AIR SAFETY WEEK

May 3, 2004

Procedures Must Be Followed to Halt Mounting Toll of Maintenance Mistakes

'All is not well,' safety board member John Goglia declares

Airline flying is becoming less safe and maintenance errors are playing
an increasing role in the reduced operational level of safety, according to a
new study of the role maintenance mistakes play in aircraft accidents and
incidents.

The examination marks an unprecedented effort to take a comprehensive
look at the impact of maintenance error on operational safety, said professor
Gary Eiff, who directed the study at Purdue University's Department of Aviation
Technology in Lafayette, Ind.

The findings - that maintenance errors were a contributory factor in far
more incidents and accidents than previously thought - come at a time when
financially-strapped airlines are contracting out more maintenance to save
money, and in a period when federal oversight of repair stations has been found
lacking (see ASW, Jan. 21, 2003, Oversight Needs to Keep Pace With Expansion

of Contract Maintenance).

Although adherence to approved procedures is one of the most effective
defenses against errors, the Purdue study found that in 76 percent of the
maintenance-related events, failure to follow established procedures was cited
as a contributing factor.

Although inspection is often the last line of defense, the study found
that inadequate or missing inspections were a contributing factor in nearly 20
percent of the maintenance-related accidents and incidents.

Organizational factors, such as inadequate procedures, documentation,
training and supervision accounted for more than 15 percent of the maintenance-
related contributions to accidents and incidents.

The study was undertaken with the encouragement of John Goglia, the
resident maintenance expert on the National Transportation Safety Board (NTSB).

The NTSB has found maintenance error contributed to a number of recent fatal

air crashes, notably the Jan. 21, 2000, fatal crash of an Alaska Airlines [ALK]

MD-83, the Feb. 16, 2000, crash of an Emery Worldwide Airlines DC-8 freighter,

and the Jan. 8, 2003, crash of an Air Midwest Beech 1900D (see ASW, March 1).

Together, these three accidents killed 112 people.

As a consequence of the investigations into these accidents, Goglia
pushed for a wider examination of maintenance error. The Purdue study is a
result of this concern that recent maintenance-related crashes are symptoms of a
more widespread and sinister trend.

"Since coming to the safety board, and seeing the raw data, my gut was
telling me the rate of maintenance-related problems was much higher than
generally believed," Goglia said. "This study tells us that all is not well."

The Purdue study found that, indeed, there are worrisome trends that
ought to prompt a hard look at the level of training and oversight of aircraft
maintenance. The safety culture in aviation maintenance also is being
scrutinized in a separate study under way at Missouri's St. Louis University
(see ASW, April 12). According to a draft copy of the Purdue study obtained by
ASW, "Maintenance contributions to aviation accidents and incidents are
increasing."

"While the increase is not at an alarming rate, we should take action now
to arrest this trend," the report said. One of the more significant revelations
in the report is that while incidents are trending toward fewer per year, the
accident trend is in the opposite direction. Thus, while the total number of
accidents and incidents remains relatively the same, the report notes that "the
criticality of [the] outcome of such events is becoming more severe."

The assertion that maintenance errors are on the rise is a contributing
factor in the larger conclusion in the Purdue study: "As a whole, the air
carrier industry is experiencing a slight trend toward becoming less safe as a
mode of transportation."

This finding flies in the face of the fervently held belief that the
safety level is improving. For example, at the recent Federal Aviation
Administration (FAA) Forecast Conference, Nicholas Sabatini, associate FAA
administrator for regulation and certification, asserted that 2003 was the
safest year ever (see ASW, March 29). However, Sabatini's assertion was

based on fatal accidents, of which there were only two in 2003, while the 50 other non-
fatal accidents for 2003 pushed the total number of air carrier accidents last
year to the highest ever recorded by the NTSB.

How the system's performance is measured can largely drive perceptions
about safety. Eiff and his graduate students, who did the grunt work in the
Purdue study, looked at all accidents and all incidents in the NTSB database
over a 20-year period from 1982 to 2002. The Purdue study focused on scheduled

passenger operations; it excluded cargo aircraft and charter operations.

The purpose of the study was to determine if maintenance played a more
prominent role in aviation accidents and incidents than was previously thought.

The Purdue researchers analyzed roughly 1,300 NTSB records of airline
accidents and incidents. The study found that maintenance problems were factors

in nearly a third (29 percent) of the events. One of the key assumptions in the study

was that mechanical failures were maintenance-related. "While the reports do not

directly link these mechanical failures to maintenance, it is reasonable to believe

that many result from maintenance shortcomings," the report said.

The study found that maintenance was a contributory factor in far more
cases than previously thought. For example, a recently published Boeing [BA]
safety summary attributes only about 3 percent of crashes to faulty maintenance.
The Purdue study suggests that maintenance problems may be some 10 times
greater.

A major reason for the difference is that the Boeing data includes only
hull losses. Not all accidents are hull losses. In addition, according to Dustin
Wilcox, one of the students involved in the Purdue study, "We looked at
incidents as well, which might be described as accidents that got lucky."

Goglia believes the findings of the Purdue study are closer to the
reality. "The students who did the work did not change the NTSB findings in all
those accident and incident reports," Goglia said. "This was a straightforward
exercise, extracting what the NTSB said and compiling it. That is significant,
and a big reason why we need to pay attention to the results of this data-mining
effort." As an example, the data mining found more than 100 cases of nose
landing gear failures to extend for landing.

In a telephone interview, Goglia said the study's findings are consistent
with other inquiries. For example, an engine manufacturer found that 50 percent
of engine-related turnbacks (return to departure airfield) were caused by
installation error, Goglia said.

What can be done? "We need to start putting some mitigations in place,"
Goglia said. He recalled that after a spate of crashes from pilot error,
primarily from a failure to follow procedures, aircrews were enjoined to stick
to procedures.

Now, for maintenance, Goglia said, "We need to pound into everybody's
head the importance of following procedures." Secondly, he said, the procedures
themselves need to be evaluated for currency, consistency and ease of
understanding. Following the Air Midwest crash, which was attributed to
improperly rigged elevator control cables, the relevant section of the Beech
1900D maintenance manual was revised significantly (see ASW, March 1, , What

a Difference an Accident Makes). >> Goglia, e-mail gogliaj@ntsb.gov; Eiff, e-mail
geiff@purdue.edu <<

Growing Concerns

Research at several airline operations suggested that:

* Growing numbers of aircraft delays were maintenance-related.
* An increasing number of aircraft turn-backs and diversions were due to

maintenance.

* Several recent accidents and incidents occurred with maintenance as a

causal factor.