1

Simulation-Based Turnaround Evaluation for Lelystad

Airport

Miguel Mujica Mota (a)*, Geert Boosten(a), Nico de Bock (a), Edgar Jimenez(b, c), Jorge Pinho
de Sousa(c)

(a)
Aviation Academy, Amsterdam University of Applied Sciences,
1097 DZ Weesperzijde 190, Amsterdam, The Netherlands

(b)
Universidad de Ibagué, Facultad de Ingeniería
Carrera 22, Calle 67, 730002, Ibagué, Colombia

(c)
INESC TEC and Faculdad de Engenharia da Universidade do Porto
Rua Dr. Roberto Frias, 4200 – 465, Porto, Portugal

* Corresponding author: m.mujica.mota@hva.nl

 2

ABSTRACT

The airport of Lelystad in North Holland will be upgraded to attract commercial traffic from Schiphol. In this paper
we present the simulation-based analysis for Lelystad Airport with the objective of identifying the most promising
configuration, identifying potential problems and capacity limitations in the system. Three layouts for the apron were
tested and we analyzed in the model the use of vehicles for the ground handling service, different demand levels and
different allocations for aircraft, in addition we included the uncertainty inherent to these systems. The results allowed to
get to the conclusion that some configurations are more attractive than others but the variability of the system might play
an important role in order to make the airport more or less attractive to commercial airlines.
Keywords: airport simulation, ground operations, turnaround time, airport performance.

1. INTRODUCTION AND BACKGROUND

The global air transport industry constantly faces the lack of capacity, particularly at major European
airports (Roosens, 2008; ACI Europe, 2010). Airports cope with the problem by developing new infrastructure
(e.g. new runways or new gates), or by optimizing existing resources in order to improve the efficiency of the
various processes involved in airport operations. Many times building new facilities seem the most logical
solution, but it is expensive, time-consuming and cannot be taken for granted in practice. Improving efficiency
is challenging, but a task in which techniques from operations research, like simulation, come handy.
Amsterdam Schiphol (AMS) is the main airport in the Netherlands and it was the fifth busiest airport in
Europe in 2014 in terms of passenger traffic (ACI Europe, 2014). AMS is also the main hub for KLM, which
provided 54% of the seats in 2013; and a major airport for the SkyTeam alliance, whose members – including
KLM – are responsible for 66,3% of the airport traffic in terms of air traffic movements (ATM) (Schiphol
Amsterdam Airport, 2015). Its role as a hub, called “Mainport” by airport management and the government, is
central to the airport strategy, especially considering the small size of the domestic market in the Netherlands
and the airport’s role as economic engine for the region.
Due to environmental reasons, the capacity at AMS is limited to 510.000 ATM per year. In 2014 there
were 438.296 movements at the airport, 86% of the imposed cap (Schiphol Amsterdam Airport, 2015). As a
result, the airport operator, Schiphol Group, would like to support the “Mainport” strategy by redistributing
traffic which does not depend directly on the hub function of the airport (mainly LCC) to other airports in the
Netherlands in order to relieve capacity at Schiphol. The preferred alternative is to upgrade Lelystad Airport
(LEY) to attract flights to European cities and regions with focus on tourist destinations.
In recent years LCCs in Europe have focused on short-haul point-to-point leisure traffic. More recently,
nonetheless, they have been targeting business travellers more actively, and some of them even offer interline
connectivity using simple hub structures (Jimenez, 2015). This means that the development process at
Lelystad should consider not only the type of passengers and airlines the operator wishes to attract, but also
the performance parameters the airport should have in order to become attractive for them. Hence, it is
important to develop tools to benchmark future performance indicators so that the airport operator can match
the proper infrastructure with the intended strategy, in order to assess the success of the project.
Lelystad is currently the largest airport for general aviation traffic in the Netherlands. It is located 56 km
from central Amsterdam, about 45 minutes by car to the East. The airport is fully owned by Schiphol Group,
which also owns Rotterdam airport (RTM) and a 51% stake in the Eindhoven airport (EIN), both in the
Amsterdam Multi-Airport System (see Fig. 1).
In relation to Eindhoven (and also to Groningen, in the northeast of the Netherlands, and Maastricht in the
south), Lelystad is considerably closer to Amsterdam and thus better located to serve as a secondary airport
for the city. In terms of distance, Lelystad airport is also closer to Amsterdam than Rotterdam airport, but
considering available connections by train and car, travel time is not so different.
 3

Fig. 1. Location of Lelystad (LEY) in the Amsterdam Multi-Airport System.

