P. 614.292.4120
F. 614.292.7852
E. smith.131@osu.edu
Institute for Ergonomics
The Ohio State University
210 Baker Systems
1971 Neil Avenue
Columbus, Ohio 43210
|
Human
Factors Issues in the Design of a More Flexible ATM Environment:
Knowledge Elicitation Based on the Expanded National Route Program
*Institute for Ergnomics, The Ohio State
University, **Department of Aviation Ohio University, ***Aviation
Research Laboratory University of Illinois at Urbana-Champaign
Introduction
| This report describes the results of a knowledge acquisition session
involving the supervisor for air traffic services and two dispatchers
from a major airline. The purpose of the session was to use a scenario
based on experiences with the expanded National Route Program (NRP)
in order to elicit the knowledge and views of these practitioners
regarding areas for improvement of the expanded NRP and for developing
new Air Traffic Management (ATM) initiatives in the future. These
improvements discussed fall into six categories: The report then ends with a consideration of some of the issues that must be dealt with to ensure safety and effective overall use of system capacity, with an emphasis on the role of the ATM system as a "referee". Methods | The goal of this knowledge acquisition session was to elicit knowledge and insights from airline AOC staff about current and future ATM system initiatives. In order to generate discussion, a number of scenarios were presented to the participants based on actual experiences with the initial implementation of the expanded NRP. These observations were based on previous research conducted by Smith, McCoy, Orasanu, et al. (1995). The rationale behind this choice of scenarios was that the expanded NRP represents the most significant attempt to date by the ATM system to provide the airlines with increased flexibility to make decisions about selecting routes for flights based on their business concerns. Consequently, experience with that program offers the potential to provide insights into the issues that need to be considered in making decisions about the design of the ATM system in the future. Subjects | Put
info on the participants here and ask each for a description of
their AOC- and ATC-related experience (jobs Procedure | This
data collection session lasted 6 hours, and was held in February
of 1996. The format for the meeting Instructions to Subjects
| The goal of the short observations or scenarios described below
is to focus attention on preflight planning at the airlines, and
on the implications of such planning for air traffic management
and air traffic control. These scenarios will be used to begin to
address the following questions: Thus, the basic question raised by this scenario is: What can we learn from experiences that arose during the initial implementation of the expanded NRP in order to guide us in future decisions about the design and implementation of the ATM system? This includes future decisions regarding the roles and responsibilities of the various people and the technologies embedded in such a future system. The Scenario
| The order for the expanded NRP took effect on January 9, 1995.
This program was phased in at progressively lower altitudes over
the following twelve months. Advisory Circular 90-91 had been in
effect for several years at that point. To produce this scenario,
we have documented a number of issues and situations that arose
during that time period (some represent improved system performance,
while others represent new or continuing problems). These issues
and situations are described below as a set of observations, supported
by anecdotal information elicited from their sources. On a particular day, the ATM system predicted that San Francisco would only be able to accommodate 50% of its normal arrival capacity in 3 hours (from 10-11 am) due to bad weather. Ground delays were therefore initiated for a number of flights at various airlines. The meterologists for one such airline, however, believed that there was a significant chance that the weather at San Francisco would clear before 10 am. Given that internal forecast, and given that the flight could be diverted to Oakland if the weather didn't clear, the responsible dispatcher concluded that it it would be preferable from a business perspective to launch a flight from DFW to SFO that was being held because of the ground delay program. The ATM System did not accommodate this request. Observation 4.
Because of a closed runway, the arrival rate into LaGuardia for
the next 3 hours was restricted to 75% of the normal rate. ATCSCC
put a ground stop program into effect, holding a number of flights
at other airports on the ground. Several affected airlines wanted
to substitute some of their delayed flights for others that were
still scheduled to leave on time, as these delayed flights were
much more time-critical in terms of passenger loads and connecting
flights. The ATM system had no easy mechanism to accommodate this
desire. You can take a 20 minute ground delay and then be assured that you can be sequenced into the flow at the northwest cornerpost, or you can take off now with a 20% chance of being fit into that sequencing and an 80% chance that you will be vectored to the northeast cornerpost instead. The airline decided for business reasons to go ahead and launch the flight. Observation 6. The expanded NRP raised certain concerns at some (but not all) Centers. As an example, one challenge arising from the expanded NRP was that airlines sometimes wanted to cross their high altitude flights over departure and arrival routes. For instance, for certain flights over the top of O'Hare, Chicago Center had always preferred that the traffic be routed over Badger in order to avoid having enroute traffic cross the departure lanes. One airline, however, preferred (and was now been filing) these flights over Iowa City-Waterloo under the expanded NRP. Such flights criss-crossed through the departure lanes, creating a "very tricky, complex operation" for ATC. This raised an interesting question: Do you let 3 or 4 planes cross at the cost of slowing departures by about 20%? This tradeoff was particularly interesting because such flights were most often slowing departures from Chicago of flights by two other airlines. In contrast, at Oakland Center such criss-crossing traffic through arrival lanes is being allowed, but not without significant concern: We're seeing a funnel effect. Where it used to be we would have 2 streams down to the Los Angelos basin, now everybody aims for Avenal. This works ok a lot of the time, except when things go wrong. When they go wrong, you suddenly have a whole herd of airplanes pointed at one point and a last minute change such as the need for an increase in the miles-in-trail or an airplane taking a wrong clearance, or any number of things, can cause a serious problem because you have a real concentration of airplanes. Suddenly you have to pull the plug to deal with it, whereas before at most you had a miles-in-trail situation, you had a bit more leaway. Even when it's workable, it puts a burden on the controller