Você está na página 1de 44

EUROPEAN ORGANISATION

FOR THE SAFETY OF AIR NAVIGATION

EUROCONTROL EXPERIMENTAL CENTRE

ATFCM Performance and Declared Capacity Study


Work Interim Results (2006)

ATFM Ground Regulation Efficiency

EEC Note No. 20/2006

Project NCD-1-CD-FLOW

Issued: December 2006

The information contained in this document is the property of the EUROCONTROL Agency and no part should be
reproduced in any form without the Agency’s permission.
The views expressed herein do not necessarily reflect the official views or policy of the Agency.
This document has been collated by mechanical means.
Should there be missing pages, please report to:

EUROCONTROL Experimental Centre


Publications Office
B.P. 15
91222 BRETIGNY-SUR-ORGE Cedex
France
DOCUMENT CHANGE RECORD
The following table records the complete history of the successive editions of the present
document.

EDITION DATE DESCRIPTION OF EVOLUTION MODIFICATIONS

0.1 October 15th2006 First Draft

0.2 December 2006 Final Note


Intentionally blank
REPORT DOCUMENTATION PAGE

Reference Security Classification


EEC Note No. 20/2006 Unclassified
Originator Originator (Corporate author) Name/Location :
EEC - NCD EUROCONTROL Experimental Centre
(Network Capacity and Demand B.P.15
management) Research Area F-91222 Brétigny-sur-Orge CEDEX
FRANCE.
Telephone: +33 (0) 1 69 88 75 00
Sponsor Sponsor (Contract Authority) Name/Location
CFMU EUROCONTROL Agency
Rue de la Fusée, 96
B-1130 BRUXELLES
Telephone: +32-(0)2-729 90 11
Title : ATFCM Performance and Declared Capacity Study -
Work Interim Results (2006)

Authors Date Pages Figs Tables Annex References


Marc Dalichampt (EEC) 12/06 34 6 1 4
ADV Systems

Project Sponsor Task No. Period


NCD-1-CD-FLOW Year 2005

Distribution Statement :
(a) Controlled by : Head of NCD (Network Capacity and Demand management)
(b) Special Limitations (if any) : None
(c) Copy to NTIS : No
Descriptors (keywords): Uncertainty, Smoothing, ATFCM, ATFM, Indicators,
CFMU, Performance, FMP, Flow, Capacity
Abstract :
Objective: This note presents the interim conclusions of a study launched in April 2006.The
study aimed at proposing a list of ATFM performance indicators that reflect the variety of the
ATFM users' objectives.

The study was part of the yearly NCD "ATFCM Studies" work programme, in support of and
funded by CFMU.

Results: The notion of ATFM performance should be associated to a wide range of users'
objectives which lies well over the traditional resolution of the hourly demand excesses
problem. Other objectives are: Reduction of the traffic bunching problem, prevention from the
forming of complex traffic clusters within the sectors, improvement of flows predictability,
optimisation of the declared capacity.

Different indicators are defined to assess whether the system is able or not to meet these
objectives. In some cases, it is possible to observe that a given regulation can be very
efficient for the achievement of one particular objective while providing modest results for the
achievement of the other objectives. That is why, it is proposed to assess the performance by
applying the whole set of indicators.
Intentionally blank
EUROCONTROL Experimental Centre
ATFCM Performance and Declared Capacity Study

TABLE OF CONTENTS

1 INTRODUCTION 1
1.1 DOCUMENT RATIONALE ................................................................................... 1
1.2 CONTEXT ....................................................................................................... 1
1.3 STRUCTURE OF THE DOCUMENT ...................................................................... 2
2 ATFM REGULATION EFFICIENCY: PERFORMANCE MEASURE
ACCORDING TO USERS’ OBJECTIVE 3
2.1 OBJECTIVE N°1: IMPROVEMENT OF THE INCOMING TRAFFIC FLOWS SMOOTHING . 3
2.2 OBJECTIVE N°2: IMPROVEMENT OF FLOWS PREDICTABILITY ............................... 4
2.3 OBJECTIVE N°3: OPTIMISATION OF CAPACITY USE ............................................. 4
2.4 CONCLUSIONS................................................................................................ 5
3 PERFORMANCE INDICATORS 6
3.1 SMOOTHING INDICATORS ................................................................................ 6
3.2 UNCERTAINTY REDUCTION INDICATORS .......................................................... 14
4 ILLUSTRATIVE APPLICATION CASE 17
4.1 SMOOTHING PERFORMANCE APPLICATION CASE: ANALYSIS OF ONE REGULATION
PERIOD (EUHL26A REGULATION - LFEUHL4 TV) .......................................... 17
4.2 UNCERTAINTY REDUCTION APPLICATION CASE: LFEUF4 TV (REIMS ACC)
ANALYSIS, OVER ONE AIRAC CYCLE .............................................................. 20
5 APPROPRIATENESS OF THE USE OF ALL-FT FILES VS. THE USE OF
SIMULATION RESULTS (TACOT) 23
5.1 RELEVANCE OF A STEP-BY-STEP SIMULATION OF THE SLOT ALLOCATION PLAN
LIFE-CYCLE .................................................................................................. 23
5.2 SIMULATIONS RESULTS ................................................................................. 24
6 CONCLUSION & NEXT STEPS 29
6.1 CONCLUSION: A SEGMENTED ATFM GROUND REGULATION PERFORMANCE
ASSESSMENT ............................................................................................... 29
6.2 NEXT STEPS ................................................................................................. 30
EUROCONTROL Experimental Centre
ATFCM Performance and Declared Capacity Study

List of Figures

Figure 1: Flow evolution curves.............................................................................................. 6

Figure 2: 2D view (average amplitude and duration of the traffic load fluctuations).......... 8

Figure 3: 2D view (average amplitude and duration of the traffic load fluctuations).......... 9

Figure 4: Flow evolution curves (measures of max. and average excesses) ........................ 11

Figure 5: Flow evolution curves (measure of the average deviation from the flow rate).. 12

Figure 6: Distributions of TV entries over the reg. period (measure of the time during
which TV entries are over a certain level) ......................................................... 13
EUROCONTROL Experimental Centre
ATFCM Performance and Declared Capacity Study

GLOSSARY

ACC Area Control Centre


AIRAC Aeronautical Information, Regulation and Control
AO Aircraft Operator
ATC Air Traffic Control
ATCO Air Traffic Controller
ATFCM Air Traffic Flow and Capacity Management
ATFM Air Traffic Flow Management
ATM Air Traffic Management
ATO Actual Time Over

CASA Computer Assisted Slot Allocation


CFMU Central Flow Management Unit
CPR Correlated Position Report
CTO Calculated Time Over
CTOT Calculates Take-Off Time

EEC EUROCONTROL Experimental Centre


ETFMS Enhanced Tactical Flow Management System
ETO Estimated Time Over

FMD Flow Management Division (CFMU) Controllers


FMP Flow Management Position (ACC)
FPL Filed Flight Plan

IFPS Integrated Initial Flight Plan Processing System

NCD Network Capacity and Demand

PRU Performance Regulatory Unit


PRR Performance Review Report

RTO Reference Time Over

TMA Terminal Manoeuvring Area


TV Traffic Volume

WP Work Package
EUROCONTROL Experimental Centre
ATFCM Performance and Declared Capacity Study

REFERENCES

EUROCONTROL / Performance Review Commission:


Performance Review Reports: n°1 (1998) to n°8 (2005)

EUROCONTROL
Tasking Support for ATFM – Impact of uncertainty and smoothing on
declared capacity – Interim results (November 2005)

EUROCONTROL
Tasking Support for ATFM – Impact of uncertainty and smoothing on
declared capacity – General framework (November 2005)

EUROCONTROL
Tasking Support for ATFM – Intermediate note on ATFM efficiency indicators
- ATFCM Performance and Declared Capacity study (July 2006)
EUROCONTROL Experimental Centre
ATFCM Performance and Declared Capacity Study

1 INTRODUCTION

1.1 Document rationale


The present document was issued in the frame of the “ATFCM Performance and
Declared Capacity study”, which has been launched by the EEC in April 2006. It
provides the interim results drawn from the conclusions of the first two WPs which
primarily intend to propose a list of ATFM performance indicators and to apply this
set of indicators on some illustrative application cases.
The document investigates also the suitability of using the “FTFM1” and “RTFM2”
profiles obtained from the processing of “after-the-fact” data, which is extracted from
the CFMU “ALL-FT” files. The results provided by the use of the “ALL-FT” files are
confronted to those obtained with the use of an ATFM fast time simulator (TACOT),
when trying to assess the performances of the ATFM process and especially ground
regulation one.
Indicators and application cases were presented and discussed with CFMU
representatives in October 2006.

