MODELING DYNAMIC TRANSITIONS IN THE
GLOBAL AIR TRANSPORTATION SYSTEM
A system dynamics tool for Competitive Industry Assessment
Denes Csala, Sgouris Sgouridis
Masdar Institute of Science and Technology, Abu Dhabi
Abstract
The air transportation system needs to change drastically if it is to transition into a sustainable state. The
stakes for the needed technology-driven change — innovation —are extremely high for the lead adopters
given the high cost of R&D in manufacturing and the need to change parts of the existing infrastructure
and operational procedures if a significant improvement in high energy efficiency is to be achieved. We
assess different transition options, using the Global Aviation Industry Dynamics Transition (GAIDT) system
dynamics model of the industry that extends a previous implementation in the ability to test how such
changes will cascade through the global aviation system accounting for airline and aircraft manufacturer
competition across a variety of aircraft product lines. The GAIDT model exhibits advanced features of
modularity, extensibility, autocalibration and a visual user interface, some considered unique in system
dynamics, such as a dynamic modular structure or compatibility with other simulation modeling
techniques. We conduct a quantitative strategy assessment using a composite indicator 1 (normalized
strategy value), that combines the economic and environmental benefits of different strategies. It has been
found that the normalized strategy values will highly depend on the fuel prices. However, if the fuel prices
continue to rise, any kind of innovation option is beneficial. When decoupled from fuel prices, it has been
found that early transition options are preferred. In order to properly quantify the impact of technology
innovation, its penetration and assess the full impact of the different transition options, further research
is needed for which the GAIDT model can serve as basis.
1. Introduction
The commercial air transportation system has been consistently growing at a rate faster than the
background global economic growth. This was achieved through the development of a tightly coupled
ecosystem of airlines, aircraft manufacturers, airport operators, regulators and the traveling public and
influenced by the exogenous but still coupled impact of fuel prices and economic growth. The endogenous
dynamics of aircraft ordering rate and airline competition induced alternating periods of over- and under-
capacity that would ideally be moderated at the second stage of the value chain — through the actual
availability of aircraft (Sgouridis, 2007).
The primary challenge faced by the air transportation system today is the transition into a viable part of a
global sustainable transportation system. The ability to do so will depend on the interplay between
demand side changes driven by fuel prices, the internalization of the greenhouse gas emissions (fossil fuel
peak / carbon price / biofuel premium) and the ability of airlines and aircraft manufacturers to reduce fuel
consumption and by implication emissions and ticket prices through operational and technological
changes (Sgouridis, Bonnefoy, & Hansmann, 2009).
Nevertheless, the yields from harvesting the low hanging fruit in the case of the air transportation system
are insufficient for the scale of transition needed. Extant lean and fairly optimized operations and aircraft
are a result of the high level of industry competitiveness in an environment where historically fuel prices
are the major cost driver. As a consequence, drastically improving the system’s efficiency requires
significant changes in the technology and by implication in the existing infrastructure. The long lead times
and high costs of aircraft development and the long aircraft lives (in some cases exceeding thirty years)
make radical redesign a risky proposition for the manufacturer and their lead adopter customer airline.
a. Problem Definition
We develop the Global Aviation Industry Dynamic Transition (GAIDT) model as a tool to aid policy and
industry decision makers for tracing the impacts of drastic transition efforts in industry behavior and
competitiveness. GAIDT explores the dynamic behavior of the competitive aviation industry under
transition including: existence of any first mover advantage, necessary extent and benefit of regulatory
mandates for certain aviation transition options, relative impact and diffusion rate of transition options on
industry financials and throughput, and the robustness of transition options under varying exogenous
parameters. Assessment of the results can help develop appropriate policies and competitive strategies to
facilitate the global aviation system sustainable transition.
The type of transition options that can be investigated are:
e BUINESS AS USUAL (BAU): Aircraft manufacturers carry on with their current development
strategies, releasing the currently planned aircraft in the currently planned timeframe(s), under
different fuel price scenarios.
e INNOVATE: One aircraft manufacturer decides to radically change the design of its offered aircraft,
while the other manufacturer stays on the predefined development track, under different fuel
scenarios.
e FOLLOW: All manufacturers decide to radically change the design of their offered aircraft, with a
difference in introduction times and fuel benefits, under different fuel price scenarios.
e INDUSTRY REFORM: All manufacturers decide to radically change their aircraft, offering substantial
fuel burn benefits and introduce it at the same predefined time in the future, under different fuel
price scenarios.
In the transition option listed above, the focus has been on new technology diffusion. Other types of
transitions can also be linked to the problem, such as penetration of biofuels, carbon taxes and emissions
regulations, change in airport and/or air traffic control procedures regulations, etc. While the GAIDT model
is prepared to handle these transitions options as well, they do not constitute the focus of the present
study.
In the above description a radical change means a modification of aircraft that brings lower fuel burn (and
higher potential development costs but also impacts the operationally relevant aircraft characteristics like
speed and range) for future aircraft than normal aircraft redesign programs would do. At this stage of the
research, only the two dominant aircraft manufacturers are considered (Morrison, Hansmann, & Sgouridis,
2
2010). We investigate whether it makes sense to opt for small incremental imporvements of current
aircraft and “wait” for the radically new technologies to become available through other participants in
the market or it is more beneficial to the competitors present in the industry, as well the industry as a
whole to innovate as soon as possible. From this point forward in the study, the aircraft employing radical
changes are commonly referred to as new aircraft, the currently available aircraft on the market are
referred to as current aircraft and the aircraft currently in development at major aircraft manufacturers
and a planned release to market in the near future are referred to as aircraft in development.
