MONTE CARLO ANALYSIS OF AIRPORT THROUGHPUT AND TRAFFIC DELAYS USING SELF SEPARATION PROCEDURES Maria Consiglio, James Sturdy** * NASA Langley Research Center, ** Raytheon Technical Services Company Keywords: Air Traffic Management, Self-Separation, Self-Spacing Abstract participating Visual Flight Rules (VFR) aircraft at all towered airports: they control aircraft on This paper presents the results of three the runway and in the controlled airspace simulation studies of throughput and delay immediately surrounding the airport. They times of arrival and departure operations coordinate the sequencing of aircraft in the performed at non-towered, non-radar airports traffic pattern and direct aircraft on how to using self-separation procedures. The studies safely land and depart at and from the airport. were conducted as part of the validation Conversely, at airports without control towers process of the Small Aircraft Transportation and radar coverage, IFR flights are limited to Systems Higher Volume Operations (SATS one operation at a time, severely reducing the HVO) concept and include an analysis of the utilization of these airports. HVO procedures predicted airport capacity using with different can increase the rate of operations at non-radar, traffic conditions and system constraints under non-towered airports by enabling multiple increasing levels of demand. Results show that simultaneous arrivals and departures in near SATS HVO procedures can dramatically all-weather conditions within a designated increase capacity at non-towered, non-radar volume of airspace where pilots have the airports and that the concept offers the responsibility for maintaining safe separation potential for increasing capacity of the overall from other traffic. A complete description of air transportation system. SATS HVO concept and procedures can be found in References 1, 2, 3 and 4. This concept 1 Introduction is particularly relevant now that the air transportation system is going through a The Small Aircraft Transportation System complete revision intended to address the Higher Volume Operations (SATS HVO) expected increased demand of the future as concept of operations was developed as part of described in the recent Joint Planning and an effort to address the capacity problem of the Development Office (JPDO) report on the Next air traffic system in the United States by Generation Air Transportation System promoting more evenly distributed air traffic (NGATS) [5]. and reducing congestion at large hub airports. The three Monte Carlo studies presented This is accomplished by increasing access to in this paper are part of the SATS HVO thousands of public use airports in the United validation process that included safety States without a major impact on the air traffic verification and formal methods analysis [6,7], controller’s workload or on overall National batch studies of performance [8,9], human-in- Airspace System (NAS) structures and the-loop experiments designed to measure principles. pilot’s and controllers’ workload and situation In today’s system air traffic controllers awareness [10,11,12], and flight tests [13]. The (ATC) provide sequencing and separation for three different experiments were designed to all Instrument Flight Rules (IFR) and 1 compare measures of delay and throughput performance, or by using an on-board self- obtained using both today’s procedure-based spacing tool. separation and SATS HVO validating the For departures, pilots must file flight plans predicted capacity improvements of the SATS with a SATS HVO departure procedure to a HVO concept. Departure Fix (DF), obtain ATC clearance, and then use onboard information/tools to find a departure window, (e.g., safe separation from 2 Overview of the SATS HVO Procedures other approaching and departing traffic). Since The SATS HVO concept relies on the departures are not sequenced by the AMM then establishment of a volume of airspace, they are not limited by the SCA configuration identified as the Self Controlled Area (SCA) or the number of ongoing approach operations. surrounding the airports within which pilots In other words, departures have no impact on assume responsibility for self-separation. arrival delays, while in today’s procedural Flights operating in the SCA, during instrument separation; departures are separated from meteorological conditions, are given approach arrivals in a one-in-one-out manner. Figure 1 sequencing information computed by a ground represents a nominal circular SCA with two based automated system referred to as the IAFs, Cathy and Annie also serve as the missed Airport Management Module (AMM). All approach holding fixes (MAHFs). The two participating aircraft are required to be equipped with Automatic Dependent Surveillance-Broadcast (ADS-B) and must be able to communicate with the AMM. The AMM functions do not include separation services, altitude assignments, or departure sequencing information. Aircraft arriving into the SATS designated airport are managed by ATC according to an IFR flight plan to a transition fix above the SCA. The transition fix