1.2 Context

The conclusions of a previous study, launched by the EEC in September 2005:


“Study of the impact of ATFM uncertainty and smoothing performance on declared
capacity” demonstrated that the notion of ATFM performance should be associated
to a wide range of users’ objectives which lies well over the traditional resolution of
the hourly demand excesses problem:
9 Ground regulation users’ secondary objectives:
• Reduction of the traffic bunching problem;
• Prevention from the forming of complex traffic clusters within the
sectors;
• Reduction of the uncertainty (improvement of flows predictability); and
• Optimisation of the declared capacity.

Therefore, the ATFM performances should reflect the variety of the ATFM users’
various objectives and especially the ones of the FMPs and FMDs. The identification
of the users’ objectives associated to ATFM performance has constituted the
baseline for the instigation of the work on ATFM performances indicators. Once the
objectives have been identified, a set of indicators has been proposed to assess the
performances of the ground regulation process for each of the identified objectives.
Those indicators have been further “tested” on sundry application cases.

1 The FTFM is the “initial” profile as it reflects the status of the demand before activation of the regulation

plan. It is computed with the latest flight plan version, sent by each AO to the CFMU/IFPS.
2 The RTFM is the “regulated” profile as it reflects the status of the demand after activation of the regulation

plan. It is computed with the latest ATFM slot (CTOT) issued to the AO, by the ground regulation system.

1
EUROCONTROL Experimental Centre
ATFCM Performance and Declared Capacity Study

1.3 Structure of the document

The document is broken down into six sections and one annex:
9 Section 1 – Introduction – presents the purpose and structure of the present
document;
9 Section 2 – ATFM regulation efficiency: Performance measure according
to users’ objective – recalls the users’ objectives identified in the previous
2005 study and presents how the performance indicators are related to them;
9 Section 3 – Performance Indicators – presents the indicators proposed to
give account of the system’s capability to fulfil the users’ objectives and traffic
counting functions on which these indicators are based;
9 Section 4 – Illustrative application cases – presents the results, which
could be obtained from the application of the indicators on a particular case;
9 Section 5 – Appropriateness of the use of ALL-FT files vs. the use of
simulation results (TACOT) – assesses the suitability in using “after the
fact” data extracted from ALL-FT files, as compared to the use of results
provided by an ATFM fast-time simulator (TACOT), for the measurement of
ATFM performances;
9 Section 6 - Conclusion and Next steps – presents the interim conclusions
on the proposed performance assessment method, the next work steps and
the associated requirements.
9 Annex 1 – ATFM smoothing performance assessment (methodology) –
presents the methodology applied for the traffic data analysis and the building
of the different observables of the ATFM slot allocation smoothing efficiency.

2
EUROCONTROL Experimental Centre
ATFCM Performance and Declared Capacity Study

2 ATFM REGULATION EFFICIENCY:


PERFORMANCE MEASURE ACCORDING TO USERS’
OBJECTIVE

As specified in the introduction of this document, our approach is essentially


established on the account of a number of users’ objectives which stretches wider
than the traditional basis upon which the ATFM performance is assessed (resolution
of the hourly demand excesses problem).
In this document, the term “users” of the ATFM system refers to the direct
beneficiaries of the ATFM ground regulation system, i.e. ATFM operational actors –
FMPs, FMDs and controllers.
A recall of the users’ objectives and how the performance indicators relate to them is
presented in the next paragraph.

2.1 Objective n°1: Improvement of the incoming traffic flows


smoothing

Three distinct smoothing purposes <> Three measure parameters

In interviewing FMPs and FMDs, it appeared that the smoothing objective could be
parted in three different sub-objectives:
9 Firstly, the traditional resolution of the hourly demand / capacity
imbalance. This refers to the original mandate attributed to the ATFM
regulation system and relies on a commitment vis-à-vis a number of
movements per hour.
The corresponding performance indicators will therefore derive from the
analysis of hourly flow counts.

9 Secondly, the reduction of the traffic bunching. The problem of bunching


has much to deal with the punctual accumulation (bunching peaks) of traffic
on controllers working positions, even in situations when the hourly traffic flow
complies with the declared capacity. FMPs regularly exploit the traffic
smoothing capability of the ground regulation process for the resolution of this
most “unpleasant” situation for ATCOs, especially when the ATCOs are
already working at the limits of the hourly sector capacity.
In that case, the corresponding performance indicator derives from the
analysis of twenty minutes flow counts. This is consistent with operational
figures a priori determinant in terms of ATCOs perception of demand
pressure; such as the average time of virtual or effective presence of a flight
into ATCO responsibility area, or the minimum time-window which the ATFM
system offers to analyse the flow throughput.

9 Last but not least, the prevention from the forming of complex traffic
clusters within the sectors. This last objective may be considered out of
ATFM scope, and is normally addressed directly by the ACCs: FMPs,
together with ATCOs collaboratively coordinate with upstream ACCs for
obtaining a regular spacing of aircraft and preventing from the forming of
complex traffic clusters within the sectors. However, we were reported that
FMPs may also wish to improve in advance the regularity of the flows

3
EUROCONTROL Experimental Centre
ATFCM Performance and Declared Capacity Study

throughput and use the “limit” smoothing capability of the ATFM system for
this “à la marge” traffic smoothing objective.
The corresponding performance indicator derives from the analysis of
instantaneous – 1 minute flow counts.

2.2 Objective n°2: Improvement of flows predictability

The second class of objective – Improvement of flows predictability – reflects a


feeling regularly expressed by FMPs and controllers that they may request the
implementation of ATFM ground regulations, even when the planned demand
excesses are low, in order to improve the predictability of the incoming flows.
One of their primary concerns is to make sure that the uncertainty, conveyed by the
stochastic nature of air transport, could be contained in between stable and reduced
margins. To this regard, the slot allocation process is considered as a mean to
improve flows predictability, since it imposes additional departure time restrictions3.
This commonly admitted explanation as to why the flows predictability is improved is
assessed using a specific performance indicator accounting for the gain in time
precision over TV entry when ATFM regulations are implemented.
From this perspective the predictability is measured in terms of individual flights time
deviation between actual and planned times of penetration into a given airspace.

2.3 Objective n°3: Optimisation of capacity use

The last class of objective is an indirect, longer term objective that is out of the
performance measurement scope presented in this document. It focuses on the
direct results of ATFM ground regulation application (smoothing and predictability
improvement) on the declared capacity over the years.
As a matter of fact, the calculation of the declared capacity could be sometimes an
empirical process. A way of assessing a suitable value for the declared capacity of a
given TV consists in issuing some ATFM ground regulations with gradual increased
values. The process allowing the suitable declared capacity value could be explained
as follows:
At first the FMP issues a low value for the declared capacity and if there is no ATCO
report on over-delivery for some times, the FMP will gain confidence in the system
and will increase gradually the declared capacity value to a point at which some
ATCOs will complain on over-deliveries.
Therefore, the objective – that draws the link between the ATFM service performance
and the capacity offer – could be assessed using a specific performance indicator
accounting for the positive evolution of capacity figures, if not over the entire life
cycle of the ATFM ground delay service, at least over months or years.

3 The regulations restrict departure times’ variations down to a 15 minutes tolerance whilst the tolerance

applicable on non-regulated traffic is 30 minutes.