Initially, all new aircraft are considered to comply with the standards of the currently existing aviation
infrastructure, therefore retaining their maximum wingspan, design cruise speed, balanced field length,
and their design payload/range. These characteristics are collectively referred to as mission specifications.
Further fuel-burn and emissions reductions of future aircraft can be obtained by modifying these mission
specifications and re-optimizing the aircraft and airline operations. In this study, potential benefits offered
by changes to the following mission specifications are considered:
e Design Cruise Speed Reduction (CSR)
o Decreasing the design cruise speed of the aircraft has an impact on airline operations:
while reducing the fuel-burn and thus the fuel costs of airlines, concomitantly increases
the flight time, thus increasing flight-time-related costs, such as crew labor. Through the
change of flight time, passenger demand is also affected.
e Wingspan Increase (WI)
o Increasing the wingspan improves the aircraft’s aerodynamic capabilities thus reducing
fuel-burn, but at the cost of limiting access to airport gates.
Changing the design range, accounting for current usage trend of aircraft operators can also yield
substantial fuel burn reductions and will be investigated in the future.
To compare between different strategies, for all of the above transition options the economic and
environmental impacts for all stakeholders are quantified and assessed.
b. Methodology: System dynamics approach
The aviation industry can be described as a “system of systems” or an “enterprise of enterprises”
(Sgouridis, 2007). Several techniques are available to translate enterprise models into simulation models.
A system dynamics approach was chosen, to best capture the global industry dynamics (Csala, 2013). To
compensate for the modeling choice’s greatest disadvantage, the rigidity of the model structure, an
experimental dynamic model structure has been developed, that can be expanded programmatically,
during model execution and also opens the door for a seamless integration into frameworks governed by
other simulation modeling methods.
c. Existing works
Several authors have studied aviation industry cost dynamics and have made causal loop diagrams or
complete system dynamics models, including the works of, (Lyneis, 1988), (Lyneis, 1999), (Lyneis, 2000),
(Liehr, Groessler, Klein, & Milling, 2001). However, most of these works concentrate on explaining the
cyclical nature of airline profits and correlating it with the industry structure. (Sgouridis, 2007), (Pierson,
2011) and (Cronrath, 2012) present models which describe the aviation industry dynamics with great
detail, including the dynamics of passenger demand, airline and manufacturer capacity management and
operating economics. These works have been used as basis for the development of the GAIDT model.
2. The Global Aviation Industry Dynamic Transition (GAIDT) Model
a. Description
The Global Aviation Industry Dynamic Transition (GAIDT) model is an advanced version of the Global
Aviation Industry Dynamics (GAID) model ( (Sgouridis, 2007), (Sgouridis, Bonnefoy, & Hansmann, 2009)),
implemented in up-to-date software (AnyLogic 6, written in JAVA language) and having a user friendly
interface. The model structure has been dynamized, allowing for easy expansion in future releases, and it
has been extended to accommodate for the industry transitions and mission specification changes
proposed in (PARTNER, 2012). Provided that we have the right data for model input, this structure allows
for greater resolution and also open the door for expansion and addition of modules, which might be
based on system dynamics or other simulation modeling methods, such as Agent-Based Modeling.
b. Structure —a dynamic modular approach
The GAIDT is built upon the modular structure of the GAID model, made up from logically bound structural
blocks. These modules are then integrated into classes to make up the GAIDT model. Each of the classes
and modules is then replicated, depending on the model’s dimensions. The model dimension types have
been predefined, however, the size of each of these dimensions does no need to be preset. The size of the
dimensions defines the level of insight we want to gain into the industry dynamics. The GAIDT model is
built over following dimensions:
e Aircraft Type: defines the aircraft types used in the GAIDT model, e.g. [Narrow Body, Wide Body]
or [A320, A330, B737, B777)
e Manufacturer Type: defines the manufacturer types used in the GAIDT model, e.g. [Airbus, Boeing]
e Airline Type: defines the airline types used in the GAIDT model, e.g. [Full Service, Low Cost] or
[Delta, Lufthansa, Emirates, Southwest]
e Passenger Type: defines the passenger types used in the GAIDT model, e.g. [Leisure, Business]
e Destination Type defines the flight leg types used in the GAIDT model, e.g. [Long Haul, Short haul]
or [1000 km, 2000 km, 3000 km]
At the current stage of the research, the following dimension sizes have been used, dependent upon data
availability for calibration:
e Aircraft Type (2)
o Narrow Body
o Wide Body
e Manufacturer Type (2)
o Boeing
o Airbus
e Airline Type (1)
e Passenger Type (1)
e Destination Type (1)
(Airline Type, Passenger Type and Destination Type dimensions have a size of 1 and are treated as
integrated entities)
Global Aviation Industry Dynamic Transition Model MODEL STRUCTURE
Airlines Class
jemand & Envitonment Module
| Destination Management Module | Finance Module
Passengers Class Manufacturers Class
Demand Module
Finance Module
Figure 1— GAIDT model structure
Figure 1 presents the GAIDT model’s modular structure. Each of these is modules is replicated across a
number of dimensions. These differentiate across the type of the data the model can handle and also
define its resolution. For example, for the Airline Capacity Management Module being 2 dimensional,
across the dimensions Aircraft Type and Manufacturer Type means that it is replicated n x m times, where
nis the size of the Aircraft Type dimension and m is the size of the Manufacturer Type dimension. This
way, the model data can be controlled individually for the Capacity Management of Aircraft Type 1,
Manufacturer Type 1 until that of Aircraft Type n, Manufacturer Type m.
