Guide to Establishing Monitoring Programs for Travel Time Reliability (2014) / Chapter Skim
Currently Skimming:
Pages 211-500
The Chapter Skim interface presents what we've algorithmically identified as the most significant single chunk of text within every page in the chapter.
Select key terms on the right to highlight them within pages of the chapter.
From page 211... ...
209 C CASE STUDIES INTRODUCTION This appendix contains fi ve case studies that demonstrate approaches to the travel time reliability monitoring techniques described in the Guide. The case studies illustrate real-world examples of using a travel time reliability monitoring system (TTRMS)
|
From page 212... ...
210 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY Figure C.2. Case study locations.
|
From page 213... ...
211 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY Case Study 1 SAN DIEGO, CALIFORNIA This case study focused on using a mature reliability monitoring system in San Diego, California, to illustrate the state of the art for existing practice. Led by its metro politan planning organization, the San Diego Association of Governments (SANDAG)
|
From page 214... ...
212 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY Methodology is the most experimental and least site specific. It is dedicated to an ongoing investigation, spread across all five case studies, to test, refine, and implement the Bayesian travel time reliability calculation methodology outlined in Chapter 3.
|
From page 215... ...
213 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY Over the past several years, transportation agencies operating within the San Diego region, through partnerships between SANDAG, Caltrans, local jurisdictions, transit agencies, and emergency responders, have been updating and integrating their traffic management systems, as well as developing new systems, under the concept of integrated corridor management (ICM)
|
From page 216... ...
214 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY To monitor its freeways, District 11 has 3,592 intelligent transportations system traffic sensors deployed at 1,210 locations that collect and transmit data in real time to a central database. Of these, 2,558 sensors are in the freeway mainline lanes, 20 are in HOV lanes, and the rest are located at on-ramps, off-ramps, or interchanges.
|
From page 217... ...
215 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY TABLE C.1. SAN DIEGO COUNTY FREEWAY DETECTION Freeway No.
|
From page 218... ...
216 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY routines on the data, including detector diagnostics, imputation, speed calculations, performance measure computations, and aggregation. These processing steps have been fully described in Chapter 3 of the Guide.
|
From page 219... ...
217 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY granularity. It reports performance measures such as speed, delay, percentage of time spent in congestion, travel time, and travel time reliability.
|
From page 220... ...
218 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY information is parsed from time-of-day signal-timing plans. A-PeMS can also integrate real-time signal-timing cycle lengths and phase green times from traffic signal controllers.
|
From page 221... ...
219 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY loop actuations allows the generation of data sets that are not found elsewhere, such as vehicle stream data, which can be used for headway studies, gap analysis, and merging studies. The BHL loops also provide individual vehicle length measurements, allowing for the classification of freeway traffic.
|
From page 222... ...
220 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY Methodological Use Cases Overview Five concepts are important in this analysis: 1. Regardless of the data source, the methodology must always generate a full TTPDF.
|
From page 223... ...
221 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY distribution between an origin and a destination on a given route over a full year. Figure C.5 illustrates an example that uses four PDFs and a transition PDF (labeled T)
|
From page 224... ...
222 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY It is further hypothesized that the individual auto travel times on links or routes can be characterized by a three-parameter (α, β, and δ) shifted gamma distribution as shown by Equation C.1: g t t e tfor , o.e.w.t, , 1β α δ δ( )
|
From page 225... ...
223 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY 1. User enters origin, destination, and a DAT of 8:40 a.m.
|
From page 226... ...
224 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY 3. The system identifies the relevant time-dependent PDF (the morning peak)
|
From page 227... ...
225 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY Use Case MC2: User Wants to Know Immediately What Route to Take and What Time to Leave for a Trip to Arrive on Time at a Destination This use case is different and more challenging to demonstrate than MC1 or MC3. It also represents the application with the highest utility from the driver's perspective because it will provide real-time information on the recommended trip start time, including the effects of incidents or other events not explicitly accounted for in historical TT-PDFs.
|
From page 228... ...
226 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY • In a real-time context, when computational speed is of the essence, the number of PDFs to be considered should be kept to a minimum. Each link or route could theoretically be considered to operate in four regimes: uncongested, transition from uncongested to congested, congested, and transition from congested to uncongested.
|
From page 229... ...
227 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY Analysis of Bluetooth Travel Times To support the methodologies presented in Use Cases MC1, MC2, and MC3, Bluetooth data from the BHL were analyzed to see what could be learned about individual vehicle travel times and their PDFs. The raw data were filtered to remove MAC addresses with six or more time stamps on either reader.
|
From page 230... ...
228 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY The resulting distribution of the Bluetooth data regime classifications is shown in Table C.3.
|
From page 231... ...
229 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY TABLE C.4. GOODNESS-OF-FIT RESULTS FOR FREE-FLOW REGIME Distribution Kolmogorov– Smirnov Anderson– Darling Chi-Squared Statistic Rank Statistic Rank Statistic Rank Pearson 5(3P)
|
From page 232... ...
230 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY TABLE C.5. GOODNESS-OF-FIT RESULTS FOR TRANSITION REGIME Distribution Kolmogorov– Smirnov Anderson– Darling Chi-Squared Statistic Rank Statistic Rank Statistic Rank Burr 0.02676 1 0.81555 1 16.645 1 Burr (4P)
|
From page 233... ...
231 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY TABLE C.6. GOODNESS-OF-FIT RESULTS FOR CONGESTED REGIME Distribution Kolmogorov– Smirnov Anderson– Darling Chi-Squared Statistic Rank Statistic Rank Statistic Rank Fatigue life (3P)
|
From page 234... ...
232 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY Conclusions This analysis examined BHL data to see if operative regimes for individual vehicle travel times can be identified from Bluetooth data. The research team concluded that this can, indeed, be done.
|
From page 235... ...
233 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY USE CASE ANALYSIS Overview Chapter 4 of the Guide and Appendix D: Use Case Analyses present dozens of use cases intended to satisfy the myriad ways that different classes of users can derive value from a reliability monitoring system. For the San Diego case study, various use cases were combined to form six high-level use cases that broadly encompass the types of reliability information that users are most interested in and that were suited for validation using the San Diego data sources.
|
From page 236... ...
234 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY methods were developed to create TT-PDFs from large data sets of travel times that occurred under each event condition. From these PDFs, summary metrics such as the median travel time and planning travel time were computed to show the variability impacts of each event condition.
|
From page 237... ...
235 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY travel time variability were investigated for weekdays between the months of November and February (when Qualcomm Stadium regularly hosts events and when San Diego experiences the most inclement weather)
|
From page 238... ...
236 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY To link travel times with the source condition active during their measurement, each 5-minute travel time was tagged with one of the following sources: baseline, incident, weather, special event, lane closure, or high demand. A travel time reliability monitoring system (TTRMS)
|
From page 239... ...
237 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY During a few time periods within each data set more than one factor was active during a single 5-minute period. In these cases, the travel time was tagged with the factor that was deemed to have the larger travel time impact (e.g., when an incident coincided with light precipitation, the travel time was tagged with incident)
|
From page 240... ...
238 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY Interestingly, it is apparent from this graph that sometimes, even when an incident is active, the travel time falls below 10 minutes. This is likely due to the fact that this analysis does not account for the severity of incidents in the travel time tagging process.
|
From page 241... ...
239 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY on the corridor, the distribution of travel times is much wider. An incident increases the median travel time on the facility by 2 minutes over the baseline condition and, with a 95th percentile travel time of 18.7 minutes, requires travelers to add a buffer time of 9.8 minutes, almost doubling their typical commute, to arrive on time during an incident.
|
From page 242... ...
240 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY TABLE C.9. WEEKDAY MORNING PEAK TRAVEL TIME VARIABILITY CAUSALITY FOR WESTBOUND I-8 Source Active When Travel Time Exceeded 85th Percentile 95th Percentile Baseline 37.7% 3.3% Incident 31.2% 41.1% Weather 30.6% 55.6% The conclusions that can be made from the morning time period analysis are that weather almost always slows down travel times significantly.
|
From page 243... ...
