Originally Posted by bucket&spade
Of course, with a bit of legwork we can find the info elsewhere ourselves but as we're on the subject and some people here have hinted at some knowledge of the subject, it's perfectly reasonable to ask for more info on this thread.
Oh dear. Can of worms opened!
I bring two perspectives to this problem. First, the perspective of someone who pursued the qualification of journeyman mathematician but never put it to 'professional' use. Second, as a professional pilot.
So, take a reasonable grounding and interest in mathematics, an imperfect knowledge of avionics and properties of electromagnetic waves, and a particular interest in not hitting other aeroplanes, and this might be the result.
or....
Caveat Emptor!
The algorithm used in TCAS computers is from the branch of mathematics some call Probability. Having first detected a likely bearing and range to a target, the TCAS unit then has to decide, using probability mathematics, which of the reported replies are likely to be aeroplanes. In the same way that VOR's may scallop, and surveillance radars might paint flocks of birds or lorries as targets, not every whisper-shout return the transponder reports to TCAS might be an aeroplane at the calculated range, bearing, and relative altitude.
The algorithm matches three or four valid replies or more, selects those that are most probably
an aeroplane travelling in a straight line.
An aeroplane that is travelling in a straight line, can be computed faster and a projection of its future location made, using fewer such returns and the solution found much faster than for an aeroplane that is turning.
Once it determines that a series of valid returns are matched using probability mathematics to be aeroplanes, the algorithm makes forward projections based on current vehicle states, and issues alerts as required.
If for some strange reason you wanted to confuse the TCAS computers of other aeroplanes in the same airspace, then a good party trick would be to either (a) accelerate or decelerate, and/or (b) make a rapid turn.
Because
then you are making your aeroplane behave in a way that the
other proximate, probably Mode S, aeroplanes will think is a 'very un-aeroplane like' manner.
That is, you no longer fit the model of 'aeroplane-like-behaviour' that other TCAS II computers use to filter out multipath returns, etc.
Also, out of the range, bearing and relative altitude information that TCAS algorithms filter, understand that it determines range not by GPS coordinates transmitted -- TCAS can independently calculate range from signal gain in the whisper-shout transmissions between transponders. Squaring the range returns, and finding the asymptote of the resulting parabolic curve is the input to range calculations and closest point of approach. And (thankfully) I can point you to one internet paper on the subject confirming in mostly plain language that asymptotic values of range-squared are too low to track a turning target.
http://www.dtic.mil/cgi-bin/GetTRDoc...c=GetTRDoc.pdf
Humans (ATC, pilots) can separate aeroplanes by changing their speed, their heading or their altitude. TCAS sought an automated solution too 'see' further than pilots and simultaneously calculate more potential threats than an ATC. When melding mathematics, electronic and avionics engineering, it was far easier to implement a fully automated solution based on vertical speed and altitude than accelerate/decelerate, lateral separation, or turning. Simpler with existing technologies.
When avionics can be made of newer computing technologies that are not simply bigger and faster Von Neumann machines (fer chrissake they still have MODES!!!), we might see other mathematical and technical solutions to collision avoidance.
As I said, that part is the formulation of a trained, but amateur, mathematician. You are free to discount it all as a load of b0llocks if you like. Now lets talk as professional pilots.
so I think a turn to AVOID an RA situation is an appropriate response.
I disagree.
When we were introduced to the rules of the air, we were taught right of way, and overtaking, who is it that has the other on his right, etc.
Conflict? Resolve by
turning. Primacy of learning, what ever you call it, or maybe its how we solved conflicts in two dimensions on the surface. Aeroplanes cannot stop,so... Turn!
Take a look at any airprox or TCAS event recorded in any State's accident/incident database and you will find they are replete with pilots that
turned in response to a threat of midair collision.
Key point: In response to.
Some of those turns made the situation worse.
As someone said before, any search on TCAS training materials (there are some excellent ones by Eurocontrol) and you will quickly find that the TCAS software tracks many more targets than it displays, and probably more than most pilots can process (which is about seven, plus or minus two, but thats another topic).
A 'lesser of Rate One or 25 deg AoB' turn probably fits, within the TCAS algorithm of moderate turn.
But anything more than that, such as the kind of turn that might be considered by a pilot that thinks they need to 'do something' in response to a TA, or 'do something' extra in an RA, is in my mind making it harder for the TCAS system to do a job that it does very, very well. Using altitude and vertical speed.
Thus far, noone here has suggested that they firewall the power levers, or slam them shut, throw out the airbrakes gear and flap, as a collision avoidance manoevre. Fair enough too -- what would that achieve?
Much the same as turning. Might work if the aeroplane you have eyeballed is the right threat, and only threat. Otherwise great potential to make things worse. Turning does not fit the definition of aeroplane like behaviour in the probability mathematics that is the heart of the system that is alerting you and protecting you.
My Operations Manual clearly states....
My training organisation has instructed self and colleagues
not to
initiate a turn
in response to a TA or RA.
As a working pilot that follows instructions from above, that is what I do. As someone with a background and an interest in the underlying maths and tech, I believe I understand some of the reasons behind that instruction.
.... "If an RA is given in a turn, the turn may be continued"
Amateur mathematician says: That sounds reasonable. Your turn onto an ATC heading or following an LNAV track is probable at or much less than rate 1 or 25 deg AoB. Moderate. And you are not considering
initiating a turn in response to a threat, you are in un-accelerated flight.
Pilot says: It is also your instructions in your FCOM. Follow those instructions.