e Mission Specification Changes Class (2D — Aircraft Type, Manufacturer Type)
e Airlines Class (1D — Aircraft Type)
© Forecast and Environment Module
o Capacity Management Module (2D — Aircraft Type (inherited), Manufacturer Type)
o Destination Management Module (3D — Aircraft Type (inherited), Manufacturer Type,
Destination Type — currently not implemented)
o Finance Module (2D — Aircraft Type (inherited), Manufacturer Type)
e Manufacturers Class (1D — Manufacturer Type)
© Production Management Module (2D — Manufacturer Type (inherited), Aircraft Type)
o Finance Module (2D — Manufacturer Type (inherited), Aircraft Type)
e Passenger Class
o Demand Module (1D — Aircraft Type)
Each of the GAIDT modules is composed of a parameterized system dynamics sub-model, having the
capability of receiving inputs either from other modules or from historical data. A navigable, visual user
interface aids model setup and operation. Figure 2 presents a snapshot of a sample module setup of the
GAIDT model and Figure 3 presents a snapshot of a sample module structure. Detailed individual modules
are presented in Appendix 1 and Appendix 2.
ula ae einai Configraten 0 Viowo oct! fulec2t MODEL SETUP
parent
3. ‘SETUP =
[> oesteatea Maneseet Module
8 <a 9 => Sen
Onet Target wackiog Years [inti order B 7
eux 2] temutier Yo 1 2 ||TaeL years 1
Onn nia Prod, Rate i I Detivery Pian
O Linearnreare i IMukisiertmutiptert os + 2 |[te9SM1 '» 9 10 || Covannnng @ cries
Osmectinresse 557 acu a0 Rae) Prod. ian umes | Proqucton Pi
O)Smooth Decrease Detay 0 Shit 0 Mutpter [Smooth [ years | o o oe || te7SM/yearl io 203 || O Backlog @ inventory
i |
OFet Production Adjustment Time
@t0%
Onediom
Oren 1
Lem 20m 2020 2.050
Deay 0 She 0 unpter
I
Target Order Share
Onnereasing
Orait
© strong Decrasse
O tiettbecrease:
(OStona meres3® | 5a pono
Otightincrease —Deay 0
Figure 2- Sample GAIDT module setup
‘ranian Medel Configuction: © View: © GRUGHED) SMIULATN RESULTS MODEL STRUCTURE
STRUCTURE ms
[a dasiatina Uacsoenent adic
8 <9 =>10
Ya nn
a
Averagevtine GOW
eran ae
" “ta oS =@ winprostina
SED JOS ZS
\ Z
Fa
DesiredProd? =
Dearne @e—
SN
DeseurodRd a
\ye . a
i er
& Trechnget
Cain) Rosits Toil
Figure 3 - Sample GAIDT module structure
c. Model calibration
Calibration of the GAIDT model has been carried out using the same principles as (Sgouridis, 2007). Since
the GAIDT model is fully backwards compatible with the GAID model, using the calibration parameters
deduced for the GAID model yield the same results when used in the GAIDT. This parameter set has been
considered as the initial calibration setup.
Furthermore, an advanced calibration module (which we will call autocalibration) has been developed that
can automatically adjust model parameters to meet a predefined output pattern for the chosen indicator
variables (with weights in brackets, for establishing relative indicator hierarchy) — therefore effectively
being able to deduce the form and size of unknown transfer functions between two connected variables.
e Passenger Demand (1.0)
e Airline Profit (0.5)
e = Airline Profit Margin (1.2)
e Fare (0.75)
e Load Factor (1.2)
e Capacity Order Rate (4 x 0.2)
e Capacity Order Backlog (4 x 1.0)
e CO2 Emission Rate (0.5)
The following parameter types are defined (as JAVA classes) in the model and used for fitting:
e Time Parameter
oA parameter which varies with time, unaffected by the other parameters or variables.
It has the following properties:
= Scenario — for deciding on the prediction table used
= Multiplier — for data scaling
= Delay — whether the data has a delayed effect in affecting the other variables of
the model
= Shift — whether the data table selected has a time shift
e Effect Parameter
o Aparameter which takes an argument and uses a transfer function to produce and effect
It has the following properties:
= Scenario — for deciding on the transfer function (table) used
= Multiplier — for effect scaling
= Delay — whether the data has a delayed effect in affecting the other variables of
the model
= ArgDelay — whether the argument of the transfer function has a delay before
applying the function
= Arginitial — initial value of the argument (where applicable)
For implementation of any delay, a first order exponential smoothing is used, unless noted otherwise.
With autocalibration, GAIDT can alternate between predefined transfer functions for variable (by changing
the Scenario parameter fields) interactions, apply weights, delays, argument delays and time shifts to them
and determine which yields the smallest deviation from the predefined dataset. In further releases it
should allow to devise what kind of transfer function should be used (already implemented for the
simulation part, where the user can choose or construct any transfer function for the effects)
Ahybrid calibration approach of per module and global calibration has been used. First, each of the classes
with their respective modules were fitted separately, using different sources for their inputs — either
previous simulation data or historical values. Then, an experimental global calibration was also conducted,
varying every module’s parameters at once. For every calibration experiment, 3 calibration scenarios were
defined, based on the datasets used for fitting. These are:
0. Calibration dataset: Output of the GAID model (1984 — 2025)
1. Calibration dataset: Actual data of the GAID model (1984 — 2005)
2. Calibration dataset: Actual data from (A4A, 2012) (1984 - 2012)
Each of these scenarios was run twice, using different random seeds, 1, and 2, respectively to ensure
consistency. Then the resulting parameter values were averaged (modal value was taken where necessary)
and then final parameter set was created to be used in simulations, alongside the default setup from the
GAID model. The Airline Profit Margin fit is presented in Figure 4. The full calibration report is presented
in Appendix 3.