The ambition to divert short-haul non-hub-related traffic, “with focus on tourism destinations”, to
Lelystad implies a stronger focus on the airlines that are able to deliver such type of traffic. In order to attract
airlines, especially LCCs, Lelystad would need to provide the following differentiation factors: availability of
slots; low aeronautical charges; incentive programs; and quick aircraft turnaround (Jimenez, 2015). Available
slots are also crucial for airlines to start new services at times that match their network configuration and are
attractive to passengers. At Lelystad the availability of slots can be hampered by general aviation traffic, in
case it is not diverted to other airports after upgrading the infrastructure for commercial use; and by the
possible conflicts with air traffic in approach and departure trajectories at Schiphol Airport.
Turnaround time (TAT), is the time measured from the moment the aircraft parks at the correspondent
stand until it is ready for taxing out towards the runway. This TAT depends a lot on the operative conditions
of the airport and the efficiency of the ground handler. Ensuring quick aircraft turnaround is essential to attract
the desired mix of airlines, but it is also crucial to better manage the capacity of the entire Multi-Airport
System. The following sections present the characteristics of the simulation model developed to evaluate TAT
performance of Lelystad airport considering different scenarios, as well as the results provided by the model.

2. METHODOLOGICAL APPROACH
Nowadays, simulation and optimization techniques are used in industry to deal with the decision making
activity by searching optimal or feasible solutions to real problems. The use of both, simulation and
optimization techniques facilitate the design and assessment of strategies reducing the risk of poor outcomes.
Simulation models have proven to be useful for examining the performance of different system
configurations, and alternative operating procedures for complex logistic or manufacturing systems, among
many applications (Longo, 2013). However, in the aviation industry the use of simulation has been
traditionally associated with the training for pilots via flight simulators. The analysis and improvement of
operative performance via these methods has been put on the scope recently by the scientific community. To
this end, the work of Wu and Caves (2004) is a good example for the use of Markov chains together with
Montecarlo simulation for investigating turnaround performance. For the same objective, the work of Adeleye
and Chung (2006) is relevant since they use discrete event modelling for understanding the bottlenecks in
turnaround processes. In addition, the research from Norin et al. (2012) use an optimization algorithm for
finding a sub optimal scheduling and then they create a discrete event system approach for testing the solution
 4

for the turnaround as well. Other work that is also relevant is the one from Okwir et al. (2017) which does not
use simulation but data analysis together with a collaborative decision making approach highlighting the
importance of improving the turnaround process in airports.
For other operations of the airport system, the work of Mujica (2015) and Mujica et al. (2014) put focus
in terminal buildings and the development of the initial airside models for small airports. Furthermore, the
work of Zuniga et al., (2011) take the perspective of the air traffic controller for improving the throughput in
the terminal maneuvering area of an airport using simulation and optimization.
It should be clear that the advantages and potential of simulation techniques are increasingly recognized in
a wide range of activities. Basically, simulation provides an environment to study the dynamic behavior of a
system under different operating conditions, using continuous, discrete or hybrid models to represent it (Banks
et al., 2010). There are different modeling techniques such as system dynamics, agent technology or discrete-
event systems (DES). System dynamics is used for systems in which the state variables change continuously
in time such as the level in a tank; agent technology is a relatively novel approach in which the power of
computers are used to simulate independent behavior and decision making of the entities within a system
(Becu et al., 2003); and DES are suitable for analyzing systems in which the entities follow a recognizable
sequence of processes interlinked between them and the state variables change at particular instants of time
based on the evolution of those processes. The states in DES do not change continuously but, rather, because
of the occurrence of events. This makes the resulting models be asynchronous, inherently concurrent, and
nonlinear, rendering their modeling and simulation different from that used in other approaches. Additionally,
the DES approach makes the simulations run much faster than the case of models developed using agent-
based technology which make DES a good approach for the development of decision-support tools in diverse
airport areas.
2.1. Model characteristics
For the case of Lelystad Airport, the authors chose between agent-based technology and DES and
decided for DES since the processes that occur in an airport follow specific sequences of operations thus the
use of agent based would not provide more value than the one gotten from DES instead it would make it more
complex. In addition, the software used, allowed the development of it through a bottom-up approach. Hence,
the simulation model for LEY is dynamic, stochastic, and asynchronous. In this way, the model is useful also
to identify potential problems in the airside of the airport as well as emergent dynamics between the model
components. Furthermore the tool selected for the development of the model (SIMIO) allowed performing not
only a bottom-up approach but also a modular one, which means the developer can focus on developing
different parts or elements of the system independently and then integrate the different models in order to have
a final one composed by the different modules.
In order to understand the potential problems for the airport in the future, it is necessary to use the model
for evaluating what the response of the system is to the different inputs (internal and external). Fig. 2
illustrates the approach for evaluating the model response.

RESTRICTIONS
Separation minima Speed limitsCross Wind Limits

INPUT OUTPUT
Volume of traffic /Traffic Mix Turnaround time
Number of Vehicles
Delay Level
Configuration of Apron
Airport Model
Total Time in Apron
Allocation of Aircraft

Fig. 2. Black box approach for the airport model.

