Project description

Operating speed is one of the most important factors in road design, evaluation and monitoring. For road designers, the operating speed has been gaining relevance over the last few decades, with several official guidelines urging to anticipate the operating speed estimation in the design process, with the aim of reflecting driver expectancy in the selection of design speed. Therefore, the operating speed is currently regarded as a means to improve design consistency and safety performance when setting road geometrics. Likewise, operating speed is a crucial performance measure during the lifespan of a roadway infrastructure. Drivers’ evaluation of travel time, cost, and convenience determine their route choice and are strongly influenced by their perceptions of the operating speeds of different routes. In turn, road managers use the operating speed as an input for multiple actions oriented to alleviate congestion, promote safety, and improve the environmental efficiency of the road system.

To address the problem of estimating road operating speeds, the research community, public authorities, and road operators have conducted increasing efforts to deliver adequate speed predictions tools and study its main drivers under different contexts. However, several authors point toward the need for further research aimed to fill the gaps in the literature. The deficiencies identified in the existing speed models are mainly related to the attention placed on certain geometric elements to the detriment of others, the preclusion of relevant variables, the limitations of traditional model formulations, and the debatable assumptions of driving behavior.

This thesis aims to enhance operating speed prediction capabilities in two-lane rural highways by delivering five new models with distinctive formulation and increased applicability. The work starts with two exploratory studies, one regarding the definition of free-flow traveling conditions in Portugal, and another, conducted in simulated environment, aimed to clarify the interaction between the cross-section characteristics and the operating speed. The findings from these studies are used in the subsequent speed model development. Models are estimated through regression techniques applied for the first time to operating speed modeling, using a database consisting of geometric and roadside interference parameters of seven Portuguese two-lane highways and the speeds of more than 23,500 observed vehicles.

The set of developed models is composed of four spot speed prediction models, applicable to individual curves and tangents of the horizontal alignment, and one segment speed prediction model, applicable to a length of roadway. The first model consists of an  85th-percentile speed prediction model, estimating the effects on spot speeds of on-site, upstream and downstream effects through an exponential regression. This model, calibrated for two-lane highways classified as National Roads in Portugal, serves as the starting point to develop improved methods to estimate percentile speeds. In this sense, this research proposes the operating speed frontier model (OSFM) formulation, representing a completely new approach to speed modeling, based on the principles of stochastic frontier production modeling used in econometric analysis. This formulation allows for the estimation of any percentile speed from a deterministic frontier representing the maximum operating speed as an exponential function of road characteristics. The OSFM formulation is introduced by the second operating speed model presented in this thesis, also calibrated for Portuguese National Roads. The third model improves the previous model specification by considering additional variables, and the fourth model enlarges the scope of applicability to roads of higher functional classification, i.e., Principal and Complementary Itineraries. Finally, an OSFM for road segments is developed, considering both road and traffic characteristics. The model, calibrated for Portuguese National Roads, is the only model to date applicable to segments of two-lane highways that was developed outside the scope of design guides.

The outcomes of this research address major gaps identified in the literature about operating speed modeling and significantly improve speed prediction capabilities. The ability to provide estimations of any user-specified percentile speed for horizontal curves, tangents, and segments of two-lane rural highways gives the models increased flexibility to assist practitioners in diverse applications related to road design and operations. This thesis provides a compilation of six scientific papers, one submitted and five published in peer-reviewed journals, corresponding to the main methodological steps of this research.

Research team

• António Lobo
• Carlos Rodrigues (supervisor)
• António Couto (co-supervisor)

Financial support

• PhD Scholarship granted by FCT

Stage of progress

• Concluded in 2017