The micro-windmills (technically called horizontal axis wind turbines) have a three-bladed rotor 1.8 mm in diameter mounted on a tower about 2 mm tall. The mount is a friction hub, but this probably becomes an air bearing when the rotor is spinning. The thickness of the windmills is about 100 microns. Despite their size, the micro-windmills can endure strong winds, owing to being constructed of a tough nickel alloy (rather than the silicon and silicon oxide layers typical of MEMS designs) and smart aerodynamic design. “The problem most MEMS designers have is that materials are too brittle,” Rao said. “With the nickel alloy, we don’t have that same issue. They’re very, very durable.” Several thousand such windmills could be made on a single 200-mm (8-inch) silicon wafer using Rao’s clever design and primarily conventional MEMS fabrication processes, resulting in very low per unit prices. Let’s get to the meat. As amazing as these devices are, can they actually be of any use? Unfortunately, detailed design information on the micro-windmills does not seem to have been published as yet, and predicting the exact performance of a horizontal-axis wind turbine is still a bit of a magic art (in the absence of serious computer modeling.) However, I have put together some rough estimates to get a feel for the potential performance levels of such micro-windmills. The first question is how much mechanical power can a micro-windmill generate from a moving flow of air. Betz’ limit tells us that the theoretical maximum conversion efficiency given an argument based on conservation principles cannot be larger than 59.3 percent of the initial kinetic energy of the airflow. Modern turbines can achieve around 45 percent conversion, but their design includes several areas of optimization that will likely to be absent in a MEMS windmill. For example, there are limits in forming the airfoil of the turbine blades, and the level of friction and rotor chatter resulting from the standard and rather inefficient MEMS rotary bearing designs will considerably reduce the windmill’s performance. However, if the conversion efficiency is reduced to 20 percent, and later windmills are designed with more blade area (and perhaps more blades), in a 10 m/s (22 mph) airflow, a micro-windmill can deliver a significant fraction of a milliwatt. If we assume an application includes a thousand micro-windmills (and that the wind is constant), the windmill array would generate perhaps 5-10 W-hr (430-860 kJ) of mechanical power per day – about as much as is contained in a cell phone battery. While the amount of power available may not be enormous, it is encouraging that it is sufficient for a host of operations, even if cell phone charging may be at the upper end of such applications. Consider bridge sensors. The sensor assemblies under development work at sub-microwatt levels most of the time, slowly accumulating data and awaiting a trigger signal to dump their information in a short burst of transmission. An average input of a few microwatts is all such a sensor requires. Many applications exist for energy harvesting, particularly in situations where revisiting sensors in place is impractical. There is a potential problem that a suspicious observer would spot. Not the issue of how to convert this small amount of mechanical power into electricity; we’ll look at that in a moment. The potential problem involves the boundary layer of air flowing by a surface. A principle of modern aerodynamics is that when air flows past a surface, the air immediately next to the surface does not move relative to that surface; that is, there is a no-slip condition between the two substances. As we know that away from the surface the airflow remains largely unaffected (assuming the airflow is directed along the surface), there must be some characteristic distance away from the surface where the airflow attains its original value. This slowed layer is called the boundary layer, and the distance at which the airflow is moving relative to the surface at 99 percent of its full velocity is called the boundary layer thickness. Here’s the potential problem: if the micro-windmill is small compared to the boundary layer thickness, the air flow will be dramatically slowed. As the mechanical power of the windmill varies as the cube of the airflow speed, windmills buried in the boundary layer would be emasculated. Fortunately, all works out well for the micro-windmills. At reasonable wind speeds, the boundary layer thickness likely to be encountered is smaller than a millimeter, or about one-third the total height of the windmill (I suspect that Chiao and Rao had done this calculation before making the windmill!). The remaining engineering challenge is to convert a fraction of a milliwatt of mechanical energy efficiently into electrical energy. Not surprisingly, this won’t be done by making a rotary electromagnetic generator half a millimeter in diameter with direct drive to the windmill’s rotor. On this size scale, electrostatic forces are, on average, far stronger than are electromagnetic forces. Nor will a thousand windmills be coupled together using gears and transmissions; the mechanisms are simply not sufficiently efficient for such tiny devices. The most likely electric generator would be an electrostatic generator in which the windmill itself is the generator. If the rotor is electrically isolated from the tower, the two form a capacitor whose value changes by a factor of (very roughly) 100 as the blades spin. If the capacitor is charged to one volt when one of the blades is over the tower (maximum capacitance), after which the charge on the windmill is isolated, the stored electrostatic energy will become 100 times larger when the rotor is turned 60 degrees by the wind to an inverted Y orientation. This extra energy comes from the torque of the rotor turning the rotor by 60 degrees. At this point, the capacitor is discharged into a storage supercapacitor or battery, and the generator is ready for the next cycle. With a coating of a high-k dielectric on the tower, perhaps 10 microwatts of electric power will be generated by a single windmill, depending on wind velocity. Note that my estimate is based on a crude model; better designs will increase the conversion efficiency considerably. In addition to the unusual development of the micro-windmill, Chiao’s lab has produced a range of MEMS devices and components that should help the development of medical micro-robotics. UT-Arlington has entered into a collaborative agreement with WinMEMS Technology of Taiwan for commercial development of these MEMS innovations. “The company was quite surprised with the micro-windmill idea when we showed the demo video of working devices,” Rao said. “It was something completely out of the blue for them and their investors.” For us as well.
