Traffic assignment is the last step in the conventional four-step transportation planning model, following trip generation, trip distribution, and mode choice. It concerns selection of routes between origins and destinations on the traffic network. Traditional traffic assignment methods do not consider trip chaining behavior. Since commuters always make daily trips in the form of trip chains, meaning a traveler’s trips are sequentially made with spatial correlation, it makes sense to develop models to feature this trip chaining behavior. Network performance in congested areas depends not only on the total daily traffic volume but also on the trip distribution over the course of a day. Therefore, this research makes an effort to propose a network traffic assignment framework featuring commuters’ trip chaining behavior. Travelers make decisions on their departure time and route choices under a capacity-constrained network. The modeling framework sequentially consists of an activity origin-destination (OD) choice model and a dynamic user equilibrium (DUE) traffic assignment model. A heuristic algorithm in an iterative process is proposed. A solution tells commuters’ daily travel patterns and departure distributions. Finally, a numerical test on a simple transportation network with simulation data is provided. In the numerical test, sensitivity analysis is additionally conducted on modeling parameters.

A general framework is presented for replacing static traffic assignment with dynamic traffic assignment within the standard four step transportation planning model. Issues including model consistency and the implementation of a proper feedback loop are explored. The new model is compared with the standard four step model in order to highlight the benefits of using dynamic traffic assignment rather than static. The model is then extended to include a term for the difference between experienced and free-flow travel times, which can be used as a proxy for travel time reliability and highlights the benefits of time-dependent DTA. Additionally, a study on improving the quality of convergence for dynamic traffic assignment is conducted in order to help facilitate the usefulness of this modeling approach in practice. A variety of equilibration techniques are tested, and analysis is performed to contrast these techniques with the method of successive averages. / text

Transportation planning and operation requires determining the state of the transportation system under different network supply and demand conditions. The most fundamental determinant of the state of a transportation system is time-varying traffic flow pattern on its roadway segments. It forms a basis for numerous engineering analyses which are used in operational- and planning-level decision-making process. Dynamic traffic assignment (DTA) models are the leading modeling tools employed to determine time-varying traffic flow pattern under changing network conditions. DTA models have matured over the past three decades, and are now being adopted by transportation planning agencies and traffic management centers. However, DTA models for large-scale regional networks require excessive computational resources. The problem becomes further compounded for other applications such as congestion pricing, capacity calibration, and network design for which DTA needs to be solved repeatedly as a sub-problem. This dissertation aims to improve the efficiency of the DTA models, and increase their viability for various planning and operational applications. To this end, a suite of computational methods based on the combinatorial approach for dynamic traffic assignment was developed in this dissertation. At first, a new polynomial run time combinatorial algorithm for DTA was developed. The combinatorial DTA (CDTA) model complements and aids simulation-based DTA models rather than replace them. This is because various policy measures and active traffic control strategies are best modeled using the simulation-based DTA models. Solution obtained from the CDTA model was provided as an initial feasible solution to a simulation-based DTA model to improve its efficiency – this process is called “warm starting” the simulation-based DTA model. To further improve the efficiency of the simulation-based DTA model, the warm start process is made more efficient through parallel computing. Parallel computing was applied to the CDTA model and the traffic simulator used for warm starting. Finally, another warm start method based on the static traffic assignment model was tested on the simulation-based DTA model. The computational methods developed in this dissertation were tested on the Anaheim, CA and Winnipeg, Canada networks. Models warm-started using the CDTA solution performed better than the purely simulation-based DTA models in terms of equilibrium convergence metrics and run time. Warm start methods using solutions from the static traffic assignment models showed similar improvements. Parallel computing was applied to the CDTA model, and it resulted in faster execution time by employing multiple computer processors. Parallel version of the traffic simulator can also be embedded into the simulation-assignment framework of the simulation-based DTA models and improve their efficiency. / text

The Multi-resolution Assignment and Loading of Transport Activities (MALTA) system is a simulation-based Dynamic Traffic Assignment model that exploits the advantages of multi-processor computing via the use of the Message Passing Interface (MPI) protocol. Spatially partitioned transportation networks are utilized to estimate travel time via alternate routes on mega-scale network models, while the concurrently run shortest path and assignment procedures evaluate traffic conditions and re-assign traffic in order to achieve traffic assignment goals such as User Optimal and/or System Optimal conditions.Performance gain is obtained via the spatial partitioning architecture that allows the simulation domains to distribute the work load based on a specially designed Recursive On-line Load Balance model (ROLB). The ROLB development describes how the transportation network is transformed into an ordered node network which serves as the basis for a minimum cost heuristic, solved using the shortest path, which solves a multi-objective NP Hard binary optimization problem. The approach to this problem contains a least-squares formulation that attempts to balance the computational load of each of the mSim domains as well as to minimize the inter-domain communication requirements. The model is developed from its formal formulation to the heuristic utilized to quickly solve the problem. As a component of the balancing model, a load forecasting technique is used, Fast Sim, to determine what the link loading of the future network in order to estimate average future link speeds enabling a good solution for the ROLB method.The runtime performance of the MALTA model is described in detail. It is shown how a 94% reduction in runtime was achieved with the Maricopa Association of Governments (MAG) network with the use of 33 CPUs. The runtime was reduced from over 60 minutes of runtime on one machine to less than 5 minutes on the 33 CPUs. The results also showed how the individual runtimes on each of the simulation domains could vary drastically with naïve partitioning methods as opposed to the balanced run-time using the ROLB method; confirming the need to have a load balancing technique for MALTA.