Ot
0.05 4
o4
-0,05
1.980 1,985 1,990 1,995, 2,000 2,005
a» Standard Current + Best feasible
Figure 4 — Airline Profit Margin Calibration Fit
In all calibration graphs, Standard is the dataset against which the model gets fitted and Current is the
model output using the current calibration parameters. Best Feasible corresponds to the parameter set
yielding the best feasible solution. In the above graph Current and Best Feasible happen to be overlapping.
3. Dynamic transitions
a. Experiment design
Using the calibrated GAIDT model, a sensitivity analysis for strategy assessment has been carried out.
Using the different preset technology scenarios identified in the International Civil Aviation Organization’s
Committee on Aviation Environmental Protection (ICAO/CAEP) Long-Term Fuel Burn Technology Goals
(LTTG) exercise and used in (PARTNER, 2012), the manufacturers’ strategies have been adjusted to apply
the technology advancements and mission specification changes on their new aircraft offerings in the
2020-2030 period, complying with the transition options described in the Problem Definition section.
Table 1 and Table 2 present the technology advancement options and potential mission specification
changes, with their respective benefits for Narrow Body (after Boeing 737-800w) and Wide Body (after
Boeing 777-200ER) aircraft (adapted from (PARTNER, 2012)).
Table 1 — Fractional fuel-burn reductions (baseline = 1.00) resulting from redesigns of new Narrow Body aircraft with different
mission specifications and different levels of technology
Technology level Li CSR (0.8625) WI (1.35) CSR + WI
Li 0.9440 0.8650 0.9323 0.8542
TS1- 2020 0.7500 0.6920 0.7413 0.6839
TS1 - 2030 0.7056 0.6527 0.6971 0.6448
TS2 - 2020 0.7056 0.6527 0.6971 0.6448
TS2 - 2030 0.6594 0.6109 0.6510 0.6031
TS3 — 2030 0.5654 0.5222 0.5606 0.5177
Table 2 — Fractional fuel-burn reductions (baseline = 1.00) resulting from redesigns of new Wide Body aircraft with different
mission specifications and different levels of technology
hnology level Li CSR (0.8333) WI (1.36) CSR + WI
Li 0.9257 0.8278 0.9257 0.8278
TS1 - 2020 0.7923 0.6826 0.7674 0.6611
TS1 - 2030 0.6802 0.6033 0.6802 0.6033
TS2 - 2020 0.7084 0.6286 0.7084 0.6286
TS2 - 2030 0.6280 0.5557 0.6280 0.5557
TS3 - 2030 0.5747 0.5078 0.5747 0.5078
In the above tables, a “Baseline” technology level means re-optimization of studied aircraft (not allowing
for in-family flexibility and modularity), TS1 means light redesign of current aircraft, continuation of
current development trends, TS2 means a higher than average technology advancement, introduction of
new technology currently under research and TS3 means significant redesign of current aircraft and
introduction of new technology currently under research or beyond, potentially necessary from pressure
from regulations or taxes. CSR stands for Cruise Speed Reduction and WI stands for Wingspan Increase.
It is clear that the chosen technology scenario sets the level of fuel-burn reduction, with mission
specification changes only offering marginal reductions. Wingspan increase does not offer by far as much
benefits in fuel-burn as cruise speed reduction would (and very small or no effect on Wide Body aircraft).
In addition to this, for comparison purposes, we need to consider the current strategy of the two
considered aircraft manufacturers, Airbus and Boeing. Table 3 presents the BAU aircraft development
plans for the two major aircraft manufacturers, with the timings of new aircraft introductions and their
fuel benefits compared to their current models, corresponding to technology scenario TS-BAU.
The technology scenario “STATUS QUO” — TS-SQ corresponds to the theoretical strategy case when both
major aircraft manufacturers stop development of new aircraft and keep their current offerings.
Table 3 ~ BAU aircraft development plan for the two major aircraft manufacturers
Aircraft Type Introduction year (EIS) Fuel-burn reduction (current model = 1.00)
Airbus A320neo 2015 0.8500
Boeing 737MAX 2017 0.8375
Airbus A350 2017 0.8266
Boeing B787 2011 0.8000
Table 4 ~ Strategy decisions for the two major aircraft manufacturers
Airl BAU INNOVATE FOLLOW
BAU BAU BOEING INNOVATES =
INNOVATE AIRBUS INNOVATES INDUSTRY REFORM BOEING FOLLOWS
FOLLOW = AIRBUS FOLLOWS a
Taking into account the industry transition options described in the Problem Definition section, and the
possible strategy decisions presented in Table 4, the possible combinations of strategy choices, technology
levels and potential mission specification changes yield a very high number of possible cases. For this
study, the following representative simulation scenarios were created:
Scenario 1:
BAU (all manufacturers follow current development plan with TS-BAU parameters)
| Boeing | Follows current development plan with TS-BAU parameter aircraft |
| Airbus | Follows current development plan with TS-BAU parameter aircraft |
Scenario 2: BOEING INNOVATES SOON
[ Boeing | TS-SQ until 2020 then TS2-2020
[Airbus | TS-BAU
Scenario 3: BOEING INNOVATES LATE
[ Boeing | TS-BAU until 2030 then TS2-2030
[Airbus | TS-BAU
Scenario 4: AIRBUS INNOVATES SOON
[ Boeing | TS-BAU
[Airbus _| TS-SQ until 2020 then TS2-2020
Scenario 5: AIRBUS INNOVATES LATE
[ Boeing | TS-BAU |
[Airbus | TS-BAU until 2030 then TS2-2030 |
Scenario 6: BOEING FOLLOWS SOON
[ Boeing | TS-SQ until 2020 + 2 then TS2-2020 + 2 |
[Airbus _| TS-SQ until 2020 then TS2-2020 |
Scenario 7: BOEING FOLLOWS LATE
[ Boeing | TS-BAU until 2030 + 2 then TS2-2030 + 2 |
