*Cognitive Systems Engineering Laboratory

Ohio State University

**Department of Aviation

Ohio University

February 16, 2001

This report was produced under funding from the

FAA Office of the Chief Scientist for Human Factors

Coded Departure Routes (CDRs) are a set of predefined alternative routes for flying between particular city pairs. They were developed by ATCSCC and ARTSCC staff in cooperation with the airlines under the auspices of the FAA’s Collaborative Decision Making Program. These prespecified routes were developed for two reasons. First, there is an 8 letter abbreviation associated with each CDR, making computer entry of that route much faster for FAA and airline staff (thus reducing workload and speeding route changes). Second, these CDRs were designed to support a collaborative process for selecting an alternative departure route for a flight when the user preferred route is not available due to weather or traffic constraints, thus reducing the departure delay that would have been incurred if the flight had to wait for departure until the weather or traffic constraint was gone.

More specifically, the initiative that led to the development of CDRs had several underlying motivations. The first was to increase efficiency in communicating changes in the departure route for a flight, speeding up such communication and reducing the associated workload. The second was to develop a collaborative process that was intended to:

1. Provide AOCs, ARTCCs, TRACONs, Towers and the Command Center with a process for working collaboratively to develop earlier plans for dealing with predicted constraints in the NAS.
2. Provide a set of prespecified alternate departure routes for specific city pairs that had been approved by all of the involved Centers in terms of the impact on typical traffic flows and constraints.
3. Give the Tower at an airport greater flexibility in responding to the often rapidly changing picture regarding available airspace during weather events, so that departure delays could be reduced.
4. Keep dispatchers in the loop through the early identification of the alternate departure routes that might be selected at the time of departure from an airport.
5. Keep all of the relevant traffic management facilities in the loop to ensure that the selected alternative departure routes were acceptable to all of them in terms of current traffic and weather constraints.

Below, we provide examples of how CDRs were actually used during Summer 2000, and identify areas where improvements are needed. This is a preliminary analysis and should not be viewed as comprehensive. It is based on interviews with staff at 5 airlines and 4 Centers, and on analyses of ATCSCC advisories and POET data for flights associated with advisories where the use of CDRs was mentioned. A more complete analysis is planned in the future. However, in order to provide timely input, we are summarizing our findings to date.

In cooperation with ZOB and ZMP, NWA has developed a highly interactive procedure for implementing the use of CDRs out of DTW and MSP. Under this process, the following steps are supposed to be followed:

1. As part of the Command Center’s Strategic Planning Team teleconferences, NWA and the relevant Center (ZOB or ZMP) both participate in discussions 2-4 hours before a weather event, identifying the potential need for the use of CDRs out of DTW or MSP in order to reduce departure delays.
2. The Command Center sends out an advisory containing the Strategic Plan of Operation (SPO), which indicates that CDRs may be used for departures out of DTW or MSP.
3. Following the release of this advisory, an ATC Coordinator at NWA and a traffic manager at ZOB or ZMP communicate by phone, discussing how the weather is developing and identifying the specific CDRs that may be viable. These phone calls may be initiated by either the airline or by the Center.
4. ZOB or ZMP also communicates with the Tower at DTW or MSP to get their input and, through the Command Center, with the other Centers that may be impacted by the CDRs under consideration. Based on these communications, a small number of CDRs (2-3) are identified for possible use to each of the destinations from DTW or MSP.
5. Prior to development of the flight plan for a specific flight, the ATC Coordinator at NWA tells the responsible dispatcher the 2-3 CDRs that are under consideration for that flight.
6. The dispatcher develops a flight plan for the route that he/she considers best given available data 90 minutes before the filed departure time. This is the route that is filed with the FAA. (Note that sometimes this route may in fact be one of the CDRs under consideration if the dispatcher feels that it is likely that the "normal" route used for that flight will be impacted by weather.) The dispatcher also, however, evaluates flight plans for the CDRs under consideration, and determines whether they are potentially acceptable as well (depending on how the weather develops).
7. On the release for the flight crew, the dispatcher provides the usual information on the route that he/she considers best and that has been filed with the FAA. In the remarks section, the dispatcher also indicates which (if any) CDRs the flight is approved to accept as alternative departure routes. As an example, on 9/13/00 NWA138 from MSP to ORD was filed by the dispatcher on MSP..RST..JVL.JVL4.ORD. The dispatcher also ran flight plans for two possible CDRs that had been identified by this collaborative process, and requested fueling for the aircraft so that it could fly the filed route or either of these two CDRs. The flight release indicated to the crew that MSP..RST..JVL.JVL4.ORD was the preferred route, but that it was ok to accept either E1 or S1 (two CDRs) if the Tower offered one of them.
8. The aircraft is fueled so that it has sufficient fuel for these alternatives as well as the filed route.
9. Close to departure time (often just prior to pushback or while the aircraft is taxiing out for departure), a decision is made based on the weather or traffic developments as to whether to use the filed route or one of the approved alternative CDRs. In many cases, this decision is made by ATC in the Tower, as they often have the best picture of how things are developing at the airport in terms of both weather and traffic constraints. In some cases, however, it could be the airline ATC coordinator or a traffic manager at the Center that requests the use of a particular departure route based on the data that they have available.
10. The Tower then asks the Center to change the filed route for that aircraft as needed. Within a minute, the Center traffic manager changes the filed route to the selected CDR. (ZOB traffic manager: "When the Tower calls and says we need to move one or more flights, it’s less than a minute before I enter it. I also tell the Command Center what we’re planning.")
11. The Tower then tells the flight crew that they are cleared to depart via that CDR.
12. The flight crew checks their release to see if they are approved on that CDR, and if so, they accept the clearance and depart.
13. If the clearance is for one of the CDRs and differs from the originally filed route, the flight crew looks up the CDR on the Jeppeson charts and sends an ACARS message to dispatch requesting the full flight plan information for that route.

