University of California Transportation Center
UCTC- FR- 2010- 26
Measuring Multimodal Transport Level of Service
Adib Kanafani and Rui Wang
University of California, Berkeley
August 2010
Measuring Multimodal Transport Level of Service
1
Final Draft
UCTC Faculty Research Report
Measuring Multimodal Transport Level of Service
Adib Kanafani and Rui Wang
University of California, Berkeley
July 2010
Abstract
One of the challenges facing intermodal integration is that the planning framework needed
for it lacks appropriate measures of level of service that cut across the modes involved and
the connections between them. In this study we develop a framework and a set of metrics of
level of service in a multimodal context. We propose a conceptual framework in which we
identify the various attributes of level of service and the method of their integration. These
measures of performance are defined from two perspectives: the user’s perspective ( the
demand side) and the provider’s perspective ( the supply side). An analytical framework is then
proposed in which a working definition of a “ multi- modal corridor” is adopted and a
methodology for defining and combining measures of performance for such a corridor is
developed. The methodology is defined in the context of evaluation for the purpose of
choosing among alternative corridors. The approach is grounded in utility theory and
quantitatively these measures of performance are defined as indirect utility functions of the
type used in choice models.
In combining the measures of performance for different elements of a multi- modal corridor,
the methodology recognizes that some are additive, either simply or with appropriate
weights, while others are not additive at all and exhibit phenomena such as weakest link, or
maximal effort. Safety is a good example of this. The basic proposition is that many level of
service metrics are non- additive and their combination for a multimodal systems requires
specific models that reflect the way the attributes impact users of different modes and during
different segments of a multimodal journey.
This study concludes by recommending some research directions to develop the models
needed for the integration of level of service measures for multi- modal corridors and for
their inclusion in indirect utility function.
Measuring Multimodal Transport Level of Service
2
Table of Contents 1 INTRODUCTION 4 2 LITERATURE REVIEW 6 3 MULTIMODAL MEASURES OF PERFORMANCE – A CONCEPTUAL
FRAMEWORK 9 3.1 A MULTIMODAL ALTERNATIVE: DEFINITION 9 3.2 DECISION PERSPECTIVES: 10 3.3 FRAMEWORK: INTEGRATION OF MEASURE OF PERFORMANCE 10 3.3.1 TRAVEL TIME 11 3.3.2 MONETARY COST 12 3.3.3 SAFETY 13 3.3.4 RELIABILITY 13 3.3.5 FLEXIBILITY 14 3.3.6 ENERGY 15 3.3.7 INVESTMENT AND AGENCY COST 15 3.3.8 EQUITY 15 3.4 ANGLE OF PERCEPTION 16 4 ANALYTICAL FRAMEWORK - USERS’ PERSPECTIVE 18 4.1 TIME COST: PERCEIVED TRAVEL TIME 18 4.2 MONETARY COST 21 4.3 RELIABILITY 21 4.4 MODE- SPECIFIC PARAMETER 25 4.5 SOCIODEMOGRAPHIC VARIABLES 26 4.6 DEMAND MODELING 26 4.7 SUMMARY 27 5 ANALYTICAL FRAMEWORK - SUPPLIERS’ PERSPECTIVE 29 5.1 EQUITY ISSUE 29 5.2 ENERGY CONSUMPTION: 29 5.3 EXTERNALITY/ EMISSION 30 5.4 LEVEL OF SERVICE: 30 5.5 MONETARY COST: INVESTMENT COST, SUBSIDY 30 5.6 SUMMARY 31 6 ON THE ISSUE OF “ THE LAST MILE” 32 6.1 SCHEDULING COORDINATION 33 6.2 PHYSICAL INTEGRATION 34 6.2.1 STRUCTURAL UTILITY 37 6.2.2 INFORMATION ASSISTANCE 37 6.2.3 FACILITY UTILITY 38 6.3 THE LAST MILE: INTEGRATION 38
Measuring Multimodal Transport Level of Service
3
7 CONCLUSION AND FUTURE WORK REFERENCES: 3419
Table of Figures
Figure 3- 1: Case Study Network 10
Figure 3- 2: Users’ Perception 16
Figure 3- 3: Government’s Perception 17
Figure 3- 4: Agency’s Perception 17
Figure 4- 1: Time 18
Figure 4- 3: Utility VS. Total Travel Time 20
Figure 4- 4: Monetary Cost 21
Figure 4- 5: Reliability 22
Figure 4- 5: Reliability Statistics 22
Figure 4- 7: Mode Factor 25
Figure 4- 8: Demand Summary 27
Figure 5- 1: Supply Summary 31
Figure 6- 1: Calibrated Coefficients 34
Figure 6- 2: JFK Plan View 35
Figure 6- 3: JFK Transfer Choice 36
Measuring Multimodal Transport Level of Service
4
1 Introduction
One of the challenges to the planning of multimodal transportation systems is the
development of measures of performance that properly reflect the level of service provided
by these systems. In thinking about multimodal level of service measures we face conceptual
and analytical challenges that stem from the need to integrate the measures of performance
for different components of a multimodal system. This is especially the case when these
measures are categorical, or non- additive, which complicates their integration into evaluation
models. Similar metrics are commonly used to reflect the level of service of different
elements ( time, cost, and so forth,) but they are considered for each mode separately, even
though the transition from a metric to a measure of level of service will take on different
scales in the different modes, as well for the intermodal connection. The resulting distortion
can mean inadequate assessment of the impact of intermodal connections on overall
multimodal performance. Simply comparing the sums of these elements among different
multimodal alternatives is not a sound basis for evaluating overall performance. This can be
seen in the issue of “ the weakest link”. As it appears in the decision- making process, and
especially for issues such as reliability, rather than looking at the average of the entire picture,
people have more concern about the link that provide the least level of guarantee. With
inadequate integration methodology, the problem may be underestimated or even
overlooked. Multimodal systems work only if the modes are integrated well at the interface,
and it is this integration that suffers the most from inadequate level of service definition and
measurement. A new analytical framework for multimodal MOP is indispensible.
Besides the challenges in the integration process, it should be also noted that two parties
would perceive the performance of the network: the demand side and the supply side. Their
perception will have different focus but still related. Measures of performance have a
multifaceted purpose and are needed in a wide variety of planning, design and policy analysis
of transportation systems. Level of service measures represent set of all the measures of
performance needed for transportation analysis and reflect primarily the attributes of the
system that affect user perception of the quality of service. In this report, the demand side is
discussed as “ the user’s perspective” ( Section 4), because it represents the response of the user
and the market; whereas the supply side being “ the supplier’s perspective” ( Section 5), level of
service perceived by a system.
There are multiple levels of integration: the integration of decision factors of one mode,
the intermodal integration, and the integration of individuals across the society. These
attributes may be additive ( such as monetary cost) or non- additive ( such as safety, reliability,
etc), or even more complex and needs to be evaluated from multi- perspectives ( such as
travel time). We will try analyzing their features more into details in later sections.
In this research we looked at the various measures of level of service for an array of
multimodal transportation systems and categorize them in terms of the feature in which they
are to be combined. With this conceptual framework in place we would move toward
defining a quantitative framework with models for combining the various measures to come
up with overall multimodal metrics. In this we would rely on choice model theory to
consider the way in which separable metrics can be combined. We relied on probability
theory to consider the combination of probabilities metrics such as reliability. We discussed
Measuring Multimodal Transport Level of Service
5
the possibility and validity means of assessing some subjective, such as comfort and
convenience. We also reviewed existing literature to come up with numerical examples to
illustrate the framework. We developed improvements of existing methodological systems of
transportation analysis rather than defining completely new methodologies. The reason for
this is to overcome the inertia of adoption and to facilitate expediency in use of our research
findings.
We produced is a framework and a set of metrics for evaluating the level of service of
transportation systems at a multimodal level. Our scope includes urban and regional
transportation and the focus, empirically, is on California. We identified innovative methods
of providing these connections, and their impact on the effectiveness of intermodalism. We
also provide innovative institutional and legislative perspectives that may contribute to the
policy debate on how to achieve a true multimodal transportation system. Furthermore, we
identified the issues that should be further investigated into in the future.
The second section of this report will be the literature review. The third section is our
conceptual framework: identifying the attributes, and their respective roles in from different
perspectives ( demand, supply). In section 4 and 5, we will introduce a simple illustrative
multimodal transportation network trying to demonstrate the analytical framework we
developed. In section 4, we will try to model the decision making process from a user’s point
of few, and come up with an indirect utility function and a demand function for multimodal
transportation network. In section 5, we will be constructing the evaluating framework from
the suppliers’ perspective. In the last section, we will make policy recommendations basing
on these analytical results.
Measuring Multimodal Transport Level of Service
6
2 Literature Review
Literature related to the topic lies in two aspects: multimodal system evaluation and the
impact of each attribute we are trying to look into.
Multimodal transportation
The work of Lyons, Harman, Austin and Duff ( 2001) is based on the U. K.’ s ten year plan
( starting from year 2000), trying to provide the public with the opportunity to compare travel
options across public and private transportation modes, seeking to offer a one- stop- shop
journey planning. By reviewing abundant literature across the globe, the work identified
critical topics, their findings and limitations. The topics mainly lie in the government’s
planning perspective, its feasibility and information gathering process. The report offers
highlights on the challenges during the government’s decision- making.
Li and Wachs ( 2000) proposes a set of inter- modal performance indicators in which
service input, service output and service consumption are measured by total cost, revenue
capacity, and unlinked passenger trips/ miles respectively based on economic principles and
evaluation objectives, and concluded that their enhanced inter- modal performance indicators
are more appropriate for comparing the efficiency and effectiveness of different modes or
combination of transit modes. The concern of inter- modal performance can be view as a
proof of the importance of differences in multimodal transportation planning process.