 5

The input variables are the ones mentioned in Fig. 2 and they will be explained in the following
subsections. The main output will be the turnaround time, the total time the aircraft spend in apron and the
difference between the end of the turnaround time and the total time in apron which might be an indication of
delay as it will be explained in the following sub sections.
2.2. Simulation Modules
This section describes the different models used for the development of the airside of the airport under
study.
Stand module
The stand module simulates the ground operations performed at any aircraft stand in the airport.
Considering the focus on low-cost traffic for the case of LEY we included the following activities to be
performed at the stand: fueling, passenger disembarking and boarding, baggage loading/unloading, water
service and cleaning.

Fig. 3. Stand module.

illustrates the physical aspect of the module used for the stand. The different nodes in the figure are the
nodes of the route that the aircraft must follow. All the logic for the turnaround is implemented in the module
and it has been done general enough to adapt it for a different configuration and/or type of aircraft. We used
probability distributions for the time spent in performing the different operations; they are presented in Table
1. The values used were identified from the discussion with operators at Lelystad airport, based on their
expertise, and since the approach is a more strategic one, the accuracy of those numbers will not influence the
outcome of the model.
Table 1. Values of the Turnaround operations for the stand module.

Process Probability Distributions (Time)

Positioning stairs TRIANGULAR (90,120,150) sec
Disembarking TRIANGULAR (3,4,5) min
Luggage out TRIANGULAR (5,7,11) min
Positioning luggage TRIANGULAR (40,60,80) sec
Luggage in TRIANGULAR (5,7,9) min
Positioning fuelling TRIANGULAR (40,60,80) sec
Fuelling TRIANGULAR (7,8,9) sec
Positioning cleaning TRIANGULAR (4,5,9) min
Cleaning TRIANGULAR (8,13,16) min
Positioning water TRIANGULAR (1,2,3) min
Water service TRIANGULAR (4,5,6) min
Positioning TRIANGULAR (1,2,3) min
Boarding TRIANGULAR (4,5,6) min
Headcount TRIANGULAR (90,120,130) sec
Pushback 2.5 min.
 6

Network Module
Once the stand module is developed, it is necessary to create instances of it and connect them to
construct the parking positions and apron, then they will be connected to nodes and network paths
that represents the taxiways and the runway. Fig. 4 illustrates an example of the network that is used
for modeling the airside of the airport. The apron is located at the top right of Fig. 4 (a). It is
composed by 16 instantiations of the stand module. This figure represents only one of the different
possible configurations analyzed in this study. The different dots in Fig. 4 (a) represent the nodes that
are part of the network. In Fig. 4 (b) a layer has been added to illustrate the parts of the network that
represent the taxiways and the runway.
The logic implemented for this model assumes that the turnaround processes start as soon as the
aircraft has parked in the stand. The vehicles that perform the ground operations are located in depots
at one extreme of the apron. Thus the model provides insight about:
 Taxi-in and Taxi-out times;
 Potential conflicts in the taxiways, runway and among vehicles;
 Turnaround times;
 Impact on TAT from the variation in the number of vehicles;
 Best runway configuration.

(a)

(b)
Fig. 4. Network model (a) for the case of a L-shape apron model (b).

Wind direction affects the operation in such a way that it changes the side of the operation, under certain
limits the landing and take off is using RWY05 (50° or from west to east in Fig. 4) and during other direction
of wind the operation changes to RWY23 (230° or from east to west in Fig. 4). Since at this point we did not
want to evaluate the effect of the switching configuration due to the wind direction we assumed that the
analysis would take as a base case the example of the month of February when the 95% of the time the
predominant wind forces the use of the configuration of RWY05.
2.3. Modeling scenarios
We developed the model to test different scenarios according to the Ondernemingsplan Lelystad
(Schiphol Group, 2014). The volume of traffic is one of the most relevant inputs. We evaluated the
performance under three different traffic figures:
 40K: 40.000 annual movements (referred as 40K).
 7

