TCAS philosophies

[ATC Watcher]

I decided to start this thread as a follow up of the “ Ueberlingen trial “ thread which was used by some to re-open what I call the “ TCAS Philosophies “

If a mod would like to move the posts in the Ueberlingen trial thread here, welcome, but not necessary.

The TCAS system is very complex, there is no more than a dozen persons in Europe and in the USA who fully grasp everything, and no more than a few more dozens that understand more than 50% of it . ( I classify myself in the second category )

What is below is not a lecture, this is basic stuff to help you understand where we are .

A few facts :

The Ueberlingen collision revealed to the public ( but not to the specialists) a few flaws in its logic , mainly that the RA sense reversal did not work. (That is when one aircraft follow the RA , but the second maneuvers in the opposite direction of its RAs and both aircraft end up towards each other ).
There was request made from Eurocontrol to re-open the work on TCAS and make a software change to correct this flaw, but the FAA refused initially as version 7 was “ final” in their eyes and that the TCAS Team had been disbanded. Nevertheless though RTCA , discussion were restarted and what became known as Change Proposal 112 ( or CP112 ) became a fact, later it became 112E ( enhanced) , and , since we were re-opening the Pandora box, we added a few more flaws corrections mainly one ( CP115) on the replacement of the RA ‘ Adjust vertical Speed ,Adjust “ to :” Level off , Level off “ and another one (CP116) on weakening RAs at low altitudes .

There is a chance now that we will get a new version 7.1 in the” future” . I do not put “ near future ” as the date is still fought between Eurocontrol which would like 2010 for all, and the FAA which says “ much later” for CP 112E ( sense reversal ) and is against CP 115 , the “Adjust/ level off “ change .

In case you are lost in the TCAS / ACAS versions here is a short recap :
TCAS is a brand name. ICAO call its Specs ACAS
2 current sofware versions of TCAS flying around : 7 and 6.04 A
Udate to 6.04A was free, 7 was not . 7 was pushed by the Europeans, FAA was against it. 7 is RVSM compliant , 6.04A is not.
If a new version comes ( 7.1 ) , will it be mandated ( i.e free) or will it be recommended ( at a cost ) is a good question to which I do not have the answer.
Initially there were 3 TCAS intended , TCAS I : TAs only , TCAS II RAS in vertical plane, TCAS III RAS in both vertical and horizontal plane. Work on TCAS III stopped as the logic proved too complex, and vertical plane RAs were more effective at avoiding a collision. So there will most probably never be any TCAS III .

This is the situation today .

Now questions asked were : is a new Ueberlingen possible today :

With TCAS sense reversal logic flaws : yes , nothing has changed since 2002. same logic. Same version flying around (7.0).
With procedures : maybe : ICAO has revised its documentation and most training material have highlighted the way to follow in case contradiction between ATC clearance and RA. But incident statistics so far both from USA and Eurocontrol show that some controllers still issue clearances contrary to RAs and that pilots still choose to follow ATC instead of RA.
( a recent case in France, with US trained and experienced 767 pilots from a major US airline , so no need to think it only affect exotic or Russian pilots )
Main reason : lack of understanding on how humans and automated machines function in reality. Prof Ladkin in his excellent paper http://www.rvs.uni-bielefeld.de/publ...rts/SCSS04.pdfon causal analysis after Ueberlingen call this “ Sociotechnical systems “ .

Now a last statistic : ( coming from Bill Thedford, of the Boston MIT LLC, one of the top 12 in my first classification earlier) Ueberlingen type encounters ( sense reversal ) occurs at 4.7 x 10 to the minus 6 per flying hour , or 58 events a year in Europe .If we expand this model, Bill predict that we could have 1 Ueberlingen –type mid air collision every 4 years due to the logic flaws.

The conclusion of this all :
Do not philosophy about TCAS : It is too complex a system. My advice : if you are a pilot : follow the RA, and if you are a controller and you have to give anti collision instructions , give a turn as well, in case TCAS comes in , it will complement your instruction instead of nullify it.