because it's not something they do every day. If you just take those aircraft crossing Coaldale at 39,000 feet, they've got to cross Modesto at 24 and you've got airplanes southbound at 29, 33 and 37, northbound at 31, 35. It's pretty much like a charge. It makes it real hard to thread that guy from 39 down to 24 and at the same time do all the things you have to be doing. If it was just one airplane to get through anybody could do it, but we're talking multiple airplanes, because it's the old push time and they're all coming at once. 20 minutes later there may not be an airplane in the sector, but for those 20 minutes it gets wild and crazy. Not having them on the routes the way they used to be has caused a lot of heartburn for the controller. They're having trouble getting them down. Another concern introduced by the expanded NRP was due to the instruction to give preference to NRP flights. As one TMO stated: "It gets cumbersome because the NRP program says youÕre not supposed to touch them. So we have to move 3 or 4 other airplanes to solve one problem. They [the Command Center] encourage you to move other traffic to leave the NRP traffic alone." A fourth example that arose as a direct result of the expanded NRP at some airports, and also as a result of an increased numbers of direct flights, had to do with balancing of loads at cornerposts: "If we get a jetstream right out of the southwest part of the country, everyone rides it [into O'Hare]. 75 percent of these airplanes are all coming in at the southwest cornerpost, creating a major volume saturation point. The old solution was to create a delay program to avoid launching too many flights into traffic, for example creating 20 minute delays at an airport, and to increase capacity by moving half a dozen flights to the northwest cornerpost. Under the expanded NRP, controllers were not allowed to use the latter solution. Controllers had no way of policing those flights, because if it said NRP it meant they were not supposed to reroute them. They were supposed to leave them alone." A similar situation arose as a result of flights filed by a single airline which "had 5 flights which originated in the LA Basin, PHX and LAS. When they were all filed to the Soutthwest cornerpost at DTW during certain arrival banks, the result was an overload at that fix. We responded by moving a couple of those flights, or other flights originating in Florida, to another fix. It would have been cheaper for that airline to file some of them to that other fix to begin with." A sixth example of an issue associated with the actual expanded NRP concerned what was happening when there were arrival rate restrictions (due to weather, etc.). For instance, in one case Kennedy had set a reduced arrival rate of 50 percent at 2 p.m. because of the weather forecast. To deal with this, Chicago Center began limiting flights bound for Kennedy that were flying the standard pref routes. In addition, however, there were flights filed under the expanded NRP that were not limited. The net result was that the capacity for Kennedy was exceeded, with many planes "winding up in high altitude airborne holding, and that's a major problem." Finally, holding itself has become more difficult: If they do go into holding, you might not have these airplanes anywhere near the fix you have to hold them at. You have to pull them in from all corners of the sector, where before they'd all automatically be going over this point and all you had to do was issue a clearance and watch for the altitude. Now you've got to transition people all the way from the north and south to a common point and probably start doing a lot of altitude juggling at the same time. Observation 7. At certain Enroute Centers such as Chicago and Cleveland, numerous examples were noted where new problems with air traffic congestion arose. It is important to note, however, that many of these concerns did not arise because of the preflight plans filed under the expanded NRP. As one TMO reported: The source of the major concerns was not the expanded NRP per se; it was the spin-offs of the expanded NRP, such as the increase in direct routes that were filed while flights were enroute. The impact was that flights were now going direct through sectors where they had not done so in the past. The major change in air traffic patterns wasnÕt due to flights filed under the new expanded NRP itself, as fewer than 5% of the flights were being filed under the new expanded NRP. It was due to this spin-off of the expanded NRP. Specifically, once airborne, pilots were requesting and getting clearances for direct routes from controllers, who were clearing them on these direct routes without any approval from the affected sectors. This was apparently interpreted by these controllers as the "Free Flight" concept. (The actual order for the expanded NRP contained no such instructions.) The result for Chicago Center, for example, was that they were inundated with direct routes on the south side of the Center, in sectors that in the past had not been impacted by such traffic levels. There was also a problem with the unpredictability of traffic loads from day to day. As an example, the flights to Newark were now coming through Chicago on various routes, often through different sectors on different days, depending on the winds. As a result: "One day a controller is inundated, the next day he's twiddling his thumbs." As one TMO put it: "The whole concept of traffic management is we don't shotgun any one controller. The guy who is working the initial point of the arrival, like Little Rock going to Blue Ridge arrival, we're now pointing a shotgun at his head and saying: 'Here come the airplanes'. Before, we had a structured flow into those fixes. Each time we melded 2 or 3 lines, a different controller was doing it, so finally we got down to one line. Our big fear now is a possibility in Memphis of a single controller having 3 of those points in a sector, so we're shotgunning him 3 times." In terms of the magnitude of this problem, 70 to 80% of the flights over many Centers were now on direct routes. For some Centers, such as Kansas City and west, this wasnÕt a major problem. For others like Chicago and east, there was a significant impact. Some aircraft had to be held at the Chicago Center, Cleveland Center, and Indianapolis Center boundaries because sequencing the resultant multiple flows became nearly impossible. Observation 8. Another potential challenge was a concern with commuter flights. With aircraft like the Challenger jets, "they want to go to 37,000 feet." How will this traffic be handled if they start to take advantage of the expanded NRP? Observation 9. As indicated above, pilots as well as controllers contributed to this sudden increase in direct flights. Pilots (with the permission of controllers) chose to fly direct instead of flying the route selected as "optimal" during preflight planning. Below are two examples of this: A dispatcher was riding jumpseat on a 757 flight from EWR to LAX which had been filed under the NRP by the dispatch staff. ATC offered the flight direct from EWC to