 45K: 45.000 annual movements (referred as 45K) which is the main number expected by the airport
once the operation takes place.
 50K: 50.000 annual movements (referred as 50K).
Regarding the allocation of aircraft in the available stands (16 stands), we evaluated two different
allocations:
 Center-Out Allocation: This allocation implies that the air traffic controller allocates the aircraft firstly
close to the center of the terminal and as long as more aircraft are arriving, they are allocated out of
that center.
 Right-Left Allocation: This allocation assumes the aircraft are located using Gate1 and then Gate 2
and so on as long as the aircraft arer arriving.
We took also into account the number of vehicles in use for ground handling operations. Since it was not
feasible to evaluate all the possible combinations of vehicles, we proposed three different combinations based
on a standard set of vehicles:
 1 fueling truck
 1 bus for passenger boarding
 1 bus for passenger disembarking
 2 stairs (for dual boarding using both aircraft doors)
 1 water truck
 1 cleaning truck
 1 baggage cart for baggage in and out
In this set we did not include the pushback truck explicitly since we considered the pushback truck
operation as deterministic with a time consumption of 2,5 min. The standard set is the one required for giving
service for at least one aircraft at the parking position. In our study we tested three different set combinations:
1) Group A: 4 sets of vehicles
2) Group B: 8 sets of vehicles
3) Group C: 8 vehicles for the operation except fuel, water and cleaning which we used 6 vehicles.
Regarding the airport facility configuration, particularly the allocation of taxiways and parking positions
are also critical for TAT performance. We took into consideration three different lay-outs, named as
Configuration A, B and C, as the next section discusses.
2.4. Airport layout configuration
Three different layout configurations for the airside elements of the airport were tested in the simulation
model. All configurations are based on the official plans presented by the airport operator in the
Ondernemingsplan Lelystad (Schiphol Group, 2014). These plans are based on an extension for the existing
single runway (05/23), with variations on the taxiways, terminal and apron layouts.
Configuration A
It is the first layout published in the Ondernemingsplan Lelystad (Schiphol Group, 2014). This
configuration consists of a partially parallel taxiway system with additional access taxiways. It has an L-shape
apron in which the aircraft use pushback trucks with the nose in the direction of the terminal. Fig. 5 illustrates
Configuration A.
The flow of inbound and outbound aircraft would change depending on the direction of the wind
according to the preferred runway direction for each case (either RWY05 or RWY23). Fig. 6 shows flow
circulations for Configuration A. The blue lines indicate the routes of departing aircraft and the red lines the
routes of arriving aircraft. The taxiways configuration also depend on the configuration of the runway:
counter-clock wise to clock wise.
 8

Gate 16

Gate 1

Fig. 5. Configuration A, L-shaped apron with partial parallel taxiway.

Fig. 6. Aircraft circulation flows (blue for departures, red for arrivals) for Configuration A using runway 05 (above)
and 23 (below).

Configuration B
It is based on a linear apron, and in comparison to the original it has a full parallel taxiway along the total
length of the runway, with supporting access taxiways. In this configuration the L-Shape terminal disappears,
instead the apron is extended towards the left. Fig. 7 illustrates configuration B in which again the aircraft
have the nose perpendicular to the runway and a push back truck is needed after the block off.
Gate 1 Gate 16

Fig. 7. Configuration B, linear terminal and apron with full-length parallel taxiway.

In this configuration, the flow of aircraft is also counter-clock wise for the use of RWY05 and clock-wise
for RWY23. The flows can be appreciated in Fig. 8.
 9

Fig. 8. Aircraft circulation flow (blue for departures, red for arrivals) for Configuration B with runway 05 (above)
and runway 23 (below).

Configuration C
It is based on the concept that the aircraft use their own engines to perform the taxi in and out in the apron
using a full-length parallel taxiway. The apron, however, has no parallel taxiways available in this design due
to the lack of space since the stands require more space for performing the operation. As Fig. 9 shows, this
configuration does not offer a lot of flexibility for the taxi routing to and from the runway and vice versa. This
scenario should in theory allow a quicker and cheaper turnaround process for the airlines due to the reduction
of vehicles needed. However one might expect problems for the neighboring stands when the engines are on
for taxiing. This potential problem was also analyzed with the model.

Gate 1
Gate 16

Fig. 9. Configuration C, taxi-in taxi-out apron layout.

This configuration might also imply more attractiveness for the potential airlines due to cheaper handling
costs. However as it has been mention, due to the size of the apron, the airport might have to invest for more
area to perform the operations.
In this case the flow is less flexible since there is only one direction for taxiing in and out as Fig. 10
shows. Therefore the taxiways cannot be used as a buffer when there are operations going on at the aprons and
it might imply some conflicts that were also analyzed.

Fig. 10. Aircraft circulation flow (blue for departures, red for arrivals) for Configuration C with runway 05 (above)
and runway 23 (below).
 10

2.5. Simulation experiment and results

Table 2 describes the experiments developed for evaluating the impact of some of the variables in the
performance of the airport.
Table 2. Characteristics of the experimental design for the configurations and scenarios modelled.

Input Description Assumptions

Airside Configuration A. L-shaped linear stands with parallel Configuration A. Based on public designs of Lelystad
configuration and non-parallel segments (with pushback airport.
operation). Configuration B. Scenarios created accordingly to the
Configuration B. Linear stands with parallel taxiway latest Ondernemingsplan (Schiphol Group, 2014).
(with pushback operations). Configuration C. Scenarios created accordingly to the
Configuration C. Linear with taxi-in taxi-out apron latest Ondernemingsplan (Schiphol Group, 2014).
configuration.
Traffic mix 737 series + A320 series (narrow-body) Schedule developed based on Eindhoven operations
due to lack of information.
Vehicle numbers Group A - 4 sets for each operation.
Group B - 8 sets for each operation.
Group C - 8 loaders, 8 baggage, 8 stairs, 6 fuel, 6 water, 6
cleaning.
Traffic limit 1. Scenario 40K: 40.000 flight movements annually. 1. Lower Bound.
2. Scenario 45K: 45.000 flight movements annually. 2. Based on the Ondernemingsplan (Schiphol Group,
3. Scenario 50K: 50.000 flight movements annually. 2014).
3. Upper Bound.
Stand allocation Allocation 1. Left-right 1. The allocation is performed from left to right stand
based on the dominant RWY use.
Allocation 2. Center-Out 2. Allocation performed nearest to the terminal so
that the passengers walking distance is minimized.

