ble of Contents
AIR TRAFFIC CONTROLLER MANPOWER PLANNING MODELLING
A learning platform for understanding the staffing process.
Mr. Cees Niesing
Human Resources Management Expert
Human Factors and Manpower Unit, EATMP, EUROCONTROL
EUROCONTROL Agency, Rue de la Fusée 96, 1130 BRUSSELS, BELGIUM
Tel.: +32 2 729 3336 Fax: +32 2 729 9149 E-mail: cees.niesing@ eurocontrol int
If you always do what you always did, you will always get what you always got.
Abstract
Air Traffic Controller (ATCO) shortages seem difficult to overcome worldwide. Some aspects of
manpower needs are probably considered at too late a stage or their impact is underestimated. An
ATCO Manpower Planning (MP) simulation model (LAMPS) has been developed to assess the
overall MP process in relation to air traffic demand over a period of 15 years. LAMPS enables the
various players/decision makers to increase awareness of the impact of MP policy decisions in the
long term, thus providing a learning platform for improved decision making.
Introduction
The main objective of Air Traffic Controller (ATCO) Manpower Planning (MP) is to ensure the
provision of the right number of ATCOs, with the right qualification, at the right time and in the
tight place.
As simple as it reads, it has proved difficult to reach this objective in practice, given the overall
perceived shortage of ATCOs, as the following results from surveys show, in Europe and in fact
worldwide.
In 1993, the Select Group for Assessing Manpower Issues (SGAMI) reported an average shortage
of ATCOs in 20 European States of 9% in 1993, 7% in 1994, 5% in 1995, 2% in 1996 and a surplus
of 0.4% in 1997 (EATCHIP 1993).
However, in 1997 an update of the SGA MI survey of 1993 showed a real shortage overall over the
period 1993 - 1997 of 9%, a forecast over the period 1998 - 2002 showed an overall shortage of
3.5% (EATCHIP 1998c).
In 2000, in a survey on required/available ATCOs, 18 European States reported ATCO shortages of
13% in 2000, 12% in 2001, 10% in 2002, 7% in 2003 and 6% in 2004 (EATMP 2000). This result
cannot be directly compared with the previous surveys, as the composition of participating States is
slightly different, but the message remains the same: the real gap between what is required and what
is available was either not detected or was in fact not expected.
Depending when it is measured, it is nevertheless astonishing how different the perception is
between required and available ATCOs. To a certain extent, these shortages can be translated into
reduced time for training activities and increased air traffic delays, thus putting pressure on the
current ATCO workforce. Possibly deteriorating working conditions and increasing delays even
further, making it a vicious circle. What does it reveal about the MP process? Is it too short-term? Is
the MP process too static? Are the training possibilities so low that making up the shortfall is not
feasible?
nth Intamatianal Canfaranca nf tha Cwctam Dimamice Caniahr Dalarma — Tly 207 1
A Different View on MP
The desired outcome of MP is to provide a sufficient number of qualified personnel for air
navigation services on a timely basis, to ensure the safe, orderly and expeditious handling of air
traffic in the first place and, secondly, to provide personnel for all other ATCO tasks. Even the posts
considered necessary to be filled by staff with an ATCO background have to be taken into account
when assessing overall ATCO requirements. The process by which this outcome is achieved can be
expressed as dynamic decision making.
Brehmer (1990, 1992) considers the dynamic decision making process as the process of achieving
control over a system in order to produce a desired outcome. In order to control the system,
decision-makers must have a mental model of the dynamics they seek to control. Mental models are
the mechanisms whereby humans are able to generate descriptions of the system purpose, explain
system functions, and predict future system states. One conclusion from research and practice is that
people can make better (and more informed) decisions in a dynamic environment, when they have a
more complete model of the system and thus know better how the system is composed and how
parts of the system are linked together and in fact interact.
Many problems concemed with MP are not easily described in written documents and thus are of no
teal help. The documents themselves tend to match in complexity the problems they are supposed to
solve. The reason is that the complexity and dynamics involved in planning, and how in practice to
cope with both, cannot be presented in a static manner.