4
EUROCONTROL Experimental Centre
ATFCM Performance and Declared Capacity Study

2.4 Conclusions

The performance of a given ATM ground regulation should reflect the objectives
sought by the users when issuing the ground regulation. That is why, we propose to
assess the overall performance of a ground regulation, by applying the whole set of
indicators on this regulation. We will probably observe that a given regulation can be
very efficient for the resolution of the bunching effect when providing modest results
for the prevention of hourly flows over-deliveries. In other cases, we will maybe find
that a given ground regulation has been issued just to reduce the uncertainty
conveyed by the stochastic nature of air transport.
In any case, the performance of a given regulation will be assessed in applying the
whole set of indicators reflecting all the users’ objectives.

5
EUROCONTROL Experimental Centre
ATFCM Performance and Declared Capacity Study

3 PERFORMANCE INDICATORS
Performance indicators are derived from a model, which describes the different input
databases, counting functions and observation parameters used for the computation
of each indicator. The model is presented in Annex 1.
Performance indicators are presented in the following chapters.

3.1 Smoothing indicators

Flow evolution curves


The ATFM activity is based on the notion of traffic deliveries (or traffic flows)
throughput. Local ATFM service users (FMPs) specify a particular request on traffic
flow throughput (flow rate) in order to meet a specific load requirement.
Therefore, the adopted indicators of performance will rest on flow evolutions curves.
The curves are obtained from CFMU data regulation records, and present the
evolution, minute per minute, of TV entries during a certain period “ahead” (i.e. the
number of aircraft flying over a sector entry point or landing at an airport in a specific
unit of time).
The time period may be 60 minutes, 20 minutes, 1 minute, depending on the
smoothing measured objective: “resolution of the hourly demand / capacity
imbalance”, “reduction of the traffic bunching”, or “prevention from the forming of
complex traffic clusters within the sectors”, respectively.

An example of flow evolution curves is presented below:


The figure on the left
presents the evolution of TV
entries for the period
4
elapsing from “t” to “t + 60”
minutes, at three different
anticipation levels: just before
the activation of the reg.
process (blue curve) and just
after the activation (pink
curve) of the reg. process,
then at flights entry time into
the regulated traffic volume
(green curve). The requested
Regulation period hourly flow rate level (47) is
pictured in black.

Figure 1: Flow evolution curves

4
A time index is represented on the (X axis). The origin (Index “0”) corresponds to the time of activation.
Index “178” corresponds to the start of regulation period. The regulation process is therefore activated at
anticipation “178” minutes, i.e. 2 h 58 min.

6
EUROCONTROL Experimental Centre
ATFCM Performance and Declared Capacity Study

In order to build a practicable smoothing performance evaluation, the shapes of flow


evolution curves have been characterised by a limited number of determinant figures,
measuring:
9 Individual evaluation of the traffic load fluctuations amplitude and duration
(analysis of the traffic fluctuations whatever the originating regulation(s)) ;
9 Scan of one regulation period (analysis of one particular regulation):
• Amplitude of the traffic excesses, over the regulation period;
• Average deviation from the flow rate (efficiency in the use of capacity);
• Dispersion around other references than the flow rate.

In the measures proposed herein, two levels of analysis are distinguished.


The first level of analysis – individual evaluation of the traffic load fluctuations – will
correspond to a discrete characterisation of the flow evolution curve. The curve is
indeed characterised by (amplitude, duration) pairs of measures, each pair being
associated to a traffic load fluctuation. Given a regulation, these measures address
each fluctuation individually, breaking the link between them (i.e. their common
regulation period). Therefore, they are a good way to look into the overall protection
of one to several regulations (e.g. protection of one TV over one AIRAC cycle), by
aggregating results whatever their originating regulation.
The second level of analysis – scan of one regulation period – put the focus on one
particular regulation period. Here, the indicators measure if the implementation of this
regulation is in line with what the FMP was expecting, i.e. if it would allow the ATCO
on his position to deal with the actual traffic and if the available capacity is efficiently
used, over the regulation period.

7
EUROCONTROL Experimental Centre
ATFCM Performance and Declared Capacity Study

Individual evaluation of the traffic load fluctuations.


Indicator I1 (reduction of the duration and average amplitude of the
fluctuations).

The primary purpose of ATFM service users (FMPs) in each ACC is to avoid
situations of unbearable ATCO workload. The objective is the elimination of the
uncontrolled loads of traffic (traffic peaks during significant periods of time).
Another objective is to make the most efficient use of the available capacity (deduced
from the flow rates requests) by minimising capacity wastes (traffic shortfalls during
significant periods of time), and to limit the burden (delays) imposed on airspace
users induced by the implementation of regulations.
Therefore, ATFM smoothing performance will be firstly evaluated with regards to the
reduction, in terms of duration and amplitude, of the traffic load fluctuations (“peaks”
and “shortfalls”) around the flow rate.
In order to discriminate between the different fluctuations, we propose to segment the
regulation period into series of shorter intervals, within which the flow evolution curve
is continuously either strictly above (peak interval) or below (shortfall interval) the
requested flow rate.
The result is a discrete characterisation of the flow evolution curve in terms of
average amplitude of a fluctuation (positive or negative) and duration of the
fluctuation.

9 Illustration n°1: results for one regulation.


A first illustration, showing the average amplitude and duration of the traffic load
fluctuations for one single regulation (case of the example above presented – figure
1) is presented below5. Fluctuation of average
Figure 2: 2D view (average amplitude and duration of the traffic load fluctuations)
amplitude +6 (+13%) and
duration 108 minutes

Flow Rate: 47
Fluctuation of average
amplitude -10 (-21%)
and duration 106 min.

5 The average amplitude is measured in terms of number of aircraft above or below the flow rate and in

proportion (%) to the flow rate.

8
EUROCONTROL Experimental Centre
ATFCM Performance and Declared Capacity Study

In the above figure, it could be shown that the blue curve (before activation of the
reg. process) is represented by one dot showing one over-delivery of average
amplitude 13% and one under-delivery of average amplitude 21%.
It could be also noticed that the pink curve (after activation of the reg. process) is
very fluctuating, and close to the flow rate as being represented by several dots close
to the “flow rate (i.e. 47)” axis.
Pink dots are inside a zone (in green) delimited by -10% and +10% below and above
the flow rate. Then, it could be inferred that the regulation is “successful”, as the
over-deliveries are fully eliminated according to a traditional7 tolerance margin
accepted by the FMP (in case of hourly counts measurements).

9 Illustration n°2: aggregated results: several regulations (10 TVs, 6 days).


As already mentioned, the 2D view presented can feature a wider array of points,
whatever their originating regulation. Thus the use of the (amplitude, duration) pair of
measures is a good way to look into the overall impact of a sample of regulations.

3 situations with regulated demand, exceeding in


amplitude (> 10%) and duration (> 20 minutes)

12 situations with initial demand, exceeding in


amplitude (> 10%) and duration (> 20 minutes)

10 situations with regulated under-delivered demand


(< -10%) and (> 20 minutes)

17 situations with significant initial under-deliveries


(< -10%) and (> 20 minutes)

Figure 3: 2D view (average amplitude and duration of the traffic load fluctuations)

The aggregated results figure6 demonstrates clearly that regulations – most of the
time – prevent or at least decrease substantially over-deliveries (most of the pink
dots inside the -10% ; +10% green zone). However, some exceptions are recorded:
3 pink dots inside the {> 10% ; > 20 minutes}7 orange zone.

6 The “Y axis” evaluates the fluctuations in proportion to the flow rate.


7 The criteria are at the convenience of the ATFM service user. A {>10% ; > 20 minutes} zone would
correspond to the outside zone of tolerance margin for the resolution of hourly over-deliveries, within which
the FMPs / FMDs are traditionally satisfied with the regulation. Regarding the assessment of capacity wastes
(under-deliveries), it could be inferred that the outside zone is symmetrical {i.e. <-10%; > 20 minutes} but this
is to be confirmed and checked with the FMPs / FMDs.