Example 1a. DTW-LAX on 9/12/00. As a first example for NWA, on 9/12/00, the dispatcher chose to file a flight from DTW-LAX on a CDR, with appropriate fueling because the dispatcher felt that the "normal" route was unacceptable in terms of the likely weather. As shown in the figures below, this filed route was used for departure, but with a 54 minute departure delay. The filed route is in green; amendments are in orange; the actual route is in black. Figures 1-4 show the data available on this flight. Figure 5 shows the actual weather.

Figure 1. Data on an NWA flight from DTW to LAX on 9/12/00.

Figure 2. Additional data on this NWA flight from DTW to LAX on 9/12/00.

Figure 3. Filed route and amendments for this NWA flight from DTW to LAX on 9/12/00.

Figure 4. Filed route (green), amendments (orange), actual route (black) and forecast weather (CCFP) for this NWA flight from DTW-LAX on 9/12/00.

Figure 5. Radar weather display for 9/12/00 at 0100 Z.

Example 1b. DTW-CLT on 7/28/00. A second example for NWA shows a case where there was uncertainty about whether the weather would move east fast enough to impact the normal route for a flight from DTW to CLT departing via CAVVS. The dispatcher therefore chose to file the flight to depart at 1500 Z along the "normal" route, but included on the release approval to depart on a CDR via WINGS. Prior to departure, the weather did impact the filed route, blocking the filed departure via CAVVS, so the flight used the CDR to depart.

The SPO sent out at 1200 Z indicated that, starting at 1400 Z for DTW "EXPANDED MIT AND GROUND STOPS EXPECTED, LAADR AND CDR WILL BE UTILIZED IF ZOB IS ABLE. EXTENDED GROUND STOPS MAY RESULT IN A GROUND DELAY PROGRAM." Additional details on this SPO are shown below.

SPO_ISSUED: 07281200

AM CONVECTIVE ACTIVITY IN MI, OH, MOVING ENE DISSIPATING.

AM CONVECTIVE ACTIVITY IN MO, IL AND SOUTHERN IN MOVING ESE GRADUALLY DISSIPATING.

ANTICIPATED_CONSTRAINTS_TERMINAL: ORD/DTW/CLE/PIT/PHL/CVG/STL/MEM CHANCE THUNDERSTORMS 1800Z AND LATER.

SHORT_PLAN_START_TIME: 07281400

…

3. DTW-EXPANDED MIT AND GROUND STOPS EXPECTED, LAADR AND CDR WILL BE UTILIZED IF ZOB IS ABLE. EXTENDED GROUND STOPS MAY RESULT IN A GROUND DELAY PROGRAM.

Figure 6. Performance data, filed route (green), amendments (orange), actual route (black) and forecast weather (CCFP) for a NWA flight from DTW-CLT on 7/28/00. The orange route is the CDR used at departure.

Figure 7. Radar weather at 1500 Z on 7/28/00.

Delta and ZTL use a very similar collaborative procedure to plan and coordinate the use of CDRs out of ATL. This procedure clearly recognizes the human factors issues central to collaborative decision making regarding the need to keep all of the participants (traffic managers at the relevant ARTCCs, TRACON and ATCSCC; dispatchers at the AOC; controllers at the airport Tower and flight crews) on the same page regarding the plan that is being implemented, After looking over the procedures used by NWA and ZOB for using CDRs out of DTW, a traffic manager from ZTL made the following observations:

"Prior coordination is a must, as this is the only way that dispatchers can be kept in the loop for advance planning."

