This is the latest from Aviation Week & Space Technology.....
Jim
Aviation Week & Space Technology
Weak Deceleration Probed
By Michael A. Dornheim
12/20/2005 03:15:58 PM
Investigators are trying to sort out what happened to wheel braking and thrust reverser action during the Southwest Airlines runway overrun at Chicago Midway airport on the night of Dec. 8. The weather was snowy, the runway may have been slippery, and the aircraft deceleration was weak.
Flight 1248 from Baltimore/Washington International Airport broke through a jet blast deflector and the airport perimeter fence and came to a rest on the adjacent public road about 300 ft. from the end of the paved runway. The airliner struck two cars, and a 6-year-old boy in one car was killed, making it Southwest's first fatal accident since the airline was founded in 1971. None of the 98 passengers or five crew was seriously injured in the 7:14 p.m. CST nighttime acccident. A Chicago Aviation Dept. official says the accident involved the first fatality at Midway in more than 30 years. She couldn't recall a prior accident in which the airport boundary was overrun.
The National Transportation Safety Board (NTSB) says the crew reported that the automatic braking system on the 1.5-year-old Boeing 737-700, tail No. N471WN, was set to maximum and that it activated shortly after touchdown. But the crew said they later felt poor braking effectiveness, prompting both the captain and first officer to apply maximum manual braking. The flight data recorder (FDR) shows brake hydraulic pressure was applied constantly throughout the landing roll. NTSB investigators are attempting to understand how the crew felt poor deceleration while brake pressure was high. All flight attendants also said deceleration felt less than usual.
In addition, the reversers apparently were slow to activate. The 59-year-old captain was flying the aircraft and told the NTSB he couldn't get the reverser levers to unstow. The 35-year-old first officer was able to activate them but only at 18 sec. after touchdown -- and just 14 sec. before the first collision with the jet blast deflector, according to the FDR.
However, a quick calculation shows that not much deceleration would have been required to stop within the runway. Using FDR data, the NTSB estimated touchdown occurred with about 4,500 ft. of runway remaining. It's not clear where the plane actually touched down because the control tower couldn't see the aircraft, owing to poor visibility. The length of the runway from its displaced threshold is 5,826 ft., plus perhaps 200 ft. of overrun, which suggests the landing point was 1,300-1,500 ft. from the threshold -- fairly close to the normal 1,000 ft. spacing.
Indicated airspeed at touchdown was 124 kt. Allowing for low-density altitude (from temperatures of 25-27F) and adding 7-8 kt. of tailwind component gives a ground speed of 128-129 kt. The FDR indicates the plane made a hard touchdown and a slight bounce, says Ellen Engleman Conners, the NTSB board member at the scene. Assuming 2 sec. of coasting after touchdown for wheel spinup and brake activation, the average deceleration required to stop within 4,500 ft. is 3.4 kt./sec., or 0.18g. This rollout would take 40 sec. from touchdown to stop. On a dry runway, an aircraft could decelerate at least twice as hard as that, and 0.18g. deceleration, using brakes alone and no reversers, lies in the range of a slippery -- but not hydroplaning -- wet runway, or a runway with patches of snow.
The airport staff was plowing and treating the runway to maintain good traction, the Chicago Aviation Dept. official says. Friction test equipment indicated "good" traction at 6:55 p.m., 19 min. before the accident, as well as after the accident, the official says. But controllers rated braking as "fair" for most of the runway and "poor" at the end, Conners says. A business jet pilot who landed about 2 min. earlier reportedly said braking was fair to poor.
All of these factors highlight the difficulty of predicting stopping distance on a contaminated runway. The fact that ground equipment shows "good" traction while pilots report "fair to poor" shows how hard it is to measure friction. The FAA-certified dry runway landing distances include a 67% distance margin, but the charts of "advisory" distances for contaminated runways may have no margin and make assumptions about runway friction that are hard to relate to actual conditions.
"Data for dry runways are proven in flight test, and a wet runway happens so often that pilots are experienced," one 737 pilot says. "Then all of a sudden it's winter and you find yourself headed for a snowed-in airport with freezing fog and so on. You dig out the advisory table. If the table says you can do it, you will have a very hard time explaining your diversion to your fleet captain."
Traction can vary widely in snow, as exemplified by the range of values of the Canadian Runway Friction Index (CRFI). To cope with contaminated runways, Canada developed the CRFI and a ground technique to measure it. The CRFI ranges from 0 to 1, with 1 being equivalent to a dry runway and 0 being no friction. Canadian literature shows CRFI can range from 0.2-0.6 in snow with 0.2 representing compacted snow above 5F, and 0.6 representing very light patches of snow on the runway.
Assuming a dry runway produces a tire friction coefficient of 0.4, which translates very roughly into 0.4g of deceleration, then the range of deceleration in snow could be 0.08-0.24 g, which is both above and below what Flight 1248 needed to stay on the runway.
Prior to takeoff, Southwest dispatchers determined that Runway 31 Center, with its deteriorated weather conditions, was approved for landing. Takeoff was delayed by 2 hr., and the crew had to hold about 35 min. at Midway. According to the NTSB, the crew said they listened to recorded Midway weather reports as they flew to the field and agreed with the dispatcher's assessment of conditions. An onboard laptop computer verified that the landing would be possible and within Southwest's procedures. Investigators plan to compare the laptop results with flight manual data.
