TAl - 11:00

SIMULATION OF TIME-CONTROL PROCEDURES FOR TERMINAL AREA FLOW MANAGEMENT

Monica Alcabin, Heinz Erzberger, Leonard TobiasNASA Ames Research CenterMoffett Field, California

P. J. O'BrienFAA Technical Center

Atlantic City, New Jersey

ABSTRACT

Simulations of a terminal area traffic-management system incorporating automated schedul-ing and time-control (four-dimensional (OD)) tech-niques conducted at NASA Ames Research Centerjointly with the Federal Aviation Administration,have shown that efficient procedures can be devel-oped for handling a mix of 4D-equipped and conven-tionally equipped aircraft. A crucial role inthis system is played by an ATC host computeralgorithm, referred to as a speed advisory, thatallows controllers to maintain accurate timeschedules of the conventionally equipped aircraftin the traffic mix. Results are of the mostrecent simulations in which two important specialcases were investigated. First, the effects of aspeed advisory on touchdown time scheduling areexamined, when unequipped aircraft are constrainedto follow fuel-optimized profiles in the near-terminal area, and rescheduling procedures aredeveloped to handle missed approaches of4D-equipped aircraft. Various performancemeasures, including controller opinion, are usedto evaluate the effectiveness of the procedures.

INTRODUCTION

A proposed element of a future air trafficcontrol (ATC) system is an on-board guidance sys-tem that can predict and control the touchdowntime of an aircraft to an accuracy of a few sec-onds throughout the descent. The performance andfeasibility of these systems, also known as four-dimensional (4D) guidance systems, have been dem-onstrated in several flight test programs (1,2).

In applying 4D guidance, the main problem isthe development of ATC procedures that can exploitthe on-board time-control capability. An earlystudy of this problem developed terminal areaprocedures under the assumption that all aircraftare MD-equipped (3). However, before reaching thegoal of a future system in which all aircraft are4D-equipped, it is necessary to study problems inthe long transition period, in which the trafficis composed of a mix of conventionally equippedand MD-equipped aircraft. The basic difficulty isthat in the 4D mode aircraft are separated bytime, whereas in the conventional mode they areseparated by distance. Developing techniques tohandle both types of aircraft effectively is acomplicated task that has been the subject of a

This pper is declared a work of the U.S. Govement adtherefore is in the public domain.

687

series of real-time simulation studies at NASAAmes Research Center.

In 1982 a real-time simulation study inves-tigated the operational problems of mixingMD-equipped and -unequipped aircraft in theterminal area (4). In that initial study, allMD-equipped or -unequipped aircraft departed thetop-of-descent feeder fix exactly at theirassigned departure times. The MD-equipped air-craft then flew their descents to touchdown with-out controller intervention. Unequipped aircraftflew profile descents to a point 30 n. mi. fromtouchdown, where they were radar-vectored onto astandard instrument landing system (ILS) approach.The study concluded that ATC procedures could bedeveloped to handle efficiently even low percent-ages of MD-equipped aircraft in the traffic mix.Moreover, the fuel efficiency of unequipped air-craft was not reduced by the special controllerprocedures adopted to accommodate MD-equippedaircraft.

The objective of this paper is to describerecent follow-on simulations of procedures forhandling more difficult traffic problems than weresimulated in the baseline study of Ref. 4. Thus,in one experiment unequipped aircraft were allowedto depart the top-of-descent feeder fix with largerandom time errors, and computer-generated adviso-ries were used to reduce the adverse effects ofthese errors. In the most recent experiments,controllers evaluated the feasibility of vectoringunequipped aircraft to follow a specified hori-zontal profile in the near-terminal area.Finally, procedures were tested to handle missedapproaches of MD-equipped aircraft.

The paper begins with a brief review of amethod for designing a scheduling system thatassigns conflict-free landing time slots. Thenthe simulation facility, controller procedures,and test variables used in the experiment aredescribed. Finally, simulation results are pre-sented by evaluating controller opinion and work-load for the various test conditions.

The study described in this paper is part ofa program of advanced flow control research con-ducted Jointly by Ames Research Center and theFederal Aviation Administration (FAA) TechnicalCenter.

