TCAS philosophies

[PBL]

FullWings,

I think everyone understands by now that you choose to follow a TCAS RA no matter what. But you have not given anybody else a reason to do so until you can answer the decision problem to which I referred in post [email protected] (currently #22) and explained again in post [email protected]
(currently #43).

You offered a response in [email protected] (currently #41) which gave decision criteria which had you going in two different directions in my problem, which I pointed out in [email protected]. You responded ([email protected], currently 45) that I had misunderstood your response. Maybe, but you still have to offer a resolution, along with reasons, of the decision problem, and it's been 11 days now. The Bakshirian crew only got 30 seconds.

bsieker has responded as I might have to some of your other points, so I won't repeat those.

PBL

[FullWings]

PBL,

Please read my analysis of the decision problem presented to the Bakshirian crew at Überlingen. They were faced with an "intruder" at 10 o'clock which they saw, and an unknown conflict at 2 o'clock which they didn't see, and for which they had an advisory to descend (that is, he was at or above their altitude).

The (sadly last) communication from the Tu-154 to Zurich was: "'Ja', we have traffic at your... 2 o’clock now at 360". Where does the idea of an "unknown conflict" come from?

What would you do? Climb towards another conflicting aircraft that you don't see but ATC does? Or avoid him, descend towards an aircraft that you do see, and hope to avoid him using visual means?

What does the TCAS have to say about it? Follow the commands to resolve the conflict. Done. To reverse the question, why would I not follow the RA? You have 18 seconds to reply...

May I respectfully suggest that you're trying to find a complicated answer to a simple problem? The sky is often full of aircraft, some you can see, some you can't (try an approach to EWR or JFK on a busy day). The sort of scenario you're talking about is not uncommon; in fact, I would go as far as to say, unremarkable. Installation & use of TCAS is compulsory in the USA and Europe in most airspace that airliners use; if we all went about 'doing our own thing' in reponse to RAs, I don't really want to imagine the consequences. Until someone proves it otherwise, I shall be 'staying out of the red', thank you!

bsieker,

I think you misunderstood. One cannot tell that these probablilities are "low enough" to disregard them. Since we do not know them, perhaps never will, they may be quite high.

Given the amount of flying hours logged and the number of incidents (and I'm talking generically about systems) one should be able to set an upper bound on the these probabilities, given certain confidence limits, just like you would with any other experimental data?

And yet, the Ueberlingen accident would not have happened, if neither aircraft had been equipped with TCAS.

Not possible I'm afraid. We all have to have it. Therefore we need a common standard in how we respond to it (which is where I came in on this discussion).

All,

I'm continuing to partake in this discussion because I don't want any impressionable pilots reading this thread and thinking that it's a good idea to start tinkering with standard responses to RAs (especially if I am in the other aircraft). I'm all for academic arguments and theories as that is the way we progress science but in many cases there has to be a set way of doing something until we build up enough evidence to justify changing it. This applies especially to aviation.

[bsieker]

Originally Posted by bsieker

And yet, the Ueberlingen accident would not have happened, if neither aircraft had been equipped with TCAS.

Originally Posted by FullWings

Not possible I'm afraid. We all have to have it.

I understand that it is mandatory. Whether or not that is actually a Good Thing is another matter.

Having two collision avoidance systems (radar/ATC/ground-based, and ground-independent aircraft-transponder-based) that may contradict each other is a real problem. Particularly and glaringly so if ATC instructs in one direction, and TCAS instructs in the opposite direction immediately afterwards. Do you reverse your manoeuvre to cross your conflict's original trajectory again?

ATC usually has a lot more information, than TCAS, and may instruct differently from TCAS to avoid further conflicts. The premise that ATC is already "out" when TCAS issues an RA does not hold.

Would you advocate doing away with the ground-based system and rely exclusively on TCAS? The trend towards Free Flight will go in that direction, but we're not there yet.

Therefore we need a common standard in how we respond to it (which is where I came in on this discussion).

