First Advisor

Miguel A. Figliozzi

Term of Graduation

Fall 2020

Date of Publication


Document Type


Degree Name

Master of Science (M.S.) in Civil & Environmental Engineering


Civil and Environmental Engineering




City traffic, Traffic flow, Bicycle commuting, Automobiles, Urban transportation



Physical Description

1 online resource (xi, 105 pages)


This thesis presents a compilation of papers exploring passenger car and bicycle speeds through their interactions with each other and with urban roadway factors.

First, following a concern raised in part of the traffic literature that a large mode shift toward bicycling may cause travel time delays and potentially exacerbate congestion instead of alleviate it unless bicycle lanes are installed, an empirical study detailing how the presence of bicycles on urban roads without bicycle lanes may affect passenger car speeds is presented. Pneumatic tube data from six predominantly low speed, low volume roads in Portland, Oregon were utilized to identify observations of passenger cars (class two vehicles) belonging to one of two vehicle following scenarios. In scenario (i), a passenger car was directly preceded by a bicycle (class one vehicle), and in scenario (ii), a passenger car was directly preceded by another passenger car. Speed distributions were examined, and the mean, the 50th, and the 85th percentile speeds of scenario (i) and scenario (ii) vehicles for both peak-hour and 24-hour traffic were compared using t-tests and confidence intervals. A few statistically significant differences between scenario (i) and scenario (ii) were found, but the actual differences in speed were generally on the order of one mile per hour or less. Thus, from a practical perspective, the presence of bicycles on these roads without bicycle lanes was deemed to have negligible effects on passenger car speeds.

Following the results of the initial study, a second study was conducted to address limitations in the initial study regarding the homogeneity of site characteristics. This second study also expanded the research to explore how oncoming (opposing direction) traffic and the availability of overtaking opportunities might affect passenger car speeds when a bicycle was present on an urban road without bicycle lanes. A large number of datasets (n = 75) from locations in Portland with a variety of geometric, roadway, and traffic characteristics were chosen for examination. As with the initial study, vehicle observations belonging to the previously defined scenarios (i) or (ii) were selected for analysis. Comparisons of the mean and 85th percentile speeds of scenarios (i) and (ii) were performed using t-tests. Relationships between scenario (i) speeds and gap times in oncoming traffic were also investigated. The results of this expanded study support the findings of the initial research in that bicycles did not reduce passenger car speeds by more than one mile per hour at most sites (92%), suggesting bicycles are not likely to cause practical speed reductions on lower speed and volume roads without bicycle lanes. The propensity for significant speed reductions was lower when adequate gaps in oncoming traffic existed for overtaking, and at sites with a lower functional class or where sharrows were present.

After exploring how bicycles might affect passenger car speeds, the focus of the third paper was shifted to the site-level determinants of bicycle speed. Bicycle speed is typically assumed to be a constant value for planning and design purposes. However, the probability of the success of projects aimed at improving bicycle infrastructure and routing may be enhanced if more accurate estimates of bicycle speed can be applied. Prior studies have attempted to model bicycle speed from a mix of site factors and factors related to the individual cyclist, requiring more complex data collection methods, and generally resulted in low R2 values. In this paper, widely utilized pneumatic tubes were once again leveraged to collect traffic data for bicycles and passenger vehicles. This traffic data was combined with additional site-level geometric and roadway data to predict mean bicycle speed using generalized linear regression. The adjusted R2 of the final model was 0.63, suggesting a reasonable fit. The regression analysis revealed that grade, negatively associated with the mean bicycle speed, is the most important determinant, accounting for 79% of the final model's explanatory power. The average passenger car speed, the segment length, the percentage of bicycle traffic, and the presence of a shared bikeway facility had statistically significant (p < 0.05) positive effects on the mean bicycle speed. On shared roads, the interaction of the bicycle facility type and the percentage of bicycles was found to have a moderating effect on the mean bicycle speed.


© 2020 Jaclyn Sue Schaefer

In Copyright. URI: This Item is protected by copyright and/or related rights. You are free to use this Item in any way that is permitted by the copyright and related rights legislation that applies to your use. For other uses you need to obtain permission from the rights-holder(s).

Persistent Identifier