The most expensive failure of a twin Inertial Reference System (IRS) occurred on the first Ariane V rocket in 1996. Thirty-seven seconds after liftoff, a software fault caused an extreme attitude change, which tore the launch vehicle apart.
Apart from guidance and control, the other output function of an IRS is navigational data. The effect of an error there can be much subtler, but similarly dangerous. Navigation errors tend to be either self-canceling or cumulative.
Days after the disappearance of Adam Air Flight KI-574 between Surabaya, located on Indonesia’s main island of Java, and Manado, on Sulawesi Island, in January, a massive search failed to turn up anything. Adam Air’s B737s are equipped with twin IRS systems; one wonders whether satellite navigation would have been more suitable. Indonesia’s archipelagic layout and minimal radar coverage dictate the normal method of combating IRS drift-rates en route is via a manual or auto-update of the Flight Management System (FMS) from a VOR (and the odd DME) navaid somewhere along or adjacent to the planned route.
Unfortunately, VOR radials are only accurate to within 4 degrees and are subject to bending and scalloping, particularly when high terrain intervenes. For many Adam Air routes, distant fixes taken on the beam will be less accurate because of this. Other accuracy wild cards include the frequency with which pilots update the FMS and the integrity of the geographic coordinates inserted at IRS initialization on the ramp. All these biasing factors presume the twin IRS itself is functioning within laid-down drift-rate limits.
In a February 2006 incident, an Adam Air 737 ended up at Tambolaka, some 400 nautical miles away from its destination of Makassar, Indonesia. That "lost" episode was attributed to the vagaries of haphazard VOR updating. The pilot of KI-768 was a senior member of the Indonesian Directorate of Air Safety.
Ex-Adam Air pilots have reported a tendency for pilots to deviate from track to avoid weather and to readily accept direct (off airway) routing, as well as lax updating of VORs. Sometimes, direct routing would take aircraft well away from en route VORs, so that updating wouldn’t happen. However, if the IRS drift-rates were within limits, normally the terminal error would still be within range of the terminal navaids and the overall size of the cumulative IRS error would be inconsequential. Radar outbound tracking for the missing KI-574 was some 7 degrees off its initial airways track. The last radar paint for the aircraft was port of its track to Manado and well to the northwest of Makassar, supposedly following two approved deviations due to weather. No distress calls were received. A short transmission of an Emergency Locator Beacon later was received in Singapore by satellite.
Pundits are claiming whatever befell KI-574 must have been sudden and cataclysmic, such as a structural failure. There’s another possible explanation. Let’s assume KI-574 was off track, in or above cloud yet believing its IRS coordinates (whether updated or not). The pilots may have been puzzled by not receiving the Manado VOR-DME fix, or hearing any replies to their calls to Manado, and descended on time and/or IRS position anyway. They may have been far enough off track that other aircraft in the area at that altitude were on other frequencies.
There were some very high winds that day. If the pilots were well off-track to port and ahead of time, once they broke cloud and sighted the ocean, they may have assumed Manado was still ahead. However if the navigation error was gross (as in the Tambalaka incident), Sulawesi Island would have been behind them and all that lay ahead would have been the Philippines, many miles distant. Speculative, perhaps, but viewed in light of Tambalaka, not out of the ballpark.
VOR updating of twin IRS installations led to corrupted navigation problems in the past. It is not inconceivable the initial search area for KI-574 was 500 nautical miles removed from a possible ditching location north of Sulawesi. It’s noteworthy that Adam Air has now banned its pilots from accepting direct routing. There’s much to be said for eventually realizing your navigation equipment limitations.
John Sampson is the director of Air Safety Week, a sister publication for Avionics Magazine. Visit www.aviationtoday.com.