Freeway level of service criteria based on travelers perception

• Autor(a) principal: Piva, Fernando José
• Data de Publicação: 2022
• Tipo de documento: Tese
• Idioma: eng
• Título da fonte: Biblioteca Digital de Teses e Dissertações da USP
• Texto Completo: https://www.teses.usp.br/teses/disponiveis/18/18144/tde-10012023-155328/
• topic: Freeway operations; Level of service; Microsimulation; Microssimulação; Nível de serviço; Online surveys; Operações de rodovias de pista dupla; Percepção do usuário; Pesquisas online; User perception

Level of service (LOS) classifications of traffic operational conditions play a large role in roadway-improvement funding decisions. Consequently, it is essential that traveler perception of LOS is consistent with the values determined through a traffic analysis. Otherwise, the public confidence in appropriate transportation agency funding decisions may be undermined. Research to measure travelers\' perceptions of traffic operational conditions is a relatively immature field and methods to obtain such data are still evolving. Methods used to collect data on the travelers\' perception in previous studies include driver interviews at rest stops, focus groups, and rating video clips recorded from the driver\'s field-of-view. These videos are recorded on real roadways for greater realism but the capture of the full range of operational conditions in a timely manner is its greatest problem. Another significant challenge that all the methods in the literature face is collecting a sufficiently large sample of responses. This dissertation describes an approach to incorporate travelers\' perception of trip quality that addresses the limitations of the previous studies by the use of a combination of realistic looking 3-dimensional traffic stream visualizations and an online survey to reach a large number of people. The use of traffic simulation software provides full control of the traffic and roadway characteristics, and allows for the creation of realistic computer generated animations over the full range of operating conditions in a fast, efficient and economical manner. Furthermore, modern microsimulation software is able to record video clips from the driver\'s viewpoint, which is likely to elicit more accurate rankings from the study participants than an overhead view. The creation of the roadway environment to produce a realistic view from the vehicle\'s cabin interior and an automated method for choosing a representative vehicle from the traffic stream are presented. Data on drivers\' perception of the quality of the trips depicted in the video clips are obtained inviting participants to respond to a web-based survey. In the web survey, participants watched a set of freeway trips under the desired range of roadway and traffic conditions and rate each trip using a visual analog scale that varies from 0 to 100. The travelers\' ratings are then discretized into the desired number of service levels using cluster analysis and the service level boundaries and the associate confidence intervals are estimated using logistic regression. The feasibility of the proposed approach was demonstrated with a case study with 977 participants that rated 10,228 trips depicting 128 different combinations of traffic density, truck percentage, posted speed limit, grade steepness, and number of lanes, chosen according to a fractional factorial design. After a series of filters, the final sample consisted in 6231 ratings by 554 participants. A statistically significant correlation between traffic density and rating was found (&#961; = &#8211;0.532, N = 6231). The case study results indicate that the HCM-7 freeway LOS boundaries are within the confidence intervals estimated using the proposed approach, in the exception of the threshold between LOS D and E. This suggests that, at least for the participants in the study, that drivers perceive LOS D and E as very similar. The effect of sociodemographic, traffic and roadway factors on the participants\' perception of trip quality was investigated using a fractional factorial design. The sample consisted of 7,004 ratings by 625 participants. A total of 16 factors were analyzed using bivariate correlation and traffic density was the only factor that showed a significant, strong correlation with rating (&#961; = 0.520, p < 0.001). Truck percent was also correlated with rankings, at a much lower level (&#961; = &#8211;0.116, p < 0.001) and no significant influence from sociodemographic factors was found. These results were confirmed by a stepwise multiple linear regression model calibrated to the data. A CART decision tree model indicated that participants\' perception of the trip quality tend to be affected by the presence of trucks in the stream when traffic density is greater than 7 veh/km/ln.