Common standards are a wonderful thing, if we can agree on what they should be for sound reasons. I don't see a silver-bullet solution that will cure all problems. It is clear that making following the RA compulsory in all cases is not it. (If it were, TCAS would be coupled to the autopilot to perform the avoidance manoeuvre automatically; I don't see anyone actively advocating that.)


Bernd

[PBL]

Originally Posted by FullWings

Where does the idea of an "unknown conflict" come from?

If you cannot see it in the Überlingen report, just take it as posited in my decision problem.

Originally Posted by FullWings

What does the TCAS have to say about it? Follow the commands to resolve the conflict. Done.

As I said in my last post, we all understand by now that that is your position.

Originally Posted by FullWings

To reverse the question, why would I not follow the RA?

Because following the RA would put you into immediate conflict with traffic which you are not painting but which ATC is.

Originally Posted by FullWings

I'm continuing to partake in this discussion because I don't want any impressionable pilots reading this thread and thinking that it's a good idea to start tinkering with standard responses to RAs

Thank you for your concern for "impressionable pilots". Speaking of impressionable pilots, I wish you could understand and resolve the decision problem which I keep asking you about, rather than simply asserting that TCAS solves it (which it obviously - to others - does not).

PBL

[joernstu]

Originally Posted by FullWings

I'm all for academic arguments and theories as that is the way we progress science but in many cases there has to be a set way of doing something until we build up enough evidence to justify changing it.

I don't think that this is an "academic argument" any more. Perhaps it was an "academic argument" when the impact of TCAS on RVSM was analysed using airtraffic data for 2 days.

There are very real problems, in which TCAS has plays a role, as two mid-air collisions since 2002 demonstrate.

If you look for more evidence you will find complains of ATC on a device that manipulate the outcome of their instructions, leaving them without a coherent image of the situation in the air.

Another problem with TCAS is that the manufacturer (and following him many other organisations) present TCAS as a technical solution in a conflict situation where every other means to avoid a collision have failed. In the Ueberlingen mid-air TCAS gave an RA when both a/c were at 7.11 NM seperation. The limit for seperation in this airspace at this night was 7 NM, usually 5 NM. This demonstrates that TCAS can act at points in time, where other means have not yet failed, rebuting the claim that TCAS only acts when everything else has already failed.

More theoretical arguments are multiple aircraft situations (for which it can be proved that TCAS cannot give advice solving the collision threat) and aircraft without transponders.

Based on these problem (some of them real, some of them more theoretical) I get the picture, that TCAS is not the silver-bullet some want it to be. Of course TCAS is better than noting (at least I hope that this is the case as it was certified for use in aviation), but I also get the image, that it is still far from optimal. The problem is, that some of the problems with TCAS can only be solved by spending lots of money and/or replacing existing TCAS units against newer models.

[joernstu]

Originally Posted by FullWings

The (sadly last) communication from the Tu-154 to Zurich was: "'Ja', we have traffic at your... 2 o’clock now at 360". Where does the idea of an "unknown conflict" come from?

This communication was not from TU154M to ATC Zurich, but from the controller in Zurich to the TU154M crew.

This means that the TU154M crew was warned of traffic from 2 o'clock (by ATC) and from 10 o'clock (by TCAS and visually confirmed). This leads to the asumption (not only by PBL, but by the accident investigators at BFU), that the TU154M crew was occupied with the search for the traffic at 2 o'clock. The report says in chapter 2.6.3.1 under "Distribution of tasks" (on page 98 in the english version, on page 103 in the german version)

It is probable that he (PNF, TU154M) at least monitored the PF as the descent was initiated, but he then trained his attention on the visual search. He did not advise the PF that they were approaching teir cleared level of FL350. At this time the PNF's attention was concentrated on the visual search, and was probably centered in the wrong sector

(emphasis added).

[PBL]

It has been worrying me that some people don't appear to understand the decision problem I have posed or why certain solutions are unsatisfactory. It occurred to me this morning while looking for reasons not to get out of bed at 06.00 that there is another way to present the dilemma.

[Begin Exercise Statement]

Scenario 1: You are flying along, level at FL 360. ATC says descend immediately to FL 350 due to conflict with traffic at 2 o'clock.

