Mid-Air Over Germany

Comment

a.  No foul play could have been involved. It's just that there are some poor English speakers and dyslexic pilots out there. With the .005khz spacings you can now have six digit frequencies to remember and dial in. (eg 126.85Mhz may now have become 126.825Mhz). Easy to transpose the latter three digits. However that's why it is always SOP to initiate a call to your new agency after a check-in. The possibility is that the Russian pilot may have mis-dialled into an active frequency and was either creating confusion on there, awaiting a chance to check in, or just taking some time to sort himself out. The only safe way is to grease-pencil the new freqs on the perspex as you hear it. There's a very simple explanation why he failed to respond to the earlier repeated calls from ATC. He just wasn't on that frequency. Usually these type mistakes, which are very common, are encountered in an approach or departure climb type scenario - where they used to be much more hazardous. But now in upper airspace Man's ingenuity has created greater hazards (see below)

b.  And of course they were up there at 36,000ft (Flt Lvl 360) in RVSM country (Reduced Vertical Separation Minima where aircraft previously had to utilise 2000ft vertical separation for opposite direction or crossing traffic - because of the accuracy limitations of the lowest common denominator barometric altimeters and autopilots).

c.  A method that used to be used to avoid this sort of fiasco was for the military to utilise 243.00mhz (International Distress UHF) and civilian aircraft to listen out on a silent common frequency of 121.5.00 (half of 243Mhz) - which is the civil VHF distress frequency. However because of the traveling goat acts that abuse silent frequencies, there is never any guarantee that someone who ATC calls in an emergent situation actually IS going to be listening out on that silent frequency.

d. So why did this automatically have to be a disaster? Pilots have been saying for ten years that the ability of modern navigational kit to accurately track the centre-line of airways was always going to cause this sort of NOCOMM collision - because the high closure speeds defeat visual acquisition, TCAS and the ability to think, realize, advise and react. With modern GPS-equipped aircraft, every time we go flying we go either directly over or under other traffic; and the big sky principle no longer applies. Airways are 10 nm wide, so why don't regulators require all operators to offset 1 mile right (or even 100m) in the cruise phase to reduce the chances of this unfortunate accident re-occurring. Even in RVSM airspace, there is sufficient space to allow this principle to be applied until the aircraft begins descent. ICAO has been "looking into it" for over ten years. The problem is described here and here.

e. Once placed in jeopardy by the situation described in d. (above), a single-party TCAS solution is still open to the variability and unpredictability of the other aircraft's last second flight-path changes (based on misinterpretation of a last-second visual acquisition or ATC urgent instruction). That seems to have been the nightmare scenario for this accident.

initial discussion

see below

ongoing discussion

see here
From a Pprune Posting

From what I can gather, the DHL pilot was responding correctly to TCAS RA to fly down. Now if the TU154 was in RVSM, shouldn't it have had TCAS installed as a requirement?; so shouldn't the two TCAS computers have co-ordinated the avoidance action?

In any case, it is an accident which should have us all reflect that even in one of the best controlled airspaces with all the modern airborne equipment etc. we still need to be scanning that sky or at least scanning the ND continously for suspect TCAS targets in order to have as early a warning as possible of potential threats.

Another thing which I hope will *not* happen after this accident is that pilots will become wary of following TCAS RAs if it transpires that the DHL pilot still got himself into trouble by following the TCAS instruction. We could end up in the same trap where pilots "analyse" GPWS warnings during an approach.

I did have a TCAS RA once during cruise (FL350, European airspace) and since it was an opposite (uncontrolled!) fast military traffic, our TCAS didn't have time to give a "TRAFFIC, TRAFFIC" warning and we were issued with the "DESCEND, DESCEND" warning immediately, accompanied by the controller "screaming" a lateral avoidance vector at the same time.
 

The reluctance of some pilots to appropriately respond to avoiding action has already been discussed at length here

 

As an ATC, the following things make me wonder....

*1.
If the aircraft were at the same level inbound from the Italian and German sectors respectively, then surely the Swiss sector should have co-ordinated a different level for one of them prior to entering the sector (less than 10 minutes before impact)?

*2.
Does the Swiss Sector have MTCD (Medium Term Conflict Detection) which would give about 10 -15 minute warning of conflict? If so, what was done with this MTCD information?