PWE (south of the user preferred trajectory as determined during preflight by the dispatch staff). The Captain looked at the dispatch-provided data, which showed that this direct route would cost additional time and fuel. Nevertheless, because the FMS gave different data than the ground computer, the pilot chose to accept the offer of a direct route. The result was added time (4 minutes) and fuel burn (500 lbs.). (In order to make up the time they were losing 2 hours later, the flight changed its flight level and cruise schedule, which increased its overburn.) The flight also went through a convective sigmet with tops to FL 450 that they would have been north of on the original route. Another example was a flight that flew from DFW direct to Parker. The responsible dispatcher commented: "If a direct route had been better, I would have filed it through the NRP. I had planned it over Albuquerque because of a favorable southerly jetstream. Flying direct to Parker, the flight was flying directly into the jetstream. The plane was 6 minutes late." A contributing factor in examples like the second one above appeared to be a lack of awareness by pilots as to when they were on an NRP flight plan. Interviews with pilots from several airlines indicated that they did not know when they were flying a flight plan filed under the expanded NRP. Another example was even more extreme. As noted above, a by-product of the expanded NRP was a sizable increase in the number of direct flights approved while enroute. One of the pilots interviewed from a major air carrier indicated that he thought "that was what the expanded NRP was all about", that when a controller now offered him a direct flight, ATC and the AOC had jointly determined that a direct flight was best for him in terms of weather and air traffic. His comment was: "I was tremendously impressed that they could achieve such coordination." (The reality was that such offers for direct flights were completely uncoordinatedÑ the controllers werenÕt even checking with the other affected Centers, let alone the airlines, regarding the impact of such direct flights.) Observation 10. A Dispatcher had a choice between filing a flight from LAX to DFW either north or south of White Sands. The northern deviation offered the shorter wind route. While enroute, however, the flight was rerouted to the southwest cornerpost because of traffic congestion. As a result, the flight burned substantially more fuel and was later than it would have been if it had originally filed the southern route to the southwest cornerpost. Observation 11. A number of airline dispatchers and ATC coordinators were completely unaware of the dramatic increase in direct flights that occurred following the implementation of the expanded NRP (even though flights from their own airlines were involved). Observation 12. The system for requesting non-pref routes through ATCSCC under the old NRP (Advisory Circular 90-91) provided an example of how procedures can be established to encourage the distribution of knowledge to relevant participants in the flight planning process. As one airline ATC coordinator stated: "When we started this [the procedure for requesting non-pref routes], even Central Flow didnÕt know where all the choke points were. But as we pressed the system and said Ônow we want to fly over hereÕ, weÕd call the Albuquerque Center and theyÕd say: ÔWell, you canÕt go eastbound over St. John at 4 oÕclock in the afternoon.Õ Well, that was tribal knowledge in the Albuquerque Center. The tribe expanded to include Central Flow; Central Flow expanded the knowledge to the airlines and we [the airlines] began to build better routes. So rather than having to fly a 2000 mile route because it didnÕt work at one point, we began joggling around and making routes that were smarter. ... Originally, weÕd call and theyÕd say no. But then it became: ÔWell, if you would just do this, if youÕd just make this minor adjustment in your flight plan, we could probably do this. It became a much more collaborative effort." Observation 13. In contrast, 6 months later, after the start of the expanded NRP, this same ATC Coordinator noted: "The problem with the expanded NRP is that thereÕs no feedback to the AOCs. NobodyÕs getting smarter. ... When we went to free flight on Jan. 9, we cut off the feedback loop for those flights filed under the expanded NRP. ... How do we get this local knowledge that the TMUs and controllers have out there for the dispatchers and pilots? ... There are problems in the ATC system that I donÕt know about. I need a mechanism to get feedback. ... How do we give the airlines more timely information? Depending on where they're going on which day, how do we get the information to everybody? How do we all get the same picture?" Observation 14. The assumption behind free flight is that "if the airlines create a bottleneck and for 3 days in a row they get delayed, they'll change. They'll change the departure time or file a different route. Under free flight, we're leaving it up to them to find a solution." The assumption, in other words, is that free flight represents a "free market" environment in which businesses will respond to problems in order to remain competitive. Upon implementation of the expanded NRP, however, new problems arose that resulted in consistent inefficiencies for certain flights. The airlines were often very slow to respond, however, taking weeks to recognize and react to such problems. Observation 15. TMO: "We donÕt even get a listing of who flew NRP the day before so we can review it and see what are the trends." Results and Discussion | This data collection session elicited a wide range of results that are of potential value in guiding future directions for research and development. These are outlined in the sections below. Benefits Associated with the Implementation of the Expanded NRP | On the whole, the AOC staff for this airline were very positive about the impact of the expanded NRP: "I think it's paid some big dividends for our airline and the whole aviation system." The basis for this assessment is data such as the following, which indicate the level of usage and the estimated savings from the use of the expanded NRP for one month by this airline. For example, during that month, the airline's ATC Coordinators requested a total of 459 NRP routes through ATCSCC, and 260 of these were approved for filing. An additional 2355 flights were filed under the expanded NRP by the responsible dispatchers. Thus, there were a total of 2615 flights filed under the NRP. (In this airline, ATC Coordinators handle any requests for NRP routes through ATCSCC, while the dispatcher responsible for a flight is in charge of filing it under the expanded NRP if desired.) The 260 flights filed through the Command Center were estimated to save $55,160 in time (estimated at $40/minute saved) and $14,218 in fuel (at $0.58 per gallon), while the remaining 2355 flights saved $339,880 in time and $94,060 in fuel. Areas for Improvement | The ensuing discussion did, however, lead to a number of suggestions for how to either further improve upon