Study on Traffic Synchronization

Transcription

1 Study on Traffic Synchronization Claus Gwiggner Masato Fujita Yutaka Fukuda Sakae Nagaoka Electronic Navigation Research Institute Tokyo EIWAC, Tokyo, November,

2 Contents 1. Traffic Synchronization 2. Delay propagation 3. Delay absorption 4. Conclusions 2

3 1. Traffic Synchronization Airspace Congestion Gates Tokyo Terminal Area Figure: Inefficient arrival flow to Tokyo Int'l Airport Top of descent Delay Fuel consumption Controller workload En-route sector 3

4 1. Traffic Synchronization Airspace Congestion Gates Speed Speed control control Dynamic Dynamic Queues queues Balance ground / enroute ground and Balance en-route delays delays Tokyo Terminal Area Traffic Synchronization Figure: Inefficient arrival flow to Tokyo Int'l Airport Definition: Tactical establishment and maintenance of a Top safe, of orderly and En-route efficient sector flow of air traffic. descent [ICAO] Benefits: Less delay Less fuel consumption Less controller workload [ICAO 2005, SESAR, NetGen] 4

5 1. Traffic Synchronization Concept of Operations Controlled time of arrival (CTA) [ i ] [ i+1 ] CTA i CTA i 1 Target precision: +/- 10 sec Traffic Window Window size: [cta-, cta+] sec Current precision: +/- 30 sec Sector Traffic Window Window position: between sectors on waypoints on merge-points (...) Flight Trials: CTA/ATC system integration studies (CASSIS) Simulation studies: Contract-based Air Transportation System project (CATS) 5 [CASSIS 09, CATS 08]

6 1. Traffic Synchronization Concept of Operations Controlled time of arrival (CTA) [ i ] [ i+1 ] CTA i CTA i 1 Open questions Target precision: +/- 10 sec Traffic Window Feasibility? Current precision: +/- Number 30 sec of constraints? Impact of uncertainty Sector Traffic Window Window size: [cta-, cta+] sec Window position: between sectors on waypoints on merge-points (...) Flight Trials: CTA/ATC system integration studies (CASSIS) Simulation studies: Contract-based Air Transportation System project (CATS) 6 [CASSIS 09, CATS 08]

7 2. Delay Propagation Delay Propagation under uncertainties m i 1 eta D eta C Follower Leader Heavy Mid/Small Heavy Mid/Small =ma a i m i,0 A i r c r a f t e t a s t a d e l a y A 1 2 : : B 1 2 : : D 1 2 : : C 1 2 : : Queueing Delays [Erzberger 95, Bayen 06, Balakrishnan 09] 7

8 2. Delay Propagation Delay Propagation under uncertainties h High altitude (fuel efficient) 1 Top of descent (TOD) Optimal descent :queueing delay of aircraft i :absorption coefficient 0 1 c h,c l : costs c h c l, [min 1 ] Low altitude (fuel inefficient) gate Profile plot m i 1 eta D eta C Follower Leader Heavy Mid/Small Heavy Mid/Small =ma a i m i,0 A i r c r a f t e t a s t a d e l a y A 1 2 : : B 1 2 : : D 1 2 : : C 1 2 : : Queueing Delays [Erzberger 95, Bayen 06, Balakrishnan 09] 8

9 2. Delay Propagation Delay Propagation under uncertainties h High altitude (fuel efficient) 1 Delayed arrival Top of descent (TOD) Optimal descent :queueing delay of aircraft i :absorption coefficient 0 1 c h,c l : costs c h c l, [min 1 ] Remaining delay on low altitude Actual time of arrival high altitude cost: c h low altitude ata cost: c l Scheduled time of arrival sta t Low altitude (fuel inefficient) gate 1 c i =[ 1 c h c l ] Delay absorption cost Profile plot Time plot m i 1 eta D eta C Follower Leader Heavy Mid/Small Heavy Mid/Small =ma a i m i,0 A i r c r a f t e t a s t a d e l a y A 1 2 : : B 1 2 : : D 1 2 : : C 1 2 : : Queueing Delays [Erzberger 95, Bayen 06, Balakrishnan 09] 9

