“It caught my eye becauseit was…different,” said Brad Brooks, a customer service agent. That difference was an angleof ascent more than 45 degrees– other than a brief correction so violent that the tail pitched over the nose. “I’ve never seen that before,” said Brooks.
Brooks stood outside the operations door at Little Rock on Oct. 14, 2004, to watch the Bombardier CL600-2B19 take off at 2121 CDT for a repositioning flight en route to Minneapolis-St. Paul. Officially the flight was designated as Pinnacle Airlines FLG3701, doing business as Northwest Airlink.
Transcripts reveal that a radar controller later asked the pilot to verify that his aircraft type was an “RJ200.” The pilot confirmed.
“Oh, I’ve never seen you guys up at 41 there,” said the controller.
“Yeah, we’re actually, uh, we don’t have any passengers on board so we decided to have a little fun and come on up here,” responded the pilot. “It’s actually our service ceiling.”
At 2213 CDT on Oct. 14, 2004, flight 3701 crashed in a residential area in Jefferson City, Mo., 2.5 miles short of the Jefferson City Airport. The two crewmembers were killed and the airplane was destroyed by the impact and fire.
On June 13, the NTSB opened three days of hearings by its Board of Inquiry, chaired by Deborah Hersman, to supplement 61 pages of cockpit transcript and hundreds of pages of interviews, complex exhibits and test results.
“This hearing is being held for the purpose of supplementing the facts, conditions and circumstances discovered on-scene and throughout our continuing investigation. No determination of cause will be rendered during these proceedings. We are not here to malign anyone for actions taken or not taken leading up to the accident,” she emphasized.
In Defense of the Pilots
As official testimony opened, the Air Line Pilots Association (ALPA), which was party to the investigation (as were the FAA, Bombardier Aerospace, General Electric and Honeywell), distributed a media release by ALPA executive air safety chairman Terry McVenes.
“The NTSB investigation makes it clear that if just one of the two stalled engines had restarted, this accident would never have occurred. The facts show that the pilots followed the proper procedure and attempted to restart their engines multiple times. But both engines failed to restart because they had suffered ‘core lock,’ a safety risk previously known only to engine and aircraft manufacturers until very recently– and about which the pilots knew nothing,” wrote McVenes.
Chairman Hersman reminded the panelists and some 200 attendees that no parties should purport to speak for the Board’s analysis or to make conclusions, either during the hearing or until the NTSB releases its determination of probable cause.
In its report, the NTSB might cite factors from a pro forma menu gleaned from post-accident action and advice from manufacturers, positioning statements by the alphabet groups, and interim communications by the FAA. These factors include pilot training regarding stall awareness; the dual-engine-failure checklist for air restarts; Pinnacle company culture and the corresponding FAA oversight capability and personality; and mechanical elements.
Among the witnesses at the June hearing were Capt. Thomas Palmer and chief pilot Terry Mefford of Pinnacle testifying about stall awareness training, company oversight and any changes since the accident; Greg Browning of Bombardier Aerospace regarding air restarts and flight tests on core lock; and Ed Orear, former CF34-3 program manager of GE Transportation for issues regarding core lock.
FAA Memphis FSDO operations supervisor Bob Cowell was there to explain principal operations inspector oversight, particularly the need for initial and recurrent type training appropriate to their specialty and relevant to their assigned air carrier.
Understanding Core Lock
On June 2, the FAA issued a Special Airworthiness Information Bulletin, encompassing research by Bombardier and GE, with methods to avoid core lock. In his letter to the NTSB, GE’s Paul Mingler defined core lock as “caused by the contact of the HPT inter-stage seal (ISS) static honeycomb component with the rotating seal teeth on the outer torque coupling, resulting in a frictional sticking (stiction) that results in the rotor being temporarily inhibited from rotating.”
“Do you have an opinion as to whether the Pinnacle Airlines accident was a core-lock event?” Hersman asked GE’s Orear.
“We don’t know. We don’t have any indications that it was a core-lock event,” replied Orear. “It’s work that we’re continuing to do.”
With or without lobbying, the Board may determine that core lock doomed engine N2 to remain at zero. When the airflow into the turbine engine is disrupted, it can cause compressor airfoils to stall, though the rotors continue to windmill due to the forward speed of the aircraft.
In a multi-spool engine, without sufficient airspeed and airflow, the inner spool or core may stop spinning. Cooling at different rates can cause different rates of contraction, resulting in binding of the core until the temperatures return to equilibrium.
Post-accident analysis showed that the cores of the two GE CF34-3B1 engines were free to rotate and there was no indication of any pre-existing problems but that engine No. 2 turbine temperature reached 1,200 degrees C, exceeding the redline limit by 300 degrees C. Engine No. 1 reached a maximum temperature approximately 100 degrees C less than redline.