We run several experiments combining the different inputs, in total we run 54 experiments with 10
replications each. Fig. 11 to Fig. 16 highlight the main important results, and Table 3 to Table 5 complement
the most important data from each airside configuration. The main outputs obtained from the experiments are:
Actual Time at Apron: Is the time measured from when the aircraft stop at the stand, perform the
turnaround and starts the taxiing out to the runway.
Turnaround time: The time between the parking at the stand and when the aircraft has finished the last
operation of the turnaround. It does not include the time it has to wait due to vehicle blocking in the parking
position.
Avg. Difference: It is the sum of pushback operation + delay, therefore when there is pushback truck the
difference with 2,5 min represent a delay due to vehicle blocking.
Configuration A: L-shaped Analysis
The most relevant results can be illustrated in Fig. 11 which shows the relationship of the time at Apron
with the traffic, allocation of stands and amount of vehicles used for the operation. In addition Fig. 12
illustrates the evolution of variability of the system under the different configurations.
The figure shows that Group C for ground vehicle allocation gives the best performance for the L-shaped
apron as the surface of the graph becomes flatter in the middle which corresponds to the zone of the Group C
of vehicles. After analyzing the causes in the simulation model, we identified that the utilization of vehicles is
higher in Group C having as a consequence that when there is shortage of vehicles, delays occur since some
aircraft need to wait for being served. On the other hand, the higher utilization indicates that the vehicles
spend less time in the vehicle depots which in occasions is translated into a better service as the vehicle jump
directly from one aircraft to the other. On the contrary, with excess of vehicles, they are routed directly from
the depot with the corresponding delay due to the time consumption of getting to the stands to serve.
 11

AVG Actual Time at APRON

Conf. A Left-Right, Center-Out

60.00
55.00
50.00 55.00-60.00
45.00 50.00-55.00
MINUTES

40.00 45.00-50.00
35.00 40.00-45.00
30.00 35.00-40.00
25.00 50k 30.00-35.00
20.00
25.00-30.00
GroupA L-R
GroupC L-R 20.00-25.00
GroupB L-R
GroupA C-O 40k
GroupC C-O
GroupB C-O
Number of Ground Handling Vehicles

Fig.11. Plot of the actual time at Apron.

AVG Half Width (Variability)

Conf. A Left-Right, Center-Out

6.00

5.00

4.00 5.00-6.00
MINUTES

3.00 4.00-5.00

2.00 3.00-4.00
2.00-3.00
1.00
50k 1.00-2.00
0.00
45k 0.00-1.00
GroupA L-R
GroupC L-R
GroupB L-R
GroupA C-O 40k
GroupC C-O
GroupB C-O
Number of Ground Handling Vehicles

Fig.12. Plot of the evolution of variation for configuration A.

Table 3 complements the main results of the experiments. It is worth noting that the statistic “Avg
Difference” measures the difference between the off block time (end of turnaround time) and the actual time
in apron (turnaround time + pushback +delay). This parameter includes the time consumed by the push back,
which is 2,5 min, so whatever value higher than it implies a delay caused by interference. The values of this
statistic suggest there is no significant interference between the vehicles and the aircraft and also that there is
minimal congestion on the taxiways parallel to the aprons, as there is minimal or in some cases null delay.
The results also suggest that the allocation of the aircraft in the stands (left-right or center-out) has a
significant impact on TAT as it can be seen that the numbers are reduced in general from center-out
allocation. The reason is that the allocation will directly influence the distance travelled by the vehicles to
reach the aircraft with the correspondent time consumption.
 12

Table 3. Results of the simulation model for Configuration A.