What do you do? Why?

Scenario 2: You are flying along, level at FL 360. You have another aircraft in sight at your 10 o'clock. It's night, so you have little range information, just direction. ATC says descend immediately to FL 350 due to conflict with traffic at 2 o'clock.

What do you do? Why?

Scenario 3: You are flying along, level at FL 360. You have another aircraft in sight at your 10 o'clock. It's night, so you have little range information, just direction. ATC says descend immediately to FL 350 due to conflict with traffic at 2 o'clock. At the same time, you get a climb RA for your visual target at 10 o'clock.

What do you do? Why?

[End Exercise Statement]

As far as I understand, for example, FullWings's answers, they would be:

1. Descend. Reason: conflict avoidance with unseen traffic.
2. Descend. Reason: conflict avoidance with unseen traffic.
3. Climb. Reason: RA advises to do so.

The difference between scenarios 2 and 3 is, however, minimal. Yet the avoidance manoeuvre is exactly opposite.
And the unseen conflicting traffic does not go away.

PBL

[FullWings]

PBL,

Replying to your latest post first:

As far as I understand, for example, FullWings's answers, they would be:

1. Descend. Reason: conflict avoidance with unseen traffic.
2. Descend. Reason: conflict avoidance with unseen traffic.
3. Climb. Reason: RA advises to do so.

The difference between scenarios 2 and 3 is, however, minimal. Yet the avoidance manoeuvre is exactly opposite. (My bolding)
And the unseen conflicting traffic does not go away.

Yes. Nothing very surprising here. Does the absolute direction of the avoiding manoeuvre matter if it resolves the conflict? In your scenario No. 3 you're assuming that the traffic you can see is the one that has generated the RA... That may not be the case, especially as "It's night, so you have little range information, just direction." The information presented to you also depends on the type of TCAS installation and what modes you have selected; TCAS is much less accurate in displaying bearings than distances, so the actual traffic could be in a significantly different direction. Also, if you can't see your 2 o'clock traffic and it isn't on TCAS then it's probably far enough away not to be an immediate risk, unlike the traffic that is causing the RA.

[PBL]

FullWings,

Originally Posted by FullWings

Yes. ... Does the absolute direction of the avoiding manoeuvre matter if it resolves the conflict?

The manoeuvre in Scenario 3 (which you affirm) obviously does *not* resolve the conflict with the unseen 2 o'clock traffic, and might well exacerbate it. It also conflicts with the principle you proposed in your [email protected], currently #41, that

Originally Posted by FullWings

The danger in aviation comes mostly from the aircraft you can't/haven't seen or aren't aware of

In contrast, the manoeuvre in Scenario 2 resolves the conflict with the unseen 2 o'clock traffic.

Originally Posted by FullWings

In your scenario No. 3 you're assuming that the traffic you can see is the one that has generated the RA...

I am not assuming anything at all.

Originally Posted by FullWings

That may not be the case .... The information presented to you also depends on the type of TCAS installation and what modes you have selected; TCAS is much less accurate in displaying bearings than distances, so the actual traffic could be in a significantly different direction

No. TCAS is said to be accurate to 15 degrees. The difference between 10 o'clock and 2 o'clock is 120 degrees, more than 30 degrees (= 2 x 15 degrees) which is the resolution necessary to resolve which aircraft is causing the RA.

Originally Posted by FullWings

Also, if you can't see your 2 o'clock traffic and it isn't on TCAS then it's probably far enough away not to be an immediate risk

Or it is a military aircraft, or a NORDO, or an older bizjet, or a Legacy whose pilots just kicked off the transponder and ATC is giving you a primary return assuming (appropriately) that he is maintaining altitude.

I am much less convinced than I was about your understanding of TCAS scenarios.

PBL

[bsieker]

Originally Posted by FullWings

you're assuming that the traffic you can see is the one that has generated the RA... That may not be the case, especially as "It's night, so you have little range information, just direction." The information presented to you also depends on the type of TCAS installation and what modes you have selected; TCAS is much less accurate in displaying bearings than distances, so the actual traffic could be in a significantly different direction.