9
EUROCONTROL Experimental Centre
ATFCM Performance and Declared Capacity Study

The observation of the blue zone {< - 10% ; > 20 minutes} depicts a more
preocupaying situation, in terms of capacity efficiency use: still significant 10 under-
deliveries records, revealing capacity wastes.
Thus, the use of the 2D views, as illustrated by the last presented figures are a good
way to discriminate between the successes patterns (a “success” zone that would
include the points for which the smoothing is successful according to a tolerance
margin, in this case: between -10% and +10%) and exceptional or failure situations
(the outside zones).

Then, a first indicator of performance (I1) could be the proportion of points


inside the “success” zone, with respect to the points that are in the “outside”
zone.

The reason for experimenting peaks of significant amplitude and duration, in zones
outside the tolerance margin within which regulations are considered “successful” by
the ATFM service user, are to be investigated using additional information recorded
at the CFMU in heavy databases (Oplog files). In these files, all the relevant ATFM
events, including regulation requests tracks are archived.
In many cases, sources of inefficiency situations are related to the system’s inertia
(regulations have not been issued before the time required by CFMU to operate
properly a regulation).

10
EUROCONTROL Experimental Centre
ATFCM Performance and Declared Capacity Study

Scan of one regulation period:


Indicators I2, I3 (prevention from over-deliveries)

Given one regulation or given several regulations, the 2D view (amplitude; duration –
Indicator I1) is a very good way to look into the success or the failure of the TV(s)
protection and use of the available capacity. This analysis can be done for several
regulations, since each fluctuation is characterised individually.
Now, when trying to figure out if one particular regulation is in line with what the FMP
was expecting; one has to figure out if the implementation of this specific regulation
could allow the ATCO on his position to deal with the actual traffic.
To give account of this, the shape of the flow evolution curve can be characterised by
the maximum peak and average traffic excesses records, over the regulation period.
The two are complementary since an excess may not climax at the maximum, for a
long time, which in turn mitigates the average excess record.

Then, another pair of indicators derives from the measures of the maximum of
the flow evolution curve (I2) and of the average of the Flow evolution curve
excesses (I3) over the regulation period.

In the illustration besides (case


of hourly counts, before and
Max (Before reg.): 59 (26%) after the activation of reg.
process), the maximum
amplitude of excesses is 59
Average (Before reg.): 53 (13%)
(excess of 26%), while the
average is 53 (excess of 13%),
for the blue curve (before reg.).
The corresponding values for
Max (After reg.): 51 (9%) the pink curve (after reg.) are
51 (excess of 9%) and 48
(excess of 2%), respectively.
Average (After reg.): 48 (2%) In this example, it could be
shown that the regulation was
of interest in lowering the
traffic excesses, over the
period.
Figure 4: Flow evolution curves (measures of max. and average excesses)

Note: In the above figure, the shape of the flow evolution curve is evaluated in terms
of numbers of aircraft and in proportion (%) to the flow rate (e.g. 53 aircraft
corresponds to (53-47)/47, i.e. 13%)

11
EUROCONTROL Experimental Centre
ATFCM Performance and Declared Capacity Study

Scan of one regulation period:


Indicator I4 (capacity use)

The other component of the ATFM smoothing performance relates to an efficient use
of the achievable capacity offered by the ATCO (deduced from the flow rate
requests).
Therefore, when looking at the performance of one single regulation, one has to look
on the ability of the system to maintain the load as close as possible to the flow rate,
over the regulation period. The system is all the more efficient as the average
deviation from the flow rate is low.
Then, a fourth indicator derives from the measure of the average deviation (in
absolute value) from the flow rate, over the regulation period.

In the illustration besides


Average deviation
(case of hourly counts,
from the flow rate
before and after the
(Before regulation):
activation of reg. process),
8 (17%)
the initial (blue curve)
average deviation from the
flow rate is equal to 8
(17%) whereas the system
output a distribution with
Average deviation average deviation 1 (2%).
from the flow rate It could be shown the
(After Regulation): efficiency of the regulation
1 (2%) in reshaping the initial
distribution very close to
the flow rate.
Figure 5: Flow evolution curves (measure of the average deviation from the flow rate)

Note: In the above figure, average deviations are evaluated in number of aircraft and
in proportion to the flow rate (e.g. an average deviation of 8 aircraft corresponds to
8/47, i.e. 17%).

12
EUROCONTROL Experimental Centre
ATFCM Performance and Declared Capacity Study

Dispersion around other references than the flow rate, capacity management
Indicator I5 (definition of a “tolerance” margin for over-deliveries)

With the last presented measures, the smoothing is characterised with respect to a
“standard” reference, the requested flow rate.
In addition to the compliance with the requested flow rate, ATFM users may have
different criteria against which their own evaluation of the system’s performance is
made. Depending on the user, the “tolerance” margin within which a traffic excess
remains acceptable could be for example of 0%, 10% or 20% above the flow rate. In
the same way, the efficiency in the capacity use could be considered as achieved
when the load is above a limit of -20%; -10% or 0% below the flow rate.
Therefore, the smoothing quality could be also evaluated when that natural reference
(the flow rate) changes.
Then, a fifth indicator derives from the measure of the amount of time
(expressed in terms of percentage of the total regulation time) during which the
traffic flow has been over a certain level. That “certain level” is the new reference
and is defined over a range of values centred on the optimum (the requested flow
rate). Then it is another way of measuring the ability to keep the regulated traffic as
close as possible to the flow rate.

New Reference:
Flow Rate: 47 In the illustration, the initial (blue
Flow Rate + 20%: 56
curve) traffic excesses account for
47% of the total regulation time
whereas the system output a
Before regulation:
distribution with excesses
% of reg. period with
accounting for only 8% of the
TV Entries > Flow
total regulation time.
Rate: 47%
Before regulation:
When the reference changes (new
% of reg. period with
reference: Flow Rate + 20%, i.e.
TV Entries >
After regulation: 56), initial traffic excesses account
Flow Rate + 20%: 5%
% of reg. period with for 5% of the total reg. time,
TV Entries > Flow while the system output a
Rate: 8% distribution with all the period
below that reference (no excesses
> 20%).
Figure 6: Distributions of TV entries over the reg. period (measure of the time during which TV
entries are over a certain level)

Note: The ideal distribution would be a regulated curve confounded with the “Y axis”.

13
EUROCONTROL Experimental Centre
ATFCM Performance and Declared Capacity Study

3.2 Uncertainty reduction indicators

Flight time deviations on arrival at the TV


The commonly admitted explanation as to why the predictability of incoming flows for
a given TV is improved with ATFM ground regulation implementation is the
application of additional time restrictions on the departing traffic, which squeeze
operational drifts prior to take-off, and hence benefit to the subsequent flight
segments, up to the TV in question.
Therefore, uncertainty reduction is here mirrored by the reduction in flight time
deviations on arrival at the TV.
Flight time deviations are evaluated by comparing time estimates over TV entry
points. The estimates are either an ETO8 in the case of non-regulated flight or a
CTO9 in the case of regulated flight:
9 Non Regulated: The flight time deviation is measured by comparing ETO
(Estimated Time Over) and ATO (Actual Time Over) on arrival at the TV;
9 Regulated: The flight time deviation is measured by comparing CTO
(Calculated Time Over) and ATO (Actual Time Over) on arrival at the TV.
The methodology is based on a global evaluation – one studied TV over one to
several AIRAC cycles (sample of analysis). Within the sample of analysis, the flights
are sorted by call-sign and grouped in two subsets: regulated and non-regulated.
For each call-sign, a twofold measurement is performed for each subset:
1. Average flight time deviation over TV entry point (mean value);
2. Dispersion around the average (standard deviation).

Call-signs are then put to the test individually to determine whether the measurement
is statistically meaningful. The “succeeding” call-signs are those for which
significance is ensured with a confidence level of at least 95%, a standard value in
statistics.
Significant call-signs are then correlated to figure out (if possible) a relationship
between the average flight time deviation without regulation and the average flight
time deviation with regulation. The type10 and coefficients of the relationship are
the indicators of the impact of ATFM measures on uncertainty reduction (I6).
A graphical illustration, recapping the main steps of the methodology, is presented
below.