In the study of Richard H. Pratt and Timothy J. Lomax ( 1996), Performance Measures for
Multimodal Transportation Systems, they observed the importance on: Match performance
measures with objectives; Understand the effect of improvements; Address people and
goods; Use common denominators; Development of measures should not be governed by
data concerns; Employ both multimodal and mode- specific measurement; and Remember
the audience.
In the work of Robert A. Johnston ( 1994), the evaluation of multimodal transportation systems for
economic efficiency and other considerations, the concern of multimodal transportation evaluation an
planning is highlighted. The work shows that the concern of multimodal transportation
planning in improving planning efficiency. The paper deals the issue more from an ITS point
of view.
In the DOT’s strategic plan 2006- 2011, the strategies, milestones, and outcome as well as
goals of the U. S. Department of Transportation is highlighted, offering the guidance of
emphasis of policy making, which is important in the suppliers’ perspective in our on- going
project. The plan stands from a more political point of view, whereas the technical
perspective is not a major issue of concern.
Modeling Attributes
Despite the abundance of literature on transportation measures of performance, there is
little that has addressed the question of the multi- modal integration of attributes of level of
service. In our literature reviewing process, we found the following issues are more of global
research interest:
Measuring Multimodal Transport Level of Service
7
• Time
In the work of Bates et. al. ( 1987), travel time variability is classified into three categories:
i. Inter- day variability caused by seasonal or day- to- day variations: demand fluctuation, accidents, road
construction and weather charges;
ii. Inter- period variability, which reflects the impact of differences in departure times and the caused
charges in construction;
iii. Inter- vehicle variability mainly due to individual driving styles and traffic signals
Noland and Polak( 2002) used similar categories to represent travel time variability;
differences in travel time from day- to- day, over the course of the day and even from vehicle
to vehicle
In Bates et al.( 2001), it is further added that on the demand side, after considering
seasonal effects, day- of week variation another systematic variations, the residual day- to- day
variations are essentially random, whereas the randomness on the supply side is mainly due
to incidents, such as vehicle breakdowns, signal failure, etc.
Bates et al. 2001, Hollander 2006, Hollander and Steer Davies Gleave 2009 emphasis the
stochastic feature of travel time variability, and defined travel time reliability as random
variation in travel time.
In Noland and Polak ( 2002) identified the difference between travel time variability and
congestion, as a transportation system with severe congestion may have very stable day- to-day
travel time.
Many measures of travel time variability have been developed in the extant literature,
while one common feature is the recognition that travel time distribution is impacted by day-to-
day fluctuation on the demand side as well as the supply side of the traffic
Travel time variability is most commonly represented by lognormal distribution ( e. g.
Rietveld et al. 2001, Giuliano, 1989). Bates et al. found that a generalized Poisson
distribution to better describe the delay distribution for train travel time.
• Reliability
In the study of Rietveld et. Al. ( 2001), the issue of reliability is studied in a multimodal
perspective, in the case of public transportation system in Netherlands. The study shows
explicit attention on one missing the connection between elements on a chain. The data
from various unimodal resources, partly thin, need to be combined. Customer valuation of
unreliability is estimated by means of a stated preference approach. The result is, however,
difficult to transfer into other countries due to the high popularity of bicycles in Netherland.
Li, Hensher and Rose ( 2010) studied the issue of reliability in terms of willingness to pay
with reviews of empirical evidences from the Europe, the U. S. and also provides new
evidence from Australia. The review focuses on car, rail and bus, each by their single mode,
showing the significance of reliability in travelers’ decision making.
• Safety
Moen studied the determinants of safety priorities in transport, from the perspective the
effect of personality, worries, attitudes and willingness to pay in his work 2007. The study is
conducted by means of survey, and receiving 1727 returned questionnaires. Several factors
Measuring Multimodal Transport Level of Service
8
were found to be important to priority of safety. The three personality assets: trust,
excitement seeking and anxiety, were measured along with measurements of optimism,
worry, attitudes and willingness to pay. They were all tested by factor and reliability analysis.
The results were satisfactory and the measurements were used in both a regression analysis
and a SEM- path analysis to predict priority of safety.
Jessup et. Al. conducted a feasibility study evaluating transportation security systems and
associated multi- modal efficiency impacts. The issue of terrorism attack is the major concern
study. In this study, a constrained transportation optimization model was developed to
estimate the effects that security related impacts had on an electronic firm’s supply chain of
televisions through the six major west coast ports. This modeling effort was developed using
primary data obtained through interviews with the firm, and maritime experts.
• Energy and Emission
Delucchi ( 2000) suggested in his Environmental Externalities of Motor- vehicle Use in
the US, that the marginal impact is increasing with the existing pollutant level. The
perceivable impact includes human illness, visibility reduction, agricultural loss, etc. Although
there remains considerable uncertainty in all stages for modeling the damage cost, the results
enriches cost- benefit and pricing analyses from a larger extent. The estimation of external
costs has been used for comparing the social costs of different transport technologies or
modes, evaluating the trade- offs between different environmental impacts, and analyzing
policies.
There are more literature on this issue from the perspective of taxing, ( Ramseur and
Parker, 2009; GAO 2010), and emphasize on the measurements of control rather than
evaluation.
Measuring Multimodal Transport Level of Service
9
3 Multimodal Measures of Performance – A Conceptual
Framework
The measure of performance of a multimodal transportation alternative describes both
its positive and negative impact. Depending on the perspective, ( demand or supply) there
can be multiple criteria and objectives. We therefore use an analytical framework grounded
in utility theory leading to the construction of indirect utility function for any multi- modal
alternatives, such that the elements of that function can vary depending on the perspective,
and the use for which the function is intended.
3.1 A Multimodal Alternative: Definition
Before introducing the metrics we adopt a working definition of a multimodal
transportation alternative. There is clearly some flexibility in defining a multimodal
transportation system. Essentially all transportation is multimodal, but a working definition
permits the focus on important elements. When more than one mode is involved, each
mode provides “ access” to the next in a chain connecting an origin to a destination.
Accessing a bus at a stop may require walking to the stop or driving to a park- and- ride site.
Whether either of these two configurations is modeled as a multimodal network or a single
mode network with walking access depends on the purpose of the model. The framework
adopted for this analysis accommodates this flexibility in model design and identifies the
following elements of a multimodal system. But first a clarification of terminology: We use
the term ‘ multimodal’ to represent the system with more than one mode, and the term
‘ intermodal’ to represent the connection between any two of these elements.
Access: The concept of access is usually mode specific. As mentioned above it is possible
to consider that each mode is providing access to the next, but for the purpose of this
analysis we reserve the term “ access” to represent the first link connecting between a trip
origin and the first mode to be used in the multimodal system. From a door- to- door
perspective this begins from the moment the user gets out of the door to the point when
he/ she enters the system of the mode. This process may include composite links ( walking,
bicycling etc.) but they are all consolidated together as an access link. If any of these links is
itself the subject of analysis, then it would have to be modeled as a separate mode. In urban
transportation modeling it is customary to consider the access process as individual travel,
such as walking or bicycling,.
Waiting: This is defined as the duration between the a user’s arrival to a transport
terminal and the actual entry into the vehicle of the mode in question.
In- vehicle travel: This is the duration of time on the vehicle.
Transfer: This is defined as the duration between one getting out of the previous mode
and arriving at the entrance point of the following mode.
Measuring Multimodal Transport Level of Service
10
Figure 3- 1: Case Study Network
The use of these definitions to model a multimodal system has to be case specific.
Firstly, overlapping will arise in cases such as a coordinated system. For example, to transfer
between flights at an airport, and if the check- in is completed already, there will not be much
distinction between transfer and waiting. This is, because the reasons to distinguish between
them – reliability, security, and etc.- lose their importance. Secondly, the consistency in scale
will help us better construct the alternative and simplify the analysis. For instance, if one
drives 20 minutes to the airport to take a 1- hour flight, the driving will certainly be
considered as the first mode in the trip. However, if the flight is 15 hrs, it makes more sense
to consider the driving as access to the transportation alternative. Therefore, the definitions
are flexible and should be adjusted depending on the specific scenario, and the purpose of
the application.
3.2 Decision Perspectives:
As mentioned earlier, multimodal measures of performance are defined from two
perspectives: the user’s perspective and the supplier’s perspective. To analyze the user’s
perspective we focus on the individual’s response emphasizing the assessment of user costs
and benefit, as measured by the perceived utility. This can be aggregated later to represent
community welfare impacts. This aggregation can be done using the usual procedures used
in social welfare analysis and based on socioeconomic distribution.
The supplier perspective reflects a producer’s optimization process, or a public agency’s
process of evaluating global impacts. In other words, the supplier can be the government or
private companies. In the context of this report, we are more inclined to consider the
government as the supplier, as their concern is on overall social welfare rather than final
profit. The user’s assessment of benefit is part of the supplier’s optimization objective. But
the supplier is also concerned with other attributes such as investment cost, agency cost,
externalities, and with weighing all these aspects as part of a planning, policy making process.
3.3 Framework: Integration of Measure of Performance
The combination of measures of level of service across modes is the main subject of
this study. An important distinction to make here is between measures that are additive and
ones that are non- additive. Most measures of a multimodal system are of the second category
Mode1 Mode2
TArcacnesssfe/ r Waiting ITnr aVveehli c le
Measuring Multimodal Transport Level of Service
11
and require a modeling framework for combining them. To illustrate this point consider the
example of comparing the door to door travel time between two multimodal systems, a high
speed rail with auto access to stations and an air transport journey with transit access to the
airports. To simply add up the travel time components for each of these multimodal systems
and then compare them ignores important differences in the incidence of these various
components and their contribution to the disutility of travel. It ignores the different impact
of access time and line- haul time. It also ignores the complex question of reliability and the
variance in travel time. It therefore misrepresents the relative performance of these two
alternatives and can lead to invalid conclusions and possibly wrong forecasts.
