follow all of the rules, even those that do not make sense
Here, gentle noder, we delve intrepidly into the murkily-nuanced world of 'ensuring separation' and 'deemed separation'. I never thought it was possible to hear thirty TATC's eyes rolling simultaneously—as they do when TATC thirty-one broaches this subject in a tutorial—but it's there with the weary murmurs and noticeable deflation.
Both said concepts have bitten me a number of times during my practical training and I present here my interpretation of both. I have little misgivings about such mild indeterminacy since 'ensured separation' is something of a fluid concept, that I and others sometimes wonder if even our instructors understand sufficiently to teach.
'Deemed separation' is something that fairly frequently keys into 'ensuring separation' (and usually failure by a student to do so during a simulator exercise), so I will describe it first. It is not a specific idea in any general guidelines for controllers, but as applied to specific contexts it has a legal basis. Before we get into it properly it's important to talk a little about how aircraft navigate.
When aircraft transit controlled airspace they are usually - if the controllers can help it at all - on what is referred to as their 'own navigation', or 'own nav'. This means the aircraft will be following its own pre-planned route through the sector, rather than flying a direction instructed by ATC. ATC will be aware of said route, since it is printed on the aircraft's flight progress strip. Let us also be clear that 'own navigation' refers to the horizontal dimension only; levels are assigned by controllers and barring emergencies or other such nonsense, aircraft must adhere to them.
Although it is nominally true that an aircraft on 'own nav' is flying a known route, it is assumed for safety purposes that it could, for example, make a 90° turn to avoid a CB cloud that has just materialised in front of them.
Conversely, aircraft locked on a heading - that is, flying a direction instructed by ATC - will, for all intents and purposes and barring things like engines falling off or being attacked by a giant winged leviathan, fly in that direction forever until they are instructed otherwise. You have to pay extra attention to the direction of aircraft on headings, not just against other aircraft but things like the edge of controlled airspace. This means we prefer not to use headings if we can avoid it. One of the things you always check when scanning through your aircraft is whether any of the aircraft you've currently got on headings can be put back on their own navigation.
So, with the 'own nav' definition made, it should not come as much surprise when I say that two aircraft flying on their own navigation must generally be separated by levels. If you had five miles of horizontal separation between two aircraft on their own navigation, this could be eroded without warning if, for example, one of the pilots turned their aircraft to avoid bad weather. There are a couple of exceptions to this, both of which fall under the heading of 'deemed separation'.
First, aircraft inside controlled airspace are deemed to be separated from any aircraft outside, regardless what actual separation exists (obviously some discretion on the part of the controller is needed here, but I won't go into that now). This is described in a bit more depth here. Second, the Civil Aviation Authority may deem two features that aircraft follow as safely separated for aircraft flying on their own navigation. This might be, say, a road and a parallel railway line or perhaps two parallel airways. Whatever they are, they will be specified in the Manual of Air Traffic Services (MATS) Part 2 for the ATC unit whose airspace those features are in (MATS Part 1 being a generic procedural manual and guide).
In our 'basic' simulator we have a section of airspace much like this:
The dotted lines mark the centre lines of two adjacent airways - N864 and R8. N864 is a westbound airway; R8 is eastbound. An airway is ten nautical miles wide, so the distance between the two centre lines is also ten miles. The solid lines are the boundaries between the airway (controlled airspace) and uncontrolled airspace. Bandit country.
Our (very small) MATS Part 2 details the 'deemed separation' that exists in our airspace. It applies to these airways. The 'deemer', as we call it, is that all Westbound aircraft on their own navigation routing along, or North of the centre line of N864 are deemed to be separated from Eastbound aircraft on their own navigation routing along, or South of the centre line of R8.
Without knowing what I said earlier about the whole 'own nav' wild card, this may have seemed a little odd. In the lesson where this issue first arose (before simulator exercises had started), the instructor said more or less what I have just written. Immediately one of the new students muttered: "'Deemed separated?' What's he on about? They are separated!"
