ThesisPDF Available

Optimal Time Point Locations on a Markovian Bus Route

Authors:
  • Klumpentown Consulting

Abstract and Figures

Transit agencies struggle daily with the randomness experienced by vehicles as they traverse their routes. Since reliability is one of the most important factors in determining the quality of a transit system, it is important that agencies use effective and data-informed strategies to combat this randomness as much as possible. One commonly used strategy is holding control, where early buses are held at specifically chosen stops known as time points. This strategy presents a trade-off between improved reliability and longer overall travel time, and leads to the central research question: for a route with varied travel times between stops and fluctuating passenger demand, what is the optimal number and location of time points along a route? This thesis develops a mathematical Markov chain model of bus behaviour on an urban bus route which can capture the effects of time point placement and is sensitive to individual inter-stop travel times. After formulating a cost function to account for the various impacts of unreliability on passengers and operators, two algorithms are presented to optimally locate time points, with an improvement on existing configurations used by Calgary Transit in Calgary, Canada. Finally, a simulation model shows that holding control drastically reduces the occurrence of the phenomenon of bus bunching.
Content may be subject to copyright.
A preview of the PDF is not available
... Practically, it may be that a transit agency has a set slack time at stops regardless of the optimal value, or that the value of τ i can be optimized separately using a separate technique such as the one developed by Wirasinghe (1993). Klumpenhouwer (2018) found that slack time adjustment is secondary in its effect on reliability compared with time point placement, and that sensitivity to slack time exists mainly for slack times in the range of 0 to 6 minutes. ...
... We present here an improved heuristic algorithm developed by Klumpenhouwer (2018) utilizing the following key assumptions and strategies: ...
... In Section 7, current practice is considered to use a slack time of three minutes while the heuristic algorithm allows slack time to vary between 0 and 6 minutes. This range was chosen based on the analysis of slack time sensitivity done by Klumpenhouwer (2018). ...
Article
Full-text available
For a scheduled bus route adopting the holding control strategy, determining the optimal number and location of time points is considered a long-standing but elusive problem. In this paper, we take a new approach to the problem by developing a Markov Chain model to accurately capture the stochastic nature of a bus as it moves along a route in mixed traffic. Transition matrices are created using theoretical distributions of travel time calibrated with stop-to-stop travel time and dwell time data. The approach captures analytically the bus behavior while still allowing the model to be informed by the unique characteristics of the route, including travel time between stops and passenger demand. This stochastic process model mimics the physical phenomenon of Brownian motion, and it is found that the compounding nature of randomness leads to greater unreliability as the route progresses. Theoretical analysis of routes allows us to demonstrate where problem points may exist on the route and can point to locations where reliability improvements may be more effective. We develop a cost function to capture the values of time of passengers including waiting time due to early and late buses, and lost time at time points. We include operating cost capturing the increased cost of travel time caused by added control, and the improved overtime costs resulting from more consistent service. Using data from automated vehicle location (AVL) and automated passenger count (APC) systems, an operational route in Calgary, Canada is optimized using the developed model and cost function. A heuristic optimization algorithm is developed to consider high-cost stops iteratively which improves the cost function compared with existing configurations and with fewer time points.
... Preliminary results from this application were developed by and presented inKlumpenhouwer (2018) ...
Preprint
Full-text available
Representing the randomness that is inherent in transportation networks is of key importance for planners, schedulers, analysts, and users. While many detailed stochastic-sensitive simulation models of transportation networks exist, they are often costly, closed, and extremely data intensive, requiring significant investment from researchers and agencies to develop and analyze. To provide a higher-order understanding of the reliability impacts of new infrastructure or operational policies, new flexible data analysis tools and modelling methods are needed. This research presents an adaptable mathematical framework for modelling transportation networks using Markov chains, and presents a new open source tool, MCRoute, which allows users to quickly prototype and analyze stochastic networks using data or theoretical distributions. Applications of the model and the tool are demonstrated in three case studies: (i) the potential impact of infrastructure improvements at a bottleneck on a railway corridor, (ii) the simulation of bus schedule adherence and bunching occurrences under various holding control scenarios, and (iii) the analysis of path-based reliability of users on a multi-stage transit journey across various modes.
Article
Full-text available
For a scheduled bus route adopting the holding control strategy, determining the optimal number and location of time points is considered a long-standing but elusive problem. In this paper, we take a new approach to the problem by developing a Markov Chain model to accurately capture the stochastic nature of a bus as it moves along a route in mixed traffic. Transition matrices are created using theoretical distributions of travel time calibrated with stop-to-stop travel time and dwell time data. The approach captures analytically the bus behavior while still allowing the model to be informed by the unique characteristics of the route, including travel time between stops and passenger demand. This stochastic process model mimics the physical phenomenon of Brownian motion, and it is found that the compounding nature of randomness leads to greater unreliability as the route progresses. Theoretical analysis of routes allows us to demonstrate where problem points may exist on the route and can point to locations where reliability improvements may be more effective. We develop a cost function to capture the values of time of passengers including waiting time due to early and late buses, and lost time at time points. We include operating cost capturing the increased cost of travel time caused by added control, and the improved overtime costs resulting from more consistent service. Using data from automated vehicle location (AVL) and automated passenger count (APC) systems, an operational route in Calgary, Canada is optimized using the developed model and cost function. A heuristic optimization algorithm is developed to consider high-cost stops iteratively which improves the cost function compared with existing configurations and with fewer time points.
Article
Full-text available
