In 1995 I was driving around Ann Arbor, Mich., one rainy day when I became fixated on my wind-shield wipers. I was then an associate professor of mechanical engineering at the University of Michigan. In the preceding years I had done several studies of what is known in industry as “design for assembly.” The goal of such a study is to reduce the number of parts in any given machine, thus reducing manufacturing and assembly costs. In the course of this work, I had begun to wonder what happened if you took design for assembly to its logical extreme. Could we design products for no assembly?
As I sat behind the wheel, it struck me that my windshield wiper was a ludicrous waste of engineering effort. The wiper frame, which holds the disposable blade, has to be highly flexible. It must keep the blade pressed against the glass as it moves back and forth across a variable contoured surface. Moreover, it must be able to do so on a number of car models, each of which has its own windshield geometry. Our response to this need for flexibility? A complicated system of rigid bars, links and pivots.
At the time I had another burgeoning interest—elastic, or compliant, design, which involves building flexible, strong machines from as few pieces as possible. My colleagues and I had already succeeded in building machines from a single piece of material. For instance, in 1993 my graduate students G. K. Ananthasuresh, Laxman Saggere and I built a no-assembly compliant stapler. But the windshield wiper struck me as a perfect test case. A one-piece, or monoform, wiper would virtually eliminate assembly. If successful, such a project would be more than an exercise in engineering minimalism. Most of the cost of manufacturing a windshield wiper goes into its assembly. It should surprise no one that the production of such assembly-intensive products moved offshore to low-wage countries long ago.
My colleagues and I did not get around to designing the one-piece windshield wiper right away. For the past two decades most of my research has focused on general principles for elastic design—developing the theoretical tools that engineers need to design and build compliant devices. But we did eventually design that windshield wiper. In fact, we have used elastic design to build miniature monoform motion amplifiers, flexible airplane wings, robot snakes, and other machines, each one an expression of a new engineering paradigm whose time has come.Living Machines
We are more familiar with compliant machines than we might think. Perhaps the earliest and most elegant example is an archer's bow. As the archer draws the bow, elastic energy is stored slowly and then released quickly to propel an arrow. This strong, flexible mechanism can be used many times with precision and without failure. A newer example is the cap of a shampoo bottle: it is a monoform device that combines an easy-opening cap and a screw-on sealing collar without a mechanical hinge. Here is another example: the disposable medical forceps widely used in hospitals, which are precise enough for an operating room but inexpensive enough to be discarded after each use.
The most successful elastic designs exist in nature. I began to realize this in 1995, when I started reading works by Steven Vogel, the renowned biologist at Duke University. In books such as Life's Devices and Cats' Paws and Catapults, Vogel masterfully explains the working of nature's designs and draws parallels to engineered devices. Tree branches, bird wings, crab legs and elephant trunks are all flexible and strong. Their components either grow out of one another or are bonded together with strong, self-regenerating interfaces. Unlike systems of gears, sliders and springs, they bend, warp and flex by exploiting their inherent elasticity.
Humans have accumulated millennia of experience designing strong and rigid structures such as bridges and buildings. For the most part, we do this by using materials that are strong and stiff. If the stresses get too high, we simply add more material to share the load or increase its stiffness. Stiffness, in this paradigm, is good; flexibility, bad. Indeed, with rigid structures, deflection—the tendency to deform, or give under stress—is desirable only if you are designing for earthquake resistance.
Compliant design, in contrast, embraces deflection. If the stress on a flex point gets too high, we make it thinner, not thicker, because the function of a compliant structure is to exploit elasticity as a mechanical or kinematic function.
In the case of the shampoo bottle cap, the stress is focused on the thin polymer section that connects the lid to the base. Disposable forceps have much the same design. When the stresses are concentrated in a thin, discrete area, the flexion is referred to as lumped compliance. Researchers have been studying lumped compliance since the 1950s. More recently, Ashok Midha of the Missouri University of Science and Technology, Larry Howell of Brigham Young University, Shorya Awtar of the University of Michigan and Martin L. Culpepper of the Massachusetts Institute of Technology have all done excellent research on the subject, demonstrating applications of lumped compliance in precision instruments and nanopositioning devices.
The archer's bow, in contrast, has no such localized flexural zone: it displays “distributed compliance” throughout its whole length. Distributed compliance is essential for building flexible machines that have to do heavy work—wings that must keep planes in the air, for example, or motors that must run for millions of cycles. When I began my work in this field, I could find no theoretical underpinnings or general methods for designing machines with distributed compliance. Naturally, that is where I focused my efforts, and it is where my interest remains.Starting Small
I started working on flexible, one-piece machines not because they seemed like intriguing novelties but because in certain applications, designing for no assembly is a necessity. I began my career studying large mechanical systems such as automotive transmissions. In the early 1990s, however, I found myself designing truly tiny machines—micro electromechanical systems (MEMS). This was largely a circumstance of that era. Telecommunications companies were starting to develop optical switches for fiber-optic networks; they would use minuscule motors to change the angle of mirrors very quickly to route an optical signal in one direction or the other. Not long after I began reading Vogel and exploring elastic design, I embarked on a project with Steven Rodgers and his team at Sandia National Laboratories' microsystems division, where a monoform design seemed perfect.