In recent years, traffic congestion in transportation networks has grown rapidly and has become an acute problem. The impetus for studying this problem has been further strengthened due to the fast growing field of Intelligent Vehicle Highway Systems (IVHS). Therefore, it is critical to investigate and understand its nature and address questions of the type: how are traffic patterns formed? and how can traffic congestion be alleviated? Understanding drivers' travel times is key behind this problem. In this paper, we present macroscopic models for determining analytical forms for travel times. We take a fluid dynamics approach by noticing that traffic macroscopically behaves like a fluid. Our contributions in this work are the following: (i) We propose two second-order non-separable macroscopic models for analytically estimating travel time functions: the Polynomial Travel Time (PTT) Model and the Exponential Travel Time (ETT) Model. These models generalize the models proposed by Kachani and Perakis as they incorporate second-order effects such as reaction of drivers to upstream and downstream congestion as well as second-order link interaction effects. (ii) Based on piecewise linear and piecewise quadratic approximations of the departure flow rates, we propose different classes of travel time functions for the first-order separable PTT and ETT models, and present the relationship between these functions. (iii) We show how the analysis of the first-order separable PTT Model extends to the second-order model with non-separable velocity functions for acyclic networks. (iv) Finally, we analyze the second-order separable ETT model where the queue propagation term - corresponding to the reaction of drivers to upstream congestion or decongestion - is not neglected. We are able to reduce the analysis to a Burgers equation and then to the more tractable heat equation. / Singapore-MIT Alliance (SMA)

In this thesis, we propose a dynamic traffic adjustment scheme for end-to-end proportional differentiated services (EEPDS) on MPLS networks. When the Egress LSR (Label Switch Router) first observes that the proportional ratio of multiple flows could not be maintained, it will locate the performance bottlenecks by sending control messages to all the traversed LSR. In the proposed scheme, the identified bottleneck routers may have to drop the packets of traffic flows that do not require any QoS assurance (for example, the best-effort flows). Moreover, if dropping packets of the best-effort flows could not meet the goals of maintaining proportional ratios, the Ingress LSR would have to shape the traffic rates of sending flows proportionally based on the throughput values observed by the Egress LSR. Once a bottleneck router regains its sufficient bandwidth released by other traffic flows, it will send out control packets to notify the Ingress LSR to recover the original sending rates. Eventually, the proposed EEPDS mechanism can meet the goals of maintaining the proportional ratios by dynamically tuning the traffic flows. For the purpose of demonstration, we implement the EEPDS scheme by using MNS simulator. According to the experimental results, the EEPDS scheme has shown that it can meet the proportional ratios of multiple traffic flows even under severe network congestions, where multiple performance bottlenecks may occur. Finally, we observe that the processing overhead of the proposed EEPDS scheme is very small, since it is triggered only when the Egress LSR detects the situation of unsatisfied proportional ratios.

A transition to low speed vehicles (LSVs), a federally-designated class of vehicles smaller, lighter and slower (limited to maximum speeds between 20 and 25 mph) than conventional automobiles, for intra-city travel offers several advantages. Their smaller size provides roadway space for other modes such as cycling and reduces the amount of land dedicated to vehicles. Their lower maximum speeds are more compatible with operation in populated areas where cars traveling at 30 mph prove deadly for pedestrians and people biking, and their energy usage and emissions are less than conventional automobiles. Communities such as Lincoln, CA, Peachtree City, GA, and those in the South Bay Cities and Western Riverside Councils of Governments in California recognize the benefits of using LSVs and actively provide infrastructure and programs to support their use. Considering the advantages of LSVs, this dissertation demonstrates potential ways to transition to LSVs and seeks to answer a question considered key to their adoption as the means of motorized travel in the city: Could LSVs also offer a travel time advantage? The basis for this seemingly paradoxical question is the observation that because of their smaller size, lower weight, and slower speed, more space- and operationally-efficient intersections, such as LSV-scaled roundabouts, overpasses and interchanges, are possible within the existing right-of-way to replace signalized intersections. The hypothesis that LSVs can offer comparable or better travel time compared to conventional automobiles assumes the removal of intersection delay will allow LSVs to make-up for their slower speeds. The methodology to test the hypothesis uses dynamic traffic assignment to compare average system, corridor and origin to destination travel times for conventional automobiles and LSVs in a subnetwork of Austin, Texas during transition periods when both vehicles are permitted and when only LSVs may be used for intra-city motorized travel. The findings indicate LSVs can offer similar and in some cases better average travel times than those for conventional automobiles, especially for the LSV-only network. However, careful planning is required during the transition stages when both vehicle types are in operation to maintain acceptable travel times for both conventional automobiles and LSVs. / text