A SCHEDULING SYSTEM FOR THEMIXED ENVIRONMENT

The 40-equipped aircraft have the capabilityof meeting a touchdown time assignment to an accu-racy of a few seconds. To use this capability toformulate efficient operational procedures for thetime scheduling of all aircraft in the terminalarea, it is necessary to 1) determine the inter-arrival time separations for two consecutiveaircraft to be used in aircraft scheduling;2) develop a scheduling algorithm for assigninglanding times; and 3) develop a rescheduling algo-rithm to fit missed approaches of 40-equippedaircraft back into the terminal area flow.

Time Separation Requirements

The present ATC system. uses radar vectors andspeed control to space aircraft so that the mini-mum separation distance rules are not violated.The rules depend on aircraft weight category(small for lowest weight category, large formedium weight category, and heavy for largestweight category) and are summarized in Fig. 1.For example, if a small aircraft is in trail fol-lowing a large aircraft in the landing sequence,these two aircraft must be separated by at least4 n. mi. during the entire length of the cooconfl ightpath.

TRAILING AIRCRAFTSMALL LARGE HEAVY

SMALL 3 3 3FIRSTTO LARGE 4 3 3LAND|L HEAVY 6 5 4

Fig. I Minimum separation distance (n. mi.).

These minimum separation distances can beconverted to minimum separation times using knowl-edge of the speed profile of each aircraft weightclass. In this way, a matrix of minimum timeseparations is determined as shown in Fig. 2.

TRAILING AIRCRAFTSMALL LARGE HEAVY

(SMALL 98 74 74FIRSTTO LARGE 138 74 74

LANDHEAVY 167 114 94

Fig. 2 Minimum separation time (sec).

It is assumed that, if there are two consecu-tive 40-equipped aircraft, the interarrival timesgiven in Fig. 2 can be used for scheduling pur-poses. However, unequipped aircraft will needadditional time buffers to prevent distance sepa-ration violations. The technique for obtainingthese buffers is discussed in Ref. 4. For thepurposes of this study, it was assumed that if one

of the two consecutive aircraft was unequipped, abuffer of 10 sec was added to the separation time;if both aircraft were unequipped, a buffer of20 sec was added.

Scheduling Algorithm

Based on knowledge of the feeder fix depar-ture time and on the desired time to traverse theroute, a desired touchdown time for each aircraftcan be determined. Using a first-come, first-served ordering protocol at touchdown and thetime-separation matrix developed in the previoussection, the time schedule at touchdown isobtained. It is possible to increase the capacityby altering the first-come, first-served order;thus, future studies will incorporate more effi-cient ordering algorithms to take advantage ofbunching of weight and speed classes (5).

In addition to setting up an initial sched-ule, algorithms are required to revise the sched-ule. For example, as shown later, missedapproaches need to be accommodated. The con-troller may need to change the aircraft arrivalrate, block out specific time periods from thecomputer schedule to accomuodate a missed approachor an emergency landing, or require a few aircraftto be scheduled in a specified order. This can beaccomplished by manipulating the existing sched-ule. Two procedures used during these simulationsto manipulate the existing schedule were the haltcomnand and the capture probe (CP) inquiry. (TheCP inquiry will be discussed in a later section.)

To explain the halt procedure, let us supposethat an initial schedule has been established forthose aircraft that have departed the feeder fix(denoted active aricraft) and those which have notyet departed the feeder fix (denoted inactiveaircraft). Controllers may need to hold the inac-tive aircraft for a time th to acconnodate amissed approach. A rescheduling algorithm leavesthe time assignments for the active aircraft unal-tered, but revises the touchdown times of inactiveaircraft by at least th. An illustration of thisis found in Fig. 3. For example, inactive air-craft (aircraft that have not yet departed thefeeder fix) El and Gl are halted while Al executesa missed approach. Aircraft are assumed to bescheduled 2 min apart. When Al is reschedulec[into the traffic flow, it is assigned to 9:14since there is an open slot at that time. Withthe 2-min halt, El would be rescheduled for 9:16,but since there is a previously scheduled activeaircraft landing at 9:16, El will be rescheduledfor 9:18 and Gi for 9:20. The effect of themissed approach is to leave active aircraft aspreviously scheduled, but to revise the inactiveaircraft schedule by 2-4 min.