is also the Initial Approach Fix (IAF) on a GPS-T instrument approach procedure. Prior to reaching the DFs, Ginny and Ellen are located transition fix, pilots must request landing Fig. 1. Nominal SCA and approach geometry sequence information from the AMM. The AMM message includes relative sequence outside the SCA. There are also two arriving information that identifies a lead aircraft to be aircraft (Red and Blue) with alternating followed. If the SCA is full, then the AMM MAHFs, and two departing aircraft (Green and sends a “stand by” message. An SCA Purple) depicted in a “snapshot” in time. configuration such as the one shown in Fig. 1 The SATS HVO concept relies on pilots allows a maximum of four aircraft on approach complying with procedures, communicating at any given time. An SCA with only one IAF their intentions and maintaining some degree of would have a maximum capacity of two synchrony during operations. While minor approaches at any given time. deviations from these rules may have no Pilots in the SCA initiate their approach negative effects, major procedure violations once adequate spacing behind the lead aircraft can be significant. A detailed description of has been met. Adequate spacing is determined the SATS HVO concept is out of the scope of through either a generic rule-based spacing this paper, but can be found in References 1, 2, procedure, safe for all combinations of aircraft 3, and 4. 2 3 Experimental Platform throughput obtained using both today’s procedure-based separation, referred to as The Air Traffic Operations Laboratory (ATOL) Baseline operations, and SATS HVO. The is a simulation facility developed by NASA model was implemented as a finite horizon Langley Research Center to evaluate new air discrete event simulation and special traffic management concepts. The General consideration was given to minimize both the Aviation (GA) simulation platform has been initial state and terminating condition biases. In specially designed to investigate issues unique all runs, the first hour of simulation was to General Aviation (GA), specifically discarded, and the runs were stopped after 11 integrating unscheduled operations into the hours of execution without allowing the system NGATS and flight operations at airports to clear the queues. without Air Traffic Control (ATC) services. The simulated airport environment The GA portion of ATOL is a distributed real consisted of a single runway non-tower, non- time simulation platform that can be run in radar airport. The approach geometry modeled single-pilot, multi-pilot, or remote-site linked was a GPS “T” approach with two IAFs and simulation experiments, all either with or two DFs located outside the SCA. A nominal without virtual traffic and/or voice round SCA was represented with a 15 nm communications. A number of unique software diameter extending from sea level up to 3000 components provide both real time and batch feet of altitude. Holding altitudes for the IAFs simulation capabilities. The General Aviation were 2000 and 3000 feet for aircraft inside the Airport Traffic (GAAT) simulator used in this SCA, and 4000 feet and above for aircraft study is a virtual traffic generator that can under ATC control. The approaching traffic operate as a real time tool for concept consisted of two airport departure routes and exploration and demonstration and as a batch four fixed airport arrival routes as shown in tool for Monte Carlo simulation experiments. Figure 2. Input traffic streams comprised The GAAT tool allows aircraft types, multiple aircraft types and different performance and characteristics to be performance characteristics. configured to simulate different airport traffic combinations. Traffic patterns can also be configured to represent different sources of arrival streams with configurable rates. Pilot DF and ATC models implement all the necessary (cid:79)IAF (cid:79) interactions to compare baseline operations (cid:79) FAF with HVO. The pilot model enables the virtual (cid:79)IAF (cid:79) aircraft to follow the sequence instructions DF given by the AMM model, maintain self- separation and proper spacing from a leading aircraft while in the SCA, as well as descend or Fig. 2. Simulated Arrival and Departure Routes climb to appropriate altitudes and maintain intended speeds. The ATC model assigns Each input traffic stream had exponentially holding altitudes outside the SCA, provides distributed inter-arrival times. Each run departure clearances to all departing aircraft collected data for ten hours of simulated traffic and provides approach clearances in non-HVO flow. scenarios. To reduce the initial condition bias associated with queue filling, data collection did not include the first hour of simulation. 