As much as 120 degrees away? Is TCAS bearing accuracy reallky that bad (even in the worst case)?

Further, you are then assuming that you have traffic, that you can see, which is not on a collision course (could be), and traffic that is causing an RA, which you cannot see.

Also, if you can't see your 2 o'clock traffic and it isn't on TCAS then it's probably far enough away not to be an immediate risk, unlike the traffic that is causing the RA.

Probably not.

If ATC gives you an "expedite!" instruction to avoid it, it is probably quite close.

Further, in contradiction to your first quote, here you assert that traffic you cannot see, is probably far enough away.

So by your own logic, whether or not you can see your traffic, is not an indicator of proximity.


Bernd

[FullWings]

joernstu,

If you look for more evidence you will find complains of ATC on a device that manipulate the outcome of their instructions, leaving them without a coherent image of the situation in the air.

If I'd set two aircraft on a collision course in error I'd want something to manipulate the outcome of my instructions! Never heard/seen any complaints myself, especially from people who understand how the systems work.

Another problem with TCAS is that the manufacturer (and following him many other organisations) present TCAS as a technical solution in a conflict situation where every other means to avoid a collision have failed. In the Ueberlingen mid-air TCAS gave an RA when both a/c were at 7.11 NM seperation. The limit for seperation in this airspace at this night was 7 NM, usually 5 NM. This demonstrates that TCAS can act at points in time, where other means have not yet failed, rebuting the claim that TCAS only acts when everything else has already failed.

OK, RA given at 7.11NM, limit 7NM. Time to close 0.11NM was less than a second, so loss of separation was assured as no realistic course change could be made in that timeframe.

TCAS only knows what is happening, not "intent" (maybe in the future with mode-s, etc.) It has to assume there will be no outside intervention and I think 20-30s before collision is not unreasonable. In the event under discussion, primary means of separation had or were shortly going to fail so the secondary systems started to activate. There is inevitably going to be an overlap between the two but in the example above it's very small.

More theoretical arguments are multiple aircraft situations (for which it can be proved that TCAS cannot give advice solving the collision threat) and aircraft without transponders.

And these cases form what percentage of the threat population in an ATC radar environment?

Of course TCAS is better than nothing (at least I hope that this is the case as it was certified for use in aviation), but I also get the image, that it is still far from optimal.

At least we agree on something.

[bsieker]

Originally Posted by FullWings

In the event under discussion, primary means of separation had or were shortly going to fail [...]

Evidently not. ATC was about to issue instructions to separate traffic.

It was neither failed, nor was it going to fail shortly. If TCAS "thinks" late ATC intervention is a failure of ATC, then maybe its requirements are flawed.


Bernd

[FullWings]

In contrast, the manoeuvre in Scenario 2 resolves the conflict with the unseen 2 o'clock traffic.

Because you haven't come close enough to the other traffic to trigger an RA yet. That's why your scenarios are different - the aircraft are in different positions...

Originally Posted by FullWings
In your scenario No. 3 you're assuming that the traffic you can see is the one that has generated the RA...

I am not assuming anything at all.

No? Then I'm having problems with the meaning of this sentence:

..you get a climb RA for your visual target at 10 o'clock.

As opposed to: "...you get a climb RA"?

No. TCAS is said to be accurate to 15 degrees. The difference between 10 o'clock and 2 o'clock is 120 degrees, more than 30 degrees (= 2 x 15 degrees) which is the resolution necessary to resolve which aircraft is causing the RA.

OK, the aircraft that's going to hit you is at 9 o'clock. You haven't spotted this but you have seen the one at 10 o'clock. You ignore the RA because "you've seen it". Crunch.

Or it is a military aircraft, or a NORDO, or an older bizjet, or a Legacy whose pilots just kicked off the transponder and ATC is giving you a primary return assuming (appropriately) that he is maintaining altitude.

Or it's a cruise missile, a weather balloon, a flying saucer, a flock of birds, etc. Very easy to find edge cases where the system may not work but how realistic are they in a positive radar environment? If you stray into the path of a UFO during an RA it's just not your day is it? (BTW, NORDO stands for "non-radio"; doesn't mean they're not squawking...)