241 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY during the data tagging process. The 16.2-minute 95th percentile travel time likely represents the short time period when the majority of people are trying to access the special event venue, and the faster special event travel times are likely from the periods before the events start, when attendees are just beginning to trickle in.
|
From page 244... ...
242 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY Table C.10 summarizes the contribution of each source condition to travel times exceeding the 85th percentile (8.9 minutes) and the 95th percentile (9.2 minutes)
|
From page 245... ...
243 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY TABLE C.10. TRAVEL TIME VARIABILITY CAUSALITY DURING AFTERNOON PEAK FOR WESTBOUND I-8 Source Active When Travel Time Exceeded 85th Percentile 95th Percentile Baseline 59.7% 15.2% Incident 13.4% 29.8% Weather 22.4% 50% Special event 3.6% 4.5% High demand 1.0% 0.6% Synthesis.
|
From page 246... ...
244 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY between the median and 95th percentile travel times. The travel times exceeding the 95th percentile predominantly occurred under incident and weather conditions.
|
From page 247... ...
245 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY Figure C.24. Weekend morning TT-PDFs for northbound I-5.
|
From page 248... ...
246 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY TABLE C.11. WEEKEND MORNING TRAVEL TIME VARIABILITY CAUSALITY FOR NORTHBOUND I-5 Source Active When Travel Time Exceeded 85th Percentile 95th Percentile Baseline 79.5% 64.2% Incident 11.3% 20.9% Weather 0.4% 0.1% High demand 6.5% 8.8% Afternoon Peak.
|
From page 249... ...
247 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY Figure C.25. Weekend afternoon distribution of travel times for northbound I-5.
|
From page 250... ...
248 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY TABLE C.12. WEEKEND AFTERNOON TRAVEL TIME VARIABILITY CAUSALITY FOR NORTHBOUND I-5 Source Active When Travel Time Exceeded 85th Percentile 95th Percentile Baseline 51.4% 20.2% Incidents 29.1% 48.2% Weather 0.0% 0.0% Special events 8.8% 25.3% High demand 10.8% 6.3% Figure C.26.
|
From page 251... ...
249 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY Synthesis. The morning weekend travel time variability on the corridor is very minor, leaving little room for improvement from planning or operational interventions.
|
From page 252... ...
250 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY for which the traveler must arrive at the destination at or before a specified time (such as a typical morning commute to work)
|
From page 253... ...
251 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY Figure C.27 Freeway Use Case 2 alternate routes. Map data © 2012 Google.
|
From page 254... ...
252 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY days of the week, as well as show the similarity in operating conditions across certain hours of the day. Regime assignment is addressed in the discussion of methodological advances in this case study, and the team is further refining its regime assignment methodologies.
|
From page 255... ...
253 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY 5-minute period of each day-of-the-week type was assigned to a regime based on average TTI during that time period. Average TTIs for each time period were calculated using 6 months of 5-minute travel time data (excluding holidays)
|
From page 256... ...
254 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY H ou r 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 Day of Week M Tu W Th F Sa Su Fi g u re C .2 9 .
|
From page 257... ...
255 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY H ou r 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 Day of Week M Tu W Th F Sa Su Fi g u re C .3 1 .
|
From page 258... ...
256 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY PDF Generation. Although regime assignments are made offline, this validation assumes that the regime-based PDFs are assembled in real time in response to a user's request for information.
|
From page 259... ...
257 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY and travel times upward of 30 minutes can occur. The most frequently occurring travel time on Route 3 is approximately 24 minutes, and the route has significant travel time variability on Fridays.
|
From page 260... ...
258 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY Diego validation, the route that is the fastest on average is not always the route that consistently gets travelers to their destination on time. Providing buffer time measures for alternate routes conveys this message to the end user, ultimately giving them more confidence in the ability of the transportation system to get them to their destination on time.
|
From page 261... ...
259 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY approach is currently used in PeMS to predict travel times along a route for the rest of the day (1, 2)
|
From page 262... ...
260 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY The distance dh between the current day travel time and the historical day travel time is calculated using Equation C.2: d x w x T x T xh h c 2( )
|
From page 263... ...
261 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY Results The travel time prediction methodology was used to compute predictive travel time ranges for the two example alternate routes between 5:35 and 5:45 p.m. on Thursday, August 12, 2010.
|
From page 265... ...
263 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY TABLE C.18. PREDICTED VERSUS ACTUAL TRAVEL TIMES FOR AUGUST 12, 2010, FOR I-15 Travel Time Measurement 5:35 p.m.
|
From page 266... ...
264 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY 20 21 22 23 24 25 26 27 28 5:05 PM 5:10 PM 5:15 PM 5:20 PM 5:25 PM 5:30 PM 5:35 PM 5:40 PM 5:45 PM Tr av el T im e (m in s)
|
From page 267... ...
265 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY TABLE C.19. PREDICTED VERSUS ACTUAL TRAVEL TIMES FOR AUGUST 12, 2010, FOR I-5 Travel Time Measurements 5:35 p.m.
|
From page 268... ...
266 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY Transit Use Case 1: Conducting Offline Analysis on the Relationship Between Travel Time Variability and the Seven Sources of Congestion Summary This use case aims to quantify the impacts on travel time variability for transit trips of the seven sources of congestion: incidents, weather, work zones, fluctuations in demand, special events, traffic control devices, and inadequate base capacity. To perform this analysis, methods were developed to extract travel times from APC bus data.
|
From page 269... ...
267 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY The third route analyzed was Route 50 southbound, which travels along I-5 into downtown San Diego. This route begins near the Clairemont Drive on-ramp to I-5, continues south along I-5 for 6.4 miles, and ends 0.8 miles later at 10th Avenue and Broadway.
|
From page 270... ...
268 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY For every weekday run for which data were available on each of the three routes described above, APC data were analyzed to determine the in-vehicle travel time from delivered service records. Passenger loadings were also extracted from the APC data.
|
From page 271... ...
269 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY Over the period of study, 129 transit trips were made on this route. Of these 129 trips, seven special event, two incident, and 16 high-demand trips were recorded.
|
From page 272... ...
270 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY Figure C.41 shows the travel time distribution over the month for midday trips. Travel times for the midday period, in contrast to those seen in the morning peak, appear clustered around the primary mode of 54.2 minutes, as shown in Figure C.39.
|
From page 273... ...
271 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY TABLE C.20. TRAVEL TIME VARIABILITY CAUSALITY FOR ROUTE 20 Source Active (%)
|
From page 274... ...
272 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY and nine were made by the same driver. Of the 11 days when this smaller travel time was not seen, 10 had no 7:13 a.m.
|
From page 275... ...
273 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY 0 2 4 6 8 10 12 14 16 18 31.02 38.06 45.10 52.14 Travel Time (min) Baseline Special event Incident High demand 0 2 4 6 8 10 12 14 16 18 31.02 38.06 45.10 52.14 Travel Time (min)
|
From page 276... ...
274 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY Table C.21 summarizes the contribution of each event condition to all travel times, to those exceeding the 85th percentile (49.1 minutes) , and to those exceeding the 95th percentile (51.4 minutes)
|
From page 277... ...
275 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY in-vehicle APC sensors as described above. Scheduled travel times for the 158 runs analyzed for this route ranged between 18 and 21 minutes, averaging 19.5 minutes.
|
From page 278... ...
276 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY Similar to the other two routes analyzed here, the midday period, shown in Figure C.49, carried the majority (four of five) of the high-demand trips on this route.
|
From page 279... ...
277 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY Table C.22 summarizes the contribution of each event condition to all travel times, to those exceeding the 85th percentile (35.9 minutes) , and to those exceeding the 95th percentile (37.1 minutes)
|
From page 280... ...
278 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY Conversely, uncertainty can also have a significant effect on the traveler experience. The agony associated with waiting for transit service has been well documented; research suggests that passengers overestimate the time they spend waiting by a factor of two to three compared with in-vehicle time (3)
|
From page 281... ...
279 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY • Door open time; • Door close time; • Number of passengers boarding; • Number of passengers alighting; and • Passenger load. Notably absent from these data is any kind of service pattern designation, which is necessary to group similar trips together for comparison purposes.
|
From page 282... ...