Transit agencies increasingly deploy planning strategies to improve service reliability and real-time operational control to mitigate the effects of travel time variability. The design of such strategies can benefit from a better understanding of the underlying causes of travel time variability. Despite a significant body of research on the topic, findings remain influenced by the approach used to analyze the data. Most studies use linear regression to characterize the relationship between travel time reliability and covariates in the context of central tendency. However, in many planning applications, the actual distribution of travel time and how it is affected by various factors is of interest, not just the condition mean. This paper describes a quantile regression approach to analyzing the impacts of the underlying determinants on the distribution of travel times rather than its central tendency, using supply and demand data from automatic vehicle location and farecard systems collected in Brisbane, Australia. Case studies revealed that the quantile regression model provides more indicative information than does the conditional mean regression method. Moreover, most of the coefficients estimated from quantile regression are significantly different from the conditional mean–based regression model in terms of coefficient values, signs, and significance levels. The findings provide information related to the impacts of planning, operational, and environmental factors on speed and its variability. On the basis of this information, transit designers and planners can design targeted strategies to improve travel time reliability effectively and efficiently.
Conference Paper
Full-text available
There are a number of strategies that transit operators may use to mitigate the randomness inherent in a bus’ movement along a route. Many of these strategies have evolved to include real-time information in order to make dynamically informed decisions. One such strategy is the holding control strategy, where certain stops are selected as time points, and early buses are instructed to wait until a scheduled departure time at these points. While extensive research has been done incorporating real-time information into bus operations, little attention has been given to potential advantages gained in using real- time information to control inter-time-point movement. An algorithm is proposed for an early bus, comparing on-board and off-board passenger costs to decide whether to continue to run early or attempt to slow down. The proposed algorithm considers two situations: when inter-time-point stops have published arrival times, and when they must be deduced by the passengers.
Book
This book should be of interest to undergraduate and postgraduate students of probability theory.
Article
Reliability and speed are arguably the most important indicators of surface transit performance for both operators and passengers. They can be influenced by a variety of factors, including service characteristics of bus routes, physical infrastructure, signal settings, traffic conditions and ridership patterns. These factors have often been analyzed individually for their impact on transit reliability or speed. Studies considering more than one factor tend to use one or two transit routes to explore their effects. The study that is the subject of this paper proposed an evaluation framework to guide the selection of an appropriate reliability measure. Regression analysis was applied subsequently to determine the factors that exhibit a statistically significant relationship with transit reliability and speed at both the route and segment levels. Automated vehicle location data of a bus route sample that is representative of the entire bus network in the City of Toronto, Ontario, Canada were used. Features significantly associated with reliability and speed were compared. The results showed that lower transit reliability and speed are significantly associated with the increase in service distance, signalized intersection density, stop density, volume of boarding and alighting passengers, and traffic volume. By segregating bus route segments on the basis of the presence of transit signal priority, the results of the segment-level model demonstrated the beneficial impact of transit signal priority on improving transit reliability.
Article
One of the key components of real-time bus information systems is knowledge of bus locations in real-time. With recent advancements in sensing and communication technologies, location data can be obtained frequently and incorporated into the estimation models. One important issue is determining the suitable time interval at which location data should be chosen as input for estimation models. This paper explores the impact of the time interval of real-time bus location data on the accuracy of bus arrival estimation. It also examines by how much, and at what distance, real-time arrival information can outperform information provided from a static schedule for a given estimation technique. This study also investigates the “cut-off horizon,” the threshold horizon beyond which real-time information no longer outperforms information based on a particular static schedule, and how the cut-off horizon changes with the time interval of bus location data. Three types of estimation models are used, and they are updated with real-time bus locations. It is shown that incorporating real-time bus location data below a 30-second time interval did not increase the accuracy of the applied estimation models. Estimations based on current and average historical bus speed outperformed other estimations only if the time interval of the location data was short. However, estimations that relied on three similar historical trips were more accurate when both the estimation horizon and time interval were longer. Estimations utilizing schedule and real-time bus location information, but not historical trip information, performed worse than the schedule-only case beyond a 7.5 kilometer horizon even for a 30-second time interval.
Chapter
As a bus travels along a route, its trip time between successive bus stops is subject to random fluctuations. If the bus should fall behind schedule because of this, then some excess passengers will have arrived at the bus stops during the late time and it will take the bus longer to load passengers. A bus, therefore, tends to travel slower and falls even further behind schedule. To compensate for this, the usual strategy of control used by bus operators is to provide some slack time in the schedule so that, normally, a bus can gain some time. The bus then operates under a rule that it may not leave a bus stop ahead of schedule, but will leave immediately if it is late. Actually, this control is not completely stable; if a bus falls so far behind schedule that the extra loading time generated by the lateness exceeds the slack time, the bus will still fall progressively further behind schedule.
Article
To better understand the core concepts of probability and to see how they affect real-world decisions about design and system performance, engineers and scientists might want to ask themselves the following questions: · What exactly is meant by probability? · What is the precise definition of the 100-year load and how is it calculated? · What is an “extremal” probability distribution? · What is the Bayesian approach? · How is utility defined? · How do games fit into probability theory? · What is entropy? · How do I apply these ideas in risk analysis? Starting from the most basic assumptions, this book develops a coherent theory of probability and broadens it into applications in decision theory, design, and risk analysis. This book is written for engineers and scientists interested in probability and risk. It can be used by undergraduates, graduate students, or practicing engineers.