Longitudinal Performance Assessment of Traffic Signal System Impacted by Long-Term Interstate Construction Diversion Using Connected Vehicle Data

Local arterials can be significantly impacted by diversions from adjacent work zones. These diversions often occur on unofficial detour routes due to guidance received on personal navigation devices. Often, these routes do not have sufficient sensing or communication equipment to obtain infrastructure-based traffic signal performance measures, so other data sources are required to identify locations being significantly affected by diversions. This paper ex-amines the network impact caused by the start of an 18-month closure of the I-65/70 interchange (North Split), which usually serves approximately 214,000 vehicles per day in Indianapolis, IN. In anticipation of some propor-tion of the public diverting from official detour routes to local streets, a connected vehicle monitoring program was established to provide daily perfor-mances measures for over 100 intersections in the area without the need for vehicle sensing equipment. This study reports on 13 of the most impacted signals on an alternative arterial to identify locations and time of day where operations are most degraded, so that decision makers have quantitative information to make informed adjustments to the system. Individual vehicle movements at the studied locations are analyzed to estimate changes in volume, split failures, downstream blockage, arrivals on green, and travel times. Over 130,000 trajectories were analyzed in an 11-week period. Weekly after-noon peak period volumes increased by approximately 455%, split failures increased 3%, downstream blockage increased 10%, arrivals on green de-creased 16%, and travel time increase 74%. The analysis performed in this paper will serve as a framework for any agency that wants to assess traffic signal performance at hundreds of locations with little or no


Introduction
Interstate maintenance and construction can significantly impact the surrounding network by creating an influx of diverging vehicles that can saturate local streets. This can lead to significant congestion and delays. According to the 2021 Urban Mobility Report [1], in 2019 there were 8.7 billion hours of congestion-related travel delay, which represented a $190 billion cost in time and wasted fuel. The Federal Highway Administration (FHWA) indicates that 10% of all congestion, and 24% of non-recurring congestion, are caused by work zones [2] [3]. Therefore, it is imperative for agencies to develop scalable monitoring tools that do not require infrastructure investments, to detect and mitigate the effects that work zones have on local streets.

Literature Review
There has been extensive research on driver behavior when presented with alternative routes by message signs [4] [5] [6] and by Advanced Traveller Information Services (ATIS) [7] [8], as well as stated surveys and modelling studies that analyze the impact of traffic on alternative routes [9] [10] [11] [12]. It is well understood at the macro model level, that diversions onto local streets will occur, but those models do not provide sufficient fidelity to characterize the daily, hourly, and even 15-minute variations in driver route choices that impact local streets.
Intelligent Transportation System (ITS) technology has been employed to assess the consequences of diversions. Bluetooth sensors have been used to identify driver route choices related to work zones [13] [14] [15]. In addition, agencies in Indiana have used traffic impact dashboards from vehicle probe data to assess mobility and queues during an unplanned 37-mile-long closure of an interstate [16].
State-of-the-practice Automated Traffic Signal Performance Measures (ATSPMs) utilize controller high-resolution data to provide insight on the efficiency of traffic signals [17] [18] [19]. However, this technology provides information on an intersection-by-intersection basis, relies on communication and sensing equipment, and requires significant initial capital investment [17].  [29] have also been calculated. Further, critical split failures and downstream blockage have also been derived from trajectory data [25] [26].
Even though traffic signal performance measures derived from CV trajectory data provide accurate results, do not depend on vehicle sending equipment, and improve scalability in comparison with ATSPMs, no studies have used this recently available dataset to assess the impact of long-term work zone diversions on local arterials.

Motivation
The motivation of this study is to demonstrate that current CV trajectory data can be integrated into real-time dashboards to assess the impact work zone diversion has on local streets. This is demonstrated using a case study based upon a 13-intersection segment impacted by a long-term closure of the I-65/70 interchange in Indianapolis. The case study performs a longitudinal analysis assessment of the changes in:  Volumes;  Split failures;  Downstream blockage;  Arrivals on green;  And travel time.

Study Contribution
This study's main contribution is a framework which utilizes techniques that:  Only use CV trajectory data to assess the effects of diversions on local arterials. This independence from infrastructure-based monitoring equipment makes the techniques very scalable for any agency that wants to assess in real-time the effects of diversions on local arterials;  Allow practitioners to identify locations that are under-performing;  Provide insight on the type of problem being experienced (saturation and/or coordination), which aids in the identification of potential solutions.