4 Experiment Design Data collection stopped at the 11th hour A set of three batch experiments were without allowing the input queues to empty. conducted to compare measures of delay and Multiple replicates were performed for each 3 configuration to achieve a 95% confidence operations (failures or emergency conditions) level that the mean arrival delay estimate was or missed approaches were simulated. The within a 2-minute error interval. In all independent variable in all cases is the average scenarios, all traffic followed procedures and input operations rate. ATC instructions, and no non-normal Average Arrivals and Departure Delays 120 s100 Baseline Departure Delays e 97.63 t Baseline Arrival Delay u n 86.53 HVO Arrival Delays Mi 80 HVO Departure Delays" n i ay 60 61.84 el D 48.02 45.92 n o 40 37.56 37.27 ti a r e p 20 O 13.75 17.09 1.37 5.43 8.26 4.97 6.02 0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 Average Operation Rate Fig. 3. Average Arrival and Departure Delays for Baseline and HVO Operations considering only holding time and not the descent path up the IAF. 4.1 Experiment 1: Balanced Traffic Load 2. Queue-Size: Number of aircraft on Stand-By In this experiment, approaches were uniformly or Hold. distributed and “balanced” (evenly distributed) 3. Departure-Delay: Computed as the time among all four sources. The average operation elapsed between a clearance request and the rate was increased for each run of 10 simulated time the pilot turns on the runway. During a hours of operations from 2 to 32 operations per SATS operation this happens when there is hour. Operations were on average 50% enough separation from traffic, at least a 5 approaches and 50% departures. miles separation from approaching traffic and The goal of the study was to analyze the either 3 or 10 miles separation from other impact of SATS HVO procedures on airport departing traffic depending on whether the capacity and traffic delays. previous departure was on the opposite or same Performance metrics collected for SATS HVO departure fix (DF). scenarios includes: 4. Throughput: Computed as the average 1. Arrival-Delay: The sum of Stand-By-Delay number of operations completed per hour. plus Time-on-Hold. Stand-by-Delay is Throughput is computed as the average of the computed as the elapsed time between a means for all replicates. This includes both “STAND-BY” message and the entry approaches and departures. notification sent by the AMM. Time-on-Hold is Performance metrics collected for zero for an aircraft that is first on approach. Baseline scenarios include: Otherwise, it is the time elapsed between 1. Total-Delay: The time elapsed between reaching the IAF and the time the pilot initiates reaching the IAF and crossing the IAF on the approach. This distinction allows approach. 4 2. Queue-Size: Number of aircraft on Stand-By operations per hour. These results seem to or Hold. indicate that SATS HVO can accept an input 3. Departure Delay: The time elapsed between rate three times that of Baseline. To experience a departure clearance request and the clearance comparable average delays the Baseline input being granted. rate must be less than six average operations 4. Throughput: Same as above. per hour, in which case SATS HVO input rate is four times higher. 4.1.1 Arrival Delay Analysis Figure 3 shows the comparative average arrival 4.1.2 Departures Delay Data and departure delays of Baseline and SATS As shown in Figure 3 departure delay results HVO operations for an airport demand ranging are consistent with the arrivals results. from 2 to 32 operations per hour. Baseline and SATS HVO departures SATS HVO arrival operations can experience comparable delays with input rates f sustain a combined average demand of 26 8 and 28 operations (4 and 14 departures) per operations (13 approaches and 13 departures) hour respectively. HVO departures still show per hour with an average delay of 5.43 minutes. acceptable departure delays at an average rate When the input rate increases to 28 operations of 30 operations per hour (15 departures) and per hour the system reaches saturation and the begin to degrade at 30 average operations per delay curves begin to grow. hour. These results show that SATS HVO Baseline arrival operations show an operations can support much higher traffic average delay of 13.75 minutes for an average loads and about four times as many departures combined input rate of 8 average operations per per hour with lower departure delays relative to hour (4 approach and 4 departure operations) baseline operations. and the system reaches saturation at 10 average Average Throughput and Queue Lengths 30 26.43 27.4 27.7 HVO Average Throughput per Hour 25.7 25 Departure Queue 24.0 Arrival Queue 22.0 r Baseline Throughput u 20.0 o 20 H 17.9 r 16.0 e p 15 13.9 15.0 s n 11.9 ratio 10 8.0 9.8 8.0 10.5 7.9 pe 5.8 9.0 9.1 9.1 O 5 4.0 4.0 1.99 1.1 1.8 0.7 0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 Average Operation Rate Fig 4. Average HVO Throughput and Queue Lengths results complement the delays results from 4.1.3 Throughput Analysis Figure 3 showing the SATS HVO throughput Figure 4 shows the SATS HVO and Baseline increasing with input rate until it reaches an throughput and queue lengths observed during average of 26 operations per hour. At that point the study under increasing traffic loads. The the system reaches saturation and the queues 5 begin to build up. In fact, the peak sustainable experiment