1 of 25

Testing Applicability of Point Merge Systems for Göteborg

Landvetter Airport

Henrik Hardell, Tatiana Polishchuk, Lucie Smetanová

The 12^th OpenSky Symposium, 8.11.2024

2 of 25

The Structure

WHY: TMAs often experience inefficiencies
WHAT: Goal is to identify the areas for improvement
HOW: Separate analysis of arrival traffic flows

Part 1

Part 2

WHY: Suggest how to redesign arrival procedures at Göteborg Landvetter
WHAT: Test whether Point Merge procedures improve performance
HOW: Point Merge simulation on Göteborg Landvetter airport’s case

3 of 25

Point Merge

Source: Y. Hong, B. Choi, K. Lee, and Y. Kim, “Dynamic robust sequencing and scheduling under uncertainty for the point merge system in terminal airspace,” IEEE Transactions on Intelligent Transportation Systems

Arrival procedure implemented by Eurocontrol EEC in 2006
Developed to reduce delays on busy airports
Aims to simplify controller tasks by reducing communication and workload
Pre-defined sequencing legs for path stretching

4 of 25

The Airport

Göteborg Landvetter airport
Second largest in Sweden
35.000-40.000 movements annually
Mid-sized airport in European measures
Shared runway for arrivals and departures
Planned reconstruction of the procedures

Current annual capacity 6 mil. passengers
Planned capacity up to 10-15 mil. by 2040

Source: the traffic library

5 of 25

The Data

Historical flight data from the OpenSky database
Arrival data to Göteborg Landvetter airport
Year 2019 (the busiest so far)
Data divided into clusters based on the traffic flows

Six clusters divided by the TMA entry points

Data for part 1:

- The busiest month: April

Data for part 2:

- The busiest day of the month: 10^th of April 2019

6 of 25

Evaluation Framework 1/2

Focus on fair and comprehensive evaluation of the TMA performance
Vertical Efficiency:

Time Flown Level
Vertical Deviation from the Reference Profile

Horizontal Efficiency:

Horizontal Spread
Distance Flown in TMA

Environmental Efficiency:

Fuel Consumption

7 of 25

Evaluation Framework 2/2

Efficiency of Metering and Spacing:

Minimum Time to Final
Spacing Deviation
Throughput
Sequencing Effort

Time Efficiency:

Time in TMA
Additional Time
ASMA Additional Time

PM-specific metrics:

PM usage
PM utilization

Source: author’s picture

We evaluate Metering and Spacing applying the following selected metrics: Minimum Time to Final - heatmap of the minimum possible travel times from the current space in TMA to the final, Spacing Deviation - information about the accuracy of spacing between aircraft, Throughput - the number of aircraft sharing the same minimum time to final value within a given time window, Sequencing Effort - intensity of the control action applied to organize the aircraft in the desired order for landing. We also use two PM-specific descriptive metric; the PM Utilization which characterizes what part, in percent, of the PM arcs are used by the aircraft. Additionally, we calculate the Time efficiency using the Time in TMA metric, which indicates the duration aircraft spend in the TMA before landing. This is complemented by the Additional Time metric, which measures the time each aircraft spends in TMA beyond the Minimum Time to Final values. Furthermore, we calculate the additional ASMA time based on the documentation from Eurocontrol [18] for comparison, but due to lack of data, we adjusted the methodology of

ASMA reference time calculation taking into consideration the data for the whole year 2019 instead of the recommended month-11 data.

