A320 - Loss of Separation departing KMSP

[mm43]

Press Release by NTSB

On September 16, 2010, about 6:49 a.m. CDT, US Airways flight 1848 (AWE 1848), an Airbus 320, was cleared for take off [Minneapolis] on runway 30R en route to Philadelphia, Pennsylvania, carrying five crew members and 90 passengers. At the same time, Bemidji Aviation Services flight 46 (BMJ46), a Beech 99 cargo flight with only the pilot aboard, was cleared for takeoff on runway 30L en route to La Crosse, Wisconsin. Weather conditions at the time were reported as a 900-foot ceiling and 10 miles visibility below the clouds.

Immediately after departure, the tower instructed the US Airways crew to turn left and head west, causing the flight to cross paths with the cargo aircraft approximately one-half mile past the end of runway 30L. Neither pilot saw the other aircraft because they were in the clouds, although the captain of the US Airways flight reported hearing the Beech 99 pass nearby. Estimates based on recorded radar data indicate that the two aircraft had 50 to 100 feet of vertical separation as they passed each other approximately 1500 feet above the ground.How close can you get?!

mm43

[ATC Watcher]

the captain of the US Airways flight reported hearing the Beech 99 pass nearby

Hard to beleive, but if that is true, (presurized a/c , during take off power and pilots on headsets) it must have been really close indeed.

[SKS777FLYER]

How close can you get? The gap can be closed until the near hit, becomes a near-miss.

[p51guy]

That had to be close if he heard the other AC. I have had some close ones within 100 ft and never heard it. The closest two were within 50 ft, one when I was head to head with an aircraft in my Cessna 340 and rolled inverted over Long Beach keeping my wings parallel to his to clear him and the other as a passenger on an MD80 in the front right seat in back watching a commuter twin coming right at us off LAX and passing less than 50 ft below our nose. As it passed the left side a passenger jumped seeing it barely miss the left wing. I knew the captain of that flight and had already warned my daughter flying with me how he would report traffic in sight so he didn't have to level out. If he heard the aircraft he is lucky to be alive.

[HEATHROW DIRECTOR]

I'm curious to know if each of those runways has a separate controller. If they do, it sounds like a liaison error of some sort.

[andrasz]

I'm wondering what TCAS displayed to both as the event was unfolding ... ?

[Hotel Tango]

Hard to beleive, but if that is true, (presurized a/c , during take off power and pilots on headsets) it must have been really close indeed.

I don't think the crew would make it up. The A320 is relatively quiet up front and you would certainly hear a turboprop on climb power 150 feet away. Furthermore, if memmory serves me right, not that many US crews operate with headsets on.

[BandAide]

KMSP has three separate gound control sectors and three separate tower frequencies. They are usually consolidated during low volume times.

At the reported time of the incident, just before the morning rush, I would think all facilities would have been split to their individual operations, each with its own controller.

I can only speak for my airline, but we must wear headsets below FL 180 on the A320.

[misd-agin]

Headsets - boom mike's required below 18,000'. Dave Clark style headsets covering your ears? No. Almost every guy has a custom molded ear piece(hearing aid type) that mounts to a thin boom mike.

Heard a plane go overhead once. Even with ear plugs and a helmet on. 1 v. 1 against an F-15. Granted he was in full afterburner so it's a bit louder. 1000+ knot closure. Seperation criteria? "We'll be level, don't hit us." :)

[Herod]

Many years ago the military were evaluating fighter-evasion tactics for helicopters (yes, really), which involved the second pilot of the helicopter recording a running commentary onto a tape recorder. At one debrief the crew just set their recorder to "play". You could clearly hear the sound of the jet, as picked up by the throat mic. Now that's close!

[DingerX]

They were on separate frequencies, but a different problem.

off the Liveatc Scanner (KMSP Twr 16 Sep. 2010 1130Z -- it records 26:24 of time, so I suppose something's wrong with the encoding or my file player). Anyway, at about 12:18 on my player I hear this.
(clipping makes it hard to get the airborne part of the last few conversations, and for all I know that could be bemidji with the "I've got traffic" call)