but the approaching traffic pattern combined throughput is reached at 24 was modified to cause overloading of one IAF. operations per hour. As is was described in In this case, the ATC model made no effort to Section 2, the frequency of arrivals have no balance the load between the two IAFs causing impact on departures in the SATS HVO the arriving traffic additional delays. As before, concept, therefore, the arrivals and departure 50% of the operations were arrivals and 50% queues reflect different points of saturation that were departures. depend on the duration of the respective In one set of runs, arrivals were operations. Baseline throughput also shown in distributed between 40% to one IAF and 10% Figure 4, increases until the input rate reaches to the opposite IAF. As shown in Figure 5, the eight operations per hour and from that point average arrival delay for SATS HVO on the system becomes saturated, with a peak operations reflect the impact of unevenly sustainable average completion rate of eight distributed arrival traffic. The average delay for and a maximum throughput of 9.1 operations an average input rate of 20 operations per hour per hour. is about five minutes, about twice the delay in the balanced load experiment. At this point the system begins to reach saturation, and delay 4.2 Experiment 2: Unbalanced Traffic Load times begin to rise. As expected, the Baseline The goal of this study was to estimate the operations were virtually unchanged since they system capacity metrics in the presence of were always one operation at a time. unevenly distributed arrivals. The study looked In the second set of runs, only a single at delay times of SATS HVO vs. Baseline IAF was used, so all the arrivals were assigned operations under increasing (unbalanced) to a single IAF. In this case the SATS HVO traffic loads. delays indicate a sustainable demand of 18 Two separate set of runs were operations (nine arrivals) per hour, still about performed. In both of them, the initial twice the sustainable Baseline demand. conditions were similar to the previous Average Arrival Delay for Unbalanced Loads 50 46.6 "HVO Average Arrival Delay" 40 n 36.4 "Baseline Average Arrival Delay" 38.2 i y Single IAF ela es30 29.2 D ut e n g Mi20 a 17.4 r e 13.7 v A 10 3.1 3.6 0.6 0 2 4 6 8 10 12 14 16 18 20 22 24 26 Average Operations per Hour Fig. 5. Average Arrival delay for unbalanced loads operations under increasing (balanced) traffic loads. In-trail spacing constraints between 4.3 Experiment 3: Spacing Constraints approaching an aircraft and its lead on SATS The goal of this study was to estimate the HVO are affected by the equipage of the impact of approach spacing constraints on aircraft. Aircraft equipped with an advisory delay times and throughput of SATS HVO 6 system such as the “Pilot Advisor” (PA) [14] HVO delays and throughput. All the initial are expected to maintain a minimum spacing of conditions were the same as the first study of 3 nm with their lead aircraft on approach. The balanced loads in section 4.2 except for the PA advises pilots on when to initiate the minimum spacing constraints that ranged from approach based on position reports and 3 nm to 10 nm of spacing between pairs of intended speed range of the lead aircraft. In leading and trailing arriving aircraft. Input rates addition, the PA monitors the spacing during ranged from 12 to 30 average operations per approach and provides pilots speed adjusting hour. advisories. More information on the PA can be As depicted in Figure 6, the results found in Reference 14 and 15. For aircraft not indicate that approach spacing constraints have equipped with this kind of automation, an effect on throughput and delays for both procedures require that pilots wait for the lead arrivals and departures. Arrivals throughput is aircraft to be at the Final Approach Fix (FAF) maximized with the minimum spacing before initiating the approach, resulting in more constraint of 3 nm, reaching an average of 14.4 conservative spacing of about 10 nm. Pilots in completed arrivals per hour while departures this case must monitor the lead aircraft on their reach an average of 13.0 completed departures Cockpit Display of Traffic Automation (CDTI) per hour. In contrast, arrivals throughput is and use their own judgment to initiate sharply reduced with 10 nm of spacing approach. reaching a maximum of 9.4 completed arrivals This study sought to estimate the impact per hour while departures reach 19.1 operations of conservative spacing constraints on SATS per hour within the tested input range. Arrivals and Departures Throughput using diferent spacing constraints 20 19.1 r u ho 18 Arrivals Throughput at 3 NM r Departures Throughput at 3 NM e p Arrivals Throughput at 10 NM s 16 Departure Throughput at 10 NM n o 14.4 ati 14 14.0 r e p O d 12 13.0 e t e pl m 10 o 9.4 C 8 12 14 16 18 20 22 24 26 28 30 Average Operations per Hour Fig. 6 Arrivals and Departure Throughput at different Spacing Constraints The delays statistics are complementary of the HVO operations and today’s procedural throughput in that at 3 nm of