In remote areas that have no radar (oceanic airspace, etc.) how is aircraft separation maintained so that airplanes don't get too close?
What do areas that normally have radar do when there is a sudden radar outage? There are airplanes going everywhere, not necessarily on standard airways, and a lot of them are being given radar vectors. What happens when the screens go dark?
For oceanic airspace, flights proceed along specified tracks and altitudes specified in their filed flight plan. From there, once they leave radar-controlled airspace, radio reports are made to ATC, who ensure that aircraft are appropriately separated if a flight needs to change its track or altitude. See the wiki article the North Atlantic tracks, which give a good overview on the track system. Specifically:
Prior to departure, airline flight dispatchers/flight operations officers will determine the best track based on destination, aircraft weight, aircraft type, prevailing winds and Air Traffic Control route charges. The aircraft will then contact the Oceanic Center controller before entering oceanic airspace and request the track giving the estimated time of arrival at the entry point. The Oceanic Controllers then calculate the required separation distances between aircraft and issue clearances to the pilots. It may be that the track is not available at that altitude or time so an alternate track or altitude will be assigned. Planes cannot change assigned course or altitude without permission.
and
Contingency plans exist within the North Atlantic Track system to account for any operational issues that occur. For example, if an aircraft can no longer maintain the speed or altitude it was assigned, the aircraft can move off the track route and fly parallel to its track, but well away from other aircraft. Also, pilots on North Atlantic Tracks are required to inform air traffic control of any deviations in altitude or speed necessitated by avoiding weather, such as thunderstorms or turbulence.
I'm going to leave the "what happens when things break" to someone else.
You are asking two very different, but related, questions:
How is separation provided in procedural airspace? (that is, airspace where no surveillance exists)
What are the emergency procedures when a surveillance system fails?
I will address them in that order.
There are three main types of separation: vertical separation, horizontal separation and separation in the vicinity of the aerodrome. I am going to assume you are not asking about separation in the vicinity of an aerodrome. We can divide horizontal separation further into three parts: lateral separation, longitudinal separation and radar separation. Obviously, radar separation does not apply in procedural airspace.
Common for all separation types, when used in procedural airspace, is that they are based on position reports made by the pilots. When we do not have transponders to broadcast the position and level of flights, the pilots have to manually tell us where they are.
Vertical separation is pretty straight forward. It is very easy to apply, and we often prefer vertical separation (even in radar airspace), simply because it is so easy. Vertical separation is obtained by requiring aircraft to operate at different levels expressed in terms of flight levels or altitudes. In most areas today, 1000 feet is the minimum vertical separation. 2000 feet is still used as minimum in some places. For more details, you should research Reduced Vertical Separation Minima (RVSM).
Lateral Separation is obtained by requiring aircraft to fly on different routes or in different geographical locations, as determined by visual observations, use of navaids or RNAV equipment. The idea is, if two aircraft are flying in two different places, then they won't collide.
One example of application of lateral separation is two aircraft flying on two different radials of the same VOR (radio beacon). If the aircraft are established on radials diverging by at least 15 degrees and at least one aircraft is 15 NM or more from the VOR, they are separated - per definition, regardless of the actual distance between them.
Another common example is approach procedures and different holding patterns. Certain procedures are designed in a way that ensures sufficient separation. For example, an NDB approach procedure might be separated from a nearby holding pattern. Again, we do not care about the actual distance between the aircraft - the two procedures are separated, by definition, if that is how they are designed.
Longitudinal Separation is used for aircraft at the same level, flying on the same, reciprocal or crossing tracks. Longitudinal separation is based on time or distance. For example, two aircraft following the same route are separated if there is 15 minutes between them. That is, if aircraft A reports overhead point XYZ at 09:12 and aircraft B reports overhead point XYZ at 09:30, they are separated.
Now, 15 minutes is a long time, and it can be reduced in certain circumstances, but these are the kind of rules we have to follow in procedural airspace (when vertical separation cannot be used). There might be a hundred miles between two aircraft with 14 minutes between, but they are not separated.
You can also base longitudinal separation on distance, if two aircraft are operating directly to or from the same DME (distance measuring equipment). In this case, the minimum separation is 20 NM (and can be reduced in some cases).
In addition to the two above rules (15 minutes or 20 NM), there are many, many rules regarding crossing tracks, reciprocal tracks, separation during level changes, separation between aircraft with different speeds, and so on. Common for all of them is that the pilots will have to report their position, and then the controller has to apply separation in accordance with the rules.