*3.
Surely it is not normal to leave a separation instruction so late that a non-compliance immediately causes a bad situation to turn into a desperate disaster..

*4
At what point should ATC leave the avoidance to the respective TCAS's, and

*5.
Will pilot's respond to ATC or TCAS as priority ?

These are questions that are raised in my mind as an ATC. No intent to blame anyone..
1. The entry levels via those two points were correct for the direction of flight of the two aircraft. The 757 , via SRN, would have been on the freq. for quite a wee while before the T154 in at KPT.

2. Yes, Zurich does have Short Term Conflict Alert , but it seems it was off for maintenance. Usual procedure to do so during the night shift.

3. I also feel very uncomfortable about the ATCO waiting approx. 1 min. prior to "loss of separation" before instructing an aircraft to change level. Compounded by the fact the pilot did not immediately respond/comply. It's a sad reality that pilots do not always do what we tell them to do immediately.

But what is emerging and is of real concern regarding TCAS is that the DHL TCAS " Descend" command came AFTER the Tu154 started to comply with the ATCO's instruction to descend to FL350. Why would it give a descent command against conflicting traffic already descending ?
TCAS alerting is based on prediction of a 'Closest Point Of Approach' (or CPA).

If the predicted CPA is too small (horizontal and vertical values depend on Altitude - the higher you are the bigger the predicted 'miss' must be for TCAS to consider it safe) then a Resolution Advisory (RA) will be generated once the aircraft are within the RA Alerting Time.

This value also varies with altitude, from (if I recall correctly - manual not to hand) some 15 seconds at low levels to 35 seconds at high level.

The RA is preceded by a Traffic Advisory, 10 - 15 seconds earlier, depending on altitude.

Since it is a time parameter, a 'nose to nose conflict in fast jetliners' is not an issue. (i.e. the greater the rate of closure, the greater the distance at which the alerts are generated.)

However, the increased time allocated to Alerts at high level does not translate into more time for 'assessment and reaction'. The reason more time is given is because:



a) Altimeter inaccuracies mean that a bigger vertical miss must be provided by TCAS to be sure of safety. I.E. aircraft could be at the same level even though in reported altitude terms they could look like they are a couple of hundred feet apart. Hence the safe separation required is effectively:

Worst Case scenario of altimeter inaccuracy, X 2, + required actual separation.

b) High altitude make the aircraft less maneuverable (IAS / TAS relationship).



The alerting times are selected so that the aircraft has enough time to go from predicted co-altitude with intruder to safe separation, plus 5 seconds of reaction time for the crew.


As another point, there are various eventualities that can delay or even prevent the TCAS response. However to go into them here would be verging on speculation.


Typically, there are two kinds of TCAS advisories, TA - Traffic Advisory and RA - Resolution Advisory.

Traffic advisory TA, is activated when another aircraft is about to intrude the TA protective envelope within approximately 40 sec. The envelope is 0.75 nm radius and 850 ft above and below your a/c. The TA will provide both visual and aural cues, ie the display will turn the color of the traffic from white to amber with the aural warning, i.e "Traffic, Traffic"

The Resolution advisory RA, is activated when another a/c is about to intrude the RA protective area within approximately 20 sec. The RA protective envelope is 0.30 nm and 600 ft above and below your a/c. The color of the symbol will now turn red and the aural warning and the maneuvering command will be displayed on the VSI with the aural warning, i.e "Climb, Climb Now!" "Climb, Climb Now!" Compliance with the commanded maneuver is required with 5 sec (2.5 for enhanced RA) of the aural warning with at least +/- 0.25 G maneuver.

"Clear of Conflict" will be announced once the traffic is well cleared.

There is a new version of TCAS called TCAS VII that is not widely used or required but it provides both vertical and lateral RA.

Notes:

1.  TCAS will not be able to detect a/c that do not have operating transponder.

2. They will not provide RA against non-altitude reporting transponder.

3.  It has an automatic function to make coordination with the other TCAS-equipped a/c in case of RA so that both a/c will not move in the same direction.

What is described above is how a TCAS II would operate and it is believed that the B-757 was probably equipped with such a system. No idea what was on the Tupolev, if any...


Quest:  Why does TCAS make you go down instead of turning left/right to avoid other aircraft?