Another way of dealing with the complexity and dynamics in ATM is using a modelling tool to
conceptualise, document, simulate, analyse and optimise a MP model. The aim of the developed
model, LAMPS (Long-term ATCO Manpower Planning Simulation), is to provide greater insight
into the behaviour of different events within MP and to explore the causes and effects that lead to
these events. All management levels, from the Director managing a Centre, to the individual ATCO
managing air traffic from the position, can learn from and create a better understanding of the
system they are handling if they would get a deeper insight into the processes and the behaviour of
the interlinked sub-systems in MP.
With LAMPS the manpower planner(s) (MPs) can provide such a leaming environment, in the
sense that all players in the MP process, normally having only a limited view of the system as a
whole, learn to view the system as a whole, identifying their part and understanding its impact on
the system.
Modelling Tool
LAMPS is built using dynamic modelling software called VENSIM (VENtana SIMulation
environment)! and SABLE? front-end software. The software allows the analyst to execute a
complete System Dynamic (SD) project from initial set-up through to simulation and analysis. The
developer is provided with a powerful array of graphical and analytical tools to facilitate the tracing
of system behaviour to causal influences and, ultimately, input the necessary data under control of
the user. A model consists of various views of the system. Diagrams are constructed for each
element of the view, describing its relationship with other elements. The relationships are further
quantified by equations.
Verification and Validation
'VENSIM™ isa trademark of VENTANA SYSTEMS, INC., HARVARD, Massachusetts.
> SABLE™ is a trademark of VENTANA SYSTEMS UK Ltd., Eastleigh, Hampshire, UK.
nth Tntarnatianal Canfaranra af tha Cvetam Dimamice Caniahr Dalarma — Inky 27 2
Coyle (1996) defines validity of a model as “well-suited to its purpose and soundly constructed”.
His key philosophy is that “the model should do the same things as the real system and for the same
reasons”. LAMPS has been tested by three European ATM providers to verify this. These tests
stated that the LAMPS tool reflects the current manpower planning practise and is consequently
well suited to one of its purpose to mirror and understand the current practices employed in ATCO
MP. As there are no objective standards with regard to MP, validation has to be done by using the
tool over time and to establish its suitebality for the purpose of achiving a better balance between
ATCOs required and available at any time. Once validated the model will be further developed into
a decision making support tool.
From Static Descriptions to Dynamic MP Modelling
The purpose of this modelling is to provide the decision maker with a better understanding of the
teal problems he or she has to cope with, and not to replace the decision maker, or make a complex
reality even more complex.
Air Traffic Demand
In simple terms, the airspace entrusted to an ATS provider is divided into control sectors containing
consoles for an executive control position, a planning control position and, if necessary, an assistant
control position. From these positions ATCOs organise a safe, orderly and expeditious flow of
traffic. Consequently, the air traffic demand drives the number of positions to be manned and the
number of hours control positions must be open. These opening hours for positions can vary from
day to day, from weekday to weekend day and from season to season. There are occasional irregular
peaks in demand, which occur at the start and end of the holiday season, before and after a
European Cup final, etc.
Based on the actual opening and closing times of sectors, various OR (Operational Requirement)
profiles can be derived. Normally the OR over a certain period of time is calculated from the
average opening and closing times, rather than the peak requirements. Once the OR has been
established, it forms the basis for the calculation of the number of ATCOs needed. It must be
emphasised that this basis is the most important factor in MP, as it is the comerstone of the
manpower need. From the viewpoint of cost, it should be established in an appropriate manner.
Calculation of Required Manpower
In Annex A of the deliverable “ATS Manpower Planning in Practice: Introduction to a Qualitative
and Quantitative Staffing Methodology” (EATCHIP, 1998c), two calculation methods for the
Tequired manpower are explained in detail. As well as these two methods, other approaches exist. It
will be a challenge for ATS providers throughout the European Civil Aviation Conference (ECAC)
tegion to arrive at a clearly understood transparent and effective way of calculating the OR and
hence the required manpower.
LAMPS integrates three approaches in MP; one by which the manpower required is driven by the
traffic demand and its growth over time, the other by which the manpower required is driven by the
OR, adjusted over time if the increase in air traffic demand reaches a certain point at which more
ATCOs should be available and the manpower required by the number of duties as reflected in a
duty roster. Each approach can be considered separately. The dynamic change in traffic demand
over time and its consequent effect on the number of ATCOs required is mapped on to the number
of ATCOs available. This availability is determined by the in-flow, through-flow and out-flow of
ATCOs.