TWR1: Cactus 1848, runway 30 Right position and hold
AWE1848: Position and hold on uh 30 Right Cactus 1848
TWR1: Cactus 1848 requesting base and tops and you can pass to us on the departure RWY 30 fly runway heading cleared for takeoff
AWE1848: runway heading cleared for takeoff give you the base and the tops Cactus 1848
TWR1: Bemidji 66 heading 300 contact departure
TWR2: Bemidji 46 turn left heading 180 runway 30 left cleared for takeoff
BMJ46: 30l cleared for takeoff
~2.5 minutes later
TWR2: Bemidji 46 are you in the turn?
BMJ46: Say again, the turn?
~1 minute
AWE1848?: Guys, that ain't cool, I've got traffic
~40 seconds
BMJ46: Tower Bemidji 46 going to departure.
TWR2: departure
BMJ46: Roger uh I just (got an airprox*)
TWR2: Okay, um, why didn't you start the turn when you were airborne?
BMJ46: Tower, I got ** sorry about that.
TWR2: Okay, Bemidji 46 contact departure
BMJ46: *

[pattern_is_full]

So an echo of the collision over the Hudson: pilot failed to copy the turn instruction, and tower failed to notice the omission on readback.

[Mr.Bloggs]

Poor airmanship. Single-pilot commercial ops in busy airspace. This happens a lot. Nearly lost myself and 345 others near Daytona. RA. 250 feet vertical and lateral separation estimated by colleague looking. (Not myself, obviously). Similar amateur flying by a small a/c.

[skysign]

That morning the ground controller for that side and tower for 30L was the same controller.
The 99 did not comply with the left turn to 180, but then his readback did not mention the turn. Why he did not turn ? is pure speculation at this time. ( did not ear, forgot,malfunction of some sort, waiting to reach a certain altitude before turning,...???)

What is sure is the controller was busy giving instruction to different A/C on the ground after the 99 T/O. And therefore did not realise that the 99 did not comply to his instructions and reissued the turn clearance.

Who is to blame ?
Pilot maybe ? ( did not turn why ? )
Controller maybe ? ( fail to catched wrong read back, did not realised that 99 did not turn ?

MSP at 0650am is already a very busy airport, having ground & tower and some time metering done by the same controller, is in my opinion is a perfect set up for the controller to miss something at one point or an other ( task saturation...).

Proper staffing is the guilty one, just my 2 cents.

[cactusbusdrvr]

I only fly westbound out of MSP and we normally are given a left turn to 270 off of 30L. Right side (30R) traffic is runway heading or if we get a left turn it is because there were not any simultaneous departures.

All the controller had to do to ensure separation would have been to give a left turn to the 30L departure and a right turn to the 30R departue. Normally MSP tower gives you the turn along with the T/O clearance. Chances are both would comply but if one did not then there would still be separation.

Single pilot IFR is common in the US, especially in light cargo ops. That is the best flying you will ever do and you will learn a hell of a lot more than sitting in a observer seat as a cadet for years before you get to fly a jet on autopilot. Don't apply EU perceptions to US operations.

[KKoran]

The incident supposedly happened 1/2 mile off the end of the runway. For a controller to assume that a departure would be turned by then is lunacy.

[HEATHROW DIRECTOR]

KKoran. I do not agree. Depends how long the runway is, a/c type, rate of climb, etc. I've seen aircraft reach 6000 ft before the end of the runway so such a general statement is wrong.

[aterpster]

Heathrow Director:

KKoran. I do not agree. Depends how long the runway is, a/c type, rate of climb, etc. I've seen aircraft reach 6000 ft before the end of the runway so such a general statement is wrong.

6,000 feet before the end of the runway? It must have been a military fighter.

For a controller to assume what height a transport aircraft will achieve by the departure end of the runway is the height of folly; that is, unless the controller has the weight & balance information for each flight transmitted to the tower, has precise performance data available for the aircraft, and has some mystical assurance there won't be an engine failure after V1.

That is what it would take to remove the ass-umption factors.

[ATPMBA]

How about controllers getting 1,500 hours of closely supervised work before they gt turned loose? And if the fail two performance reviews they get the boot.

[Oakape]

6,000 feet before the end of the runway? It must have been a military fighter.

I once crossed the upwind end of a 2,800m runway at 5,000' AGL in a B767-200. So no, it wasn't necessarily a military fighter!

[Two's in]

If positive separation was determined by the single pilot ops making the left turn after take-off, the controller missed the first opportunity to avoid this when BMJ46 failed to read-back the turn instruction before take off, and then again when 2:30 elapsed before he queried the lack of turn. With parallel take-offs it seems that the positive clearance mechanism (one of them turning immediately) was not positive enough, especially given they were IMC.