I am much less convinced than I was about your understanding of TCAS scenarios.

Out of interest (and I'm not trying to "one-up" you or anything), how many real or simulated (as in sat at the controls) TCAS encounters have you experienced?

[FullWings]

Evidently not. ATC was about to issue instructions to separate traffic. It was neither failed, nor was it going to fail shortly. If TCAS "thinks" late ATC intervention is a failure of ATC, then maybe its requirements are flawed.

Yes but those instructions came too late to stop the aircraft coming within 7NM of each other at the same level, on a collision course. That, by definition, is a "failure of separation". "Late ATC intervention" is a "failure of separation", involving much form filling, suspension of the individuals involved, official reports, etc. I'm coming to think your definition of "separation" is "not actually hitting each other", which is not an industry standard!

[bsieker]

Yes but those instructions came too late to stop the aircraft coming within 7NM of each other at the same level, on a collision course. That, by definition, is a "failure of separation". "Late ATC intervention" is a "failure of separation", involving much form filling, suspension of the individuals involved, official reports, etc. I'm coming to think your definition of "separation" is "not actually hitting each other", which is not an industry standard!

Following your argument through would mean that if ATC fails to separate traffic before it comes inside the 7nm limit, the controller should shut up, and leave it to TCAS to do the job (or not). Better to take the risks of TCAS failure, than to violate an industry standard!

I don't think you're advertising that course of action.

But what, then, is your point?


Bernd

[FullWings]

Following your argument through would mean that if ATC fails to separate traffic before it comes inside the 7nm limit, the controller should shut up, and leave it to TCAS to do the job (or not).

No, the controller carries on trying to separate the traffic. If, in extremis, there comes a point where there are conflicting instructions from ATC and TCAS, TCAS is followed.

The order things happen is something like this:
1. ATC do their job and for virtually all the time get it right; but they are as human as the pilots so...
2. The two groups between themselves may, every now-and-then, trigger a STCA. This may lead to a change of plan or giving of further instructions.
3. A TCAS TA might be issued so that we can get the hosties off our laps.
4. If it gets to the point of an RA, then TCAS "controls the vertical" and ATC are informed ASAP of a deviation. Climb or descent instructions from ATC are ignored as their information is not as up-to-date as that possessed by the TCAS computers. (Amongst other reasons.)
5. When we've all missed each other, we initiate a return to our cleared levels and await further instructions from ATC...

But what, then, is your point?

My point is that TCAS is here, the installation and use of it is mandated and because of this, the way it is used has also been mandated. Until there is some real evidence that the way it is being used is wrong/dangerous, then this state of affairs will continue.

An open question: Can you point me in the direction of any hard evidence in support of the 'danger' hypothesis? Links, etc.?

[PBL]

Originally Posted by PBL

In contrast, the manoeuvre in Scenario 2 resolves the conflict with the unseen 2 o'clock traffic.

Originally Posted by FullWings

Because you haven't come close enough to the other traffic to trigger an RA yet.

Exercise 1: Modify the scenario to make your reply inappropriate.

Originally Posted by FullWings

In your scenario No. 3 you're assuming that the traffic you can see is the one that has generated the RA...

Originally Posted by PBL

I am not assuming anything at all.

Originally Posted by FullWings

No? Then I'm having problems with the meaning of this sentence:

Originally Posted by PBL

..you get a climb RA for your visual target at 10 o'clock.

Exercise 2: Explain the difference between an assumption and a statement.

Originally Posted by FullWings

OK, the aircraft that's going to hit you is at 9 o'clock. You haven't spotted this but you have seen the one at 10 o'clock. You ignore the RA because "you've seen it". Crunch.

The TCAS screen apparently was painted two targets at 9 o'clock and at 10 o'clock, and the 9 o'clock was lit up red. That was not a scenario I set.

Originally Posted by PBL

Or it is a military aircraft, or a NORDO, or an older bizjet, or a Legacy whose pilots just kicked off the transponder ...