Even the most obviously additive measure, out- of- pocket cost, which is uniformly
measured in money terms and can be simply added for the components of a multimodal
service, can sometimes behave in a non- additive manner when the different elements of cost
have different tax implications, or subsidy potential. An example of this could be line haul
costs that may be tax- deductible or subsidized by employers but terminal costs ( parking or
access to station) that may not be. Or different cost elements that can be paid at different
times thereby having different impact of the user’s budget. The challenge for dealing with
multimodal measures of performance is how to address the issue of non- additive metrics.
Some of these are separable in that they may be combinable with some linear function. An
example would be line- haul time and access time combined according to incidence
parameters that are estimated from a discrete choice model. Others are non- separable in that
they combine in nonlinear ways and require complex functions. An example of this can be
found in reliability and safety measures that reflect the performance of the weakest link of the
multimodal system. Another example comes from the various measures that reflect variance
of individual metrics requiring the nonlinear combination of standard deviations. The
detailed description of the various measures of level of service is in the following session.
3.3.1 Travel Time
The two main components of the cost of traveled that is incurred by individual users are:
time and money. We begin with travel time. The measurement of travel time is fairly easy,
but the difficulty is in quantifying user’s perception of time especially as it varies between the
different components of the total travel time. . On the individual level, the evaluation of
perceived travel time is based on the definition in the beginning of this section: access,
waiting, transfer, and in- vehicle traveling.
IN VEHICLE TRAVEL TIME
The weight of in- vehicle travel time of different modes in total perceived travel
time depends on the overall level of service provided by the mode, such as level of
comfort, sense of security, etc. For example, the perceived travel time of taking a bus is
usually longer than driving, for the same time of actual time spent. Therefore, again,
these can be interrelated or even overlapping. In addition to adding the different
components of time together with appropriate weights, it is not unusual to define
weights that reflect some attributes that affect the perception of time, such as comfort,
and security. But that would require detailed information as might be obtained from in-
Measuring Multimodal Transport Level of Service
12
depth user surveys. . In the absence of such detailed information the alternative is to we
introduce these attributes as stand- alone categorical variables.
TRANSFER TIME
Due to the uncertainties that may arise during transfer, the transfer time is usually
more heavily weighted than in- vehicle traveling time. The higher weight reflects 1)
effort involved, such as walking; 2) reliability issue and the uncertainty regarding the
intermodal connection being made. Therefore, if one enters into a coordinated system
then the weight should be less than in an uncoordinated connection between different
modes.
ACCESS TIME/ ACCESSIBILITY
From our definition, access time acts in a similar manner as the transfer time in
user’s perception. Meanwhile, access time is a measure of accessibility, especially in
public transportation. Maintaining adequate public transit accessibility is a responsibility
of the government, and also a concern of the transit agency. The issue of accessibility
also involves the equity issue.
WAITING TIME
The waiting time is usually heavily weighted, partly because people usually perceive
it as a waste of time. At the same time, safety problem may arise during waiting,
especially for modes such as bus with a significant crime rate at the stop, which gives
more reason to assign heavier weight to waiting time in decision- making. Scheduling
and coordination may help improve the negative impact of waiting.
Also as the concept is widely adopted, in public transportation, the waiting time is
usually directly related to transit headway or service frequency. However, from the
perspective of the agencies, what is perceived is the increased cost of increasing service
frequency.
TOTAL TRAVEL TIME
The definition of each element allows some overlapping and should be case- sensitive.
Even with all the weighting factors being considered, the evaluation of total travel time in
the utility function may still follow non- linear pattern. As it appears in existing literature,
people’s budget ( money and time) per day is limited, and therefore, when the travel time
exists a certain level, there can be a sudden increase in value of time. The same issue applies
to monetary cost. However, the problem is never analyzed into depth from this perspective,
and should be a point of research interest in future.
3.3.2 Monetary Cost
The measurement of monetary cost is rather direct, and clearly additive. Complications
involved in the issue mainly lies in how the user perceives to the cost. The monetary cost
can be directly out of pocket, or less perceivable. At the same time, bundle between agencies
may affect the perception too.
Measuring Multimodal Transport Level of Service
13
OUT OF POCKET COST
Out of pocket cost includes taxes, ravel tickets, tolling fees, etc. It is the part of
cost incurs directly to the user during one’s travel. The cost is usually on a one- trip
basis.
INDIRECT COST
Indirect cost can be partly hidden, such as gasoline, car insurance, mileage cost,
maintenance etc. The cost is usually paid collectively, and not during the trip. As a
result, the traveler may become less sensitive to it, and perceive partially.
BUNDLE BETWEEN AGENCIES
Bundle between agencies usually occurs as discount. For example, a traveler
may get a free shuttle bus ticket when he/ she books from a certain airline.
Furthermore, bundle may imply coordination between agencies
From the user’s perspective, the perceived monetary cost is the one that matters and
appears in the utility function. However, from the supplier’s perspective, all the costs
collected are weighted equally enters the revenue..
3.3.3 Safety
Besides its impact on perceived time by travelers, the safety factor may have other
impacts on the user’s decision making before planning the travel. For example, while the
travel time is perceived during the trip, before deciding which alternative to take, the user
may evaluate the issue of safety basing on his/ her knowledge, probably information from
newspapers. As a result, safety has the property of “ the weakest link”, as when it comes to a
multimodal transportation alternative, no matter how the safety factors of the other modes
are, the safety factor of the entire trip is only determined by the worst link.
The evaluation of safety is perceived as the risk taken due to insecurity. However, the
risk has a wide range, from monetary cost to fatal risk, and therefore quite difficult to
translate onto the same scale. Besides, user’s perception to risk cannot be assumed to be
linear. So the evaluation of safety is a problem that is not easily quantified. We will discuss
about this issue later in the analytical framework.
3.3.4 Reliability
No transportation mode can be perfectly reliable. In a multimodal transportation
system, the arrival time of one link is the starting time ( which is decided by previous modes)
plus the travel time ( which is generally independent of previous modes). Both of the
elements would not be accurate, but usually follows some probabilistic distribution. The
reliability is the possibility that the arrival time falls into a certain confidence interval.
Existing work on the reliability issue usually look from the perspective of willingness to
pay, which studies the reviewed preference between reliability and other travel disutility
( usually time or monetary cost). The assessment of reliability can be achieved from historical
data of each mode within the alternative. Trying to achieve more insights of the issue, in this
report, we considered the reliability problem as the expected penalty from delay.
Measuring Multimodal Transport Level of Service
14
Unreliability causes extra waiting or delay, the later of which is more heavily weighted.
Reliability can be partly considered as a “ weakest link” factor as well. However, the arrival
time distribution of the last link requires intense analytical work. As when the risk of missing
the next link becomes part of our concern, the risk of delay performs rather non- linearly.
The scenario also depends on the degree of coordination between modes. The most
common example exists when the user need to transfer between flights. If the two
connected flights belong to the same airline, the connection can be considered as perfectly
coordinated ( comparing with other types of connection). Whatever accidental events occur
to the first flight, the airline company is responsible for arranging the connected flight. A
certain level of service is guaranteed. Whereas if the flights are from different companies,
there is no guarantee exist. If the first flight is delayed, the user is responsible for all the
monetary and time cost that rises from it. The difference can be also referred to as “ inside
the system” versus “ get out of the system”, where system refers to the condition when two
connected modes are coordinated. However, coordination means investment cost. In some
cases, coordination may not necessarily improve the situation. For example, to transfer
between the airtrain and the flight does not require coordination, especially when the airtrain
has a high service frequency. We will further demonstrate this quantitatively in the following
section.
3.3.5 Flexibility
In our definition of a multimodal transportation alternative, the illustrative structure is a
corridor. When it comes to a transportation network, the overlapping between alternatives
raises complication in numeration. The number of available alternatives may even exceed the
total number of links involved. The network structure may provide convenience, especially
in the case when one link in an alternative is shut down, the user may switch to another
alternative without sacrificing much of the utility. For example, to transfer from a flight to
the train, which means from an airport terminal to the air train station serving it, a user may
choose from several options, such as walking, taking a shuttle bus, or though a linking
bridge. Even though the modes proceeding and following this link are the same. Therefore
each of these choices indicates a transportation alternative, as we defined. From the point of
view of utility, one of the alternatives may be dominant ( shuttle bus), and the rest may be
considered as redundancy ( walking). But in case when shuttle bus is not in service ( due to
accidents or service hour limits), users planning to choose that alternative have the flexibility
of switching onto another one ( walking) without much delay. In other words, if there are not
walking routes or any other substitutes available, the user will need to go all the way back to
the directory to search for a taxi.
We define flexibility as an indicator of the level of service It describes the ability of a
network to adapt to changes ( either due to user’s behavioral change or the network setup
change). It is overlapping with the reliability issue, as they are both concerned with accident
possibility. To some extent, flexibility may contradict with other indicator. For example,
direct service provides better LOS in terms of time cost, but sacrificing flexibility. This is
also why when direct service and indirect service are usually both offered.
The concern of flexibility also accounts for the fact that some transportation modes,
especially public transportation, has limited service hour. And therefore, the flexibility of
using the mode, or the alternative, is measured by the service hour. Flexibility has similar
properties as accessibility, which is also an alternative specific “ weakest link” factor.
Measuring Multimodal Transport Level of Service
15
Also as it is mentioned before, in a multimodal network design, redundancy must be
included to provide flexibility. The redundancy should be provided such that, when one link
in an alternative is broken ( due to foreseen or unforeseen reasons), the user will not be
“ stuck” in the alternative. He/ she can switch onto another alternative and continue with
his/ her trip with none ( or little) sacrifice of utility. It is difficult to accommodate flexibility in
common utility- based analysis, because flexibility is not an independent indicator between
alternatives. A common method to include flexibility in the design without much
mathematical challenge would be adding redundancy to the network after the arteries are
designed. On the other hand, in terms of evaluation, the value of flexibility requires more
strategic and numerical effort, which could be a field of interest in future research.
