The Voice of the Future – Part 2 of 3
Recently, we had the opportunity to talk to a group of advanced architecture students at UCLAunder the instrucion of Julia Koerner, who had participated in an unusual architectural competition. They submitted three innovative projects to a challenge made by ACMA (the American Composites Manufacturing Association.) These projects all involved creating FRP architecural panels without he use of a mold.
In Part 1, last week, we discussed the first of the three entries. This week, we present to two remaining entries.
Team: Ruslan Antonenko, Ruolin Xu, Yitao Chen, Ting Xu, Chunxiao Wang, Uriel Alexander Lopez
“Our initial inspiration was from the textile industry,” explained team member Uriel Alexander Lopez, who presented the project during our interview. “We looked at weaving and overlapping patterns. We thought that was interesting because the composite were working with is e-cloth. We thought maybe a weave would give us greater strength.”
One of their early experiments was a woven model with styrofoam pads interleaved for formwork. The idea was to burn out the styrofoam afterwards, but that was not completely successful. However, it led to idea of making a large panel and then splitting it after curing.
The panels were formed by placing a long strip of fiberglass fabric in the vacuum bag, and having each end of the bag held by one of the robot arms. Between the two robots is a jig, a horizontal bar with a curved surface, mounted on a sawhorse. The robots hold the bag in tension, and lower it onto the jig, creating a curved panel inside the bag. The curvature depends on how far they pull the bag down across the jig, and how much tension they apply.
“It’s sort of like wire-bending technique. We can use one jig that’s exactly the same shape to produce multiple curvatures, depending on the tension and angle of the robots.”
The initial product of this process they dubbed the “mega-panel.” After removal from the bag, this mega-panel is sliced in two, longitudinally, making two strips with identical curvature. One is flipped over, yielding an up-arch and down-arch pair. A series of these arches form a translucent skylight that diffuses incoming light. It’s designed to be topped with a sheet of flat glass that seals the skylight against the elements.
“What’s nice is that the two robots are really precise. Down to the millimeter, more than to the millimeter. We got pretty precise in the end. When you organize it like that, you know exactly where you’re going to end up. One of the benefits was that you could create parametric design by just using the positions of the robots, without having to make a mold every time that you have to CNC and then waste afterwards.”
“A big part of this class was using digital simulation techniques to rapidly figure out how the material would behave. What really took the most work was just dialing in the settings to make the simulation act like the infusion bag. The motion itself is pretty simple.”
There were a lot of different aspects to this project, many different skills brought into play. Software simulation, bag-stress analysis, light analysis, and all the physical aspects of making the composite parts. The team divided into sub-teams with specializations: cutter, resin mixer, etc. This was precisely the strategy instructor Julia Koerner had in mind when she divided the class into relatively large groups.
“The end product we made,” continues Lopez, “was a skylight system that was partially structurally. Towards the end of the project we started to think, How can the curvature be performative? A lot of the things that come out of the composite world have curvature built in, so that you develop strength through the geometry. We thought, Wouldn’t it be cool if the skylight system we’re putting in could also help with the structural load of the glass?”
The panel consists of two layers of 25 oz fiberglass cloth, with a layer of sorc in between. “Once they were together, you couldn’t flex it. So we came up with a hybrid system: the glass is structural, but it’s also supplemented by the strength of the fiberglass in compression. The fiberglass is not taking all the load, because we tried to be realistic in our approach, but it’s an idea for how you could begin approaching that.”
“Towards the end we started doing tilts, so that we could get these nice openings, undulations. That was actually really hard, because once you start turning the bag like that, one side becomes longer than the other. We had to do a little bit of math to make sure the robot was slightly rotated, so that when it start putting on the tension it doesn’t rip the bag apart. Once it’s programmed, there’s just one button, we hit play and pause it when it gets tight.”
Team: Anna Kudashkina, Yifan Wu, Yuekan Yu, Shahr Razi, Simi Shenoy, Marcelo Marcos
Undulating Gills was the result of a creative group taking aim at the problem of mass production. Usually, in composites fabrication, mass production means making a mold, and then casting multiple copies of the element from that mold. The evolution of the Undulating Gills transcends that production concept, but it is not exactly mass customization, either. Rather, it is mass production with a twist, literally, and in this system, the fabrication method is an integral part of the design process.
“For this project we were looking at creating multiple large scale panels in a single vacuum bagging process,” explains Simi Shenoy. “That was our main strategy for this. We’re calling the technique we used the ‘sandwiching technique.’ We put multiple layers of fiberglass in one vacuum bag, and we separated them using peel-ply. So we would get three panels in one shot.”
The bag with the package of panels was held in tension between two enormous Kuka robot arms. The bag was infused with resin. Then, the robots twisted the bag, to that one end was rotated about 90 degrees out of plane with the other end. The robots held the bag in that position while the resin cured.
“We were using robotics as a tool to achieve the design intent that we had from the start,” continues Simi, “that is, this twisting gesture. Initially, it started with testing to see how it would work, or if it would work at all. We figured out that these panels came out completely identical to one another. Mass producing, saving time, was the intent behind that.”
