Learn › The Auto-Controller

How the Auto-Controller works

An automated approach controller that works the whole sky — and shows you how it is done.

Switch on the Auto-Controller in the ATC Radar Simulator and an automated controller takes the frequency. It assumes the arriving flights, vectors them into an orderly stream, clears them for the ILS, hands them to the tower, climbs the departures out — and keeps every aircraft safely apart while it does it. It is not a trick of the game engine: it controls the traffic with exactly the tools you have, through the same radio, under the same rules. That makes it a strong partner when the sky gets busy, and an even better teacher.

Turning it on

On a computer, tick Auto-Controller in the Simulation section of the left-hand panel. On a phone, open the ☰ menu and switch it on under Simulation. It works in Approach sessions at every airport; the Area Control Centre has no Auto-Controller. It pauses with the simulation, and runs at whatever simulation rate you choose — at 8× you can watch a whole rush hour unfold in a few minutes.

Only a controller's tools

The Auto-Controller has no special powers. It cannot move an aircraft or change how it flies; it can only give the same instructions a human controller gives: headings (radar vectors), levels, speeds, direct-to clearances, holding, the ILS clearance, and assume and hand-off. Every instruction goes through the same command system as the ones you type, appears in the R/T log tagged ▲ AUTO, and is read back by the pilot. Aircraft then fly it with the simulator's normal flight physics — turn rates, climb and descent rates, acceleration — so a clearance takes time to work, exactly as it does in the real world.

The rules it keeps

• Any two aircraft are kept at least 1,000 ft apart vertically or 5 NM apart horizontally.
• Aircraft landing on the same runway are kept at least 10 NM apart on final.
• Arrivals within 10 NM of each other are also stacked 1,000 ft apart, each one above the one ahead of it in the sequence.

These are wider than the 3 NM / 1,000 ft minimum the game's conflict alert enforces — the Auto-Controller deliberately leaves itself a margin, the way a real controller works to a buffer rather than to the absolute minimum.

How it thinks: three minutes ahead

Every two seconds of simulation time, the Auto-Controller looks at the picture and asks one question: where will everyone be over the next three minutes? To answer it, it runs a throwaway copy of the simulation — the same flight physics the game uses — and plays every aircraft forward along its current clearance. Before it issues any instruction, it tests it in that copy: a clearance is only given if the aircraft stays separated from everyone else's predicted path. A clearance that would solve one problem but create a sooner one with a third aircraft is rejected.

When the prediction shows two aircraft heading for a conflict within the next two minutes, it resolves it in a fixed order of preference: first a level change (1,000 ft of vertical separation is quick and leaves the routes alone), then a speed change, and only then a heading. Further ahead than two minutes it waits: arrivals will have turned base, joined the downwind or been cleared lower by then, and a heading given for a conflict that never happens is just noise on the frequency.

Sequencing the arrivals

The heart of approach control is turning a scattered crowd of arrivals into an even stream on final, and the Auto-Controller does it the way modern arrival-management tools do: in time, not distance. Each arrival is given a slot a fixed interval behind the one ahead, worked out along its planned path and in the wind, so that when the leader touches down the follower is at least about 10.6 NM out. Planning in time matters because a stream slows down as it nears the runway — the leader is at 140 kt while the follower is still at 180 — and a gap planned in miles keeps shrinking.

To make the gaps, it flies a classic pattern rather than a scramble of headings:

• Levels go straight to the lowest available — typically 4,000 ft on the downwind and 3,000 ft for base and final, raised at high airports such as Sana'a — never step by step.
• Speeds are the standard ones: 250, 220, 180 kt, only ever reducing — and, unless it is well behind its slot, never faster than the aircraft ahead.
• Headings are few: one to join the downwind behind the traffic ahead, one for the base turn, one to intercept the localizer at a gentle angle.
• Delay is absorbed by extending the downwind — the "trombone" — and by speed, never by orbits.

Once an arrival is established on the localizer, it is handed to the tower and leaves the frequency.

Departures

Departures are cleared to climb only as high as stays clear of everyone else — the arrivals above all — and are given the rest of their climb once the traffic is past. If an arrival and a departure are in conflict, the departure gives way: it is far easier to stop a climb than to hold up an aircraft on its way down to a runway.

A busy sky

When ten or more arrivals are on the frequency, the trombone runs out of room and the Auto-Controller starts to hold: it sends the surplus into stacks from 5,000 ft upwards and pulls them out one by one, each when it can still slot in far enough behind the aircraft ahead. Combined with auto-generated traffic (a Premium option that keeps new flights coming at up to 20 a minute), you can watch it manage a sustained rush — the closest thing in the game to a real morning at a hub.

Emergencies

An aircraft squawking 7700 goes to the front of the sequence, behind only those already turning in. A 7600 radio failure cannot hear any instruction, so the Auto-Controller never talks to it: it works out when that aircraft will arrive on its own lost-communications procedure and builds the rest of the sequence around it, keeping everyone else out of its way. (See squawk codes for what those codes mean in real life.)

Working alongside it

You do not have to choose between controlling and watching. With the Auto-Controller on, you can still give any aircraft an instruction — and when you do, it steps back from that flight for 90 seconds, leaving it in your hands. Take the departures while it runs the arrivals, or step in when you would have done something differently and see how the picture develops. One rule keeps personal bests honest: a session in which the Auto-Controller has worked is marked assisted — its score is shown, but it does not count as a personal best.

How to learn from it

• Watch at 8×, then at 1×. At speed you see the shape of the plan — the downwind stretching and shrinking, the stream forming. At normal speed you see the individual decisions.
• Read the R/T log. Every instruction tagged ▲ AUTO is a decision with a reason. Ask yourself why that aircraft got that level, at that moment.
• Notice what it does not do. Few headings, no orbits, levels in one go, speeds only ever reducing — economy on the frequency is a skill in itself.
• Take over one aircraft. Give it your own vector and see whether your plan or the Auto-Controller's would have landed it sooner.

Real controllers and automation

Real air traffic control uses a great deal of automation — arrival managers that propose a landing order and the delay each flight must absorb, conflict probes that look minutes ahead for trouble, and the short-term conflict alert that warns of imminent losses of separation. But in real life those tools advise: the instructions are still given, and the responsibility still carried, by a human controller. The Auto-Controller goes further, because a game can — and in doing so it puts on display the same principles a controller learns: predict, plan in time, separate early, and keep the frequency quiet.

New to the job it is doing? Start with how approach radar control works, then read about holding patterns, the ILS approach and wake turbulence.