280 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY are missing or do not make physical sense. For example, if a given transit stop has no passengers waiting at it, and no riding passengers have requested a stop there, it is common for the transit vehicle to skip this stop.
|
From page 283... ...
281 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY 1. A section of 8.13 miles of Route 30 northbound (from the Grand Avenue exit on Highway 5 along the coast to the intersection of Torrey Pines Road and La Jolla Shores Drive)
|
From page 284... ...
282 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY TABLE C.23. DEPARTURE TIMES AND TRAVEL TIMES ON ROUTE 30 NORTHBOUND 75th Percentile Departure Time 75th Percentile Travel Time 85th Percentile Departure Time 85th Percentile Travel Time 95th Percentile Departure Time 95th Percentile Travel Time Arrival Time 6:54 a.m.
|
From page 285... ...
283 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY Figure C.53. Marey graph (top)
|
From page 286... ...
284 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY Conclusion The most direct analysis would be achieved by restricting the date range to dates with identical schedules; however, in practice it can be rare to find days with the exact same schedule. Regardless, because for routes with headways smaller than 10 minutes it is common for passengers to arrive at bus stops independently of the schedule, the constant arrival pattern used in this simulation may be more meaningful.
|
From page 287... ...
285 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY Use Case 3: Analyzing the Effects of Transfers on the Travel Time Reliability of Transit Trips Summary The goal of this use case is to demonstrate a methodology for quantifying the effects of missed transfers on travel time (and travel time reliability) for a particular transit trip.
|
From page 288... ...
286 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY deviations) , passengers may be affected, even if the transfer was officially untimed.
|
From page 289... ...
287 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY Methods This section describes the data preparation and trip time methodologies. Data Preparation.
|
From page 290... ...
288 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY The APC data contribute distributions of arrival schedule adherence, departure schedule adherence, and travel times for the relevant buses and stops. In order to construct these distributions accurately, only data from runs that serve both the origin and transfer (or transfer and destination)
|
From page 291... ...
289 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY Several of these distributions correlate with each other, affecting how the samples are drawn in each simulation; Figure C.57 shows an example. On both routes, there was some correlation between Bus A's departure time at the origin, Bus A's travel time, and Bus A's departure time at the transfer point.
|
From page 292... ...
290 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY Figure C.58. Procedure followed to generate travel times.
|
From page 293... ...
291 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY trips on which a connection was missed. Further insight into travel times on this route can be gained by dividing the simulated trips into those that made or missed each bus.
|
From page 294... ...
292 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY On Route B, whether either Bus A or Bus B, or both, were missed or caught was tracked for the purposes of exploring the effects of missed transfers on travel time. Travel time histograms corresponding to each scenario are plotted in Figure C.61 and described in Table C.27.
|
From page 295... ...
293 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY Figure C.61. Travel times when catching and missing buses on Route B
|
From page 297... ...
295 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY more passengers missed the second bus, with each of those passengers experiencing a delay of one Bus B headway. However, the mean travel time in this scenario only increased by 1 minute.
|
From page 298... ...
296 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY In terms of trip characteristics, freight and overall travel have spatial differences, temporal differences, and facility differences. Spatial differences refer to the fact that origins and destinations with the heaviest freight traffic do not necessarily also have the highest overall traffic volumes.
|
From page 299... ...
297 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY Highway Administration (FHWA) described in the following section, but these efforts are still in the research phase and are not feasible for public sector agencies to put into operational practice.
|
From page 300... ...
298 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY Figure C.62. Otay Mesa border crossing district map.
|
From page 301... ...
299 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY Figure C.64. Otay Mesa trips spanning multiple districts.
|
From page 302... ...
300 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY District Reliability. To understand which geographical segments of the border crossing have the most travel time variability, the research team assembled the TT-PDFs for trips within each of the 10 individual districts and for two trips spanning multiple districts.
|
From page 303... ...
301 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY median, standard deviation, and 95th percentile travel times by district in Table C.29. From the plots, the district that notably stands out as having the most travel time variability is District 7 (U.S.
|
From page 304... ...
302 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY TABLE C.29. DISTRICT-BY-DISTRICT TRAVEL TIMES AND VARIABILITY District Median Travel Time (min)
|
From page 305... ...
303 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY Figure C.67. Cross-district travel time PDFs.
|
From page 306... ...
304 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY drivers are resting or because there is less staff available to perform inspections. The peak number of trucks use the FAST lanes at around noon and between 4:00 and 6:00 p.m.
|
From page 307... ...
305 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY 0 100 200 300 400 500 600 0 20 40 60 80 100 120 140 160 Tr uc k C ou nt Tr av el T im e (m in s) Count Average SD Figure C.69.
|
From page 308... ...
306 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY and reliability, managers can begin to take steps to improve operations: for example, adding lanes to capacity-restricted locations or adding staff to checkpoints that affect reliability during peak hours of the day. Extensions of the district-level analysis would group travel times by hour of the day to explain not just where travel time reliability is high, but when it is high, as well.
|
From page 309... ...
307 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY After the identification and removal of these data points, assembling route-based reliability statistics using a drastically reduced subset of good data presented the next challenge. This limited the number of routes that the research team could consider, as not all trips on all routes are made by equipped buses, and trips made by equipped buses contain holes due to erroneous data records.
|
From page 310... ...
308 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY 4. Berkow, M., A
|
From page 311... ...
309 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY Case Study 2 NORTHERN VIRGINIA This case study provides an example of a more traditional transportation data collection network operating in a mixture of urban and suburban environments. Northern Virginia was selected as a case study site because it provided an opportunity to integrate a reliability monitoring system into a preexisting, extensive data collection network.
|
From page 312... ...
310 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY calculation methodology outlined in Chapter 3 and introduced in the San Diego case study will resume as part of the final three case studies. Use cases are less theoretical and more site specific.
|
From page 313... ...
311 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY Traffic operations in the district are overseen from the NOVA Traffic Operations Center (TOC) , which manages more than 100 miles of instrumented roadways, including high-occupancy vehicle (HOV)
|
From page 314... ...
312 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY VDOT's management strategy has undergone a dramatic change in the last few years, transitioning from a two-pronged build–maintain regime to a three-pronged build–operate–maintain scheme. As such, VDOT is evolving into a customer-driven organization with a focus on outcomes and a 24/7 performance orientation.
|
From page 315... ...
313 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY limited-access highways include I-495 (the Capital Beltway) , I-95, I-395, and I-66; the Fairfax County Parkway and Franconia–Springfield Parkway; the George Washington Memorial Parkway; and the Dulles Toll Road.
|
From page 316... ...
314 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY is being deployed for the first time. As a result of a combination of older loop detector station failures, ongoing roadway construction, and the need to configure many of the newer radar-based units, data are available for only about 75 of the detectors.
|
From page 317... ...
315 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY • VSP has used the data gateway to share VSP data since 2004. Data entered into the VSP computer-aided dispatch system are shared in real time with all participating TOCs.
|
From page 318... ...
316 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY data XML documents are the primary traffic data source for NOVA PeMS. When data quality, largely due to recent construction on monitored roadways, proved to be a major issue impeding the study of reliability on the 2011 data, the research team also acquired a database dump of detector data along I-66 and I-395 for the entire year of 2009 from the CATT lab.
|
From page 319... ...
317 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY average 5-minute occupancy for detectors that transmit flow and speed. These data are stored in a 5-minute detector database table.
|
From page 320... ...
318 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY Metadata PeMS needs to acquire two types of metadata before traffic data can be stored in the database: roadway network information and equipment configuration data. To represent the monitored roadway network and draw it on maps, PeMS needs to have geographic information system (GIS)
|
From page 321... ...
319 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY composed of |
From page 322... ...
320 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY Once the file was obtained, the next step was to fit the data into the PeMS configuration framework. The third step was to parse the XML file, insert relevant fields into the PeMS database, and write a program to automatically download the configuration file from the RITIS website and populate relevant information into the database whenever the file updated.
|
From page 323... ...
321 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY operation meant that sensors monitored different directions of travel based on the time of day, which PeMS also could not accurately configure. For this reason, shoulder and RHOV stations were not stored in the PeMS database.
|
From page 324... ...