[Airbus _| TS-BAU until 2030 then TS2-2030 |
Scenario 8: AIRBUS FOLLOWS SOON
[ Boeing | TS-SQ until 2020 then TS2-2020 |
[Airbus | TS-SQ until 2020 + 2 then TS2-2020 + 2 |
Scenario 9: AIRBUS FOLLOWS LATE
[ Boeing | TS-BAU until 2030 then TS-2030 |
[Airbus | TS-BAU until 2030 + 2 then TS2-2030 + 2 |
Scenario 10: INDUSTRY REFORM SOON
[ Boeing | TS-SQ until 2020 then TS2-2020 |
[Airbus _| TS-SQ until 2020 then TS2-2020 |
Scenario 11: INDUSTRY REFORM LATE
[ Boeing | TS-BAU until 2030 then TS-2030
[Airbus | TS-BAU until 2030 then TS-2030 |
In the process of designing the above scenarios, it was assumed that if new aircraft enters service in 2020,
the TS-SQ scenario is followed and the aircraft in development does not get introduced (this includes the
Boeing B787 entered into service in 2011, as the model starting year is set to 2010), until the introduction
time of the new aircraft, 2020. However, if the innovated aircraft enters service in 2030, the TS-BAU is
followed and the aircraft in development gets introduced at the planned time and the new aircraft is
introduced in 2030 as a replacement. This decision was based on taking into consideration the typical
duration of aircraft development programs (Morrison, Hansmann, & Sgouridis, 2010). For all scenarios
technology advancement level TS2 was chosen.
Since all transitions yield fuel-burn reduction of aircraft, their operating benefit is clearly dependent on
fuel price. For this reason, the simulations have been conducted for 3 different fuel-price scenarios:
e Current, high fuel price (fuel price stays at 2012 levels until 2040 — on par with Airbus’ (Airbus,
2012) and FAA’s (FAA, 2010) predictions)
e High fuel price (current fuel price will double by 2020 then stagnate until 2040)
e Low fuel price (current fuel price will drop by 20% by 2020 then stagnate until 2040)
After the simulation runs, the most promising manufacturer strategy scenarios for each fuel-price scenario
can be chosen and mission specification changes’ effects might be investigated.
The strategy decisions presented in Table 4 are taken separately for the Narrow Body and Wide Body,
markets. Likewise the considered transitions are cross-dimensional and can be classified as:
¢ One-dimensional transition
o Only Narrow Body
o Only Wide Body
¢ = Multi-dimensional transition
o _Both Narrow Body and Wide Body (all dimensions)
In order to reflect dimensional changes, the Narrow Body and Wide Body markets, respectively the
combination of them, were analyzed separately and are presented as separate experiments in the results
section. For multi-dimensional transitions, it is assumed that the changes get into effect at the same time
across all dimensions, as having different entry times would add significant complexity to the experiment
design, offering only marginal accuracy improvement (deduced empirically).
b. Simulation Results
The experiments defined and described in the Experiment design section were simulated, using the GAIDT
model. This section summarizes the simulation results.
During the simulations, the cumulative airline profits, the cumulative manufacturer profits and cumulative
total industry profits (all 1984 dollars), cumulative emissions (Gtons of CO2) and cumulative passenger
demand (Revenue Passenger Miles — RPM) for the 2010-2040 period were taken as key comparison
indicators. Cumulative per unit industry profit and cumulative per unit emissions were calculated, where
the unit is RPM. Then a geometric composite indicator V was created, further referred to as strategy value:
v= “Mle. (2) (1)
where: P,, is the per unit industry profit, desired to be high
E, is the per unit CO2 emissions, desired to be low
In this study, equal emphasis was placed on economic and environmental criteria, thus a = B = 1.
In order to aid comparison, all of the above mentioned measures have been normalized with respect to
the TS-BAU scenario to give information about the relative advantage or disadvantage of a certain scenario
compared to the normal industry development plan. This yields the normalized strategy value:
(2)
where: P, is the normalized per unit industry profit
E,, is the normalized per unit CO2 emissions
Narrow Body market
Table 5 presents scenario winners and Table 6 summarizes the results of the simulations when the
transitions are only applied to the Narrow Body Market. Detailed results are to be found in Appendix 4.
Table 5 — Narrow Body market winner by scenario
Fuel Winner by Si. s2 $3
Profits Boeing | Boeing |)
Current : : : :
Orders | Boeing | Boeing | Boeing Boeing | Boeing | Boeing | Boeing | Boeing
aan Profits | Boeing | Boeing | Boeing Boeing | Boeing | Boeing | Boeing | Boeing
i
. Orders | Boeing | Boeing | Boeing Boeing | Boeing | Boeing | Boeing | Boeing
Profits Boeing | Boeing
Low 7 = 5
Orders Boeing | Boeing | Boeing
By referring back to the simulation scenario definitions in the Experiment design section and consulting
Table 5, the complex dynamics of the aviation industry are revealed. It can be seen that the innovators
gain more profits, especially in the case when the competitor does not innovate. (Orders are affected by
Boeing
Boeing | Boeing | Boeing | Boeing | Boeing | Boeing | Boeing
other endogenous mechanisms and no specific emphasis should be placed on them at this point) In fact,
if the fuel-prices stay at current levels or go down, Boeing can only win if he chooses to innovate sooner
or later ($2, S3). However, if fuel prices continues to rise, the only viable option for Airbus is to innovate
(84, $5).