TEST PROCEDURE

Simulation Facility

The simulations were conducted using the AmesATC Simulation Facility. It includes two airtraffic controller positions, each having its own

688

SCHEDULEDTOUCHDOWN

TIME (HRS: MIN)

9:06

9:089:109:129:14

9:169:18

EVENT:

Al EXECUTES A

MISSED APPROACH.

CONTROLLER ISSUES

HALT FOR

Th=2MIN

REVISED SCHEDULED1* - ATif%rkl AlMf%OACrlU = At;

AIRCRAFTID

B1iC?*Dl-Al*

F1?El

Gl

I PVC AIflLflr ij

SCHEDULEDTOUCHDOWN

TIME (HRS: MIN)

9:089:109:12

9:149:169:18

9:20

Fig. 3 Sample scheduling revision.

color computer graphics display; one was desig-nated arrival control and the other, final con-trol. In proximity to the color displays, therewas a keyboard with which the ATC-display-relatedrequests were entered into the controller displaysand the simulation computer. Such inputs includedchanging the position of an aircraft identifica-tion tag, transferring an aircraft between controlsectors, or stopping and restarting the flow oftraffic at the feeder fixes.

Each keyboard pilot position was able tocontrol up to 10 omputer-generated aircraftsimultaneously. The clearance vocabulary includedstandard heading, speed, and altitude clearancesas well as special clearances for 1W-equippedaircraft. In this study, three keyboard pilotpositions were used; one was responsible for allaircraft in the arrival sector while the other twodivided responsibility for aircraft in the finalsector. Previous studies have used one or twopiloted simulators connected by voice and datalink to the ATC Simulation Facility. However, inthis study, no piloted simulator was used. It isplanned to include an airline-quality simulator aswell as a helicopter simulator in future studiesof the mixed environment.

Scenario and Controller Procedures

The simulated terminal area is based on theJohn F. Kennedy (JFK) International Airport inNew York. The route structure and runway config-uration investigated as seen by the controllersare shown in Fig. 4. It is assumed that instru-ment flight rule conditions prevail and that allaircraft use runway 4R; furthermore, no departureflights, winds, or navigation errors are simu-lated. Two routes, Ellis from the north and Satesfrom the south, are high-altitude routes flown bylarge or heavy jet-transport aircraft. Both4D-equipped and -unequipped aircraft on theseroutes fly profile-descent, fuel-conservative

procedures, providing a mix of the same speedclass on the same route. Low-performance aircraftflew the Deerpark route from the east, but usedthe same final approach and landed on the samerunway as the jet traffic. The Deerpark trafficwas unequipped and always constituted 25% of thetraffic mix.

During this study, an extended terminal areawas used. Aircraft entered the extended terminalarea at the feeder fix points and were at cruisespeed and altitude. The total distance to beflown along each of the jet routes was 120 n. mi.and that flown by low-performance aircraft was60 n. mi. Two air traffic controller positionswere established, arrival control and final con-trol. The arrival controller controlled aircraftfrom all three feeder fixes and transferred traf-fic to the final controller at approximately30 n. mi. from touchdown.

Control procedures differed for 4W-equippedand -unequipped aircraft. Controllers wereinstructed to monitor the progress of 26D-equippedaircraft after the time assignment had been estab-lished, and to override the ground-computer sched-uling system only if necessary to ensure safeseparation. Any 4D-equipped aircraft could alsobe controlled by conventional methods and treatedas unequipped. Alternatively, a 4D-equipped air-craft which had been taken off the 4D route couldbe given a waypoint to recapture a 4D route, andbe given a revised landing time. Unequipped air-craft were considered to be navigating in theconventional manner via very-high-frequency omni-directional range procedures, with altitude clear-ances, radar vectors, and speed control.