“We always started small,” points out Marcelo Marcos, “and if they worked, we did them bigger. We were really interested in having many panels in one infusion so you save time and materials. We were really interested in mass production.”
The twisting produced an interesting panel, but it also had a significant structural weakness at the twist.
“We did some stressflow analysis and pressure testing in Karamba simulations in Rhino and Grasshopper,” says Simi. “The part where the opening is had a lot of stress. We had a couple of extra carbon fibers and we got the idea to put them together to reinforce that part. It [carbon fiber] was included in the infusion bag.”
The flow of the seminar goes very well between the digital explorations and the physical ones,” puts in Anna Kudashkina. “There was a set of panels that came out very nice but quite weak. We started to explore what happened and how we could change it. One of the results, based on the Karamba simulation, was just to add the carbon fibers, but there were other simulations which were just to change the twisting angle, and that also worked. Both of those techniques were developed for the other prototype.”
“We studied the difference between fiber glass and other materials,” adds Shahr Razi, “the tensile strength, and sheer strength, and ratio of elasticity of fiber glass compared to other materials. We actually were taught the perfect ratio of resin to fiberglass, to get perfect strength. We found out that glass fiber is really good, but you have to use carbon fiber to get the maximum strength panels.”
“We got to the idea,” relates Marcelo, “what if the panels could be different in the same infusion?”
“We started to experiment how we could shift the panels,” explains Anna, “so that in the final façade, we could achieve this parametrized effect.”
The team began shifting the position of the layers in the bag with respect to one another. Then, when the robots twisted the bag, the curvature of the twist occurred at different positions in each panel. There was a close relationship between the different panels, a running curve that seemed to flow from one to the next, which led them to name the project Undulating Gills. That “gill” curve was even enhanced by cutting the outline of each piece fiberglass panel into a shape that expanded the reach of the curve.
“As we created multiple panels,” continues Simi, “they started to create a parametric relationship with one another. That’s something we looked at very closely, the relationship of these panels to one another across the façade. In our further development, we started to change the outline of the panels so that they’re not identical but all sort of shifting, so the relationship between them changes.”
The process was not without its bugs, however. The twisting gesture did not always produce smooth curves.
“We got these weird creases and wrinkles,” recalls Anna. “We tried different techniques. In the beginning we were looking at profiling techniques, but the profiles came out with wrinkles kind of, the profiles themselves created creases.”
Their solution was to add a jig outside the bag, somewhat similar to the one used by the Slicer/Gradient team. This jig was a cone. The robots could hold the bag against the jig, at right or oblique angles to its axis, and add twist across the jig.
“We found out that in order to have a smooth surface and better control, and predict the shape that you will have,” adds Jorel (who was not even on the Undulating Gills team), “you have to have some sort of structure that the shape will take place on. But we didn’t like the idea that you needed to create a mold for one single object, like a one-off, and then destroy it, or that you have to produce exactly the same object with that same mold.”
Marcelo enthusiastically points out that with the jig technique, “you could have a hundred shapes with one mold. That was also very interesting for us, how can you get one shape make different panels, depending on how you bend or approach the shape.”
They also received some professional input about the real-world considerations. “We had some meetings with Walter P. Moore and their facade consultants,” continues Marcelo, “and they reviewed our panels and how could they be mounted into existing wall mullions or into the slab. We did two different mounts, how they can be bolted into an existing lacing mullion.”
However, inventing an entirely new way to simultaneously form multiple FRP architectural panels without a mold was not enough. ‘We were like, how can we eliminate all post production at the end?” explains Marcelo. “Can you have a final piece, with the finished coating, and you don’t need to paint or sand at the end?
On a molded panel, the gelcoat finish could be applied to the mold. Without a mold, the finish material would have to be part of the laminate package.
“We put some fabrics inside the infusion bag and tested them” Marcelo goes on. “One melted, because the fabric was not good enough for the resin.” But eventually, “the result was really nice. So you can have a final product after infusing without post production.”
The robot track at the UCLA IDEAS studio is seven meters long, so that defined the limit of the panel height they could produce. However, they believe the technology will scale. They could make a curved section about two stories high on the equipment at UCLA, and add simple flat sections on either end of that, tall enough for the façade of about a five-story building. And out in the real world, there are longer robot tracks.
This project one first prize in the ACMA competition.
During the interview, I asked the student if anybody else in the world (that they know of) is doing this kind of panel-forming. Several of the students grinned a little, and said, “No.” They’ve got a right to grin, because I think they are now the planetary experts on these techniques.
Molding has been the dominant method for forming FRP composites pretty much since the invention of the material. Pultrusion – using a cut-out profile to squeeze the material through in a continuous forming process – is probably second. Both methods involve making the shape first, and then forming the composite to it.
These two UCLA projects represent an entirely different approach. The shape of the final product is variable, and is determined during the casting process. This kind of approach mirrors some of the speed and versatility advantages enjoyed by additive manufacturing (e.g. 3-D printing), but retains the engineering advantages of directional fabric reinforcement. It is a major technological and conceptual advance with exciting architectural potential.
And, as I mentioned last week, these insightful, knowledgeable, creative young architects are looking for jobs.
Next Week, Part 3: How They See the Future of Architectural Composites