Table 6 — Narrow Body transitions strategy values
Fuel Best > Worst
Py
High eg, -& sf is [sz sysop tse siz ssf sio fsa
y ets [os | iss _{ su_{ _se_{ sio_[ se _{ _2 _[ 3 _| si_|
yp, Ests { s {os [osu [os [os [ss { so | os [ss _|
[ s [| s« [| s [ s | 2 [| sx [ s | su [| ss [ ss | so |
Current | Ey,
Ls |s« || |s|s | s | «| s | s | 2 |
7 Ls s || | = | | ss | so [ss | s [ a _|
"
w be Ls ss |s |s | 9 [= [ou | os | os [so |
A
SL 82 $3 $5 36 7 $8 39 S10 Sil
Average
Table 6 presents normalized profit, emissions and strategy values. The High, Current and Low represent
fuel scenarios and the color coding reflects is the value order within the table. Profits and strategies are
preferred to have a high value and emissions are preferred to have a low value. The Average row defines
the value of a certain strategy decoupled from the fuel price by taking the geometric average of the three
studied fuel scenarios.
A dominant pattern clearly emerges from the above table: as the fuel price gets lower and lower, the
technology innovations begin to lose their value. For example if the fuel price is low, the industry profits
can be as much as 11% less than following the current development plan (S2). High fuel prices clearly favor
the technology innovations. In scenario $4 with high fuel price, profits more than double compared to the
BAU case.
It is important to remark, that, as expected, the emissions are in all innovation cases less than in the BAU
case, for all fuel scenarios.
Another interesting remark is that when decoupled from fuel price and averaged, all strategy scenarios
have a strategy value higher than 1. This leads to the conclusion, that all Narrow Body innovation strategies
are more beneficial than following the normal development path, regardless of fuel prices. Clearly, the
best option is for Airbus to innovate late then for Boeing either to follow or to innovate at the same time
(S5, $7, $11). However, the strategy choice is not straightforward as later in time transition strategies yield
the most cumulative profits, but also the least amount of emissions reductions.
Wide Body market
Asimilar analysis is conducted for the Wide Body market, preserving the previously introduced notations’
consistency.
Table 7 presents scenario winners and Table 8 summarizes the results of the simulations when the
transitions are only applied to the Wide Body market. Detailed results are to be found in Appendix 5.
Table 7 — Wide Body market winner by scenario
Fuel Winner by
Profits
Current
=
oe
Low
Orders
In the Wide Body market, Airbus emerges as the winner in most of the cases, with Boeing needing high
fuel prices and to innovate in order to gain more profits (S2, S3, as it is the case in BAU scenario as well).
Also, it can be seen that Boeing wins both of the FOLLOW LATE scenarios, no matter if he or Airbus is the
first-mover. This can be largely attributed to its strong BAU development plan, as justified by the outcome
of S1. In case of current of low fuel prices, Airbus always emerges as winner, except when he decides to
INNOVATE SOON
Table 8 - Wide Body transitions strategy values
Fuel
High
Current
Average
The same pattern can be observed in the above Table 8, as in Table 6: as the fuel price gets lower and
lower, the technology innovations begin to lose their value. However, it is important to notice that in case
of Wide Body aircraft, the deviation for the BAU case is, on average, much smaller than in the case of
Narrow Bodies. This clearly indicates a higher relative value for Narrow Body innovations than Wide Body
ones and gives evidence of robust current development plans for Wide Body aircraft.
Profit gains can be as much as 100% less and emissions are on average 5% less than for Narrow Bodies.
When decoupled from fuel price, the INDUSTRY REFORM LATE strategy (S11) has the highest value with
the FOLLOW LATE strategies closely following. This clearly indicates that the current Wide Boyd aircraft
development programs are robust and new aircraft need to be introduced only in the far future.
This is in harmony with the findings in the Narrow Body market.
All dimensions
Table 9 present the market winners and Table 10 summarizes the results of the simulations when the
transitions are applied across all dimensions of the model, i.e. to the entire industry. Detailed results are
to be found in Appendix 6. Because of endogenous feedback mechanisms the multi-dimensional effect on
the industry is more complex than just the sum of the two one-dimensional effects (due to cross-
dimensional phenomena, such as brand loyalty).
Table 9 — All dimensions market winner by scenario
Fuel Winner by Si s2 $3
Profits Boeing | Boeing |)
Orders Boeing | Boeing | Boeing |
Profits Boeing | Boeing | Boeing
Orders Boeing | Boeing | Boeing
Profits Boeing | Boeing
Orders Boeing | Boeing | Boeing |
Current
High
Low
If transitions are applied across dimensions, the true complexity of the problem emerges. In this case,
there is no majority winner anymore and the individual strategy winners apparently began to depend more
on the chosen strategy and /ess on the fuel price.
It can be seen from Table 9 that, as expected from the one-dimensional cases, if both manufacturers follow
the BAU strategy (S1), high fuel prices favor Boeing and current or low fuel prices favor Airbus.
Innovators are rewarded regardless of fuel price, as both Boeing and Airbus win their INNOVATE strategies,
regardless of fuel price (S2, S3 and S4, S5), and regardless of the introduction times of their new aircraft.
In the FOLLOW strategies, first-mover Airbus wins if Boeing decides to follow soon ($6), in all fuel scenarios,
but Boeing can turn over the game and win if he decided to follow late (S7), if the fuel prices are
conveniently high. If Boeing moves first, he can only win if the fuel prices are high, but then he does so
regardless of the introduction of the new aircraft ($8, S9).
If an INDUSTRY REFORM is considered, Airbus wins if the introduction of new aircraft is soon (S10 all) and
Boeing can turn it over only with late new aircraft introduction times and high fuel prices (S11 High).
