Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
TCRP Web-Only Document 53 78 CHAPTER 5: RISK ANALYSIS This chapter describes the risk analysis conducted for the test of higher speed LRV operations. The risk analysis identifies and measures potential safety impacts of the test, including: ï· What is the probability that a crash (or crashes) will occur? ï· What types of crashes are likely to occur? ï· How severe will the crashes be? In order to answer these questions, it is useful to consider the rates reflected in the past (i.e., observed crashes, near-miss incidents, and risky behaviors). These historical data, described in detail in Chapter 3, are a strong predictor of future events. Additionally, it is imperative to consider all the variables of change that could impact future outcomes. In the case of this test, change can result from the following: ï· The installation of alternative traffic control devices; ï· The addition up new traffic control devices or safety measures; and ï· An increase in LRV speed. Other factors that could change over time during the testing period that could impact future outcomes might include factors such as traffic volumes, traffic patterns, pedestrian volumes, and signal timing plans. Early on, it was envisioned that a probabilistic approach using regression analysis would be used for the risk analysis to predict the impacts and level of severity of crashes associated with an increase in LRV speed. However, the frequency of LRV-related crashes at the three study intersections over the past 3 years was very low, and there were no fatalities or serious injuries associated with any of the LRV-related crashes at the intersections. Therefore, it is not possible to correlate the crash data, traffic volumes, and observed risky behaviors at the three intersections. As a result, a frequency-based analysis was determined to be a more appropriate approach to the risk analysis considering the 3-year history. CRASH DATA The city data indicated that there were 6 LRV-related crashes in the past 3 years at the 3 study intersections. Four (67 percent) of the six crashes occurred at Brokaw Rd. One LRV-related crash was reported at each of the other two intersections. Of the LRV-related crashes, none involved pedestrians or bicyclists. One-half of the crashes (three of the six) involved vehicles making left turns from North First St. onto the cross-streets. The crash data provided by VTA indicate that there were 11 LRV-related crashes over the same 3-year period. Six (55 percent) of the 11 crashes occurred at Brokaw Rd. There were two LRV- related crashes reported at Charcot Ave. and three LRV-related crashes reported at Trimble Rd. in the past 3 years. As with the city data, none of the LRV-related crashes involved pedestrians or bicyclists. Nearly all of the crashes (10 of the 11 crashes, or 91 percent) involved vehicles making left turns from North First St. onto the cross-streets. These data are summarized in Table 22.
TCRP Web-Only Document 53 79 Table 22. Summary of Historical LRV-Related Crash Data Intersection of N. First St. Total Crashes* Ped or Bike Left Turn Red- Light Running LRV Signal Violation Other City VTA City VTA City VTA City VTA City VTA Brokaw Rd. 4 6 0 0 2 5 0 0 2 1 Charcot Ave. 1 2 0 0 1 1 0 0 0 1 Trimble Rd. 1 3 0 0 0 1 0 1 1 1 Total (n) 6 11 0 0 3 7 0 1 3 3 Percent of total 0% 0% 50% 64% 0% 9% 50% 27% Average 8.5 0 (0%) 5 (57%) 0.5 (4.5%) 3 (38.5%) * No fatal crashes reported. Three crashes involved minor injuries. These data show that the crash rate hovered between 2.0 and 3.7 crashes per year across the three intersections, depending on the data source used. If all factors were to remain constant, it would be possible to expect similar crash rates for the 1-year test period being proposed. Conservatively, using the VTA data, it would be expected that there could be approximately four crashes during the test year; two of these would be expected to occur at Brokaw Rd., and one would be expected to occur at the other two intersections. Based on the historical data, it is also likely that there would be no pedestrian or bicycle crashes observed in the 1-year test period. Most (if not all) of the observed crashes during the test year would be the result of a left-turn red- light violation from North First St.. It is important to note that history is merely predictive. A variety of factors could influence actual crash rates, which could presumably result in abnormally low or high crash rates during the 1-year test period. These factors include anomalies due to weather, construction, fuel costs, and others that could affect road conditions in such a way as to make a crash more likely or more severe or that could affect the amount of driving people do (and thus their exposure to crash opportunities). RISKY BEHAVIOR The historical crash data described above are predictive of future crash rates. However, risky behaviors provide a more detailed understanding of the potential for events that may not have been captured in a snapshot of crashes. For instance, in the 3 years of crash data reviewed, there were no LRV-pedestrian crashes. Relying solely on crash data to predict outcomes during the test year would predict zero LRV-pedestrian crashes. However, such a perspective would be naive. Examining risky behavior provides a more meaningful and realistic understanding of potential safety problems.
