3/28/97

CONTACT: Stanford University News Service (650) 723-2558

Arrays of hundreds of identical millimeter-sized machines may replace conventional mechanical devices

Some day, aircraft may be powered by an array of hundreds of tiny jet turbines each a fraction of an inch wide, rather than by a single large jet engine.

That idea is among the blue-sky possibilities suggested by a new approach to mechanical design called "massively parallel mechanical systems" being pioneered at Stanford University's Rapid Prototyping Laboratory.

Although replacing a jet engine is well beyond the current state of the art, the scientists propose demonstrating the value of this approach by building several simpler but still useful devices.

Illustration of a flow control device for aircraft wings. In order to prevent the wing from stalling ­ a dangerous condition ­ critical areas of a wing are covered with an array of thousands of millimeter-sized holes, spaced about 10 millimeters apart. Each hole is connected to a microvalve that can release jets of pressurized air. Upstream of each hole is a pressure sensor. When the sensor detects the condition that causes stalling, it instructs the valve to open. The resulting jet of air inhibits the stall.

One such device is a system to keep aircraft wings from stalling, a condition that causes the wing to lose the upward force that keeps it in the air. Such a system would cover critical parts of a wing with thousands of tiny holes each about 1/25th of an inch in diameter and separated by 2/5th of an inch. In front of each hole is a tiny pressure sensor. When a sensor detects the conditions that precede a stall, it instructs a tiny valve to open, which allows a jet of pressurized air to blow out through the tiny hole behind it. If properly triggered, such jets can prevent a stall from developing.

This kind of system employs mechanical devices that range from a millimeter (1/25th of an inch) to the width of a human hair. Until recently it has not been possible to construct mechanisms in this "mesoscale" size range.

"Right now there is a big gap in the size of the mechanical systems that we can construct," said Friedrich B. "Fritz" Prinz, the Rodney H. Adams Professor of Mechanical Engineering and Materials Science. Normal manufacturing methods create objects a centimeter or larger, and micro-mechanical devices that measure a few microns (about a tenth the width of a human hair) are made using semiconductor manufacturing techniques.

Prinz and his colleagues are developing methods to efficiently make large numbers of mesoscale-sized mechanical devices. They are doing so by combining two different types of techniques. On the one hand, they are miniaturizing traditional manufacturing methods. On the other, they are scaling up techniques used in the semiconductor industry to create millions of transistors on a silicon chip. This allows them to efficiently produce large numbers of mesoscale devices out of metal, ceramics or plastic.

"There are major opportunities in creating new and attractive devices in this intermediate 'mesoscale' size regime," Prinz said. He and Robert Merz, an engineering research associate, described this approach on Wednesday, April 2, at the Seventh International Conference on Rapid Prototyping in San Francisco.

Replacing a few large actuators, turbines, valves and other mechanical devices with large numbers of much smaller devices has a number of potential advantages. For certain applications, key performance parameters, such as power density and response time, improve substantially as the size drops. Large numbers of small, redundant devices also have an inherent edge in reliability when compared to a few large ones.

Prinz, working with Lee Weiss, director of the Shape Deposition Laboratory at Carnegie Mellon University, has developed a layered manufacturing technology, called Shape Deposition Manufacturing (SDM), that makes it possible to create large mechanical arrays in much the same way that computer chips are made.

(a) Mesoscale nickel wheel assembly containing nine wheels mounted on axles. Each wheel is one third of a millimeter thick and five millimeters in diameter. (b) Four-bladed propeller is five millimeters in diameter: It would take 50 lined up end to end to span an inch.

The ability to make entire devices in place, without any assembly required, is a critical requirement for creating massively parallel mechanical systems. The researchers have fabricated an array of nine nickel wheels, each one a third of a millimeter thick and five millimeters in diameter, mounted on nickel axles to demonstrate that SDM can make entire mechanical devices in place, without any assembly. Similarly, they have made a four-bladed propeller that is five millimeters in diameter: It would take five of these, lined up end to end, to span an inch.

Prinz has proposed to the Defense Advanced Research Project Agency that it support the development of the aircraft flow control system described above and two other projects. If the proposal is approved he will be collaborating with a number of other Stanford researchers, including Merz; Paul Losleben, senior research scientist; Mark A. Cappelli, associate professor of mechanical engineering; John Fessler, acting assistant professor of mechanical engineering; John K. Eaton, professor of mechanical engineering; and Michael Binnard, a graduate student who works with Mark Cutkosky, professor of mechanical engineering.

Illustration of satellite thruster unit. Each unit would consist of an array of 10,000 millimeter-sized nozzles. Electrodes in each nozzle create an electrical field of up to 10,000 volts that accelerates small drops of weakly conducting colloid fluid at very high velocities. The array is designed for the ultra-low thrust systems used to keep orbiting satellites in position.

The two additional projects include a satellite thruster system and a tactile interface for virtual reality and teleoperation systems:

• The satellite thruster system is called a colloidal thruster. It is a type of solar-electric propulsion that has generated considerable interest as solar panel and battery technology has improved. Its major advantage is that it does not require as much propulsion mass as conventional thruster systems. Manufacture of conventional colloidal thrusters involves painstaking manual assembly, and they operate only at a single thrust level, which greatly limits their usefulness. The thruster array proposed by Prinz would consist of 1,000 nozzles one tenth of a millimeter wide and spaced a millimeter apart. Electrodes in each nozzle create a 10,000-volt potential that accelerates droplets of a colloid propellant to very high velocities. Output of the thruster can be controlled by turning individual nozzles on and off. Scientists from the Boeing Company will also participate in the project.

  
  Illustration of a "haptic" display that could provide users of virtual display and teleoperation systems with a sense of shape or texture. By resting a finger or hand on a pad made up of thousands of independently controlled millimeter-sized needles, the user could "feel" the roughness or shape of computer-generated surfaces.
  
  

• Force feedback mechanisms currently are used to increase the realism of virtual reality systems and give teleoperators a better feel for what they are manipulating. But these mechanisms are relatively crude and cannot give the user a sense of the shape or texture of the virtual objects with which they are interacting. Providing this additional level of realism is the goal of the proposed "haptic" interface. It would be something like a flat pin cushion, consisting of a dense array of millimeter-square pins attached to actuators that would position and push them up and down with a controllable amount of force. The millimeter spacing between individual pins would make the interface feel almost like a solid surface when all the pins are positioned at the same level. Under computer control, however, the surface could be programmed to imitate the shape and hardness of different surfaces. Researchers from micro-valve maker Redwood Microsystems will participate in both this project and in constructing the aircraft flow control array.

As to the idea of a jet aircraft propelled by an array of mesoscale turbines, the massively parallel approach would have some definite theoretical advantages, particularly in power density: It would weigh significantly less than a conventional jet engine of the same power. But an array of tiny turbines would also have some offsetting disadvantages, primarily in increased energy loss caused by siphoning air through hundreds of small holes rather than one large one. "Unfortunately, we don't yet see any clever way to get around these limitations," Prinz said.

For more information on the World Wide Web:
Stanford Rapid Prototyping Laboratory  http://rpl.stanford.edu/
Fritz Prinz home page  http://rpl.stanford.edu/fritz.html