Table 10 - All dimensions transitions strategy values
Fuel
High
Current
BI 82 $3 S4 $5 sé 7 $8 39 $10 Sil
Average
As expected from the previous, one-dimensional simulations, the global transition strategy values are
situated between the Narrow Body and Wide Body values. It is clear, that transitions are needed and late
transitions are preferred. Rooting from the benefits for the Narrow Body case for S4, this appears to bare
the highest strategy value. However, for the Wide Body simulations, $4 was one of the worst strategies.
So, in order to decide on the best strategy, it is preferable to account for the scale of the variations that is
different across dimensions, then decide aided by Table 10. Likewise, strategy the variations have been
normalized with respect to the difference between greatest and lowest strategy value in each dimension.
This way an equal comparison benchmark can be constructed. This normalization yields the following
percent strategy values:
Table 11- i Hi ions strategy values
Narrow | Average V
Average
From Table 11 clearly emerges which are the best transition options, and the findings are on par with our
predictions by consulting the previous strategy value tables and scenario winner tables.
When considered decoupled from fuel prices, FOLLOW LATE (S7, S9) strategies have the greatest value,
with INDUSTRY REFORM (S10, S11) strategies closely behind. If a new technology transition is for
consideration in the near future (SOON strategies), then the option for Airbus is to start the innovation
with Boeing keeping his BAU (S4) scenario or following (S6).
c. Implications
Figure 5 summarizes all simulation results. The normalized strategy value is represented for all simulation
cases and all fuel price scenarios.
11
2 3 4 5 6 8
Strategy Scenario
mNarow High mWideHigh @TotalHigh mNarowCurrent mWideCurent mTotalCurent mNarowLow mWidelow sm Totallow
Figure 5 — Simulation results summary
It can be seen that the maximum improvement potential is in the Narrow Body market, especially for the
case of high fuel prices. Also, if the fuel prices stay at today’s levels or lower, the current development
strategies (BAU) have a higher value than other innovating transition options.
Figure 6 presents a summary of all simulation runs, displaying the strategies by normalized cumulative
profits and normalized cumulative emissions.
e _Alldimensions | LJ Narrow Body l a Wide Body
@ Current All @ High @ Low @ Current NB @ High @ Low 4 Current WB 4 High 4 Low
15
s.8 e
a & Low Emissions s4e S5S5, $5
2 r)
3 11,5
& = ese >” w 854 055 ass
=, 05 = é sf
z 10 a sys ™S4S10
ou
5, [we gists 5° s S45
2
& 3 S10 @ S8 Ss
Si a
ae
0.95
i}
8 a a:
0.9. | Airbus moves first: $4, $5, S6, S7 mS3
: Early transitions: $2, S4, $6, $8, S10 @s3
Sy a5 8% emissions savings, 15% profit gains (compared to BAU) Strategy Value Map
20.
0.92 0.93 0.94 0.95 0.96 0.97 0.98 0.99 1
Normalized Cumulative Per Unit Emissions, E,
Figure 6 - Summary strategy distribution map
On Figure 6 we are looking for data points in the top left corner of the map, featuring high-profit — low-
emission strategies. We can observe that the strategies are concentrated around a vertical and a horizontal
axis, meaning that they are either economically or environmentally beneficial, the combination of both is
hard to achieve. It is also easily observable that the scale of the Wide Body transitions’ strategy values is
much smaller than that of the Narrow Bodies’.
The summary strategy value map presents that emission reduction of up to 8% can be achieved compared
to the base (BAU) case, i.e. opting for radical innovation instead of following incremental. Early transition
strategies S6, S8, S10 lead to the highest emissions reductions. Profit gains of up to 15% can be realized,
when only Airbus innovates either soon or late (S4, $5). Good compromise strategies are the follow
strategies (S6, S7, S8, S9), but with no clear better option, with Airbus moving first resulting in higher
profits, but with Boeing moving first resulting in higher emissions reductions.
4. Conclusions
The commercial air transportation system has been consistently growing at a rate faster than the
background global economic growth, achieved through the development of a tightly coupled ecosystem
of airlines, aircraft manufacturers, airport operators, regulators and the traveling public and influenced by
the exogenous but still coupled impact of fuel prices and economic growth. However, as fuel prices and
economic growth remain highly volatile, the aviation industry will be greatly affected and possible
innovation strategies have to be developed, analyzed and their values clearly quantified in order for it to
transition into an economically sustainable state. As air transportation accounts for =2.8% of global CO2
emissions (Raper, 2009), reduction of greenhouse gas emissions are also important for the aviation
industry and are among the top priorities of the main aircraft manufacturers.
Possible transition options have been identified, on par with ongoing research in field and the Global
Aviation Industry Dynamic Transition (GAIDT) system dynamic model was developed for strategy
assessment. The model exhibits advanced features (modularity, extensibility, autocalibration), some
considered unique in system dynamics, such a dynamic module structure or compatibility with other
simulation modeling techniques. For easier use, a visual user interface has also been developed for the
GAIDT model.
In order to assess the value of the different strategies investigated and decide between options, a
composite indicator V (normalized strategy value) that combines the economic and environmental
benefits of different strategies has been created and used throughout the analysis.
Then strategy assessment has been conducted through a sensitivity analysis with the GAIDT model. This
has been carried out separately for one-dimensional (Narrow Body and Wide Body) and multi-dimensional
changes. In all cases, the strategy values compared to the Business as Usual (BAU) strategy (where aircraft
manufacturers keep their current development strategies and release the currently planned aircraft at the
currently planned time in the future) have been quantified and analyzed. The entire analysis has been
carried out for 3 different fuel price scenarios, High, Current and Low. The winners between the
manufacturers of the transition strategies have also been identified.