"While our coordination is not quite as interactive as the process for NWA at DTW,

the STMC or TMC coordinating the use of CDRs will typically call the ATC Desk at Delta, along with ATCSCC and the first tier facilities, letting them know what we are planning. This coordination is usually composed of the directions out of ATL that could be blocked by weather, the time frame involved, and which way we expect to go with the CDRs."

"Atlanta is different from DTW in that we have only 2 CDRs for any given city, one 90 degrees left and one 90 degrees right of the "normal" direction for departure for that destination. By keeping it simple, there is less for the dispatcher to have to consider for any given flight. We may add a third one that is 180 degrees from the "normal" direction in the future, but that remains to be seen. By keeping it simple we reduce the amount of communication required with the airline’s ATC Desk."

"As far as what happens when the flight actually pushes for departure, the decision on whether or not to put the flight on a CDR is made close to departure, and is based on the real-time weather status at the airport and en route. One of two things happens:

1. The TMC at the Tower at ATL calls the TMC at ZTL and gives information on the flight. The TMC at ZTL then amends the flight plan for that flight and the Tower issues it a clearance on that CDR.
2. The TMCs at the Tower and Center have already done their coordination and there was agreement on the wholesale movement of flights departing out a specific direction. In this case, the TMC at the Tower will enter the amendments for the flight plans, clear the flights and keep the log of CDRs used.

In either case, ATL and ZTL can move 1 or 51 flights with the same ease and move only those flights that are necessary, thus saving our customers unneeded reroutes and delays."

USAirways takes a much less interactive approach to the use of CDRs. They participate in the SPT telecons to provide input in deciding whether CDRs should be considered at particular airports. When a SPO (Strategic Plan of Operation) is then issued indicating that CDRs may be used at a particular airport, that advisory is reviewed by the appropriate dispatcher(s). The dispatcher then uses the airline’s flight planning software to determine the longest wind route among the alternative CDRs in order to ascertain the maximum fueling requirements. If the fueling requirements for the longest wind route can be met, then the dispatcher files the route that he/she thinks is best (which in some cases could be one of the CDRs), and then indicates in remarks on the flight release the specific CDRs that are also approved for acceptance by the flight crew. (If fueling or some other constraint makes one or more of the CDRs unacceptable, the dispatcher does not include those CDRs on the list of approved routes in the remarks section of the flight release.)

If the crew receives a clearance from the Tower for a particular CDR, the Captain looks in the remarks section of the release to make sure the dispatcher has validated the Tower-given CDR for fuel and MEL considerations. The bottom of the release then provides the detailed routing information for each of the pre-approved CDRs, so that the flight crew can look at the necessary details for the specific routing that they have been given. After accepting the clearance, the crew is then supposed to inform dispatch about the route they have accepted, so that the dispatcher stays in the loop.

Reactions to the use of CDRs over the past year have been very mixed. Some airlines and Centers (such as Delta with ZTL and NWA with ZOB and ZMP) have indicated that their overall experience was very good, while others either didn’t use them at all or were less than pleased with the results because of significant problems. In general, the most successful cases were those where AOC, ATCSCC, ARTCC and airport Tower staff had worked with each other to develop specific procedures for communication and coordination at specific airports on days when CDRs might be useful. Even in the most successful cases, however, the dispatchers and traffic managers indicated there were areas where significant improvements could be made. These areas are discussed below.

The Use of CDRs to Reduce Communications Workloads. One of the potential benefits from the use of CDRs is the use of the associated abbreviations to reduce the workload associated with entering a route into the computer or with communicating a route clearance to another person. The FAA traffic managers were unanimously enthusiastic about the value of this "shorthand" as a way to enter routes into their computer systems, indicating that it significantly reduced the time required to enter a route selection. Some, but not all, airline staff were similarly positive about this benefit.

The concerns fell into four categories:

The Use of CDRs as Part of a Collaborative Procedure. The section above focuses on the issues associated with the use of CDRs as an abbreviated code to speed up route entries on the computer and to make communications more efficient. It is important to understand, however, that the proposal to use CDRs also included a procedure to support collaboration between FAA traffic managers and airline AOCs.

