Benchmarking Airport Airside Performance: FRA vs. EWR Amedeo Odoni, Thomas Morisset, MIT Alexander Zock, Wilhelm Drotleff, ECAD ATM Research & Development Seminar Berlin, June 14, 2011
Compare airside performance characterisjcs: European vs. US Airports Part of FAA EUROCONTROL inijajve. MIT effort funded by FAA through NEXTOR Stage 1: 34 top airports in Europe vs. 34 top in US Stage 2: Deep dive at selected airports (FRA vs. EWR) Impacts of: (i) use of VMC procedures, weather permi^ng, at US airports (ii) limited or, mostly, non- use of declared capacity limits at US airports on the performance of US vs. European airports w.r.t. capacity, delays and reliability of flight schedules 2
Based on combinajon of data analysis (primarily from 2007) and modeling (throughputs, and delays) Databases: US: ASPM (34 busiest airports, complete data) EUR: CODA (34 busiest airports, parjal data collected from airlines on a voluntary basis) Germany: DFS, Fraport, LH, German Airport Coordinator (GAC) Models: MACAD, MIT DELAYS, MIT AND (Airport Network Delays) [not in this talk] 3
FRA vs. EWR: Main Characteris>cs US & GERMAN AIP 2007 Layout FRA & EWR Compara>ve overview: FRA EWR Number of pax 54,2 mln 36,4 mln No. of movements pax/ movement 479.874 443.952 113 82 No. Of runways 3 3 Cargo volume (t) 2,2 mln. 0,9 mln % intern. pax 85% 29% Carrier Lulhansa ConJnental ACI World Traffic Report 2007 Both airports are considered congested Number of runways and configuration are similar Number of movements are comparable Both airports serve as hubs for a network airline 4
FRA vs. EWR - airport vicinity EWR and its surrounding airports FRA and its surrounding airports EWR is embedded in a metroplex environment wherein the airspace is shared with six other airports two of which are major international airports within 25 nm of EWR FRA has almost no constraints deriving from other airports in its neighborhood 5
In US: Opera>ng Policies (somewhat simplified) In Visual Meteorological CondiJons VMC) use Visual Flight Rules (VFR); this may increase throughput significantly compared to use of Instrument Flight Rules (IFR) which are used in Instrument Meteorological CondiJons (IMC) With very few excepjons, no caps on number of movements that can be scheduled Almost everywhere else (including Europe): Operate (officially) under IFR all the Jme Set the declared capacity (=number of slots) at or below the IFR capacity of the airport (but LHR, LGW, FRA, MUC, other?) 6
Some Consequences US airports operate with VFR for about 80% of the Jme on average As a result, they increase their average capacity (weighted by weather condijons) by about 25% compared to their IFR capacity Given extensive use of VFR, US airlines typically schedule airport operajons with reference to the VMC capacity of airports Even when caps on schedules have been set by the FAA, these caps are close to the VFR capacijes of the respecjve airports In Europe, the number of scheduled movements typically does not exceed the IFR capacity of the airport 7
EWR vs FRA: Scheduling Limita>ons Newark (since 2008) Frankfurt (Summers 2010 and 2011) Hourly*: 81-84 total; 41-44 arrs; 43-50 deps 30-min: 43 total; 23 arrs; 25 deps Hourly: 81 scheduled movements + allowance for 2 unscheduled movts Slots specified within 30-minute time windows (6:00 6:29, 6:30 6:59, etc.) 10-min: 16 total; 9 arrs; 9 deps * Hourly limits vary by time of day depending on arrival departure mix, widebody operations Slots specified within 5-minute time windows EWR schedule limitations administered by FAA Slot Coordination Office; initiated 5/15/2008, extended to expiration date of 10/29/2011 FRA schedule limitations administered by German Airport Coordinator (GAC); 15 full-time staff, coordinates 16 airports with total of about 2 million annual movements; no expiration dates for schedule limitations
FRA Average daily schedule by month (2007) Daily demand profile movements / hour 90 80 70 60 50 40 30 20 10 0 05 06 07 08 09 10 11 12 13 14 15 16 17 18 19 20 21 22 23 time January February March April Mai June July August September October November December Evenly distributed demand profile from 07:00 to 21:00 Hourly demand peaks at 84-movement hourly slot limit 6/3/11 FAA - FRA airport operajon analysis 9
FRA Max throughput es>mates (95%- ile) (movements per 15 minutes) Hourly capacity: 84 96 movements 10
FRA Average daily schedule by month (2007) Daily demand profile movements / hour 90 80 70 60 50 40 30 20 10 0 Scheduling limit VMC Capacity IMC Capacity 05 06 07 08 09 10 11 12 13 14 15 16 17 18 19 20 21 22 23 time January February March April Mai June July August September October November December Evenly distributed demand profile from 07:00 to 21:00 Hourly demand peaks at 84-movement hourly slot limit 6/3/11 FAA - FRA airport operajon analysis 11