3.3.6 Energy
Although energy consumption is a monetary cost concern to the user, the energy issue
under discussion here is from the supplier’s perspective, highlighting the concern of
externality. Therefore to evaluate a certain source of energy, we need to discuss four aspects:
the cost of generation, the renewability, the efficiency during production and its emission.
The result, however, is subject to technological constraint. So there may be innumerous
number of existing resources, our discussion always falls between electricity and gasoline
( sometimes diesel).
The environmental impact is largely a government evaluation concern. Individual
travelers may be aware of the issue, but it rarely affects their decision- making. With the
increasing urgency of environmental concern in the contemporary society, the issue of
environmental impact is taking on increasing weight in the government’s decision making.
Emission is the subsequent impact of fuel combustion, which includes non- pollutant such as
CO2, which causes green house effect, and pollutant such as NOx, SO2, VOC and etc, which
are detrimental to human health. Emission should be evaluated by unit volume/ pax, for
passenger movement, and unit volume/ unit weight for freight movement. Generally
speaking, it is directly electricity is related to fossil fuel combustion, and therefore, the fuel
type; and is also related to fuel efficiency; which is why electricity is usually considered as a
clean energy in terms of emission. Emission is usually a government concern, as the
commuters and transportation agencies cannot perceive it directly.
3.3.7 Investment and Agency Cost
The investment and agency cost is the major constraint to the supply side. In the case of
public transport service, the objective of the agency is to provide social welfare, and usually
cannot cover all its expense. Therefore the agencies rely on the government’s subsidy. The
government’s decision of giving its financial support is based on the measurement discussed
before. Because the focus of this report is not on the business strategy of private companies,
in the framework we are introducing, the government is responsible of all the expenditure
that cannot be covered by the revenue.
3.3.8 Equity
The issue of equity is relatively complicated; as it involves factors such as travel cost,
accessibility, and emissions, under a specific sociodemographic distribution. At the same
time, due to the concern of equity, the transportation agency may offer lower fares to the
low- income group, and may be obligated to provide adequate accessibility to each area.
Measuring Multimodal Transport Level of Service
16
Therefore, equity is an issue that experienced by travelers in a non- uniform fashion, occurs
largely under government regulation, and the relative consideration is implemented by
transportation agency. Meanwhile, when it comes to investment, the resultant scenario of the
improved service should also be evaluated from the perspective of equity. The evaluation
can be achieved from a small- scale implementation.
The equity issue can be quantified from the indirect utility model. If we can obtain
sufficient information about the social make- up of the community and the respective
indirect utility perceived by each group, we are able to observe the variance of utility in the
community, so as to obtain the evaluation of equity. For example, equity can be described by
the standard deviation of indirect utility of a community.
3.4 Angle of Perception
USER’S PERSPECTIVE
Users’ perception of a composite of travel modes is usually captured by the utility
function. As analyzed in the previous session, standing at the users’ perspective, the factors
perceptive includes travel cost, safety, reliability, travel flexibility and etc.
Concerning multimodal alternative , including modes
Indicator ( data input) Elements
Access
Waiting
In Vehicle Traveling
Time( s)
Transfer
Out Of Pocket
Indirect
Disaggregate Factors
Money($)
Bundle
Safety Risk of Insecurity
Reliability Risk of Delay and Waiting Aggregate Factors
Flexibility Feasible Duration of Taking
the Mode
Figure 3- 2: Users’ Perception
The resultant demand will be following market response, such that the demand will
increase with the level of service.
SUPPLIER’S PERSPECTIVE
There are two groups of suppliers: the government – the policy maker, and the agencies
– the actual implementer.
Measuring Multimodal Transport Level of Service
17
GOVERNMENT’S PERSPECTIVE
From the perspective of the government, the issues that are of her concern will be
LINK j
Indicator ( data
input)
Property
Equity Variance of Utility
Efficiency
Energy
Renewability
Emission Weighted Volume
Monetary Cost
Level of Service
Figure 3- 3: Government’s Perception
The issues under the government’s concern can be considered as independent, and their
relative weight in the final decision largely depends on the current need and the policy
maker’s final call. Therefore, these weighting factors are out of the scope of the discussion in
this report.
AGENCY’S PERSPECTIVE
From the perspective of an agency providing the mode i, its profit is affected by
Property Assessment
Agency Cost 􀀁 Agency cost Level of Service
Revenue
Subsidy
Figure 3- 4: Agency’s Perception
Therefore, during the operation, it should be at least guaranteed that the agency should
be able to maintain the balance of its income and expense. The government is responsible to
provide the subsidy covering the gap between revenue and agency cost. So from our
conceptual framework, the agencies only act as a budget constraint to the government’s
policy making. To simplify our analysis, in the following analysis, the supplier’s perspective is
represented by the government’s perspective.
In the following two sections, we would introduce the analytical framework we
developed, from the user’s perspective and supplier’s perspective respectively. The analytical
framework will be a numerical representation of the conceptual framework we are discussing
over here, and based on a simplified network.
Measuring Multimodal Transport Level of Service
18
4 Analytical Framework - Users’ Perspective
In this section we construct a user indirect utility function for the intermodal network
defined in Fig. 3.1, and shown below. The simplification to two modes should not result in
any loss of generality. As is commonly done in transportation utility analysis, we consider the
perceptions of level of service attributes to be deterministic when measures at a dis- aggregate
individual level. The aggregation to a group of users introduces stochastic effects resulting in
a variance of perceptions across the relevant population. We build the elements of the utility
function for each attribute ( time, money, etc.) and then these are combined to form the total
function for a multimodal alternative. The structure of the indirect utility function is
assumed to be basically a linear function of the attributes, with the parameter of each
attributes reflecting its relative influence on the utility function. As such, some attributes may
be allowed to share parameters, as is shown below.
4.1 Time Cost: Perceived Travel Time
For the transportation alternative defined as above, the travel time elements and their
weighting factors are defined as:
Figure 4- 1: Time
Therefore, the indirect utility function for travel time ( factorized travel time) is
Where is a negative, monotonically decreasing function of , ,
The non- factorized total travel time is
Mode 1 Mode 2
Attribute Access
Time 1
Waiting
Time 1
IVTT
1
Transfer
Time
Waiting
Time 2
IVTT 2 Access to
Destination
Denotation
Time
Parameter
Mode1 Mode2
eArc cTeismse/ Transf Waiting Time ITTnirm aVveee hliicnlge
Measuring Multimodal Transport Level of Service
19
For simplification, denote , , , , and the
parameter is named accordingly.
It is noted that with different compositions of T, may increase with increasing . The
condition of interest would be:
or
, such that
An interesting research question would be to explore situations under which this
condition obtains, which means situations when the trade- off between transfer and line haul
times is such that an individual would prefer a longer total travel time to avoid a certain
amount of transfer time. This is illustrated by the following example.
EXAMPLE:
We consider a case when and Also . We define:
If changes by , changes by
Start with , if ,
A reasonable assumption is that , i. e. .
Therefore, when ,
Otherwise,
For example, as mostly appears in the existing literature, the waiting time has twice the
weight of IVTT, when , . So when the waiting time is
reduced by more than half the increase in vehicle travel time, the passengers’ perceived
utility may be increasing as the total travel time increases.
Measuring Multimodal Transport Level of Service
20
The resultant increase in total travel time starts from 0, may reach up to . Also if
we consider the uncertainty in travel time, as the weight of waiting time increases, the
schedule may be more relaxed without sacrificing the overall utility CASE STUDY: US ENVIRONMENTAL PROTECTION AGENCY ( 2000)
The work reviewed 50 US studies and concluded that the perceived time of walking
time is 2.0 to 2.72 times that of in- vehicle time, i. e. , the possible increase in
utility with respect to increase in total travel time and different values of can be plotted as
Figure 4- 2: Utility VS. Total Travel Time
Measuring Multimodal Transport Level of Service
21
4.2 Monetary Cost
For the simplest case, assume all the cost is 100% perceived by the user. And therefore,
there is usually no perception difference in monetary cost for the same customer, so there is
only one parameter, which can be applied to all the monetary costs.
Figure 4- 3: Monetary Cost
Also in the usual case, there is no monetary cost for accessing/ waiting/ transfer, so
The total monetary cost can be calculated as
There are, still, several complicating issues involved in the evaluation in monetary cost, the
two major ones are:
a. Bundling: to achieve optimal social welfare, agencies sometimes bundle, or coordinate
between transportation services provided by different suppliers. This results in
discounts of fare during transfer.
b. Discount for a specific groups: In situations when pricing schemes are not uniform and
depend in part on some socioeconomic characteristics of the users, such as senior or
student fares, the assessment of cost will have to be differentiated on the bases of these
characteristics.
4.3 Reliability
Travel time is should be considered to be stochastic due to the uncertainties
surrounding it. While considering stochastic effects, early arrivals may not be a major
concern, as the earliness usually results in shifting of waiting time rather than the
reduction of the total travel time. But delayed arrival is a major issue, and also leads to
the discussion of travel time reliability. In this session, we discuss reliability mainly in
terms of the risk of delay. Travel time variability is most commonly represented by a
lognormal lateness distribution ( e. g. Rietveld et al. 2001, Giuliano, 1989). Generally, the
distribution of time ( starting time, line- haul time, waiting time, transfer time, etc.) can be
characterized by a distribution that is described by two major parameters: : the
probability of delay; and : the variance of delay; The cumulative distribution may be
described by:
Mode 1 Mode 2
Link Access
Time 1
Waiting
Time 1
IVTT 1 Transfer
Time
Waiting
Time 2
IVTT
2
Access to
Destination
Measure of
Performance
Monetary
Cost
Parameter
Measuring Multimodal Transport Level of Service
22
Mode 1 Mode 2
Link Access
Time
1
Waiting
Time 1
IVTT 1 Transfe
r Time
Waiting
Time 2
IVTT 2 Access to
Destination
Measure of
Performance
Reliability
Parameter
Figure 4- 4: Reliability
In practice, perfect reliability of access time, waiting time and transfer time can be
assumed, as it is not within the planner’s control. Therefore,
, and the respective reliability factor drop out of our
discussion.