Group A Group C Group B

Half Half Half
Average Min Max Average Min Max Average Min Max
width width width
Turn Around
Left-Right 40,03 33,81 45,52 3,10 32,09 30,12 33,91 0,82 32,45 30,02 35,10 1,17
Time
Avg Difference
40K (Pushback + 2,46 2,00 3,00 0,27 2,16 1,82 2,63 0,19 2,17 1,82 2,67 0,19
delay)
Turn Around
Left-Right 37,97 33,70 44,94 2,57 32,16 29,62 34,92 1,22 32,09 29,97 34,37 0,99
Time
Difference
45K (Pushback + 3,62 1,94 3,06 0,29 2,22 1,84 2,67 0,22 2,20 1,82 2,58 0,19
delay)
Turn Around
Left-Right 40,19 35,94 45,42 3,99 31,78 29,30 34,12 1,36 31,78 30,06 33,88 1,24
Time
Avg Difference
50K (Pushback + 2,26 1,90 2,68 0,28 2,22 1,94 2,61 0,18 2,25 1,88 2,61 0,34
delay)
Center- Turn Around
34,65 29,36 38,45 2,32 30,49 28,57 32,35 0,83 30,56 28,47 32,90 1,06
Out Time
Avg Difference
40K (Pushback + 2,62 1,92 2,62 0,18 2,13 1,92 2,46 0,13 2,24 1,94 2,58 0,17
delay)
Center- Turn Around
35,62 32,50 38,82 1,57 30,31 28,50 31,81 0,81 30,47 28,93 31,91 0,87
Out Time
Avg Difference
45K (Pushback + 2,68 1,99 2,65 0,18 1,97 2,62 2,65 0,17 2,26 2,00 2,54 0,16
delay)
Center- Turn Around
37,54 34,63 39,54 1,26 32,80 30,97 34,39 0,79 32,92 31,01 35,40 1,12
Out Time
Avg Difference
50K (Pushback + 2,26 2,02 2,53 0,12 2,21 1,98 2,55 0,13 2,23 2,03 2,46 0,11
delay)

Configuration B
The results of configuration B show that regarding the total time in Apron, the vehicles’ Group C again
performs best due to the same reasons as for Configuration A. This effect can be appreciated if one pays
attention to the Group C plots in Fig. 13. Using this group the actual time at apron is the least using Group C
in most of the cases, in addition one can appreciate that the variability is minimum (Fig. 14) as well with the
same group of vehicles.

AVG Actual Time at APRON

Conf. B Left-Right, Center-Out

60.00
55.00
50.00 55.00-60.00
45.00
MINUTES

50.00-55.00
40.00 45.00-50.00
35.00 40.00-45.00
30.00 35.00-40.00
25.00 50k 30.00-35.00
20.00
25.00-30.00
GroupA L-R
GroupC L-R 20.00-25.00
GroupB L-R
GroupA C-O 40k
GroupC C-O
GroupB C-O
Number of Ground Handling Vehicles

Fig.13. Actual time at Apron for configuration B.

 13

AVG Half Width (Variability)

Conf. B Left-Right, Center-Out

6.00

5.00

4.00 5.00-6.00
MINUTES

3.00 4.00-5.00

2.00 3.00-4.00
2.00-3.00
1.00
50k 1.00-2.00
0.00
45k 0.00-1.00
GroupA L-R
GroupC L-R
GroupB L-R
GroupA C-O 40k
GroupC C-O
GroupB C-O
Number of Ground Handling Vehicles

Fig.14. Variability for configuration B.

Table 4 complements the information from the previous figures, and as a general remark we can say that
there is some delay due to the values present in the statistic “Avg Difference”, for example the max values of
Group A with 40K using Left-Right allocation. In this configuration the effect of the allocation of aircraft in
the stands has a significant impact.
By comparing Figure 11 with Figure 13 we can conclude that Configuration B outperforms Configuration
A since there is high reduction of the operation time in which the time reduction in turnaround time ranges
from 2 to 4 or more minutes in some cases.
Table 4. Results of the simulation model for Configuration B.

Group A Group C Group B

Half Half Half
Average Min Max Average Min Max Average Min Max
width width width
Turn Around
Left-Right 35,53 30,42 38,27 1,97 30,42 28,95 32,7 0,86 30,26 28,24 31,99 0,92
Time
Avg Difference
40K (Pushback + 2,79 2 3,18 0,31 2,34 2,01 2,73 0,17 2,29 1,91 2,8 0,2
delay)
Turn Around
Left-Right 35,71 32,79 38,37 1,58 30,67 28,27 33,54 1,26 30,75 27,86 33,9 1,43
Time
Difference
45K (Pushback + 2,47 2,08 2,93 0,22 2,43 2,05 2,87 0,21 2,42 2,09 2,88 0,19
delay)
Turn Around
Left-Right 35,47 32,22 38,26 1,67 30,31 28,02 32,44 0,99 30,28 28,44 33,39 1,13
Time
Avg Difference
50K (Pushback + 2,79 2,08 3,02 0,25 2,34 2,09 2,72 0,15 2,36 2,04 2,7 0,16
delay)
Turn Around
Center-Out 29,15 26,58 33,11 0,26 27,21 25,26 28,93 0,2 26,85 25,16 28,45 0,18
Time
Avg Difference
40K (Pushback + 2,38 1,93 3,01 1,69 2,22 1,89 2,73 0,95 2,21 1,89 2,67 0,81
delay)
Turn Around
Center-Out 29,39 26,1 33,24 2,08 30,67 28,09 33,58 1,25 30,95 28,86 33,83 1,16
Time
Avg Difference
45K (Pushback + 2,51 1,93 2,79 0,21 2,92 2,12 5,6 0,86 3,56 2,13 7,89 1,45
delay)
Turn Around
Center-Out 29,76 26,14 33,78 0,14 30,91 29,26 33,93 0,13 27,28 25,43 30,22 0,14
Time
50K Avg Difference 2,22 1,95 2,6 1,81 2,18 1,97 2,49 1,18 2,24 1,99 2,64 1,16
 14