The different implementations of this collaborative procedure by NWA, DAL and USA were described earlier in this report. The important features shared by all of these implementations, however, are:

1. Through the use of teleconferences held several times each day by the Strategic Planning Team at ATCSCC (which include participation by ARTCC and AOC staff), advance plans are developed that identify airports where alternative departure routes may be needed because of developing weather or traffic constraints. At the conclusion of each such teleconference, a Strategic Plan of Operation (SPO) is sent out as an advisory to all of the FAA facilities and airline AOCs. This SPO is generally sent out 2 hours ahead of time. If this plan includes the possible use of CDRs, the SPO will say something like "CDRs may be used for flights departing south out of DTW."
2. In many cases, this is the only notification to the airlines that CDRs may be used for that airport.
3. The procedure for responding to this SPO differs among the airlines. As described below, for flights out of DTW, once a SPO provides an alert that CDRs could be used at DTW, the ATC Coordinator at NWA works out details via the phone with a traffic manager at ZOB, and then provides the relevant dispatchers with the information they need to plan individual flights. This traffic manager at ZOB further coordinates with the Tower at DTW and with ATCSCC and the surrounding Centers. At the other extreme, USA leaves the responsibility for responding to the SPO to the relevant dispatchers, and does not routinely communicate any further with FAA traffic managers about the details of the plan.
4. Under the approaches used at NWA and USA, the individual dispatchers run flight plans for the airline-preferred route as well as for all of the CDRs associated with the departure airport that were identified in the SPO (the procedure at USA) or for a more limited set of CDRs identified through additional communication with the relevant ARTCC (the procedure at NWA for DTW and MSP). Note that many SPOs simply say things like "CDRs may be used for departures out of EWR", leaving a large number of CDRs to be considered during flight planning. Based on these flight plan calculations, the dispatcher determines which routes are viable, requests fueling to handle them all, and informs the pilots (via the flight release), that they can accept a clearance to fly any of the listed CDRs as well as the filed route.
5. The Tower, in collaboration with the associated ARTCC, then makes a determination for each flight close to take-off as to whether it can be cleared on the filed route or whether it should be cleared on one of the CDRs. If the filed route isn’t to be used, either ARTCC or Tower staff change the flight plan to the selected CDR in the FAA computer, and the Tower reads that clearance to the flight crew.
6. The flight crew then either accepts that clearance on the CDR (if it is in the list of approved routes on their release) and departs on that route, or declines to accept it and must wait for a different clearance that they can accept.

Note that this procedure has a number of interesting features. First, it is collaborative in the sense that, if the airlines and ARTCCs participate in determining the contents of the SPO, then they have a chance to influence the strategic decision to use CDRs as a tool to reduce departure delays. Second, although the locus of control for the final (tactical) decision regarding which route to use for departure is in the hands of the Tower controller (in communication with the ARTCC), the dispatcher has a chance to influence this tactical decision by providing the flight crew with an advance list of acceptable alternative routes in terms of fueling and Minimum Equipment Lists (MELs), and the flight crew has the opportunity to accept or decline the route offered by the Tower. Note also that this procedure is predicated on the assumption that, in most cases, Tower and ARTCC staff have the best real time information about the availability of alternative departure routes and are therefore in the best position to make the final decision (subject to constraints placed on that decision by the dispatcher and the flight crew), thus expediting departures.

Overall, dispatch staff at NWA and DAL spoke very positively about the use of CDRs to allow better real-time decisions about what routes to use to expedite departures from an airport. NWA indicated that there were cases where they were able to depart with almost no delay because of the use of CDRs. Traffic management staff at ZOB, ZTL and ZMP similarly indicated that this collaborative process was a significant improvement over past procedures for dealing with weather and traffic constraints, and that they expect performance to be even better next year as they gain more experience in the use of this process.

Staff at the other airlines were less positive about the effectiveness of this collaborative procedure. Their reactions ranged from "we don’t participate in the use of CDRs" to confusion about how and whether CDRs were being used at their major airports. Everyone agreed, though, that whether their experience with CDRs in 2000 was positive or negative, there is room for significant improvements in the collaborative process. These concerns are outlined below.

Training and Staffing. One problem associated with the use of CDRs is the need for improved training of FAA staff. In some cases, for instance, it appears that some Tower staff do not adequately understand the established procedures for using CDRs, resulting in significant departure delays. In other cases, there appear to be differing understandings by the ATCSCC severe weather specialists who put out the advisories. For example, toward the end of the summer, one specialist was still under the impression that when the possible use of CDRs was outlined in a SPO, a follow-up advisory then had to be distributed that identified the one specific CDR that the airlines should file, thus eliminating the intended flexibility. These different understandings are so pervasive, and are so clearly illustrated in the different SPOs and reroute advisories that are sent out, that it is difficult to claim that there is an established procedure for how to make use of CDRs.

Similar problems exist in the training of dispatchers and pilots regarding this collaborative process. At some airlines, for example, it appears that very few of the dispatchers are even aware of the intended collaborative process.