Originally Posted by FullWings

Or it's a cruise missile, a weather balloon, a flying saucer, a flock of birds, etc. Very easy to find edge cases where the system may not work but how realistic are they in a positive radar environment?

According to Flight International, bizjets are quite realistic in the positive radar environment round London, indeed are said by the airlines to be causing a disproportionate increase in traffic. I would imagine in other places as well.

I don't think there is much point in continuing discussion unless you are going to take it seriously. Either you see the point of my scenarios, or you don't. If you do, it would be nice to get the answers. If you don't, it would be nice to have you say so loud and clear.

PBL

[FullWings]

The TCAS screen apparently was painted two targets at 9 o'clock and at 10 o'clock, and the 9 o'clock was lit up red. That was not a scenario I set.

No, I'm saying there's only one target on the TCAS display - it's been misidentified as the aircraft you can see (which is not in conflict at present).

I think I might be starting to understand why we seem to be having a bit of an irreconcilable argument. You seem to be looking at your "scenarios" from the point of an experimental set-up, where you have a "God's eye" view and have set the pieces in place and started the clock running. You know what the optimal response should be because you know everything about this imaginary universe.

I'm coming from the real world and thinking I'm sat there with only a limited amount of information to go on. I know there's conflicting traffic somewhere and I also know there has been a breakdown of ATC. I can't be sure of where these aircraft are or indeed how many are involved. I *can* see something out of the window but hey, that's not unusual in itself. I get an RA: I know I'm not in possession of all the facts so I have very little choice but to follow it - I have very little reason to disbelieve the proffered escape advice; even if I did, do I think that with my limited view out and lack of a complete picture that I could do any better than TCAS? Not really.

There are known unknowns and unknown unknowns...

[bsieker]

FullWings,

I'm coming from the real world and thinking I'm sat there with only a limited amount of information to go on. I know there's conflicting traffic somewhere and I also know there has been a breakdown of ATC.

Industry standard or no, silence from ATC up until the point of the RA does not mean that you know ATC has broken down.

I get an RA: I know I'm not in possession of all the facts so I have very little choice but to follow it - I have very little reason to disbelieve the proffered escape advice; even if I did, do I think that with my limited view out and lack of a complete picture that I could do any better than TCAS? Not really.

I don't think anybody is debating that scenario.

And I would dare say that all agree that if there is no word from ATC, by all means, follow the RA. It's the best information you have, and in all likelihood will avoid the traffic.

The problem is when you get conflicting advice from ATC and TCAS, particularly if in the order:

- ATC advice
- starting following ATC
- TCAS RA

Which I do not see you considering at all. You seem to be stuck in the scenario where ATC has in fact failed. But the premise that that has always happened at the time TCAS issues an RA does not hold.

We know late ATC intervention happens, and just because it is not "industry standard", you cannot simply ignore it when looking at the entire socio-technical system, which includes two flight crews, two aircraft with TCAS, and ATC with primary and secondary radar.


Bernd

[FullWings]

The problem is when you get conflicting advice from ATC and TCAS, particularly if in the order:

- ATC advice
- starting following ATC
- TCAS RA

Which I do not see you considering at all. You seem to be stuck in the scenario where ATC has in fact failed. But the premise that that has always happened at the time TCAS issues an RA does not hold.

From a pilot's POV I have to assume that there has been a failure of ATC when in receipt of an RA and act appropriately. Why? If I assume that ATC has not failed and carry on with their advice (possibly against an RA) and I am wrong in that assumption, I could be part of an accident. If I follow the RA, it will keep me clear of traffic, whether ATC are giving valid commands or not...

I one of my previous posts I said:

No, the controller carries on trying to separate the traffic. If, in extremis, there comes a point where there are conflicting instructions from ATC and TCAS, TCAS is followed.

and

If it gets to the point of an RA, then TCAS "controls the vertical" and ATC are informed ASAP of a deviation. Climb or descent instructions from ATC are ignored as their information is not as up-to-date as that possessed by the TCAS computers. (Amongst other reasons.)

I think that covers the conflict case?

Off flying across the Atlantic now with my trusty TCAS, EGPWS and sextant.