spacing, arrivals separation. The study showed that HVO experience the least delays and departures the operations can increase dramatically the highest while at 10 nm the opposite happens, capacity of non-tower, no-radar airports during arrivals experience the highest delays and periods of IMC and in turn potentially increase departures the least ones. the overall capacity of the NAS. The impact of the SATS developed 5 Conclusion concepts and technologies is reflected in the This paper presented the results of three Monte interest generated in the aviation community Carlo studies of throughput and delay of SATS and some policy making organizations such as 7 the Joint Planning and Development Office [6] Munoz, C., Dowek, G., and Carreno, V., “Modeling (JPDO) that recognized SATS in a recent and Verification of an Air Traffic Concept of Operations”, Software Engineering Notes, Proc. report on the plans for the Next Generation Air International Symposium on Software Testing and Transportation System (NGATS). In particular, Analysis, Boston, Massachusetts, 2004 the 2005 NGATS Integrated Plan [5], mentions [7] Carreno, V., Munoz, C., “Safety Verification of the the SATS demonstration in June 2005 as a Small Aircraft Transportation System Concept of highly successful event and a very important Operations”, AIAA Paper 2005-7423; Proc. 5th milestone achieved by NASA in cooperation Aviation Technology, Integration, and Operations Conference, Arlington, VA, September 2005. with the FAA and industry partners. “A whole [8] Williams D., Consiglio M., Murdoch J., and Adams new generation of safe and affordable small C., “Preliminary validation of the Small Aircraft new aircraft will take advantage of the SATS Transportation Systems Higher Volume Operations enabling technologies and start delivering (SATS HVO) Concept”, Proc. 24th International service where there is little or none before, Congress of the Aeronautical Sciences, Yokohama, Japan, August 2004. thereby taking the pressure off busy airports [9] Consiglio, M., Munoz, C., and Carreno, V., while conveying other economic and quality of “Conflict Detection and Alerting in a Self- life benefits to literally thousands of smaller Controlled Terminal Area”, Proc. 24th International communities”. The report goes on to say that Congress of the Aeronautical Sciences, ICAS 2004, the assessment of future system performance Yokohama, Japan, 2004 allows for both hub-and-spoke operations and a [10] Consiglio, M., Williams, D., Murdoch J., and shift to the use of smaller regional airports for Adams, C., “SATS HVO Normal Operations Concept Validation Simulation Experiment”, AIAA point-to-point operations. No matter what the 2005-7314, Proc. Aviation Technology Integration expected demand growth, results indicate that and Operations Conference, Arlington, Virginia, the baseline system will not provide enough September 2005. capacity to accommodate the levels and types [11] Magyarits S., Racine, N., Hadley, J., “Air Traffic of demand in future years. Control Feasibility Assessment of Small Aircraft Transportation System (SATS) Higher Volume Operations (HVO),” DOR/FAA/CT-05/26, May 6 References 2005. [1] Abbott, T., Jones K., Consiglio, M., Williams, D., [12] Consiglio, M., Conway, S., Adams, C., and Syed, and Adams, C., “Small Aircraft Transportation H.,”SATS HVO Procedures for Priority Landings System, Higher Volume Operations Concept: and Mixed VFR/IFR Operations at Non-Towered, Normal Operations”, NASA/TM-2004-213022, Non-Radar Airports “, Proc. 24th Digital avionics 2004 Systems Conference, Arlington, Virginia, October, 2005. [2] Abbott, T., Consiglio, M., Baxley, B., Williams, D., and Conway, S., “Small Aircraft Transportation [13] Williams, D., Murdoch, J., Adams C., Consiglio, System, Higher Volume Operations Concept: Off- M. Sturdy J. and Peplow K., “The Small Aircraft Nominal Operations”, NASA/TM-2005-213914; Transportation System Higher Volume Operations September 2005. Flight Experiment using NASA’s Cirrus SR22”. Proc. Aviation Technology Integration and [3] Baxley, B., Williams, D., Consiglio M., Adams, C., Operations Conference, Arlington, Virginia, and Abbott, T., “SATS HVO Concept of September 2005. Operations”, AIAA-2005-7379, Proc. AIAA Aviation Technology Integration and Operations, [14] Adams, C., Consiglio, M., Conway, S., and Syed, Arlington VA., September, 2005. H., “The Pilot Advisor: Assessing the Need for a Procedural Advisory Tool”, Proc. 24th Digital [4] Baxley, B., Williams, D., Consiglio M., Adams, C., avionics Systems Conference, Arlington, Virginia, “SATS HVO Off-Nominal Operations”, AIAA- October, 2005. 2005-7461, Proc. AIAA Aviation Technology Integration and Operations, Arlington VA [15] Consiglio, M., Munos, C., Carreno, V., and September, 2005. Williams, D., “Conflict Prevention and Separation Assurance in the SATS HVO Concept of [5] Next Generation Air Transportation System Operations”, AIAA-2005-7463, Proc. AIAA Integrated Plan”, 2005 Progress Report, Aviation Technology Integration and Operations, http://www.jpdo.aero/site_content/index.html Arlington, Virginia, September 2005. 8