Conners says the 737 hit the jet blast fence just off the end of the runway at about 40 kt. That works out to an average deceleration of 0.15g from touchdown 32 sec. earlier -- a crucial amount less than the 0.18g required to stay on the runway.
Extrapolating from the deceleration observed on the FDR -- with actual use of brakes and reversers -- the NTSB calculated that the landing roll would have been 5,300 ft. on an unobstructed runway. That also equates to an average 0.15g deceleration assuming 2 sec. of initial coasting. The actual distance was about 5,000 ft. with the aircraft being slowed by going through both fences.
The NTSB states that "autobrakes were active and provided high brake pressure upon touchdown. Autobrakes and manual braking continued to provide high brake pressure throughout the landing roll." It's still possible there was a mechanical problem with braking but it would have to be in the narrow area between hydraulic pressure and the brake discs. The NTSB found the brakes in good condition with adequate wear remaining. The main landing gear tires had acceptable tread depth and no indication of flat spots. The crew said the aircraft was "completely normal" before landing.
One area that will be examined is Southwest's policy to not use autobrakes, the crew's apparent violation of that directive, and why the airline has the policy. Some organizations, including Boeing, believe pilots may not apply sufficient braking in critical situations and recommend that autobrakes be used. A Southwest official declined to explain the reasoning behind its policy, citing the ongoing investigation.
Southwest had a runway overrun incident in 2000 at the Burbank, Calif., airport when a 737 crew accepted a quick descent from air traffic control and landed long and fast. Like Midway, Burbank is tightly surrounded and the aircraft rolled out onto the public road, stopping short of a gasoline station (AW&ST Mar. 13, 2000, p. 40).
The balky thrust reversers will be another area of investigation. Thrust reverser levers sprout off the throttles and are normally down in the stowed, forward thrust position. To activate them, the throttles are placed in idle, and the pilot pulls the levers up to the reverse idle stop. They can't move beyond that point until the reverser mechanism has slid into position, then the pilot can pull them up farther for more reverse thrust.
The levers can't be moved at all unless the throttles are pulled all the way back to idle. That seems to be the case because the autobrakes appear to have worked, which requires that the throttles remain at idle. The reversers aren't armed unless radio altitude is below 10 ft. or aircraft logic determines it is on the ground. It's not clear yet what was causing the reversers to balk. The FDR shows all reversers were deployed until after the aircraft left the paved runway overrun. Spoilers are important for deceleration but investigators say they were working.
Hourly weather reports bracketing the accident gave 0.5-mi. visibility in snow and freezing fog with ceiling ranging from 400 ft. broken to 200 ft. indefinite. Temperature was 25-27F. The reports remarked that the snowfall was increasing rapidly at 1 in./hr., with 10 in. already on the ground.
Runway visual range (RVR) was 4,000-4,500 ft. on the accident runway 31C. Its approach minimums when using the instrument landing system are a 250-ft. decision height and 4,000-ft. RVR. The 737 was equipped with a head-up display that can provide guidance to a synthesized runway aimpoint. The opposite direction runway, 13C, has the same decision height but a longer 5,000-ft. minimum RVR, which made it unusable for the measured conditions.
The reported winds at 6:53 p.m., 21 min. before the accident, were 11 kt. from a heading of 100 deg., or a 9.5 kt. tailwind component to Runway 31. Pilots and controllers were discussing the weather just before the accident and the NTSB is trying to sort out these transmissions, an agency official says. The discussion may have included whether to change landing direction to Runway 13C. The stopping distance would have been about 1,000 ft. less into the headwind, the NTSB calculates. But again, that runway was not usable for the measured RVR.
The airport was closed after the accident, stranding about 600 passengers for the night. It was reopened at 6 a.m. the next day using other of its five runways. The impact knocked out the localizer antenna for Runway 31C, but it was replaced on Dec. 13.
Midway opened in 1927 and since has become tightly surrounded on all sides by homes and businesses. This has resulted in the airport fence being only 200 ft. beyond the paved overrun at the departure end of Runway 31C, measured along the centerline. The fence comes as close as 82 ft. to the left side of the 150-ft.-wide runway.
The FAA recommends 1,000 ft. of "runway safety area" (RSA) clear space beyond the end of a runway, but this is essentially impossible at Midway. A September 2000 FAA study of Midway listed alternatives for RSA compliance, but said none were feasible or practical. The Chicago Aviation Dept. conducted its own study in May 2004, which verified the FAA report, the department official says.
One alternative in place at several other airports includes the use of a bed of crushable concrete to arrest overrunning aircraft. The top surface of the engineered material arresting system (EMAS) is weak enough that airliner wheels punch through for quick deceleration, but it is strong enough for passengers to walk on and to support rescue vehicles. The custom-designed EMAS is made by Engineered Arresting Systems Corp.
But EMAS was one of the alternatives that both the 2000 and 2004 studies found to not be feasible for Midway, the aviation department official says. "We meet and talk with the FAA regularly to try to identify new technologies," she says.
Major airports will be required to meet FAA runway safety area standards or install an alternative such as an arresting system by 2015, as mandated by legislation sponsored by Sen. Frank Launtenberg (D-N.J.) that was signed into law on Nov. 30.
"We hope that the federal government and others step up to the plate for this national problem," the aviation department official says. "Midway is just one of about 300 airports in the country with this problem."
