Data Acquisition and Analysis

Data collection on the VDOT HfL project consisted of acquiring and comparing data on safety, construction congestion, quality, and user satisfaction before, during, and after construction. The primary objective of acquiring these types of data was to provide HfL with sufficient performance information to support the feasibility of the proposed innovations and to demonstrate that ABC technologies can be used to do the following:

• Achieve a safer work zone environment for the traveling public and workers.
• Reduce construction time and minimize traffic interruptions.
• Produce greater user satisfaction

This section discusses how well the VDOT project met the specific HfL performance goals related to these areas.

Safety

If traditional construction methods had been used, the project would have impacted highway users for an estimated 100 days and nights. The revised construction scheme used to build the project reduced this impact significantly, reducing crash potential and improving safety during construction. Constructing prefabricated modular segments offsite also enhanced worker safety because they were not working adjacent to traffic.

The safety goals for the project included both worker safety and motorist measures. The worker safety goal was an incident rate of 4.0 or less as measured by the OSHA 300. The motorist goal during construction was a crash rate equal to or less than the preconstruction crash rate. No worker injuries occurred during the project. Six incidents involving motorists with flat tires occurred during construction because of a temporary patch on the abutment backwall, but no injuries or other vehicle damages occurred. Therefore, the safety goals were met on the project.

The replaced superstructure improves the safety of the existing bridge by adding 8 ft (2.4 m) to the face–to–face curb dimensions to increase the width of the substandard shoulders. The exterior shoulder was widened from 2 ft (0.6 m) to 8 ft (2.4 m) and the median shoulder from 2 ft (0.6 m) to 4 ft (1.2 m). The substandard railings on the bridge were replaced with crash–tested Kansas Corral railings attached to the prefabricated modular segments.

Contruction Congestion

Introduction

The project's primary congestion goal was to reduce construction impact on motorists by 50 percent compared to conventional construction methods. Using conventional cast–in–place construction would have impacted motorists for an estimated 100 days and caused diurnal nonrecoverable queuing problems of about 1.5 to 2 mi (2.4 to 3.2 km) during rush–hour peaks. VDOT employed innovative A + (B x C) bidding to incentivize and accelerate the construction. The use of prefabricated superstructure elements reduced the construction impact to only three weekends.

The three weekends of full road closures of the southbound lanes were conducted on U.S. 15/29 at Buckland in fall 2008. During each weekend closure, officials diverted traffic onto two primary defined detours. Local traffic southbound was directed to leave the highway at the U.S. 29–U.S. 15 intersection, travel north on U.S. 15 to SR 55 (John Marshall Highway), turn westbound onto SR 55 to Beverly's Mill Road, turn south on Beverly's Mill Road, and rejoin U.S. 15/29. The northbound local detour was the opposite. Nonlocal traffic (including trucks) traveled the southbound detour by following Interstate 66 westbound to U.S. 17 and turning south on U.S. 17 to rejoin U.S. 15/29 (the northbound nonlocal detour was the opposite of this). Figure 12 illustrates the defined detours.

The implications of this traffic management approach on travel times needed to be quantified, particularly to compare conventional construction techniques to the three–weekend closure scheme. Therefore, travel time studies were performed in August and September 2008. One set of studies was conducted on a weekend when the highway was closed and all traffic was diverted to the detour. Another set was conducted on weekend days when the highway was open to traffic. The results of these studies showed increases in travel times of up to 14 minutes and 7 minutes for local and through traffic, respectively, traveling in the southbound direction. This produced a calculated total delay for the three weekend closures of 9,461 vehicle–hours. This compares favorably to the estimated total delay of 720,000 vehicle–hours that would have resulted from conventional construction over a 100–day (daytime and nighttime) construction period.

Data Collection

Researchers used the floating vehicle methodology to collect travel times, attempting to mimic the typical driving speed of other vehicles along the various roadway segments. Data were collected only during daytime hours, since at night traffic demands would be lower and any effects of the total roadway closure would be smaller. Researchers collected data during a scheduled full roadway closure on August 23 and 24, 2008 (Saturday and Sunday). Researchers returned to the site on two subsequent weekends (September 13 and September 21) to obtain Saturday and Sunday travel times when U.S. 15/29 was not closed. A continuous 90–mi (144.8–km) loop on U.S. 15/29 and detour routes was designed, and data collection personnel traveled it repeatedly on each day of data collection. Four circuits were completed on each day of data collection for the normal (nonclosure) weekend for a total of eight circuits. Three circuits were completed for each day of data collection during the total roadway closure weekend for a total of six circuits. After discussions with the data collection crew and review of project diary information, it was determined that the data collected on August 23 occurred while U.S. 15/29 remained open. Therefore, only the data collected on August 24 was used to estimate the effects of the full roadway closure.