In addition to problems with training, there are also problems with adequate staffing. Clearly, with current levels of technological support, the use of CDRs as part of the collaborative procedure(s) described above (as distinct from their use to speed communications, as described earlier in this report) increases workload, both at the airlines and at FAA facilities, and it is important to provide necessary staffing to make it possible. In particular, Center TMU staff indicated that it is critical to have someone at the ATC Tower who is the focal point for this coordination, and that this was not always the case this past summer. Without such staffing, it is very difficult to achieve the necessary level of coordination in a timely fashion. (Traffic Manager at ZOB: "We need the TMU at DTW to concentrate more on coordination during bad weather. We need someone who is the focal point. Cleveland Hopkins is even more of a problem. They don’t even have a traffic management unit with a focal point for weather coordination with the airlines.")

Less Restrictive Departure Routes. There is agreement among both dispatchers and traffic managers that the current CDRs, which are defined as complete routes from origin to destination, are unnecessarily restrictive. As one dispatcher expressed it:

"CDR routes should be nothing more than severe weather SIDS that get our flights safely around significant weather in and around the terminal area in an orderly flow.  Everyone would benefit if that were the true practice.  Instead of being severe weather SIDS, CDRs determine what your flight will do for the remainder of its life span, sometimes even routing into severe weather further down the route.  That is wrong.  Tell me, as a dispatcher, what works best for ATC to get out and around the severe weather but not what I need to do for the remainder [of the flight plan for that flight]."

One traffic manager indicated agreement with this general approach, but also suggested that the major constraint to accomplishing this is a limitation with the FAA’s host computer:

"I totally agree with the idea that CDRs should serve to get the flight off in a timely fashion and around the weather affecting the normal departure route. Right now it’s a limitation in the Host coding. We need to deal with that, as we also can’t include Internationals [International flights] in this program until that’s done."

One additional consideration, however, is that if the departure routing leads to an overall route that is significantly different from those alternatives that are normally filed for that flight, it may encounter traffic congestion that the dispatcher is unaware of. This suggests that, in the not too long run, there is a need for better information displays for dispatchers to check on potential traffic constraints. It also, however, raises a question about whether, even if the dispatcher is given more flexibility in designing the route beyond some initial "CDR departure fix", there will still be a need for coordination with traffic managers in order to ensure that a flight is not being filed into a significant traffic bottleneck.

Slow Response Times. One Center indicated that there is sometimes a problem with slow response times from the Command Center, making it hard to respond in a timely fashion to changing weather conditions. Similarly, one airline indicated that there are times when they would like to request the use of a CDR because of quickly changing conditions. Their experience has been that the time to get a response from the traffic management system in such cases is so slow that the opportunity to use that CDR passes before approval is given. (These slow response times are in part due to the reliance on the telephone to communicate and coordinate regarding the use of CDRs.)

Lack of Coordination Among Centers. Both airline and Center staff cited a problem with reroutes or airborne holding because a Center can’t accept a flight that has been put on a CDR. As one traffic manager put it:

"If the CDR gets back on the pref before it leaves his airspace, no problem. But that doesn’t necessarily happen. The problem then is the loading for the accepting Center. An example is problems with Washington Center accepting CDRs from Atlanta."

Three issues are implicit in this discussion. First, in creating the CDR database, CDRs were supposed to be checked by all affected Centers in order to ensure that the routings do not run into predictable traffic bottlenecks. This has not been done for all of the CDRs. Second, when CDRs were first proposed, it was recommended that they always be routings that tied back into the "normal" routes before the flight left the departure Center. Many Centers have concluded that, if they are to develop a set of CDRs to provide sufficient flexibility to deal with possible weather scenarios, then some of those CDRs can’t tie back into the "normal" routes within the departure Center, because the flights would have to cross other traffic flows in order to do so. To simply treat these as SWAP routes would be to lose the flexibility offered by the CDR process. A third issue is the need for real-time coordination across Centers when there are plans to use CDRs in order to ensure that, even if the CDRs in the database were originally checked by all the Centers, there isn’t some specific bottleneck that has arisen on a given day.

Inflexibility in the Design of Sectors. At present, CDRs are designed for and used in an environment where, with the exception of combining adjacent sectors, sector boundaries are fixed. These fixed sector definitions limit the opportunity to increase throughput because of workload limitations imposed by controller staffing of any given sector. It might be possible to refine the use of CDRs and further increase throughput by the use of dynamic resectorization along those portions of a CDR where capacity is being strained. If, for instance, a sector is now saturated due to an increase in the number of flights filed on a CDR through that sector, it might be possible to further increase throughput by dynamically adjusting the boundaries of the overloaded sector, thus distributing the workload among a larger number of controllers.

