Before they made it, they broke the mold.

Before they made it, they broke the mold.

Pleated Shell Structures 1 – Part 1 of 2.
Part 2, an interview with designer Shajay Bhooshan, will post on Thursday, Dec 6.

The Pleated Shell Structures recently exhibited at SCIArc (Southern California Institute of Architecture) are a pair of free-standing, canopy-like structures made primarily of FRP.  They are the product of an experiment in fabric-forming of FRP instead of using hard molds, a method that uses the stretching properties of fabrics to define a minimal surface meeting certain boundary conditions.

Forms were fabricated and cast at Kreysler & Associates in American Canyon, CA, north of San Francisco, and the project was assembled onsite at SCIArc in Los Angeles, all in just 8 days. It was a joint effort by Shajay Bhooshan of Zaha Hadid Architects (ZHA), Josh Zabel of Kreysler & Associates, and a group of SCIArc students.    Students from UC Berkeley and the California College of the Arts also participated in the fabrication process at Kreysler.  While ZHA is a design firm, and Kreysler is an FRP fabricator, both companies were involved in design, and both Bhooshan and Zabel were hands-on in molding and assembly.

The importance of the project is its exploration of shapes defined by edge conditions, and their translation into an alternative method for forming FRP.  Both ends of the process – the computational creation of shapes and physical fabrication – were experimental.

The method under exploration is to make frames, composed of flat pieces of plywood, whose curvilinear edges have fabric stretched over them.  The curved surface taken by the fabric becomes the basis of a hard FRP panel.

To achieve this, it is necessary to make software that simulates the response of fabric to being stretched over a contoured frame.  That is used to define surfaces – panels – that are parts of a free-standing structure.

The frame shapes defined in software are cut out of flat pieces of plywood on a CNC router.  The pieces are attached to a flat base to form a box: the top is open, and the edge is composed of the computer-cut shapes.

When fabric is stretched over that curvy top edge, it takes on a uniquely curved shape.  Stretching-tension is the principle force at work; the influence of gravity is ideally kept to a minimum.  In this case, a Lycra-like fabric that could be stretched 200%-300% is used to form the surfaces.

To “freeze” the surface permanently, FRP is applied.   A mat of glass fiber is wetted down with resin, and then applied to the stretched fabric.  Additional layers are added to build up strength.  The stretched fabric becomes part of the final product.  Foam implants in the corners of the legs create tubes of FRP around them in casting, forming a more structurally robust shape.

Once the resin has cured and hardened, the plywood form is removed.  The composite panel retains the stretched fabric’s contour, and is self-supporting.

That’s the theory.

For Josh Zabel, the most valuable aspect proved to be the move from theory to practice. “Finding out whether you can actually form fiberglass this way, and do it predictably, that is, to predict in the computer what the form will be when stretched and cast.”

In practice, gravity remained a very small factor.  “We tested with a flat frame, four feet square, to see how much the fabric would sag.  Over 4 feet, it sagged maybe 1/8 inch. It stretched to 200 or 300 %.  One guy pulling ‘kind-of-hard’ was enough to put it in sufficient tension.”

Zabel muses on the unexpected aspects of taking theory into the physical world. “I was leading myself to believe it was possible to do this computationally.  Shajay spent a lot of time working on a way to create a realistic digital model of a minimal surface. It breaks a surface down into a bunch of tiny points, makes little simulated springs between the points, and tries to minimize energy in all of the springs simultaneously.  That’s what fabric does naturally when stretched. It’s easy to simulate, but it turns out it’s not easy to simulate very accurately, especially when considering the variations in physical properties from one material to the next – or even one fabric to the next.”

In practice, when they cast panels, “we discovered there’s actually a great deal of influence you have in the act of making it, specifically in putting the fabric onto the form.  Machine stretching would be more repeatable.  The material itself makes a big difference, the difference between the shape that’s yielded by Lycra vs. canvas, for example. It seems obvious in retrospect, but that was the big discovery for me.  Of course, the next step is to configure the computer model so it better mimics the structures of a specific material.”

Because of the very tight fabrication schedule, the design changed.  The wooden structures now seen behind the FRP shells were originally going to be FRP themselves.  When it became evident that there would not be time to mold them all, they were re-conceived as plywood skeletal structures.  All elements are cut from flat pieces of plywood on the CNC router, similar to the molding forms.

The plywood edging on the FRP shells was also cut this way.  These thin wood strips were primarily an aid to aligning the panels in 3-D space during assembly.  “They are probably not necessary structurally,” says Zabel.

He points out that this method is only for creating minimal surfaces, but within that application, it is very practical.  “It’s a very inexpensive way to make a fiberglass mold. This method can happen faster because you’re not waiting for the CNC machine to mill the whole surface. EPS molds take a lot of manual preparation.   Between the two methods, the man-hours are probably about equal… although there is currently very little trade or institutional knowledge about forming FRP without a mold.  A real comparison of the labor would take closer study.”

Photos courtesy of Kreysler Associates and Zaha Hadid Architects.

Fabrication at Kreylser & Associates, assembly at SCIArc.