This article was downloaded by: [182.70.89.

246] On: 14 October 2020, At: 08:09

Publisher: Institute for Operations Research and the Management Sciences (INFORMS)
INFORMS is located in Maryland, USA

Service Science
Publication details, including instructions for authors and subscription information:
http://pubsonline.informs.org

Interactive Method for Service Design Using Computer

Simulation
Yuki Makino, Kazuo Furuta, Taro Kanno, Shigeki Yoshihara, Takamichi Mase,

To cite this article:

Yuki Makino, Kazuo Furuta, Taro Kanno, Shigeki Yoshihara, Takamichi Mase, (2009) Interactive Method for Service Design
Using Computer Simulation. Service Science 1(2):121-134. https://doi.org/10.1287/serv.1.2.121

Full terms and conditions of use: https://pubsonline.informs.org/Publications/Librarians-Portal/PubsOnLine-Terms-and-

Conditions

This article may be used only for the purposes of research, teaching, and/or private study. Commercial use
or systematic downloading (by robots or other automatic processes) is prohibited without explicit Publisher
approval, unless otherwise noted. For more information, contact permissions@informs.org.

The Publisher does not warrant or guarantee the article’s accuracy, completeness, merchantability, fitness
for a particular purpose, or non-infringement. Descriptions of, or references to, products or publications, or
inclusion of an advertisement in this article, neither constitutes nor implies a guarantee, endorsement, or
support of claims made of that product, publication, or service.

Copyright © 2009, INFORMS

Please scroll down for article—it is on subsequent pages

With 12,500 members from nearly 90 countries, INFORMS is the largest international association of operations research (O.R.)
and analytics professionals and students. INFORMS provides unique networking and learning opportunities for individual
professionals, and organizations of all types and sizes, to better understand and use O.R. and analytics tools and methods to
transform strategic visions and achieve better outcomes.
For more information on INFORMS, its publications, membership, or meetings visit http://www.informs.org
 Makino, Furuta, Kanno, Yoshihara, and Mase: Interactive Method for Service Design Using Computer Simulation
Service Science 1(2), pp. 121-134, ©2009 SSG

Interactive Method for Service Design Using Computer Simulation

Yuki Makino, Kazuo Furuta, Taro Kanno

The University of Tokyo
7-3-1 Hongo, Bunkyo-ku
Tokyo, 113-8656, Japan
makino@sys.t.u-tokyo.ac.jp, furuta@sys.t.u-tokyo.ac.jp, kanno@sys.t.u-tokyo.ac.jp

Shigeki Yoshihara, Takamichi Mase

ANA Strategic Research Institute Co., Ltd.
1-5-2 Higashi-shimbashi, Minato-ku
Tokyo, 105-7133, Japan
s.yoshihara@ana.co.jp, ta.mase@ana.co.jp

A n interactive method for service design has been proposed for services that heavily depend on human
expertise and performance. In this method a simulation model of the service processes is to be
constructed based on ethnographic field observation, and then the model is to be validated by showing
simulation results to the field experts in a visualized form. In the course of proposing and assessing design
options, opinions are repeatedly acquired from the field experts also by showing simulation results. Their
expertise can be reflected thereby in the final design through such an interactive design process. In order to
demonstrate the effectiveness and usefulness of the proposed design method, the method was applied to
ground aircraft operations at a large airport. A simulation model of the service processes at Tokyo
International (Haneda) Airport was constructed, and a simulation system was developed on a Java platform
for Windows PCs. It was then demonstrated that the simulation could well replicate the observed data of
ground aircraft operations. It was also shown that the proposed design method was useful to create new
design options for ground aircraft operations and comparatively assess them for improving service
performance.

Key words: service design; interactive design method; computer simulation; ground aircraft operations;
ethnography; visualization
History: Received June 18, 2009; Received in revised form Sept. 22, 2009; Accepted Sept. 25, 2009;
Online first publication Sept. 30, 2009

1. Introduction
The center of industries is gradually shifting toward services rather than agriculture or manufacturing particularly in
industrialized countries, and the productivity of services has a great impact on the international competitiveness of
these countries (Council on Competitiveness 2004; Spohrer and Maglio 2008). The concept of a Product-Service
System (PSS), which is an integrated form of products and services capable of fulfilling customer’s needs, is being
accepted in the recent business scene (Oliva and Kallenberg 2003; Baines et al. 2007). Meanwhile some authors
focus on the potential of PSSs for innovation by improving the sustainability or eco-efficiency of businesses
(Manzini and Vezzoli 2003; Tukker 2004). Planning and designing service systems properly are therefore highly
expected.
Intangibility, perishability, simultaneity, and heterogeneity feature services compared with other industries or
products. In addition, services usually rely heavily on human expertise and performance. These features make it
difficult to apply the design methods developed for physical products directly to services; some new approach of
service design is necessary. Performance measurement from a comprehensive viewpoint is another critical issue in
service design. Neely et al. (1997 2005) have proposed a framework for designing performance measures and
demonstrated the practical validity and utility of the framework (Neely et al. 1997 2005). How to model and account
human performance in performance measurement is still a non-trivial issue.
A variety of approaches have been proposed so far for analysis and design of service systems. Analysis methods
based on video ethnography (Buur et al. 2000) or observation of human performance (Kumar 2004) are useful for
understanding customers’ behavior in services. Shostack (1982 1984) proposed a method called service blueprint to