Flight Data Table

To assist the controller in integrating the4D-equipped and -unequipped traffic, a flight datatable (FDT) was provided on each controller

689

INITIAL SCHEDULE( = ACTIVE AIRCRAFT)

AIRCRAFTID

Al*

B1*C14DI'El

F1*

*u - - w u -- - - -- - - - - - - -- - - - - -

M2

ELLIS

13: 27:12

ID TYPE RT STA

Rl1IL2JATiA2E2M2

G2

4H4SUU4

HUU

HU

H4

SASADPELSAELSAEL

SA

3700382441314304429471549295103

5247

JADY CAS00

-50 11535 2600

-80 230120 290100 280

2849

LZ

? DEERPARK

LEGS

11

Ti

rE2

Fig. 4 Controller display showing route structure and flight data table.

display. A typical arrival controller display isshown in Fig. 4. The map-portion of the displayprovides a horizontal display of traffic in theterminal area. Each aircraft position is shown bya triangular symbol, and the block of data next toeach aircraft provides the aircraft identifica-tion, type, altitude, and speed. The flight datatable in the upper left portion of the displayprovides schedule information for all aircraft inthe approach control sector. At the top of thetable, the time is shown in hours, minutes, andseconds. The first column shows the aircraftidentification (ACID), such as "Rl." The secondcolumn provides aircraft type (TYPE) whichincludes a) weight category (small (S), large(blank), or heavy (H)); and b) 4D status (equipped(4) or unequipped (U)). The third column providesthe assigned route (RT). Also shown is the sched-uled time of arrival (STA) at the runway in mrin-utes and seconds. Thus, RI is scheduled to touchdown at 13:37:00. Note that touchdown times areshown for all aircraft, whether 4D-equipped or-unequipped. For the 4D-equipped aircraft, thisis the time assigned by the ground-based computersystem to touchdown. For the unequipped aircraft,no time assignment is given to the aircraft;rather, the controller is to use this information,the positions of the 4D-equipped aircraft as theytraverse their routes, and the speed advisories to

generate appropriate vectors to the unequippedaircraft so that they touch down at the time indi-cated. The next column is the expected delay(DY), where the expected delay at touchdown is inseconds. In an effort to simulate an environmentin which an advanced en-route metering system isnot present, the unequipped aircraft were assumedto depart with an -initial time error uniformlydistributed in the range ±20 sec.' Thus, if anaircraft departed the feeder fix 90 see late, a DYof 90 would be displayed, indicating that unlesscontroller action was taken, the aircraft wouldtouch down 90 sec late. Early arrivals were indi-cated by a negative value in the DY column; latearrivals by a positive value. All 49-equippedaircraft departed the feeder fix at the scheduled

'These delays are considered large, since,for example, two consecutive OD-equipped aircraftcould arrive in the final control sector inreversed order if there are consecutive departureerrors of +60 sec and -60 sec. Also, some infor-mal testing of controllers by the FAA in the seg-ment area indicated that, with practice, control-lers could get aircraft over a designated arrivalfix within ±1 min of an assigned time, without anycomputer-generated assists.

690

departure time. In flight tests, it has beenshown that 4D-equipped aircraft can meet timeschedules within ±5 sec; hence, these small errorswere neglected (1,2).

Finally, aircraft below the dotted line inFig. 6 are aircraft which will depart the feederfix within the next 5 min (shown in the right-mostcolumn) indicated by the feeder fix-departure timein minutes and seconds.

Speed Advisory System

In the scheduling process all aircraft,4D-equipped or -unequipped, are assigned a touch-down time. For the LD-equipped aircraft, thistime assignment generated on the ground is trans-mitted to the aircraft, which uses its on-board 4Dsystem to land at the assigned time. However,unequipped aircraft must be controlled by vectorsand speed clearances. It is the task of the con-troller to issue these clearances so as to meetthe assigned time. However, in the arrivalsector, 120 n. mi. from touchdown, the controllercannot predict with accuracy when an aircraft willland and how it will fit into the landingsequence.

Thus, an algorithm known as a speed advisorysystem was developed to give the arrival control-ler a tool for controlling the landing time ofunequipped aircraft. With inputs of current posi-tion, altitude, and speed, the speed advisorysystem computes and displays to the controller thecalibrated airspeed (CAS) that is required to makethe aircraft land at the desired landing time (6).