Mode 1 Mode 2
Reliability Distribution Start Time
Travel Time
Ending Time:
Start Time
Travel Time
Ending Time:
Figure 4- 5: Reliability Statistics
Denote the lateness penalty per unit time as L, usually , as L has significantly
larger weight in negative utility. The problem of reliability can be accessed by the expectation
of delay, denoted as . If the passenger misses his/ her desired run in mode 2, he/ she
needs to wait for the next one, causing a delay of .
Mode1 Mode2
eArc cTeismse/ Transf Waiting Time ITTnirm aVveee hliicnlge
Measuring Multimodal Transport Level of Service
23
EXAMPLE: UNIFORM DISTRIBUTION
As a demonstration of the impact of different parameters, we start our analysis by
assuming that the delay is uniformly distributed between 0 and
Where can be considered as a description of , and the distribution applies to
.
The expectation of delay is
Upon calculation, and assuming the line- haul traveling time and starting time are
independent, it can be achieved that the delay of follows
The expectation of ending time delay is
The reliability issue of mode 1 may affect not the duration of the trip ( may even shorten
it), but it will affect passengers’ possibility of missing the following mode. From experiment,
researchers should be able to find the transfer time required by the passengers, which can be
assumed as a deterministic value A, .
Based on the previous assumptions, the distribution of ET is relatively complicated, but
in order to simplify the analysis, we assume the distribution of delay in ET to be uniform.
The assumption is reasonable in the sense that, 1) the objective here is to demonstrate some
variation characteristics; 2) we start up with series of assumptions in the first place; and 3)
even if we follow the previous assumptions, the resultant distribution is discontinuous,
which is even less reasonable than uniform distribution. The distribution of ending time
delay follows
UNCOORDINATED CASE
In the case that mode 1 and mode 2 are not coordinated, i. e. ET1 and ST2 are
independent, and the scenario can be discussed as following:
In Vehicle Delay
Given the schedule of the multimodal transportation network, the expected penalty
due to travel time uncertainty is determined by the weakest link i, such that is the
maximum for all i. ( The weakest link is in terms of ET.) Denote the weakest link as ,
as calculated before, the expected delay penalty is
Measuring Multimodal Transport Level of Service
24
Transfer Delay
Usually there is more variance in ending time delay, so also assume .
The expected penalty of missing the next run of mode 2 is
……………………….………...( 2)
The total expected delay penalty is , where
COORDINATED CASE
In the case of coordinated modes when the first mode is delayed, the second will be
delayed accordingly, in order to ensure passenger connections. Therefore, there is neither
the risk of missing the next run, nor the probability of catching up with the schedule
during the transfer. The expected delay is then give by
And the delay penalty is
Comparing the result of coordinated and uncoordinated cases, it can be observed that:
If there is: 1) an clear “ weakest link”, i. e. the expected delay from which is significantly
Measuring Multimodal Transport Level of Service
25
higher than the others, or 2) one of the links has high missing penalty/ large headway,
( such as in air travel), then there is an advantage for the transportation agencies to be
coordinated ( without considering the relative difficulty of coordination). The result is
intuitive since it is rational to take the mode of high variance at the end of the traveling
process.
4.4 Mode- specific Parameter
Besides all the alternative- specific characteristics discussed above, we may also introduce
another set of characteristics of each mode that may not be easily quantified, but still affects
the decision making process, such as safety, comfort, etc. All these factors may be essential
to the level of utility, but too subjective to be measured on an objective scale.
D Potoglou et. al.( 2010) studied the issue of safety as a tradeoff with privacy and liberty.
The work assessed the people’s preference across privacy, liberty and security, in the case of
rail in UK, through discrete choice modeling. The results indicate that generally, respondents
show higher willingness to pay for security improvements, and thus the issues of liberty and
privacy are outweighed by their preferences for security. Meanwhile, they did identify
segments in- the- sample that are against measures that sacrificing liberty and privacy, for
example, the presence of uniformed military at rail stations.
The ranking of safety may not be useful from the system’s perspective, as the safety of a
system is not an issue that can be compromised. The level of safety is always at the best
possible level. However, from the users’ perspective, if we try to quantify the utility of safety
among modes, one of the most obvious approaches is to rank the safety from users’
perception, or from historical data.
However, the impacts of these factors cannot be ignored. Existing study shows that with
all the mode- specific factors being equal, users still shows preference of one mode over
another ( train over plane). The result may be due to the impression of unsafe, or merely due
to visual pleasure, which cannot be decided for certain; on the other hand, the situation may
also change with time, for example, people’s reluctance of choosing air traveling may be
decreasing with its improvement of performance.
The impact of these factors will be described by a mode- specific factor M, therefore, in
our case study, the mode specific factors will be
Mode 1 Mode 2
Link Access
Time 1
Waiting
Time 1
IVTT 1 Transfer
Time
Waiting
Time 2
IVTT 2 Access to
Destination
Mode
Factor
Measure of
Performance
0 0 M1 0 0 M2 0
Figure 4- 6: Mode Factor
Mi will be a weakest link factor, such that for the entire travel alternative, the resultant utility
impact will be
€
Vm = MIN( M1, M2)
Measuring Multimodal Transport Level of Service
26
4.5 Sociodemographic Variables
Other than the alternative- specific variables, the other group of variables is
sociodemographic variables. The sociodemographic characteristics of an individual
determine his/ her perception of the alternative specific attributes. Sociodemographic
variables may cover the impact of different travel purpose ( business, leisure, etc.), different
income groups, age groups, and etc. Sociodemographic variables may be additive to the
utility function, or can be considered as a variable that affects the evaluation of all the
variables. For example, the income factor may affect the weight of monetary cost in ones
decision, whereas the value of time may follow another pattern while varying with income.
This feature also follows the survey result that the weighting of each factor in the utility
function follows a certain distribution. While considering the impact of sociodemographic
variables, what the transportation agencies may do is to design the survey such that the
characteristics of each group can be captured. Therefore, the resultant utility function can be
following certain sociodemographic distribution. For the community that is concerned, the
utility distribution is
€
P( U = Ui) = f ( Ui )
Therefore the expected value and variance of the distribution may be also observed for
further use, such as equity, which will be discussed in next section.
4.6 Demand Modeling
The indirect utility ( for a certain individual of a certain series of characteristics) is the
summation of the utility of all the aspects mentioned before.
€
V = Vt + Vc + Vd + VM
The assumption of indirect utility function follows the Logit model in mode choice
modeling. Therefore, from the utility function calculated above, assuming there are other
traveling alternatives with similar form of utility function, the probability of passenger
choosing one alternative would be
€
Pi = eVi
eV j
j Σ
The total travel demand can be assumed to be a function of the expected indirect utility
of an alternative as given by the log sum of indirect utilities across the whole choice set,
such that
€
Q = fn log eV j
j Σ


 


 








For simplicity, we may also assume the function to be linear, so
€
Q = a + b⋅ log eV j
j Σ


 


 
Therefore, assuming all the revenue collected comes from the fee, the revenue will be
€
πi = Q⋅ Pi ⋅ Ci
Measuring Multimodal Transport Level of Service
27
4.7 Summary
Summarizing the previous analysis we obtain the following table of combined measures of
performance as perceived by the user. The sum of the indirect utilities from the various
components enters into the demand function, for example as a logsum in the case of a
choice model formulation.
Mode 1 Mode 2
Feature Access
Time 1
Waiting
Time 1
IVTT 1 Transfer
Time
Waiting
Time 2
IVTT 2 Access to
Destination
Denotation
Parameter
Time
Composited
Measure of
Performance
Parameter
Monetary
Cost
Composited
Measure of
Performance
Parameter
Uncoordinate
d
( Uniform Distribution)
Reliability
Coordinated
( Uniform Distribution)
Measure of
Performance
Mode 0 0 M1 0 0 M2 0
Factor
Composited
€
Vm = MIN( M1, M2)
Figure 4- 7: Demand Summary
The total indirect utility is given by
€
V = Vt + Vc + Vd + VM . . Although this is quite simplified
example, it can still demonstrate the framework of multimodal transportation evaluation, and
the challenges that are involved in this process. From the utility function we can observe
that:
1) The evaluation of multimodal transportation involves significantly more than the
summation of attributes for the different modes involved.
2) The evaluation of different categories of travel time can be the most complicated
aspect of multimodal evaluation. This is because times combine in many different
ways and sometimes in ways that are counter- intuitive. Therefore a thorough
Measuring Multimodal Transport Level of Service
28
understanding of the trade- offs people make between different elements of travel
time need special attention in travel surveys and behavioral studies.
3) Reliability in multimodal transportation can be an even more serious issue than in the
case of a single modal. Multimodal reliability is always less or equal to that of the
least reliable mode in a multimodal system.
4) Coordination between modes can change the scenario, however, not necessarily in a
positive way from the user’s perspective;
5) We believe that some of the attributes, such as safety, are not necessarily examined
separately from the users’ perspective, as their utility is very subjective. Therefore, we
recommend treating all the other impact factors as a combined, mode specific factor;
6) In our framework, sociodemographic variables are important determinants of
perception of attributes and result in the variance of the indirect utility function.
They should be treated at as dis- aggregate a level as possible.
Also from the analysis, we are able to identify some of the issues may be of interest in
future studies:
• Tradeoff between waiting time and in- vehicle traveling time:
As illustrated in the case study, the change of proportion between different travel-time
elements can be utilized in planning to improve the level of service. One advantage
of it is the planning process does not involve much investment on facility. As a result,
this issue may involve relatively short studying process and promising outcome.