Lack of Coordination between Dispatchers and Flight Crews. Dispatchers at a number of the airlines contacted as part of this review indicated that there is a serious weakness in this process in terms of keeping the dispatcher in the loop. They indicated that, with some frequency, flight crews fail to inform their dispatchers when they are given a CDR as a reroute. As a result, the dispatcher may be unaware of the amended route of flight (which may be significantly different from the originally filed route). The frequency of occurrence of this problem appears to vary greatly from airline to airline. For the worst case, however, a quick survey at one airline indicated:

"Dispatchers on the floor today estimated that only 20% of the crews call back and tell them the actual CDR route that ATC has given them."

Given some CDR routes could have impacts such as moving aircraft that are not over-water equipped onto CDRs that involve over-water routings, this breakdown in the normal safety net is of concern.

SPOs and Reroute Advisories that are too Long and Difficult to Process. Both FAA traffic managers and airline dispatchers have indicated that the length and lack of adequate structure of the SPOs and reroute advisories make it very difficult and time consuming to determine whether there is anything relevant to a particular flight. The lack of structure also makes it difficult and time consuming to interpret the advisory even if it is located.

As one ATCSCC specialist put it:

"We have the same problem reading the SPO in Severe Weather that you have at your airline."

The interviews with traffic managers and dispatchers further indicated that there is no single best way of structuring the advisories. Different people need to search the list of advisories in different ways. This suggests that the structuring of the advisories should make it possible to sort the advisories according to different criteria.

Lack of Adequate Support Tools. As the collaborative process for using CDRs was described above, it is clear that this is still a very manual process for many of the steps, relying on voice communications for coordination. There are numerous ways in which this process could be supported with software and communications tools. However, to be fully effective, specifications need to be developed to ensure easy integration among these tools, as well as integration with existing airline and FAA tools (or future modified versions of these tools).

These new tools should make it easy for FAA and airline staff to identify and view alternative CDRs in both text format and on map displays that show potential weather and traffic constraints. It should also be easy to select specific CDRs for inclusion in a plan via some point and click interface. If CDRs are modified in the future to be defined in terms of departure SIDs instead of as complete routes (as discussed above), then it will be important to design the systems to make it easy to use such a SID as the initial input to airline flight planning systems that would then (with dispatcher input) identify alternatives for the remainder of the route. (Such tools will need to be integrated with existing software such as CRCT, the enhanced Traffic Situation Display and the Common Constraint Situation Display.)

Finally, since the selection and use of CDRs is meant to be a collaborative process, there is also a need to support more effective synchronous and asynchronous communications among FAA and airline staff in support of discussions about what CDRs to consider in a particular situation. This calls for the design of shared visual displays and other rich communications media and the support of computer-to-computer interactions to allow automatic computation of alternative flight plans. It also means the bandwidth to enable such interactions must be available. It should be remembered that some of these interactions may involve interactions with the flight deck (ground-to-air communications), as well as interactions among ground-based staff (AOC staff, traffic managers and controllers).

To summarize this concern, one dispatcher noted:

"There is no automated interface between the ATC Host computer, ATCSCC ETMS flow management systems, dispatch flight planning and tracking software and the pilots’ onboard flight management system. For this to be a viable system, these pieces must work together."

The importance of such an infrastructure was illustrated by the unintended impact of the current manual process on performance at one AOC. At that airline, many dispatchers routinely ignore the SPOs. If they don’t see the CDR in a reroute advisory (and it is usually the case that the use of CDRs is only mentioned in the SPOs and not in the reroute advisories), then they decide for themselves what contingencies are relevant, and fuel the flight based on their individual assessments. Fortunately, this strategy will usually provide the flight with enough fuel to deal with the use of a CDR as a reroute, even though no calculations were done for that specific route.

The reason for this is in part the unacceptably high additional workload that would have been added if they had to hunt through the SPOs for information about CDRs and had to then enter each alternative CDR into the flight planning computer and evaluate the results. (It may also be due in part to a lack of awareness by dispatchers at that airline that the use of CDRs is often only discussed in a SPO, and is not later included in any reroute advisories.) It is worth noting that although improved software tools would be one possible solution to reduce this workload, the procedures discussed above regarding how NWA handles the use of CDRs at DTW and MSP demonstrates an alternative or complementary approach. NWA redistributes the work, assigning the ATC Coordinator the task of working with traffic managers at ZOB to identify a smaller number of CDRs for consideration, and then having the ATC Coordinator communicate this smaller list to the appropriate dispatchers.