The data collection circuit was identified by a series of 21 different nodes where interim travel time readings were taken. Table 1 identifies each node followed in sequence during each data collection circuit, as well as the approximate travel distance to that node from the previous node. The use of multiple data collection nodes allowed researchers to directly compare individual roadway segments (with and without a full roadway closure in place) and to combine those segments in various ways to quantify the impacts on the entire length of the two detour routes.

Table 1. Node definitions for travel time data collection circuit.

The travel time data were entered on a spreadsheet for reduction and analysis. Data from each travel time circuit were combined in an overall average for each study period and compared. Tables 2 and 3 summarize these data. A slight increase (7 percent) in the total travel times on the second day of the non–full roadway closure travel time study was noted. Data collection personnel indicated there was a strong police presence on the corridor and alternative routes during the second day of the full roadway closure, which may have reduced speeds and caused this slight increase. Consequently, an average travel time across both days was used as a conservative estimate of the effect of the closure.

Table 2. Travel time data in before (no full roadway closure) condition.

Table 3. Travel time data during full roadway closure conditions.

Travel Time Comparison Results

Differences in Operating Conditions on Detour Route Segments

Table 4 summarizes a segment–by–segment comparison of the average travel times when the full roadway closure was in place on U.S. 15/29 (i.e., during) to when the full closure was not in place (i.e., before). Simple t–tests were performed to determine statistical significance of any differences. As table 2 indicates, the full roadway closure had significant effects on several detour route segments for southbound U.S. 15/29 traffic in the corridor:

• U.S. 15 northbound from U.S. 29 to SR 55—a 225–second travel time increase (107 percent)
• SR 55 westbound from U.S. 15 to Beverly's Mill Road—a 36–second increase (14 percent)
• Beverly's Mill Road southbound from SR 55 to U.S. 15/29—a 36–second increase (11 percent)
• I–66 westbound from the U.S. 15 overpass to the U.S. 17 interchange—a 119–second increase (19 percent)
• U.S. 17 southbound from I–66 to U.S. 211/29—a 53 second increase (8 percent)

Table 4. Comparison of segment travel times, before versus during roadway closure.

*Signicicantly Different (a=0.05).

Although there were some travel time increases on detour routes serving the northbound U.S. 15/29 traffic, these were not substantial enough in most cases to be detected as statistically significant. There was one exception: SR 55 eastbound from Beverly's Mill Rd to U.S. 15—a 59–second increase (23 percent).

Meanwhile, one segment on U.S. 15/29 northbound between U.S. 17 and Beverly's Mill Road actually experienced a 61–second (21 percent) decrease.

Effect of Full Roadway Closure on Route Travel Times

Table 5 summarizes the increase in total travel times caused by the total roadway closure of U.S. 15/29. VDOT defined two types of detour routes, one for local traffic attempting to travel on the U.S. 15/29 segment affected by the closure and another for through traffic (including trucks) using the highway. For both routes, total travel times that would have been possible by using U.S. 15/29 were compared to the defined alternative route.

The local detour route, because of its length compared to the U.S. 15/29 segment, increased travel times by 848 seconds (14 minutes) per diverted trip southbound and by 734 seconds (12 minutes) northbound. For through traffic, the effect of the detour was much less pronounced. Through traffic in the southbound direction that followed the detour along I–66 and U.S. 17 rather than using U.S. 15/29 experienced travel time increases of 398 seconds (nearly 7 minutes). In the northbound direction, the through traffic detour travel time was 186 seconds (slightly more than 3 minutes) longer than if the U.S. 15/29 highway had been available for use. All of the increases shown in table 5 were highly significant.

Table 5. Comparison of normal (before) travel times to detour travel times during full roadway closure.

*Signicicantly Different (a=0.05)

Quantification of Total Delays Generated by the Full Roadway Closure

As expected, the travel time studies indicated that traffic diverted from U.S. 15/29 experienced significant increases in travel time. In addition, this diverted traffic caused additional congestion on certain segments of the detour routes and increased travel times for those drivers normally using those segments. Full quantification of the total delays caused by the full roadway closure requires estimating the amount of diverted traffic from U.S. 15/29 that used either the local diversion route or the through diversion route.

Researchers were able to obtain traffic count data from several VDOT sensors in the study corridor:

• U.S. 15/29 between Beverly's Mill Road and Vint Hill Road
• U.S. 17 just south of I–66
• I–66 just east of U.S. 17

Together, these three sensor locations allowed for a detailed analysis of diversion behaviors in the corridor. The U.S. 15/29 sensor location was located in the defined local detour segment. As such, it provided an indication of the amount of local traffic not diverting at all (i.e., those with local destinations not affected by the total road closure). Meanwhile, the I–66 and U.S. 17 sensor locations allowed for an assessment of traffic choosing to use the through or truck detour defined above.