121
 Makino, Furuta, Kanno, Yoshihara, and Mase: Interactive Method for Service Design Using Computer Simulation
Service Science 1(2), pp. 121-134, ©2009 SSG

describe activities in a chart to show the cooperation between service providers and customers. Arai and Shimomura
(2004) proposed a service CAD (Computer-Aided Design) system where one can define and describe an overview
of service flow to design and assess service systems. As for modeling human performance in computer simulation,
Baines and Benedettini (2007) proposed a practical framework to incorporate workers’ performance into discrete
event simulation for manufacturing systems design.
Though these methods are useful for understanding the reality and architecture of services, they are so weak in
describing dynamic service processes that they are sometimes little useful for detailed design or quantitative
assessment of service systems. In addition, they prescribe neither methodological guidelines for acquiring human
expertise that often plays crucial roles in services, nor exact models for relating human performance to the
influencing factors of human performance.
An example of service where an effective design method is desired is operation of a large airport. The quality of
airport service has some problems recently such as increasing delays in arrival and departure of aircraft due to heavy
air traffic. Extension of runways, terminal buildings, and other ground facilities is planned or under way at many
large airports worldwide. The expanded capacity, however, does not function as expected, unless operations of
airport service are appropriately designed and managed. Though computerization and automation have already been
introduced into airport service, its considerable parts still depend on human expertise and performance, because a lot
of contextual factors affect airport service. This situation makes it impossible to model the service processes
mathematically and then to apply conventional optimization methods.
The aim of this work is to develop an interactive method for service design, where human expertise and
performance play crucial roles. We will propose a human-centered design process, where intensive use of computer
simulation enables one to design service systems interactively with the help of field experts. In addition to
ethnographic analysis, which is suitable for revealing the reality of service processes and extracting knowledge from
the field experts, an agent-based approach for modeling and simulation is used. Agent-based simulation can replicate
dynamic service processes with relatively less approximation and the model used can be validated in comparison
with field data.
We will then demonstrate the effectiveness and usefulness of the proposed method by applying it to airport
service: aircraft operations on the ground of a large airport. In aerospace industries, human performance modeling
and simulation environments have been developed for design, visualization, and assessment of complex human-
machine systems for cockpit design (NASA 2009), but such technologies have hardly been applied to airport service
where interactions among many actors play crucial roles. Some previous studies dealt with ground aircraft
operations in preparation for extension of the runway capacity (Kazda and Caves 2007). Almost no studies, however,
considered microscopic interactions of aircraft like delays caused by inappropriate ordering of pushback operations
in departure. Our interview to the field experts of an airline suggested that such interactions influence considerably
the efficiency of aircraft operations. In this study, we will try to construct a simulation model for ground aircraft
operations from a microscopic view so that we can assess and design the service system considering dynamic and
more detailed service processes.
The service design method to be proposed will be explained in the next chapter and then construction of a
simulation model for the application domain will be discussed in Chapter 4. The next chapter discusses
implementation of the simulation model and then shows the results of test simulation to demonstrate the validity of
simulation. In Chapter 6, how the simulation system developed in this study is used for proposing and assessing new
design options will be presented, and finally conclusions will be given.

2. Interactive Design Method

2.1 Overview
We will propose a human-centered and interactive design method for service systems where the expertise of service
providers can be reflected in service design. As shown in Figure 1, the method consists of three phases: service
modeling, service simulation, and service planning.

2.1.1 Service Modeling

Firstly a model of the service processes is to be constructed through ethnographic field observation.
Ethnomethodology is a study method of sociology to find out some implicit orders, rules, or norms behind human
behavior through observation in the actual work environment. Suchman (1987) pointed out that an ethnographic
approach is required when a question is to be answered what knowledge and experience people use in a cooperative
work. Service modeling should start by recording tasks, decisions, behavioral patterns of service staff in the actual

122
 Makino, Furuta, Kanno, Yoshihara, and Mase: Interactive Method for Service Design Using Computer Simulation
Service Science 1(2), pp. 121-134, ©2009 SSG

service operations, interviewing the staff, and analyzing the obtained data. This approach enables one to extract the
knowledge of field experts that is required to create, modify, and improve the model. It is a key process for human-
centered design of service systems, where human factors should be properly considered and reflected.

2.1.2 Service Simulation

In service simulation, the whole service processes are replicated in a virtual environment within a time period
shorter than the real time. Various data obtained from the simulation in addition to the simulation model itself are
visualized in order to easily understand problems in the service system. Showing the visualized simulation results to
the field experts belonging to different sections of service provider will enhance understanding and sharing of the
problems among them. Solution of the problems from a comprehensive viewpoint is thereby expected. Such an
interactive and cooperative process is very useful for decision-making in service design. It is also effective for
validation of the simulation model from a viewpoint of the field experts.