8 of 25

Part 1: Per Cluster Analysis

Average Horizontal Spread 13% (36% for Cluster 2)
Maximum values of Min. Time to Final range between 603 (c6) and 1378 (c2) seconds

9 of 25

Part 1: Per Cluster Analysis

Similar shape of Spacing Deviation Evolution curves expect Cluster 2
Low dispersion around 70 seconds to final 🡪 successful sequencing
Sequencing Effort shapes highlight the peaks of the SD values
Highest control value for Cluster 2 was 183 while the average among other clusters is around 120 🡪 Cluster 2 requires significantly more attention

10 of 25

Part 1: Per Cluster Analysis

High dispersion of Time on Levels values for Cluster 2
High Time on Levels values for Cluster 4 despite the number of arrivals 🡪 vertical inefficiency
ASMA Reference transit time computed for the whole year 2019 (only for three clusters)
Additional time and ASMA Additional time provide similar results despite different calculation methodology
Arrivals from Cluster 2 spent more Time in TMA because of the asymmetrical TMA shape

11 of 25

Part 2: Testing PM Applicability

Evaluation of a proposed PM design used within an ongoing airspace project�
Utilization of our previously developed optimization framework [1]

Applied to a real, busy-day scenario, based on OpenSky Network data

[1] H. Hardell, T. Polishchuk, and L. Smetanová. “Automated Traffic Scheduling in TMA with Point Merge to Enable Greener Descents”. In: ICRAT, Tampa, June 19-23, 2022.

12 of 25

PM Design

Three entry points per arc�
Arc entry point selected based on location of TMA entry�
Descent allowed along the arcs�
Sequencing legs discretized

13 of 25

Acquisition from OpenSky Data

Arrivals

Aircraft type
TMA entry time
TMA entry coordinates
TMA entry speed (GS -> CAS)
Cruise altitude

Departures

Aircraft type
Timestamp of first recording

14 of 25

Generation of Arrival Routes

E6/W6: 16 route options

E3/W3: 7 route options

E1/W1: 1 route option (can also use E3/W3)

15 of 25

Vertical Profile

EUROCONTROL BADA v4.2
Wind and temperature from ECMWF ERA5
CDO at idle thrust
Total Energy Model (TEM)
CAS speed profile:

TMA entry: speed identical to actual flight
<FL100: 250 kt
PM arcs: 220 kt
IF: 200 kt

16 of 25

Separation Requirements

11.25 NM

Time separation between two arrivals [2]:

69 s between M and M/H
157 s between H and M

Estimation of flight time between IF �and runway threshold
Time separation between arrival/departure [2]:

60 s

[2] Richard de. Neufville and Amedeo R. Odoni. Airport Systems. Planning, Design, and Management. Mc Graw Hill, 2013

17 of 25

Optimization Framework

Based on mixed-integer programming (MIP)
Provides optimal assignment of arrival profiles to each aircraft

Identifying conflicting arrival profiles

Optimization framework

(min fuel)

Conflicting profiles may not be selected simultaneously

Assign to each aircraft one profile

Provide necessary spacing between arrivals and departures

Optimal assignment of arrival profiles

Constraints

18 of 25

Dataset

April 10, 2019
90 arrivals (84 M, 6 H)
95 departures (91 M, 4 H)
988 arrival profile options
1407 arrival profile pairs conflicts
Takeoff time estimated based �on location of first timestamp

19 of 25

Optimized Arrival Schedule

20 of 25

Optimized Routes

21 of 25

Time – Distance - Fuel

5% fuel saved on average per flight
Time and distance increased by ~10%
All clusters had an increase in time and distance
Large variation in number of flights – difficult to compare clusters

22 of 25

Vertical Performance

Median vertical deviation 4450 ft×min
Ranges between 2190 and 7660 ft×min

23 of 25

Sequencing and Spacing

Similar max of min. time to final
Average min. time to final�increased for optimized scenario
Spacing deviation very disperse at �900 s to final
Aircraft in optimized scenario �organized for landing earlier
Throughput results similar – �matches with optimized �spacing deviation at ~300 s

24 of 25

Conclusions

Part 1

Assessment of cluster-wise flight performance at Göteborg-Landvetter airport in April 2019
Performance variations when comparing the six clusters

Part 2

PM design evaluated with an optimization framework
Small decrease in fuel consumption – increase in time and distance
PM used by <10% of the flights – PM systems size could be reduced
More efficient flow organization – flights organized earlier

25 of 25

Thank you for listening!�Any questions?

lucie.smetanova@liu.se

tatiana.polishchuk@liu.se

henrik.hardell@liu.se