TCRP Web-Only Document 53 80 The repeated occurrence of a particular driver and/or pedestrian behavior could explain why a specific type of an incident or a crash would occur. As Chapter 3 reported in detail, there were 29 near-miss incidents reported by train operators at the 3 study intersections during the past 3 years. Seventeen (57 percent) of these near-miss incidents were reported at Brokaw Rd., while 6 (21 percent) were reported at each of the other two intersections. A majority of the near-miss incidents (21, or 72 percent) involved left- or U-turning vehicles, many of which were classified as a âleft-turn violation.â There were two near-miss incidents, both at Brokaw Rd. that involved a pedestrian crossing in front of a train. In addition to the historical safety data, the team conducted field observational studies as a means of assessing the operational and safety aspects of the existing traffic control devices. These data, also detailed in Chapter 3, showed that the most frequently-observed risky behavior was a left- turn during the yellow change or all-red clearance interval, and in several cases completing the turn at the end of the all-red clearance interval. This behavior occurred consistently at all three intersections. Observed pedestrian risky behavior did not demonstrate a high potential of severe implications. Pedestrian activity was low at all three intersections. Several pedestrians crossed the intersection either during the Donât Walk interval, or continued to cross after the Flashing Donât Walk had expired. Risky behaviors associated with cross-street motorists entering the intersection on red were also minimal; a mere 9 violations were observed during the 160 hours of data collection. The highest frequency of violations, overall, was observed at the intersection of North First St. at Brokaw Rd. The highest frequency of red-light violations (five of the nine) was observed for the westbound movement at Charcot Ave. Overall, however, none of the three intersections demonstrated unusual risky behaviors. This observational study data represents approximately 2 percent of operations in a year (1 out of 52 weeks). Extrapolating to a full year, the overall number of risky behaviors would likely increase. For example, there were 19 mainline left-turn change and clearance interval violations observed during the 1-week field data collection. Assuming a linear relationship and multiplying by a factor of 52, as many as 988 violations may be observed in a year. Similarly, there were 4 instances of pedestrians standing on the tracks. Thus, in one year, there might be approximately 208 events of pedestrians standing on the tracks. Fortunately, as seen in the low rates of crashes noted in the historical data, not every risky behavior leads to a crash. Also, the observed risky behaviors do not indicate the particular likelihood of a crash occurring that had not been reported in the crash data. Examination of the risky behavior data indicates that left-turn violations are the behaviors most likely to lead to crashes. Other crashes, such as those involving pedestrians remain unlikely (although not impossible, as pedestrians are occasionally present). IMPACT OF CHANGES The major changes that will occur during the test, and that were not reflected in the historical safety-related data or the observational behavior data, are: ï· Installation of alternative traffic control device (e.g., replacing the W10-7 sign to an alternating version of the sign, as previously described); ï· Installation of supplemental traffic control devices and safety countermeasures to change driver behavior (specifically, to increase awareness of LRV presence, discourage turning violations, and provide positive guidance to mitigate against track intrusions); and ï· Increase LRV speed from 35 to 40 mph.