322 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY requires a large set of historical travel times, the team wanted to minimize the delay in acquiring traffic data. For this reason, as soon as the metadata were inserted into PeMS, the team implemented a program to download the traffic data XML file from the RITIS website every minute and save it so that data could be parsed from the files and placed into the PeMS database as soon as the file format was thoroughly understood.
|
From page 325... ...
323 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY NOVA |
From page 326... ...
324 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY in the RITIS XML file. From this, it was concluded that the data collection system in the field is doing the preprocessing, but the team was not able to ascertain exactly what was being done.
|
From page 327... ...
325 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY for each detector. There are a few common problems with detection infrastructure, and they manifest themselves in distinct ways in transmitted data, allowing for an automated quality control process.
|
From page 328... ...
326 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY 0 5 10 15 20 25 30 35 40 45 50 3/1/11 3/8/11 3/15/11 3/22/11 3/29/11 4/5/11 4/12/11 4/19/11 4/26/11 5/3/11 Pe rc en t o f D et ec to rs Good Stuck No Data Figure C.76. Daily detector health status, NOVA PeMS deployment, 2011.
|
From page 329... ...
327 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY only crude estimates of its flow values based on flows observed at nearby detectors. This meant that PeMS repeated the same flow, occupancy, and speed data for a given hour from week to week.
|
From page 330... ...
328 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY the percentage of detectors falling into the leading two error categories: no data and stuck. During 2009, the number of working detectors was significantly higher than in 2011, generally hovering above 70% for most of the year.
|
From page 331... ...
329 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY Figure C.80. Hourly travel times, westbound I-66, March 1 through April 30, 2011.
|
From page 332... ...
330 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY incidents, construction, weather, or fluctuations in demand. This concept is illustrated in Figure C.81, which shows the distribution of weekday travel times on a corridor in Northern Virginia.
|
From page 333... ...
331 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY also provides an opportunity to test a methodology that was developed for modeling the distribution of individual vehicle travel times on aggregated travel times calculated from loop detectors. Site Description A multistate model was developed for a 26-mile stretch of eastbound I-66 from Manassas to Arlington, Virginia.
|
From page 334... ...
332 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY Method The goal of this study was to generate, for each hour of the day, two outputs: the percentage chance that the traveler would encounter a certain condition; and for each condition, the average and 95th percentile travel time. The mathematical details of these steps are explained by Guo et al.
|
From page 335... ...
333 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY is selected as the optimal model, and each data point is given an initial probability of belonging to each state. The outputs of this step (the model type, number of states, and initial probabilities of a data point belonging to each state)
|
From page 336... ...
334 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY TABLE C.31. SELECTION OF STATES Hour Optimal Constrained Final No.
|
From page 337... ...
335 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY active during these hours, the mean travel time is still generally less than 30 minutes, and the 95th percentile travel time is generally less than 35 minutes. At the beginning of the afternoon peak (4:30 to 5:30 p.m.)
|
From page 338... ...
336 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY likelihood is around 25% during the other 2 hours. The severity of congestion in this state differs across each hour.
|
From page 339... ...
337 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY TABLE C.33. PROBABILITY, MEAN TRAVEL TIME, STANDARD DEVIATION, AND 95TH PERCENTILE TRAVEL TIME BY STATE Time Probability (%)
|
From page 340... ...
338 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY Morning Peak As discussed in previous sections, a three-state normal-mixture model was selected to measure reliability statistics for the four morning peak hours. Figure C.85 provides a visual comparison of the relative model fits of the three-state normal-mixture model, a two-state normal-mixture model, and a lognormal distribution model.
|
From page 341... ...
339 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY 8:30 a.m. hour, the fits between the three-state and two-state mixture models appear comparable, and their BICs are essentially equivalent.
|
From page 342... ...
340 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY Figure C.86. Travel time distributions and states, morning peak.
|
From page 343... ...
341 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY PROBE VEHICLE COMPARISONS Introduction To better understand the implications of the data quality issues on travel times, the team performed a quality control procedure. Probe vehicle runs were conducted along I-66 to amass ground-truth data that could be compared with the sensor data.
|
From page 344... ...
342 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY Figure C.87. Display of functioning versus nonfunctioning sensor stations.
|
From page 345... ...
343 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY • Speeds reported by this sensor are approximately 27 mph at all times of day except during the middle of the night, when traffic speeds increase significantly. • The reported traffic flows appear fairly normal (with the exception of an apparent issue occurring between approximately 1 and 5 p.m.
|
From page 346... ...
344 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY It is likely that some portions of these data-related issues are the result of the high percentage of imputed detector data being used to represent conditions at many detector stations (e.g., 59% of data used to generate the contents of Figure C.88 are imputed rather than observed)
|
From page 347... ...
345 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY Data Analysis Along I-66 Inside of I-495 Eastbound Figures C.90, C.91, and C.92 show plots of the instantaneous speeds recorded by the vehicle probe as it traversed I-66 eastbound inside of I-495 at three times on Tuesday, April 19, 2011, plotted against the speeds reported by the detectors (804, 822, 808, and 817) along that stretch of roadway at that those same times.
|
From page 348... ...
346 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY Comparison of the probe speeds with the sensor-based speeds suggests the following: • Data generated by Sensor 804 (Mileposts 68.5 to 70.0) were not consistent with the probe data collected along this roadway segment.
|
From page 349... ...
347 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY Figure C.93. Segment C > D, Run 4, I-66 westbound at 3:27 p.m., Tuesday, April 19, 2011.
|
From page 350... ...
348 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY Figure C.95. Segment C > D, Run 6, I-66 westbound at 6:38 p.m., Tuesday, April 19, 2011.
|
From page 351... ...
349 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY Figure C.97. Segment E > F, Run 8, I-66 eastbound at 10:20 a.m., April 20, 2011.
|
From page 352... ...
350 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY Comparison of these probe data with the sensor-based speeds suggests the following: • Only 15% of the speeds reported by Sensor 1139 (Mileposts 54.4 to 54.9) were actually observed.
|
From page 353... ...
351 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY Figure C.100. Segment G > H, Run 11, I-66 westbound at 9:53 a.m., April 20, 2011.
|
From page 354... ...
352 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY Comparison of these probe data with the sensor-based speeds suggests the following: • Only 15% of the speeds generated by Sensor 1143 (Mileposts 56.3 to 55.7) were actually observed.
|
From page 356... ...
354 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY calculating travel times, it must be remembered that two of the sensors generated suspect speed data -- in this case, very low freeway speeds. Incorporating these speeds into the travel time estimation appears to have offset the higher roadway speeds generated by the other two roadway sensors, speeds that were generally much higher than those reported by the probe vehicle.
|
From page 357... ...
355 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY LESSONS LEARNED Overview The team selected Northern Virginia as a case study site because it provided an opportunity to integrate a reliability monitoring system into a preexisting, extensive data collection network. The data collected on NOVA roadways are already passed to a number of external systems, including RITIS at the University of Maryland, the archived data management system at the University of Virginia, and the statewide 511 system.
|
From page 358... ...
356 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY One of the ways that agencies can facilitate the distribution of data from their data collection system is by establishing one or more data feeds. As discussed in the first chapter, different parties will want to acquire data processed to different levels, depending on the intended use.
|
From page 359... ...
357 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY available to them. As part of this use case, the team examined the data available from a network of fixed infrastructure sensors (a combination of single loops and radar-based sensors)
|
From page 360... ...
358 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY Another dynamic that affects the comparison of sensor data with probe vehicle data stems from a basic difference between these data sets: • Sensor data represent 5-minute, average conditions across all lanes of travel observed at the sensor location; but • Probe data represent the movement of a single vehicle through one lane of travel across the segment being evaluated. These differences may potentially result in significant differences in speed and estimated travel time between the two data sources if one lane of travel experiences significant congestion, but the other lanes do not.
|
From page 361... ...
359 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY Case Study 3 SACRAMENTO–LAKE TAHOE, CALIFORNIA The Sacramento–Lake Tahoe region of northern California was selected for this case study because it provided an example of a rural transportation network with sparse data collection infrastructure. The study region was of additional interest because it includes urban, suburban, and rural areas and has routes with heavy recreational traffic and areas where adverse weather can have a major effect on travel time reliability.
|
From page 362... ...