Sandia needed to build a linear motor with sufficient output displacement to do work—at least 10 microns. Yet the fabrication constraints of electrostatic motors limit their motion to two microns. I knew I could not simply miniaturize, say, a geared transmission. Even if we could find someone with steady enough hands to assemble gears, hinges and shafts with dimensions in the one- to two-micron range, the resulting machine would be too sloppy for modern engineering. At MEMS scale, a machine with a tenth of a micron of clearance is about as useful as a Tinkertoy. Besides, MEMS devices are batch-fabricated much the same way as integrated circuits, tens of thousands in an area the size of a thumbnail. Given all that, I designed a monoform motion amplifier to generate 20 microns of output motion when integrated with the electrostatic motor.
By 1998 we had the motor and amplifier humming away. I clearly remember standing in the laboratory, marveling at the tiny device. It had been running for more than 10 billion cycles with no end in sight. But to my mind, the most impressive thing was that the entire motion amplifier, with all its complexity and flexibility, consisted of a single piece of polysilicon.Flexible Fliers
Of all of the reasons that I have chosen to study compliant design, the one I find most compelling is shape adaptation, or “morphing.” The ability to alter the geometry of a structure in real time enables nature's machines to operate with the utmost efficiency. Compare this adaptability with the fixed geometries of the engineered world—automotive drivetrains, airplane wings, engines, compressors, fans, and so on. These and practically all other conventionally designed machines are most efficient under very specific conditions. They operate suboptimally the rest of the time. An aircraft, for example, experiences a variety of flight conditions as it goes from point A to point B —changing altitude, speed, even weight as its fuel is consumed—which means that it is almost constantly operating less efficiently than it could. Birds, on the other hand, can take off, land, hover and dive by effortlessly adjusting the configuration or shape of their wings on demand.
Back in the mid-1990s, I wondered if anyone had ever attempted to change a wing's shape (camber) during flight to improve performance. I was amazed to discover that the Wright brothers had pioneered a different type of wing morphing—wing twist—in their original flier. I later learned that changing a wing's camber to meet different flight conditions on a modern aircraft had remained an elusive goal for decades. So one night I sat down at my dining room table and got to work on a design.
After a few months of study, I came across a small blurb in a newspaper about flexible-wing research that was conducted in the late 1980s at Wright-Patterson Air Force Base in Ohio. The engineers there called their goal a mission-adaptive wing (MAW). I knew nothing about the outcome of their work, but I understood that a morphing wing was not a wacky idea, so I contacted the researchers to ask whether they might be interested in reviewing my design. Their reaction was overwhelming.
They explained that most, if not all, past attempts to create a morphing wing have employed rigid structures—complex, heavy mechanisms with scores of powerful actuators to make a wing structure flex to different geometries. One time, for example, engineers modified the wing of an F-111 fighter jet with flexible panels. Their adaptive wing showed aerodynamic promise, but the structure was deemed too heavy and complex for practical application.
This did not surprise me. Designing a practical variable-geometry wing would involve satisfying many conflicting requirements. The wing must be lightweight, strong enough to withstand thousands of kilograms of air loads, reliable enough to operate for hundreds of thousands of hours, easy to manufacture and maintain, and durable enough to withstand chemical exposure, ultraviolet radiation and significant temperature changes. The conceptual and software tools in use at the time were never intended to design monoform machines, let alone ones that satisfied so many competing demands.
The flexible-wing design I submitted to Wright-Patterson exploited the elasticity of the test components, which were completely conventional aerospace-grade materials. The wing had an internal structure designed to deform easily when a compact internal motor applied force, and it still remained stiff when powerful forces were exerted externally in the wind-tunnel test. The senior engineers at Wright-Patterson were excited about the design, and so was I. In fact, I was so enthusiastic that in December 2000 I founded a company, FlexSys, to develop practical applications of compliant design.
Six years later, after much development and several successful wind-tunnel tests, we managed to get a prototype of the flexible wing affixed to the underside of a Scaled Composites White Knight aircraft for flight tests in the Mojave Desert. The wing was mounted below the jet's body and fully instrumented to measure lift and drag. Its coefficient of lift varied from 0.1 to 1.1 without increasing drag; that translates to a fuel-efficiency boost of up to 12 percent in a wing designed to take full advantage of the new flexible flap. (Flexible flaps retrofitted to existing wings would give a boost of 4 percent or more.) Considering that U.S. airlines consume about 16 billion gallons of jet fuel every year, these seemingly small percentages could be significant. The wing was also simpler, with no moving parts in the morphing mechanism. As a result, it would be more reliable and have a better weight-to-power ratio.