8
The ETO is the expected time of penetration within the TV, when the flight is not regulated. It is obtained
from the latest flight plan sent by the AO to the CFMU.
9
The CTO is the expected time of penetration within the TV, when the flight is regulated. It is deduced from
the latest slot (CTOT) issued to the AO.
10 Premininary results seem to derive a linear relationship (see § 4.2)

14
EUROCONTROL Experimental Centre
ATFCM Performance and Declared Capacity Study

TV TV TV All movements are captured


1/ Collection
Day 1 Day 2 Day - through TV, over timeframe

NO REG
Movements are sorted by:
2/ Reshuffle 1/ call-sign
REG 2/ status (regulated or not)
Call-sign 1 Call-sign 2 Call-sign -

Call-sign Movements TV Distributions

NO REG ∆ = | ATO - ETO |


3/ Processing

REG ∆ = | ATO - CTO |

∆ vs. occurrences
Call-signs are processed one-by-one

4/ Analysis Call-signs are tested one-by-one: are they statistically significant?


[The scope extension is an important enabler of statistical significance]

Yes No

5/ Integration
Non-significant call-signs
Mean REG

Significant call-signs

Correlation coefficient
Mean NO REG

15
EUROCONTROL Experimental Centre
ATFCM Performance and Declared Capacity Study

1/ Data Collection:
The studied TV is put under the microscope over a significant amount of time,
typically several AIRAC cycles. Every movement on record is taken into account
within that scope.

2/ Data Reshuffle:
All movements sharing the same call-sign are clustered. Each cluster is partitioned:
non-regulated movements vs. regulated movements (whatever the origin of the most
penalising regulation).

3/ Data Processing:
For each cluster, the two subsets are the baseline samples from which the
distributions of operational deviations are drawn:
9 Deviations from ETO11 for non-regulated movements;
9 Deviations from CTO12 for regulated movements.

4/ Data Analysis:
The two distributions are weighed one against the other to determine whether the
cluster is statistically significant. Is it a meaningful indicator of how the call-sign
behaves routinely, whether or not it is submitted to restrictions? Does it denote a
deep-rooted operational change? A confidence level of 95% is retained to make a
decision. Presumably: the broader the initial scope, the greater the number of
significant call-signs.

5/ Data Integration:
Significant call-signs are “benchmarked” to try and set out a trend. In other words, if
the correlation between them is strong enough, then a relationship can be drawn to
model the impact of ATFM measures on predictability.

11 The ETO is the expected time of penetration within the TV, when the flight is not regulated. It is obtained
from the latest flight plan sent by the AO to the CFMU.
12 The CTO is the expected time of penetration within the TV, when the flight is regulated. It is deduced from

the latest slot (CTOT) issued to the AO.

16
EUROCONTROL Experimental Centre
ATFCM Performance and Declared Capacity Study

4 ILLUSTRATIVE APPLICATION CASE


This chapter presents the results of the application of the previous indicators on a
particular case – Reims ACC, together with first indications that can be derived from
the indicators analysis.

4.1 Smoothing performance application case: analysis of one


regulation period (EUHL26A regulation - LFEUHL4 TV)

The case of the French North East ACC is interesting since located in an area with
high concentration of traffic and complex flows crossing. A particular point on this
case is the mix of cruising flows, of low complexity from an Air Traffic Controller
viewpoint, with the climbing or descending flows from or to the airports of Zurich,
Geneva, etc. which are of higher complexity.
Hence, in order to avoid the uncontrolled accumulation of complex aircraft, FMPs
frequently request regulations applied to the complex flows, even when the planned
demand hourly excesses are low. Here, the objective is more to tackle the bunching
threat of the climbing or descending aircraft on the working positions, than to
maintain the hourly load below a given barrier.
Because the resolution of the bunching problem is one of the intended objectives, the
emphasis here is put on the assessment of that performance of the regulation system
as well as on the (mandatory) performance of resolution of the hourly demand /
capacity imbalance. As the illustration is focused on one single regulation case, the
analysis is performed using the indicators I2, I3, I4 and I5.

First Objective: Resolution of the hourly demand / capacity imbalance

Analysis of the load evolution for initial (before activation of the reg. process) and
“after regulation” (just after activation of the reg. process) traffic (60 min. counting
step).

Maximum amplitude of over-deliveries (before reg.) (I2): 22%

Average amplitude of over-deliveries


(before reg.) (I3): 9%

Flow Rate

Average amplitude of over-deliveries (I3) Flow Rate


(After regulation): 0% % of reg. period
with TV Entries
> Flow Rate (I5):
56%

17
EUROCONTROL Experimental Centre
ATFCM Performance and Declared Capacity Study

It could be noticed that the hourly situation is perfectly corrected (I2 and I3 after reg.
– pink curve), albeit not critical initially (I2 and I3 before reg. – blue curve) with an
average excess below 10%.

Second Objective: Reduction of the traffic bunching

Analysis of the load evolution for initial (before activation of the reg. process) and
“after regulation” (just after activation of the reg. process) traffic (20 min. counting
step) 20 min counts depict a more preoccupying situation, unveiling greater demand
excesses (I2 and I3 before reg.) than expected with 1 h counts.

Maximum amplitude of over-deliveries (before reg.) (I2): 78%

Average amplitude of over- Average amplitude of over-deliveries (I3)


deliveries (before reg.) (I3): (after regulation): 4%
39%

Flow Rate
% of reg. period
with TV Entries >
Flow Rate + 10%:
1% (after reg.)

By the way, 20 min excesses and 1 h excess are not in phase. The regulation
process proves to be efficient for the bunching objective; traffic is smoothened below
10%.

18
EUROCONTROL Experimental Centre
ATFCM Performance and Declared Capacity Study

Synthesis table: 60 min. vs 20 min. counts

% 60 min. 20 min.
Before reg. After reg. Before reg. After reg.
I2 22 0 78 11
I3 9 0 39 4
I4 9 16 37 17
I5 (0%) 56 0 30 8
I5 (10%) 19 0 27 1
I5 (20%) 4 0 21 0

Summing up values is interesting to find trends. Measures indicate that:


9 From the (I2 ; I3) perspective (amplitude of the excesses), performance is
perfect for the 60 min. case and very good for the 20 min. case (e.g. from
39% initial average excesses to 4%, while looking at the after reg. result);
9 Not surprisingly, from the I5 perspective, the performance is better, not to say
perfect, for significant excess (> 20%) because smoothing first tackles those
ones. All excesses > 20%, dealing with an hourly basis are cut down for both
60 min. and 20 min. bases;
9 On the other hand, from the (I4) perspective (use of capacity), the impact of
the regulation is negative (from 9% initially to 16%, while looking at the after
reg. result), in the case of 60 min. counts, but positive, when associated to the
objective of bunching reduction (from 37% initially to 17%, while looking at the
after reg. result).
Therefore, the indicators confirm the added-value of the system for the objective of
bunching reduction. On the other hand, they point out a certain inefficiency of the
regulation for the hourly situation. Even, if the hourly excesses are fully eliminated,
the situation is not critical initially and capacity is wasted with the implementation of
the regulation.
In the example presented, it could be noticed that the performance of a regulation
should be assessed in applying the whole set of indicators, for the different count
parameters reflecting the objectives sought by the users. In this regulation case, it
could be seen that the regulation was issued for the reduction of the bunching
essentially.