Add to that split tower frequencies, so they were both blissfully unaware of the other's intentions and it's another classic, "it used to work fine right up until the big fireball" scenario. Even a basic risk assessement would identify positive acknowledgement of the clearing turn as an absolute minimum requirement before issuing the take off clearance. I'm sure it was, but in this case it got missed and a lot of people nearly had a bad day.

[Mr Optimistic]

Out of hours so had to nightstop before flying back the next day - empty flight and achieved 5000ft by end of runway having some fun!

[aterpster]

ATPMBA:

How about controllers getting 1,500 hours of closely supervised work before they gt turned loose? And if the fail two performance reviews they get the boot.

How about semi-annual recurrent training and pass-or-fail proficiency checks?

One they are journeymen there is no more formal training. Most of the training after initial qualification is of the "frick teaches frack" variety.

[aterpster]

oakape:

I once crossed the upwind end of a 2,800m runway at 5,000' AGL in a B767-200. So no, it wasn't necessarily a military fighter!

Once isn't the norm.

Crossing the end of the runway at 5,000 feet, agl, isn't the norm either.

I am sure you had fun, though.:D

[HEATHROW DIRECTOR]

<<6,000 feet before the end of the runway? It must have been a military fighter.>>

Learjet.

<<How about controllers getting 1,500 hours of closely supervised work before they gt turned loose? And if the fail two performance reviews they get the boot.>>

In the UK controllers get a hell of a lot of time fully supervised before they take exams to go solo. They're also subject to regular checks, just like pilots.

[PaperTiger]

Add to that split tower frequencies, so they were both blissfully unaware of the other's intentions and it's another classic, "it used to work fine right up until the big fireball" scenario.Which will be the only thing to wake up the FAA :mad: .

Not like it hasn't (nearly) happened before (BOS, ORD, JFK etc.) as google will attest.

[HEATHROW DIRECTOR]

<<Add to that split tower frequencies, so they were both blissfully unaware of the other's intentions and it's another classic>>

I can't speak for the USA, but in the UK different frequencies are used for different runways at busy airfields. The controllers are sat close to each other and they work as a team, not "blissfully unaware" of what the other guy is doing..

[LN-KGL]

I'm not a pilot and nor do I work in air traffic control, I'm just a regular aviation enthusiast with a safety background. I find it interesting that the air traffic controllers at MSP assign headings for each flight and not use the powerful safety tool of standard departure patterns out from their runways. These standard patterns are much more thought through than any heading command given orally from an air traffic controller to a pilot, and in addition with no risk of miscommunication as it clearly was in this close call.

All departures from my local airport (not that much smaller than MSP - two runway, mixed operation, will probably end up around 20 million passengers/220,000 movements this year) follow these standard patterns. As an example we got a clearance for TOMBO 5A departure from runway 01L in February 2006 (I say we because my boarding card said seat 0C for this short flight from OSL/ENGM to TRD/ENVA). Below is a short description of a TOMBO 5A departure (the flight is initially cleared for 7,000 feet):

Climb on RWY track 015° to 1200 FT, then intercept and follow R-015 GRM to DME 4.5 GRM, then turn left and track 348° . At DME 6 GRM, turn right and track 053° to intercept and proceed on QDR 003° BGU to TOMBO.
FMS/RNAV (overlay): GM381 - GM383 - On track 053° to GM387 - TOMBO.

Most operators at our airport has pre-programmed these departures in their FMC/FMS, and the picture below I took as we were DME 5.3 GRM at 4,000 feet with heading switching between 350° and 349°. Both primary flight displays (PFD) show the TOMBA 5A pattern as a solid magenta line.

http://www.plane-spotter.com/Jpegs/OSL-TRD-OSL/SK4152/15-37-08.jpg

How come that many US airports don't use standard departure patterns or even standard climb out direction to increase separation between aircraft to a safe/acceptable distance?

[pattern_is_full]

Well, many US airports do have SIDs (standard instrument departures) routes.

Standard instrument departure - Wikipedia, the free encyclopedia (http://en.wikipedia.org/wiki/Standard_instrument_departure)

However:

a) note that some SIDs are "vectored" - i.e. under verbal ATC instructions as appears to have been the case for both craft in this event

b) if the pilot misses the instructions, it doesn't really matter if the instructions were "turn left to 180" or "cleared ALPOE 6 departure" - (s)he is still going to head off in the wrong direction.

c) if the SID that works for a given final course (e.g. eastbound) is the same for two planes being cleared for take-off at the same moment from different runways - they may still end up in close proximity.

