Lets get right to it, shall we. Be advised, this may get a little complicated.
If you kept exploring around that FAA website, perhaps you found another Traffic Management "product": the AADC or Airport Arrival Demand Chart. It's difficult to explain how to get the most out of this chart, as the data is constantly changing for each airport. The idea is to double click on your airport, and the chart will populate with data. The horizontal white line is the Arrival Rate, as discussed in the previous post. Occasionally, the columns of arrivals will peak above the white line, indicating that the arrival demand exceeds the acceptance rate of the airport for that hour. If you select the 15 button instead of 30 or 60 to the left of the airport selection box, the data will show columns of arrivals every 15 minutes. You'll notice how arrival demands increase dramatically in very short periods of time. Also, you'll notice how the spikes in demand occur more frequently in the 15 minute increments, as they would in the 60 minute graphic. As I write this, LGA shows 2 hours of demand higher than capacity in the 60 minute display, but when showing the 15 minute display, there are 12 periods that exceed capacity, or 3 hours worth. So, between 11am-2pm, there are 12 instances when flights will incur a guaranteed delay. Proof of this is something that really caught my attention when I first started training: The SAME flights often are victim of the same delay every day due to this lack of capacity. The same flights are delayed on the ground or asked to hold in the air on an almost daily basis.
So, these few flights are delayed daily, regardless of the weather, because the airport can NEVER handle as many flights as are scheduled. So, when the weather becomes less than ideal, even more flights must be delayed. With proper planning, these flights will be held on the ground. Lets discuss how the length of the delay is calculated.
In the last post, I mentioned how the weather affects the two main variables in relation to airport acceptance rate. The overall visibility and weather, which determines spacing on final approach, and the wind, which mainly determines the runway configuration, thus setting the actual acceptance rate. With these two values set for the period of time that the weather will require use of said approaches and runways, Traffic Management Unit (TMU to save bandwidth) then determines the Miles in Trail (MIT) requirement for each arrival fix. To explain, each approach control airspace has 2-4 arrival fixes (fixes are imaginary points in space that aircraft can determine using their on board computers, though sometimes they are simply well-placed ground based navigational aids). All traffic landing at the main airport are always routed along charted arrival routes which organizes traffic flow over these main arrival fixes. Depending on the number of arrival fixes a MIT requirement for each fix is determined. If traffic on final approach must be 5 miles apart due to cloudy weather, and there is only one runway in use for landing, and there are 4 arrival fixes, TMU would require 20 miles between aircraft arriving from the same fix. Aircraft are sequenced to these arrival fixes by the Center, and the approach control (TRACON) would then take the 4 streams of traffic from different directions (each with planes 20 miles in trail of each other) and line them up on final approach 5 miles apart. Centers are much more adept to using vectors and speed control to achieve the 20 MIT needed, as the sectors are much bigger. Also, the 20 MIT requirement is often achieved hundreds of miles from the arrival fix, if the sectors close to the arrival fix are too busy or not big enough to get the spacing needed. I'll discuss that later on.
As we move farther from the airport, the spacing requirement will often grow. While the traffic flow may be organized in 4 flows close to the destination, this organization can't exist forever. Aircraft approach the airport from every direction, on a fairly direct route from their departure airport. At a point a few hundred miles from the arrival airport, the routes become standardized and they join the Standard Terminal Arrival Route (STAR). Each STAR may contain transitions which aide the pilot in determining the best way file his/her flight plan to join the STAR. There will be one Center sector that will be in charge of determining the sequence of all arrivals to one of the STARs. Traffic on each transition, or a set of transitions, would be worked by a different sector. These outlaying sectors would have their own MIT requirement based on their percentage of the overall traffic flow to that arrival fix. So, assuming the main sequencing sector is required to achieve 20 miles between planes to feed into the approach control, and there were 2 transition routes into the main sector, both with equal amounts of inbound traffic, TMU would determine that each outlaying sector should provide the main sector with 40 miles in trail. The main sector would vector the two flows together into a single line 20 miles in trail. This is rather simplified, but is meant to show the theory behind miles in trail. Often times, its actually reverse simplified. Every sector in the center working arrivals to that airport would have to provide the next sector with 20 miles in trail. This would cause the delays to happen slowly over the course of the flight, but would require every sector along the way to vector planes and delay them slightly. Flights would be constantly adjusted for each sector to accomplish its MIT requirement, as opposed to the plane being turned and slowed once or twice. This is where the ground delay comes in to solve this problem.
TMU has established a program that determines the amount of delay required for each flight so that it can depart into a predetermined slot so that it will experience the smallest airborne delay possible. Lets take a step back first.
We have explored MIT requirements per arrival fix, but lets look at it from a system-wide perspective. While, we know we'll need 20 miles in trail over a certain fix, the idea is still to determine a sequence into the airport from all 4 fixes together. We don't really care which direction they're coming from, only when they are going to arrive. Also, we'll need to figure out when they're going to depart and how long it will take to get to where they are going. To do this, the TMU at the arrival airport will determine the MIT per arrival fix. The TMU at the departure airport will delay aircraft on the ground so that they depart at a time that will allow the Center a fighting chance at achieving that spacing with minimal vectors.
I discussed how the arrival flows coming into each airport is organized into 4 flows. The same is true with departures. In Boston Center, for example, we have 5 fixes that aircraft fly over when they depart the Center. There are more, in reality, but these 5 are the fixes that aircraft use when they are flying to airports that need miles in trail. So, TMU has determined the amount of time it takes for airliners to fly from each airport to each of those 5 fixes. Oh, by the way, these fixes are SYR, HNK, HTO, SAX, and SBJ. Look 'um up. With these flying times in hand, they separate all departures by their arrival airport. All aircraft flying to Atlanta would go together, all those flying to Ohare would go together, etc. If we need 20 miles in trail (airlines fly about 7-8 miles a minute) 3 minutes between planes is the goal of the program. Each control tower is advised that every aircraft destined to the airports specified as needing MIT would need special permission to depart. The tower would call TMU on the phone, advise them of when the plane wants to depart, and TMU calculates the best time for that plane to depart, based on the other departures from the other airports in the center. The solution, of course, isn't simply to space the departure times of each departure by 3 minutes. A departure from Bangor, ME (BGR) can departure well ahead of the Providence, RI (PVD) departure, but the Providence plane will still be in front. The key is not to allow the Providence departure to leave at a time that will tie up the Bangor departure time wise. There would be a 6 minute window when the PVD departure can't go. Add in BDL (Hartford), ALB (Albany), BTV (Burlington), MHT (Manchester), PWM (Portland), and BOS (Boston), and things get complicated. Meanwhile, the airlines schedule their flights to depart at slightly different times so that they ARE TIED as they leave the center. The schedule is designed to do that. In the Hub-and-Spoke system of the major airlines, the goal for the airline is for everyone to arrival simultaneously at the hub airport so that everyone can connect flights and then leave simultaneously. This of course works out well, until we have to space you on final. Darnnit!
As the final controller spaces planes on final approach, the gap between flights increases. This gap expands out in time over the whole country. Each plane gets moved back further and further to increase spacing. Eventually, planes are delayed to a point where the last plane in line hasn't even left yet. You can wait for your place in line at the gate, or you can get vectored all over the sky. I'll take the ground delay, thanks (both as a controller AND as a passenger).
Next up, I'll discuss the impact of en-route weather and how sector capacity is a limiting factor as well as runways.