19
EUROCONTROL Experimental Centre
ATFCM Performance and Declared Capacity Study

4.2 Uncertainty reduction application case: LFEUF4 TV (Reims ACC)


analysis, over one AIRAC cycle13

As mentioned in the previous sections of this document, users’ expectations towards


the ATFM service go beyond the improvement of the traffic load smoothing.
Interviewed FMPs reported that, even when the planned demand excess is low, the
decision can be made to implement an ATFM regulation in order to reduce the
uncertainty on flights at entry into the sectors.
A particular point on this case is the LFEUF4 Reims ACC TV, feeding LFEUF sector
(upper En Route sector, above FL 195), which was regulated 12 days out of 28
during the studied AIRAC cycle 265. Reims FMP has confirmed that the objective
pursued when regulating that TV is also to maintain the uncertainty under control and
hence to ensure that actual loads will remain consistent with the prediction.
In order to assess the uncertainty reduction, when the TV is regulated, a data mining
process was launched, over the entire AIRAC cycle 265. The scan of the 28 days,
allowed identifying about 400 call-signs with an average of 10 non-regulated and 5
regulated movements each.
For each call-sign, an analysis of the statistical significance of the results was
performed (using statistical tests), in order to determine whether or not the results are
representative of how the call-sign behaves routinely when it is regulated and when it
is not. For some part, statistical significance depends on the size of the call-sign
sample (either regulated or non-regulated samples).
Many “underrepresented” call-signs were therefore set aside so far, whereas the
remainder is showcased hereafter. For these fifteen or so call-signs, the relevance of
the results is secured since statistical significance is ensured with a high degree of
confidence (95%)14.

Average deviation: Regulated movements vs. Non-regulated movements


Each data point stands for a call-sign and pinpoints the average deviation, for non-
regulated vis-à-vis regulated movements.

13Application case: LFEUF4 TV over AIRAC cycle 265


14 As many call-signs are considered non eligible for the analysis, there is a need to consider a larger time
frame for the analysis. Therefore, the next step is to collect data over six AIRAC cycles, to say the least. As
more movements are pumped into the sample, more call-signs will acquire statistical significance, becoming
meaningful data points in their own right. They should give further evidence on the trend obtained so far.

20
EUROCONTROL Experimental Centre
ATFCM Performance and Declared Capacity Study

18

16

14
Regulated movement

12

10

0
0 2 4 6 8 10 12 14 16 18
Non-regulated m ovem ent

The data points lie on the same side of the pink solid line, pointing to a reduction of
operational drifts when regulations are applied. Furthermore, they tend to flock to the
green dashed line, profiling a linear relationship with a twofold reduction in deviation
when the flight is submitted to restrictions.
Such preliminary results echo the initial assumptions15; they seem to derive a trend,
to be confirmed by the consolidation of the study, which will address more TVs, over
longer time periods.

Average deviation and Standard deviation reduction evidence

Uncertainty has different components. One of them is the average deviation of flight
time estimates over TV entry, which seems to be half-reduced when regulations are
applied. The other component of uncertainty is the dispersion of the values around
the average deviation, which under a statistical approach, is referred as the standard
deviation.
The standard deviation characterises uncertainty in terms of predictability and can be
very small even with a significant average deviation.
Therefore; in order to assess the predictability improvement for regulated flights of
the studied application case, a measure of the standard deviation for the call-signs
has also been performed. Results are illustrated below.

15Namely that uncertainty reduction seems to be half-reduced as the shortening of the ground tolerance
window, which is twofold.

21
EUROCONTROL Experimental Centre
ATFCM Performance and Declared Capacity Study

18

16

14

12 Non-regulated movement
Standard deviation

10

8
Regulated movement
6

0
0 2 4 6 8 10 12 14 16 18
Mean

The same call-signs are spotted twice hereunder, in the non-regulated (diamonds)
and regulated (triangles) modes. The standard deviation is exhibited in addition to the
mean.
A slip towards the origin is noticeable for the whole cluster. The magnitude of the
shift is a rough marker of uncertainty reduction. The attracting power of the origin
points out the overall impact of ATFM measures in that sense. Ideally, a call-sign
lying on the focal point would become more predictable, with no operational
deviation.

Conclusion on the uncertainty reduction: linear relationship of slope 0.5

In the case study presented, the comparison between the average flight time
deviation with regulation and the average flight time deviation without
regulation derives a trend: a linear relationship of slope 0.5.
The present result needs to be confronted with an analysis on other TVs and longer
periods so as to assess if this is a regular pattern and/or if it depends from the TVs
characteristics such as e.g. the distances to the departure aerodromes.
The correlation coefficient will indicate the strength of the relationship.

22
EUROCONTROL Experimental Centre
ATFCM Performance and Declared Capacity Study

5 APPROPRIATENESS OF THE USE OF ALL-FT


FILES VS. THE USE OF SIMULATION RESULTS (TACOT)

The previously outlined indicators are herein run on data records to measure
smoothing performance from an analytical perspective. However different options are
available to shape the data into a purposeful observation model. They range from the
static batch of “after-the-fact” flight models (ALL-FT-based)16 to the dynamic
simulation of the evolution of the slot allocation plan.
The former is a fast-track (permanent availability of ALL-FT files) but partial way to
measure performance, as the observation model does not strictly reflect the traffic
load situation that FMP / FMD would have observed on the field.
The latter is a time-consuming solution, but it enriches considerably the performance
assessment, and guarantees quality of the results since it provides the observables
at different time events of the regulation life-cycle, when the FMP / FMD decides to
activate, monitor, modify or cancel a regulation.
The following chapters present the results of a first set of simulations launched in
order to “test” the quality of the observations obtained when using the ALL-FT model.

5.1 Relevance of a step-by-step simulation of the slot allocation plan


life-cycle
Until now, our reports have presented preliminary results from the analysis of
different observables, at three different statuses: “initial (FTFM)17”, “regulated
(RTFM)18” and “actual (CTFM)19”. Those were obtained through the processing of
“after-the fact” data, namely data extracted from CFMU “ALL-FT” files.
However, these views do not strictly reflect the traffic load situation that FMP / FMD
would have observed on the field, at different time events of the regulation life-cycle,
when deciding to activate, monitor, modify or cancel a regulation.
Indeed, flight plans are sent at different times by the AOs. Furthermore, slots are
frozen at different times (function of the EOBT of each aircraft). Also, traffic departing
from remote airports can be mixed with still on-ground traffic departing from closer
sources, etc.
In particular, some doubts could be raised concerning the results provided by the
analysis of the “regulated (RTFM)” vision. For instance, because of the “true revision
process” (ETFMS dynamic process), slots that are not frozen yet can be improved. In
addition, since the slots are frozen at different times (function of the EOBT of the
aircraft), one slot that is already frozen, can be reallocated to another aircraft still on
the ground, which hence benefits from the true revision process. The consequence is
a risk of “counting” twice the slot that has been reallocated (aggregation of data,
issued at different times).

16 “FTFM” & “RTFM” fields of the ALL_FT files


17 “Initial” reflects the status of the demand before activation of the regulation plan. It is computed with the
latest flight plan version, sent by each AO to the CFMU/IFPS.
18 “Regulated” reflects the status of the demand after activation of the regulation plan. It is computed with the

latest ATFM slot (CTOT) issued to the AO, by the ground regulation system.
19 “Actual” integrates the actual entry time of the flights in the regulated TV. It is computed with the Radar

Data sent by ACCs to CFMU/ETFMS.

23
EUROCONTROL Experimental Centre
ATFCM Performance and Declared Capacity Study

Therefore, the fast-time ATFM simulator – TACOT – has been used in


order to rebuild the slot allocation plan observable, at different time events, and for
our ATFM smoothing performance assessment model.
The model is presented in the annex 1.
The following section provides the results that were obtained so far.

5.2 Simulations results

A first set of simulations was launched in order to find out about the demand status
before the implementation of a regulation, as well as to derive initial conclusions on
the slot allocation process performance.
The simulation results were also confronted to results obtained with the processing of
“ALL-FT” data so as to “test” the quality of the observations obtained when using that
ALL-FT model.

9 First results: observation of the upcoming traffic load (initial demand),


as seen by the user before a request [Pred-]

Observation of the drivers of the decision-making process

For one day of simulation and one TV under analysis, the flow evolution is depicted
below, for the two observation models:
9 The one output by the simulator, at the following prediction time: just before
the implementation of the regulation, herein, 113 minutes19 prior to the period
start (Pred- curve): this is the demand, as figured out by the user when he or
she decides to take action;
9 The one that is derived from the ALL-FT file (FTFM profile): this is the
demand that is known “after the fact”, i.e. that integrates subsequent flight
plan cancellations or later flight plan submissions.