2.1.3 Service Planning

In the final phase of service planning, possible solutions are to be proposed to the problems found in the previous
phase, assessed by simulation, and design decisions will be made. Design options are assessed from various
viewpoints: customer and staff satisfaction, task difficulty, costs, and so on. The result of assessment is to be
reviewed by the field experts again. As a consequence, new problems such as external costs and safety concerns
may be found and further improvement may be added to early options. It is sometimes necessary to repeat this
process to gradually brush up design options to be adopted finally.
The model constructed at the beginning will undergo modification and improvement from the needs of field
experts. It is realized by repeating simulation, presentation of the results to the field experts, and modification of the
model considering the obtained feedbacks. As described in the above, the design method proposed in this work is
not straight forward, but it is carried out as a cycle of proposal, assessment, and review. Simulation and presentation
of its results to the field experts are repeated in this cycle, and we call the method interactive in this sense.

Figure 1 Interactive Design Method for Service Systems

2.2 Model Development

An agent-based approach (Ferber 1999) has been adopted for modeling the service processes. An agent-based
approach is a method of studying phenomena in a complex system by observing the macroscopic system behavior
that emerges from microscopic interactions between agents. An agent is defined as an autonomous entity that can
perceive the surrounding environment, make a decision, and change behavior adapting to the environment. The
environment here stands for the world outside of the system. A collective system modeled as a set of many agents is
called a multi-agent system.

123
 Makino, Furuta, Kanno, Yoshihara, and Mase: Interactive Method for Service Design Using Computer Simulation
Service Science 1(2), pp. 121-134, ©2009 SSG

An agent-based approach has the following merits as a method of study. Firstly, it can deal with the phenomena
emerging from microscopic behavior of system components. Some order exists in emergent behavior of a complex
system even without any central or top-down controls. It is usually the case with service systems and an agent-based
approach is suited for modeling service systems. Secondly, a multi-agent system can be a natural representation of a
complex system when agents can be defined corresponding to actual system components. As for service systems, it
is comprehensible if each provider, receiver, and resource is modeled as an agent. Finally, a multi-agent system is
easy to modify and expand, because agents can be developed, tested, added and modified separately. It is preferable
not only for rapid prototyping of the model but also for the interactive design process.
Model development progresses in seven steps shown in Figure 2. Each step is performed as explained below.

(1) Definition of System Boundary

The boundary that separates the target service system and the environment is to be defined. The world outside
of the boundary is described with boundary conditions, i.e., fixed simulation scenarios.
(2) Identification of Agents
The persons and organizations that have relations to the target service are to be identified, defined, and
represented as agents. Service providers and receivers are the people to be considered first, but other
stakeholders may exist.
Figure 2 Progress of Model Development

(3) Identification of Services

Services exchanged within the system are to be identified and described. Intra-organizational services should be
included as well. The providers and receivers for each service, the roles they play, the channels and media they
use are to be identified and described.
(4) Identification of the Environment
The components of the environment that exist outside of the system boundary but affect the service processes
are to be identified and described. Their behavior is given as prescribed scenarios, or they are often modeled as
static agents.

124
 Makino, Furuta, Kanno, Yoshihara, and Mase: Interactive Method for Service Design Using Computer Simulation
Service Science 1(2), pp. 121-134, ©2009 SSG

(5) Description of Agent Behavior

Behavior of each agent is to be described. An agent usually has some internal mechanism to change actions of
its own depending on the status of the environment and other agents. The rules used for selecting actions and for
changing the internal state of agent are described.
(6) Description of Communication
Communication among agents is to be described. Every interaction among agents that occurs to achieve specific
goals in the service processes is represented as communication.
(7) Model Validation
The simulation model is to be validated by implementing the model as an executable simulation program,
carrying out simulation, and checking obtained simulation results. Visualization of simulation results is very
useful also for model validation.

3. Service Modeling for Ground Aircraft Operations

The proposed method was applied to ground aircraft operations at a large airport, Tokyo International (Haneda)
Airport, for demonstration of its effectiveness and usefulness. In this chapter, it is explained how a model of the
service processes was constructed for the concrete application domain.

3.1 Knowledge Acquisition Through Field Visit and Observation

Through visit to Haneda Airport and interview to the staff in various sections there, we investigated the actual
service processes and their situations at the site of work. The places visited include ground facilities, the operation
control center of an airline, towing cars, and so on. The interviewees were operators of towing cars, pilots, and
operation controllers. As a result, the flow of activities and the roles of field staff were clarified. Documents on the
layout and specs of ground facilities were separately obtained.
In next step, we video-recorded the movements of aircraft from different locations, in particular around the
second terminal building of Haneda Airport, and voice-recorded the radio communication between pilots and ground
controllers. From the video and voice data, the time required for arrival and that for departure were evaluated and
the taxiing routes instructed by ground controllers were identified. These data were then transcribed and used as a
basis for acquiring the rules that pilots and ground controllers use in their work. Since an aircraft cannot move
backward by itself in departure from the spot, a towing car must push it back. This operation is called “pushback.”
The pushback patterns that are used for each spot following ground controller’s instruction were examined and
tabulated together with their application conditions.
Preliminary results of simulation were shown to operation controllers, pilots, ex-pilots, and ex-controllers for
qualitative check in the course of model development by prototyping. The obtained comments were fed back for
redesign of the simulation model. This process was repeated a few times.