The arrival controller, presented with atable of speed advisories for the unequipped air-craft on his or her display, is responsible forissuing the advisories to pilots starting theirdescents. During the descent the speed advisorysystem updates the time error by comparing thepredicted position with the actual position of theaircraft. If the error exceeds ±20 sec, the speedadvisory is updated on the controller's display.This feature helps to reduce time errors caused byunknown winds and pilot tracking inaccuracies.

Once the CAS advisory is issued and the pilotadjusts the speed accordingly, the gap between thereference position and the actual aircraft posi-tion narrows and the magnitude of the delay isreduced. If the magnitude of the delay is lessthan 20 sec, or the aircraft is less than25 n. mi. from touchdown, the CAS advisory isremoved. This discourages the inappropriate useof the advisory. In the final control sector, thecontroller uses conventional vectoring techniquesto correct any remaining spacing errors.

Capture Probe

Another controller assist that was providedto handle missed approaches was a CP inquiry whichassisted in determining a feasible revised touch-down time for the missed-approach aircraft. The

approach controller provided the ID of the missed-approach aircraft and a specific waypoint along aroute to the ground-based computer system. Thecomputer then generated a 4D trajectory whichwould take the missed-approach aircraft from itspresent course (away from the runway) to a way-point on a landing route and then schedule it fora second approach. On the approach controller'sdisplay, the scheduled time of arrival for thataircraft (if it were to change its course andcapture that waypoint) appeared in the FDT in theappropriate time order. The time and positionsequence in the FDT would be updated every 10 secas the aircraft proceeded along its presentcourse. When the revised scheduled time ofarrival was not in conflict with the scheduledtime for other aircraft, the approach controllerissued a command to the pseudopilot of thataircraft to capture that waypoint. At this time,the aircraft made the change and proceeded tofollow its 4D route from that waypoint, withoutcontroller assistance.

Test Conditions

In all test cases, unequipped aircraftdeparted the feeder fix with random departureerrors uniformly distributed in the range±120 sec. The traffic mixes examined were 0%,25%, and 50% 4D-equipped aircraft. In all cases,the time information (STA) was displayed. Someruns displayed the CAS speed advisory and DY andsome did not. In addition, simulation runs wereconducted for the 0% case when STA, DY, and CASwere not displayed, which corresponds to presentoperations.

In a series of runs, controllers wereinstructed to keep unequipped aircraft on fuel-optimized profiles for cases in which 25% and 50%of the traffic was 4D-equipped aircraft. Casesdistinguished between runs in which the CAS speedadvisory was displayed to the controller and thosein which it was not.

Another series of runs involved reschedulingmissed approaches of 4D-equipped aircraft for both25% and 50% 4D-equipped aircraft mixes. In theseruns, all information contained in the FDT wasdisplayed. In all runs the aircraft arrival rateinto the terminal area was chosen so that a fullschedule with no gaps was generated.

No departure traffic was simulated, and windsor navigation errors were not modeled explic-itly. The tracking inaccuracies caused by navi-gation errors and wind uncertainties were assumedto be incorporated in the feeder fix departureerrors.

Forty-eight data runs were made, each approx-imately 80 min long. Three air traffic control-lers from the FAA Technical Center, Atlantic City,N.J., participated as subjects in these studies.

691

RESULTS AND DISCUSSION

Controller Evaluations of Mix Conditions and CASAdvisory

TheEmix conditions and the CAS advisory wereevaluated in a simulation experiment conducted ayear before this experiment investigating follow-ing exact routes and missed-approach procedures.Results from the earlier experiment, which werereported previously in Ref. 7, are brieflyreviewed here because of their close relationshipto those of the most recent experiment.

The 25% 49-equipped case was rated by con-trollers as the condition with the heaviest work-load. The main difficulty seemed to be that thecontrollers were establishing distance spacing forthe majority of the traffic, and they felt that bynot altering the flightpath of the ID-equippedaircraft, they were occasionally losing some slottime. However, they were pleased with the 50%4D-equipped case, which allowed for easy controlof the unequipped aircraft. One controller com-mented that it was the best ratio; he could "workwithout being overtaxed." In this mode, fewercomnunications were required, the traffic flow wasmore orderly, and it was easy to fill the gapsbetween the 4D-equipped aircraft with theunequipped aircraft. The baseline case of 0%4D-equipped aircraft was regarded as reasonable,but not because of a lessening of workload;rather, because it was the most familiar.