24
EUROCONTROL Experimental Centre
ATFCM Performance and Declared Capacity Study

FTFM < Pred-


FTFM > Pred-

It could be observed from the figure that the prediction simulated curve (Pred-) can
present significant deviations from the ALL-FT (FTFM) one. It could be derived the
following two trends:
9 An overestimation of the demand (FTFM < Pred-), for anticipation periods20
of up to 9 hours (540 minutes). For that this trend, our assumption (to be
confirmed by further analyses) is that the impact of flight plans subsequent
cancellations or modifications is predominant in the deviations observed
between the prediction and the “after-the-fact” data;
9 An underestimation of the demand (FTFM > Pred-), for anticipation
periods19 of more than 9 hours. For this trend, our assumption (to be
confirmed by further analyses) is that a significant proportion of the flight
plans has not been submitted yet, leading to the deviations observed between
the prediction and the “after-the-fact” data.
As the time of events is getting closer, the two curves naturally reconcile one
another.
Therefore, it could be shown from this example that the ALL-FT curve (FTFM) may
deviate from the curve that the FMP would have observed on field, when deciding to
create a regulation process.
In some cases, deviations are of significant amplitude, which lead to the following
question: Could we estimate the appropriateness or inappropriateness of
implementing ground regulations, only by the observation of the FTFM profile? In
particular, a decision-making criterion which will recommend that ground regulations

20 A time index is represented on the (X axis). The origin (Index “0”) corresponds to the time where the user is
taking the action of creation. Index “113” corresponds to the start of regulation period. The regulation process
is therefore activated at anticipation “113” minutes, i.e. 1 h 53 min.

25
EUROCONTROL Experimental Centre
ATFCM Performance and Declared Capacity Study

should not be activated for demand excesses lower than 10%, could be challenged if
the observation is only based on the use of the ALL-FT files.
The response to this question leads to push further the investigations about the
deviations between FTFM and prediction curves, and to launch an analysis for other
regulation cases.
As a preliminary result from the analysis, we found in particular that for 33% of the
studied sample of analysis (44 regulation cases); the FTFM curve was deviating
more than 10% in average with respect to the prediction curve (Pred-).
Such result tends to confirm the issues raised by the using of “FTFM” curves only:
frequent deviations of significant amplitude with what the FMP has observed on field,
when deciding the regulation settings.

9 Second class of observables: the upcoming traffic load, as seen by the


user just after a request [Pred+]

Smoothing performance of the Slot Allocation Process


Sample of analysis
They were 10 TVs under analysis, over 6 different days.
In order to get an initial overall evaluation of the ATFM regulations performance, we
launched a first analysis by aggregating all the results corresponding to one specific
indicator, derived from the measurement of “Traffic load fluctuations average amplitude”.

Paris CDG
(4 TVs)
Aggregated results (average amplitude
and duration of the traffic load
fluctuations)
Madrid
(1 TV)

Amsterd.
(2 TVs)

Lon. Gat.
(1 TV)

Reims
(2 TVs)

26
EUROCONTROL Experimental Centre
ATFCM Performance and Declared Capacity Study

Results analysis

In the analysis of the results, two different scenarios have been considered:
1. Scenario Pred(+): evaluation of the full sample of analysis (6 days of
simulation, 10 TVs), obtained with the data output by the simulator, at T+
prediction time, i.e. just after the activation of the user’s regulation request –
distribution output by the system;
2. Scenario ALL_FT (RTFM): evaluation of the equivalent sample (6 same days,
10 TVs) obtained with the processing of ALL-FT data (RTFM – regulated
profile), so as to measure the deviations in the results of this ALL-FT model
with the results obtained by simulations.

Traffic load fluctuations average amplitude is evaluated in proportion to the flow rate
(e.g. 0.05 means 5% above the flow rate, -0.05 means 5% below the flow rate).
An illustration is presented below. The curve represents the number of records (in
proportion to the total number) comprised in a] x-1%; x%] interval (step of 1%).

]-0.01; 0] interval:
- Scenario Pred (+): 43%

]-0.01; 0] interval:
- Scenario RTFM: 24%

]-0.05; 0.05] interval:


]-0.05; 0.05] interval: Scenario Pred (+):
Scenario RTFM: 87%
66%

The analysis clearly confirms that ALL-FT RTFM profiles under-estimate the
performance of the system. In the ALL FT scenario, only 66% of the traffic load
fluctuations average amplitude contained within an interval of ]-5%; 5%], whereas the
simulation shows that 87% of the distribution output by the system is contained within
that interval.

27
EUROCONTROL Experimental Centre
ATFCM Performance and Declared Capacity Study

When we strictly focus on the reference, i.e. the flow rate, the difference is equally
important: 43% of the traffic load fluctuations average amplitude is contained within
an interval of ]-1%; 0%] in the case of the Pred(+) scenario, 24% of the traffic load
fluctuations average amplitude is contained within an interval of ]-1%; 0%] in the case
of the ALL FT RTFM scenario.

It could be shown that the use of the RTFM files does not guarantee the quality of the
results, in the analysis of the performance of the slot allocation process. As a matter
of fact, RTFM data are altered by the real-time events occurring after T+ (as it takes
into account the latest slot issued to the AO, which can be improved in function of
subsequent real-time events, such as airborne deviations from the flight plan, etc.).

Therefore, performance should be better assessed through the observation of the


slot allocation plan, rebuilt at T+, i.e. just after a user’s request.

28
EUROCONTROL Experimental Centre
ATFCM Performance and Declared Capacity Study

6 CONCLUSION & NEXT STEPS

6.1 Conclusion: a segmented ATFM ground regulation performance


assessment

As it has been demonstrated by the previous study and by the interviews conducted
with the users of the ATFM ground regulation system, FMPs / FMDs have developed
methods of using the ATFM regulation process to fulfil a number of various
operational objectives that goes beyond the traditional objective of resolving the
hourly demand / capacity imbalances.
ATFM service users will still look for the elimination of the hourly over-deliveries, but
some will also look for increased traffic predictability, and others for the elimination of
the bunching peaks. Assessing whether the system is able or not to meet these
objectives therefore required to specify an appropriate ATFM performance
assessment matrix of indicators.
That matrix of indicators21 has the following axes of measurements:

9 Traffic load fluctuations amplitude and duration, to measure whether or not


the elimination of the uncontrolled loads of traffic (from an ATCO’s view point)
is effective when a regulation is applied;
9 Average traffic load deviation from the flow rate, to measure whether or not
the regulation has reshaped the initial distribution of traffic as closely as
possible to the flow rate. This evaluates how the available capacity is
efficiently used;
9 Dispersion around other references than the flow rate, to measure the
performance of the system according to “customised” criteria, i.e. to the
“tolerance” margin that the user has defined, while qualifying an excess as
acceptable (this can be e.g. of 0%, 10% or 20% depending on the user);
9 Flight time estimates deviation over TV entry points, to measure the
uncertainty reduction, and predictability improvement on the flight time
estimates over TV entry, when the flight is regulated.

In the four above categories of indicators, the three first categories are referring to
the smoothing performance of the ATFM ground regulation solution, and the fourth to
the uncertainty reduction performance.
It is also relevant to notice that in the class of the smoothing indicators, the indicators
are associated to different TV entry counting parameters depending on the
smoothing objective that is measured: 60 minutes (hourly over-deliveries resolution),
20 minutes (traffic bunching reduction) or 1 minute (prevention from the forming of
complex traffic clusters within the sectors).
In various cases, we could observe that a given regulation could be very efficient for
one given objective, when providing modest results for another objective. Therefore,
the regulation performance is to be assessed globally by the application of the whole
set of indicators reflecting the various users’ objectives.

21 A full description of the indicators developed (incl. formulae) is available on a referenced report
(Intermediate note on ATFM efficiency indicators).