3.2 Simulation Model for Ground Aircraft Operations

The service processes were next modeled based on the data obtained through the field visit and observation. The
agents identified in the application domain are aircraft (pilots), ground controllers (air traffic controllers), operation
controllers, and passengers. An overview of the ground aircraft operations can be illustrated in Figure 3, which
consists primarily of air traffic control service, transportation service, and in-company operation service. Service
here stands for some sort of activities through which the provider changes the physical or mental state of the receiver
to fulfill receiver’s needs, and a service agent is the provider or receiver of the service.
Every service is performed between providers and receivers. As for the air traffic control service, the providers
are ground controllers and the receivers are aircraft. Radio is used as the physical communication channel, where
messages are exchanged between the providers and receivers. It is the basic pattern of communication in this service
that aircraft request permission to make a move and the ground controllers give instructions on the move.
An aircraft plays a role of the receiver in the air traffic control service, but it provides the transportation service
to passengers at the same time. An aircraft is the medium of this service, with which physical transportation of
passengers is realized. Since assessment of passenger satisfaction was beyond the scope of this study, we did not
explicitly model the transportation service and passengers.
The operation controllers of an airline provide the operation service to aircraft. They also use radio as the
physical communication channel for this service. The operation controllers determine the timing of departure
considering the flight schedule and the status of aircraft. If the pilot of departing aircraft has received an instruction

125
 Makino, Furuta, Kanno, Yoshihara, and Mase: Interactive Method for Service Design Using Computer Simulation
Service Science 1(2), pp. 121-134, ©2009 SSG

from the operation controller, he will request a departure clearance to the ground controller. The operation controller
carries out his/her tasks following some scheduling policy of the airline.

Figure 3 Overview of the Service System.

Operation Ground
controller controller

Operation Air traffic

service control service

Aircraft

Weather
Service
Transportation
service
Service agent Ground

Passenger
Environment

3.3 Agent Model

Relevant actors in the service system are modeled as agents. Each agent has some internal states defined by the
agent class, and an agent changes behavior in different states. One can handle the action rules of agent in a modular
form by defining agent’s internal states. The action rules as well as the state transition rules were extracted from
some documents or from the data obtained through the field observation and implemented into the agent models.
State transition is determined by the current state of agent and the event that triggers the transition, i.e., the message
that the agent receives. The agent is therefore modeled as an automaton.
At the moment, errors or malfunctions are not considered in modeling agent behavior, because the simulator
first aims at replicating system performance under normative conditions. Service resources such as space, time,
manpower, and equipment are not considered explicitly in decision-making of agents, but some constraints on these
resources are reflected in simulation scenarios. The spot already occupied by some aircraft, for example, can never
be assigned as the destination of an arriving aircraft in duplicate. Neither the cost nor the priority of tasks except the
length of taxiing route is considered, either, but the action rules and state transition rules obtained from the field data
implicitly reflect these factors.
Explanations of the primary agent models are given below.

3.3.1 Aircraft Model

The agent representing an aircraft has nine internal states: landing, arrival, stay, towing, pushback, stay after
pushback, departure, lineup, and takeoff.
An arriving aircraft is generated as an aircraft agent at the end of runway, and its state is set “landing” at the
beginning. Having exited from the runway, the aircraft agent requests the taxiing route to the ground controller agent,
changes its state to “arrival,” and moves along the taxiing route to the spot of destination. When the aircraft agent
arrives and stops at the final destination, it changes its state to “stay.”
A staying aircraft agent does nothing, but it requests a clearance to the ground controller agent and changes its
state if it receives any instruction from the operation controller agent. An operation to move an aircraft from the spot
staying at present to another using a towing car is called towing. If a staying aircraft agent receives a towing
instruction, it is towed in the state of “towing” to the spot of destination and stays there again.
If the operation controller agent gives a pushback instruction, a staying aircraft agent changes its state to
“pushback” and it is pushed back following the pushback pattern instructed by the ground controller agent. Having

126
 Makino, Furuta, Kanno, Yoshihara, and Mase: Interactive Method for Service Design Using Computer Simulation
Service Science 1(2), pp. 121-134, ©2009 SSG

finished the pushback operation, the aircraft agent changes its state to “stay after pushback.” After separation of the
towing car and start-up of engines in this state, the aircraft agent requests a taxi clearance to the ground controller
agent.
A departing aircraft agent that has been instructed the taxiing route changes its state to “departure,” and starts to
move along the taxiing route. When the aircraft agent enters a runway after having been approved entrance to
runway, it lines up and waits for a takeoff clearance in “lineup.” Having obtained the clearance, the aircraft agent
changes its state to “takeoff” and starts a takeoff action. When the aircraft agent in takeoff reaches the system
boundary, it is eliminated from the system model.
An aircraft agent holds state variables like the current position, moving direction, and velocity in addition to
static parameters like the hardware specs of aircraft. An aircraft agent moves on taxiways and runways following
instructions from the ground controller agent. An aircraft agent moves toward the preliminary destination point
along the taxiing route, and it moves further to the next destination point if it has passed the previous one. Such
behavior to follow consecutive points along the taxiing route is consistent with the actual movement of an aircraft.
The steering and velocity control model determines movement of an aircraft agent. An aircraft agent takes
obstacle avoidance behavior to slow down its velocity if another aircraft agent enters the area of pilot’s cognition,
which is defined for each aircraft agent. The velocity control model used in this work is basically equivalent to the
optimal velocity model (Bando et al. 1995), which has been proposed as a microscopic velocity control model of
vehicles in traffic congestion.