(Pushback +
delay)

Configuration C
From Fig. 15 and Fig. 16 we note that this configuration is very unstable or has high variability since the
range of values is big compared to the previous scenarios once we modify the inputs [25 - 43 min.]. As it has
been identified, the “Center-Out” allocation of aircraft performs better that the “Left-Right” allocation. Fig. 16
suggests that in this configuration the interaction between aircraft and vehicles play an important role in the
efficiency of the total operation. Concerning the average values, the highest and the lowest values of the three
configurations can be obtained with this one. It is also worth noting that the shortest turnarounds can be
obtained with this configuration which might be very attractive for potential airlines that wanted to locate their
operations at this airport. However in order to get stable turnaround times it might be necessary to efficiently
coordinate the operation for avoiding the conflicts between aircraft and vehicles.

AVG Actual Time at APRON

Conf. C Left-Right, Center-Out

60.00
55.00
50.00 55.00-60.00
45.00 50.00-55.00
MINUTES

40.00 45.00-50.00
35.00 40.00-45.00
30.00 35.00-40.00
25.00 50k 30.00-35.00
20.00
25.00-30.00
GroupA L-R
GroupC L-R 20.00-25.00
GroupB L-R
GroupA C-O 40k
GroupC C-O
GroupB C-O
Number of Ground Handling Vehicles

Fig.15. Actual time at Apron for configuration B.

AVG Half Width (Variability)

Conf. C Left-Right, Center-Out

6.00

5.00

4.00 5.00-6.00
MINUTES

3.00 4.00-5.00

2.00 3.00-4.00
2.00-3.00
1.00
50k 1.00-2.00
0.00
45k 0.00-1.00
GroupA L-R
GroupC L-R
GroupB L-R
GroupA C-O 40k
GroupC C-O
GroupB C-O
Number of Ground Handling Vehicles

Fig.16. Variability for configuration C.

 15

Table 5 complements the results of Fig. 15 and Fig. 16. Besides the aforementioned findings, it can be
appreciated from Table 5 that the TAT is highly dependable on the number of vehicles at the apron and also
the allocation of stands as the turnaround times differ from different scenarios. As it has been mentioned,
under this configuration the interaction between the aircraft and the vehicles is high since the aircraft perform
the taxi-in and taxi-out using their own engines and the vehicles need to cross the apron to get to the different
stands. This has as a consequence that in some cases the vehicles serving stands have to wait for an aircraft
performing the taxiing out or in with the correspondent delay in the service performed but on the other hand, it
is also feasible under this configuration to achieve the minimum turnaround time which in this case could be
as short as 27 minutes (Group C, Center-out, 40K or 45K).

Table 5. Results of the simulation model for Configuration C.

Group A Group C Group B
Half Half Half
Average Min Max Average Min Max Average Min Max
width width width
Turn Around
Left-Right 40,03 32,47 45,58 3,48 37,46 31,13 49,57 4,90 39,65 31,09 48,79 4,30
Time
Avg Difference
40K (Pushback + 2,21 1,93 2,97 0,28 2,45 1,90 2,97 0,30 2,39 2,00 2,84 0,23
delay)
Turn Around
Left-Right 34,72 31,72 37,65 2,65 31,51 27,91 35,38 2,86 30,40 27,85 34,19 1,75
Time
Difference
45K (Pushback + 2,22 2,00 2,43 0,19 2,21 1,90 2,56 0,23 2,18 1,97 2,50 0,16
delay)
Turn Around
Left-Right 33,00 29,62 40,18 2,37 32,14 28,16 41,95 3,33 29,75 27,21 32,35 1,16
Time
Avg Difference
50K (Pushback + 2,29 2,10 2,94 0,31 2,17 1,91 2,45 0,12 2,14 1,89 2,44 0,13
delay)
Turn Around
Center-Out 30,69 28,32 33,70 2,30 32,29 27,49 38,78 5,32 29,70 27,45 33,06 1,55
Time
Avg Difference
40K (Pushback + 2,26 1,96 2,59 0,25 2,19 1,89 2,50 0,20 2,18 1,91 2,54 0,17
delay)
Turn Around
Center-Out 30,60 28,26 33,70 1,93 31,85 27,48 38,78 4,32 29,66 27,31 33,06 1,37
Time
Avg Difference
45K (Pushback + 2,33 1,94 2,85 0,38 2,39 1,83 2,86 0,36 2,17 1,83 2,83 0,32
delay)
Turn Around
Center-Out 35,16 31,33 40,01 2,01 32,13 28,27 37,65 2,20 30,56 28,12 33,94 1,34
Time
Avg Difference
50K (Pushback + 2.22 1.99 2.43 0.20 2.2 1.904 2.558 0.19 2.183 1.967 2.5 0.14
delay)