The following Figure 1 shows the relation between air traffic demand and the way the requirement
of ATCOs is determined from that demand in the current LAMPS model.
nth Tntarnatianal Canfaranra af tha Cvetam Dimamice Caniahr Dalarma — Inky 27 2
Air Traffic and ATCOs Required
amumal traffic Bask ”
growth percent—_, initial traffic in each sector
monthly traffic
anmal traftie-—" Hn a OSE altel wncaib lly tall
con — ‘ile
before seasonality
igo ae monthly traffic deseasonalised initial
monthly weight to before seasonalit = monthly traffic in sector
traffik
meme ———
monthly traffic traffic in sector
in sector
traffic seasonality
owitch ta daily traffic
im sector,
ATCO hours required
* Swe: icone manning for traffic demand
hours per day - per pe
Traffic
Initial tafe hanaeig™™ DAY rainimmma oe
capability
manning switch
initial time leakage allowangetime leakage
percent allowance ATCOs required
Figure 1. Air Traffic demand and ATCOs required.
Traffic is controlled by so-called Sector Groups, which consist of a number of individual Sectors
manned by an executive ATCO and, nomally, a planning ATCO. The initial monthly traffic for all
Sector Groups has consequently to be adapted for the part of traffic flying through more than one
Sector Group and for seasonality.
Taking that into account, the daily traffic in a Sector Group averaged out over a 24 hours period is
determined. Based on that it is determined how much ATCO hours are required, in conjunction with
the Traffic Handling capability - i.e. the number of aircraft movements a Sector Group can handle -
to cope with that traffic on a monthly basis.
Depending on the “manning switch” it is calculated how many ATCOs are required on a yearly
basis if:
- the number of ATCOs required should follow the air traffic demand,
- the number of ATCOs required should follow the earlier explained Operational Requirement or
- the number of ATCOs required should follow the number of duties to perform each day.
The following Graph 1 shows the result for an imaginary Sector Group 1.
nth Intamatianal Canfaranca nf tha Cwctam Dimamice Caniahr Dalarma — Tly 207 A
Graph for ATCOs required
200
150
100
50
0 18 36 54 72 90 108 126 144 162 180
Time (Month)
ATCOs required[Sectors 1]: SDS person
ATCOs required[Sectors 1]: SDS2 person
ATCOs required[Sectors 1]: SDS1. — — — person
Graph 1: ATCOs required for Sector Group 1
The (green) striped line shows the requirement for ATCOs following the development of air traffic
demand (in this example increasing with 5% per year).
The (red) dotted line shows the requirement for ATCOs following the Operational Requirement, the
number of control hours needed being automatically adjusted by the model.
The (blue) straight line shows the requirement for ATCOs following the number of duties in the
monthly duty roster needed to cover the air traffic demand.
The latter doesn’t change over time, as it is constant and can only be changed via gaming or by
entering a time series of different number of duties as external data.
As the training from Ab initio ATCO to fully licensed ATCO may take three to four years, it is very
important to know the requirement of ATCOs compared to the available ones over a longer time
period.
The methodology of determination of the required number of ATCOs over time is still under
consideration by various Air Traffic Service Providers, and the way demonstrated here is an
approach to stimulate awareness and discussion.
Balancing In-flow, Through-flow and Out-flow
For control systems to be functional the relationship between what is registered as input and
assessed as output has to be clarified. In general four classes of functional relationship exist:
Deterministic: the interaction between input and output can be defined precisely. Examples:
number of hours worked, cost of employment.
Probabilistic: a complex linear or non-linear relationship exists that fluctuates randomly across
a range of values. Examples: the relationship between the number of ab initio trainees accepted
nth Intamatianal Canfaranca nf tha Cwctam Dimamice Caniahr Dalarma — Tly 207 5
by an ATS Academy, the number of licensed controllers that emerge with a rating, due to the
probability of failing during training.
Dynamic: multiple specific inputs have to be present before an output can at all be determined.
Very often there will be a time lag before the ‘true’ relationship becomes evident. Example:
deterioration of working conditions and social climate may influence the perceived outflow of
ATCOs, e.g. resignations and/or requests for transfer.