3.3.2 Ground Controller Model

Seven states are defined for the ground controller agent: standby, arrival, pushback, departure, towing, approach,
and takeoff. Shifting among these states, the ground controller agent instructs many aircraft agents to move on the
ground of airport without any interference.
In “standby,” the ground controller agent has no specific tasks but monitoring the status of aircraft and the
ground. The controller agent is continuously scanning the current locations of aircraft agents and predicting their
movements. If the controller agent receives a request for taxi from an arriving aircraft agent, it plans the taxiing
route toward the arriving spot and gives necessary instructions in “arrival.”
When the aircraft agent that has got ready to depart requests a pushback clearance, the ground controller agent
determines the pushback pattern to be used and gives a pushback instruction in “pushback.” The selection rules of
pushback pattern acquired from the field observation implicitly reflect the taxiing cost afterwards. Having been
requested a clearance for taxi, the controller agent plans the taxiing route to the runway and gives instructions on the
route in “departure.” The controller agent that has received a towing request plans the towing route and gives
instructions on the route in “towing.“ The controller agent gives permission to enter the runway in “approach,” and
then gives a takeoff clearance to an aircraft agent awaiting takeoff in “takeoff.”
The agent representing ground controllers performs its tasks in three steps: recognition of the environment,
prediction of aircraft movements, and decision-making for instruction. The controller agent recognizes the present
situation based on the information collected from the external environment: the status of the ground, locations of
aircraft agents, weather conditions, messages from aircraft agents, and so on. It then predicts aircraft movements in
the near future, makes a decision by the internal reasoning mechanism, and then gives instructions to aircraft agents.
The reasoning mechanism is based on a conventional rule-base system that uses the action rules extracted from
the field observation. In addition, the controller agent plans taxiing routes by A* algorithm considering route length
as the taxiing cost, but taxiing routes are restricted within specific alternatives that are usually used in the actual
operations. The controller agent handles ground traffic in a FCFS (First Come First Served) style, which is the same
as in the actual operations. This service style together with the obstacle avoidance behavior of aircraft agents can
implicitly resolve the constraints on space and time.

3.3.3 Operation Controller and Environment Model

The operation controller agent gives instructions to aircraft agents referring to the flight schedule and considering
the present status of aircraft agents. Two types of instructions are possible: departure and towing instruction. The
operation controller agent of this study decides the start timing of move with a simple rule to follow the flight
schedule. The operation controller agent therefore is just an event generator, and the priority of flights is implicitly
assumed in the prescribed flight schedule.
The environment model consists of the ground model and the weather model. It represents the field and the
surroundings where the airport service is provided. The ground model consists of various ground facilities of airport:
runways, taxiways, and spots. A taxiway is a path where aircraft move between runways and spots. Each taxiway
consists of many path segments and joints. A spot is a location where an aircraft parks for embarkation,

127
 Makino, Furuta, Kanno, Yoshihara, and Mase: Interactive Method for Service Design Using Computer Simulation
Service Science 1(2), pp. 121-134, ©2009 SSG

disembarkation, or maintenance. The wind direction is the only component of the weather model at present. It
affects the usage of runways, and the ground controller agent decides how to use runways depending on the current
status of wind direction. Two patterns are now in use for the runways at Haneda Airport: one for a north wind and
the other for a south wind.

3.3.4 Service Process Model

Having designed relevant agents for the service system, the protocol and processes of communication were
determined. Aircraft agents carry out the service processes by radio communication with the ground controller and
operation controller agent. It is the basic process of the air traffic control service that aircraft agents request service
to the ground controller agent, and the latter delivers instructions in reply. In addition, the ground controller agent
may spontaneously gives instructions judging from the situation of aircraft agents on the ground. Figure 4 shows the
flow of service processes.

4. Service Simulation
4.1 Architecture of Simulation System
A simulation system has been developed for ground aircraft operations based on the agent-based simulation model
explained so far. The simulator development was done on a Java platform for Windows PCs. Most of the
information required for setting up simulation cases is given as input data to the simulator so that assumptions for
simulation can be modified easily without changing the simulation model. The input data include the layout of
airport facilities, flight schedule, weather conditions, and some of the action rules of agents. These data also specify
design options of the airport service. The output data of simulator include chronological actions of agents.
The simulation controller runs many agents in turn, but the cycle time of simulation, 100 ms, is short enough
that agent actions including communication look asynchronous in a long time span. No task interruptions are
considered, and the agents handle events sequentially. In addition, it is assumed that agent’s internal process is
carried out within a single simulation cycle. These approximations, however, cause no limitations so far for the
present simulation purposes, because agent actions are organized in a small unit compared with the whole duration
of simulation scenario and because the time required for some object to move in space or for some condition to
obtain dominates event timing.
Results of simulation are visualized with a graphical user interface shown in Figure 5. Movements of aircraft
are displayed by animation of icons on the airport map. Flight data including aircraft type, call sign, agent’s internal
state, destination, and so on are shown in text near the icon. The area of pilot’s cognition and the planned taxiing
route of each aircraft are also displayed. The messages exchanged between agents are listed in the separate
communication window.
The visualization makes it easy to find out relevant problems not only for the designers but also for the field
experts as a sense of disagreement with the reality, which could not have been detected just by interview. It
contributes to brush-up of the simulation model. For example, we modified the model so that unrealistic movements
of aircraft and unrealistic controller’s instructions were fixed, and that the agents made decisions depending more on
the ground situation than at the early stage of model development. We could also fix unrealistic design options that
are unfeasible due to constraints just implicitly shared by the experts.
The visualization is also effective to detect problems hardly recognized in specific field sections; one can notice
new issues from a comprehensive viewpoint by looking at a variety of information provided by simulation. It
therefore can be a good media through which staff members in different service sections communicate each other
and share problems and solutions beyond the boundaries between different professions.