360 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY Bluetooth- and electronic toll collection (ETC) –based systems deployed in rural areas as part of this case study.
|
From page 363... ...
361 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY Caltrans and its regional partners are pursuing the creation of corridor system management plans (CSMPs) , defined by Caltrans as follows: A CSMP is a comprehensive, integrated management plan for increasing transportation options, decreasing congestion, and improving travel times in a transportation corridor.
|
From page 364... ...
362 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY Figure C.104. Map of electronic toll collection and Bluetooth readers deployed in Caltrans District 3.
|
From page 365... ...
363 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY TABLE C.41. BREAKDOWN OF DEPLOYED BLUETOOTH READERS IN STUDY AREA BTR ID from Figure C.104 Roadway and Direction of Travel Bluetooth Reader ID Nearest Crossroad Postmile Bluetooth 1 I-5 NB 1005 Elk Grove 506.4 Bluetooth 2 I-5 NB 1011 Pocket 511.5 Bluetooth 3 I-5 SB 2101 Florin 512.4 Bluetooth 4 I-5 SB 2009 Gloria 513.5 Bluetooth 5 I-5 NB 1039 Vallejo 517.2 Bluetooth 6 I-5 NB 1004 L Street 518.9 Bluetooth 7 US-50 EB 1054 Placerville 48.4 Bluetooth 8 US-50 EB 2055 Twin Bridges 87.1 Bluetooth 9 US-50 WB 2058 Echo Summit 94.9 Bluetooth 10 US-50 EB 2056 Meyers 98.7 Note: NB = northbound; SB = southbound.
|
From page 366... ...
364 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY Data Management The primary data management software system in the District 3 region is Caltrans' Performance Measurement System (PeMS)
|
From page 367... ...
365 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY • Real-time traffic data in a constant format at a constant frequency (such as every 30 seconds or every minute)
|
From page 368... ...
366 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY Integration of District 3 Case Study Data Sources into PeMS The two sources of data used in support of this case study, which are based on the movement of vehicles equipped with ETC and Bluetooth devices, are extremely new and are not currently integrated into Caltrans District 3's existing PeMS data feed. Consequently, it was necessary to incorporate these data sets into project-specific instances of PeMS for analysis as part of this project.
|
From page 369... ...
367 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY • Minimum: the fastest vehicles that traveled across a roadway segment during a given period of time; • 25th percentile: the 25th percentile travel time during a given period of time; • Mean: the mean travel time during a given period of time; • Median: the median travel time during a given period of time; • 75th percentile: the 75th percentile travel time during a given period of time; and • Maximum: the slowest-moving vehicles that traveled across a roadway segment during a given period of time. It is likely that much (if not all)
|
From page 370... ...
368 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY the ETC antenna. However, BTRs can record any device generating a Bluetooth signal within its sensing radius, sometimes from 100 meters away.
|
From page 371... ...
369 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY Bluetooth Device Data Impact of Bluetooth Reader Hardware on Available Data The characteristics of Bluetooth-based data available for analysis are determined largely by the capabilities of the BTR deployed at the roadside. For example, only 5 of the 10 BTRs deployed by Caltrans had the ability to read and store signal strength measurements for each observation.
|
From page 372... ...
370 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY Figure C.105 depicts the detection zones generated by ETC- and Bluetooth-based data collection technologies. Table C.43 provides examples of mean, maximum, and standard deviation of mobile device signal strengths collected by the various BTRs involved in this study.
|
From page 373... ...
371 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY TABLE C.43. BTR SIGNAL STRENGTH CHARACTERISTICS Bluetooth Reader ID No.
|
From page 374... ...
372 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY for approximately 11 minutes within the detection zone. The second vehicle (middle)
|
From page 375... ...
373 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY Figure C.107. Node traversal time frequency distribution for BTRs 2 (top)
|
From page 376... ...
374 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY Calculating Travel Times Based on Bluetooth Device Data The primary goal of BTR-based data analysis is to characterize segment travel times between BTRs based on the reidentification of observations derived from unique mobile devices. Generally, the data processing procedures associated with the calculation of BTR-to-BTR travel times can be broadly broken down into three processes, as shown below.
|
From page 377... ...
375 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY Process 1: Identification of Passage Times The first step in the process of calculating segment travel time PDFs (TT-PDFs) for a roadway is the calculation of segment travel times for individual vehicles.
|
From page 378... ...
376 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY The method used to aggregate visits is a causal sliding time window filter. This method is a filter in that it removes unnecessary observations during the aggregation process.
|
From page 379... ...
377 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY • Observation underaggregation. Incorrectly subdividing observations from a single visit into multiple visits may result in the incorrect calculation of passage time, depending on the method used.
|
From page 380... ...
378 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY Vehicles that are continuously within a BTR's detection zone (and generating observations) are either in travel mode (e.g.
|
From page 381... ...
379 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY In addition to considering the impact of BTR passage times, users of Bluetooth data must consider that the accurate calculation of segment travel time is a function of the relationship between BTR-to-BTR distance and the maximum speed error. Following on the analysis performed by Haghani et al.
|
From page 382... ...
380 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY Figure C.114. Relationship between maximum speed error and BTR-to-BTR distance with ΔT > 0.
|
From page 383... ...
381 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY effect of distance errors. A time error of 4 seconds was used based on clock synchronization error; associated with Caltrans' method for synchronizing BTRs when local time differed from network time by more than 2 seconds.
|
From page 384... ...
382 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY Figure C.115. Trips generated from all visit permutations.
|
From page 385... ...
383 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY This method can be applied to the data displayed in Table C.46, which show three origin visits in Rows 1, 2, and 3. The question is whether any of these visits can be aggregated or whether each should be considered a valid origin departure.
|
From page 386... ...
384 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY This section of the methodology describes a four-step technique for filtering travel times, presents travel time histograms before and after filtering, and compares the effects of two passage time pairing techniques (Method 2 and Method 3, described above) on the resulting travel time histograms.
|
From page 387... ...
385 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY distribution of the speeds (with a recommended radius of 4 mph)
|
From page 388... ...
386 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY was set to the maximum observed speed (the minimum of the moving average above the modal speed) , and the lower threshold was set to 25 mph (the minimum of the moving average below the modal speed)
|
From page 389... ...
387 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY Figure C.120. Four-step filtering on passage time pairing Method 2 with 5-minute intervals.
|
From page 390... ...
388 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY Figure C.121. Filtered travel times (5-minute intervals)
|
From page 391... ...
389 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY SUMMARY This section evaluates various methodological approaches and processes for estimating ground-truth segment travel times based on Bluetooth data. The characteristics of Bluetooth data at each node were found to vary significantly as a function of the surrounding roadway configuration.
|
From page 392... ...
390 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY USE CASE ANALYSIS This case study explores the use of ETC- and Bluetooth-based vehicle reidentification technologies in support of travel time reliability monitoring in a rural setting. These data collection technologies work by sampling the population of vehicles along the roadway and subsequently matching unique toll tag IDs or Bluetooth MAC addresses between contiguous reader stations.
|
From page 393... ...
391 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY Impact of Electronic Toll Collection Reader Deployment Configuration on Data Quality This use case details the findings of the research team's investigation into the impact of the configuration of the Lake Tahoe ETC network on the quality of travel time data collected. The ETC detection network consisted of eight FasTrak readers located on I-80 between the eastern outskirts of Sacramento and North Lake Tahoe.
|
From page 394... ...
392 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY The final question relates to the quality of travel times being reported. As a higher percentage of matches will be more likely to result in a more accurate travel time estimate, the research team assessed the percentage of tags matched between all possible combinations of upstream and downstream ETC readers.
|
From page 395... ...
393 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY Figure C.124. Electronic toll collection locations.
|
From page 396... ...
394 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY The team next evaluated the minimum travel times reported between each pair of readers to ensure they were reasonable given the distances involved. All travel times were determined to be reasonable with the exception of trips that originated or ended at the Kingvale reader, stated by Caltrans as being located on I-80 westbound, adjacent to the Rainbow reader on I-80 eastbound.
|
From page 397... ...