• Reliability issue in multimodal transportation:
Although we modeled the issue of reliability in the previous case study, the model is
obviously an oversimplification. However, we can still observe that, the reliability issue in
multimodal transportation is quite different from the single- modal case, and involves
more challenges. It is a weakest- link type of issue, but one that is made more difficult by
the complexity with which users perceive the cumulative reliability impacts of a number
of modes integrated into one system. This is a subject that is ripe for further research.
• Threshold Value in time attribute:
Our treatment of travel time in this analysis has assumed the monotonic, continuous
impact of time on utility. In reality is often arises that the impact of time is discontinuous
at some threshold values. These are values that have to do with time of day, possibility of
round trip in one day, and other impacts that arise in transportation. Despite the rich
literature on time value and time budgets, little has been done on how the discontinuities
in travel time valuation can be combined and used in the indirect utility functions of a
multimodal transportation alternative. This topic is therefore a promising one for further
research into measures of performance.
Measuring Multimodal Transport Level of Service
29
5 Analytical Framework - Suppliers’ Perspective
The modeling of the government’s decision- making process is relatively simpler as the
evaluation seems more straightforward, as the complications between modes are not as
significant as from the travelers’ perspective. However, it can be more challenging as 1) there
are no universal weighting factor for the measures of performance involved, neither there
are well- justified judging criteria for evaluating the evaluation criteria; which makes the
decision- making seems more arbitrary than it is from the users’ perspective; 2) the actual
scenario is difficult to model analytically, because especially when private companies take
into play, the process will involve business strategy, which falls out of the scope of our
discussion. Therefore, we try to make this discussion more from a government and public
agency’s point of view, where profit is not the dominant concern.
As analyzed in the previous sessions, there are five major issues that may affect the
government’s evaluation of a transportation system: equity, energy, and externality, level of
service and cost.
5.1 Equity Issue
The issue of equity usually cannot be evaluated directly, but may require a dedicated
studying process. In evaluating the equity issue of a multimodal transportation alternative,
there can be two major challenges
QUANTIFYING THE VALUE OF EQUALITY
As illustrates in the utility function, there are a series of sociodemographic
variables that represents the complexity of the construction of the society, each
group of which has its own utility value. As a result, the equity issue can be observed
from the difference of utility of different user groups.
The variance of utility among all the users ( and non- users, if can be quantified) can
be used as an indicator of the equity issue. Such that
€
Sequity = VAR( Vi )
Meanwhile, there is another issue that attracts our attention: the disutility of non-users.
The emission problem, for example, is a serious concern that affects the social
welfare of non- users, but these impacts are difficult to quantify.
MULTIMODALITY:
As the scope of this report is multimodal transportation, the resultant equity can
be quite different from what has been discussed alone for each mode. There is
possibility that the composite transportation alternative has an equity problem that is
less serious than each of the modes. The problem of equity of each mode may add
up as well as cancel out when they are composed, depending on specific situation,
which also shows the necessity of evaluating a multimodal transportation alternative
as a different issue than the direct summation of each single mode.
5.2 Energy Consumption:
The issue of energy conservation, as discussed before, is more of the government’s
concern. Due to the differences in production process, the renewability, the cost of one type
of energy can be, therefore, weighted. As the monetary cost will not be the only issue that
Measuring Multimodal Transport Level of Service
30
the government is concerned with. The monetary cost of energy resources will be included
in the agency cost. Therefore the indicator here only accounts for the extra weight of energy
due to the type.
€
SE = β i
Ei
i Σ
where
€
β i
is the weighting factor of energy by type 5.3 Externality/ Emission
In our case, the concern of negative externality lies in pollution, or more specifically, the
emission of vehicles, and therefore, usually monetized. The monetization of emission can be
quite complicated, as its impact varies rapidly with the pollutant situation of the
circumstances. When the environment is already polluted, the marginal cost of emitting
increases sharply, as the pollutant can be considered as more detrimental. In the work of
Delucchi ( 2000), the cost of emission is studied as the cost of eliminating the amount of
pollutant, and he simplified the environmental circumstances as two scenarios: highly
polluted and lower polluted, each of which can be a reference of the government’s interest.
The externality can be indicated as the weighted summation of all types of pollutant:
€
SP = ε iPi
i Σ
where
€
ε i is the weighting factor of pollutants by type.
5.4 Level of Service:
The level of service of the system can be calculated as the log- sum of utility value of all
different transportation alternatives that are involved in the transportation system we are
concerned with. The utility value is what we have been calculating from the users’
perspective.
€
LOS = log eV j
j Σ


 


 
5.5 Monetary Cost: investment cost, subsidy
The government may be responsible for the investment of a transportation system, or part
of it. Furthermore, the government may need to keep the balances of transportation
agencies’ operational budget. In most of the cases, for public transportation agencies, with
the objective of which being providing social welfare, and not profit driven, the government
will need to provide subsidy to cover its cost. The subsidy needed can be calculated as the
difference between the total agency cost and the revenue, which is usually collected from the
users. The government may need to cover all the extra expenses:
€
Ss = ( Ii − π i)
i Σ
The actual scenario will be quite case- specific. For example, when transportation agencies
shares mutual benefit, but these are usually the situation when profit- driven companies
operate the mode, which can be beyond the scope of our discussion.
Measuring Multimodal Transport Level of Service
31
There are no universal applied weighting factors among these issues. The factors adopted
are largely based on government’s decision, depending on the specific situation. However, if
all the previous mentioned information collected, the government will be able to construct
its own optimization process, based on her goals. The relative weight between each criterion
can be adjusted.
5.6 Summary
To sum up, from the government’s perspective, the issues of concern can be evaluated as:
Measure of Performance Evaluation
Equity
€
Sequity = VAR( Vi )
Energy
€
SE = β i
Ei
i Σ
Externality
€
SP = ε iPi
i Σ
Level of Performance
€
LOS = log eV j
j Σ


 


 
Monetary Cost
€
Ss = ( Ii − π i)
i Σ
Figure 5- 1: Supply Summary
It can be observed that:
1) The evaluation result can be significantly different from the summation of each
single mode, especially for the issue of equity;
2) Cooperation among agencies can be essential from the government’s perspective
( although it is not necessarily the case from the users’ perspective, as discussed
before);
3) An optimization framework can be constructed, where the government’s “ prioritize”
decision is the final link of the loop;
4) The measures of performance can be also utilized as monitoring criteria, in case of
one of the issue becomes significantly problematic.
Issues that may be of interest in future research are:
• The disutility problem of non- users when a transportation system is constructed,
which enables us to treat the issue of equity more rationally. Incorporating non-users’
in our evaluation process can be a challenging topic itself.
Measuring Multimodal Transport Level of Service
32
6 On the Issue of “ The last mile”
The perceived cost of transfer time is typically found to be much higher than in- link
( linehaul) time. In some cases ( e. g. PBQD Inc. 1993), the penalty of transfer is found to be
so significant that it can be weighted up to the scale of hundreds of in- vehicle traveling
time, which makes the transfer critical in the overall level of service. However, the
traditional metrics of measure of performance, which usually emphasize on travel cost, are
by all means oversimplified. Besides the readily quantifiable measure of performance, the
negative utility of transfer is resultant from other elements that contributes to the
perception of the existing measure of performance. For example, when people are
transferring from a rail station to an airport, their preference of people mover or moving
walk to walking is obvious. The preference, or the relative indirect utility, cannot be
quantified by travel cost alone. Also because transfer is always a small potion in terms of
measurable cost, its significance cannot be described by the traditional metrics. Another
example would be that user sometimes rule out the option once out- of- system activity
( crossing the street to transfer between rail lines) is involved due to the risk of confusion.
In such cases, the problem of perception of travel cost does not even have the chance to
arise, as the decision is made before any other measure of performance takes place.
Therefore, people’s perception of intermodal activity attributes is different from that of
online activities. We analyzed the rationale behind the traveler behavior, especially the
reluctance to transfer. However, to analyze and forecast the marginal benefit while
designing the interface between modes, it is critical to understand user’s behavior. In
existing literature, the studies on user’s reluctance to transfer are empirical, and the impact
of transfer in utility function is consolidated into one dummy variable. Due to lack of
analytical understanding, the dummy variable can only be quantified with large variance,
which adds to numerical complication. We refined the study of transfer by looking into the
analytical background of user’s behavior, and to develop a conceptual framework. Although
abundant empirical research is required to realize such framework, at this point, it will help
investigating into the problem of “ the last mile”, so as to identify the critical factors in
investment decision- making.
A categorical passenger service standard of rail access to airport ( Kivett and Parsons,
2004) includes:
1. Number of mode changes en route;
2. Proximity of rail station to primary airport functions;
3. Walking distances, level changes and travel times;
4. Fare collection/ control systems;
5. Baggage check- in and retrieval systems;
6. Signing, graphics and flight information;
7. Marketing, PR and IT systems;
8. Aesthetics
Of the eight elements, number 2, 3 and 4 are relatively objective standards, and widely
adopted in existing metrics. No. 2 and No. 3 are overlapping, as distance is perceived by
travel time, while No. 5 and 8 are out of the scope of this discussion. We investigated into
more insights of the other elements, by constructing a qualitative analysis of the demand-supply
mechanism of the last mile.