4.2 Test Simulation for Model Validation

A test simulation was performed to check the validity of the simulation model. The time required for arrival and that
for departure were chosen as the performance measure, because they determine whether or not flights are on time
and then most affect passengers’ convenience in the airline business. Fuel savings from reduction of delays also
contribute to cuts in CO2 emission as well as operation costs. The time required for arrival from entering the field
boundary of observation to stopping at the spot, and that for departure from requesting pushback to exit from the
field boundary were evaluated from the video taken at the airport for three hours during daytime, December 10,
2007. The simulation scenario used for the test was set up from the flight schedule and the weather conditions
during the same time interval.

128
 Makino, Furuta, Kanno, Yoshihara, and Mase: Interactive Method for Service Design Using Computer Simulation
Service Science 1(2), pp. 121-134, ©2009 SSG

Figure 4 Flow of Service Process

Operation controller Passenger Aircraft (Pilot) Ground controller

Landing
Landing
request
Request taxi Plan taxiing route

Arrival
Receive taxiing route Instruct taxiing route
delivery
Arrive at spot

Stop at spot
Stay
Embarkation or
disembarkation Standby

Instruct
departure or towing Receive instruction

Towing
request
Plan pushback
Request pushback pattern
Pushback

Receive pushback Instruct pushback

pattern pattern
delivery
Separate towing car
pushback
Stay after

& start engines

request
Request taxi Plan taxiing route

Receive taxiing route Instruct taxiing route

Departure

delivery
Arrive at runway end

Request entry request

to runway
Entry clearance
to runway
Enter runway delivery
Lineup

Request takeoff request

Takeoff clearance
Takeoff delivery
Takeoff

129
 Makino, Furuta, Kanno, Yoshihara, and Mase: Interactive Method for Service Design Using Computer Simulation
Service Science 1(2), pp. 121-134, ©2009 SSG

Figure 5 Graphical User Interface of Simulator

Area of pilot's cognition
Communication window
Wind direction Clock

Airport map Flight data

Time control slider
Planned taxiing route

Figure 6 Comparison of Time Required for Arrival and Departure

Time predicted by simulation (s)

300

200

100
y = 1.016 x
r = 0.88
0
0 100 200 300
Time evaluated by field observation (s)

The time predicted by simulation was compared with that evaluated from the video. As mentioned before, the
simulation model had been improved to eliminate unrealistic agent behaviors by showing simulation results to the
field experts and collecting their comments until the final results were obtained. Figure 6 shows that a correlation

130
 Makino, Furuta, Kanno, Yoshihara, and Mase: Interactive Method for Service Design Using Computer Simulation
Service Science 1(2), pp. 121-134, ©2009 SSG

coefficient of 0.88 and a regression coefficient of 1.016 were attained between the simulated and measured data; the
simulation could well predict the time required for arrival and departure.
The simulation system can temporally simulate aircraft movements on the two-dimensional plane of airport and
agent actions by the elementary task of ground aircraft operations. Model validation in a microscopic view is
therefore possible, where the agent actions simulated are compared with those actually observed in the field from
moment to moment. The macroscopic performance measure of time required for arrival and departure was adopted
in this study, because the airline is highly concerned about this measure in terms of service quality and it could
fulfill the preliminary objective of service design.

5. Service Planning
The service simulation presented in the previous chapter revealed problems in the present situation of ground
aircraft operations at Haneda Airport. The concave shape of the second terminal building and the four open spots,
which are located apart from the building, cause interference between aircraft and consequent delays in arrival and
departure of aircraft. Figure 7 schematically shows the layout of ground facilities in front of the second terminal
building. The aircraft pushed back from any of the open spots obstruct the taxiway in front of the terminal building.
To solve this problem, alternative design options were proposed for usage of the open spots from Spot 81 to 84
considering the comments from those who looked at the simulation results. The time required for arrival and
departure was the performance measure also for assessing the proposed options by simulation. Here the time
required for arrival was defined as that between exit from the runway and stop at the spot, and the time required for
departure between requesting pushback and entrance to the runway.

