Software Tackles Inflatable Structures

When children attending the annual Macy’s Thanksgiving Day Parade gaze in wonder as a giant Mickey Mouse floats above their heads, they probably are not thinking about how such a balloon is built. Researchers at Disney, however, are very concerned with the optimal way to design and build such inflatable structures—not just parade balloons, but smaller air-filled characters, inflatable furniture, and even portable architectural structures.

Much design work these days for all sorts of items is done using computer-aided design (CAD) software, and one of the key issues for CAD programs is how to "flatten" a product, creating the correct shapes for the two-dimensional patches that will be assembled into the 3-D object. Flattening is particularly challenging with inflatables, because the designer has to account for how air pressure changes the shape of the material at the same time while being concerned about how the placement of seams between different pieces of material will affect the final look of the piece.

"You need to figure out how do these patches need to look and how do you join them together," says Bernd Bickel, a scientist at Disney Research Zurich who works on such issues. "This is actually quite a complex problem."

Melina Skouras, a Ph.D. student in computer graphics at the Swiss Federal Institute of Technology in Zurich (ETH Zurich)—along with Bickel and colleagues from Disney and Columbia University in New York—presented a design tool they’ve developed for inflatable structures at the recent SIGGRAPH 2014 conference in Vancouver. The aim of the software, Bickel says, is to give the designer complete control over the aesthetic aspects of the object and to make the program simple enough that almost anyone can use it. The computer does the optimization work in the background, computing the shape of two-dimensional patches using a model based on tension field theory.

To use the software, the designer starts with a target object, perhaps created by modeling software or copied from a real-world object. He then gives the system an initial guess on where the seams should go. Some of the seams are structurally important, but others–say one across a creature’s eye to delineate the eyelid—are aesthetic. The computer can then move some of the seams around to optimize the shape of the patches. "You can tell it where you want to keep the seams at a particular location, and where you can allow it more wiggle room," Bickel says.

The user can also create internal connections to create the structure he wants. For instance, if he needs a concave shape to create the inner curvature of a creature’s ear, he can add an internal connection that will prevent that piece from puffing outward.

Because changing one piece of the overall structure can lead to other large changes, optimization can be challenging. The team decided to use an iterative algorithm to handle the process, which can usually complete the redesign in a few seconds, allowing the designer to try different approaches. "It allows you to very quickly explore the design options," Bickel says.

"The problem of stretch is a very, very difficult challenge," says Eaton Donald, CEO of Tri-D Technologies, Toronto, Canada. "The fabric side of the 3D CAD space is far behind the mechanical CAD space."

His company makes ExactFlat, software for flattening 3D CAD designs for objects made with fabrics, such as car seats. The software has been used to design various inflatables, including a hot air balloon shaped like Darth Vader’s head. The software relies on libraries of physical characteristics of various stretchable materials, including their Poisson’s ratio—a measure of how they stretch in different directions—and their Young’s modulus—a measure of stiffness. "We’ve had quite a bit of success with very complicated surfaces," Donald says.

One challenge Bickel would like to tackle is adding a way to automatically minimize wrinkling. Because stretching material in one direction necessarily compresses it in another, it is difficult to completely eliminate wrinkles, he says, adding that it also is challenging for the computer to model them in real time. "If you want to simulate the wrinkles very accurately, it won’t be an interactive system," he says. At best, "we sort of know in which region the wrinkles are going to happen."

Donald expects the issue of how to design structures that stretch to become more important as manufacturing replaces traditional processes. "Textiles are evolving rapidly to replace things that are commonly metal," he says. "The whole category of manufacturing products we use in our daily lives from flat pieces of material is an important question."

Neil Savage is a science and technology writer based in Lowell, MA.