Lead Presenter Biography

Dr. Xuesong (Simon) Zhou, a faculty member at Arizona State University (ASU), is an expert in the methodological development of transportation network modelling and developing open-source tools for multimodal transportation planning applications. Simon directs the ASU Transportation+AI Lab, which has produced several open-source packages with over 100K downloads. With a total of 9,000 citations in Google Scholar, Dr. Zhou's work has significantly contributed to the field. He also serves as subcommittee chair of Network Models in Practice in the TRB Committee on Transportation Network Modeling (AEP40).

Co-Authors

Presentation Description

As an important modeling process to capture long-term demand and supply interactions, planners need to map the volume during the analysis period, such as AM, MD, and PM, to peak hour volume/demand. This is typically called the traffic peak spreading analysis step. In the current practice, the 15 min volume observed from loop detectors is typically used to determine related temporal profile factors such as K-factors.

Two challenges are associated with the current common approach, under heavy congestion conditions.

1) Simply using observed 15 min or hourly volume under-estimate the true demand behind the bottleneck, as the discharge rate on the supply side is lower than the ultimate capacity during the peak hour.
2) Simply summing up the volume over the entire congestion duration measures also overestimate peak-hour demand since when congestion duration is longer than 1 hour, it could include many travelers with preferred arrival time different from peak hour.

In this talk, we introduce a realizable peak hour load based on the concept from electrical engineering to offer a more conservative and realistic measure, with a data-driven process to guide travel demand spreading planning in a practical and robust fashion.

First, we will talk about the peak hour factor (PHF), which is used to convert the peak hour traffic volume into the flow rate representing the busiest 15 minutes of the rush hour in the assignment period (e.g., AM or PM). It can be used to quantify the effects of short-term traffic peaking, leading to congestion. The K factor is the proportion of annual average daily traffic (AADT) occurring in the peak hour. Then, we will systematically connect the definition of PLF with well-accepted research and terminology.

In addition, we will derive the PLF. In the National Cooperative Highway Research Program (NCHRP) report (Horowitz et al., 2014), a link-based peak spreading approach analyzes the traffic occurring in the peak hour at a given location and ships excessive traffic onto the shoulder hours, mainly based on the overloaded demand-to-capacity ratio, while the trip-based approach modifies the departure-time based trip tables to reflect the shifting of trips to other times of the day. Thus, we will review Newell's continuous-time fluid-based polynomial arrival rate (PAQ) model to connect it with PLF and show the audience how we derive the PLF utilizing the quadratic formulation. The assumption we made here is that the arrival rate follows a quadratic form.

Finally, we will extend the PLF to practical applications and future research.
1) Foundation for transportation systems analysis. Peak load volume represents a critical module in traffic assignment for future infrastructure investments.
2) Connections with queueing models to measure delay performance with a suitable volume-delay function and calibrate related parameters (Pan et al., 2022; Zhou et al, 2022).
3) Connections with capacity-expansion factor to compute the period-based capacity in traffic assignment.
4) Performance measurement under both oversupply and undersupply conditions.
5) Sustainable multimodal systems. A multimodal transportation system requires the prediction of pedestrian flow, bike volume, and railway multicommodity traffic. 

Presentation File