Update on 3407

Status
Not open for further replies.

SteveC

"Laconic"
Staff member
************************************************************
NTSB ADVISORY
************************************************************

National Transportation Safety Board
Washington, DC 20594

March 25, 2009

************************************************************

UPDATE ON NTSB INVESTIGATION INTO CRASH OF COLGAN AIR
DASH-8 NEAR BUFFALO, NEW YORK; PUBLIC HEARING SCHEDULED

************************************************************

In its continuing investigation into the crash of Colgan Air
flight 3407 in Clarence Center, New York, the National
Transportation Safety Board has released the following
factual information.

On February 12, 2009, about 10:17 p.m. Eastern Standard Time
(EST), a Colgan Air Inc., Bombardier Dash 8-Q400, N200WQ,
d.b.a. Continental Connection flight 3407, crashed during an
instrument approach to runway 23 at the Buffalo-Niagara
International Airport (BUF), Buffalo, New York. The crash
site was approximately 5 nautical miles northeast of the
airport in Clarence Center, New York, and mostly confined to
one residential house. The 4 crew members and 45 passengers
were fatally injured and the airplane was destroyed by
impact forces and post crash fire. There was one ground
fatality. Night visual meteorological conditions prevailed
at the time of the accident. The flight was a Code of
Federal Regulations (CFR) Part 121 scheduled passenger
flight from Liberty International Airport (EWR), Newark, New
Jersey to Buffalo.

The NTSB has voted to conduct a public hearing on this
accident. The hearing, which will be held May 12 - 14,
2009, at the NTSB's Board Room and Conference Center in
Washington, D.C., will cover a wide range of safety issues
including: icing effect on the airplane's performance, cold
weather operations, sterile cockpit rules, crew experience,
fatigue management, and stall recovery training. The public
hearing is part of the Safety Board's efforts to develop all
appropriate facts for the investigation.

"The tragedy of flight 3407 is the deadliest transportation
accident in the United States in more than 7 years," Acting
Chairman Mark V. Rosenker, who will chair the hearing, said.
"The circumstances of the crash have raised several issues
that go well beyond the widely discussed matter of airframe
icing, and we will explore these issues in our investigative
fact-finding hearing."

The hearing will be held "en banc," meaning that all Members
of the NTSB will sit on the Board of Inquiry. Parties that
will participate in the hearing will be announced at a later
time.

The aircraft wreckage has been moved from the accident site
to a secure location for follow-on inspections as may be
needed.

A preliminary examination of the airplane systems has
revealed no indication of pre-impact system failures or
anomalies. Investigators will perform additional
examinations on the dual distribution valves installed in
the airplane's de-ice system. The de-ice system removes ice
accumulation from the leading edges of the wings, horizontal
tail, and vertical tail through the use of pneumatic boots.
The dual distribution valves, which transfer air between
the main bleed air distribution ducts and the pneumatic
boots, were removed from the airplane for the examination.
The airplane maintenance records have been reviewed and no
significant findings have been identified at this time.
The ATC group has completed a review of recordings of
controller communications with the flight crew during the
accident flight and conducted interviews with air traffic
controllers on duty at the time of the accident. The group
has no further work planned at this time.

Further review of the weather conditions on the night of the
accident revealed the presence of variable periods of snow
and light to moderate icing during the accident airplane's
approach to the Buffalo airport.

Examination of the FDR data and preliminary evaluation of
airplane performance models shows that some ice accumulation
was likely present on the airplane prior to the initial
upset event, but that the airplane continued to respond as
expected to flight control inputs throughout the accident
flight. The FDR data also shows that the stall warning and
protection system, which includes the stick shaker and stick
pusher, activated at an airspeed and angle-of-attack (AOA)
consistent with that expected for normal operations when the
de-ice protection system is active. The airplane's stick
shaker will normally activate several knots above the actual
airplane stall speed in order to provide the flight crew
with a sufficient safety margin and time to initiate stall
recovery procedures. As a result of ice accumulation on the
airframe, an airplane's stall airspeed increases. To
account for this potential increase in stall speed in icing
conditions, the Dash 8-Q400's stall warning system activates
at a higher airspeed than normal when the de-ice system is
active in-flight to provide the flight crew with adequate
stall warning if ice accumulation is present.

