Implementing ASMA and Taxi Out Additional Times algorithms: a practical experience

Transcription

1 INTRODUCTION: SOME (MODERN) HISTORY... Interest of AENA in having local implementation of Performance Indicators Implementing ASMA and Taxi Out Additional Times algorithms: a practical experience Operational ANS Performance at Airports Workshop Brussels, 19 February 2014 José Manuel Cordero presented by José Miguel de Pablo ASMA Additional Time selected at a initial stage (late 2012) - Taxi Out Additional Time immediately after (early 2013) Fundamental premise during development: maximum adherence to PRU definition and implementation of algorithm - Need of strong coordination with PRU Currently fully deployed locally - Lot of experience gained in the process Extensive analysis of data has allowed to raise some points to improve results 2 Performance at Airports Workshop -19 February Brussels 1

2 THE IMPLEMENTATION PROCESS: The beginning Descriptions of the indicators are available in a wiki ( - Necessary to count on highly reliable description However - Very specific details are obviously not included, and may bring errors in results ASMA and Taxi Out Additional Times at AENA THE IMPLEMENTATION PROCESS - Some of the values used are very local, and depending on ad-hoc analysis (i.e., determining the ASMA entry sectors) - Many questions arise if 100% adherence to PRU is desired! Then contact PRU for support! - Special thanks to Philippe Enaud, Aurelie Nieuborg and Guglielmo Guastalla 3 Performance at Airports Workshop -19 February Brussels 4 Performance at Airports Workshop-19 February 2014-Brussels 2

3 THE IMPLEMENTATION PROCESS: Hands-on For local implementations, it is necessary to count on the necessary data to fill the algorithms (Check!) - AENA/CRIDA has great experience processing real data, and they are available for these indicators - ASMA is calculated with very accurate data (radar tracks), even better than the PRU (CPRs and flight plans) - TXOT depends on AOBT: not automated for every airport, adds some uncertainty. - Fully automation on data generation would be highly desirable for proper implementation Iterative development process - Difficulties for modules calibration as no intermediate values were available; brainstorming with PRU experts New specific details may arise THE IMPLEMENTATION PROCESS: Results validation ResultsmaydifferfromPRUones(eveniftryingtobe fully compliant with algorithm): - Different quality of data - Subtle differences in the implementation - Other reasons? How to calibrate results or explain differences? - Main difficulty, as no access to intermediate results or detailed sample of flights - Debugging needs for indicators reliability - Still not fully solved! Key for the reliability of the Performance Scheme 5 Performance at Airports Workshop-19 February 2014-Brussels 6 Performance at Airports Workshop-19 February 2014-Brussels 3

4 FEEDBACK ABOUT THE WORKFLOW LESSONS LEARNT/ PROPOSAL FOR IMPROVEMENTS IN WORKFLOW Difference in results when calibrating, requiring further investigation (not fully addressed yet) Access to calibration data Increases confidence in the indicators Enables local implementations Eases the debugging process Clear, accurate and reachable definition of the algorithm Updates must be properly distributed Technical forums/working groups to discuss indicators improvement/ evolution Global, per KPA, per KPI - In current scenario, very time-consuming and difficult 7 Performance at Airports Workshop-19 February 2014-Brussels 8 Performance at Airports Workshop-19 February 2014-Brussels 4

5 ABOUT THE ALGORITHMS Local implementation has allowed in-depth understanding of the algorithms This knowledge is essential to use the indicator! Difference in results leads to investigation on results, and to raise some issues related to the algorithms ASMA Additional Time ASMA and Taxi Out Additional Times at AENA ALGORITHMS DISCUSSION 40 NM radius First/Last ASMA Entry Entry sector categorization Taxi Out Additional Time Stand grouping Common Recalculation of unimpeded times 9 Performance at Airports Workshop -19 February Brussels 10 Performance at Airports Workshop-19 February 2014-Brussels 5

