When a plane crashes - like the USAir jet outside Pittsburgh - journalists are quick to reveal the aircraft maintenance records maintained by the Federal Aviation Administration. But a huge piece of the puzzle is the FAA's own sloppy record keeping, and the way airlines compile information. The reporters went back to the FAA, back to airlines, back to overseers, mechanics, pilots, manufacturers and experts.
What they found were problems everywhere. Perhaps there is no better way to dramatize the gaps than this: In July 1992, a TWA L-1011 was taking off at New York's JFK International Airport. The pilot said he saw a warning system telling him that the engines were stalling. He aborted the takeoff, veered off the runway and the plane erupted into flames. Fifty-one passengers suffered minor injuries. Investigators started checking on L-1011 warning systems.
They found only two FAA reports of stall-warning system failures. But then they went further. Lockheed, the maker of the jet, had its own files. Those showed an additional 14 instances where warning systems went wrong. Investigators then looked at TWA. The airline had records of another 10 malfunctions. In other words, the FAA was missing records for 24 out of 26 incidents where warning systems went wrong a problem that appears to have contributed to a high-profile airplane accident.
Cleveland's Plain Dealer has been sifting through FAA records since November and published its first investigation last month. The results show the TWA accident does not represent the exception. It embodies the way of life at an agency that's supposed to regulate airline safety. The newspaper has been exploring the FAA's service difficulty reports, often referred to by their initials, SDR. The records are not supposed to be some kind of bureaucratic wasteland.
They're designed to play a vital role - helping the FAA troubleshoot airplanes by looking for trends. But bad data and a lack of staff mean the FAA can't do the job. The funny thing is, that isn't the story Plain Dealer set out to do. The newspaper sent two reporters - Elizabeth Marchak and Keith Epstein - to find which airlines are the safest and least safe. They were also told to do the same for types of aircraft.
And, to drive the point home, they were told to concentrate on aircraft flying in and out of Cleveland's Hopkins International Airport. Marchak and Epstein used the FAA's SDR database as a starting point. But it was just that: a starting point. FoxPro proved a powerful tool. The reporters conducted a variety of searches, categorizing and ranking records by aircraft and maintenance problem. Here are some of the stumbling blocks, and how Marchak and Epstein worked to get things right.
The SDR database tracks planes with tail numbers and serial numbers. But airlines over the years may repaint identification numbers. That rendered tail numbers almost meaningless as a way to rummage through the SDR database. Airlines also don't consistently report airplane serial numbers. In addition, the airlines are supposed to rank the seriousness of the maintenance problem on a scale of 1 to 5.
However, the Plain Dealer found those rankings are often haphazard. Marchak spotted the shifting serial numbers while performing data analysis. Sometimes they were three digits. In later years, they were six digits. Marchak went into the data line by line to get serial numbers to match. Despite the problems encountered, serial numbers are still the easiest way to follow an aircraft's history. At the same time, the Plain Dealer asked those airlines flying in and out of Cleveland to send lists of the aircraft they were using, including tail numbers.
And the newspaper sent a reporter to spend a day at Hopkins, jotting down all the tail numbers. Marchak later checked those numbers against the SDR databases. And a lot of numbers were not matching up. The problem wasn't obvious at first. The gap became apparent not by checking serial numbers, but the dates in the SDR database for when the planes underwent repairs. That's when Marchak noticed a disturbing mismatch: a lot of jets flying out of Hopkins showed no maintenance updates since 1992.
The reporters went back to the FAA, back to airlines, back to overseers, mechanics, pilots, manufacturers and experts. What they found were problems everywhere. In some instances, the airlines were submitting the data to the FAA - sometimes doing the job electronically - yet the information couldn't always be found in SDR. Many times, the airlines were not reporting maintenance as many as nine of 10 repairs went unreported The problem was particularly acute among commuter airlines.
And the FAA, left with weak provisions to enforce the rules, can do little to force airlines to change their ways. In addition, records were sloppy- Repairs were often mislabeled. The judgment was left to airline mechanics, who sometimes softpeddle the seriousness of a repair. Marchak and Epstein also looked for what wasn't tracked at all in the database, and found key numbers were missing - the number of hours flown, and number of takeoffs.
The stuff that would better track the repairs versus wear and tear. Along with poor documentation and no enforcement, the Plain Dealer used the database as a springboard to show how the FAA poorly analyzes its own records. Agency officials admitted that, in the wake of budget cutbacks, spotting dangerous trends with SDR doesn't get done. The kicker: FAA employees often don't use the agency's database.
A private consultant is among those who buys SDR data. He repackages the information and resells the data. Among his biggest customers is - that's right - the FAA. Agency workers find its easier to sift through his version, instead of the FAA's own compilation. For journalists using FAA data, the Plain Dealer's analysis offers another lesson. Learning an aircraft's maintenance record sometimes reveals less than meets the eye.
The Plain Dealer demonstrated that by profiling a MidWest Express aircraft with a raft of repairs. The newspaper compared that with the USAir jet that crashed outside Pittsburgh. The MidWest Express passengers were on an airplane that underwent more than 144 repairs, but there appears nothing that should have raised concerns. The same went for the USAir flight's 37 repairs. 'Trivial' was the way one expert described the maintenance record.
In the article Marchak and Epstein wrote: 'The two planes' histories illustrate the perils - for regulators, inspectors, reporters and the public - of drawing conclusions about relative safety from information in federal government computers. Even though that's why the computerized information exists in the first place.' For more information, the reporters can be reached by phone at 202-638-1366. To send e-mail, reach Epstein at 70404.247@compuserve.com and Marchak at Marchak@digex.com.