The total roadway closures were performed on three nonconsecutive weekends:

• 9 p.m. on Friday, August 15, through 10 p.m. on Saturday, August 16
• 7 p.m. on Saturday, August 23, through 12:05 p.m. on Sunday, August 24
• 7 p.m. on Saturday, September 6, through 2:45 p.m. on Sunday, September 7

Table 6 summarizes the comparison of hourly traffic counts occurring on the other nonholiday weekends in August and September to the hourly counts during these total roadway closure hours. Northbound counts (and I–66 eastbound counts) were generally unaffected by the total roadway closure on U.S. 15/29. It appears that most drivers did not divert at the Beverly's Mill Road detour, but continued on U.S. 15/29 to a destination within the segment before the bridge closure or onto Vint Hill Road (SR 215, which serves subdivisions in the area).

Table 6. Traffic count comparison during total roadway closure.

The situation was markedly different in the southbound direction on U.S. 15/29 (and westbound on I–66). Only a small portion of traffic normally using the route in the southbound direction in the vicinity of the sensors actually did so during the hours of the total closure. These vehicles presumably came from Vint Hill Road or from businesses on U.S. 15/29 south of the total bridge closure point. Some of the vehicles (between 5,598 and 8,368) that did not travel down U.S. 15/29 showed up in higher I–66 and U.S. 17 counts. The remainder of the diverted traffic presumably used the local detour route. Applying the increased travel time estimates shown in tables 4 and 5 to these traffic counts yields an overall estimate of the delays generated by the closures.

Through traffic southbound (delays from table 5):

(5,598 + 8,368)/2 * (398/3,600) = 6,983* 0.111 = 772 vehicle–hours

Local traffic southbound (delays from table 5):

(37,316 – 6983)*(848/3,600) = 30,333*0.236 = 7,145 vehicle–hours

Additional delays incurred by other westbound I–66 drivers (delays from table 4):

37,767*(119/3,600) = 1,248 vehicle–hours

Additional delays incurred by other southbound U.S. 17 traffic (delays from table 4):

20,117* (53/3,600) = 296 vehicle–hours

TOTAL = 772 + 7,145 + 1,248 + 296 = 9,461 vehicle–hours

Comparison to Original Phasing Scheme

Since the construction phasing scheme was changed after the project started from twelve weeknight closures to full closures over three weekends, it is prudent to compare the traffic delay impact between these two schemes.

Using VDOT traffic count data from U.S. 29, it is estimated that (0.89*6572=) 5,489 vehicles southbound and (0.055*4748=) 261 vehicles northbound would be impacted during each weeknight of full roadway closure. Assuming the same estimates of increased travel time to this U.S. 29 traffic because of northbound and southbound detours, it is anticipated that the use of weeknight full roadway closures would have yielded 1,346.1 vehicle–hours of delay each night. Compared to the three weekend full roadway closures that were ultimately used which generated a total of 9,461 vehicle–hours of delay, it would have taken only about seven of the scheduled twelve weeknights of closures to reach the same amount of total delay that occurred over the three weekend full closures. Perhaps more important, if the nighttime full roadway closure had not been removed each morning by 5 a.m. (because of construction problems), there was a potential for gridlock to develop because peak period (5 to 9 a.m.) traffic volumes on U.S. 29 exceed 9,000 vehicles northbound and 3,500 vehicles southbound. These volumes alone exceed those occurring over the entire nighttime work period. Although not specifically estimated in this analysis, the queuing that likely would have developed on the detour routes would have escalated exponentially, generating more delay in one peak period alone than occurred during the entire series of weekend closures used in this project. Therefore, the modified construction work scheme not only reduced the estimated traffic impact by almost one–half, it also greatly reduced the risk of traffic impact during critical peak–flow periods.

Quality

The project goals for quality were to improve the ride quality, reduce the noise, and provide a new and durable bridge. The existing deteriorated bridge deck required continuous maintenance and patching of potholes, creating frequent traffic congestion. The rough deck surface created a safety hazard and wear and tear on vehicles. The new superstructure provides a durable, maintenance–free bridge deck by using lightweight, high–performance concrete materials and corrosion–resistant reinforcing steel, all prefabricated in a controlled manufacturing environment. The durability is further improved by the application of a waterproofing membrane and hot–mix asphalt overlay. The asphalt overlay and membrane on the bridge is extended over the multiple pier joints with a special detail to deter reflective cracking. This also reduces the noise intensity compared to riding on a bridge deck with multiple expansion joints. A slab extension over the backwalls at the abutments eliminates the joints at the abutments, further minimizing noise, improving ride quality, and reducing water infiltration at the backwall.