Planned versus Actual Fuel Burns. Some of the CDRs for certain city pairs involve considerably longer routes than the "normal" routes used during good weather. Furthermore, even when geographic distances are not too different, the fuel requirements can be quite different because of winds. This presents a challenge, as a flight cannot safely land if it has too much remaining fuel at its destination. (What constitutes "too much" requires a computation considering the aircraft type and load, runway lengths etc.)

Two issues arise regarding this issue. First, under the collaborative process described above, the aircraft is fueled for a number of alternate routes, and the actual route is selected near departure time. This implies that one of the constraints on the acceptability of alternate routes (CDRs) is whether they require a fuel level that would leave the flight too heavy to land if it in fact it is ultimately routed on the airline preferred route (which is what the airline is supposed to file). A similar issue arises if route amendments are made while the flight is enroute. If, for example, a CDR is used to expedite departure, but weather or traffic further into the flight makes it possible to give that flight a "shorter" route (in terms of "wind distance"), it could arrive at the destination with excess fuel. In order to avoid an overweight landing, the flight must then be put into airborne holding to burn off the excess fuel before landing.

Such considerations have implications for the design of CDRs. They also suggest the need for more careful consideration of the procedures to be used when weather (or traffic congestion) doesn’t materialize as predicted.

Delays Even When CDRs are Used. It might be expected that, even with the use of CDRs, flights will experience some delays because more flights have to depart via that alternative departure fix or because there is a constraint at the destination airport. (The assumption is that, without the use of CDRs the delays would have been even longer.) However, a preliminary analysis suggests that in many cases, these delays average 50-60 minutes or longer. It would be worthwhile to complete a more detailed analysis to determine whether the process for using CDRs can be streamlined to reduce such delays.

Non-Participation by Some Airlines, Centers and Airports. Not all airlines, Centers and airports are participating fully (or at all) in the use of CDRs. As one example, a Center traffic manager noted that the airline is simply filing departure routes without any coordination with the airport Tower or the affected Center. "They start pumping them out. That’s a hurt." As another example, one airline indicated that one of their major hubs was not using CDRs at all, even though it appeared to the airline that they would be very helpful in reducing delays. Finally, at times some Centers are choosing to use CDRs, but without the normal coordination with the airline or with other Centers (via the Command Center). An example of this is provided in the figures below.

This example involves a flight from DFW to BOS that was put on a CDR without any coordination with the airline’s AOC, with the Command Center, or with the other involved Centers. The flight crew was offered this CDR by the ATC Tower as they taxied out for departure. The crew did not clear this new route with dispatch prior to takeoff. However, while enroute, they decided to contact dispatch and learned that this new route was undesirable. Although it served to move them around a storm cell in Arkansas (thus enabling departure of the flight), it then routed them into a major storm in ZDC. After the flight crew contacted dispatch, a new route was planned and cleared with ATC which put them back on their original flight plan after deviating around the storm in Arkansas. Note that this sequence involves a breakdown in communication and coordination among FAA facilities as well as between the flight crew and dispatch. This story is told graphically in the following figures.

Figure 8. Flight information for a flight from DFW to BOS that was given a major reroute using a CDR.

Figure 9. Filed vs. flown routes (and amendments) for this flight from DFW to BOS.

The figure below shows the same map display, but with the CDR highlighted in the table (Amendment 1) and on the map.

Figure 10. Amendments for this flight from DFW to BOS.

The figure below show the actual weather on this day (8/31/00) at 2100Z.

Figure 11. Actual weather for this flight from DFW to BOS.

As noted at the beginning of this report, CDRs have been developed for two distinct purposes:

1. To serve as an abbreviation or coding for route entry and communication.
2. To support a collaborative process for reducing departure delays.

During the severe weather season this past summer, there were significant success stories for both of these distinct uses of CDRs. There were also, a number of areas where their design and use could be improved.

The Use of CDRs to Reduce Communications Workloads. From a human factors perspective, CDRs offer a tremendous opportunity to help reduce the workloads of traffic managers, controllers, dispatchers and pilots. The use of the codings for CDRs as abbreviations for route entries or selections reduces workload and improves efficiency when entering a route into the computer. Similarly, the use of the abbreviation for verbal clearances serves to improve the efficiency of human-human communication.

This application received unanimous support in terms of its potential value. A number of areas for improvement were also identified, however, including the challenges of:

1. Training a distributed workforce effectively so that everyone understands the abbreviations.
2. Developing processes to maintain consistency when the CDRs are stored in multiple databases at different sites.
3. Designing the program so that it takes into account the equipment limitations of different user groups (such as the limited capacity of onboard flight management systems in aircraft to store large sets of CDRs).