Measuring Multimodal Transport Level of Service
33
Meanwhile, there is also the supplier’s perspective on this issue. First of all, we need to
understand “ what the suppliers can do”. In order to provide better connectivity between
modes, the supplier ( or suppliers) can design the system by offering
• Scheduling coordination
• Physical Integration
6.1 Scheduling coordination
This is what we have been talking about by “ coordination” in previous sections. The
coordination in scheduling requires long- term cooperation and agreement between modes,
and hence more easily implemented when the same company owns the both. For instance,
travelers always prefer to book connected flights from the same airline; so that if the first
flight is delayed, the airline is responsible of providing an alternative connection, and the
traveler will not take the risk of misconnecting. We can consider the two coordinated
modes a “ system”; and in this case, once the traveler checks in, he/ she is considered “ inside
the system”, such that a certain through level of service is guaranteed at least as far as reliability
is concerned. Most of the elements of scheduling coordination are discussed in the previous
sections, and its related user’s behavior is usually quantifiable by the conventional units –
time and monetary cost, as well as probability and utility theory. As demonstrated in the
previous case study, scheduling coordination is not enough to mitigate the disutility in
transfer. There remains the element of physical integration at the intermodal connection.
EXAMPLE
In this example we adopt the result of Kanafani and Ghobrial ( 1985) who used a
dummy variable to differentiate between different levels of intermodal connectivity. The
utility of intermodal connection is reflected in their indirect utility function of route choice
as follows:
€
V( r, j) = α1Trj + α2Frj + α3 frj + α4Drj + α5Crj
where,
T: travel time
F: daily frequency of service
f: one- way economy class airfare
D: dummy variable for aircraft size
C: dummy variable for connectivity service ( 0 for non- stop service, 0.5 for online
service with stops, 1.0 for indirect connecting service)
r: denotes the route, j as the city pair.
€
α : is a coefficient.
In the context of multimodal transportation, and by our definition online service
represents perfect scheduling coordination, while indirect connecting represents no
coordination ( but with physical integration, which will be discussed later). The coefficients
were calibrated from data of 19 city pairs:
Measuring Multimodal Transport Level of Service
34
Attribute Estimated
Coefficient T- value
Time ( hour) - 0.897 - 14.77
Daily Frequency 0.239 38.21
Fare ( dollar) - 0.278 - 6.7
Aircraft Type 1.461 15.4
Connectivity
Pattern - 1.557 14
Chi- Square Value 3114
Log- Likelihood Value - 34.617
Figure 6- 1: Calibrated Coefficients
Although the data are rather old, we use this result only as an example of how to reflect
passengers’ response. It can be observed that the connectivity pattern is critical in terms of
weight. We will use some simple algebraic example here to illustrate its impact.
Imagine two transportation alternatives each made up of two connecting flights, one
alternative with an online connection ( alternative 1), and the other with indirect connecting
service ( alternative 2). The indirect utilities of the connectivity pattern are respectively:
€
Vc1 = α5C1 = 0.5α5
€
Vc2 = α5C2 = α5
€
α5 = − 1.557
With all the other factors being equal, the ratio of market share of online connection to
indirect connection is
€
eVc1
eVc2 = e0.5α5
eα5 = 2.178
TIME VS TRANSFER
For a utility maximizing passenger, the online service can take higher transfer time and
still have a market share advantage. The advantage remains if
€
α1T1 + α5C1 ≥ α1T2 + α5C2
α1( T1 − T2) ≥ 0.5α5
T1 ≤ 0.5α5
α1
+ T2 = 0.868 + T2
Therefore, the online connection flight can have 0.868 hrs extra transfer time to reach equal
market share with indirect connection. If alternative 2 has transfer time of 1 hr, that of
alternative 2 may reach up to 1.868 hrs.
6.2 Physical integration
This refers to the degree of geographic, instrumental and structural involvement between
stations, and may as well lies in the architectural design. Physical integration can also
provide the sense of “ inside the system”, however, not on an easily quantifiable base. For
Measuring Multimodal Transport Level of Service
35
example, user walking on a linking bridge with direction sign may consider him/ herself
inside the system. But the conventional metrics cannot capture the improvement on the
level of service; as the difference in terms of travel cost is limited, neither does it imply
long- term cooperation relationship between the modes. Even within the same terminal,
there can be multiple types of physical integration. Here we use the case of JFK ( John F.
Kennedy International Airport New York) to illustrate different levels of physical
integration.
Figure 6- 2: JFK Plan View
TRANSFER INSIDE TERMINAL
TRANSFER SEMi- INSIDE TERMINAL
TRANSFER OUTSIDE TERMINAL
Measuring Multimodal Transport Level of Service
36
Figure 6- 3: Transfer Choices
The impact of physical integration on user’s behavior is rather abstract, and difficult to
evaluate. As is illustrated in the previous example, part of the impact can be psychological,
but the understanding of user’s reaction to it can be essential, because it reflects the
mechanism of multimodal level of service integration, and user’s reluctance to transfer. On
the other hand, better demand forecasting will better justify the investment cost, and help
supplier making planning decision. Physical integration can incur different level of
investment cost, but less operating cost comparing with scheduling coordination. As a
result, it is a challenging and rewarding task to estimate the marginal benefit with improving
physical integration. Due to all the reasons above, the primary goal of this session is to
identify the elements in user’s reaction, and mapping them to different levels of physical
integration.
Scheduling coordination and physical integration cannot be completely separated. The
two methods are always utilized in accordance, as it exists in most transportation facilities.
But because in this report we want to emphasize on marginal benefit, we analyzed the
impact by elements. It is still important to keep in mind that, as the transportation
alternative is an integrated process, no attribute can be completely separated with another.
There are different types and levels of physical integration. In this report, the utility
resultant from physical integration are classified into two categories: 1) structural utility: the
basic level of utility due to the sheltering effect of physical integration, which can be
considered as constant with respect to level of integration; 2) facility utility: the utility due to
the facilities involved in the transfer ( people mover, moving walks, shuttle bus, etc).
Therefore, the facility utility is variable with change of integration methods. The overall
utility of physical integration is the sum of the two.
Choice 1: Transfer inside the
building, with dedicated linking
structure
Choice 2: Transfer outside the
building, exit the system and then
reenter it
Measuring Multimodal Transport Level of Service
37
6.2.1 Structural Utility
Studies on transfer environment tend to focus on the either pedestrian environment in
open space and on the impact of facility assistance, such as the presence of escalator. The
impact of the presence of a dedicated link alone is not well analyzed. For example, while
users are choosing a route of transfer between stations, it makes a difference whether one
needs to enter the open space and cross the street and access the following terminal, or
whether there is a dedicated linking structure for transfer. We refer to this part of utility as
“ structural utility”. Structural utility is critical in terms of passenger utility. The impact of
structural utility on user perception is subtle and includes elements such as the sense of
security, orientation, and the perception that one is being cared for inside the system.
IMPROVED LEVEL OF SECURITY
Physical integration offers user the coverage from the elements and gives a sense
of safety and security. But as discussed earlier, the evaluation of level of security is
too complicated to be realized analytically. But we still have reason to believe that
physical integration offers a measurable attribute of level of service in an intermodal
environment.
REDUCED RISK OF CONFUSION
The transfer terminal can serve as guidance to the user, and eliminate much of the
uncertainty and confusion during transfer. This effect may be different on first time
traveler and regular travelers. But it, too should have a measurable impact on the
utility as perceived by passengers in an intermodal system. Although not commonly
done, attributes such as physical integration shouldn’t be neglected in user’s mode
choice modeling.
PERCEIVED INTEGRATION OF SERVICE
Physical integration conveys the additional, positively perceived attribute, of
integrated services. The sense of being inside the system detracts from the disutility
of a connection activity. Coordinated schedules and “ online” transfers often go hand
in hand with physical integration since they are often offered by the same provider.
6.2.2 Information Assistance
The definition of information assistance is relatively vague. It starts from the basic
orientation function and range up to all the information that helps improve the level of
service. Form of information assistance includes graphics, signing, IT systems, on- line
information availability, and etc. For example, when a traveler is entering an airport from a
train station, he/ she needs direction sign to find the terminal building location, which also
provides the feeling that he/ she is inside the system. It will further soothe the traveler’s
anxiety if there is flight information available inside the connector. On- line information
availability about the train station, the airport, the scheduling, and the bundling scheme ( if
bundle ticket is available) will further improve the level of service. In our definition, not all
of the information availability mentioned above should be necessarily integrated as part of
structural utility. Part of the assistance is also counted as scheduling coordination, on- line
information about transfer, for instance. But it should be noted that structural utility
couldn’t be guaranteed without adequate information assistance, such that certain level of
Measuring Multimodal Transport Level of Service
38
information assistance is an inseparable part of physical integration. For instance, as
discussed in the previous session, we have presumed that the linking structure includes signs
and graphics that provide the orientation to the terminal building.
To clarify the concept, here we define information other than direction sign is
considered as facility utility, because fundamental level of information is indispensible. Of
course, the definition of “ fundamental” is still abstract, and scenario- specific. The concept
is, as long as physical integration is concerned, it is the supplier’s responsibility to provide
orientation, cleaning, security and etc, to maintain the basic level of service.
6.2.3 Facility Utility
The facility utility depends on the level of equipment and technological support, and
thus may vary rapidly with the design and also the investment. The utility is in forms of
comfort, as the user perceives it. Facility that is concerned commonly includes escalator,
elevator, people mover, moving walks, shuttle bus, and etc. The evaluation of facility can be
categorical. From experience, the facilities are usually offered in a combined fashion. The
facility utility also differentiates across modes. For example, the transfer between buses
usually doesn’t involve any of the pre- mentioned equipment ( though there can be
technological support during boarding and unboarding, it is not considered as part of the
transfer process), and the impact of facility support during the transfer is not easily
perceived; whereas, in the case of entrance to the airport, due to the burden of baggage and
anxiety caused by the schedule, users are more sensitive to facility utility. The task of
accounting for facility utility is challenging, but there are abundant empirical resources on
this issue. But few generalized results across all modes can be achieved, as the utility
perception varies with transfer scenario ( baggage, first- time traveler, etc).
In addition, bundling also implies the requirement of facility. Because the bundling
increases the complexity of service, without adequate guidance, it may cause user’s
confusion. However, the complexity and confusion can be meditated by information
assistance, as discussed previously.