If you found this too long , this was the ( very) short version.
Safe flying.

[airmen]

We have very clear company procedures which are to follow the TCAS and not the controller, and I do follow the orders, they saved already my live once at least!
However, do not use the TCAS screen as a radar, target position are innacurate.
I have seen too much pilots using their screen for avoiding and or criticise the ATC...
The thing is to use the system as it should be used and the training and recurrent given actually will tell you exactly that.

[ray cosmic]

Thank you for the information. Actually I am quite astonished to hear conflicting situations still exist. Wasn't THE lesson of Ueberlingen that pilots shall follow TCAS alert in preference of ATC instructions?
I don't exactly know how much technical info one needs to have to operate in unison with the system. Its more like in flightschool: don't argue, just do it!

[berkut]

TCAS action

first of all, thank you very much for the information.

In my opinion it should be very clear. FOLLOW THE TCAS COMMAND. If you see, that a conflict can arise in a couple minutes, because you see it on the screen, than an advice from the ATC is helpful. If the "target" has already reached the TCAS warning area its clear. Follow the commands. The ATC controller is not a good help at this time as he has in the worst case a position indication on his screen which is 6 or 12 seconds old, depending on the radar.

There are only two scenarios which we have to ignor the TCAS command. In the event of a EGPWS call out (hard warning) or a windsheer warning.


Überlingen was a tragic and I hope we never have it again.

The guys on the flights DL 37 / CO 25 on 1987 July 8th had a lucky day, the missed each other by centimeters over the atlantic.

With the new technique (TACS III) and a proper pilot education it should be possible to avoid such scenarios in the future.


Happy flying and trust your avionics (with good airmenship)

[joernstu]

@ATC Watcher

Depicting the situation today, you said:

But incident statistics so far both from USA and Eurocontrol show that some controllers still issue clearances contrary to RAs and that pilots still choose to follow ATC instead of RA.

I think that one can argue about the controller in the Ueberlingen accident, if he really has issued a clearance contrary to the TCAS RA.
He had to maintain separation of the aircraft, which (if I have read the report correctly) normally meant 5NM and - due to inactive STCA - at the night of the accident had been set to 7NM.

The TCAS RA came before the aircraft underrun separation (at 7.11NM) and the controller had just finished wording his clearance when the aircraft under-run separation.


Do you (or anyone else in this threat) know, if there is work going on to give the controller feedback on the RAs issued by TCAS?

[Capt Pit Bull]

OK,

Pre Ueberlingen, many operators had ops manual phraseology allowing the crew to disregard an RA under certain circumstances. Post Ueberlingen that option no longer exists.

Much comment has been made regarding the differences in training philosophy that led to the accident.

The thing is, even before Ueberlingen, every source I ever saw gave specific advice that an aircraft should NEVER be manouevred in the opposite sense to an RA.

Loss of flexibility about following an RA (pilots discretion) is, imho, a bad thing. I've talked about this before on several threads, and I am not going to rehash the arguements or get into a pissing contest with anyone who slags me off for saying that following an RA may not be the safest course of action. The simple fact is that TCAS has good info about some threats to your aircraft, but there are a bunch of things it doesn't know about. As such 99.9% of the time its a good bet to follow the RA but as Pilots we should retain final control of our aircraft to cover the 0.1%. Thats why they pay us the big bucks, if you don't like it get off the jet and let them replace us entirely with computers.

At Ueberlingen, what did the Tupolev crew need to do to avoid the collision?

Well...... nothing in fact. Thats right, not a thing.

All they had to do was NOT manouevre in the opposite sense to their RA. Unfortunately they did, not once, but twice.

Ueberlingen type events had happened before. They were standing out in the JAA TCAS transition program newletters (but fortunately with horizontal separation saving the day). The simple fact is that most TCAS nasties contain at least one, and often several, manouevres opposite to the RA. This is the key point, focussing on the need to follow an RA is missing the point, and sooner or later somebody will find themselves in a situation where they can't follow it, or to so so will be unsafe, and then they'll be in a double bind.
ATC Watcher: Are there still 6.04 users flying around?