395 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY Figure C.127. Travel times between origin I-80 eastbound at Auburn–Bell and destination I-80 westbound at Prosser Village.
|
From page 398... ...
396 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY indicated in Figure C.127, the minimum travel times between Auburn–Bell on I-80 eastbound and Prosser Village on I-80 westbound range between 60 and 80 minutes, which is reasonable given the 60-mile distance between them. The 75th percentile travel times are higher, likely reflecting the travel times of vehicles detected passing the Prosser Village reader on I-80 westbound as part of a round trip after having first passed both the Auburn–Bell reader, as well as the Prosser Village reader, but not being detected by it, while traveling east.
|
From page 399... ...
397 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY Comparing average hourly hit rates for all of the readers on I-80 eastbound from Tuesday through Friday makes it clear that some readers are sampling a significantly higher percentage of traffic than others. An examination of photographs of the signs on which each reader was mounted provides no clear explanation for why the hit rates at some readers are approximately double those at other readers.
|
From page 400... ...
398 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY To gauge the sampling rate in another way, the team also looked at the raw number of hourly tag reads reported by each reader, the results of which are displayed in Figure C.129 for the eastbound direction of travel. Despite its low percentage of tag reads, the reader at Auburn–Bell Road still records a large number of reads, simply because the traffic volumes are higher here than at any other reader.
|
From page 401... ...
399 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY patterns between the Bay Area and Lake Tahoe. Again, despite being the second farthest reader from the origin, the Kingvale reader often sees the most matches on I-80 eastbound.
|
From page 402... ...
400 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY Findings Two primary variables affect hit rate: the total number of ETC tags in the population of vehicles that pass a reader and the number of tags actually read by a specific reader. The product of these factors has a significant influence on the accuracy of travel time data generated between any two ETC readers.
|
From page 403... ...
401 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY As expected, the hit rate for an individual reader has a profound impact on that reader's ability to reidentify vehicles initially detected at upstream readers. For example, as shown in Table C.49, even though the Auburn–Bell Road reader is 45 miles and 24 exits from the downstream reader at Rainbow, the high hit rate at this downstream reader enables it to reidentify 83% of vehicles initially detected at Auburn–Bell Road.
|
From page 404... ...
402 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY The first question was particularly important for this use case as the BTRs were deployed as part of a test, and not as permanent data collection infrastructure. As a result, each BTR changed locations multiple times over a span of several months.
|
From page 405... ...
403 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY Which Facilities Is Each Reader Monitoring? The next step in understanding the impact of each BTR's configuration on the nature of the data collected was to determine which readers might be capturing traffic data for multiple directions of travel.
|
From page 406... ...
404 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY 79 2014.04.23 12 L02 Guide Appendix C Part 4_ final for composition.docx US-50/Echo Summit: not visible in photograph US-50/Meyers: off US-50 eastbound US-50/Twin Bridges: off US-50 eastbound US-50/Placerville: off US-50 eastbound I-5/Vallejo: off I-5 northbound Figure C.132. Bluetooth reader cabinet locations.
|
From page 407... ...
405 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY 80 2014.04.23 12 L02 Guide Appendix C Part 4_ final for composition.docx Figure C.132. Bluetooth reader cabinet locations.
|
From page 408... ...
406 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY To calculate the percentage of vehicles sampled at each BTR, the research team compared Bluetooth mobile device reads with the traffic flows measured at nearby loop detectors; the result is referred to as the hit rate. Hit rates were computed for the four readers on I-5 (there were no working loop detectors near the US-50 readers)
|
From page 409... ...
407 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY reaching nearly 1,000 reads per hour during the weekday afternoon peak. In contrast, the reader at I-5 and Pocket has both the lowest hit rate and the lowest number of reads, with between 500 and 600 MAC address reads per hour during the peak hours and only 300 to 400 per hour during the midday.
|
From page 410... ...
408 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY one another) near South Lake Tahoe, the number of hourly reads are fairly similar, and are quite low (30 to 50 per hour, or two to four per 5 minutes)
|
From page 411... ...
409 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY 85 2014.04.23 12 L02 Guide Appendix C Part 4_ final for composition.docx For the Placerville (westernmost) reader, 22% of the reads were reidentified downstream at Twin Bridges.
|
From page 412... ...
410 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY Based on these high reidentification rates, the team concluded that the readers on US-50 were capable of detecting and reidentifying a very high proportion of the Bluetooth devices that pass through their detection zones, likely due to the narrow roadway width at these locations and the limited options available to exit the roadway. The next technique for evaluating Bluetooth device reidentification between readers was to examine the raw number of matches between readers to assess whether the match volumes were sufficient to yield accurate average travel times.
|
From page 413... ...
411 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY • As Pocket had the lowest hit rate of the readers on I-5, a smaller percentage of vehicles was available for reidentification when using this reader as an origin. • The numbers of matches at each of the three destination readers were similar, and generally differed by less than 25 per hour, representing a difference of about 10%.
|
From page 414... ...
412 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY • Although the number of matches decreased with distance from the origin reader, the number of matches was similar between the three destinations due to a significant amount of traffic on US-50 traveling its entire length from Lake Tahoe to Sacramento. • The number of matches was much lower than along I-5, likely due to the rural characteristics and lower traffic volumes on US-50.
|
From page 415... ...
413 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY 100 per hour (8 per 5 minutes or 25 per 15 minutes) , likely enough to compute average and median travel times at a 5- or 15-minute granularity.
|
From page 416... ...
414 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY direction of Lake Tahoe. The number of matches peaked at between 6 and 12 per hour on Friday afternoon, and was also higher on Saturday morning, at around 8 per hour.
|
From page 417... ...
415 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY Finally, this use case enabled the research team to compare hit rates and matching percentages for readers located in both urban and rural environments. In this study, as is typical for urban versus rural settings, the biggest differences between the readers deployed on I-5 and US-50 included the number of lanes at each reader, the distance between readers, and the traffic volumes at each reader.
|
From page 418... ...
416 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY the day, none of the readers reidentify more than 50% of the hits from upstream readers. In contrast, as Table C.51 shows, 68% of the hits from the Twin Bridges reader on rural US-50 are reidentified at the Placerville reader (39 miles away)
|
From page 419... ...
417 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY traffic. However, the team also found that the readers are most effectively deployed in locations where they only monitor traffic in the mainline lanes.
|
From page 420... ...
418 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY traffic data. Operators and analysts will benefit from a discussion of the quality and typical characteristics of this type of data.
|
From page 421... ...
419 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY Figure C.145. Example of Site 1 on I-80.
|
From page 422... ...
420 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY These two sites were selected due to their strong weekend traffic patterns, as well as their proximity to local weather observation stations. They were made as short as possible (within the constraints of the detection infrastructure)
|
From page 423... ...
421 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY • Minimum travel time; • Average travel time; • Maximum travel time; • 25th, 50th, and 75th percentile travel times; and • Flow (number of vehicles observed during the window)
|
From page 424... ...
422 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY by those values, the research team used the median travel time for each 5-minute interval. In this case, working with the median as opposed to the mean has a significant effect on the analysis, reducing the appearance of implausible extreme values.
|
From page 425... ...
423 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY Travel Time Analysis: Site 1 Site 1, which lies on I-80 westbound and begins just north of Lake Tahoe, is known to receive heavy traffic from vehicles returning to the Bay Area from weekend trips on Sunday evenings. The breakdown of travel times by day of the week from April 25 to June 29, 2011, shown in Figure C.149 indicates that the Sunday 95th percentile travel time exceeds that of a normal weekday by approximately 34%.
|
From page 426... ...
424 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY was plotted as shown in Figure C.150. This figure demonstrates that no significant timeof-day trends exist on this route.
|
From page 427... ...
425 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY Figure C.151. Site 1 TT-PDFs during various weather conditions.
|
From page 428... ...
426 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY It is clear from Figure C.152 that snow, low-to-moderate visibility, and precipitation have a measurable effect on travel time reliability. The 95th percentile travel times during those weather conditions are significantly higher than their median travel times, indicating that the distribution of travel times is skewed toward the high end.
|
From page 429... ...