The Use of CDRs as Part of a Collaborative Procedure. The human factors issues associated with the use of CDRs to improve collaborative decision making are even richer. As with their use to speed communications, there is a great need to develop methods for more effectively and efficiently training a distributed workforce about new collaborative procedures. Because many people within different organizations at distributed locations must work together to make such collaborations successful, it is essential that they all have a common shared understanding of the process and their roles in it. Our review makes it clear that the training provided this past year, both within the FAA (for traffic managers and controllers) and within the airlines (AOC staff and flight crews), was not fully successful. There were numerous examples illustrating cases where one or more of the system participants exhibited either a lack of understanding or an incorrect understanding of how the collaborative process was meant to be conducted. As a result:

1. Many individuals and organizations chose not to participate.
2. Many individuals participated inappropriately, using CDRs in ideosyncratic ways that were not intended, and that resulted in a lack of coordination among different system participants.

Equally important, when redefining the roles and responsibilities of different individuals, it is critical to consider the impacts on their workloads, and on the level and extent of the expertise and staffing required for them to complete their tasks. When such considerations are not made in defining new procedures, the individuals are likely to either resist these changes or adopt satisficing strategies that allow them to cope with the resultant new complexities, but without fully complying with the defined procedures. The result of such satisficing behaviors is generally less efficient or less effective performance than was expected.

Related to this concern over workload is the availability of adequate technological support. If well-designed tools are not available to appropriately support each person in his or her role, and if these tools are not sufficiently integrated to allow effective synchronous and asynchronous interactions among these individuals, then well-intentioned new procedures may not be viable. Equally important, it is important to be realistic, recognizing the designers of such technological support tools are not likely to anticipate all of the scenarios that can arise. For this reason, it is important to take a human-centered approach to the design of such tools, designing them to assist rather than supplant the user, and allowing the user the flexibility to adapt to unanticipated situations (Billings, 1997).

Finally, it is important to realize that no process is perfect. There may be rough edges in the original design or implementation, or it may "wander" away from the original design over time. As a result, it is critical to provide adequate tools and procedures to provide feedback and quality assurance, thus ensuring that a high level of performance is maintained.

In conclusion, in order to develop effective approaches for collaborative decision making in general, and for the use of CDRs in particular, it is essential that we design an architecture for distributing work (Smith, Billings, et al., 1997; Smith, McCoy, et al., 1997; Smith, McCoy, et al., J., 1998) that:

1. Allocates responsibilities and defines procedures that are viable in terms of the workload demands (Kerns, K, Smith, P.J., et al., 1999) placed on the participating individuals.
2. Allocates responsibilities and defines procedures that are viable in terms of the extent of expertise (both breadth and depth) required of the participating individuals to perform their functions (Smith, Woods, et al., 1997).
3. Allocates responsibilities and defines procedures in such a way that safety nets exist to protect against human error (Fraser, Smith, et al., 1992) and the failure of technological tools (Smith, McCoy and Layton, 1997).
4. Allocates responsibilities and defines procedures that are viable within the organizational structures that exist and that must manage both the evolution to this new architecture and maintain its effective functioning once operational.
5. Considers the availability of resources to effectively implement and operate the revised system, and to effectively evolve from the current system to this new revised architecture.

Billings, C., 1997, Aviation Automation: The Search for a Human-Centered Approach. Erlbaum, Mahwah, NJ.

Kerns, K, Smith, P.J., McCoy, C.E., and Orasanu, J. (1999). Ergonomic issues in

air traffic management. In W. Marras and W. Karwowski (eds.). Handbook

of Industrial Ergonomics. New York: Marcel Dekker, Inc., 1979-2003.

Fraser, J. M., Smith, P. J., and Smith, J. W. (1992). A catalog of errors. International Journal of Man-Machine Studies, 37, 265-307.

Smith, P.J., McCoy, E. and Layton, C. (1997). Brittleness in the design of cooperative problem-solving systems: The effects on user performance. IEEE

Transactions on Systems, Man and Cybernetics, 27, 360-371.

Naturalistic Decision Making, Warrenton VA.

Smith, P. J., McCoy, E., Orasanu, J., Billings, C., Denning, R., Rodvold, M., Gee, T., and Van Horn, A. (1997). Control by permission: A case study of cooperative problem-solving in the interactions of airline dispatchers and ATCSCC. Air Traffic Control Quarterly, 4, 229-247.

Smith, P.J., Woods, D., McCoy, E., Billings, C., Sarter, N., Denning, R. and Dekker, S. (1998). Using forecasts of future incidents to evaluate future ATM system designs. Air Traffic Control Quarterly, 6, 71-86.