I agree with the 'add a turn' concept. It used to be part of the training offerred to controllers at LATCC (I spent a fair bit of time talking to their terminal training team when I put our companies TCAS program together). However I believe that they don't suggest this anymore.

Airmen: Absolutely. Over use of the traffic display is a common error.

joernstu: Mode S contains the basic architecture and message formats to downlink RA data to ATC. Apologies if you already knew that. Last I heard (which is a couple of years at least out of date) it was being trialed at a couple of places. I don't know what the current state of play is, anyone know?

pb

[ATC Watcher]

I think that one can argue about the controller in the Ueberlingen accident, if he really has issued a clearance contrary to the TCAS RA.

No, he did not. In Ueberlingen , the controller did not issue a clearance contradictory to TCAS. He issued the clearance BEFORE the TCAS annunciated the RA.
But , unknown to him , TCAS " choose " (for various reasons too long to explain here )to make the TU154 Climb and the 757 descend.
That put the TU154 crew into conflict/doubts . The crew choose to follow ATC.
Had the RA configuration be different ( TU154 to Descend and B757 to climb, there would have been no collision, as the RA would have been in the same sense as the ATC instruction.

is there is work going on to give the controller feedback on the RAs issued by TCAS?

Yes, but on down-linking the RA message to ATC only ( Advanced studies made by Eurocontrol , called FARADS ( Feasibility RA down Link) and RADE (RA Downlink simulations in Bretigny)
Consequences of D/L RA to ATC are complex and many think ( inlc. me ) that it may induce more problems that it will solve.

[PBL]

Originally Posted by airmen, referring to TCAS

they saved already my live once at least!

There is a phenomenon here worth thinking about. When I voiced some of my concerns about TCAS on a pilot's mailing list some years ago, I was deluged with affidavits from pilots who claimed that TCAS had saved their lives. If all of these claims had been true, that would have amounted to some 20 or so midair collisions in a sample of a few hundred pilots (the members of the mailing list) over the course of, say, ten years. Let's assume 10 "life-savers", that is, otherwise-midairs, for 500 pilots in 10 years. That is an average of one "midair collision" per 500 pilots per year.

There are about 15,000 aircraft from Airbus, Boeing, MD, Lockheed still in service (calculated from figures in Flight International, 24-30 October 2006). So that's not counting the "commuters", or the Tupolevs. Those aircraft are in service, let's say, 14 hours per day, for a total of 210,000 hours per day, or 5,880,000 hours in 28 days. Duty rosters take 100 flight hours for a pilot in 28 days to be relatively high, but let's use this figure, since it will lead to a conservative estimate. It means there are about 59,000 captains needed to fly those 5,900,000 hours every 28 days. With one "midair" per 500 captains per year, and accounting for the fact that it takes 2 captains to have a "midair", we come to a figure of 59 "midairs" per year.

Compare this with actual midairs amongst airline transport aircraft. 1959 Grand Canyon, 1976 Zagreb, 1978 San Diego, 1985 Cerritos. And then, since the advent of TCAS, 1995 Namibia (involving a German military machine which I do not believe was TCAS-equipped), 1996 New Delhi, 2002 Überlingen and 2006 Amazonas. That is 4 before the advent of TCAS (mandated 1991, I believe) and 4 since (16 years).

Those crediting TCAS with all these "saves" need to explain why, before the advent of TCAS, there were a statistically-negligible number of collisions in the history of airline flying, and upon the introduction of TCAS there are suddenly almost 60 per year. Even considering the growth of air travel, that is an unbelievable jump.

The obvious answer is, of course, that most of those "saves" would not have resulted in collisions, despite what their pilots thought or think.

Consider the following, as a thought experiment. Up to the Congressional mandate for TCAS, there were three collisions involving airliners in the U.S.: 1959 Grand Canyon, 1978 San Diego and 1985 Cerritos. Suppose TCAS initiated RAs, not for 2 TCAS-equipped aircraft, but only when one had TCAS and the other only Mode C. That would have satisfied the issue worrying the Congresspeople, namely incursions of GA aircraft into zones of intensive airline operations, and it would have saved one of the midairs since (Überlingen). The other three midairs in the TCAS era occurred to non-TCAS-using aircraft.