Answer:  Because TCAS uses Mode 'c' to arrange deconfliction. At the moment it does not have sufficiently good lateral resolution to ensure avoidance.

Swiss ATC media conference just held

Today’s mid-air collision over Southern Germany is, according to Skyguide (the Swiss ATC), the result of simultaneously initiated descents by both the aircraft that eventually collided. In the case of the Russian Tupolev TU-154, the descent was initiated based on ATC instructions, whereas the descent of the B-757 freighter was triggered by its TCAS.

Anton Maag, Chief of Zurich-Kloten Control Tower, and Skyguide Spokesman Patrick Herr, held a press conference today, Tuesday, to review the events of the previous night.

Between 2300 and 2400, there were no unusual occurrences in Swiss air control space, with only a few aircraft in the air. Swiss ATC was handed over the Russian TU-154, which was on a East-to-West routing by German ATC at around 2330, whilst the B-757 freighter, on a South-to-North routing, was handed over by Italian ATC to Swiss ATC at 2323. Both planes were at an altitude of 36000 feet (around 11500 meters).

The Swiss controller thus advised the Russian aircraft to descend to a lower flight level. According to Skyguide’s Anton Maag, it required three attempts before this advice was confirmed by the Russian crew. As a result, the Russian aircraft began its descent very late. At the same time, the crew of the B-757 freighter was advised for reasons unclear by its TCAS to descend as well. Obviously, according to Maag, this command has to be followed immediately.

The eventual collision happened at an altitude of 35000 feet between 2335 and 2336. Both aircraft remained visible for a number of sweeps on the radar screen. Maag says there were no linguistic problems between the Swiss controller and the Russian crew, whilst Herr mentioned two “sticking points” in the context of this crash. “First, why did the Russian pilot not react immediately? Second, why did the TCAS on board the B-757 advise its crew to dive?” So apart from the question as to why the TCAS on board the B-757 did not recognize the other aircraft's descent, both aircraft were tuned to the same radio frequency, so they could both simultaneously hear their respective interaction with Zurich-Kloten tower.

 

The answer is that if the Tu-154 was not TCAS-equipped, the 757 TCAS instruction would have been to a predetermined schedule and not an absolute determination computed from both aircraft's TCAS interaction....and would therefore seem to be a basic failing of TCAS. In other words a single party TCAS RA might lead you into danger rather than out of it when the other party is late responding to ATC instructions.

It sounds very much as if the Tu-154 was avoiding visually in response to ATC and the 757 was following the TCAS instructions for a singe station RA (Resolution Alert). Probably one (or neither?) aircraft turned right (the old rules).

Since Swiss ATC have put their side, I feel the following thoughts:
1. 8-10NM (at 90deg) is far too late to describe as even "less than ideal". As someone else said, it will be interesting to see what urgency prefixes were applied to the calls.
2. A Plan B is required here - the second call maybe, and certainly the third call should surely have been to the DHL ac? There are a number of legitimate reasons why the Russian aircraft did not, or could not respond. There is, for instance, a known issue with some radios in our fleets where they "go to sleep" and aircraft have not responded for minutes to call. Advice to us - if you don't hear something for a long time (which would be along time at night), click the Tx key...
3. IMHO, the Russian pilot could only be "blamed" if he read back the instruction, and only THEN did not comply.
4. Leaving it so late to start avoiding action, not only increases the possibility of collision by no response (which did not happen here), but TCAS getting involved.
5. Hard to blame TCAS ?? Basically, the 2 anti collision systems (ATC and TCAS) were both "working" in a "just in time" manner - and unfortunately took the 50:50 worst case. TCAS is designed to be a backstop to ATC, and hence is "just in time". ATC should not be...

I do not like pointing the finger, but am very concerned by Swiss ATC throwing the mud. IMHO, in that they have started throwing it, its alright to throw some back...
 
ISSUES for Consideration
  • Respond to TCAS resolution advisories only to the extent to which the cues are provided on your EADI or EVSI. An RA will often require no more than 1500 fpm climb or descent to deconflict the traffic. At high Mach numbers and flown manually, the pitch change is remarkably small. A panic large input may cause additional problems or even an upset.
     