Mingler said that if core lock is encountered at production-acceptance flight test, the condition is eliminated by an in-flight break-in procedure referred to as a seal “grind-in,” and that air turbine starter (ATS) torque has been shown to be adequate to overcome stiction. To overcome the limitations of a sea-level grind-in, production CF34-3A1 and CF34-3B/3B1 engines are now flown to FL310, stabilized, shut down until airspeed drifts down to 190 kias and rotation is zero, then pushed over to 320 kias to allow a windmill restart.
If the windmill does not achieve rotation, the ATS start is performed and the aircraft again climbs to FL310. Airspeed is drifted down to 240 kias to keep the core rotating at 4 percent N2.
“This grind-in procedure wears in the seals at the more aft axial location encountered during windmill operation,” said Mingler. “The windmill start procedure is repeated to confirm the core is free to rotate. To our knowledge, no engine has exhibited core lock subsequent to the aircraft acceptance process. Further, the ATS starter has successfully rotated all engines that have experienced core lock.”
If core lock was the culprit for N2 staying at zero in the accident airplane, the flight data recorder shows that airspeed was insufficient for a windmill restart. Beginning at descent through 20,000 feet the crew was deliberately decelerating to–or accidentally deteriorating toward–the best glide of 170 knots rather than the minimum of 300 knots for a windmill.
When airspeed matched the next available checklist option, an APU-assisted start at 13,000 feet, the procedure was unsuccessful after four attempts in seven minutes, given N2 below 28 percent. The NTSB is investigating whether the APU was in operating condition.
Central to the investigation is whether aggressive flight compromised safe operation to this ceiling, although it was within the [original] manufacturer and company performance envelopes.
Pinnacle Airlines has since placed a company-mandated ceiling of 37,000 feet for its CL600-2B19s. For its part, on December 14 last year Bombardier issued notice number 860 to all operators for high- altitude operations, saying that the CRJ is capable at 41,000 feet (FL410) but that a number of issues must be considered and appropriate procedures must be followed.
Bombardier advised operators to pay particular attention to weight-and-balance and ambient conditions; climb profiles in the manual; and maintaining the profile speed so as not to fall behind the energy curve. The company also advised operators to apply the appropriate engine thrust using the charts and not the FMS above 36,000 feet; to use the autopilot to accelerate to cruise speed once altitude is achieved and maintain situational awareness.
Finally, Bombardier warned operators not to wait until the onset of continuous ignition or stick shaker before attempting a descent from an altitude where continued operation is not possible.
Lorenda Ward, the NTSB investigator-in-charge, reported that the aircraft load at takeoff was 1.8gs from an abrupt pitch-up maneuver, which activated the aircraft stall-protection system. At 14,000 feet the crew engaged the autopilot for 80 seconds and changed seats.
The crew disengaged the autopilot at 15,000 feet and pulled back on the control column until the aircraft pitched 17 degrees for a load of 2.3 gs, then forward, so that the loads dropped to 0.3g. The crew then pulled back again, applying “large” rudder inputs to the left and right before again engaging the autopilot.
At 25,000 feet, the crew disengaged autopilot and pulled back then released the control column several times during the span of 30 seconds, increasing the vertical speed from 600 fpm to 5,000 fpm for several seconds before requesting 41,000 feet and climbing at 500 fpm. The airplane was at FL410 for three minutes when the right stick shaker activated, which disengaged the autopilot.
Airspeed had deteriorated continuously from FL380, to 165 kias at FL410, shortly after disconnecting the autopilot at 150 knots. Four shaker and pusher activations followed as well as an uncommanded 85-degree left bank and heading variations typical of a Dutch roll.
According to the flight data recorder, both engines stopped almost simultaneously at FL410. In short sequence, the flight crew asked for a lower altitude, declared an emergency, reported an engine failure then a double engine failure, requested a direct route to any airport, and reported the Jefferson City runway end in sight before radar contact was lost at 900 feet agl.
The crew changed seats at least twice during the accident flight, first during climb then again 12 minutes after declaring an emergency. In May Pinnacle revised its flight operations manual to mandate the captain sit in the left seat and the first officer in the right throughout operation.
The captain tried several times to transmit to Kansas City Center but his audio panel might have been in emergency mode, meaning that communications bypassed the AECU and routed directly to VHF1, which was set to the wrong ATC frequency.
On the day of the accident, the aircraft underwent maintenance to replace the 14th-stage bleed-air sensing loop on the right engine.
Several witnesses at the hearing last month suggested that training standards should be limited to the onset of stick shaker, which is a testable item, and not the use of the pusher, and whether such training ought to be performed in a level-D simulator because actual flight stalls are dangerous.
The full account of the accident and continuing updates are available at www. ntsb.gov, reference NTSB identification number DCA05MA003.