This is one of those things that makes less sense under examination; much like the six-foot concrete wall that the controlled airspace boundary is treated as. No, there is nothing more to stop aircraft following N864 from veering to avoid weather than those following any other airway. Nor does the controlled airspace boundary physically protect aircraft inside from aircraft outside any more than the solid white line delineating a motorway hard shoulder prevents people using it as a driving lane, but in both cases we treat it as if it does. A pair of Concordes could be flying in opposite directions respectively along N864 and R8 at mach 2.02 - which is a closing speed of about 3,000mph - on their own navigation only ~10nm apart, at the same altitude, and would be deemed to be separated. If this makes you uneasy, it should do. These kinds of 'deemers' exist all over the place and the separation they provide is entirely theoretical.
Okay, we're getting close. 'Deemed separation' and 'ensuring separation' both refer to separation in the horizontal dimension. If you don't have vertical separation - though since screwing around in the vertical dimension is so much easier, that is our preference - you have to have horizontal. With the previous comment in mind, we generally only use horizontal separation when vertical separation is not possible for some reason: e.g. one aircraft wants to climb through the level of another.
As far as horizontal separation is concerned, if you don't have 'Deemed separation' you have to have separation using radar vectoring: this means locking aircraft onto headings. Horizontal separation has to be 'constant or increasing'; that means tracks ('track' being the path an aircraft follows over the ground) that are parallel or diverging. For example, two aircraft five nautical miles abreast, both flying a heading of 180°, are separated horizontally. So are two aircraft whose tracks are five miles abreast, one flying a heading of 180° and the other flying a heading of 360°.
If the tracks aren't parallel they should really be diverging; you can get away with converging headings if you're very, very certain about what you're doing. If the two aircraft are far enough away from each other that you can see they will have vertical separation long before being less than five miles apart, or if you set the headings up so one aircraft will go at least five miles behind the other, then it's fine. If there is any uncertainty, just go for parallel or diverging headings.
* * * * *
I will give you an example of how I fell foul of all of this, or apparently 'failed to ensure' separation. As you might gather, a number of us trainees have our own views on the seemingly-arbitrary application of this concept. Anyhoo, refer to the diagram below. It isn't an accurate representation of our airspace by any means but it is sufficient for this. Notice the spur: this is another airway, T45, which intersects with N864 and R8 from the North.
\ \. \ AAL67
\ \. /250 E
\ \. / \
\ \* \ UAL91
__\ \______________________________________ /180 W
<------------ ~50nm ------------> /
We, or rather I, have American 67 and United 91 on frequency. The American is at 25,000ft, heading down T45 to join R8 and exit to the East, routing A -> B -> C -> D, these being navigation beacons. The United is heading West along N864, routing towards A, its next waypoint. I have given it a climb to 22,000ft (again I'm not going to distinguish altitude, flight level and height for this writeup) and it's on the way up. We have an agreement with the sector to the East about the levels aircraft enter their airspace at, so the American needs to descend to 12,000ft before reaching D.
Although I've got the strips in front of me and spot it that way, even without them you should be able to see a nominal problem here: the American needs to descend through the level of the United. After looking at the radar I decide I'm fine with the situation; by the time the United reaches A the American will be way past it, probably over halfway between C and D. I give the American its descent.
American six-seven, descend flight level one-two-zero, be level by D.(yes, the beacons do have slightly more useful names in reality)
Descend flight level one-two-zero level by
D, leaving flight level two-five-zero, American six-seven.
Now, being the obliging and thoughtful lot we are, we do cut a few miles off aircraft routes where we can. For one, it means we can dump them on the next sector sooner, which as I've said before should be the goal of every air traffic controller. We can tell aircraft to go directly to a particular waypoint on their route, cutting out any preceding waypoints. This allows us to make a certain degree of adjustment to their direction but still leave them on their own navigation. Since it avoids using headings, this is teh awesome.
The picture is now like this:
\ \. \ AAL67
\ \. /238 E UAL91
__\ \. / \_____________________________________ /195 W
I decide to cut the corner off the American's route, just because I can:
American six-seven, route direct C.
Direct C, American six-seven.
American 67 now turns towards C. The two aircraft are about 55 miles apart at this time. I do not have the aforementioned 'deemer' in this situation, so am relying on the radar and estimates of aircraft climb/descent rates to get vertical separation. After the level swap has occurred and the American is safely 1,000ft below the United, the picture looks something like this:
\ \ \
\ \ \ UAL91
__\ \ \_________________________________/195 W
The two aircraft are now vertically separated, 40-45 miles apart, on what is considered to be the 'same' track. This is where the track of one aircraft diverges from the other by less than 45° or, if the two are heading in broadly opposite directions, where the track of one aircraft diverges from the reciprocal of the other by less than 45°, which is what this is. I never had a problem with any of this. About thirty seconds later, AAL67 was on R8 and the two aircraft subsequently had 'deemed' separation anyway, not that they needed it.