the expanded NRP, or to develop new programs that provide the airlines with additional flexibility to meet their business needs. Flexibility | In terms of a desire for greater flexibility, one concern is that the expanded NRP applies to only a subset of the routes flown by this airline: "There are all kinds of other routings that we are mandated to fly and we have very few options. We have really fewer options in those markets than we do in the markets with preferred routes." In particular, this airline flew only about 500 flights per day that met the guidelines for the expanded NRP at that point, out of a total of 2690 flights. (Many of these additional flights are too short to be covered by the official preferred route system. Nevertheless, many of them have routes constrainted by the ATM system.) An example of one of these other highly constrained routes is: "Stewart-Newburg to Atlanta. We are required to come out of Stewart, north over Weard Intersection into Boston Center, back in over Lake Henry and Phillipsburg, New York Center, into Cleveland Center, then back into Indianapolis Center and back into Atlanta Center. Sometimes we get into Washington Center, sometimes we go through six Centers to get there. Nobody wants us in any of those areas. We've been trying to deal with New York Tracon to take the flights across TRACON airspace, which we think is much more efficient. They have their own problems, but we have flown that route for 4 or 5 years, which is 60 or 70 miles further than we want to go and in the wrong direction. So, although that is not a published ATC preferred route, we have no flexibility in what we can do and what we can't do." A second concern is the 200 mile constraint imposed by the expanded NRP, limiting flexibility (for the first 200 miles and the last 200 miles of a flight) in planning economical routes: "We would like to see more work to do away with the 200 mile regulation. That would give us a lot of flexibility." A third concern deals with groundstops and delays. The expanded NRP provides flexibility in the routing, but not the launching of flights: "Where we get into problems is with program delays. Half an hour here, 45 minutes there, because the arrival rate is reduced in a city. That's the kind of thing where we would rather launch and keep the pressure on the system up there and make them land us, rather than have gaps in there and be 45 minutes late because they held us at the gate." A fourth concern is prioritization: "If you're going to divert 3 of out 12 flights to Newark, we want to tell you which 3 we want to go. We don't want you to tell us which 3." Regarding prioritization, however, there was a recognition that there are additional concerns that make it infeasible to respect the airlines' economic concerns: "A lot of times ATC has no choice. You may be at 10,000 feet and I may be at 20,000 feet. Or the flights may be coming in over different fixes." A fifth concern is a desire to take advantage of variable speeds. Like the issue of prioritization, this issue requires enhancements at the airlines and within the ATC system: "Another consideration is the ability to slow airplanes down or speed them up? We are already doing that to some extent. In fact, our new flight planning system will be based on that." "As an indication of its significance, we just got a little blurb saying that our systems analysis staff identified 600 flights the other day that were available to slow down a little bit because they were estimated to arrive 10 minutes early because of the winds. We weren't able to slowany of them down in part because the ATC system doesn't like us slowing them down." "No, they like .80 and they like .80 for everybody." "They do, and they recognize an airplane as a certain speed, or operating at a certain speed. And we try to operate at that speed in the terminal area, in the arrival and departure areas. Enroute, however, I think you should be able to get away with slowing down or speeding up. If you're in line for arrival with everybody else and you slow down, obviously everybody behind you is going to have to do that. With current technology, ATC doesn't like that." A sixth concern is runway assignment. Solutions again would require changes within the airlines as well as the ATM system: "As dispatchers, we don't really get involved in ground operations that much, but that's one of the real big cost items that we have. Taxi times, landing on runways where the plane has to taxi a long way, I think that we need to look at that in more detail than we do." Feedback | The participants also suggested that, in terms of internal airline operations, as well as for overall program evaluation, there was a need for better data collection and evaluation: "We suffer from not really having a method to track what we are actually doing. We know what we plan. We know what savings we expect from what we plan, but we haven't really come up with good ways to determine if we're really achieving that." There is a similar need to provide better immediate feedback to help individual dispatchers modify flight plans on a given day and to learn in order to make future improvements: Moderator: "What do dispatchers know about what resulted from the expanded NRP?" Dispatcher: "If I'm not too busy doing other things, I might glance at my graphical screen to see where everything is, but if I don't do that, I really don't know what's happened to a flight unless the pilot calls. You know you've tried to plan the flight the most economical way. If they haven't called you, you don't really know whether they are on that route or whether they have changed the route but have chosen not to call you." Computer Support and Information Exchange | There are a number of areas where dispatchers, pilots, traffic managers or controllers need better tools to make better use of the NAS. Flight Planning Tools | One area where improvements are needed is in the design of flight planning systems that fully take advantage of the flexibility offered by the expanded NRP: "We are limited by our computer system. We still are limited in filing land-based NAVAIDS. We do directs in between them, but we don't have the capabilities of doing Lat-Longs. I see some very big benefits in doing that. We're hoping to have a system at some point in the near future that will allow us to do that. Right now, however, we're still filing VORS, maybe directs in between them, but that still requires you to do a certain amount of zig-zagging instead of using the optimal profiles." A similar problem arises for flight crews while enroute. When offered a direct route by ATC, they have no accurate source of information onboard in order to compare their current route with a direct routing. Even if they do talk with the dispatcher: "It's a very cumbersome task for the dispatcher to give them direct numbers. It's unfortunate. You actually have to insert the route one piece at a time. We also run into problems because, if it exceeds 900 miles, you're getting an average wind in the middle. You may think that you're getting some good information, but in reality you're