It has been found that the strategy values will highly depend on the fuel prices. However, if the fuel prices
continue to rise, making any kind of innovation is beneficial. When decoupled from fuel prices, it has been
found that late transition options (innovation in 2030, BAU strategy until then) are preferred. It has also
been found that in most of the cases when one manufacturers decides to innovate and the other
manufacturers follow with a delay, the first-movers will have an advantage.
We have found that radical innovation is more beneficial both when looked from the innovators
perspective in a competitive market, as well as for the industry as whole. Yet, the industry does not seem
to be in harmony with this, with innovation in commercial air transportation being a very slow and
incremental process, with new aircraft rollouts occurring decades apart and not bringing very substantial
improvements over previous incarnations. One might wonder what is the possible explanation for this,
which leads us to our analysis’ assumptions.
One of the most volatile parts of the study, but also a very crucial one is estimating the cost of research
and development of new aircraft. Since the new technology is not entirely known or even forecastable,
19
estimating further costs is highly uncertain. Also, other, ancillary technologies which can impact its
development or costs might become available, about which there is also high uncertainty. Moreover, it is
very challenging to decipher the research and development costs of past aircraft development programs
of aircraft manufacturers, as the data is highly protected by them. This has been extensively studied by
Norris (2005) (2009).
In order to properly quantify the impact of technology innovation, its penetration and assess the value of
different transition options, further research is needed. Dynamics of market with 3 major manufacturers
could also be investigated. Furthermore, the airline (and lessor) competition dynamics need to be also
modeled for a complete industry assessment. The GAIDT model has been developed with this intent in
mind and the option to expand and incorporate dynamics airlines has been preserved. Also, shifting the
bias between the climate and economy factors in the normalized strategy value indicator can lead to
different results, depending on the players’, stakeholders’ and regulators’ purposes.
5. Bibliography
AAA. (2012). Airline for America. Retrieved from Traffic & Financial Results:
http://airlines.org/Pages/Econ_Financial_Results.aspx
Airbus. (2012). Global Market Forecast 2012-2031. Retrieved from
http://www.airbus.com/presscentre/pressroom/presentations-
speeches/?elD=dam_frontend_push&doclD=25776
Cronrath, E.-M. (2012). Understanding the Airline Profit Cycle. System Dynamics Conference, PhD
Collogium. Retrieved from http://student.systemdynamics.org/wp/wp-
content/ulf/colloquia/2012-proceedings/2012_Cronrath_SD.pdf
Csala, D. (2013). Modeling Dynamic Transitions in the Global Air Transportation System. MSc Thesis,
Masdar Insitute, Abu Dhabi.
EIA, U. E. (2013). Petroleum & Other Liquids pric proejctions. United States. Retrieved from
http://www.eia.gov/dnav/pet/pet_pri_spt_s1_m.htm
FAA, F. A. (2010). Aerospace Forecast 2010-2030. FAA. Retrieved from
http://www.faa.gov/data_research/aviation/aerospace_forecasts/2010-
2030/media/2010%20Forecast%20Doc.pdf
Liehr, M., Groessler, A., Klein, M., & Milling, P. (2001). Cycles in the sky: understanding and managing
business cycles in the airline market. System Dynamics Review.
Lyneis, J. M. (1988). System Dynamics in Business Forecasting: A Case Study of the Commercial Jet
Aircraft Industry. System Dynamics Review.
Lyneis, J. M. (1999). System dynamics for business strategy: a phased approach. System Dynamics
Review, 37-70.
Lyneis, J. M. (2000). System dynamics for market forecasting and structural analysis. System Dynamics
Review.
20
Morrison, J. D., Hansmann, R. J., & Sgouridis, S. (2010). Game Theory Analysis of the Impact of Single
Aisle Aircraft Competition on Fleet Emissions.
Norris, G., & Wagner, M. (2005). Airbus A380: Superjumbo of the 21st century. Zenith Press.
Norris, G., & Wagner, M. (2009). Boeing 787 Dreamliner. Zenith Press.
PARTNER. (2012). Project 43.
Pierson, K. (2011). Cyclican dynamics of Airline Industry Profits. PhD Thesis, MIT Sloan School of
Management.
Raper, S. (2009). The Rising Effect Of Aviation On Climate. Manchester Metropolitan University’s Centre
for Air Transport .
Sgouridis, S. (2007). Symbiotic Strategies in Enterprise Ecology: Modeling Commercial Aviation as an
Enterprise of Enterprises. PhD Thesis, MIT.
Sgouridis, S., Bonnefoy, P. A., & Hansmann, R. J. (2009). Air Transportation in a carbon constrained world:
Long-term dynamics of policies and strategies for mitigating the carbon footprint of comercial
aviation. Transportation Research Part A.
6. Appendixes and other supporting materials
When creating the GAIDT model described in the attached paper, AnyLogic 6 software has been used.
Furthermore, a live working version of the model can be accessed at the following link:
http://tiny.cc/sdconf2013csala_model (It requires JAVA to run).
For your convenience, | all of the appendices can be found in the cloud folder
http://tiny.cc/sdconf2013csala_files.
Here is a list of files that have been included in the cloud folder:
Nr. Name Type Size (MB)
1. Appendix 1 * pdf, Portable Document Format file 1.04
2. Appendix 2 * pdf, Portable Document Format 1.11
3. Appendix 3 * xlsm, Macro Enabled Excel 2007 file 0.35
4. Appendix 4 * pdf, Portable Document Format file 1.31
5: Appendix 5 * pdf, Portable Document Format file 1.26
6. Appendix 6 * pdf, Portable Document Format file 1.25
21