10 2. Delay Propagation Delay Propagation under uncertainties h High altitude (fuel efficient) 1 Delayed arrival Top of descent (TOD) Optimal descent :queueing delay of aircraft i :absorption coefficient 0 1 c h,c l : costs c h c l, [min 1 ] Remaining delay on low altitude Actual time of arrival high altitude cost: c h low altitude ata cost: c l Scheduled time of arrival sta t Low altitude (fuel inefficient) Profile plot gate 1 c i =[ 1 c h c l ] Time plot Delay absorption cost m i a i sta g m i 1 (scheduled) t eta D eta C Follower Leader Heavy Mid/Small Heavy Mid/Small =ma a i m i,0 A i r c r a f t e t a s t a d e l a y A 1 2 : : B 1 2 : : D 1 2 : : C 1 2 : : Queueing Delays [Erzberger 95, Bayen 06, Balakrishnan 09] 10

11 2. Delay Propagation Delay Propagation under uncertainties h High altitude (fuel efficient) 1 Delayed arrival Top of descent (TOD) Optimal descent :queueing delay of aircraft i :absorption coefficient 0 1 c h,c l : costs c h c l, [min 1 ] Remaining delay on low altitude Actual time of arrival high altitude cost: c h low altitude ata cost: c l Scheduled time of arrival sta t Low altitude (fuel inefficient) Profile plot gate 1 c i =[ 1 c h c l ] Time plot Delay absorption cost m i a i sta g m i 1 (scheduled) t eta D eta C Follower Leader Heavy Mid/Small Heavy Mid/Small =ma a i m i,0 A i r c r a f t e t a s t a d e l a y A 1 2 : : B 1 2 : : D 1 2 : : C 1 2 : : ata g (actual) i Queueing Delays [Erzberger 95, Bayen 06, Balakrishnan 09] 11

12 2. Delay Propagation Delay Propagation under uncertainties h High altitude (fuel efficient) 1 Delayed arrival Top of descent (TOD) Optimal descent :queueing delay of aircraft i :absorption coefficient 0 1 c h,c l : costs c h c l, [min 1 ] Remaining delay on low altitude Actual time of arrival high altitude cost: c h low altitude ata cost: c l Scheduled time of arrival sta t Low altitude (fuel inefficient) Profile plot gate 1 c i =[ 1 c h c l ] Time plot Delay absorption cost m i a i sta g m i 1 (scheduled) t eta D eta C Follower Leader Heavy Mid/Small Heavy Mid/Small =ma a i m i,0 A i r c r a f t e t a s t a d e l a y A 1 2 : : B 1 2 : : D 1 2 : : C 1 2 : : Queueing Delays [Erzberger 95, Bayen 06, Balakrishnan 09] ata g (actual) sta g ' re-scheduled i i a i m i a i i i Delay propagation 12

13 2. Delay Propagation Delay Propagation under uncertainties h High altitude (fuel efficient) 1 Delayed arrival Low altitude (fuel inefficient) Profile plot Top of descent (TOD) :queueing delay of aircraft i :absorption coefficient 0 1 c h,c l : costs c h c l, [min 1 ] Actual time of arrival high altitude Optimal descent cost: c Remaining delay h Traffic Synchronization Problem low altitude queueing delays and trajectory prediction errors i. low altitude gate Given : 1 c i =[ 1 c h c l ]d Flow of pre-scheduled aircraft sta i 1 sta 2... sta n with Time plot ata cost: c l Scheduled time of arrival sta Delay absorption cost t m i 1 sta g Find : Optimal strategy. (scheduled) m i a i t eta D eta C Follower Leader Heavy Mid/Small Heavy Mid/Small =ma a i m i,0 A i r c r a f t e t a s t a d e l a y A 1 2 : : B 1 2 : : D 1 2 : : C 1 2 : : Queueing Delays [Erzberger 95, Bayen 06, Balakrishnan 09] ata g (actual) sta g ' re-scheduled i i a i m i a i i i Delay propagation 13