not." nformation on Special Use Airspace | One particular area where better information exchange would be helpful had to do with special use airspace: "Jet 110 goes straight across the middle of White Sands, into the military airspace around Edwards. There is a little tiny note on the chart that says: 'Not available most days.' We essentially avoid it. I did some work with Albuquerque Center a while back and they were going to provide us with that information on a daily basis. We never could get it worked out so that we could really utilize that. It's just a small part of the total puzzle, but it cost us money. We file north of it or south of it all the time. Even though we don't have a real big operation in that region as compare with American, it still costs us money." Coordination and Information Exchange | Experience with the expanded NRP has also indicated a need for better coordination between AOCs and the ATM system. One such issue has to do with predictability. To be effective, dispatchers need to know when Centers are going to reject certain routes: "There are times where I will sit for 4 or 5 hours and file NRPs and watch Miami Center or Jacksonville Center reroute them back to the preferred route. They get them out of another Center and they'll say 'no, we want them over here,' and they'll move them over. I'll see half a days worth of this and then say to myself: 'This is making extra radio work for me, with the crews calling me back and forth. I'm just going to put them on the preferred route. That's what the ATC system is going to do with them anyway.' " While some of these problems might be predicted by AOCs by looking at the information aavailable on weather displays and the ASD, others clearly require input from the involved Centers: "It may not be weather in that part of their Center. It may be in a different part that's compacting traffic all the way over, or it could be an occasion where it's just a manpower problem, where they don't have time to process the heavy traffic with the manpower they have, so they want to line everybody up and bring them through." "We really don't have a good handle on the traffic bottlenecks when we do have problems. It all gets back to not having the data on what they're actually doing. We need to know how they are reacting to a situation." "I need to know from ATC where the choke points are, where they're going to move somebody or where's the bad spot is today, and whether a bad spot today is a bad spot tomorrow. I need a forecast from ATC that says: 'Hey, we're compacted here, but you can have this NRP route all the way to there.' We need to know that ahead of time, as much as they can predict that for us." "What we really need is an NRP forecast for the day from them. Our workload increases if we have to change our flight plan six times on an NRP route. It's an aggravation to have to go back and change every one of those. You've got to cancel your old strip, probably add fuel to your flight plan. If we had a little bit better forecast, it would help. A lot of times, thunderstorms in the summer are not very predictable, but there are obviously times when there are lines moving that let you get a pretty good feeling that ATC won't want anybody there today. If ATC could just let us know ahead of time about these cases, we'd work around them. But we need to know how they're going to deal with a problem." "They don't have to pinpoint things for us. If they could just say 'today's NRPs we prefer south of Walnut Ridge or north of Walnut Ridge,' then we could evaluate it from there. Or you know: 'Fly east of this line today, or fly west of it, that's our forecast,' We certainly can't hold them to that, but it would help if they would just share their ideas. The guys in the Center know when they're going to get hurt. It would help if they could just give us a clue as to when they're going to get hurt and what they're going to do with about it, we'd try to work with them on that." "Right now we sometimes think we're doing something good when actually we're doing something bad. For example, we may be bringing a flight in over a fix where they're having trouble fix balancing. Consequently the flight gets put into a holding pattern. We think we're doing an economical route when we're not. We don't have enough information." "I don't know about how much ahead of time we would be able to do that, but even enroute if they are a half hour away from the big bank in Atlanta and they see Macey's going to be inundated, that they're going to have to hold six flights out at Macey, that's good information for us to have. Anytime I can tell the crew what's going to happen to them before they get there benefits us greatly. Then the flight crew is on our side to begin with. They're going to hold longer. They're going to be more comfortable with what we're telling them rather than screaming at us: 'How come we're holding?'" An important concern with such forecasts, however, is that "ATC tends to be very conservative and say: 'Hey, we might have a problem here today so we're just going to say that you guys shouldn't file that today.' If they could provide realistic data when there was a real need to avoid something, we wouldn't do it." An example where this type of information exchange is happening now is with an "economy" route that this airline has worked out with three Centers: "In our market from Atlanta to South Florida, essentially what we do is we fly two routes. We've got the preferred route and an economy route negotiated with Atlanta Center, Jacksonville Center and Miami Center because we know that it's more efficient to go down the East Coast than it is to go down the West Coast. It's automatically approved and they allow us to file it under most conditions. We do have certain conditions where we can't. If there is a missile launch at the Eastern Test Range, they don't allow us to file it. The same thing applies if there are thunderstorms. So there are times when they'll send us a message and say, 'go back to the preferred route,' and we comply with that. They usually do let us know when there's a problem through the Command Center." Shared Displays | In addition to providing information such as NRP forecasts, it was also suggested that there is a need to provide shared displays so that AOCS and ATM staff are looking at the same information when exploring and negotiating routes: "There are big benefits from us looking at the same information that ATC is looking at. We don't always interpret it the same way, but looking at the same product has a lot of benefits." Training | The implementation of the expanded NRP has highlighted another barrier to actually achieving the full benefits from air traffic initiatives: Inadequate training. Specifically, many pilots do not sufficiently understand the implications of filing under the expanded NRP: "I would estimate that only 60-70% of our pilots understand the expanded NRP." The net result is that pilots frequently accept direct routings from ATC once enroute, even though the AOC has determined that, based on the best available data, the filed NRP route should