely secure it. Note that you can probably assemble the structure by only loosely securing each line through the hole and maybe several slots. After the entire structure is built each line need to be further attended to… more wrappings around the slots and then tied off through the  hole.

This will only work if the line or the hole or the slots have some grip to them and are not perfectly smooth.  There is the type of line called spectra which can be purchased braided. It’s pretty strong and has some texture to it so I think it would work.

The hole and the slots have to be close in diameter to the thickness of the line. Obviously the line has to pass through the hole… (incidentally the hole is to prevent the whole thing from coming apart which might happen if there was just slots).  After each end is complete it is possible to add an end cap to cover the entire mechanism  further securing all four lines.  It’s also probable that an end cap that accepted more than four lines could be designed. I’m not sure if ABS plastic would be the substrate – it maybe too slippery so some texture would have to be added to the slots to keep the line from slipping.

    Strut cap



Tensegrity end fittings with 10-32 threadcuts


End cap, screen shots dated Oct 11, 2012




Can use sliding hitches to adjust tension, but assembly is tedious

(May 17, 2017; Tom is responding to Kevin Zuern about his plans to  build a tensegrity structure using static cord, with Magnus Hitches or Trucker’s Hitches placed so that the tension on each line can be adjusted. Kevin’s expanded octahedron tensegrity is shown in this video.) I am a climber and sailor so am familiar with knots and have used many over the years with tensegrity constructions. I encourage you to play around with different materials and construction methods to see what works and I think you are on the right track.

Let me point out a few salient issues. Tensegrities work because every node involves multiple tension lines meeting and being secured to the end (or sometimes the middle of a compression member (strut, rod etc.) This quickly adds up to a great many tension elements each of which must be fixed so it cannot slide or come loose but must be available for adjustment after assembly. This alone makes tensegrity assembly time consuming and sometimes infuriatingly tedious.  While I have built many tensegrities using fixed length tension and compression members (i.e. cables with swaging or clamps, ropes with knots, chain etc.) I found long ago that using high quality elastic cord that slides through slots in the end of wooden dowels to be the most efficient way to quickly build tensegrities. The slots capture the cord with enough friction that they do not slide easily yet can be fit into the slots with little difficulty. The elastic cord can be cut and knotted (either overhand knots on a length of cord or a loop joined with a square knot) such that there is no need to be precise about the length – the elasticity means a range of tension works. Also because of the friction in the slots one length can pass through one or more nodes which saves cutting and assembly. This works well for structures up to several feet in diameter and useful for toys, biotensegrity models etc.

The problem arises when attempting to build more robust longer lasting structures like robots, furniture, architectural and sculptural pieces. If it is a matter of ‘one off’ assembly then slow and cumbersome assembly methods suffice but if I am prototyping some form (which requires multiple variations to be built and trialed) there needs to be a way to speed the process up and allow for adjustments on the fly. Your idea of using a variety of sliding hitches is a reasonable approach. But it will still be time consuming and may be easier to accomplish with larger structures. I like the idea of using your knots but note that when building tensegrities it can get very messy when there are a lot of lines and knots close to each other – strut and line congestion is a problem with tensegrities… Prussic hitches make sense depending on material as do truckers hitches.

This is a field that will open up in the next few decades

(May 17, 2017) I have a few suggestions and ideas for building structures and mechanisms I have been playing with for years. For example employing some kind of end cap that can accept multiple tension lines and the ability to tune them much like a guitar or violin is tuned seems a reasonable approach. Especially with the advent of 3D printing it becomes possible to design and print a generic cap that has a lot of fine detailed features. Attached (at left) is an image from sketch up which I thought showed promise but is only one approach. Essentially the problem is: design a system that is fail proof, tidy, easy to use, quickly adjustable, strong and mass producible. It seems clear to me that this is a new field and no one has really made much headway in it – chiefly because up until now there have been no uses that demand it. It’s a bootstrapping problem. Design a complex  sophisticated system for variable tensioning lines that is scalable, and robust and applications will start to appear. I envision robots, prostheses, smart deformable walls, furniture, even bicycles and other vehicles will come into existence if a way to make them well can be found. Tensegrity is just too important a discovery with too many advantages to be ignored for much longer. This is a field that will open up in the next few decades.

Many people are developing new construction methods for tensegrity model building. For example, there is very interesting discussion of how to build large-scale tensegrities at Gerald de Jong demonstrates exciting new approaches to tensegrity building at with a studio tour in this Feb 2021 BiotensegriTea Party video.