• Home
• windfarm structure
• suspension building
• component system
• modelwork
• synergetics info
• ABOUT
• CONTACT

After creating a number of computer models, I realized that, ultimately, there could be no substitute for an actual physical model of the suspension building. Computer models provided a good picture, but I needed to know how the structure feels. This was especially true because the geometry is so completely new and different from standard structures. Consequently I embarked on a project to fabricate prototype components at 20% scale. At this scale the model would behave much like a full-sized version and yet be small enough to transport fairly easily. Still, the tower and base together stand almost 14 feet high (4.25 meters).

Image 1 (below) shows the hub implementation, which can be compared to the design in the components section (see tab above). The hub "risers" were cut from sections of 2x4 kiln-dried douglas fir. The horizontal plates were cut from 0.1" thick Lexan (polycarbonate) sheet.

The large model needed a base to simulate ground supports, so I worked out the joinery for a pentagonal frame of 2x4s, combined with a cross leading to the center. All joints (except for one) correspond to a base point for the model; thus the same bolt that reinforces the joint also attaches the model.

Image 2 (below) shows the base construction. The 2x4s extend out from the pentagon, which were intended to serve as handles; these were later found to be not needed and were getting in the way, so they were eventually cut back.

Vertical struts were made with engineering (6061) aluminum tubing cut by hand with a pipe-cutter. Horizontal members were made of 1x2 pine cut with a table saw. Of course, all holes were drilled with a drill press.

Image 3 (below) shows the the first three levels being assembled. Because of the high degree of triangulation it was necessary to exert a high degree of precision fabricating components essentially by hand.

The "almost geodesic cap dome" on the top of the tower was fabricated with aluminum tubes joined with short sections of flexible vinyl tubing. Joined sections were then drilled and fastened together with bolts. Floor panels were cut from 4'x8' sheets of 6mm Celtec (expanded PVC) and fastened to the frame with small screws.

Image 4 (below) shows the complete tower frame and flooring displayed at a local "mini" Maker Faire. The model was found to be very strong and extremely stable, with no appreciable modes of vibration.

The 20% model received a lot of interest at the Maker Faires. It would have been good, of course, to be able to display the entire suspension building. However, fabricating the outer floors and creating the cable suspension system was beyond current resources. Also, transporting and and displaying such a large model would have been prohibitively difficult.

Image 5 (below) shows an unobstructed view of the tower.

Image 6 (below) Even though the outer floors were not added to the tower, I took time out to see how a single one of the largest floors would look. At 10 feet (3 meters) in diameter, it is a large model indeed. These floors would comprise most of the open floor areas in the suspension building design.

A primary consideration of the Towerdome system is structural efficiency due to joints that that only function in tension or compression (i.e. a truss). There is no reliance on joint stiffness. In order to test stability, I created a model with joints that are completely flexible, made with vinyl tubing. Hardwood dowel rods are pushed into the tubing and locked by galvanized steel wire pieces that go through and around the rod and tubing.

Image 7 (below) shows the complete suspension building. The outer floor suspension cables are modeled with steel wire wrapped around outside fasteners and kept from sliding off using flat washers.

It was originally thought that this type of a model with this many members might be impossible, since the stacking of tubes at the joints is very imprecise and warps the angles. However, as construction proceeded it was found to be indeed possible and actually a fairly close approximation to the correct shape.

Image 8 (below) shows a close-up view of the cap dome and upper portion of the model. In particular, the stacking of tubes is clearly visible at the center of the cap dome floor. At this point there are 5 struts coming up from below, 10 floor beams and 5 cables, all coming in to the same joint!

Image 9 (below) provides a close view of a single joint. It may appear that the joint would be loose and mushy, but the vinyl tubing, although completely flexible, is rather stiff in compression and tension. The end result is a model that is surprisingly light, fairly strong and extremely stable. Visitors to the Maker Faire were impressed that we could "simulate" earthquakes by sudden movements of the model base; because it is so light it could easily move with the base and feel completely tight.

Regarding structural efficiency, it is extremely easy to create a truss that is light and strong; simply make it fully triangulated. However, that tends to completely obstruct all usable floor space. The key innovation in the Towerdome system is the creation of "sparse trusses" that have the strength (or close to the strength) of a full truss but maintain internal voids for living spaces.

To test this idea (and avoid $5000+ for analysis software) I wrote a program that calculates a basic static analysis and displays the results by coloring the members according to the resultant stresses on them.

Image 10 (below) is the result of an even distribution of weight over all floors in the building. I call this the "cocktail party", a load approximately equal to 275 average humans (@ 175 lbs) spaced evenly at about the distance they would be at a party. This is about 50,000 lbs (about 22,500 kg), or about 7 lbs per square ft (34.25 kg per square meter).

The display color scheme is as follows: white color indicates a small amount of compression or tension, with color mixing into the white up to 1750 lbs for full yellow compression or fully green for tension. From 1750 to 3500 lbs color ramps to fully orange for compression, fully cyan for tension. From 3500 to 5200 lbs the color ramps to fully red for compression, fully blue for tension.

The analysis shows that tension load on the cable supports is fairly even, increasing to about 2000 lbs at the long "top" cables that go to the center. The compression load on the top level (between floors 5 and 6) showing primarily yellow, maximum at about 2100 lbs. On the next level down, support concentrates on the 5 upper extensions showing red, about 5200 lbs. The center "waist" section of the tower distributes the load over 10 members, about 4000 lbs apiece. The lower part of the tower has some concentration in the lower extensions (mirroring the upper extensions) but a substantial load focuses on the center ground support, 5 members @ 4400 lbs each.

Conclusion: the structure nicely distributes load stresses to keep any particular member from being overstressed. The upper and lower extensions take the most stress; if one of them fails stesses are routed around the breakage and taken up by surrounding members. Results for this "built-in" redundancy are too extensive to be shown at this time.

An exciting possibility is the idea of putting sensors into the hubs, with stress computations conducted in real time to flag situations where stresses happen to go outside of a "safe range".