6 ASMA Additional time 40 NM radius: In LEMD, the 40NM radius for ASMA is not including holdings, thereby their effect is not considered in this indicator. First/Last entry in ASMA radius: Situation generated when the holding is over the ASMA frontier (several crossing points), as LEMD. - Initially unespecified in the published description of the algorithm which entry was considered AENA considered the first PRU considered the last Entry Sector categorisation: The ASMA entry sectors are obtained by PRU through statistical analysis and it calculation process is neither automated, nor defined - Automation? Changes to be distributed to the involved local actor (or agreed)? 11 Performance at Airports Workshop -19 February Brussels AENA data First ASMA Entry PRU Data: Fourth ASMA Entry 1st Entry 2nd Entry 3rd Entry 4th Entry 21:32:21 21:45:16 21:53:55 22:00:33 AENA Difference of more than 28 minutes PRU Taxi Out Additional time: Stand Grouping The algorithm considers stand grouping, in order to establish categories for unimpeded time When available (in Spain only for LEMD), this grouping is generated from statistical analysis at the PRU - Modifications of this grouping has impact on the calculated values The calculation process is neither automated, nor defined (similar to ASMA) Local actors and PRU should share values - Automated calculation through defined process? - Dissemination from PRU? - Submission from every airport/national authority? 12 Performance at Airports Workshop -19 February Brussels 6

7 Common: Recalculation of Unimpeded Times (I) ASMA and TXOT Additional Times are relative indicators - Unimpeded time acting as a reference An increase in the unimpeded time may mean a reduction of the additional time - Both values are relevant for Performance Monitoring There is a direct impact of unimpeded times in additional time values When unimpeded time should be recalculated? Common: Recalculation of Unimpeded Times (II) AENA is currently calculating it in two different ways: - Every natural year - Monthly, with a yearly sliding window for values It has been measured that the error when keeping stable unimpeded times for a year can be very relevant - In AENA, in a real scenario, a lineal modification of 1.37 minutes in unimpeded time has been observed during a year - The monthly values calculated are very affected if values for unimpeded are constant (up to 1 minute of additional time) Procedure for recalculation is not automated, or fully defined SPECIMEN DATA 13 Performance at Airports Workshop -19 February Brussels 14 Performance at Airports Workshop -19 February Brussels 7

8 ASMA and Taxi Out Additional Times at AENA CONCLUSIONS Workflow Good cooperation process with PRU The value and need of local implementations Room for improvement Calibration data Updated algorithms Working groups Algorithms Opportunity for automation and full definition of processes Unimpeded recalculation Statistical analysis Technical aspects to be discussed (ASMA entry, ) Objective Increase Reliability of Performance Monitoring through transparency of methods, understanding of the indicators and confidence in the results 15 Performance at Airports Workshop -19 February Brussels 16 Performance at Airports Workshop -19 February Brussels 8

9 ASMA I: 40 NM radius In LEMD, the 40NM radius for ASMA is not including holdings, thereby their effect is not considered in this indicator. Thank you for your attention For internal purposes, AENA calculates ASMA additional time considering 40NM and 60NM radius, that includes the holdings. - In airports with holdings inside 40NM radius, Additional time does not change significantly - In LEMD, noticeable increase (that makes the indicator more useful for performance monitoring) jmcordero@e-crida.aena.es jmdepablo@aena.es 18 Performance at Airports Workshop -19 February Brussels 9

10 ASMA II: First/Last entry in ASMA radius ASMA II: First/Last entry in ASMA radius (example) Found during calibration (after receiving a sample of 1 month) Situation generated when the holding is over the ASMA frontier (several crossing points), as LEMD AENA data 21:32:21 First ASMA Entry 1st Entry 2nd Entry 3rd Entry 4th Entry 21:32:21 21:45:16 21:53:55 22:00:33 Initially unespecified in the published description of the algorithm which entry was considered AENA Difference of more than 28 minutes PRU - AENA considered the first - PRU considered the last The results of both approaches have been calculated - Aprox. 0.5 minutes increment of ASMA Additional Time for LEMD when considering the first entry - Other airports unaffected 19 Performance at Airports Workshop -19 February Brussels 21:45:16 Second ASMA Entry 21:53:55 Thrid ASMA Entry PRU Data: 22:00:33 Fourth ASMA Entry 20 Performance at Airports Workshop -19 February Brussels 10

11 ASMA III: Entry Sector categorisation The entry sectors that are considered in the categories for ASMA are obtained by PRU through statistical analysis - They were provided for local implementation for Spanish airports The calculation process is neither automated, nor defined - Actors implementing indicators willing to attach to PRU algorithm will find discrepancies in results if not considering the same Automation? Changes to be distributed to the involved local actor (or agreed)? Values provided by every airport? - Avoiding discrepancies and potential unreliabilities in results 21 Performance at Airports Workshop -19 February Brussels 11