be more efficient: "They'll accept a lot of directs. We'll have filed them on a pressure pattern economy route but they think direct is always better, or higher is always better. We've tried to get that information out to them, but you know, there is a mindset out there among the pilots that we've got to fight against." "A lot of times the flight crew will report to us on a long haul flight. They'll report over some point where they are not supposed to be. We'll call and say: 'Why are you there?' They'll respond: 'Well, ATC offered me direct.' When ATC says they'll offer a flight crew direct, it's like holding up a $100 bill. We've caused diversions before because of that. They've refiled direct while enroute and gone way off the filed NRP route, losing the wind pattern, with the result that they are 3,000 or 4,000 pounds short. Those flights have had to stop short and get fuel." This problem is further aggravated by the inconsistencies in the data available to the flight crew through their FMS and the data available to the dispatcher: Dispatcher: "They are very comfortable with the FMS. We know that our flight planning system is more accurate than their on-board system, unless they have plugged in the winds down line that are on the flight plan. I don't think that happens a whole lot." Moderator: "Do you ever get into discussions where you're saying 'my data says this' and they say 'theirs says that'?" Dispatcher: "Oh yeah. That happens often. We may have an average wind component of 50 knots and they maybe just started between that fixed pair and have 80 or 90 knots in their face. They'll say: 'This is wrong.' You have to explain to them that this is an average of the entire fixed pair, that the wind should decrease as they go further. Sure enough, you don't hear from them again, so it's obviously what happened. They are much more comfortable with what they see in their own box than they are relying on us. I guess that's just human nature." Other Capacity Constraints | A number of the issues discussed above have implications for determining system capacity or the efficient use of system capacity. Although it was not the primary focus of this session, the participants did outline other capacity constraints of concern. First, they emphasized the critical importance of capacity limitations in considering future ATM initiatives: "I think they should continue to look at and develop ways to increase capacity because that's really our limiting factor. Weather as well, maybe, but in essence it's capacity." Relevant issues included gate and runway capacities, as well as sector and cornerpost loadings. One important question raised had to do with when such limitations are due to controller workload limitations rather than physical or procedural constraints. As an example: "It would be nice if ATC had the flexibility where they could move their resources to handle different areas, rather than have four cornerposts, to be able to move some of those resources when they have a heavier demand coming in over Rome, having two controllers with two sectors working that area, as opposed to one. I don't know how that would work, but it would be nice if they could put manpower in that area when it's needed, with less manpower in other areas." Organizational Problems | Another concern raised during this data collection session had to do with inconsistencies in the ways different Enroute Centers deal with air traffic: "There are 20 Centers out there and every Center might as well be in a different country, because they all have different moves and different regulations and they don't necessarily get along with each other. They have their letters of agreement (MOUS) out there that we know nothing about, and it's just a big mess. I think that sometime in the future, they need to think about consolidating some of those Centers and running them more consistently. It's a big problem." "When we deal with them in one Center we'll deal with an airspace and procedures guy. Then, in the next one, it's a TMO and in the next one, it's somebody else. They don't talk to each other. so if you work out something in one Center, that's fine with that Center, but now you've got to go to the next Center and you've got to work with them. We need a means where we tell them what we want to do and then they work it out, as opposed to us trying to work out many, many different agreements." An example of the value of working out problems with several Centers at once was then discussed: "We recently took a group from Cleveland Center and a group from Indianapolis Center and some representatives from their tower in Cincinnati, along with military representatives. We all sat down in Springfield, Ohio and talked about the problems we have in Cincinnati. They came up with solutions that were amazing because they had the authority to make some decisions. We came up with some real good things there. We came up with some new departures out of Cincinnati. We reduced the mileage on a lot of our routes. We decreased some of the runway problems we had by developing new departures. Those things can be done, it's just that you need to somehow get these groups together so that they can talk to each other." Dispatcher Workload
Problems | Finally, another barrier to full use of the expanded
NRP, and a potential problem with other future programs that give
the airlines more flexibility, deals with dispatcher workload Furthermore, since the airline is taking more responsibility when filing a flight under the expanded NRP (since the ATM system doesn't review such flight plans for potential problems), the dispatcher routinely has a greater workload under this program: "It's a high maintenance program. You've got to talk to the crews more frequently. You're taking a little bit more responsibility on yourself when you're filing NRP routes, so you've got a little bit more checking to do. If I'm going to file somebody at 35 on an NRP program, and 35 gets bumpy, now he's got to drop down to 31 at Indy Center and they don't like that, so they'll want to scoot him inbounds or give him a 90¡ turn somewhere. Then it's me that's held accountable, not the pilots. It's high maintenance. I'm paying more attention to those flights. I'm talking on the radio more frequently." The importance of this extra work is further emphasized by the following example: "If I've got a 1,200 pound savings on an NRP route , and I've reduced his fuel and just fueled him with minimal fuel for that NRP route and then all of a sudden he's switched back over to the pref route, then I'm taking a delay at the gate or there is a pilot who wants to stop somewhere." Regulation of Airline
Performance in New ATM Initiatives | A number of the points
made above suggest the desirability of giving the airlines greater
flexibility in making decisions about when to launch flights, how
to route them, etc. In discussing such revisions to the design of
the ATM system, however, a critical question was raised: What rules
and procedures will provide greater flexibility while at the same
time ensuring safety and overall efficient use of the NAS capacity?