Let us compare TCAS with another highly-lauded system for avoiding accidents, GPWS/EGPWS. Also a system introduced primarily by one manufacturer (again Honeywell). And in contrast to TCAS addressing a real, continuing problem which still leads to many aircraft losses per year, namely CFIT. EGPWS is essentially a private development (stemming from the work of Don Bateman and colleagues).

It is interesting in this regard to notice how TCAS is self-advertising and EGPWS not. A pilot can say "TCAS saved me" without it reflecting on himher self, but rather on some air traffic controller establishment somewhere. No pilot can say "EGPWS saved me" without someone else asking how heshe lost situational awareness in the first place (with rare exceptions in which it is clear how, such as with British Mediterranean at Addis Ababa).

So the statistics were never there to say "TCAS is a good thing", and it is self-advertising, through pilot "save" stories. But there are smart people in aerospace who, no matter whether they think TCAS is a net gain or a net problem, are able to look at the facts and the situation just the same as I am now. Airline CEOs, for example. So the question is: why are collision avoidance mechanisms still being promoted, researched and supported to such a great extent?

I think there are two reasons. One is that they are one attempt at a solution to a problem which no one knows yet how to assess accurately, namely the avoidance of midair collisions, and no one is willing to risk misjudging (underestimating) the likelihood of those. The other reason is that effective on-board systems make collision avoidance a contract between the two (or more) aircraft involved, which transfers the costs of collision avoidance away from large ATC organisations. Since the airlines and their passengers ultimately fund these organisations, one could see the financial benefits to airlines of bringing this function in-house (after the initial cost of installing the equipment, which has n any case been mandated). Add to this that collision avoidance is one of the brakes on moving to "free flight" (self-routing IFR at high altitudes), which airlines believe would bring them great cost reductions, and one can see that TCAS+free flight is a politically sellable package which could reduce the costs of air travel even further. Without some ATC-independent collision-avoidance system regarded as effective, this move could not happen, and TCAS is the only technology in town.

PBL

[brain fade]

Would it be possible to have the TCAS generate the radio calls? Or at least comunicate to the controller in some way what is actually happening.

Seems to me that it can be quite tricky to get the call off especially as there are, for obvious reasons, often other calls being made at the time. Even a short duration chime or tone might do the trick.

[Beakor]

Originally Posted by PBL

Those crediting TCAS with all these "saves" need to explain why, before the advent of TCAS, there were a statistically-negligible number of collisions in the history of airline flying, and upon the introduction of TCAS there are suddenly almost 60 per year. Even considering the growth of air travel, that is an unbelievable jump.
PBL

You also need to consider the increase in navigational accuracy and altimetry. Events which would previously have been close airproxes due to kit inaccuracy could well now be collisions with everyone within a few metres of the airway centreline and a few feet of their FL.

Even so, still agree with PBL, it is an unbelievable jump

[PBL]

Since people here are also advising always follow the RA, without exception, it might be appropriate to point again to my paper, which ATC Watcher referenced, in which I note that

* the TCAS algorithm is known to resolve all conflicts between two aircraft (Lynch, Lygeros, MIT 1997)
* It is still not known whether the TCAS algorithm resolves all conflicts between three aircraft

and I have offered (since 2002) a PhD to anyone who can definitively answer the question whether the TCAS algorithm resolves all conflicts between three aircraft.

I also pointed out that, in the Überlingen collision, the Russian commander believed they were in a possible three-aircraft conflict (I wondered about that in a paper I put on the WWW in August 2002, and it was confirmed by the report in 2004). Further, in the decision situation they were given, it was rational to choose to descend (I also went to some pains to point out that saying "it was rational" and supposing that that was the reason he continued to command a descent are not the same thing at all).