Preliminary airplane performance modeling and simulation
efforts indicate that icing had a minimal impact on the
stall speed of the airplane. The FDR data indicates that
the stick shaker activated at 130 knots, which is consistent
with the de-ice system being engaged. FDR data further
indicate that when the stick shaker activated, there was a
25-pound pull force on the control column, followed by an up
elevator deflection and increase in pitch, angle of attack,
and Gs. The data indicate a likely separation of the
airflow over the wing and ensuing roll two seconds after the
stick shaker activated while the aircraft was slowing
through 125 knots and while at a flight load of 1.42 Gs.
The predicted stall speed at a load factor of 1 G would be
about 105 knots. Airplane performance work is continuing.
Since returning from on-scene, the Operations & Human
Performance group has conducted additional interviews with
flight crew members who had recently flown with and/or
provided instruction to the accident crew, as well as
personnel at Colgan Air responsible for providing training
of flight crews and overseeing the management and safety
operations at the airline. The group also conducted
interviews with FAA personnel responsible for oversight of
the Colgan certificate, which included the Principal
Operations Inspector (POI) and aircrew program manager for
the Dash 8 Q-400. The team has also continued its review of
documentation, manuals, and other guidance pertaining to the
operation of the Dash 8 Q-400 and training materials
provided to the Colgan Air flight crews.

The Operations & Human Performance group continues to
investigate and review documentation associated with the
flight crew's flight training history and professional
development during their employment at Colgan as well as
prior to joining the company.

Post-accident toxicological testing of the flight crew was
performed by the FAA Civil Aerospace Medical Institute
(CAMI) toxicology lab. Specimens taken from the first
officer were negative for alcohol, illicit substances, and a
wide range of prescription and over the counter medications.
Specimens taken from the captain were negative for alcohol
and illicit substances, and positive for diltiazem, a
prescription blood pressure medication that had been
reported to and approved for his use by the Federal Aviation
Administration.

The Safety Board is also examining several other areas
potentially related to the accident, including:

? The circumstances of a recent event involving a Dash
8-Q400, operated by Colgan Air, in which the
airplane's stick shaker activated during approach to
the Burlington International Airport (BTV) in
Burlington, Vermont. A preliminary review of the FDR
data from that flight shows the momentary onset of the
stick shaker during the approach phase of flight. The
airplane subsequently landed without incident. NTSB
investigators have conducted interviews with the
pilots and check airman on board this flight and will
continue to investigate the incident.

? Reports of airplane deviations resulting from
distortion of the instrument landing system (ILS)
signal for runway 23 at BUF. There is an existing
Notice to Airmen (NOTAM) related to this distortion
condition. To date, investigation into these reports
has not revealed any connection to the accident
flight.
-30-

Media Contact: Keith Holloway, (202) XXX-XXXX
 
The NTSB has voted to conduct a public hearing on this accident. The hearing, which will be held May 12 - 14, 2009, at the NTSB's Board Room and Conference Center in Washington, D.C., will cover a wide range of safety issues including: icing effect on the airplane's performance, cold weather operations, sterile cockpit rules, crew experience, fatigue management, and stall recovery training. The public hearing is part of the Safety Board's efforts to develop all appropriate facts for the investigation.

I would hope they also discuss the autopilot behavior at localizer intercept when in an approach mode. Sometimes the aircraft will aggressively climb to get on the glideslope if the deviation is not pegged or a false G/S is sensed.
 
Big Boilerplate: First, maybe this should be its own topic. Second, this is in no way intended to denigrate the memories of the dead (or living) Colgan crews. At all. Seriously. There but for the grace of God go we all, etc etc.

That said: I speak from complete ignorance on the subject, since while our planes have a GS function on the autopilot, the planes and systems are old and while we have to be checked on using the A/P for an approach, in the field you're never going to do it for reasons of not trusting equipment that's older than you are with your life. But, then, with two pilots, one of whose job is just to fly the plane...why do it at all? With the obvious exception of operations where the A/P is necessary (Cat 2 and 3, I think?), why not just hand fly as a general rule? I'm sure there is some guidance in the Opsecs or whatever the 121 equivalent is that suggests that using the A/P is a good idea, but what is the logic?

Again, this is an honest question, not an opinion or a mean spirted cut.
 
So according to the NTSB they pulled back when it stalled?

That's kind of what it looks like. Can it really be that easy to stall one of these??

NTSB said:
Preliminary airplane performance modeling and simulation efforts indicate that icing had a minimal impact on the stall speed of the airplane. The FDR data indicates that the stick shaker activated at 130 knots, which is consistent with the de-ice system being engaged. FDR data further indicate that when the stick shaker activated, there was a 25-pound pull force on the control column, followed by an up elevator deflection and increase in pitch, angle of attack, and Gs. The data indicate a likely separation of the airflow over the wing and ensuing roll two seconds after the stick shaker activated while the aircraft was slowing through 125 knots and while at a flight load of 1.42 Gs. The predicted stall speed at a load factor of 1 G would be about 105 knots.
 
Big Boilerplate: First, maybe this should be its own topic. Second, this is in no way intended to denigrate the memories of the dead (or living) Colgan crews. At all. Seriously. There but for the grace of God go we all, etc etc.