Study Location
The I-65/70 interchange located in downtown Indianapolis (Figure 1), also known as North Split, was closed on May 15th, 2021, and is planned to remain closed for another 18 months. The North Split usually serves approximately 214,000 vehicles per day. As this volume of vehicles utilizes local streets as detour, the overall network performance gets degraded.
Thirteen of the most affected intersections are studied. They are all located on West St, a parallel arterial to the North Split ( Figure 1). Their names and allowed mainline direction of travel (southbound: SB, and northbound: NB) are shown in Table 1. It is important to mention that, as an outlier, intersection number 7 (West St @ Robert D. Orr Plaza) has a constant green light for vehicles traveling SB through.

Data Description
Private sector CV trajectory data for the weekdays between May 3rd and July 16th, 2021was used in this study. The CV trajectory data consists of individual vehicle waypoints with a reporting interval of 3 seconds and a positional accuracy of a 1.5-meter radius. Every waypoint has the following attributes: GPS lo-  Figure 1 were extracted from the Indiana dataset for further analysis.

Indiana Traffic Signal Performance Monitoring
In early 2020, a traffic signal performance measure monitoring program was im- The details of that automated movement assignment are described in [30].

Interstate Diversion Impact
The following points are covered in this section:  In summary, all the presented performance measures worsened after the North Split closure, specially from 15:00 to the 18:00 hrs. As this period seems the most critical, further analysis will focus on that time-range.

Volumes
As the 214,000 vehicles that used the North Split on a daily basis have to travel using alternative routes, a significant increase in the studied location's volumes is expected. Figure 5 shows

Corridor-Wide Trajectories and Performance Measures by Intersection
To better illustrate the operational dynamics at the studied intersections, trajectories of vehicles traveling SB through are plotted in Figure 6(a) (week before the beginning of the closure) and Figure 6(b) (week after the beginning of the closure). Next to the trajectories, downstream blockage, split failures, and arrivals on green results are displayed. The performance measures are placed in such a way that they match the segment of the trajectories which they represent (AOG and SF for the upstream section, and DSB for the downstream section). From performing a before and after qualitative comparison, the following points can be stated:  By comparing Figure 6(a) and Figure 6(b), not only the increase in volume is noticeable, but also the increment in the number of stops and longer time required to transverse the corridor.  By comparing Figure 6(c) and Figure 6(d), significant increments on downstream blockage occurred from intersections 3 to 7. However, this critical problem seems to abruptly end after intersection 8. This suggests that the downstream blockage identified at upstream locations may be a consequence of intersection 8 having queue spillback. If that is the case, by fixing the congestion at intersection 8, the state of operation at the upstream locations could be improved as well.  By comparing Figure 6(e) and Figure 6(f), it is clear that an important increase in split failures at intersections 2 and 4 occurred. However, as intersection 4 also showed significant downstream blockage, this is not necessarily an indication that the location is operating at overcapacity, but there is a possibility that its split failures are a result of downstream queue spillback.

Travel Times
A valuable, and commonly used metric to assess the performance of a corridor is

Results
Corridor-wide weighted average DSB, SF, and AOG results are shown in Figure  9. As expected, most performance measures worsened significantly after the North Split closure, specifically for the SB direction of travel. The maximum changes in performance for before and after the closure are 10% increase in DSB, 3% increase in SF, and a 16% decrease in AOG (all for the SB direction). Results for the corridor travel time are presented in Figure 10. For the SB direction of travel, the Interquartile Range (IQR) increased by up to 140%, and the median travel time rose 74% when comparing before and after the closure. For the NB direction of travel, IQR increased by up to 83%, and the median travel time rose 19%.

Discussion and Conclusions
This study estimated traffic signal performance measures from CV trajectory data with a 3-second reporting interval to assess the effects that Indianapolis' North Split closure on May 15th, 2021, had on a 13-intersection segment of West St., an alternative route. From the over 130,000 trajectories analyzed during an 11-week period from 15:00-18:00 hrs, the following results were observed ( Figure 5, Figure 9, and Figure 10). Journal of Transportation Technologies  Specific operational failure modes that contributed to this increase in travel time include:  3% increase on split failures, indicating an increment of traffic signals operating at overcapacity;  10% increase on downstream blockage, indicating there was queue spillback from downstream traffic signals;  16% decrease on arrivals on green, indicating there were opportunities to improve traffic signal coordination. From Figure 6, intersection 8 (Washington St) was identified as a location that influences the operational state at upstream intersections. This is an example of how agencies can use these dashboards to identify critical intersections that affect the entire network.
The calculated performance measures can be applied to any location in the world where connected vehicle trajectory data is available. Additionally, the trajectory movement identification method utilized to calculate performance measures does not require road geofencing, which enhances scalability. With this new approach, agencies could analyze hundreds of traffic signals in a time and cost-effective manner without the need for any vehicle sensing equipment.