While I understand the time pressures for both pilots and controllers at a hub like MSP during a "rush hour," I've got to question trying to cram too many instructions into one radio call.

It may add 15 seconds to each "movement" to say "Turn left 180 after liftoff" - and then wait for a positive readback - before than adding "Cleared for takeoff 30L." And those 15 seconds add up if there are 35 planes trying to depart in, say, one 20-minute window. But it may better ensure understanding.

The essence of communication is not what you say (and how fast you can say it) - but what the other person hears and understands.

[aterpster]

http://i201.photobucket.com/albums/aa214/aterpster/ATLRNAVSIDAdvisory-1.jpgLN

http://i201.photobucket.com/albums/aa214/aterpster/ATLRNAVSIDAdvisory-2.jpg

[BOAC]

Looking at LN-KGL's pic, I ask what had they 'backed up'?:)

[PENKO]

BOAC, they have the VOR needles pointing to different VOR's, so they're monitoring as far as I can tell! Let LNAV do the rest. :E

[DFC]

I can't speak for the USA, but in the UK different frequencies are used for different runways at busy airfields. The controllers are sat close to each other and they work as a team, not "blissfully unaware" of what the other guy is doing..


Indeed, and at Heathrow where they use one for landing and the other for departures they still managed to get a departure very close to a missed approach on the other runway.

What is different is that the result was a a long overdue overhaul of the missed approach procedures etc. I don't think much is going to be done here except blame someone for making a mistake.

In the absence of obstacles, most operators will have an SOP to fly straight ahead if there is an engine failure. I have not come across a situation where this changed because there was a departure on the other runway that the crew are not aware of.

This is just as likely to have happened to a two crew heavy as it is to a single pilot light operation. The requirement to have two crew (or more) in certain aircraft is not in any way linked to the flight rules they operate under or the classes of airspace they fly in.

In the end, until ATC receive a correct readback of the clearance issued then they can't rely on the pilot to do what they expect. This is becoming a more and more frequent issue - London City dep vs Heathrow Arrival - and many more.

It is not the training hours that controllers receive that is an issue. Yes they are very small - typically (in very round terms) about 200 for an aerodrome controller and about twice that for radar and I would think that 600 is the absolute maximum before the ATC provider cuts their losses (and this is an important issue).

ATC training has low minimum hours that have stood the test of time (and traffic expansion) however, it is one area where ability is the determining factor in initial qualification rather than simply hours in a logbook (experience). I do wonder if the same system applied to flight training how many people would not be flying?

The call for better recurrent training and testing in ATC may be a good one. However, if ATC training starts to follow the Pilot version and starts giving people 1500 hours experience and a simple couple of tests to become qualified then it would be a severe retrograte step.

[LN-KGL]

BOAC
Looking at LN-KGL's pic, I ask what had they 'backed up'?PENKO
BOAC, they have the VOR needles pointing to different VOR's, so they're monitoring as far as I can tell! Let LNAV do the rest.Norway isn't that saturated with VORs and the country isn't as flat as UK. For the flight to Trondheim only one other VOR are available en route (TRM just north of Trondheim Airport and then you have to fly for around 100 nautical miles to get within range - the distance between both airports is 196nm). Therefore both VOR1 and VOR2 are pointing to GRM as all waypoints within Oslo TMA and Oslo Sector North refer to distance/angle from GRM DVOR/DME (except waypoint GM401).

Here are the SID patterns used at OSL with winds from the north:
http://www.plane-spotter.com/Jpegs/OSL/SID_01L.jpg

http://www.plane-spotter.com/Jpegs/OSL/SID_01R.jpg

To see a Thai A346 do the OPA 3B departure is just fantastic - a striking similarity to the rwy 06 departure at the old (and now closed) Oslo Airport, Fornebu.

[BOAC]

Penko - another look perhaps? Perhaps also explain the 016+ADF on both. Not the way I would do it. COTML?

LN- thanks - I have the chart.

[LN-KGL]

COTML? Could you explain BOAC?

[BOAC]

Children of the Magenta Line:)

[PENKO]

BOAC, not exactly shure what the problem is. They have VOR selected and identified, no idea about their ADF, but at least they are monitoring the initial part of the departure. And what's wrong with following the magenta line...it's an RNAV departure anyway, you might have no choice!

[LessThanSte]

I've seen aircraft reach 6000 ft before the end of the runway

did you also use to work at Cape Canaveral?