That said: I speak from complete ignorance on the subject, since while our planes have a GS function on the autopilot, the planes and systems are old and while we have to be checked on using the A/P for an approach, in the field you're never going to do it for reasons of not trusting equipment that's older than you are with your life. But, then, with two pilots, one of whose job is just to fly the plane...why do it at all? With the obvious exception of operations where the A/P is necessary (Cat 2 and 3, I think?), why not just hand fly as a general rule? I'm sure there is some guidance in the Opsecs or whatever the 121 equivalent is that suggests that using the A/P is a good idea, but what is the logic?

Again, this is an honest question, not an opinion or a mean spirted cut.

To answer your question, our FOM now prohibits us from hand flying approaches if the wx is below 1000-3 (I think...somewhere around that number). The A/P must be left on to DH even if you go visual prior to that.

Yeah, I know.

It came about as a knee-jerk reaction to the Providence, RI accident where the FO was flying a raw data ILS down to pretty close minimums, got off the GS and LOC, and the CA tried to force the airplane down. Ended up going off the side of the runway and needed a new wing.
 
Big Boilerplate: First, maybe this should be its own topic. Second, this is in no way intended to denigrate the memories of the dead (or living) Colgan crews. At all. Seriously. There but for the grace of God go we all, etc etc.

That said: I speak from complete ignorance on the subject, since while our planes have a GS function on the autopilot, the planes and systems are old and while we have to be checked on using the A/P for an approach, in the field you're never going to do it for reasons of not trusting equipment that's older than you are with your life. But, then, with two pilots, one of whose job is just to fly the plane...why do it at all? With the obvious exception of operations where the A/P is necessary (Cat 2 and 3, I think?), why not just hand fly as a general rule? I'm sure there is some guidance in the Opsecs or whatever the 121 equivalent is that suggests that using the A/P is a good idea, but what is the logic?

Again, this is an honest question, not an opinion or a mean spirted cut.


I completely understand. Well, here over at the "sister" company, we weren't exactly taught hand-fly vs no-hand-fly scenarios. It just seems to be a given that you follow the company manual and have the autopilot on at a minimum 600 feet (it's in our profile). However, for landings, visual, single engine, non-precision, and precision, no mention of clicking the autopilot off is ever made. Only in our CAT-II does it say "A/P off 80 ft AGL". So.... A/P on until ground contact???? not really. But its certainly not spelled out in the gospel.

Then again, our training dept is a lot different than the real world. Seems like the training guys were allergic to having us hand fly, or even do so without the flight director.

I remember a story from one of my buds in initial: They had disconnected the A/P and intercepted the LOC and GS for an approach, using the F/D. Well, as the approach continued, the F/D began to act funky, it started a climbing turn, so the PF said "remove my flight director" while continuing to follow the G/S and LOC indications. The instructor at that point had a fit about flying without the F/D.

I have noticed, many are allergic to hand flying over here.
 
Our manual tells us we need to be proficient with all levels of automation, which actually includes no AP, no FD. Personally, I use the AP to reduce workload or when I'm just feeling lazy. That being said, I watch the thing like a hawk b/c it tends to get a little wonky sometimes. Sometimes it'll do just fine, sometimes it'll do crazy stuff. As you gain experience with the airplane, you can spot the places and situations that you know the AP is gonna have trouble. I tend to just disconnect and hand fly if I see one of those situations creeping up.

Our AP also tends to try to climb to capture the GS. That's another one of those times when I say "Where are you going, George?" as I hit the red button.
 
Our manual tells us we need to be proficient with all levels of automation, which actually includes no AP, no FD. Personally, I use the AP to reduce workload or when I'm just feeling lazy. That being said, I watch the thing like a hawk b/c it tends to get a little wonky sometimes. Sometimes it'll do just fine, sometimes it'll do crazy stuff. As you gain experience with the airplane, you can spot the places and situations that you know the AP is gonna have trouble. I tend to just disconnect and hand fly if I see one of those situations creeping up.

Our AP also tends to try to climb to capture the GS. That's another one of those times when I say "Where are you going, George?" as I hit the red button.

Hmm well yeah there's what the manual says and what we're taught....
 
I completely understand. Well, here over at the "sister" company, we weren't exactly taught hand-fly vs no-hand-fly scenarios. It just seems to be a given that you follow the company manual and have the autopilot on at a minimum 600 feet (it's in our profile). However, for landings, visual, single engine, non-precision, and precision, no mention of clicking the autopilot off is ever made. Only in our CAT-II does it say "A/P off 80 ft AGL". So.... A/P on until ground contact???? not really. But its certainly not spelled out in the gospel.

I think that's part of the "intentionally vague" nature of our manuals. If something happens, the company can just shove the blame off on the pilot since they never really gave guidance one way or the other.

Then again, our training dept is a lot different than the real world. Seems like the training guys were allergic to having us hand fly, or even do so without the flight director.

I find that the instructors that feel this way are the ones that have never flown the airplane. Any of the instructors that actually have time in the CRJ know the AP/FD is schizo.
 