  • Resist the temptation, especially at night, to try and second-guess the TCAS RA. In RVSM, an aircraft 1000 feet below you might look as if it’s actually above and vice versa. Trust the TCAS and use it to enhance your mental model of your bit of airspace. You may think you’ve worked out the best avoidance, but the TCAS has a contract with the conflicting traffic and has worked out the best avoidance manoeuvre. Remember, the other target may be simultaneously involved in avoiding another aircraft.
     
  • If your aircraft is fitted with an early type of display, remember that the azimuthal error in TCAS can be up to 20º, which means that the intruder could be either left of the nose or right of the nose. TCAS is designed to provide only vertical deconfliction, so again don’t try to second guess it.
     
  • Listen out on your frequency, try and plot other aircraft in your mental model, and think about where conflicts might arise. If necessary inform ATC of any concerns you have about another aircraft. They will be more than happy to put your mind at rest.
     
  • Make sure that at least one of you is monitoring 121.5 at all times. Someone may have urgent news for you, and you may not be talking to them at the time
Pprune Posting:

One of our 737s returning from bjv was in the vicinity at the time.
The capt has told me that he heard an American voice suddenly
calling TCAS DOWN, and looking to their right, saw the fireball
at around 36,000 feet, lighting up the sky all around.
Presumably this was the 757 calling.
Both crew a little shaken to have seen it.

The B757-200SF was reg: A9C-DHL, c/n 22175 - on flight DHX611. ( a PF operated by DHL Bahrain BAH/BGY/BRU. (Bahrein/Bergamo/Brussels))

The DHL plane had a TCAS collision avoidance system on board, which U.S. regulators require for passenger aircraft and have ordered phased in for cargo planes.

Tupolev registration is RA-85816
The Russian jet was headed to Barcelona, Spain, from Moscow, originating in Ufa, Russia, and had just made a stop in the German city of Munich.

I doubt that the T154 had RVSM compliance and BOTH aircraft require compliance to use 1000' separation above FL290. However it appears both were at the same cleared level. One of the a/c (probably the TU134) had just passed from Munich to Zurich ATC prior to the collision. Hope the TU-154 did not start descending in response to ATC, simultaneously as the DHL started a TCAS descent...
 

It's amazing the number of ATC frequency changes to transit this area of Europe; you can cross the States on less.

Obviously it goes without saying that our thoughts here go out to those involved and I for one hope the cause can be found and action taken to prevent such a tragic loss of young lives in the future.

Much speculation has been made in the Media (and elsewhere) as to why (if?) the TCAS system directed the Boeing pilot to descend when the Tupolev was already in descent.

From doing a google search it seems that there have been a number of occasions where a TCAS system has incorrectly directed a pilot to make a maneuver that has put two aircraft on a direct collision course. It is only through the skill and judgment of the pilots (And perhaps good visibility) that a disaster has been averted.

It should also be remembered that last night's tragedy occurred during darkness..maybe if visibility had been better disaster could have been narrowly avoided.

Without wishing to pre-judge any investigation, serious attention needs to be paid to the TCAS system on board the aircraft and pilots need to be made aware that these systems are far from infallible.

More on previous TCAS "incidents" here:

http://www.aaib.dft.gov.uk/bulletin/jun01/cggwd.htm http://www.rpi.edu/~woodhe/docs/wsj/flawed.txt
http://www.atsb.gov.au/aviation/rec/r19990159.cfm http://news.airwise.com/stories/99/11/942324329.html



 

Midair Collision Danger

 

We spent hours and hours cruising happily over continents like India, China, Russia and other places where the communication could be improved (This is an understatement!) and it is frightening to see todays accuracy of our on board navigation-systems with opposite traffic. In the older days (Before GPS/IRS DME update) you could spot this traffic somewhere left or right of your own track. These days there is no more "left or right", it is right on the nose.

Now, taking into consideration that some of this opposite traffic has no TCAS, in some cases no transponder, and in the worst case scenario flies with an altimeter in meters, using a manual slide ruler to "translate" meters into feet, it needs no further explanation that it would be nice to have an additional horizontal separation from your mere 1,000 - 2,000 feet vertical separation.

We would therefore like you to consider the following article perhaps for publication in some sort since the implementation of the procedure is of such a simplicity that we wonder why this hasn't been done before.

After having circulated this idea amongst some of our colleagues we realized that some companies are already following this "offset system" as a sort of in-house SOP.

