Computers Take Flight: A History of NASA's Pioneering Digital Fly-By-Wire Project (review)

For understanding new technologies that directly shape our lives, this work scores high, especially for anyone who has flown in an Airbus 320 or Boeing 777 or has any connection with American military aircraft from the F-16 and B-2 on. Computers Take Flight addresses the development of flight control systems with no mechanical connections between the pilot and the moveable aerodynamic surfaces that determine whether an aircraft goes up or down, right or left—or out of control and into oblivion. The first aircraft totally dependent on electronic, fly-by-wire controls flew on 25 May 1972, and since then digital fly-by-wire control systems—and the considerable physical and economic advantages they bestow—have become central to aviation. This was anything but preordained. How it came to pass that a small group of NASA engineers led by Melvin E. Burke convinced an inherently conservative aviation industry that mechanical linkages could be eliminated is James Tomayko's topic.

The story begins at the NASA Flight Research Center at Edwards Air Force Base, California, in the late 1960s, when Burke's group, their work on the Apollo Program winding down, began brainstorming future projects that might have significant impact and came up with the idea of applying fly-by-wire to aircraft. The advantages of fly-by-wire were many: replacing cables, pushrods, and hydraulic lines with wires saved weight; more importantly, a reactive electronic system using gyros and accelerometers for orientation could correct excursions in pitch, roll, and yaw long before a human pilot could detect them. This had the dual advantage of making inherently unstable aircraft flyable and of making it possible to shrink control surfaces dramatically, thus reducing both weight and drag.

Electronic control systems were not new: the German V-2 rocket had an electronic analog guidance system, the Canadian Avro CF-105 interceptor had used a digital yaw damper to control rudder oscillations, and digital fly-by-wire systems were standard for manned spacecraft. But no one had flown an aircraft without mechanical linkages. Burke's group proposed to do exactly that.

The question was whether to go analog or digital. Analog systems convert pilot control inputs to a voltage proportionate to the pressure applied, then use the resultant voltage to index an actuator that positions the flight control surface, usually by means of hydraulic servos. Digital systems convert the analog control input into a stream of digitized electronic information, then back to analog to position the servo. Analog was a more mature technology and at first seemed the logical choice, but changes in control-system logic require physically rewiring the system. By contrast, digital systems can be modified by rewriting the software. Aided by the availability of surplus Apollo guidance and control computers, Burke's group decided to go digital. A major issue—and the key to commercial applications—was reliability: eliminating mechanical backups would maximize the benefits of electronic control, but would demand unprecedented reliability. The answer was found in the pioneering work of John von Neumann, voting circuits that compared the output of multiple parallel systems and rejected anomalous signals.

The test vehicle was an F-8 supersonic fighter with flight-control computers filling the empty gun bays (three digital systems and a triple analog backup), a fortuitous choice in that the broad flight envelope permitted ample scope for experimentation and added credibility to the results. By the final flight in December 1985, Burke's team had made their point: properly designed digital fly-by-wire was incredibly flexible and reliable. Even though the program was cost efficient—the total cost was only $12 million—all was not smooth sailing; software proved to be the critical node, writing it more an art than a science, and debugging a tedious process. Still...