Figure 7 Layout of Ground Facilities in Front of the Second Terminal Building

E6 E5
Taxiway
Spot
I-TWY

56 J6
81 82 Z 83 84 J4 68
57 R3 R2 67

58 66
59 60 61 62 63 64 65

Second Terminal Building

Two alternatives were first examined: Plan A and B. In Plan A, Spot 82 and 83 are removed and converted into
taxiways. In Plan B, all of the open spots are removed and the apace is used as pushback areas. Simulations were
carried out for the two design options using the scenario based on the actual record of operations for about two hours
on June 27th, 2008, which corresponds to rush hour of the day.
Figure 8 shows the total reduction in time required for arrival and departure compared with the present
operation scheme. The time required for arrival did not decrease greatly, but the total reduction in time required for
arrival and departure was around 20 minutes with Plan A and around 34 minutes with Plan B. If the cost of
reconstruction is ignorable, Plan B is more attractive than Plan A.

131
 Makino, Furuta, Kanno, Yoshihara, and Mase: Interactive Method for Service Design Using Computer Simulation
Service Science 1(2), pp. 121-134, ©2009 SSG

Figure 8 Time Required for Arrival and Departure (Plan A and B)

2000

Time Required for Departure (s)

Arrival 1200
Reduction in Time Required (s)

184
Departure
1500
23
800
1000
1650

1187 400
500 Present
Plan A
Plan B
0 0
Plan A Plan B 1 3 5 7 9 11 13 15
Flight Number

The field experts pointed out, however, that the cost of reconstruction for adopting Plan A or B is considerable,
and that reduction of available spots is unfavorable because of the planned extension of the airport. Two other
alternative plans were therefore considered: Plan C and D. In Plan C, the space of the open spots is used as pushback
areas for Spot 58 through 66 during daytime. In Plan D, the aircraft leaving from Spot 56 or 57 is pushed back to
Spot 81, and that from Spot 67 or 68 to Spot 84, in addition to the same operation scheme of Plan C.
Plan C and D were assessed with a simulation scenario of operations for about two hours on October 7th, 2008.
The total reduction in time required for arrival and departure was about 11 minutes with Plan C, and about 20
minutes with Plan D as shown in Figure 9. It was confirmed by the computer simulation that use of the open spot
space as pushback areas is effective to avoid interference of aircraft in front of the second terminal building and
contributes to speedup of departure. The field experts who looked at the aircraft movements predicted for Plan D
stated concerns on safety in terms of the aircraft pushed back to Spot 81 and 84. Consequently Plan C was adopted
for further consideration, which is not only effective for efficient aircraft operations but also free from problems in
terms of reconstruction cost and safety.

Figure 9 Time Required for Arrival and Departure (Plan C and D)

1200
1500
Time Required for Departure (s)

Arrival
Reduction in Time Required (s)

11
Departure
1000
800

1229
500
653
400
0 Present
-17
Plan C
Plan D
-500 0
Plan C Plan D 1 3 5 7 9 11 13
Flight Number

As described above, an interactive process of identifying problems in the present services, proposing alternative
design options to solve the problems, assessing the alternatives, and checking the results of assessment is very
effective for human-centered design of service systems. Computer simulation and visualization of its results are key

132
 Makino, Furuta, Kanno, Yoshihara, and Mase: Interactive Method for Service Design Using Computer Simulation
Service Science 1(2), pp. 121-134, ©2009 SSG

technologies to realize this process, because the field experts as well as the designers can easily understand
simulation results and recognize implicit problems in design options.

6. Conclusion
A human-centered design method for service systems that consists of three phases of service modeling, service
simulation, and service planning has been proposed in this study. This method relies on ethnographic field
observation and computer simulation. An agent-based approach has been adopted for computer simulation, which
approach is suitable for representing complex service systems. An interactive process where visualized simulation
results are shown to the field experts repetitively enhances rapid prototyping and obtaining valuable feedbacks from
them. Construction, modification, and validation of agent models are thereby performed considering the knowledge
of field experts. In addition, identifying problems in the present service system, proposing alternative design options,
and their assessment are also carried out in a similar interactive process.
The proposed method was applied to a specific domain of ground aircraft operations at Haneda Airport for
demonstration. A model of the service processes in the application domain was constructed by field visit and
observation. The agents relevant to the services were defined and their internal mechanisms were implemented that
fit to behaviors observed in the actual operations. A test simulation was performed and comparison of the simulation
results with the field data demonstrated that computer simulation of ground aircraft operations is feasible.
As a result of test simulation, it was also revealed that the present layout of ground facilities at the airport
results in interference between aircraft. Four alternative solutions were proposed and assessed by simulation. It was
evaluated that the design option finally chosen can reduce the time required for departure by about 11 minutes for
two hours of rush hour operations without concerns on reconstruction cost and safety. This case study showed that
the computer simulation was very useful for proposing and assessing design options to solve the problems identified.
The interactive method was effective also in this course, because the tacit knowledge of field experts could be
extracted and considered in design decisions.
Though a specific domain, ground aircraft operations, was the only application domain in this study, the design
method is sufficiently general so that it is expected applicable to a variety of service domains that depend on human
expertise and performance. It will contribute to realizing rational service design that can consider various dynamic
and microscopic factors affecting service processes and predicting service performance in advance.
It will probably be a limitation of this method that it can hardly deal with infrequent or inexperienced cases of
situations, because it heavily depends on ethnography and experts’ knowledge. Neither errors nor unanticipated
events are considered in the present simulation model, either. Simulation will be possible even for such non-normal
cases, but feedbacks from the field experts are less expectable and the validity of simulation will be imperfect. Since
no other approaches exist, the interactive design method is still useful to a considerable extent within these
limitations.