It would be a good idea if all companies would follow this procedure over the above mentioned areas or at least be aware of the fact that some companies are already following this procedure.

If we would have to draw a parallel with the present situation we would construct a nice super highspeed highway with just one white line in the middle and all the traffic in both directions centering on this line, hoping that nobody comes roaring around the corner.


 

FLY RIGHT

The recent mid-air between a Kazahk Il-76 and a Saudia 747 highlights a concern which we have been discussing between ourselves for months. While all the details may not apply; this disaster does dramatically illustrate our concern.

The modern onboard GPS and/or DME updated IRS/INS navigational equipment has greatly enhanced the ease and accuracy of aircraft navigation. However, under certain circumstances, this accuracy could become a flight safety hazard. Airline pilots spent untold hours EXACTLY in the centre of their 10 mile wide airways guaranteed only 1000 or 2000 feet separation from opposite direction traffic. If there is a mistake during an altitude change by the crew of either aircraft, by the controller, with the communication between the two, or an autoflight equipment failure -- a disaster may occur.

FMS navigation computers should be off set just 1 or 2 miles to the right of track. This would guarantee 2 to 4 miles lateral separation between opposite direction traffic while all aircraft would still remain well within airways. This would utilise the extreme accuracy of onboard navigational systems to both remain within airways and to provide additional traffic separation. Had both aircraft been equipped with TCAS, the Delhi disaster would have been prevented. However, all aircraft world-wide are not so equipped.

Airline pilots almost daily will pass another aircraft with this 1000 feet clearance. Wouldn't it be much safer to also have at least 2 miles lateral separation? Since equipment and people do fail, why not implement this fail-safe technique?


See also 'LESSONS FROM DELHI'

back to Comment

LESSONS FROM DELHI

 

12 months ago now, the IFALPA Flight Safety Committee recommended that, (and I quote): "- all aviation GPS and other air navigation equipment of comparable accuracy contain a small embedded offset to the right that will protect against this increased risk of head-on collision. This offset should be built directly into the equipment, so that no operator procedure is required. The offset would be cancelled only for approaches, at a suitable point in the approach."

In the same article, the Committee notes that "..for the first time, the probable navigation error may well be contained within the dimensions of the airframe"

and "If two aircraft are flying on reciprocal tracks using GPS, and they do not have adequate vertical separation, they will almost assuredly collide."

Anyone who has seen aircraft repeatedly go directly over or under them up to a thousand miles from the nearest navigation beacon will have no difficulty agreeing with this very sobering assessment.

Now, given the proximity to Delhi and the types of aircraft involved, I accept that IRS/GPS accuracy was unlikely to have been a major factor in this incident (if a factor at all). However I would suggest that a high proportion of pilots reading this fly aircraft that are IRS or GPS-equipped. Therefore, can I suggest that you ALL start pressuring your individual airline's management and Flight Safety Departments to institute this suggestion by IFALPA without delay? - and go one pressuring them until they do something about it?

Having something like this approved by the many regulatory authorities who will want to become involved will take some time, (I'd suggest years!), even before such a change can be written into commercial FMS databases. In the meantime, why not speak to your Fleet Captain and Flight Safety Departments about having your airline amend their SOPs to have crews write in a ONE MILE RIGHT OFFSET into their IRS/GPS on passing through 10,000' in the climb. If everyone with IRS/GPS does it, we will be doubling the separation. I have spoken to ICAO regulator about this, and he is very much in favour of it, and I know a number of pilots in my own airline are doing it - including this little black duck.

I know there are a lot of you out there who don't fly IRS-equipped aircraft and I also appreciate that this is not a panacea. Nothing is, but anything that reduces the risk of another disaster like the Saudi/Kazhak collision has to be worth pursuing.

Chris Young

Associated article

back to Comment

Disaster pilot failed to change altitude: official

July 2 2002

wreckage of the Tupolev


AFP
 

The pilot of a Russian passenger plane that collided in mid-air with a cargo aircraft over southern Germany, killing at least 71 people, failed to respond to repeated calls to change altitude, a state official said today.

Most of the victims were children under 18, the duty manager at Moscow's Domodedovo airport was reported by Reuters as saying.

``All in all, there were about 50 children, including eight aged up to 12 years old,'' Stanislas Borovov told the agency.