The real test for shape-adaptive aircraft wings will come when flexible-control surfaces completely replace conventional flaps. We are putting the finishing touches on just such an endeavor. Working with U.S. Air Force research labs, FlexSys designed and built a continuous surface that bends (cambers) and twists spanwise to maximize aerodynamic performance in place of drag-producing trailing-edge flaps. We have retrofitted a Gulfstream Aerospace GIII business jet with our FlexFoil variable-geometry-control surfaces instead of conventional flaps. In addition to significant fuel savings, our design is expected to reduce aircraft noise: according to nasa, much of the noise involved in landing a plane is caused by vortices generated at the sharp edges and gaps between the deployed trailing-edge flaps and the fixed parts of the wing. We have included transition surfaces to eliminate these gaps. Flight tests at nasa's Neil A. Armstrong Flight Research Center are scheduled to take place in July.Creepers and Crawlers
In the past few years my graduate students Joshua Bishop-Moser, Girish Krishnan and I have begun conducting elastic-design research inspired by the most flexible natural machines on earth—animals with no apparent skeletons. The most otherworldly among these life-forms, such as annelids and nematodes, conduct their business in ways that we are just starting to understand. More familiar examples, such as octopuses, provide an ideal for elastic engineers to strive for.
Soft-bodied animals such as worms and octopuses lack any apparent skeletal structure, and yet they can move vigorously and gracefully. For the most part, they accomplish this through what is called elastofluidics. In engineering terms, their bodies are hydrostats—they consist of an arrangement of connective tissue fibers and muscles surrounding a pressurized, liquid-filled cavity. A study of the anatomy of these creatures commonly reveals a cross-helical arrangement of fibers and muscles surrounding the internal organs, which occupy the liquid-filled core. The cross-helical fibers serve as antagonists against the fluid pressure generated by muscle contraction; the orientation of the fibers determines the range of motion.
Many variants of hydrostatic skeletons exist throughout the animal world. The arms of an octopus are muscular hydrostats. An elephant's trunk employs tightly packed muscle fibers around a hydrostatic body. An eel's fiber-reinforced skin acts like an external tendon, enabling the animal to generate a powerful propulsive force for swimming.
Our research on elastofluidics is still in its infancy, but our hypothesis is that these elements could serve as components for constructing “soft robots” and other devices that can safely interact with humans and the environment. The earliest applications, however, will most likely be in the field of orthotics. For instance, patients suffering from arm contracture caused by muscle hardening, joint deformity or joint rigidity could use a flexible orthotic device that gently forces their arm back into functional position for daily activities.Compliance Is Appreciated
With the assistance of many talented graduate students at the University of Michigan's Compliant Systems Design Laboratory, the basic research we started in 1992 has resulted in a trove of useful insights and systematic design methods. Those graduate students, too numerous to mention here, are now doing work of their own on elastic design at Pennsylvania State University, the University of Illinois at Urbana-Champaign, the University of Illinois at Chicago, Bucknell University, the nasa Jet Propulsion Laboratory, Sandia National Laboratories, Air Force Research Laboratory, KLA-Tencor, Ford Motor Company, FlexSys, Raytheon and Intel. Thanks to the talented engineers at FlexSys, some of the devices we have developed over the years are nearing commercialization. We have completed weather testing and finished the production mold for our monoform windshield wiper frame, and discussions are under way with automakers and suppliers for implementing it as a rear wiper. The monoform wiper is made of glass-filled thermoplastic polymer and works properly in both frigid and hot conditions. It will not snap or twist even when breaking loose ice and snow. When it comes to market, it should be much more durable and reliable and cheaper to manufacture than any competing device.
Our flexible aircraft wings are technically ready for commercial implementation right now. Replacing the outer 15 percent of an existing flap with a variable-geometry subflap for cruise trim alone could save 5 percent in jet fuel. Replacing the entire flap with a seamless FlexFoil offers about 12 percent fuel savings on new designs. It might be another couple of years before we get certification from the Federal Aviation Administration, but once the industry gains confidence in flexible wings, we believe it is likely that they will replace hinged flaps completely in future fixed-wing aircraft of all types.
Cases abound in the automotive, appliance, medical and consumer sectors where elastic design could drastically reduce the number of parts used in any given device. The biggest challenge is getting the word out to industrial designers. Widespread use of novel products such as our compliant wiper should help make the argument for elastic design. Even then, however, a challenge remains: there are currently no easy-to-use software tools available for exploring elastic design. With a contract from the National Science Foundation, FlexSys is developing software along these lines.
It will take several years before elastic design reaches any kind of critical mass, but we feel that its widespread adoption is inevitable. The strength, precision, versatility and efficiency that elasticity offers will give engineers in many fields an entirely new set of tools to work with, and soon we will all start to appreciate the power of being flexible.