EWR Average daily schedule by month (2007) Daily demand profile Unevenly distributed demand profile with extended afternoon/ evening peak Average hourly demand peaks at about 90 movements 6/3/11 FAA - FRA airport operajon analysis 12
EWR Max throughput es>mates (95%- ile) (movements per 15 minutes) Hourly capacity: 72 84 movements 13
EWR Average daily schedule by month (2007) Daily demand profile VMC Capacity IMC Capacity Unevenly distributed demand profile with extended afternoon/ evening peak Average hourly demand peaks at about 90 movements 6/3/11 FAA - FRA airport operajon analysis 14
Due to: (i) intensive scheduling (close or above VFR capacity) (ii) variability of the operajng capacity with weather, delays (relajve to schedule) at EWR are high, in general, strongly variable with weather Increasing over the course of the day and very high in IMC and during the peak evening hours 15
FRA - all runway configura>ons: schedule and delays Scheduled vs. actual movements vs. delays all weather (2007) 90 90 80 80 70 70 movements/hour 60 50 40 30 60 50 40 30 Minutes 20 20 10 10 0 05 06 07 08 09 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 hour clock ARR delay min DEP and TAXI out delay min Scheduled movements actual movements 0 ARR delays and DEP delays up to +300min (including punctual movements with a delay of 0) are taken into calculation. Average calculation includes hours before 5am. Those are simply not displayed. Taxi out delays are not in accordance to the standard definition of Eurocontrols PRU, due to lack of data. 6/3/11 FAA - FRA airport operajon analysis 16
EWR - all runway configura>ons: schedule and delays Scheduled vs. actual movements vs. delays all weather (2007) 90 90 80 80 70 70 movements/hour 60 50 40 30 60 50 40 30 Minutes 20 20 10 10 0 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 hour clock sched ARR delay min DEP and TAXI out delay min Series3 actual movements movts 0 ARR delays and DEP delays up to +300min (including punctual movements with a delay of 0) are taken into calculation. Average calculation includes hours before 5am. Those are simply not displayed. 17
Average Delays (min, vs. schedule)
Reliability/Predictability Due to (i) variability in performance with weather and (ii) airline scheduling pracjces, schedule reliability and predictability is much lower at US airports than at European airports Reliability of schedules in US declines sharply over the course of a day, parjcularly in the presence of poor weather 19
FRA - delays do not worsen over the course of the day; roughly constant schedule reliability ARR delay at FRA 3.5 average delay standard devia>on Frequency (%) 3 2.5 2 1.5 8am- 9am 0,9 min 24,9 min 12pm- 1pm 4,1 min 30,6 min 4pm- 5pm 1,7 min 28,4 min 5pm- 6pm 2,3 min 35,5 min 1 0.5 0-100 -50 0 50 100 150 ARR delay minutes 6/3/11 FAA - FRA airport operajon analysis 20
EWR - delays worsen during the course of the day; schedule reliability declines ARR delay at EWR 5.0% average delay standard devia>on 4.0% 8am- 9am - 0.2 min 29.0 min 12pm- 1pm 8.1 min 38.1 min frequency 3.0% 2.0% 4pm- 5pm 30.1 min 55.0 min 1.0% 0.0% -60-40 -20 0 20 40 60 80 100 120 140 160 180 arrival delays (1-minute intervals) 6/3/11 FAA - FRA airport operajon analysis 21
JFK delays worsen during the course of the day; schedule reliability declines ARR delay at EWR 5.0% average delay standard devia>on 4.0% 8am- 9am 7.6 min 46.6 min 12pm- 1pm 10.8 min 49.3 min frequency 3.0% 2.0% 4pm- 5pm 23.2 min 56.9 min 8pm- 9pm 34.7 min 65.4 min 1.0% 0.0% -60-40 -20 0 20 40 60 80 100 120 140 160 180 arrival delays (1-minute intervals) 6/3/11 FAA - FRA airport operajon analysis 22
Delay under IFR is much greater than delay under VFR Delay increases steadily during the day 23
24
But! since 2007 Roughly 6-8% reduction in demand at JFK and EWR between 2007 and 2010 Roughly 35% reduction in departure delay relative to schedule
DOT Inspector General s Report RecommendaJon 1: (10/28/10) We recommend that FAA re- examine flight caps at JFK, LGA, and EWR, basing the caps on more realisjc airport operajng condijons, air carrier scheduling pracjces, and a goal towards reducing delays to an acceptable rate. 28
The Bofom Line Scheduling limits or market- based demand management schemes are desirable at some airports to avoid over- scheduling and extreme condijons But the approach currently ujlized to set scheduling limits and declared capacijes lacks sophisjcajon and may yield very conservajve results, wasjng valuable potenjal capacity and excessively constraining access to some major airports Need a bever approach that considers: the interests of diverse stakeholders full range of tradeoffs between throughput and delay 29
The se^ng of declared capacijes (or of target levels of demand for market- based schemes) involves finding an opjmal tradeoff: throughput vs. delays and predictability/reliability [An issue which, quite surprisingly, has not been addressed in sufficient depth to date!] 30