The Area Flow Multi-Sector Planner: A Fast-Time Study of MSP Coordination Activities. Kenny Martin ISA Software

Transcription

1 The Area Flow Multi-Sector Planner: A Fast-Time Study of MSP Coordination Activities Kenny Martin ISA Software

2 Topics MSP Concepts Modeling Approach MSP Roles & Responsibilities Modeling Platform MSP Results

3 MSP Concepts Follow-up of RTS 1 of two MSP concepts: Multi-D MSP MSP manages flows and provides medium-term conflict resolutions within the multi-sector area (MSA) (8-15 minutes in advance of conflict) Coordinates with adjacent sectors and with aircraft (after executive coordination) Area Flow MSP MSP plans reroutes to avoid congestion or weather Lookahead minutes in advance of MSA entry Coordinates with own executives and adjacent MSPs Does not contact aircraft Concept may be extended to include MSP contact with aircraft via datalink 1 Kevin Corker, SJSU 2006

4 RTS Results Based on RTS results and analysis of other research efforts, FAA determined: The MSP acting in an Area Flow Manager role showed the most promise. RTS participants felt Area Flow was more efficient (solve more problems further in advance) Multi-D could become unmanageable with high traffic loads potential for loss of situational awareness Next analysis phase should consider coordination activities for multiple MSPs over a wider area.

5 Modeling Approach 1: Airspace Busy 3-center area selected for study: Fort Worth, Kansas City and Memphis centers 150 sectors, long inter-center boundaries, complex mix of traffic Dallas-Fort Worth, St. Louis and Memphis airports Atlanta, Chicago, Houston, Denver adjacent

6 Modeling Approach 2: Multi-Sector Areas 50 MSAs were chosen, based on analysis of traffic flows High airspace (FL240-FL340) Superhigh airspace (FL240-FL340) Low altitude MSAs not shown. Major TMAs were excluded from MSAs.

7 Modeling Approach 3: Traffic Traffic Samples Baseline 15,500 flights (2007 peak day) Plus20% 18,700 flights (2013 forecast) Plus40% 21,800 flights (2018 forecast) 6 Scenarios 3 traffic samples Executive only Executive + MSP

8 CHILL Modeling Platform Chill is a multi-agent, distributed modeling platform Supports RTS and FTS agents Provides key systems and services Simulates the entire NAS in a few minutes AOC DCB MIL MSP ATC APT MONACO

9 MSP Modeling Platform Setup RAMS Plus: 150 executive controllers 4 Major airports in region (DFM, DAL, STL, MEM) CHILL-MSP: 50 MSPs SIM-C (SWIM) SENS for service discovery / message exchange

10 Automated Reroutes CHILL-MSP provides 3 types of reroutes: Local Playbook reroutes use preplanned reroutes to mimic local expertise and standard rerouting strategies. Geometric reroutes use a bounding-box algorithm to avoid the sector entirely. Vertical reroutes use level-capping or early descent.

11 Capacity Metric FAA Monitor Alert Parameter (MAP) used to predict overloads Peak instantaneous aircraft count, 15 minute period Each sector has individual threshold value determined by flow management CHILL-MSP looks for 2 minute periods exceeding the MAP Other metric(s) can be used Workload, complexity,.

12 MSP Roles and Responsibilities MSP is assigned to an MSA (usually 3 sectors) MSP has knowledge of traffic 45 minutes in advance If overload predicted, the MSP: Finds flight(s) in sector during overload Attempts to find suitable reroutes to avoid overload If successful, must coordinate with impacted MSPs Upstream MSP, if reroute begins in another MSA Downstream MSPs, if reroute enters a sector not previously in flight plan, within 40 minutes of start of reroute Other MSPs will always accept reroute unless: Reroute creates or worsens an overload in other MSA Other MSP is too busy Executive controller action (conflict resolution) always cancels pending MSP trial plans

13 Example of MSP Impact on a Sector NFlights In Sector :45 13:15 13:45 14:15 14:45 15:15 15:45 16:15 ZKC30 Timeline Plus40 MSP/NoMSP Scenarios 16:45 17:15 17:45 18:15 18:45 19:15 19:45 20:15 20:45 21:15 21:45 22:15 22:45 23:15 23:45 0:15 Traffic balance improved with MSP: Stdev of peak % of MAP reduced by 50%. NoMSP MSP MAP (=19) Change in Peak Flights :45 13:15 13:45 14:15 14:45 15:15 Change in Peak Number of Flights Per 15 Minute Period ZKC30, Plus40 NoMSP to MSP Scenarios Change as % of MAP Change as % of MAP Change as % of MAP 15:45 16:15 16:45 17:15 17:45 18:15 18:45 19:15 19:45 20:15 20:45 21:15 21:45 22:15 22:45 23:15 23:45 0:15 45% 35% 25% 15% 5% -5% -15% -25% -35% -45%