The controllers were provided with time-scheduling information in the FDT previouslydescribed. The controllers indicated that theonly information they used was the time-orderedlisting of aircraft from which the relstive orderof traffic on the downwind leg and the trafficfrom Sates was determined. Using this informa-tion, and by not altering the 4D-equipped air-craft, controllers were able to radar vector theunequipped aircraft to their assigned landingslots.

Controllers were asked to evaluate the speedadvisory. Controllers compared operations bothwith and without the speed advisory and evaluatedthe format and operational procedures establishedfor the advisory. They commented that without thedelay or advisory information displayed and withlarge feeder fix departure errors present, thesystem yielded traffic surges. Speed advisoriesresulted in less bunching of traffic and fewer"ties" in the merging area. With the advisories,the traffic seemed to blend together smoothly andrequire fewer vectors.

Following Exact Routes

In this set of experimental conditions, con-trollers were instructed to keep 4D-equipped air-craft along the nominal base leg route (Fig. 4).Controllers were asked to turn aircraft onto thebase leg at the same point, and thereafter useonly speed adjustments to control spacing. Thisprocedure sought to keep unequipped aircraft ontheir planned profile descent paths longer to

minimize fuel consumption. The procedure was alsobelieved to be helpful for controlling time errorsby reducing the variability of turning to baseleg. One important question to be answered bythese tests was whether controllers could changefrom path control to speed control as an alterna-tive for making small spacing adjustments in thebase leg region.

Controllers were initially hesitant whenasked to keep unequipped aircraft on specifiedflightpaths, but, after an initial trial periodwith speed adjustments only, they said it seemedto go well. Nevertheless, their preference was torevert to path-stretching techniques because theywere more accustomed to those procedures. Theeffect of the CAS advisory turned out to be impor-tant in the cases tested. In both the 25% and 50%4D-equipped cases, it was judged easier to followexact routes when CAS was provided than when CASwas not provided. The 25% 4D-equipped cases werefound to be more difficult than the 50%1W-equipped cases, in accordance with previousresults. A comparison was made between the air-space used when controllers were and were notconstrained to keep unequipped aircraft on speci-fied flightpaths. Figures 5a and 5b are the enve-lope of composite plots of the flightpaths flownfor these two cases. The procedure of followingthe exact routes in Fig. 5a has significantlyreduced the area in the airspace envelope comparedto that in Fig. 5b. Although the area has notbeen reduced to zero, it establishes the limit inaccuracy of following a prescribed route usingconventional vectoring techniques.

In summary, the use of heading vectors toguide the aircraft along a prescribed route andspeed control to make fine adjustments in spacingis a workable alternative to path stretching. Thetechnique is acceptable, though not preferred, bythe final controller, provided that time errorsare kept small at the hand-off point betweenapproach controller and final controller.

Missed-Approach Procedures

The procedure used to handle missedapproaches was that once a missed approach wasexecuted, aircraft that had not yet departed thefeeder fix were halted for 2 min. Since a landingslot had- been lost because of the missed approach,a new one had to be created. This 2-min slot timewas the amount decided upon to create a new slotfor the mised-approach aircraft with the leastamount of disruption to the rest of the traffic.Meanwhile, the missed approach proceeded along themissed-approach route toward waypoint 25(Fig. 6). Once the 2 min had elapsed, theapproach controller issued a CP inquiry for themissed approach to capture waypoint 19. When asuitable conflict-free time appeared in the FDTschedule, the approach controller instructed thepseudopilot of the missed-approach aircraft tocapture waypoint 19. The new time appeared on theFDT in the proper sequence, and the OD-equippedaircraft proceeded on its flightpath without con-troller assistance.

692

a) 50% 4D-equipped with CAS following exact routes.

b) 50% 4D-equipped with conventional vectoring.

Fig. 5 Composite plot of flightpaths flown.