Static: data like the number of training places, On-the-Job training possibilities, in general, data
that does not change over time or is only occasionally adjusted.
All four relationships need to be considered together to balance over time the required number of
ATCOs with the available number of ATCOs. Any discrepancy between the two is reflected in
either a controller surplus, which, if too high, might be considered uneconomical, or controller
shortage, which might lead to overtime worked, delayed traffic and, if delay exceeds a certain level,
unsatisfied traffic.
The in-flow, through-flow and out-flow of ATCOs is modelled in LAMPS in a similar way as the
ATCO requirement part, taking into account issues as;
Training cycles,
ATCOs availability, after leave, training, sickness, other absences from the Operations room,
the probability of matemity and paternity leave,
the probability of outflow through invalidity and death,
part-time work,
yearly distribution of leave and sickness days,
age distribution,
retirements,
etc.
In the context of this paper it would go too far to try to show all the other views of the LAMPS
model in which the in-flow, through-flow and out-flow is modelled.
An example
As traffic grows, ATCO shortages may arise because the number of ab initios that can enter
academy training is constrained by the number of training places available or because insufficient
positions are available in the simulator or OJT. The question to address is WHEN and HOW
MUCH additional places or positions need to be supplied.
The following assumptions are made in this example:
- 15 places are available for each intake entering academy training.
the period between intakes is 7 months.
12 position possibilities in the simulator.
3 Sectors Groups are in use, and 4 people can receive OJT in Sector Group 1, 4 in Sector Group
2, and 2 in Sector Group 3.
Traffic is growing at a rate of 5% per annum.
It is assumed here, for ease of demonstration, that all other system variables have been widely
discussed and considered viable.
LAMPS can be run to show how the monthly ATCO margin changes over a period of 15 years. The
margin is the difference between the number of ATCOs employed and the number required to
handle the traffic volume. It becomes negative when there is an ATCO shortage. The following
nth Tntarnatianal Canfaranra af tha Cvetam Dimamice Caniahr Dalarma — Inky 27 a
Graph 2 shows the situation using the assumptions listed above. The ATCO margin has been
aggregated across the three sector groups. It shows that the number of ATCOs employed meets the
requirement at the beginning of the simulated period but that a shortage develops after eighteen
months, which grows further as the traffic volume grows.
150 ATCOs
300,000 Aircraft
75 ATCOs
225,000 Aircraft
0 ATCOs
150,000 Aircraft
-75 ATCOs
75,000 Aircraft
-150 ATCOs
0 Aircraft
0 24 48 72 96 120 144 168
Month
ATCO margin ATCOs
Monthly traffic Aircraft
Graph 2 - Monthly traffic and ATCO margin
Please note that the situation in Graph 2 would arise if one would continue as described in the
assumptions. LAMPS can now be used to create scenarios and simulate them to avoid the shortage.
Result
160 ATCOs
48 Course places
80 ATCOs
36 Course places
0 ATCOs
24 Course places
-80 ATCOs
12 Course places
-160 ATCOs
0 Course places
0 24 48 72 96 120 144 168
Month
ATCO margin ATCOs
Places filled Course places
Places offered Course places
Graph 3 - Two levels of institutional course places offered and filled
As shown in Graph 3 above, the result is that by running various scenario’s and probing various
combinations of training places/positions the following scenario seems to hold: increasing the
number of training places to 27 for three years, increasing it then to 36, reducing it to 28 after 8
years for the remainder of the simulation; increasing the number of simulator positions to 18 after
two years and to 26 positions after 5 years and increasing the number of OJT positions to 5,6,5 after
thirty months. This scenario provides a good balance around the ATCO margin zero line.
nth Tntarnatianal Canfaranra af tha Cvetam Dimamice Caniahr Dalarma — Inky 27 7
LAMPS asa Learning Platform
It was also recognised that to implement the LAMPS tool in an ATM organisation and to realise the
maximum benefits of providing a "modelling and simulating for learning" platform required that the
tool is embedded within a collaborative process, thus, enabling an improvement in how ATCO
manpower planning issues are discussed and to avoid isolated solutions been taken. In this context
see the paper of Una Mellett, “A Strategic and Integrated Approach to Human Issues in Air Traffic
Management”, discussing this collaborative approach called CHAMP (Collaborative and
Harmonised ATCO Manpower planning Process).