427 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY TABLE C.54. WEATHER CONDITIONS ACTIVE DURING LONG TRAVEL TIMES Condition Active Active When Travel Time Exceeded 85th Percentile 95th Percentile No precipitation 90.3% 84.3% 76.6% Precipitation 9.7% 15.7% 23.4% Baseline 90.7% 74.9% 54.7% Snow event 5.8% 23.0% 45.3% Rain event 1.6% 1.6% 0.00% Fog event 1.7% 0.5% 0.00% Thunderstorm event 0.3% 0.00% 0.00% High visibility 84.7% 67.5% 48.4% Medium visibility 4.8% 11.0% 11.0% Low visibility 3.8% 14.1% 31.3% Travel Time Analysis: Site 2 Site 2 is similar to Site 1 in that it is subject to periodic spikes in demand from weekend travel.
|
From page 430... ...
428 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY Time of Day Figure C.153. US-50 westbound average travel time by time of day (Site 2)
|
From page 431... ...
429 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY Day of Week Figure C.155. Weekly flow on US-50 eastbound (Site 2)
|
From page 432... ...
430 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY Figure C.157. Week-long distribution of travel times on US-50 eastbound (Site 2)
|
From page 433... ...
431 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY PRIVACY CONSIDERATIONS Innovations in data collection technology are providing exciting opportunities in the area of roadway travel time measurement. At the same time, use of these technologies is not without challenges, some technical, others related to protecting the confidentiality of personal information contained in ETC toll tag and Bluetooth mobile device data sets.
|
From page 434... ...
432 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY over privacy, this section provides examples of the types of privacy protection policies and procedures currently in use by both public agencies and private sector companies to guard against the misuse of drivers' personal information. Electronic Toll Tag–Based Data Collection Overview of Personal Privacy Concerns When used for toll collection purposes, toll transponders are automatically identified whenever they pass within the detection zone of a compatible ETC reader.
|
From page 435... ...
433 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY Metropolitan Transportation Commission In support of its 511-traveler information service, MTC operates a travel time data collection system based on information collected from the region's FasTrak toll system. As part of this effort, MTC takes the following steps to ensure the protection of toll tag users' personal information (7)
|
From page 436... ...
434 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY (i.e., MAC address) is much less straightforward.
|
From page 437... ...
435 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY • Use data processing safeguards (encryption and randomization) to prevent the recovery of unique MAC addresses: -- Encryption methods transform MAC address data (at the sensor level)
|
From page 438... ...
436 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY reidentification observations and calculating segment travel times. The results show that smart filtering and processing of Bluetooth data to better identify likely segment trips increase the quality of calculated segment travel time data.
|
From page 439... ...
437 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY given location. The positioning of BTRs, which have a large detection radius, dictates whether ramp, parallel facility, or multimodal traffic is also sampled.
|
From page 440... ...
438 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY (e.g., ITS America's Fair Information and Privacy Principles) aimed at maintaining the anonymity of specific users.
|
From page 441... ...
439 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY Case Study 4 ATLANTA, GEORGIA The team selected the Atlanta, Georgia, metropolitan region to provide an example of a mixed urban and suburban site that primarily relies on video detection cameras for real-time travel information. The main objectives of the Atlanta case study were to • Demonstrate methods to resolve integration issues by using real-time data from Atlanta's traffic management system for travel time reliability monitoring.
|
From page 442... ...
440 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY MONITORING SYSTEM Site Overview With a population of approximately five-and-a-half million people, Atlanta is the ninth largest metropolitan area in the United States. The layout of the freeway network follows a radial pattern.
|
From page 443... ...
441 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY freeway network also contains 90 miles of high-occupancy vehicle (HOV) lanes that operate 24 hours a day, 7 days a week on the following facilities: • I-75 inside the I-285 loop; • The Downtown Connector; • I-20 east of the Downtown Connector; and • I-85 between Brookwood and SR-20.
|
From page 444... ...
442 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY and incidents. To minimize the disruption of traffic caused by lane-blocking incidents, TMC staff can dispatch highway emergency response operator patrols.
|
From page 445... ...
443 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY Traficon sensors are placed with a very dense spacing of about one-third of a mile. Autoscope cameras monitor a small portion of I-85 near the Hartsfield– Jackson Atlanta International Airport with a spacing comparable to that of the Traficon cameras.
|
From page 446... ...
444 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY Data Management The primary data management system used in the Atlanta region is the Georgia DOT's Navigator system. Navigator was initially deployed in metropolitan Atlanta in preparation for the 1996 Summer Olympic Games.
|
From page 447... ...
445 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY The field data acquisition subsystem, which is responsible for device communication and management, acquires data from CMS, detector stations, ramp meters, a parking management system, and highway advisory radio. The management services system helps TMC staff analyze data to determine conditions and develop response plans; it includes the Navigator graphical user interface, congestion and incident detection and management services, response plan management, and the historical logging of detector data.
|
From page 448... ...
446 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY PeMS is a traffic data collection, processing, and analysis tool that extracts information from real-time intelligent transportation systems, saves it permanently in a data warehouse, and presents it in various forms to users via the web. To report performance measures such as travel time reliability, PeMS requires three types of information from the data source system: • Metadata on the roadway line work of facilities being monitored; • Metadata on the detection infrastructure, including the types of data collected and the locations of equipment; and • Real-time traffic data in a constant format at a constant frequency (such as every 30 seconds or every minute)
|
From page 449... ...
447 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY • Current travel time; • Free-flow travel time; • Current speed; • Free-flow speed; • Jam factor; • Jam factor trend; and • Confidence. The lengths of the traffic message channel segments vary but are generally between 0.3 and 2 miles long.
|
From page 450... ...
448 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY The San Diego and Lake Tahoe case studies focused on estimating PDFs for travel times measured during instances of nonrecurrent congestion. These distributions help distinguish between the natural travel time variability of a facility due to the complex interactions between demand and capacity and the travel time variability during specific events.
|
From page 451... ...
449 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY Method The method to develop the regimes and estimate the impacts of nonrecurrent congestion events consists of three steps: 1. Regime characterization, to estimate the number and characteristics of each travel time regime measured along the facility; 2.
|
From page 452... ...
450 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY state. This output is used to drive the nonrecurrent congestion reliability analysis, which is described in the following subsection.
|
From page 453... ...
451 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY end times are rarely explicit and have to be assumed. In this study, a travel time was tagged with special event if it occurred up to 1 hour before the event start time and in the hour after the estimated end time.
|
From page 454... ...
452 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY samples. Because ATMS data are conventionally used only for real-time operations, the acceptable level of data quality is much lower than it is for the analysis of archived data.
|
From page 455... ...
453 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY are all used to denote on-ramp detectors. This multiplicity of terms required the development of a mapping structure to appropriately categorize Navigator detectors in PeMS, as shown in Table C.57.
|
From page 456... ...
454 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY report flow, occupancy, and speed, the frequency at which they report varies. GDOT stores the most recent 30 minutes of data from each active detector in a database table.
|
From page 457... ...
455 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY 1. PeMS performs detector diagnostics at the end of every day.
|
From page 458... ...
456 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY TABLE C.59. EVENT DATA FORMAT Column Name Description Example 1 ID Unique ID 244835 2 Primary Road Freeway number I-75 3 Dir Direction of travel N 4 MM Mile marker 228 5 Cross Cross street Jonesboro Rd 6 County County Clayton 7 Start Event start date and time 09/01/2011 01:00 8 End Event end date and time 09/02/2011 06:15 9 Type Type of event Accident/Crash 10 Status Status of event Terminated 11 Blockage Number of lanes blocked 2 The breakdown of events by type in the data set, grouped and summed into event types in similar categories, is shown in Table C.60.
|
From page 459... ...
457 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY • The same freeway was given different names in the Primary Road column. • Mileposts were missing from some events.
|
From page 460... ...
458 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY Use Case 2: Determining Travel Time Regimes and the Impact of the Seven Sources of Congestion Summary The Northern Virginia case study analyses developed methodologies for modeling the multimodal nature of travel time distributions to determine the operating regimes of a facility. The San Diego case study analyses validated ways to evaluate the impact of the seven sources of congestion on travel time variability.
|
From page 461... ...
