Improved transit service through real-time operations control and traveler information

Managers of modern transport systems face the challenge of serving an ever increasing demand for travel. At the same time, they are expected to curb congestion and address the undesired effects of travel: maintain sustainable levels of energy consumption and pollution from vehicles, and limit the land and other resources consumed by the transport system. Over the years, it has become evident that one of the most effective strategies to achieve these potentially conflicting goals is to develop and operate Public Transport Systems (PTS) that offer high level of service and to promote their use. To this end, transport planners and policy-makers worldwide are designing systems and policies that would enable provision of efficient and robust PTS. Efficiency means that travel should be fast, convenient and reliable under normal operating conditions. Robustness means that the system should be able to withstand or quickly recover from inherent uncertainties (e.g. unexpected traffic delays and congestion, sudden surges in the passenger demand profile, incidents and events) as well as irregular service disruptions (e.g. vehicle malfunctions and operator absentees).

Even an optimally-designed PTS may not perform as expected. To realize its potential real-time control and operations of the PTS is required. This is needed in order to adapt the service to varying traffic flow conditions, passenger demand, incidents and events, service disruptions and other interruption and irregularities. Strategies and algorithms for real-time PTS control and operations are designed to obtain the highest possible benefits to travellers, operators and the transport system as a whole. Examples of PTS control strategies include holding, expressing, skip-stop operations, short-turning and deadheading. All are aimed to maintain regular service and recover from disruptions and unexpected events.

The integration of Information Technology (IT) in PTS facilitates the implementation of these control strategies through real-time collection, transmission, processing and analysis of data that describes that state of the transport system as a whole (e.g. the current locations of transit vehicles, traffic conditions). The increasing availability of such data is rapidly reshaping PTS operations. Innovations in this area relate both to the user side (e.g. personalised, real-time information) and to the supply side (e.g. real-time operations, control and planning) of PTS.

Despite the importance of PTS operations to service quality and cost, much less attention has been given to real-time operations control when compared to service planning and design. This deficiency holds for the basic research to develop and refine control strategies as well as the identification of the conditions under which they would perform well (or fail). The overall objective of the proposed project is to develop a decision-support system that will facilitate proactive and predictive public transport operations and traveller experience. The proposed decision-support system will implement strategies and algorithms for real-time transit control and operations that are designed to maximize the benefits to travellers, operators and the transport system as a whole. Figure 1 below illustrates the framework of the proposed project. The main characteristics and features of the proposed decision-support system are:

A simulation-based modelling capability to predict the short-term dynamics of the state of the public transport system under various traffic flow and demand conditions
Methods and tools to estimate the current state of the systems and to predict future states of the public transportation service by taking into consideration both historical data and real-time measurements of the general traffic flows, public transport vehicles and passenger flows.
Use of the simulation-based model to assess the impacts of alternative TPS operation strategies such as holding, skip-stop, transfer coordination and expressing. These assessments will result in recommendations for intervention by the service control operators to mitigate and recover from the effects of service disruptions.
When making predictions, the model would take into account the response of travellers’ to the information provided to them and its impact on the predicted conditions within the system.
The inclusion of the effects of information provision on the state of the transport system enables the model to also generate consistent prescriptive and predictive real-time travellers’ information that can be disseminated to travellers in response to varying conditions and disruptions. Thus, allowing travellers to make more informed travel decisions and improve their travel experience.
The model will consider multimodal door-to-door passenger trips when evaluating alternative strategies. This will allow capturing the effects on the entire trip, taking into account walking segments, transfers and their coordination.

Figure 1: Conceptual framework of the ADAPT-IT project

The development of a decision support system to facilitate adaptive operations and traveller experience requires the dynamic modelling of the transport system. The main research fields addressed in ADAPT-IT include traffic and public transport simulation, dynamic route choice, traffic predictions, public transport operations and route choice modelling.

The integration of Information Technology (IT) facilitates the real-time collection, transmission and analysis of data that is rapidly changing public transport operations. The rapid deployment of IT in the public transport industry facilitates the design of more elaborate real-time operational strategies (FTA, 2000 and 2006). Public transport operational strategies consist of a wide variety of operational methods aimed to improve PTS performance and level of service. This includes measures to improve transfer coordination, control strategies, signal priority and real-time travel information provision through signs, apps and social media. Public transport fleet management strategies include expressing (which implies skipping stops), short-turning (requiring the vehicle to terminate its service) and deadheading (redirecting vehicles between terminals and depots).