Baden Wuerttemberg Transport Minister Ulrich Mueller said the Bashkirian Airlines pilot had failed to respond to repeated calls to change altitude.

Quoting details given by Swiss air traffic controllers tracking the planes, he said: "So we have to suppose that the pilot made a mistake."


The Tu-154 with 69 people aboard collided with a DHL Boeing 757 transport plane cargo and two crew at an altitude of around 11,000 metres above Lake Konstanz on the German-Swiss border shortly before midnight last night.

Evasive action

Mueller said the pilot of the cargo plane had tried to take evasive action but was unable to avoid the crash.

Carlo Bernasconi, the head of the Skyguide company charged with air traffic in Switzerland, would not confirm Mueller's statement but said that an inquiry was under way.

Mueller said the Russian plane's flight recorder had been found and that no dangerous materials were being transported in the cargo plane. "So there is no danger to the population," he said.

The Bashkirian Airlines Tu-154 had been en-route to Barcelona from Moscow, when it collided with the DHL Boeing 757 cargo aircraft with two crew at an altitude of around 11,000 metres over Lake Konstanz last night.

Witnesses near the lake, which borders Switzerland and Austria, reported hearing a loud explosion and seeing flaming pieces of wreckage crashing towards the ground.

'Sky was on fire'

"The sky became bright all of a sudden. It looked as if the sky was on fire," said Klaus Barinka, 42, a ferry boat captain working at Lake Konstanz.

State officials said it appeared the pilot of the passenger plane had failed to respond to repeated calls to change altitude by Swiss air traffic controllers.

Unconfirmed reports said that the cargo plane crashed into the lake and that an engine from one of the aircraft landed in a garden.

The smell of aviation fuel was thick at the scene, where hundreds of rescue service workers and firefighters were at work while police combed an area north of the town of Ueberlingen with helicopters.

Hospitals in the region were also on standby.

In Moscow, Interfax news agency quoted the emergencies ministry as saying that 57 passengers and 12 crew were aboard the Tu-154, which flew out of Moscow at 10:48 pm.
 

The DHL Cargo Jet


The ministry also said eight children were among the passengers.

German officials had earlier feared as many as 150 people were on board the Russian plane.

An airport official in Barcelona said today that the plane was carrying Russian tourists heading for Spain's Costa Brava coastal resort area.

Wreckage spread over 30k

Wreckage and bodies were reported to be spread over a 30-kilometre area in what was largely agricultural land.

"We have to fear the worst, that is that all the people are dead, said local official Siegfried Tann.

Tann said spot fires broke out as flaming wreckage hit the ground, but that most of them have been brought under control.

Police said several houses, a farm and a school had been set alight.

Bodies found

Officials at an emergency centre at the scene said at least 11 bodies had been found and that a flight data recorder from one of the two planes had been recovered in woodland by the lake.

The DHL Boeing cargo plane, operated by Bahrain Aviation and flying from Bergamo in northern Italy to Brussels with only two crew aboard, and the passenger aircraft collided shortly after 11:30 pm (0730 AEST Tuesday), police said.

Air traffic control alarm

At press conference early today, Baden-Wuerttemberg state Transport Minister Ulrich Mueller said Swiss air traffic controllers had been following the two aircraft and noticed that they were flying at the same altitude.

The Tupolev failed to react to repeated calls to change altitude, while the captain of the Boeing tried without success to avoid a collision, Mueller said.

The Tupolev TU-154 carried 69 passengers



The Tu-154 had earlier made a stopover in the southern city of Munich.

In a statement to the press, DHL said it was far too early to speculate on the possible cause of the accident but that a full and thorough investigation was under way.

DHL said its thoughts were with families and friends of those who have either lost their lives or been seriously injured in such tragic circumstances.

The fleet of Bashkirian Airlines, one of the companies that has spun off from the national carrier Aeroflot, is largely made up of Tu-154 aircraft and is the flag carrier for the Republic of Bashkortostan, part of the Russian Federation.

High-altitude mid-air collisions are rare, despite Europe's increasingly crowded skies.

Most large aircraft flying at such high cruising altitudes carry transponders such as the Traffic Alert and Collision Avoidance System (TCAS), which ensures pilots keep a safe distance from other planes in their vicinity.


to Safety Menu