TURBULENCE TAMERS | |
| Picture this: you're balancing a broom handle on your palm, constantly adjusting your hand and arm to keep the broom from tipping over. Now imagine writing a computer program to mimic the balancing act, and you've got the basic idea behind a classic project that Professor John Baillieul, chairman of the Department of Aerospace and Mechanical Engineering, likes to assign to students. It's known as the inverted pendulum experiment, a case of building a control system that needs real-time computation and constant adjustment using a feedback loop. |  John Baillieul |
| And it's a good jumping off point to talk about Baillieul's current research in intelligent mechatronics. While the word mechatronics hasn't made it into the Oxford English Dictionary, it probably will soon, as more everyday devices and appliances include built-in electronics -- that's what intelligent mechatronics is about. Baillieul is trying to make high-performance aircraft wings and turbine engines adapt in real time to potentially dangerous conditions. Scale is key here. While an inverted pendulum experiment tries to control a single object, Baillieul and his colleague Tom Bifano, chairman of the Manufacturing Engineering Department, are working on arrays of thousands of tiny actuators, called micro-electro mechanical systems, or MEMS. The actuators have very small moving parts -- diaphragms and cantilevered beams, for instance-that move in concert in response to commands fed by a sensor in a feedback loop. Some of these small-scale devices are intended to control airflow over aircraft wings and through the complex blading of gas turbine engines, all in real time. Smooth sailing, so to speak, happens when there is laminar flow-fluid streaming smoothly over an airfoil or through the compressor stages of an engine. But when aircraft or engines are operating near peak performance, the flow is at risk of becoming turbulent, and vortices can form. "Then you get flow that doesn't produce lift on the wing or does not lead to the kinds of pressure you're trying to achieve in the compressor stage of a gas turbine engine. And when these things occur, you say that the wing or the engine is in stall," Baillieul says. "And you don't want that to happen." The proposed solution? Small-scale actuator arrays on the wings or in the engine -- each actuator only a little more than a millimeter in size, each moving in a distinct pattern-to try to effectively restore laminar flow. "By doing rather minimal things, if you do them cleverly, you can invest very small amounts of energy and use the natural dynamics of the fluids to do things like stabilize and separate flow," he says. "By making minor perturbations in the flow in some cases, you can effectively reintroduce the laminar flow. But it's fairly complicated to do that -- and turbulence is not well understood. What types of actuation, where to put the actuation, and how much authority each actuator requires are questions that we're looking at." Baillieul and Bifano have also discussed putting small sensors on a turbine blade row to detect changes in pressure over the surface. The idea is to hook them up with other microdevices such as airjets or microflaps to introduce closed-loop control of the flow -- essentially using small-scale turbulence to control large-scale turbulence. "If you structure the turbulence correctly, you can get a mixing of fluids," Baillieul says, effectively restoring the flow. A multi-component digital valve is another of the fluid control applications using actuator arrays that they are working on. Think of the valve system that controls the rate of flow through a pipe by opening and closing a certain number of tiny component valves. The amount of flow passing through the array at any given time is directly proportional to the number of valve actuators that are open. To be most effective, such array systems need to have up to one million actuators. And that makes feedback control in real time tricky, because it isn't possible to have a separate wire running from the computer controller to each actuator. "As you go to denser and denser arrays, there's not going to be enough real estate left on the chip for wires to each little device," he notes. Control signals must travel over shared communication links. Thus, feedback control of large-scale device arrays unavoidably requires real-time allocation of communication bandwidth, and this requires "new and novel technology that hasn't been developed yet. But it's going to come very quickly, because it's a compelling need. "Much of the research has very practical objectives, but in a larger sense it's just aimed at satisfying our curiosity, just like trying to see how small a pendulum can be balanced." So that swaying broom might come in handy yet. | |
[Table of Contents] [Current Issue][Archives] [ENG Home]
Copyright
Trustees of Boston University
Last updated on:
.