3. CONCLUSIONS AND FUTURE WORK

The article presents the analysis of a dynamic model of an airport that will be upgraded. The objective was
put on the analysis of turnaround time and on the identification of variables that affect the future performance
of it. We could identify factors that play important roles in the performance of the future airport. The most
relevant ones were first the layout of the apron which might affect the smooth operation of aircraft. In addition
we identified that variability plays an important role which must be considered during the process of selecting
the best layout.
From the operational standpoint, the most stable one was Configuration B which requires the use of
pushback trucks. On the other hand, Configuration C which has no pushback trucks generated higher average
values for the time at apron; nevertheless, paying attention to the whole picture, this configuration had also the
shortest turnaround times of all configurations.
In order to conclude which configuration would be the most efficient it is suggested to perform a cost-
benefit analysis from the airport and airline perspective. If the airport wanted to obtain short turnaround times
 16

then Configuration C with a good coordination of resources could be the best. On the other hand if the airport
is looking for a more resilient one with the counter effect of losing economic attractiveness to the airlines due
to the turnaround time achieved, then Configuration B would be suggested. Furthermore, it is also suggested
that the stand allocation is done as close to the center as possible in order to make the vehicles travel the least
possible distance and achieve the best turnaround time.
As a future work, the model will be further improved taking into account economic factors so that it is
possible to perform a multi-objective analysis in order to provide a more informed solution from different
angles. In addition, the logic of the use of the handling operations will be on a schedule basis instead of on-
demand in order to confirm that a coordinated solution will reduce the variability of the system with the
consequence of increased performance.
Acknowledgements
The authors would like to thank the Aviation Academy of the Amsterdam University of Applied Sciences,
the Airport of Lelystad for the exchange of impressions and the INESC TEC for the support to perform this
study.

REFERENCES
ACI Europe, 2010. An Outlook for Europe’s Airports. ACI Europe. Retrieved from https://www.aci-
europe.org/component/downloads/downloads/70.html
ACI Europe, 2014. World Airport Traffic Data. Retrieved from
https://www.aci-europe.org/component/content/article/29-article/103-world-airport-traffic-data.html
Becu, N., Bousquet, F., Barreteau, O., Perez, P., Walker, A., 2003. A methodology for eliciting and
modelling stakeholders’ representations with agent based modeling, in Multi-Agent-Based Simulation III,
Springer.
Banks, J., Carson J.S., Nelson, B., Nicol, D.M., 2010. Discrete-Even system Simulation, 5th edition,
Pearson.
Chung, C., Adeleye, S., 2006. A simulation based approach for Contingency Planning for Aircraft
Turnaround Operation System Activities in Airline Hubs. J. of Air Transportation, Vol. 11 (1)
Jimenez, E., 2015. Airport strategic planning in the context of low-cost carriers ascendency: insights from
the European experience (PhD Thesis). Universidade do Porto, Porto, Portugal. Retrieved from
http://repositorio-aberto.up.pt/handle/10216/78345
Longo F., 2013. On the short period production planning in Industrial Plants: A real Case Study. Intl. J. of
Simulation & Process Modelling, 8, 17-28.
Mujica, M. M., 2015. Check-in allocation improvements through the use of a simulation–optimization
approach. Transportation Research Part A, 77, 320-335.
Mujica, M. M., Scala, P., Boosten, G., 2014. Simulation-based capacity analysis for a future airport,
Computer Aided System Engineering (APCASE), 2014 Asia-Pacific Conference, 97-101. IEEE. February 10-
12, Bali, Indonesia.
Norin A., Granberg, T.A., Yuan, D., Varbrand, P., 2012. Airport logistics – A case study of the turn-
around process. J. of Air Transport Management, vol. 20 (May)
Okwir, S., Ulfvengren, P., Angelis, J., Ruiz, F., Yilsy, M., Guerrero N., 2017. Managing turnaround
performance through Collaborative Decision Making. Journal of Air Transport Management
Roosens, P., 2008. Congestion and Air Transport: a challenging phenomenon. European J. of Transport
and Infrastructure Research, 8(2), 137–146.
Schiphol Amsterdam Airport, 2015. 2014 Traffic Review. Retrieved from
http://trafficreview2014.schipholmagazines.nl/
 17

Schiphol Group, 2014. Ondernemingsplan Lelystad Airport.

Wu, Cheng-Lung, Caves, R.E., 2004. Modelling and simulation of aircraft turnaround operations at
airports. J. of Transportation Planning and Technology, Vol.27 (1)
Zuniga C., Delahaye D., Piera M., 2011. Integrating and Sequencing Flows in Terminal Maneuvering
Area by Evolutionary Algorithms. DASC 2011, 30th IEEE/AIAA Digital Avionics Systems Conference,
Seattle, United States.