Capt Pit Bull suggests that following the RA is appropriate some X percentage of the time, and maybe inappropriate some other percentage (100-X). (He chose X = 0.1, but I presume he was illustrating his point, rather than suggesting a real value for X.) One of the main issues with TCAS, for me, is that no one knows what the value of X is.
What if it turns out to be 50?

PBL

[ATC Watcher]

Capt Pit Bull :
excellent remark about focusing on the fact that an aircraft should NEVER be manouevred in the opposite sense to an RA. . If this only would be understood and followed by all pilots, most of our problems will be solved ( except for encounters with SSR-only intruders of course)

Quick reply :

ATC Watcher: Are there still 6.04 users flying around?

Still a lot in the USA to start with. In a recent publication ( March 2007) the FAA ATO reported that :
" US aircraft were not mandated to equip with Version 7, but Europe elected to mandate this version. Therefore, US aircraft flying into European airspace must have TCAS Version 7 avionics. Currently, two thirds of the US commercial fleet is TCAS Version 7 equipped.."
Which means 1/3 of commercial fleet is still on 6.04 and some of General aviation jets and turboprops on top of that I guess .
Rest of the world : status unknown on the domestic fleets. All in all , still a fairly large number, I would say.

[Capt Pit Bull]

Capt Pit Bull suggests that following the RA is appropriate some X percentage of the time, and maybe inappropriate some other percentage (100-X). (He chose X = 0.1, but I presume he was illustrating his point, rather than suggesting a real value for X.) One of the main issues with TCAS, for me, is that no one knows what the value of X is.
What if it turns out to be 50?

Peter,

You're quite correct, I was not attempting to be scientifically accurate but rather speaking conceptually, (I'm not even remotely qualified to try and put a precise numerical value on it).

regards,

pb

[antic81]

Hi there

When looking at collisions, or near collisions, in the modern, largely TCAS II equipped era, we must also remember that routes are now being flown through far more precise and exacting corridors of airspace using GPS and IRS navigation systems, and these routes are being flown through some very poorly manned and ill equipped airspace (i.e. Africa) therefore increased chances of colliding with other aircraft is higher.

I am of the opinion that when looking at post-TCAS vs pre-TCAS collisions, we need to take that into account as it does not give an accurate portrayal of its effectiveness.

Cheers
Anthony

[Kit d'Rection KG]

1. Make the legislation fit the technology;

2. Accept that it's an imperfect system driven by greed (for cheap travel, higher share prices, etc);

3. 'Accidents' will happen;






















4. Invest in Kenyon

[Kit d'Rection KG]

Oh, and by the way, there's at least one very well respected academic in Europe who can show (by 'proof') that the TCAS probabilities are bunkum anyway, and providence is adjusting the fatality rate without our knowledge...

[alf5071h]

Re Überlingen, one of the alternative views of this accident was that it involved ‘poor’ worldwide/industry communication.
ACAS / TCAS evolved from a US concept (to solve a specific US problem) into something that the industry found to be of great value and thus was approved by ICAO. The ‘failure’ was that the procedures (operational assumptions and crew actions) that evolved in the US were not communicated worldwide – they are now via ICAO. I suspect that there are still many operators / countries who do not appreciate the need to follow the system’s instructions. I recall that at the time of the accident one PPrune poster suggest that in his country (Africa), every pilot would turn when responding to TCAS!

I wonder if the semi-circular / quadrangle flight level system would have prevented Überlingen? Was the lack of this feature a failure to ‘defend in depth’, - over reliance on radar systems, need to handle more aircraft in a crowded airspace, or the human desire to help others by giving them direct routing?

I note similarities with the TAM overrun accident re worldwide communications; i.e. the knowledge of, adoption, and training for the revised crew procedure for the MEL single thrust lever operation.

Re EGPWS; several ‘Saves’ are in the paper “Celebrating TAWS ‘Saves’: But lessons still to be learnt”. To my knowledge none of these events have been claimed as saves by the crew, primarily as the pilots were unaware of the severity of situation that they encountered. Many of the HF contributions / human behaviors in these incidents can be seen in TAM and other overrun accidents, and the issues of situation unawareness and the failure in ‘the last defense’ - the crew - failing to react or reacting inappropriately to warnings, arises in the discussion in this thread.