14 Overall MSP Results Minutes Above the MAP All Centers Combined Minute Periods Minute Periods ETMSInitial Baseline Plus20 Plus ETMSInitial Baseline Plus20 Plus40 Tactical No MSP Only With MSP MSP 321% 234% Increase in minutes above MAP with no MSP

15 Changes in 15-minute Peak Traffic Count Plus40 NoMSP Plus40 MSP S E C T O R S 15-Minute Periods (Peak 12 Hours) 25% of all 15-minute periods reduced peak % of MAP with MSP (average reduction 16.8%). 26% increased peak %MAP (average increase 9.6%). The average benefit of MSP activities (reduced %MAP) was greater than the average penalty to nearby sectors (those accepting excess traffic). % of MAP % 60-80% 40-60% 20-40% 0-20% % % % % % %

16 Trial Plan Coordinations Required Half of the TPs with 0 coordinations were accepted by ghost MSPs outside the 3 center area. If MSPs were active everywhere, these would require at least 1 coordination. Our coordination rules not operationally realistic (coord up to 40 minutes ahead). Research needed to refine MSP roles and coordination rules. 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% Trial Plans Requiring Coordination with Other MSPs 30% 22% All MSP Scenarios 5% 7% 8% 21% 21% 23% 33% 32% 18% 16% 23% 21% 21% BaselineMSP Plus20MSP Plus40MSP 4 or more MSPs 3 MSPs 2 MSPs 1 MSP 0 (No coord)

17 Reasons for Trial Plan Rejection Sector within coordinating MSA is predicted to be above the MAP Time-outs 4%) Coordinating MSP too busy Tactical update occurs during planning/coordination (conflict resolution, ). Executive actions always cancel pending MSP trial plans.

18 Trial Plans Rejected: Sector Overloaded Trial Plan Rejections due to Busy Sector 80% % of Rejections 70% 60% 50% 40% 30% 20% 10% 0% Up to 110% 120% 130% 140% 150% 160% + % Over MAP Baseline Plus20MSP Plus40MSP

19 Time to Busy Sector Rejections Half the TP rejections were due to predicted overload more than 20 minutes in advance Time to Overloaded Sector Entry TPs Rejected due to Sector Over MAP Plus40MSP Scenario Number of Rejections Uplink Sector Busy (9.7%) Cumulative % 100% 80% 60% 40% % % Minutes Before Hot Sector Entry

20 Trial Plans Rejected: MSP Busy No MSP workload metric available yet We used: N Flts to Reroute (all predicted overloads) N Pending Trial Plans (TPs must be monitored with retry if coord rejected) 3 of our MSAs were too busy: consider sector traffic loads & peak periods when designing MSAs Distribution of "MSPBusy" Measure at Trial Plan Rejection Plus40MSP Scenario No. TPs Rejected due to MSPBusy MSAs 3 MSAs MSPBusy: NFltsToReroute + NPendingTPs

21 Trial Plan Acceptance/Rejection TP Acceptance dropped: Baseline: 88.9% Plus20: 83.3% Plus40: 70.5% Primary cause: more Busy Sector and Busy MSP rejections 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% Reasons for Trial Plan Cancellation Rejected: ATC Update Rejected: MSP Busy Rejected: Timed Out Rejected: Sector Busy Accepted 0% BaselineMSP Plus20MSP Plus40MSP

22 Inter-Center Coordination ZFW Percent of Trial Plans Uplinked by Center for each Initiating Center ZKC ZME 17% 1% 0% 4% 6% 2% 17% 18% 3% 18% 8% 8% 51% 17% 63% 1% 5% 1% 1% 5% 7% 49% ZFW ZKC ZME ZAB ZAU ZDV ZHU ZID ZMP ZTL Effective Inter-Center coordination is vital for the MSP concept.

23 Next Steps? Further RTS with results fed back to FTS Refine MSP roles, responsibilities and coordination procedures Consider data-link from MSP to flight deck under certain conditions Additional FTS study incorporating all of NAS and additional refinements from RTS