14 2. Delay Propagation Main Results 1 Analytical model Flown distance with reduced speed Average propagated delay (normalized) Monte Carlo simulation j w i l j0 i0 s e lv i k i v i with j : speed control delay for aircraft j w i : metering delay for aircraft i = j-1 E D i = k=0 with u/ k 1 u v P k u, v f u g v dvdu u=0 v=0 f, g:probability density function of, d P :distribution of length of propagation 14 [Gwiggner et al. 09, 10]

15 2. Delay Propagation Main Results In words: 1 Analytical model Speed control itself causes no delay propagation. Average propagated delay (normalized) The reason for delay propagation: Flown distance with trajectory reduced speed prediction errors Monte Carlo simulation j w i l j0 i0 s e lv i k i v i with j : speed control delay for aircraft j w i : metering delay for aircraft i = j-1 E D i = k=0 with u/ k 1 u v P k u, v f u g v dvdu u=0 v=0 f, g:probability density function of, d P :distribution of length of propagation 15 [Gwiggner et al. 09, 10]

16 3. Delay absorption Delay absorption strategy Delay absorption strategy Trade-off total cost 全体コスト additional delay 追加遅延時間 baseline cost 燃料コスト High altitude High propagated delay Fuel efficiency Workload sharing Low altitude High fuel consumption High altitude only 高高度のみ Fraction of metering delay absorben low altitude 低高度での滞留時間の割合 c =[c L c H 1 ]d d =d 0 d p, Low altitude only 低高度のみ average delay + propagation where c L c H :cost of delay absorbtion (kg/min) d 0 :average queueing delay d p : propagated delay Consequences: Even in the future, there is a need for radar vectoring. Sequencing strategies under uncertainties should be studied. 16

17 4. Conclusions Conclusions Traffic Synchronization Tactical management of queues of aircraft Delay Propagation Delay propagation due to trajectory prediction errors Delay absorption strategy Trade-off between high altitude (fuel efficient) and low altitude (fuel inefficient) Even when the objective is to minimize fuel (!) 17

18 4. Conclusions Future work Fundamental Research Conditions for eistence of minimum Delay propagation in transportation networks Operational Concept Ground delay vs. en-route delays Delay management strategies 18

19 Thank you. 19

20 Traffic Synchronization Tactical phase Global ATM Concept (ICAO) Demand/Capacity Balancing 需要 / 容量均衡化 Traffic Synchronization Departure Arrival Interaction Conflict Management コンフリクト管理 Traffic Management Coordinator (TMC) 航空交通管理管制官 Air Traffic Controller (ATCo) 航空管制官 [ICAO 2005, ATM Masterplan] 20

21 Aircraft sequencing C B A m i 1 D eta D eta C 1 =ma a i m i,0 Basic Operations Sequencing First-come-first-served (FCSF) Constrained position shifting (CPS) Metering Flow control with separation constraints m i m i m i m i 1 Queueing Delay A i r c r a f t e t a s t a d e l a y A 1 2 : :0 1 0 B 1 2 : :0 3 1 D 1 2 : :0 5 1 C 1 2 : :0 7 2 FCFS delay: 4 min CPS delay: 3.2 min Follower Leader Heavy Mid/Small Heavy Mid/Small [Erzberger 95, Bayen 06, Balakrishnan 09]

22 2. Delay propagation Delay propagation Condition for delay propagation: i i a i 1... k i a i j=1 a i j k 1 (1) Delay triggered by aircraft i: D p,i =[k i a i k 1 a i 1... a i k ] k =k i k j a i j j=0 0 k j 1 j=0 k where k is smallest number, such that (1) is smaller than d k k j 1 j=0 (2) (3) Propagation approimation (high-congestion): D p,i { n i i, i 0, else (4) = d d Epectation: E D p,i = n i u=0 with u/ v=0 f, g: probability density function of, d u v f u g v dvdu (5) 22