Sound Intensity Testing

Preconstruction noise testing was done on July 10, 2007. Noise data was not collected after construction because of the short length of the bridge. Onboard sound intensity (OBSI) measurements were obtained from the bridges at the posted highway speed of 45 miles per hour (mi/h) (72.4 kilometers per hour (km/h)).

Sound intensity (SI) measurements were made using the current accepted onboard sound intensity (OBSI) technique AASHTO TP 76–08, which includes dual vertical sound intensity probes and an ASTM standard reference test tire (SRTT). The sound measurements were recorded using the Bruel and Kjaer PULSE software and data collection system. Three runs were made in the right wheelpath of the outer lane of each bridge. The two microphone probes simultaneously captured noise data from the leading and trailing tire–pavement contact areas. Figure 13 shows the dual probe instrumentation and the tread pattern of the SRTT.

Figure 13. OBSI dual probe system and the SRTT.

The average of the front and rear SI values was computed with the Bruel & Kjaer PULSE software to analyze the raw data signals over the full length of the bridge decks to produce sound intensity values. Raw noise data were normalized for the ambient air temperature and barometric pressure at the time of testing. The resulting mean sound intensity levels are A–weighted to produce the noise–frequency spectra in one–third octave bands, as shown in figures 14 and 15.

Global noise levels were calculated by using logarithmic addition of the third octave band frequencies between 315 and 4,000 hertz (Hz). The global noise levels were 100.6 and 98.6 dB(A) for the northbound and southbound prerehabilitated bridges, respectively. The original portland cement concrete bridge deck surfaces were distressed and weathered and likely to be noisier than the newly constructed surfaces. Noise values from the two bridges are relatively similar in frequencies and overall decibel levels. For reference, a 3.0–decibel difference in noise is considered noticeable. Postconstruction noise levels were not measured for this project.

Figure 14. Northbound bridge mean A–weighted sound intensity frequency spectra.

Figure 15. Southbound bridge mean A–weighted sound intensity frequency spectra.

Smoothness Mearsurement

Smoothness testing was done in conjunction with noise testing using a laser profiler manufactured by International Cybernetics Corporation built in to the noise test vehicle. Figure 16 shows the test vehicle with the laser positioned in line with the right rear wheel.

Three test runs in each wheelpath in each direction were conducted. The left and right wheelpath test runs were averaged to produce a singe International Roughness Index (IRI) value with units of inches per mile (in/mi). Resulting IRI values are plotted in figure 17 at 10–ft (3–m) intervals.

Figure 16. Laser profiler mounted behind the test vehicle.

Figure 17. Mean IRI values.

The overall IRI values are 161 and 215 in/mi for the prerehabilitated northbound and southbound bridges, respectively. Postconstruction IRI is anticipated to be lower because a new hot–mix asphalt concrete overlay will be applied. However, postconstruction ride quality was not measured. Figure 17 shows large peak values at the ends of each bridge corresponding to pavement distress near the expansion joints. Figure 18 shows the existing southbound bridge deck.

Figure 18. Prerehabilitated southbound bridge deck.

User Satisfaction

The Buckland Preservation Society's interests were considered and incorporated into the project. A memorandum of agreement signed between the Section 106 consulting parties and VDOT specifically required the use of the accelerated nighttime construction method for the superstructure replacement, as discussed previously in this report. In addition, VDOT sought to provide the best long–term solution for this bridge crossing. VDOT did not send survey letters as part of this process. As stated previously, significant work was done with the residents of the Buckland Historic District during the design phase. The following quote was sent to VDOT from a representative of that group:

"On behalf of the Buckland Preservation Society and neighbors at Buckland . . . Congratulations . . . Excellent Work! . . . .Very Special Thanks to Nick Roper for all his initial efforts!"

Public officials at the ribbon cutting ceremony (see agenda in Appendix B) held on October 14, 2008 highlighted the positive benefits that the rehabilitated bridge would have on their community and also VDOT's outreach efforts to the public during the bridge construction phase to keep them informed of the projects progress. The citizens at the ceremony expressed their approval of the new bridge and how it was constructed to minimize impact on the users. The project's Partnering Charter had two public relations objectives. The first objective was to cultivate and open and honest relationship with the public to effectively communicate with the public about the purpose of the project and project progress. The second objective was to respectfully respond to the public's inquires, comments, and concerns.