29
EUROCONTROL Experimental Centre
ATFCM Performance and Declared Capacity Study

6.2 Next steps

The previous indicators have all an operational significance, as they are


representative of the added-value that the system is able to offer to the day-to-day
work of ATCOs, and as a more indirect result, to the capacity that those controllers
are ready to offer to the airspace users.
In particular, the present indicators set the scene for all the subsequent works that
would need to be undertaken in the assessment of the ATFM performance.
In the case of traffic smoothing, we shall firstly assess the properly achievable
capabilities of the slot allocation mechanism itself, by isolating the “external” source
of alterations born from the reintegration of Real-Time events into the system’s
functioning. In a second step, we would need to investigate into the “external”
sources of alterations and assess their impact on the systems’ performance.
In the case of uncertainty, we shall extend the analysis to other TVs, in order to look
at dependencies with the TV characteristics, such as the distances to the departure
aerodromes.
In any case, an extrapolation of the results already obtained, from our first application
of the indicators on a reduced set of cases, on a more extensive analysis of recorded
regulation cases is required in order to guarantee the quality of our results.

It shall be borne in mind that this “extrapolation” will certainly require the use of
simulators (TACOT, available in the EEC/NCD), in order to rebuild the slot allocation
plans “life-cycle” and to make an assessment of the slot allocation plan evolution with
regards to the events inherent to the functioning of the system (actions of creation,
modification, cancellation of a reg. process and also the continuous adjustments of
the plan triggered by the real-time events, e.g. airborne deviations to the flight plans,
etc.).

30
EUROCONTROL Experimental Centre
ATFCM Performance and Declared Capacity Study

ANNEX 1: ATFM SMOOTHING PERFO. ASSESSMENT


(METHODOLOGY)

TRAFFIC DATA ANALYSIS - METHODOLOGY

The measure of the ATFM smoothing performance requires extracting and


processing CFMU raw data from the archive database. A structuring model has
therefore been developed in order to isolate the relevant data subsets for each
regulation case.
The model is based on:
9 the selection of the appropriate observables for the building of a slot
allocation plan observable;
9 an analytical model where different traffic pictures are drawn, for different
statuses and/or at different key events of the slot allocation plan life cycle.
The model indicates how the ATFM performance is evaluated according to
these pictures and using the indicators.

1. Selection of the appropriate observables


The analysis of ATFM ground regulation effectiveness requires studying the slot
allocation plan “life-cycle”, i.e. for different statuses and/or at different key events, in
order to make an assessment of the slot allocation plan evolution with regards to the
regulation settings (flow rates, regulation periods).
In order to reconstitute the slot allocation plan that would have been observed on
field and produce the results required for the analyses, we use an ATFM fast time
simulator – TACOT, available in the EEC/NCD.
Note: In order to work without having to use a simulator, another option would be to
build observables directly available from the processing of “after-the-fact” data,
namely data extracted from CFMU “ALL-FT” files (initial22, regulated23 and actual24
flight profiles). Nevertheless, and at the present stage of our work, we have observed
that the conclusions on the ATFM smoothing performance may be altered
significantly by the use of the “ALL-FT” files (as described in the § 5 of this
document).

22 “Initial” reflects the status of the demand before activation of the regulation plan. It is computed with the
latest flight plan version, sent by each AO to the CFMU/IFPS.
23 “Regulated” reflects the status of the demand after activation of the regulation plan. It is computed with the

latest ATFM slot (CTOT) issued to the AO, by the ground regulation system
24 “Actual” integrates the actual entry time of the flights in the regulated TV. It is computed with the Radar

Data sent by ACCs to CFMU/ETFMS

31
EUROCONTROL Experimental Centre
ATFCM Performance and Declared Capacity Study

• Time events:

The objective is to build the views that FMP / FMD would observe on the field, at
different times events of the regulation life-cycle, when deciding to create, modify or
cancel a regulation.
For each event – activation, modification or cancellation – two time events are
considered:
9 Event T-: latest flight list version available in the system before the creation,
modification or cancellation of the regulation process;
9 Event T+: flight list version output by the system, just after the creation,
modification or cancellation of the regulation process.

The “T- view” reflects the status of the demand, at a time when a decision for
regulating and for setting the regulation parameters is made by the FMP / FMD.
The “T+ view” reflects the status of the demand, output by the system and according
to the regulation settings. At this time, ideally, the traffic load should be aligned with
the flow rate(s) requested.

In order to identify the different time events, an analysis of regulation reports –


extracted from the “Oplogs” files available within the CFMU archive database – is
performed.

• Extraction of the relevant regulation observables: TACOT

In the EEC, TACOT is available for fast time ATFM simulations. TACOT integrates
an implementation of the CASA slot allocation algorithm. We are able to generate,
with this tool, slot lists at the different time events here above identified (T-; T+) and
reconstitute the slot allocation plan, that would have been observed in operations by
FMPs / FMDs.
The following data, as required for the analyses, is extracted from the simulations:

9 The “Flight List” status of each TV under study, and at the different times
here above mentioned (T-, T+ for the processes of creation, modifications
and cancellation).
The “Flight List” reports time estimates. For each flight of the TV, it gives the
information related to the latest flight status, i.e. ETO25 if the flight is not
regulated, eventually CTO26 if the flight is regulated and eventually ATO27 if
the flight has already taken off.

9 The “Flight List” status of each TV under study, at the end of simulation
day. Such “Flight List” is equivalent to the “CTFM” view (real flown profile)
extracted from the processing of “ALL-FT” files.

25 The ETO is the expected time of penetration within the TV, when the flight is not regulated. It is obtained
from flight plan sent by the AO to the CFMU.
26 The CTO is the expected time of penetration within the TV, when the flight is regulated. It is deduced from

the slot (CTOT) allocated to the flight.


27 The ATO is the real time of penetration within the TV. It is obtained from radar data (CPR – Correlated

Position Reports)

32
EUROCONTROL Experimental Centre
ATFCM Performance and Declared Capacity Study

For each regulation case, these data aim to build different traffic pictures before and
after an event of creation, modification or cancellation of the regulation process. The
data obtained at the end of simulation day is used for the building of the real traffic
pictures, as perceived by the controller on his working position.

2. Analytical model

The analytical model is granted on the processing of each “Flight List” (TACOT
outputs), at the different time events here above identified: T-, T+. Each of these
events triggers ATFM information updates (creation, modifications or cancellations).
Different traffic pictures are drawn before and after the time event, and compared to
the regulation settings.

Reg. parameters:
Flow Rate, Period

Initial Reg. Settings


(Creation)

Pred - (T-) Slot Allocation System Pred + (T+)


(CASA)

Reg. settings update


Real time events Flight Plans update (system inputs)

Real-time traffic (uncertainty)

Actual (ACT)

33
EUROCONTROL Experimental Centre
ATFCM Performance and Declared Capacity Study

Pred - (T-), Pred + (T+) and ACT (Actual) reflects predictions at different statuses and
anticipation levels:

9 Pred - (T-): Flight List status before activation of a regulation process


(creation / modification / cancellation);

9 Pred + (T+): Flight List status after activation of a regulation process (creation
/ modification / cancellation). It reflects the status of the demand after action
of the system, according to the regulation settings;

9 ACT (ACT): Flight List status at the end of the day. It integrates the actual
entry time of flights in the regulated TV.

The Real-time events are reintegrated into the system and are a source of alteration
of the distribution initially output by the system (Pred + (T+)). Real-time events can be
grouped into the following categories:

9 Reg. settings: The evolution of the regulation settings (subsequent


modifications to the initial settings).
This category refers to all the subsequent modifications to initial reg. settings.
They may have an incidence on the performance of the slot allocation system
if the anticipation for the activation of such modifications is too short (system’s
inertia to adapt traffic distributions to updated settings);

9 Flight plans: The uncertainty vis-à-vis the initial demand profile in the
regulated TV.
Part of the uncertainty on the traffic loads also results from the late filing and
sending of flight plans to the CFMU. Also, a significant proportion of the flight
plans is subjected to subsequent modifications and cancellations, leading to
modifications of the initial demand profile.

9 Real-time traffic: The real-time deviations from the flight plan (also evaluated
through the analysis of flight time estimates on TV entry deviations, as
detailed in the “uncertainty reduction” sections of this document).
The category refers to the operational deviations from the flight plan, inherent
to the real-time execution of flights (pilot / ATC instructions).

34

Você também pode gostar