459 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY Figure C.163. Morning multistate normal PDFs.
|
From page 462... ...
460 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY Figure C.165. Afternoon multistate PDFs.
|
From page 463... ...
461 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY Data Fusion In the data fusion step, the seven sources data described in the methodological advancements section were fused with the 5-minute travel times. Table C.62 summarizes the number and percentage of travel time samples by source within each time period.
|
From page 464... ...
462 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY During the afternoon peak time period, the congested state that happens 93% of the time (State 1) contains nearly all of the congestion source travel times.
|
From page 465... ...
463 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY 44 2014.04.23 13 L02 Guide Appendix C Part 5_final for composition.docx Figure C.167. Midday peak travel times by source.
|
From page 466... ...
464 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY TABLE C.63. SOURCE CONTRIBUTIONS TO MORNING PEAK REGIMES State 1 State 2 Parameter Probability (%)
|
From page 467... ...
465 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY TABLE C.65. SOURCE CONTRIBUTIONS TO AFTERNOON PEAK REGIMES State 1 State 2 Parameter Probability (%)
|
From page 468... ...
466 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY measurements, which was the distance between each pair of compared detectors, was analyzed and found to be moderately significant. Data from multiple sources, if properly understood, can be aggregated to provide a rich set of performance monitoring information.
|
From page 469... ...
467 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY Data Characteristics This use case compares two types of traffic data: speed data from vehicle probes, provided by NavTeq, and speed data from Traficon video detectors. The vehicle probe data come from GPS chips residing within individual vehicles, directly measuring their speed and location.
|
From page 470... ...
468 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY differ, they represent the same stretch of roadway (see Figure C.169)
|
From page 471... ...
469 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY Methods The comparison of the probe and video speed data began with the procurement of that data. PeMS began collecting live Traficon video detector data in the Atlanta region on September 9, 2011.
|
From page 472... ...
470 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY To properly compare the two data sets it is immediately necessary to convert them to a common time standard. As obtained from PeMS, the video data and probe data have different time ranges and different sampling frequencies.
|
From page 473... ...
471 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY focused solely on the data generated by the sensors, only 100% observed data points were included. After this filtering, between 40% and 50% of 5-minute periods contained data for most Traficon video detectors.
|
From page 474... ...
472 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY Figure C.172. Comparison of speeds from video (black)
|
From page 475... ...
473 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY Inspection of Figure C.173 reveals that the probe data at this location appear to lag slightly behind the video detector data. This lag can be quantified by computing the cross-correlation of the two data sets.
|
From page 476... ...
474 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY I-285 Southbound Results Correlations in speed measurements from the southbound direction of travel were slightly weaker than in the northbound direction, ranging from 0.69 to 0.87. The range of correlations was greater in this direction of travel, perhaps because of the larger number of pairs.
|
From page 477... ...
475 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY relationship that was seen in the southbound direction reemerges, with a slightly lower correlation coefficient (R2 = 0.43)
|
From page 478... ...
476 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY the speed difference appears to fluctuate around zero indicates further that this pair is still a good match. Since the detectors agree on the general duration and speed profile of congestion and their difference is centered at zero, their correlation will likely improve as the data are rolled up to coarser levels of temporal aggregation.
|
From page 479... ...
477 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY 3. Traffic data may not be received at constant sampling rates.
|
From page 480... ...
478 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY REFERENCE 1. Federal Highway Administration.
|
From page 481... ...
479 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY Case Study 5: NEW YORK/NEW JERSEY The New York City metropolitan area in the states of New York and New Jersey was chosen to provide insight into travel time monitoring in a high-density urban location. The main objectives of the New York/New Jersey case study included • Obtaining time-of-day travel time distributions for a study route from probe data; • Identifying the cause of bimodal travel time distributions on certain links; and • Exploring the causal factors for travel times that vary significantly from the mean conditions.
|
From page 482... ...
480 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY Lessons learned summarizes the lessons learned during this case study with regard to all aspects of travel time reliability monitoring: sensor systems, software systems, calculation methodology, and use. These lessons will be integrated into the final Guide for practitioners.
|
From page 483... ...
481 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY route travel times from point speeds. For these probe data, the network configuration is made up of links defined by ALK.
|
From page 484... ...
482 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY most or all raw GPS points, matched to ALK's link-based network configuration. The third data set, called one monument, is an aggregation of the GGD data set.
|
From page 485... ...
483 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY Interstate, Interstate without ramps, divided road, primary road, ferry, secondary road, ramps, and local road. Local roads make up the vast majority of the links in the network configuration.
|
From page 486... ...
484 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY The first step in route analysis was to determine which ALK links make up the study route, which was done by visually identifying the ALK grids through which the route travels. From there, it was possible to map all Interstate-class links contained in the relevant grids and visually identify the links that make up the route.
|
From page 487... ...
485 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY data coverage means that analysis requiring data partitioning, such as comparing weekday and weekend speeds, will likely not yield rich results. The three freeway segments have the best data coverage; coverage is sparser on the arterials near the origin, the freeway connectors, and the airport roads at the destination.
|
From page 488... ...
486 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY To understand the traffic conditions represented in the data set, time-of-day–based speed distributions on a single link can be plotted. Figure C.181 depicts hourly PDFs of speeds observed on Link 38 in the route (near the I-278/I-495 interchange)
|
From page 489... ...
487 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY Next the team looked at how speeds vary across the route throughout the whole day, again considering the entire speed distribution on each link–hour. Speed measurements on each link during each hour of the day were extracted from the one monument data set, and the 25th percentile, median, and 75th percentile speeds for each link–hour were computed.
|
From page 490... ...
488 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY Results Using the quartile speeds for each link throughout the day, it is possible to simulate trip trajectories along the route for any slice of the speed distribution. This is done by fi rst choosing a virtual trip start time and then moving along the route link by link, simulating the arrival time at the next link based on the speed and length of the current link.
|
From page 491... ...
489 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY USE CASE ANALYSIS The single use case evaluated in this case study is a site-specifi c application of the probe data processing and analysis techniques described in the methodology section. The motivation for this use case was to generate and compare travel time distributions along a route at different times of day using only probe data.
|
From page 492... ...
490 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY monument data set. For this analysis, the most important variables in the data set are time stamp, speed, trip ID, and indexed position within a trip, if any (many trips are made up of a single point on the route)
|
From page 493... ...
491 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY Distribution Construction Stage The distribution construction stage builds up the full travel time distribution along the route link by link in four steps. Each iteration of the steps adds the subsequent downstream link into the route travel time distribution.
|
From page 494... ...
492 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY Method 2 The second method for computing route TT-PDFs ignores the linear dependence between consecutive links and directly computes the route travel time distribution as if all link travel times were independent. This method is based entirely on directly observed link speeds, discarding the time stamp differences between points in the same trip.
|
From page 495... ...
493 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY The speeds between 5 and 7 p.m. appear to be shifted by roughly the same amount as seen in Method 1.
|
From page 496... ...
494 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY data for each link. In this use case, 10 bins between 0 and 80 mph (each bin is 8 mph wide)
|
From page 497... ...
495 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY Conclusions Each of the three methods presented for assembling route TT-PDFs from probe vehicle data are enabled by the techniques introduced in the methodology section. Constructing these PDFs requires identification of the data points corresponding to a particular route, separation of data by time of day when possible, and an understanding of the relationships between link speed distributions and route speed distributions.
|
From page 498... ...
496 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY Although results were not validated with a different data source, the PDFs generated using Methods 1 and 2 appear to match expectations. An online trip planner estimates the travel time on this route to be 28 minutes, which generally agrees with the distributions seen here.
|
From page 499... ...
497 GUIDE TO ESTABLISHING MONITORING PROGRAMS FOR TRAVEL TIME RELIABILITY trend analysis. Transportation planners and operators often require an understanding of how route travel times vary on a day-by-day, week-by-week, and month-by-month basis.
|
Key Terms
This material may be derived from roughly machine-read images, and so is provided only to facilitate research.
More
information on Chapter Skim is available.