The major focus of this discussion was the problem that what is
a good decision for one airline will not necessarily always be a
good decision for the overall system. Two related questions are: While these questions go beyond the focus of this particular data collection session, some pertinent insights were provided: "That's where I have a problem with collaborative decision making and with a flow based on arrival times. When and if this ever comes into effect, it will be the carrier's responsibility to keep a flight at its departure until it believes the flight can leave and actually land at its destination. I think we're going to find it hard to delay departures, because we want to get airplanes in the air. If they get in early and somebody spanks our hand, well, okay." "Let's say we start doing that. We wait for the feedback that we're having problems with our strategy for launching flights. The issue is, are we also cutting our safety margins? You could say: 'It's just a divert.' But is that changing something about the decision to divert? Is it saying: 'I'm going to risk hanging on a little bit longer because I really want to get into that airport?' We're changing the pressures on the people making the decisions. Where is that strictly an economic decision, and when does it start to be gradually more and more of a safety issue? Because it's not a clear cut line between the two." As an example it was suggested: "Suppose ATC's monitoring the overall safety of the system, and let's say the FAA has a criterion that says that we don't want people to be making diversion decisions with low amounts of fuel all of the time since that's not a good policy from a safety point of view. They would have to track how often we are doing this, and if they see that a carrier is pushing the edge of the throughput margins, they'd have to say: 'Wait a minute. We want you to back off on that a little bit.' How would that negotiation work? Do you see them coming back and saying to you: 'Limit your throughput.' Since they have a responsiblility for overall system safety, how are they going to decide?" It was also pointed out, however, that the airlines already demonstrate some ability to self-regulate their decisions effectively: "Don't get the impression that we don't hold flights for poor weather. When we're going to Dayton or Columbus or South Bend or Grand Rapids, frequently they'll be below minimums with no ground stop on. We still choose to stay on the ground. So we already do a considerable amount of self-regulation ourselves." Alternative Control Parameters | The discussion above suggests one (possibly controversial) means of control by the FAA to help ensure safety: Regulating average landing fuel levels and monitoring actual landing fuels to ensure that airlines are not routinely pushing too close to the edge in terms of safety. Another type of control parameter was also discussed in the context of cornerpost loadings when there is a reduction in capacity: "The FAA could make percentage calls. If Delta is 72% of the operation in Atlanta, if they would like to move ten flights from Macey to Rome over the next three hours, Delta could move seven and ValuJet could move one flight, and so on. With this strategy, you would start to assigning slots to each of those cornerposts, with each airline having a percentages of those slots if they chose to take them." Potential concerns with this type of control parameter were also discussed, however, based on past experience with such a strategy: "They've tried to fix loading and we really didn't like it too much. That was the fix loading program that we had for Atlanta. We had delays every day. It did not work because it wasn't managed well. They used it as a way to manage the capacity and it wasn't valid, it just didn't work. They would take our flights every day and delay them. Say we had a push of 60 airplanes coming into Atlanta. They'd delay twenty of them because of demand over the fix. They did that every day. In contrast to that, now we've got a scenario where we do some airborne holding. We encouraged everyone to use holding to a certain extent, because we want to have airplanes available when the capacity permits them to land." In short, the experience was that the estimates used by the ATM system to initiate ground delays were too conservative: "If, instead of holding those twenty flights on the ground, we had launched all twenty, we would have actually had only five that would have had some airborne holding. Why should we delay twenty flights when in fact there would only be five of them late. We can't do anything with them if they are sitting on the ground in Raleigh or Greensboro or whatever. If we've got them up there holding and there is a slot available, if somebody else is late or whatever, then we keep capacity up. Maybe the ATC specialists suffer from a lack of real time information over there. If they had real time information and knew exactly when everybody was going to come over those fixes, maybe they could manage better." A suggestion to help deal with this conservatism and this lack of access to real time information was to make decision-making more collaborative: "We could be involved in the planning with them and say: 'Hey, don't cut at this number, let's use this other number because we don't care if some of the flights have to hold. We'll handle that ourselves. How about if I come up there with an hour and a half of hold fuel and if you get me in, you get me in?'" "You've got some traffic manager who's looking at all the data he has available and he says: 'Delta, you're taking up 50% of our traffic for the next two hours and we predict there is going to be reduced capacity. You can either send all your planes as long as we agree ahead of about time what they're going to do if they can't land there, making sure there is a safe place for them to go. That way, you have a choice, but in the worst case, you pay the penalty and can safely divert. But my advice, based on the information I've got, suggests you ought to cut back.' Now it's my choice. They're saying that, if they're right as a traffic cop, when the planes start arriving, I'm not going to get 100% of mine in there. American's going to get their percentage in and I'm going to get my percentage in. As another example, they might come up and say: 'We recommend you delay Flight #345 for 2 hours and 40 minutes.' I could then say: 'How about if I delay that one an hour and a half and stick another hour and a half on Flight #262 over here.' This way they're giving me options to work with." "Separate the roles here. One is information: 'There is a potential problem here.' Let us judge whether we concur with that. The other is: 'Here's a recommendation.' Let me decide how I want to deal with it. They become an advisor, like some sort of counselor." Summary and Conclusions | This knowledge acquisition session provided a rich set of data for identifying issues related to the design of the current and future ATM systems. The issues discussed included procedural changes as well as areas for technology development, and dealt with concerns over organizational behavior as well as individual performance. For the convenience of the reader, we summarize the issues here in the order in which they are discussed in the report, rather than attempting to prioritize them. Scenario Development | It is clear from this knowledge acquisition session, as well as others we have conducted, that specific scenarios, properly used, help elicit very useful information from subject-matter experts. One of the challenges in developing future systems will be to idenfity all of the