At the end of the exercise, the instructor raises this situation. Says he wasn't happy with what I did there, and that I hadn't ensured separation. If the radar failed, I would have no way of knowing if one of them suddenly did a wild turn that put it head-on with the other. I contend they were never less than forty miles apart (which is the minimum procedural - i.e. not using radar - distance-based horizontal separation for aircraft on opposite-direction tracks). From looking at the radar I was happy that they would not hit each other, even though I wasn't using headings.
The instructor then invokes the universal question:
"So your family's on American 67. It's here (he points), the United is here (he points again) and you're not using headings - are you going to give the American its descent?"
"Yes. The only way those two are even going to get near each other is if United suddenly teleports about thirty miles West. If my entire extended family, their parents, their parents and their collective offspring for three generations filled both of those planes I would still do it. Cast-iron, one hundred per cent."
"You still can't do it. You haven't ensured separation"
In retrospect, even though this incident only happened a couple of weeks ago and I still gripe and grumble about the somewhat specious nature of the whole thing, I can sort of see the point. I maintain its application in this particular circumstance is a little tenuous, but since to pass the course I need to observe it 'correctly' should similar situations arise, I will do so.
In this situation, it is my opinion the 'deemer' wasn't an issue since when the two didn't have vertical separation (i.e. they were less than 1,000ft apart vertically) they were over 45 miles apart. Even if the United had done a turn that put it head-on with the American, they would have achieved vertical separation before closing to less than 40 miles. When the 'deemer' was present (i.e. the American had got onto the R8 airway) it wasn't needed, since the two aircraft were over 8,000ft apart vertically.
This was all clarified later in a tutorial session, where a colleague raised a situation virtually identical to the one I just described. The answer from the instructor, who is also our course Verifier (if we have technical queries and are getting different answers from instructors, the Verifier's answer is gospel) was that if you're using airways R8 and N864 and don't have the deemer, you have to use headings if you don't have vertical separation. Always. Meh. So, reproducing the earlier picture at the time I gave the American its descent:
\ \. \ AAL67
\ \. /250 E
\ \. / \
\ \* \ UAL91
__\ \______________________________________ /180 W
So, the best thing to have done with the previous comments in mind would have been to tell both aircraft to continue on their current headings, maybe turn the American right 10°. This would move the crossing point of the two aircraft West by about 10 miles, though that's of debatable use since the American is in the United's 1 o'clock at this time and crossing right to left, still 50 miles away. The headings are just to make sure that the separation that has been set up isn't eroded.
Of course the whole thing falls down if you consider that if the radio or radar failed now, the American is going to merrily trot outside controlled airspace if it continues on that heading for long enough, and depending on the airway structure outside the limited picture I've produced here the United might do as well.
'Ensured separation' seems to be a somewhat contrived and short-term concept, further complicated by the slightly differing interpretations of some instructors. A sort of 'you know it when you see it' sort of thing. You can happily consider two aircraft that are 80 miles apart, on their own navigation travelling head-on to be horizontally separated, but with lesser distances you have to think about using headings.
The point at this early stage - where you have to demonstrate understanding and application of all the rules in order to be permitted to bend them in later training - is not to issue instructions if you can't be confident they're safe without using radar ('safe' as in no aircraft lose separation with each other). If for example, you get two aircraft to exchange levels and put them on converging headings to do it (bad idea anyway), it is likely you'll have to keep a diligent eye on the radar to make sure those headings don't put them too close together, to say nothing of possible unscheduled forays outside controlled airspace. If the radar failed you wouldn't be able to verify that, and if your radio or the radio of either aircraft failed you wouldn't be immediately able to 'fix' those headings if needed.
The faintly absurd assumption that we can't rely on the radar raises comments from TATCs not devoid of merit, like "what's the point of having it, then? I thought this was a radar skills course," particularly since we know that 'ensuring separation' goes largely out the window in subsequent training courses. For now that's neither here nor there; we follow the rules, even those we don't understand, and hope we pass.
(timidly) Any questions?
<<Lesson #1 | Lesson #3>>