References
Arai, T., Y. Shimomura. 2004. Proposal of Service CAD System - A Tool for Service Engineering -. Annals of CIRP
53(1) 397-400.
Baines, T. S., et al. 2007. State-of-the-art in product-service systems. IMechE Part B: Journal of Engineering
Manufacture 221 1543-1552.
Baines, T. S., O. Benedettini. 2007. Modelling human performance within manufacturing systems design: from a
theoretical towards a practical framework. Journal of Computer Simulation 1(2) 121-130.
Bando, M., K. Hasebe, A. Nakayama, A. Shibata, Y. Sugiyama. 1995. Dynamic model of traffic congestion and
numerical simulation. Physical Review E51 1035-1042.
Buur, J., T. Binder, E. Brandt. 2000. Taking Video beyond ‘Hard Data’ in User Centered Design. Participatory
Design Conference 2000, 121-131.
Council on Competitiveness. 2004. Innovate America. National Innovation Initiative Report.
Ferber, J. 1999. Multi-Agent Systems: An Introduction to Distributed Artificial Intelligence. Addison Wesley,
Harlow UK.
Kazda, A., R. E. Caves. 2007. Airport Design and Operation. Elsevier, Oxford UK.
Kumar, V. 2004. User insights tool: a sharable database for global research. Institute of Design, Illinois Institute of
Technology.

133
 Makino, Furuta, Kanno, Yoshihara, and Mase: Interactive Method for Service Design Using Computer Simulation
Service Science 1(2), pp. 121-134, ©2009 SSG

Manzini, E., C. Vezzoli. 2003. A strategic design approach to develop sustainable product service systems:
examples taken from the ‘environmentally friendly innovation’ Italian prize. Journal of Cleaner Production 11
851-857.
NASA. 2009. MIDAS: Man-Machine Integration Design and Analysis System. Retrieved on June 4, 2009,
http://humansystems.arc.nasa.gov/groups/midas/index.html
Neely, A., H. Richards, J. Mills, K. Platts, M. Bourne. 1997. Designing performance measures: a structured
approach. International Journal of Operations & Production Management 17(11) 1131-1152.
Neely, A., M. Gregory, K. Platts. 2005. Performance measurement system design: A literature review and research
agenda. International Journal of Operations & Production Management 25(12) 1228-1263.
Oliva, R., R. Kallenberg. 2003. Managing the transition from products to services. International Journal of Service
Industry Management 14(2) 160-172.
Shostack, L. G. 1982. How to Design a Service. European Journal of Marketing 16(1) 49-63.
Shostack, L. G. 1984. Design Services that Deliver. Harvard Business Review 62(1) 133-139.
Spohrer, J., P. P. Maglio. 2008. The emergence of service science: Toward systematic service innovations to
accelerate co-creation of value. Production and Operations Management 17(3) 1-9.
Suchman, L. 1987. Plans and Situated Actions. Cambridge University Press, Cambridge UK.
Tukker, A. 2004. Eight-Types of Product-Service System: Eight Ways to Sustainability? Experiences from
SusProNet. Business Strategy and the Environment 13(4) 246-260.

Biographical Notes
Yuki Makino was graduated and obtained B.Eng. from the Department of Architecture, the University of Tokyo in
2007. He was graduated and obtained M.Eng. from the Department of Quantum Engineering and Systems
Science, the University of Tokyo in 2009. He is now working for IBM Japan, Ltd.
Kazuo Furuta was graduated from the Department of Nuclear Engineering, the University of Tokyo, and obtained
B.Eng. in 1981, M.Eng. in 1984, and Dr.Eng. in 1986. He was a researcher, the Central Research Institute of
Electric Power Industry (CRIEPI), a lecturer, and associate professor, the University of Tokyo. He has been a
professor, the Graduate School of Engineering, the University of Tokyo since 2004. His research interests are
now on cognitive systems engineering, social design for safety, and service design.
Taro Kanno is an Associate Professor in the Department of Systems Innovation, the University of Tokyo. He
received his M.Eng. and Dr.Eng. in cognitive systems engineering from the University of Tokyo. His research
interests include human factors and cognitive ergonomics, service cognition, and the modeling and simulation
of team and organizational cooperation and coordination for service systems design.
Shigeki Yoshihara was graduated from Waseda University in corporate strategy and joined All Nippon Airways
(ANA) Co., Ltd. in 1991. He was the manager of the head office on public relations, and then on government
and industry affairs. He has been the deputy director on corporate planning since 2005 and at the same time
belongs to the ANA Strategic Research Institute Co., Ltd. He is also a panel member of Japanese government
on safety transportation of dangerous goods, the International Civil Aviation Organization (ICAO).
Takamichi Mase is a chief researcher of the Division of Research and Development, ANA Strategic Research
Institute Co., Ltd

134