Actually our non-precision approaches do call for A/P & F/D off when you say "leaving MDA" .... :eek:

I find that the instructors that feel this way are the ones that have never flown the airplane. Any of the instructors that actually have time in the CRJ know the AP/FD is schizo.

I guess I've been really lucky in both airframes. Only one A/P disconnect not because I selected the red button. Other than that, the usual, overshooting the ILS with PRMs in progress, overcontrolling the bejesus out of the aircraft, etc.

How many times have you seen and wondered why the the A/P is pumping the controls very violently just before you click off the A/P?
 
You guys have the same planes we have right? How are your autopilots so messed up? The only time I can think where it gets confused is when you turn to intercept a localizer on a specific heading. It tends to stop the turn prior to getting on the 30* intercept the controller wants you on. That's about it. Other than that it's pretty flawless in my 3000 hours in the plane.
 
You guys have the same planes we have right? How are your autopilots so messed up? The only time I can think where it gets confused is when you turn to intercept a localizer on a specific heading. It tends to stop the turn prior to getting on the 30* intercept the controller wants you on. That's about it. Other than that it's pretty flawless in my 3000 hours in the plane.

The airplane flies a lot better with the A/P deferred.

Seriously - the CRJ900 has a .82 mach limitation in RVSM because of the A/P. Even at <.82, the A/P has oscillated and shown +-600 FPM attempting to maintain level flight. Luckily these changes happen so "fast" we've only been 150 feet off at the most. Ugh!
 
The data indicate a likely separation of the airflow over the wing and ensuing roll two seconds after the stick shaker activated while the aircraft was slowing through 125 knots and while at a flight load of 1.42 Gs. The predicted stall speed at a load factor of 1 G would be about 105 knots.

Let's see, stall speed at a load factor of 1.42 would be sqrt(1.42)*105 = 125 knots.
 
I know airline training can sometimes "cloud" basic airmanship with all of the procedures and flows. I know my training at PDT became somewhat overwhelming and I had to constantly remind myself of basic airmanship, especially when it came to flying with the autopilot so much.

I can't for the life of me understand why a pilot, when confronted with stick shaker and stall warning would pull up on the controls. As a CFI I tought and had it drilled into me, to push to recover from a stall. You must reduce the AOA.

Maybe the pilot mistook the stall warning for ground prox warning and he thought he was too low? Possible when the pilot had such low time in the aircraft?

When I flew the Dash-8-100/200/300 I had to remind myself all the time to not rely on the automation 100%, but to be ready to hand fly, and to remember the fundamentals, as a new FO on the airplane.

I'm sad to see this accident looks as though it was pilot error, and an error that goes back to the basics of flying an airplane.
 
Maybe the pilot mistook the stall warning for ground prox warning and he thought he was too low? Possible when the pilot had such low time in the aircraft?

When I flew the Dash-8-100/200/300 I had to remind myself all the time to not rely on the automation 100%, but to be ready to hand fly, and to remember the fundamentals, as a new FO on the airplane.

I'm sad to see this accident looks as though it was pilot error, and an error that goes back to the basics of flying an airplane.

How does one confuse "CAUTION, TERRAIN", "TOO LOW, TERRAIN" or "TERRAIN TERAIN, PULL UP" with the shaker..... =(

I sure hope not. I think pilot-error is a cop-out though. The industry won't get any safer that way.
 
I know airline training can sometimes "cloud" basic airmanship with all of the procedures and flows. I know my training at PDT became somewhat overwhelming and I had to constantly remind myself of basic airmanship, especially when it came to flying with the autopilot so much.

I can't for the life of me understand why a pilot, when confronted with stick shaker and stall warning would pull up on the controls. As a CFI I tought and had it drilled into me, to push to recover from a stall. You must reduce the AOA.

Maybe the pilot mistook the stall warning for ground prox warning and he thought he was too low? Possible when the pilot had such low time in the aircraft?

When I flew the Dash-8-100/200/300 I had to remind myself all the time to not rely on the automation 100%, but to be ready to hand fly, and to remember the fundamentals, as a new FO on the airplane.

I'm sad to see this accident looks as though it was pilot error, and an error that goes back to the basics of flying an airplane.

As crazy as it is, our training calls for full power and to pull back on the stick to minimize altitude loss.
 
How does one confuse "CAUTION, TERRAIN", "TOO LOW, TERRAIN" or "TERRAIN TERAIN, PULL UP" with the shaker..... =(

I sure hope not. I think pilot-error is a cop-out though. The industry won't get any safer that way.

Well, sure it does.

If it wasn't for the errors of our peers we wouldn't have certain training programs and procedures that protect us from performing the same "errors."

But, yes, pilot error is usually a cop-out and is joyfully praised by aircraft manufactures.
 
Status
Not open for further replies.
Back
Top