Subject:
Systemic Redundancy (Press Button A, or was it B?, to remove all doubts on this) |
Date: Wed, 20 Jan 1999 22:20:17 +0800
From: IASA Safety <safety@iasa.com.au>
To: Barry Mews <barry.mews@eclipse.com.au>
References: 1
Barry Yes, no doubt the bus-tie sensing relay is epicentric both to the system and the SR111 problem. I don't think it's by-passed when the Smk/Elec/Air switch is rotated. That control just manually switches the bus-tie so that the Gens (and their assoc busses) are out of the auto-switching circuitry. The problem with modern glass-cockpitted airliners is that their electronics can't abide by spiky, surging brownouts - like the older generation could (with its more robust electronics). This means that wiring faults are more likely to induce component failures and even generator trips because of the sensitivity of the monitoring. The Valujet crew, in an older generation jet, got a fire (lethal enough) but the Swissair crew were vulnerable upon two counts - whilst fighting the smoke their systems caved in, including their EIS displays of flight instrumentation. To get the best analogy of the modern airliner's electronics (and contingent computer systems) and their innate vulnerability you've only got to look at the specs for a redundant hydraulic system. It doesn't share accumulators and reservoirs and common filters, bypass valves nor hoses. It is purely redundant up to the actuators. In a similar fashion a triple redundant INS only shares a common keypad (but that is key-locked for logic to avoid "garbage in"). The fuel system has separate tanks, pumps, lines, filters, cross-transfer and crossfeed capabilities. If all else fails two engines will gravity feed - and anyways there are three engines. The only vital airliner system that sports UNredundant commonality throughout is the electrical system- and its got lots of it. The designers would have you believe that three generators, an ADG, three busses, a few rectifiers, an inverter and a couple of so-called EMER busses constitute system redundancy. It's actually a hybrid composite that achieves INCREASED vulnerability (i.e. more points of possible failure and more shared componentry). If you could get a written guarantee that the system-common components such as the 260 kms of bundled wiring, bus-ties, switching circuitry, relays, junction boxes, batteries, battery chargers and CB's were technically inviolate you'd still not have redundancy -because of the software that drives the systems controllers. The software's always changing due to enhancements, bug-fixes, upgrades - and to keep those software-writers in control of their own destinies. We always have to accept (as an article of faith) that those gallant 1500 programmers were all of great calibre and consummate knowledge - without a vengeful bone in their encoding bodies. But, unmentionably, when the unthinkable happens and our aircraft's Machiavellian electronic system fails catastrophically, it is highly likely to do so with toxic smoke and a distracting inferno - not just a simple computer crash. What's worse we cannot simply pull the plug. We've got to ride it out - for better or for....... If you look at the attached Word6 file you'll note that: a. The MD-11 systems controller's computer transfers fuel to maintain longitudinal stability. It is capable of achieving a very aft Centre of Gravity so McDonell Douglas designers decided to minimise pitch trim drag by making the tailplane 40% smaller than the DC-10's and utilising this feature instead. But what happens (or could?) when the volts go mad and fuel is pumped uncontrolled over this long moment arm? b. All automatic systems incorporate BITE (built-in self-test). But that's electrical isn't it ?.....so what happens when the volts go mad? c. If the BITE detects a fault the problems are annunciated to the crew. Can you imagine what a cacophony of EICAS audio and Xmas tree caption lights SR111 had - before that too failed? Trying to listen to two concurrent audio alarms is confusing enough. d. When the BITE detects a fault, as well as annunciating it, the systems controller tries to rectify it and then, if it can't, tries to reconfigure the aircraft systems to compensate. Well and good whilst the electrical system isn't having impure thoughts. Once it's started its electronic epileptic fit, it could well nigh do anything -and totally unpredictably. ("We didn't design it to do that" instead of "We never designed it so's it couldn't do that") e. As a reassurance you're told throughout this (attached) blurb that manual operation of systems can be achieved by the press of an (electric) button, i.e. to re-achieve manual (but still electrical rather than computerised logic) control. But what happens when it's all gone tits up electron-wise? f. One of the features of the AFS (autopilot) is the Longitudinal Stability Augmentation System (LSAS - which is normally on). It incorporates the autothrottle, stick-shaker and auto stall recovery modes. Without that (i.e. once the autopilot fails) you're into a different stability regime with none of those protective features that you're used to. That's roughly similar to FIRST (or Direct) Law in the Airbus control system. Interestingly, in the Airbus system, if you run out of hydraulics you can only do a power-controlled ditching (i.e. they recommend against trying to land). The MD11 was designed in the mid 80's when every-one was agog about the increasing power of computer control capabilities. In eliminating the Flight Engineer they wanted to justify that by affording the pilots a protective layer of automation - a security blanket. What they produced was an electronic master-piece that was totally software dependent and never keyed to cope with a multiplicity of major and minor malfunctions stemming from electrical system outages. Unbeknownst to them at the time, by utilising Kapton wiring, they'd created the instrument of their own undoing. By incorporating the metallized mylar thermal acoustic batts they also ensured a thorough propagation of the offspring of that bad electric Karma. |