: is a new Ueberlingen possible today : alas ‘yes’ as humans are part of the system.

[airmen]

TCAS is designed to work according manufacturer manual, it is stipulated here that pilots have to follow orders to escape, so why argue with that and prefer to follow ATC orders?
Maybe those guys need to have a good Sim session to understand by doing it wrong to be able to learn something?

PBL wrote:

The obvious answer is, of course, that most of those "saves" would not have resulted in collisions, despite what their pilots thought or think.

I saw the conflicting aircraft (a Beech Baron) at the last minute (sun in the back) during the escape manoeuver, he was very close and was flying opposite course at the same altitude (controller error) and I can tell you that the controller went by himself to meet us after landing, he told us that he saw nothing and as such gave us no instructions. Off course we followed the TCAS but I can not tell you what the other pilot did...

[4Greens]

There are many analogies with GPWS warnings. If you get either of them go. Both these responses need to be trained in the simulator so that they become automatic. The warnings of course need to be unexpected.

[PBL]

Perceptive comments as usual from alf.

Originally Posted by alf5071h

one of the alternative views of [the Überlingen] accident was that it involved ‘poor’ worldwide/industry communication.

I wouldn't characterise the view as "alternative"; it was thoroughly investigated and all but proved in the report.

Originally Posted by alf5071h

ACAS / TCAS evolved from a US concept (to solve a specific US problem) into something that the industry found to be of great value and thus was approved by ICAO

While this is generally true, I am not sure that TCAS was introduced to solve a specific problem. I was around in California at the time of the 1978 PSA midair and already flying around the state at the time of the 1985 Cerritos accident. TCAS had been under development then for quite a while, by the U.S. research contracting corporation Mitre, on government money. The precise history eludes me at the moment, but Honeywell claims on their WWW site (where there used to be a history) "over 40 years of TCAS development experience". That would make at least 18 years up to Cerritos. In other words, there was a system there waiting to be sold.

So there are two polar-opposite ways of construing developments. One is that some far-sighted industrialist knew that an opportunity would arise, developed a system, and grasped the opportunity with aid of the appropriate lobbying. Another is that some far-sighted industrialist anticipated the future need for some such system two decades before it became necessary. The difference between the two views is only that of whether one believes that such a system was/is necessary. The evidence was obviously lacking at the time of Cerritos; I don't know that there is any way to decide the question now.

The feature of TCAS which most disturbs me, and which I believe should continue to disturb anyone, is that we now have two institutions whose (main) function is collision avoidance, TCAS and ATC, and the former works *against* the latter. With 2,000 ft vertical separation, 1,500 fpm vertical manoeuvring (at least on first RA) and 30 seconds to CPA, the math says that there is no obvious conflict. However, with 1,000 ft vertical separation, a TCAS-manoeuvring aircraft could get potentially within just over 100 vertical feet of another, thus initiating a further RA. This used not to be true at cruise altitudes, but is so now with the introduction of RVSM.

Eurocontrol did a study, called ACASA, which amongst other things looked at simulated RVSM/TCAS interactions. They massaged about two days' worth of radar data to give aircraft RVSM-type vertical separation, and scanned it for potential TCAS interactions, and saw none. Eurocontrol then claimed in their safety case for RVSM that there were no identified risks with TCAS and RVSM (citing the ACASA results as proof, in personal communication). The problem is that they were aiming for a far higher TLS than looking at
the statistical equivalent of two days' radar data would allow . (General considerations from statistical theory suggest that it will be impractical to assess TCAS interactions statistically to currently-required levels of safety.)

In contrast, EGPWS, while requiring a manoeuvre, does not operate in conflict with ATC, because an aircraft responding to an EGPWS warning is already well below minimum altitude and therefore well out of airspace in which there are other controlled aircraft.

PBL