A comprehensive approach is conducted to better utilize dynamic traffic assignment (DTA) in transportation planning by investigating its role in: (1) high-order functions, (2) project evaluation, and (3) traffic assignment. A method is proposed to integrate DTA and the four-step planning model such that traffic assignment is conducted at the subnetwork level while the feedback process occurs at the regional level. By allowing interaction between the subnetwork and regional area, the method is shown to be more beneficial than previous integration structures. Additionally, DTA is applied to a case study involving the proposed urban rail system in Austin, TX. The case study showcases the benefits and capabilities of DTA when analyzing traffic impacts caused by transit rail facilities. Multiple equilibria are shown to arise in simulation-based DTA models due to simplified fundamental diagrams. Piecewise linear diagrams are introduced to eliminate unlikely equilibria. Game theory is also applied to DTA; it is shown that an equilibrium solution is guaranteed to exist for general networks in mixed strategies, and unrealistic equilibria are reduced using the trembling hand refinement. / text

A traffic network can be disrupted by work zones and incidents. Calculating diversion rate is a core issue for estimating demand changes, which is needed to select a suitable work zone configuration and work schedule. An urban network can provide multiple alternative routes, so traffic assignment is the best tool to analyze diversion rates on network level and the local level. Compared with the results from static traffic assignment, dynamic traffic assignment predicts a higher network diversion rate in the morning peak period and off-peak period, a lower local diversion rate in the morning peak period. Additionally, travelers may benefit from knowing real-time traffic condition to avoid the traffic incident areas. Deploying variable message signs (VMSs) is one possible solution. One key issue is optimizing locations of VMSs. A planning model is created to solve the problem. The objective is minimize total system travel time. The link transmission model is used to evaluate the performance of the network, and bounded rational behavior is used to represent drivers' response to VMSs. A self-adapting genetic algorithm (GA) is formulated to solve the problem. This model selects the best locations to provide VMSs, typically places are that allow travelers to switch to alternative routes. Results show that adding more VMSs beyond a certain threshold level does not further reduce travel time. / text

Dynamic traffic assignment (DTA) can be used to model impacts of network modification scenarios, including traffic control plans (TCPs), on traffic flow. However, using DTA for modeling construction project impacts is limited by the computational time required to simulate entire roadway networks. DTA modeling of a portion of the larger network surrounding these work zones can decrease the overall run time. However, impacts are likely to extend beyond typical boundaries, and determining the proper extents to be analyzed is necessary. Therefore, a methodology for selecting an adequate portion to analyze using DTA, along with provision for properly analyzing the resultant subnetwork, is necessary to determine the magnitude of construction impacts. The primary objectives of this research center on evaluating subnetwork sizes to determine the appropriate extents required to analyze network modifications and developing a strategy to account for impacts extending beyond the subnetwork boundary. The first objective is accomplished through an in-depth review of subnetwork sizes relative to multiple impact scenarios. Three statistical measures are implemented to evaluate the adequacy of a chosen subnetwork relative to the derived impact scenarios based on an assessment of boundary demand. Ultimately, the root mean squared error is used successfully to provide a series of recommended subnetwork sizes associated with an array of possible impact scenarios. These recommendations are validated, and application of the proposed methodology demonstrated, using five scenarios selected from real-world network modifications observed in the field. When a subnetwork is not large enough and impacts to inbound trips pass beyond the boundary, there is a change in flow at this location that can be represented by a change in the demand assigned to the subnetwork at each entry point. As such, two strategies for adjusting the demand at subnetwork boundaries are implemented and evaluated. This includes use of results from static traffic assignment (STA) models to identify where flow changes occur, and implementation of a logit formulation to estimate demand adjustments based on differences in internal travel times between base and impact scenario models. Based on preliminary results, the logit method was selected for large-scale implementation and testing. In the end, an inconsistent performance of the logit method for full implementation highlights the limitations of the methodology as applied for this study. However, the results suggest that a refined strategy that builds on the foundation established could work more effectively and produce valuable subnetwork demand estimates in the future. This research is used to provide recommendations for selecting and analyzing subnetworks using DTA for an array of common impact scenarios involving network modifications. The tradeoffs between improved efficiency and reduced accuracy associated with using subnetworks are thoroughly demonstrated. It is shown that a considerable amount of computational time and space, as well as effort on the part of an analyst, can be saved. A number of limitations associated with subnetworks are also identified and discussed. The proposed methodology is implemented and evaluated using several software programs and as a result, a number of useful tools and software scripts are developed as part of the research. Ultimately, the valuable experience gained from performing an extensive review of subnetwork analysis using DTA can be used as a basis from which to develop future research initiatives. / text