693

7.0 -

z

16.5-

6.0-

5.517.0 7.5 8.0

Y AXI

Fig. 6 Composite plot showing f

Controllers were generally pleased with thisprocedure. Controllers found that if the missedapproach was rescheduled so that it was positionedbehind another MD-equipped aircraft, the control-ler was able to achieve the proper spacing betweenthem. But, if the controller was forced to posi-tion the missed approach after an unequipped air-craft because of the schedule, he or she had tovector the missed approach for a longer period oftime before a waypoint was captured. Workload tohandle a missed approach was considered the samefor both 25% and 50% MD-equipped runs. Althoughonly MD-equipped aircraft executed missedapproaches, they did not cause an increase inclearances to the unequipped traffic. In usingthis procedure, a third controller, designated theflow controller, assisted the approach control-ler. With the level of automation simulated here,the flow controller (who was able to view bothcontrollers' FDTs) was invaluable in handling thelogistics of rescheduling the missed approach.

Figure 6 is a typical composite plot of a 50%MD-equipped run in which there were two missedapproaches. Waypoints 19 and 25 are shown as arethe routes taken by the two missed approaches.The "Capture Waypoint 19" commands were issued atpoints A and B in the figure. The capture trajec-tory generated by the on-board 4D system egtendsfrom points A and B to waypoint 19.

I I

8.5 9.0 9.5ISIN 100,000ftflightpaths of two missed approaches.

CONCLUDING REMARKS

A series of operational procedures for han-dling a mix of 4D-equipped and unequipped aircraftin the terminal area was investigated. The basicrule not to alter the 4D-equipped aircraft oncethey were assigned a landing time, resulted in thecontrollers learning to use the MD-equippedaircraft positions to effectively vector theunequipped aircraft to their assigned landingslots.

The speed advisory algorithm proved to be aneffective operational procedure for nullifyingtouchdown times of unequipped aircraft. Thisoperational procedure did not create additionalworkload for the approach controller and allowedthe final controller to handle the traffic in thesame manner as if there were no initial timeerrors.

Handling unequipped aircraft that were con-strained to follow fuel-optimized profiles in theterminal area was accomplished by using a speedadvisory which helped in derandomizing the flow.Composite plots of the flightpaths flown show thatcontrol of unequipped aircraft in a 4D environmentcan be restricted to specified horizontal routes.

A simple rescheduling procedure was developedto handle missed approaches of MD-equipped air-craft. This involved halting inactive aircraft

694

for a prespecified time to create an extra slot inthe terminal flow, and then scanning a series ofpossible landing times for the missed approach.When a suitable touchdown time was decided upon bythe approach and flow controllers, the 4D-equippedaircraft recaptured the route and no longer neededto be monitored by the air traffic controller.

Several issues raised by these simulationsindicate a need for further investigation. First,a simulation should be conducted to include apiloted simulator to investigate a pilot'sresponse to speed advisories. Second, furtherdevelopment of efficient ground-based 4D algo-rithms for unequipped aircraft is needed. Andthird, algorithms for optimized, revised timeschedules should be developed.

REFERENCES

(1) Lee, H. Q., Neuman, F., and Hardy, G. H., "t4DArea Navigation System Description and FlightTest Results," NASA TN D-7874, 1975.

(2) Knox, C. E., "Fuel Conservative Descents in aTime Based Metering Environment," Proceedingsof the 18th IEEE Conference on Decision andControl, Fort Lauderdale, Fla., Dec. 1979..

(3) Tobias, L. and O'Brien, P. J., "Real TimeManned Simulation of Advanced Terminal AreaGuidance Concepts for Short-Haul Operations,"NASA TN D-8499, 1977.

(4) Tobias, L., "Time Scheduling a Mix of 4DEquipped and Unequipped Aircraft," NASATM-84327, 1983.

(5) Dear, R., "Increasing Capacity with Computer-Assisted Decision Making," Proceedings of theInternational Air Transportation Conference,Vol. 2, May 1979.

(6) Erzberger, H. and Chapel, J. D., "Conceptsand Algorithms for Terminal Area TrafficManagement," Proceedings of the 1984 AmericanControl Conference, San Diego, Calif., June1984.

(7) Tobias, L. and O'Brien, P. J., "Simulation ofTime-Control Procedures for the Advanced AirTraffic Control System," Proceedings of the1984 American Control Conference, San Diego,Calif., June 1984.

695