Benefits
What we have achieved in this new approach for ATCO staff planning is a shift from a number
crunching, isolated, short term, reactive process to an integrated, strategic approach embedded in a
collaborative process which brings all players together to plan and decide on solutions to staffing
issues. The main benefits of this approach are:
LAMPS by modelling the flow of ATCOs through the ATM system offers a framework to
communicate about dynamic complexities and interdependencies of the many variables
influencing ATCOs required with ATCOs available.
Focusing on the internal structure of the processes for ATCO in-flow, through-flow and out-
flow enhances the possibilities of identifying the root causes of problems. This should result in
solutions, which can have long-term positive impact rather then short-term ineffective quick
fixes.
Simulating scenario’s enables the many players involved in staff planning to debate the
implications of policy change and to link an ATM’s performance (e.g. meeting traffic demand)
to its operating structure (e.g. staff planning processes).
CHAMP by providing staff planners with skills and processes to work with other players
involved in staff planning is building a shared understanding of strategic issues in staffing by
closely examining and discussing "what if" scenario’ s.
Thus, LAMPS through the CHAMP process enables the players involved in staff planning to
structure informed debate about staffing issues. Therefore, the long-term benefit is that decision
making on staff planning issues within ATM organisations will improve.
The introduction of this approach during an information session for Air Traffic Management
Organisations over Europe in November 2000 in Brussels has attracted a lot of interest and resulted
in pilot projects for implementing this approach in Denmark, the Netherlands, the CEATS States
(ie., Austria, Bosnia and Herzegovina, Croatia, Czech Republic, Hungary, Italy, Slovak Republic
and Slovenia) and the Maastricht UAC of EUROCONTROL.
Linked to these pilots is an overall review of the model in order to adapt in a generic way to the
slight differences in the ATCO Manpower practices in the organisations mentioned above and to
establish a better defined validation process than the one mentioned before.
The results of these pilot projects are envisaged to be available in the beginning of 2003.
nth Tntarnatianal Canfaranra af tha Cvetam Dimamice Caniahr Dalarma — Inky 27 a
Conclusion
In conclusion, LAMPS, embedded in a collaborative process, provides a learning platform to enable
all players in the MP process to involve themselves actively in this process, where pro-active and
more integrated MP policies can emerge. This should improve the decision making process for
ATCO MP and allow for a better balance between anticipated ATCO requirements and availability.
This also is an example where experts, covering a variety of expertise, can come together to give
their inputs into the LAMPS model, to widen its use and applicability for the overall benefit of the
ATM community.
References
Brehmer, B. (1992). Dynamic decision making: Human control of complex systems.
Acta Psychlogica 81, 211-241.
Coyle, R.G. (1996). System Dynamics Modelling: A Practical Approach.
Chapman and Hall:London.
EATCHIP Human Resources Team (1996). Report on Issues in ATCO Manpower Planning.
HUM.ET1.STO3.1000-REP-01 Brussels : EUROCONTROL.
EATCHIP Human Resources Team (1998b). Heuristic Modelling and Policies in ATCO Manpower
Shortage or Surplus Management. HUM.ET1.ST03.1000-REP-02 Brussels: EUROCONTROL.
EATCHIP Human Resources Team (1998c). A Systems View of Manpower Planning and
Management. HUM.ET1.STO3.1000-REP-03 Brussels : EUROCONTROL.
EATCHIP Human Resources Team (1998d). ATS Manpower Planning in Practice: Introduction to
a Qualitative and Quantitative Staffing Methodology. HUM.ET1.STO2.2000-REP-01
Brussels : EUROCONTROL.
Curriculum-Vitae
Cees Niesing is a Human Resources Expert. He started off his career in 1964 in the Dutch Royal
Air Force where he was involved in financial and logistic matters. He joined the Maastricht UAC of
EUROCONTROL in 1973 where he has been working in the Personnel Section until he joined the
Human Factors and Manpower Unit of EATMP (European Air Traffic Management Programme) in
January 1997. He is the Work Package Leader for advanced MP Methods and Tools in the
Manpower Sub-Programme of the Human Resources Programme.
Back to the Top
nth Intamatianal Canfaranca nf tha Cwctam Dimamice Caniahr Dalarma — Tly 207 a