6.3 The last Mile: Integration
In this section, we identified the importance of the “ last mile”, the concept of which is
consistent with that of “ the weakest link”, as the last mile of mode interface is always a
critical issue that requires integrated planning and service. The challenges in the last mile
issue not only lie in that it requires integrated design, but are also due to the fact that its
impact on user’s perception is subtle, and to a large extent, psychological. So we developed
a comprehensive framework that provides a conceptual illustration of the decision- making
process, from the supplier and the user perspectives. The user’s response to the issue is in
terms of sense of security, being inside the system, orientation etc; the supplier can resolve
these issues by providing schedule coordination and physical integration.
There are differences in investment and operating mechanism between the two
measures. In transportation systems requiring higher level of services, the suppliers usually
provide an integrated version of the two. As discussed before, information assistance is
overlapping with multiple aspects of the level of service, and thus is an element that links all
these measures of performances. Or more generally, we can say that information is an
essential element when it comes to interfaces between or integration of transportation
modes. When a certain service is provided passengers may or may not be aware of it. We
suggest that when physical integration is provided, the supplier is responsible for the
Measuring Multimodal Transport Level of Service
39
fundamental level of information assistance, such as signage. This level of assistance is
considered so fundamental that it will not be counted as extra facility utility. When it comes
to a larger picture, such as integrated scheduling and physical coordination, then issue of
information assistance is even more critical, and requires larger scale of planning. Factors
such as operational skills are also indispensible. There is also the issue of “ structural utility”,
which determines primary impact of physical integration. How all of this enters into the
evaluation of intermodal services requires much more research than is currently available.
7 Conclusion and Future Work
In this report, we identified the attributes and their role in multimodal transportation,
both from the demand and supply perspectives. From the supply side we focused on
public transportation agencies, although a similar analysis could be done from the
perspective of a private provider. But multimodal transportation tends to be the domain
of public agencies given its multiplicity of infrastructures involved. This study
demonstrated that both the process and the result of multimodal transportation could be
quite different from the summation of results across each mode. The intermodal
integration is a challenging process that requires additional research. In order to get a
better understanding of this problem, we started out with a conceptual framework, in
which the integration methodology of each attributes in the measure of performance is
identified. This formed the basis for an analytical framework for quantitative modeling.
We developed an indirect utility function that quantifies the user’s perception of level of
service. With also developed a set of metrics to quantify the measures of performance
from the supplier’s perspective. The main conclusions of this work can be summarized
as follows:
1. Measuring the performance of multimodal transportation systems requires a far
more complex integration than the simple addition of metrics across elements of
these systems.
2. The appropriate method of integration of metrics is itself a function of the
context of evaluation, the perspective taken and the role of each stakeholder. In
particular, the integration of measures of performance as perceived by users is
quite different when seen from the perspective of the supplier or the public
agency. Externalities and other “ publicly” valued attributed enter into the latter
but not necessarily the former.
3. Many attributes exhibit discontinuities in their impact and the way they are
perceived. This requires careful modeling to arrive at an appropriate measure of
performance.
4. A major element of multimodal transportation is the intermodal interface
between two or more modes. The contribution of this element to the overall
performance of a multimodal system is complex and challenging to model. More
than in the context of line- haul operation, intermodal systems are subject to
perceptions that are rather subjective in nature and that are influences by the
degree of operational coordination between modes, as well as the extent of actual
physical integration that is provided in the intermodal terminals.
Measuring Multimodal Transport Level of Service
40
Finally we identify four topics that are in need of further research:
a. The threshold value of attributes in demand modeling,
b. The reliability problem in multimodal transportation,
c. Externalities and the disutility of non- users.
d. The Problem of the “ last mile” and the impact of structural integration on
the perception of the level of service of intermodal connections.
Measuring Multimodal Transport Level of Service
41
References:
BANERJEEM, YE AND PENDYALA ( 2006). Understanding travel time expenditures around the
world: exploring the notion of a travel time frontier. Transportation ( 2007) 34: 51– 65.
BATES, J., DIX, M., MAY, T., 1987. Travel time variability and its effect on time of day choice
for the journey to work. In: Transportation Planning Methods, Proceedings of Seminar C
held at the PTRC Summer Annual Meeting, vol. P290, pp. 293– 311.
BATES, J., POLAK, J., JONES, P., COOK, A., 2001. The valuation of reliability for personal
travel. Transport Research Part E 37, 191– 229.
B.- E. MOEN ( 2007). Determinants of safety priorities in transport – The effect of personality,
worry, optimism, attitudes and willingness to pay Safety Science 45 ( 2007) 848– 863;
DELUCCHI, M. ( 2000). Environmental Externalities of Motor- Vehicle Use in the US.? Journal
of Transport Economics and Policy 34( 2): 135- 168.
DEPARTMENT OF TRANSPORTATION ( 2006). DOT Strategic Plan 2006- 2011
WASHINGTON, D. C. 20590.
D. POTOGLOU ET AL. ( 2010) Quantifying individuals’ trade- offs between privacy, liberty and
security: The case of rail travel in UK. Transportation Research Part A 44 ( 2010) 169– 181.
ERIC JESSUP, GRANT MONSON, KEN CASAVANT ( 2006). A Feasibility Study Evaluating
Transportation Security Systems and Associated Multi- Modal Efficiency Impacts. Final
Technical Report TNW2006- 13 TransNow Budget No. 145836, Task 1.
EZRA HAUER ( 1999). Safety and the Choice of Degree of Curve. Transportation Research
Record 1665 ( 22- 27);
GIULIANO, G, ( 1989). New Directions for Understanding Transportation and Land Use,
Environment and Planning A, 21, ( 145- 159).
ISEKI, HIROYUKI AND TAYLOR, BRIAN D.( 2009) ' Not All Transfers Are Created Equal:
Towards a Framework Relating Transfer Connectivity to Travel Behaviour', Transport
Reviews, 29: 6, 777 — 800.
KATO AND ONODA ( 2009). Transportation Research Record: Journal of the Transportation
Research Board, No. 2135, Transportation Research Board of the National Academies,
Washington, D. C., 2009, pp. 10– 16.
KENYON, S. AND G. LYONS. 2003. The value of integrated multimodal information and its
potential contribution to modal change. Transportation Research Part F, 6: 1– 21.
Measuring Multimodal Transport Level of Service
42
KINGHAM, S., J. DICKINSON, AND S. COPSEY. 2001. Travelling to work: Will people move out
of their cars? Transport Policy B, 8( 151– 160).
Lyons, G., Harman, R., Austin, J., & Duff, A. ( 2001). Traveller information systems research:
a review and recommendations for transport direct. Available at:
http:// www. dtlr. gov. uk/ itwp/ transdirect/ travinfo/ index. htm, current 15/ 04/ 02.
LI AND WACHS ( 2000). A Test of Inter- modal Performance Measurement of Transit
Investment Decisions. Transportation Vol. 27 no 3 ( 243- 267).
NCHRP Report 433 Guidelines for developing and maintaining successful partnerships for
multimodal transportation projects
NCHRP Research Results Digest ( July 1998, No. 226): Multimodal Transportation:
Developing of a Performance- Based Planning Process;
NOLAND, R. B., 1999. Information in a two- route network with recurrent and non- recurrent
congestion. In: Emmerink, R., Nijkamp, P. ( Eds.), Behavioural and Network Impacts of
Driver Information Systems. Ashgate Publishing, Aldershot.
NOLAND, R. B., POLAK, J. W., 2002. Travel time variability: a review of theoretical and
empirical issues. Transport Reviews 22 ( 1), 39– 93.
NOLAND, R. B., SMALL, K. A., 1995. Travel- time uncertainty, departure time choice, and the
cost of morning commutes. Transportation Research Record 1493, 150– 158.
NOLAND, R. B., SMALL, K. A., KOSKENOJIA, P., CHU, X., 1998. Simulating travel reliability.
Regional Science and Urban Economics 28, 535– 564.
PETER G. FURTH AND THEO H. J. MULLER ( 2006). Service Reliability and Hidden Waiting
Time Insights from Automatic Vehicle Location Data. Transportation Research Record:
Journal of the Transportation Research Board, No. 1955, Transportation Research Board of
the National Academies, Washington, D. C., 2006, pp. 79– 87.
RAMSEUR AND PARKER ( 2009). Carbon Tax and Greenhouse Gas Control: Options and
Considerations for Congress. Congressional Research Service 7- 5700.
RICHARD H. PRATT AND TIMOTHY J. LOMAX ( 1996), Performance Measures for Multimodal
Transportation Systems, Transportation Research Record 1518 ( 85 – 93)
ROBERT A. JOHNSTON ( 1994). The Evaluation of Multimodal Transportation Systems for
Economic Efficiency and Other Impact, University of California Transportation Center,
Working Paper No. 272;
S. C. HSU ( 2010). Determinants of passenger transfer waiting time at multi- modal connecting
stations. Transportation Research Part E 46 ( 2010) 404– 413;
Measuring Multimodal Transport Level of Service
43
WILLIAMS AND SEGGERMAN ( 2004). Final Report Model Regulations and Plan Amendments
for Multimodal Transportation Districts. National Center for Transportation Research,
Center for Urban Transportation Research, University of South Florida;
WU, LEVINSON, AND LIU ( 2009) Perception of Waiting Time at Signalized Intersections.
Transportation Research Record: Journal of the Transportation Research Board, No. 2135,
Transportation Research Board of the National Academies, Washington, D. C., 2009, pp. 52–
59.
YE, KONDURI, PENDYALA, AND SANA ( 2009). Formulation of an Activity- Based Utility
Measure of Time Use Application to Understanding the Influence of Constraints,
Transportation Research Record: Journal of the Transportation Research Board, No. 2135;
Z. LI ET AL. ( 2010). Willingness to pay for travel time reliability in passenger transport: A
review and some new empirical evidence. Transportation Research Part E 46 ( 2010) 384–
403;
􀀁