That sums up the separation methods and minima in procedural airspace. Now for your second question: what if radar airspace suddenly becomes procedural airspace as a result of a system failure?
Needless to say, this is extremely rare. Systems are very reliable, and there are backups, and backups of backups. That aside:
The first thing to realise is that this would most likely be considered an emergency situation, which means that normal rules and regulations do not apply. Many pilots don't realise this, but the word "mayday" is not actually reserved for pilot use only. You will likely hear a transmission from the ATC unit that sounds something like "Mayday, mayday, mayday, all stations, radar service terminated due to equipment failure, stop transmitting, standby for further instructions". At this point, you should make sure the volume of your TCAS is turned up...
However, it is important to remember that controllers are trained to handle situations like this. Local procedures are different, but the controllers will either attempt to establish procedural separation, or transfer traffic to adjacent units, if they have radar available. Even when the system fails, the controller will have some way (besides memory) to keep tracks of all the flights in their airspace. This could be physical flight strips, or an electronic equivalent.
The first thing we will do is probably try to establish lateral and/or vertical separation. The idea being, if all aircraft are flying in different locations, or at different levels, they will not collide. We are allowed to use half the normal minimum vertical separation (that is, 500 feet) in emergencies, and this essentially doubles the number of vertically separated levels available. Do not be surprised if you are instructed to fly at flight level 375.
What happens next depends entirely on the details of the outage. Traffic toward the affected areas will be kept on ground or rerouted, and adjacent ATC units with radar coverage will assist in evacuation of the airspace, which will most likely remain empty until the situation is solved. Controllers operating in radar airspace are probably not certified to provide procedural separation, so once the airspace is evacuated, it will remain empty until the system is back up - it cannot simply be changed to procedural airspace and normal operations resumed.
When the screens go dark ATC breaks out the flight progress strips (and possibly the shrimp boats or other airplane-substitutes to lay on a map & push around) and uses brain power to substitute for the computer and radar. Check out one of my favorite "Say Again?" columns over on AvWeb which talks a little bit about it (and you can find lots more if you browse the archives).
Broadly, there's a few implications to non-radar operations - both routine and "when stuff breaks" - most of which are concessions to safety. The one most pilots will feel, and the biggest safety item, is that "random RNAV routes can only be approved in a radar environment" (you'll find that gem in the AIM, but the translation is this: "GPS Direct? YOU NO CAN HAS!").
To facilitate separation in a non-radar environment the controllers are going to chuck you (and every other IFR flight) on airways, at standard IFR altitudes, quite probably with a specified airspeed, and they're going to route you navaid-to-navaid.
The idea here is simple: "You gotta put them together to keep them apart": By ensuring all flights are on known airways, at known altitudes, and at known speeds controllers have a good idea of who is where in their airspace. They will supplement this mental picture by asking pilots to report crossing certain fixes (intersections/radials), so that they know exactly where that flight is when they make their report (which will help them account for winds aloft, and give them an idea of what the flight's ground speed is in addition to the airspeed they're asking them to fly).
All of the above provides IFR-to-IFR separation, which is what the controller is concerned with. IFR-to-VFR separation (which is normally supplemented by radar) becomes the responsibility of the pilots ("See and Avoid") - this means it's largely a factor of altitude (that 500-foot difference between IFR and VFR cruising altitudes and those cloud clearance requirements should start to seem pretty important right about now).
ATC aircraft separation is built upon the assumption of no radar by way position reports. At compulsory reporting points (black triangle on enroute charts), an IFR aircraft (unless radar identified) is required to report (see 5-3-2 of the AIM):
(a) Identification;
(b) Position;
(c) Time;
(d) Altitude or flight level (include actual altitude or flight level when operating on a clearance specifying VFR-on-top);
(e) Type of flight plan (not required in IFR position reports made directly to ARTCCs or approach control);
(f) ETA and name of next reporting point;
(g) The name only of the next succeeding reporting point along the route of flight; and
(h) Pertinent remarks.
ATC uses the position and ETA to keep aircraft separated. Most of the US is covered by radar, but remote parts of Alaska and oceanic still use position reports. If ATC radar is not functioning (rare), then the system reverts to position reports.
Moving forward, ADS-B allows aircraft to self broadcast its position. A network of ground and satellite ADS-B receivers can build up a picture of the traffic.