relevant scenarios and to ensure that the full range of relevant expertise is accessed. Benefits Associated with the Implementation of the Expanded NRP | The data presented indicate that, for a single carrier during a single month, the total estimated dollar savings from the NRP amounted to over $500,000. This fact alone, if substantiated by data from other airspace users, indicates the potential gains to be realized from providing carriers with greater flexibility (see below). Areas for Improvement | Flexibility | The quotations presented earlier exemplify some of the constraints imposed on flights in the present system. While some may be the result of ingrained conservatism in the ATM system, some are clearly the result of mismatches between the present strategically-based system and the loosening of strategic constraints under the expanded NRP. What we are seeing with the expanded NRP is one of the first attempts to implement a hybrid system, in which management may be either by directive, by permission or by exception. It is clear that, although this is viewed as a step in the right direction, users clearly feel they need more flexibility, expecially with respect to timing, diversion options, speeds, and runway assignments and routings. Dispatchers recognized the problems ATC would have in granting greater flexibility, but they also point out the costs of operations under present constraints. Feedback | The issues of information management and information sharing have been pervasive in the data we have collected in support. This is not only a problem in ATM/airline communication; it is as serious a problem within the airlines themselves, and within the ATM system. As authority begins to be diffused throughout the system, more information is required in more places, to be used by more people for decision-making. This is seen in many of the following sections of the report. Computer Support and Information Exchange | Better tools are needed if human operators in the system are to make better use of its resources. The examples presented above identify decision support tools that are needed to accomplish specific dispatching tasks, with a particular emphasis on the need to support information dissemination and cooperative problem-solving. Given the large amount of information presently resident in the system, and the limited capacity of human operators to deal with it at present, we believe that the area of information dissemination and representation is perhaps the most critical and potentially productive area for intensive exploration and development of new aiding technologies for a future ATM system. Such information exchange is critical to both individual and group problem-solving. Training | The observations from this group refer to the training and education of pilots, but this is suggestive of broader issues as well. It is clear, for example, that the Air Traffic Management system has an incomplete understanding of the needs and priorities of air carriers. The comments in this session also suggest that some ATC facilities may also lack knowledge of how other facilities have coped more effectively with the similar problems. (See Results: Capacity Constraints; Organizational Problems). Other Capacity Constraints | The issue of extending the scope of flexibility to surface movements is one that may require different strategies as well as different technologies for its solution. Organizational Issues | It is regrettable, in some respects, that the original NRP, which required a good deal of discussion and interaction among all system participants, has given way to the expanded NRP in which these participants are expected simply to make decisions without such consultation and discussion. These data suggest that ways must be found to enhance and reward a more communicative, consultative, and therefore cooperative style of operations in the NAS. Dispatcher Workload Problems | The issue of simply transferring, rather than mitigating, operator workload is discussed under this heading. Care must be taken to alleviate workload under high-intensity conditions, not to transfer it from one set of system participants to another. One type of solution is for long-distance conflict prediction software tools, when they become available, to be made accessible to both ATM and AOCs. Regulation of Airline Performance in New ATM Initiatives | The overarching issue of mediation among participants in a more flexible system must be approached by system designers. Given the competing objectives of users, does there need to be a neutral "referee"? There is no question that the ultimate locus of control and the authority to assure safe separation of aircraft in a future system will reside with ATM staff. Just as aircraft automation can foreclose or render less effective pilot authority, however, the rules, procedures and software for a more flexible ATM system, unless carefully designed, may foreclose or make less effective the authority of air traffic managers and controllers under some circumstances. This is a system integration issue that requires further research, but the dimensions of the issue are evident even in the research done to date. Cooperative decision-making was suggested as one way to reduce problems with real-time decision-making in this area; post-operations analyses of system performance was identified as a second, complementary approach. It is clear, however, that the functional requirements for a future system must include at least an outline of the policies and procedures under which the system will function, and that these functional requirements must be in place as a guide for the design of the equipment which will assist human operators in the operation and management of the system. Though the philosophy of human- or user-centered automation has been accepted by at least some system proponents, such a general philosophy must be particularized in terms of the ATM system before the design of its architecture can be made final. Recommendations | Based on these results and our other research to date, we make the following recommendations for further work. • Expand research on the overall functional requirements for a future advanced ATM system embodying management by permission and management by exception, to serve as guides for the development of aiding technologies to be implemented in such a system. • Accelerate and expand research on information management for a future ATM system in which responsibilities and authority are more widely distributed among system participants. This research should include: - Information requirements under a management
by permission paradigm • Search for ways to improve collaborative decision-making among system participants in addition to studying technologies that will more effectively share relevant information among those participants. • Continue and if possible expand research into ways of providing more flexibility for airline operations. • Explore the issues raised by the probable need for a neutral "referee" or arbitrator in a more distributed future ATM system, and examine proprietary and competitive issues embedded in information requirements, since these are the issues most likely to inhibit sharing of information in a more distributed system. Acknowledgements | This work was funded by the Advanced Air Transportation Technologies program at NASA Ames Research Center under Grant Number NAG2-995. We would like to express our appreciation to Judith Orasanu and to the AOC staff who assisted us in completing the work.
|
||||||||||||
 |
||
 |
 |
 |
 |
||
 |
||
 |
||
 |
||
 |
||
 |
||
 |
||
 |
||
 |
||
 |
||
 |
||
 |
||
 |
||