Previous studies analysed the impact of such strategies for a single line in isolation under ‘normal operations’ using either analytical or simulation models. Furthermore, measures intervening in how the service is run were considers separately from measures directed towards travellers and their travel choices. However, transport supply and demand are interrelated and the design of real-time strategies should consider their mutual effects. For example, the expected arrival time at downstream stops should incorporate the impact of delay due to waiting for a connecting transfer service. Moreover, prescriptive information about alternative connections in case of service disruption should take into consideration the impact of such information in order to avoid overloading an already highly saturated service. Mitigation strategies designed to improve service reliability should therefore investigate the cascading effects of service discrepancies and their impacts on travelling patterns. This is particularly relevant for multi-modal transport hubs.

Transport systems involve complex dynamics and evolve through the interaction of various entities. The analysis of system performance requires emulating the dynamic loading of travellers and their interaction with the underlying transport infrastructure and services. Traffic simulation models are the primary modelling approach used to represent traffic dynamics and to test the performance of the transport system under alternative scenarios. Simulation environments can exploit available data (e.g. vehicle positioning) concerning the current state of the system in order to evaluate, in real-time, the impacts of alternative scenarios. Several simulation models such as MITSIMLab and DynaMIT were developed to facilitate the analysis of ITS operating strategies.

Mesoscopic traffic simulation models are the most suitable simulation approach for this purposes since the movement of individual vehicles enables the reproduction of traffic flow dynamics while representing public transport related entities and operational strategies within large-scale applications. Multi-agent simulations aim to mimic the emergence of global spontaneous order from numerous inter-dependent local decisions. DYMOBUS (project financed in “Future Travellers” program 2007-2011 by VINNOVA) is a multi-agent transport operations and assignment model which captures supply uncertainties and adaptive travellers’ decisions. The model involves the development and integration of several modules including traffic simulation, public transport operations and control, dynamic path choice model and real-time information generator. The multimodal traffic simulation represents automobile traffic and public transport networks and individual travellers. Mezzo, a mesoscopic transport simulation model is used as the platform for implementation.

Field experiments (RETT2 and RETT3) have recently demonstrated the benefits from applying a new real-time control strategy designed to improve service reliability. The field experiments realized the expected benefits from introducing the proposed strategy that were first using the DYMOBUS simulation-based framework.

Predictions of the downstream effects of alternative strategies would enable a more proactive and system-wide approach to operations and management. The prediction of public transport vehicle trajectories can be based on historical data, passenger counts and real-time AVL data. The latter can potentially result in more accurate estimations of traffic conditions. Previous studies applied various methods for bus arrival predictions as regression models, artificial neural networks (ANN), Kalman filter and statistical pattern recognition. A patent search has yielded a number of relevant patents that develop methods to provide travel time predictions (WO 2002008922) and in particular in the context of public transportation information (US 6137425 A and WO 2007142462 A1) and travel journey planner (CN 101308555 A).

The Federal Transit Administration in the US has recognized the potential benefits in developing a decision support system for control centre operations. The TODSS project (FTA-IL-26-7009-2009.2) was initiated in 2006 and designed the core functional requirements and a prototype concept of a decision support system for traffic dispatchers including modules to identify service disruptions and a rule-based recommendation for response strategy. However, it did not consider travel information strategies and the consideration of impacts on travellers.

ADAPT-IT will develop a decision support system to systematically evaluate alternative strategies to adjust public transport services and disseminate travel information in real-time. The project will hence involve the integration of advanced in traffic simulation models, real-time information and operational strategies, prediction models and behavioural modelling in the form of a decision support system. The project will contribute to bridging the gap between existing automated data collection methods and developments in the transport management and travel information domains.

ADAPT-IT will develop a decision support system (DDS) that will be deployed by the transport operations management as illustrated in Figure 1 above. Real-time data such as public transport vehicle positions, passenger counts and traffic counts that are available from transport system that are processed and analysed by the control centre systems will be given as input to the DDS in order to replicate the current system performance. An agent-based simulation environment will then be used for mimicking how the transport system is expected to evolve. Alternative real-time fleet and travel information strategies will be evaluated by the DDS. The DDS will then evaluate each candidate strategy as well as the do-nothing strategy based on a set of measures considering passengers and operational perspectives. In order to investigate solution robustness, the prediction will be performed under various scenarios (e.g. demand levels, traffic delays). The DDS will report the most promising strategies to control centre operators who will implement the chosen strategy through the respective control centre systems (e.g. dispatching management system, radio communication, information dissemination channels, journey planners). This will ultimately improve travel experience and improve consequent system performance.