TCRP Web-Only Document 53 81 ï· Closely monitor the increase in traffic volumes over the duration of the test period to track any significant fluctuation. Address any major volume changes through signal timing adjustments in order to avoid excessive queuing and subsequent blockage of the track which may increase the likelihood of a light rail vehicle/automobile conflict or crash. These changes during the test period will likely impact the rate and severity of crashes that do occur; however, they are likely to have effects that may counteract one another. For instance, increasing LRV speed may increase the severity of an injury should a crash occur. But fewer crashes may occur due to the new traffic control devices and safety measures. This may produce a net result of fewer but potentially more severe crashes. Impact of Safety Countermeasures The alternative traffic control devices and supplemental safety measures being installed during this test period have been well described in this report. These traffic control devices and safety measures focus on mitigating left-turn crashes from North First St., as well as mitigating track intrusions. Thus, it is expected that the crash rate for these types of crashes will be the most impacted as a result of the countermeasures. Conversely, it is also possible that some element of the installed countermeasures will have a detrimental effect on risky behaviors, near-miss incidents, or even crash rates. For instance, signs and markings, especially unfamiliar ones, can be misunderstood by drivers or may seem confusing, leading to an increase in unsafe actions and possibly even crashes. While the research team feels this outcome is unlikely, it is for this reason that the test and evaluation has been structured to include multiple stages. By first measuring the risky behaviors, near-miss incidents, and crash rates in the Baseline Conditions, there is a basis for comparison for the After Safety Improvement Conditions, which do not include an increase in LRV speed. Only after it has been established that there have been no unanticipated outcomes as a result of the safety improvements will the LRV speeds be raised to 40 mph. This is one of the primary ways of mitigating the risk associated with the test. Impact of Increased LRV Speed Theoretically, an increase in LRV speed from 35 to 40 mph should not increase the probability of crash occurrence per se. This small increase in speed should be transparent to drivers and therefore should not impact driver decision-making or behaviors. Assuming the intersection operations (track circuitry and traffic signal timing) have been modified to accommodate the speed increase, there is little reason to anticipate an increased likelihood of crashes at the intersections. The one exception may be that during a driver violation (such as running a red light during a left turn) the slightly faster LRV speed will result in decreased time for the vehicle to clear the intersection before impact, as well as increased braking distance for the LRV to come to a stop. These factors could, in some cases, result in crashes occurring that would have resulted in near-miss incidents at 35 mph. In examining the near-miss incidents over the past 3 years, there were 12 incidents where the driver applied the maximum brake (approximately four incidents per year). It is in these four cases that the speed increase may play a factor in whether or not a crash might occur. What can also be anticipated is a probable increase in crash severity should a crash occur during the test. The question then becomes, what impact does the 5 mph increase in speed have on
TCRP Web-Only Document 53 82 crash severity? Considering the 11 LRV-motor vehicle crashes reported in the past 3 years, none of the crashes resulted in a fatality, and three of the 11 crashes (27 percent) resulted in minor injuries. According to the European Road Safety Observatory, when collision speed increases, the amount of energy that is released increases as well, and part of this energy will be absorbed by the human body. When the amount of external forces exceeds the physical threshold a human body can tolerate, serious or fatal injury will occur. This is particularly true for occupants of light vehicles, when colliding with more heavy vehicles and for unprotected road users, such as pedestrians and cyclists when colliding with motorized vehicles.1 Based on this, Nilsson developed the following equation, which describes the effects of a speed increase on the rate of minor injury crashes:2 A2 = A1 (v2 / v1)2 where: A2 = the number of injury crashes after the speed change; A1 = the number of injury crashes before the speed change; v1 = the original velocity; and v2 = the increased velocity. Applying this equation to the situation at hand, A2 = 3 (40 mph /35 mph) 2 = 4 injury crashes Therefore, according to this equation, one could expect approximately 5 injury crashes, as opposed to 3 injury crashes, in a 3-year period after the increase in speed, or approximately 1.23injury crashes in the 1-year test period (as opposed to 1 injury crash per year before the speed increase). As this equation does relate to motor vehicle crashes, not LRV-motor vehicle crashes, the increase in the rate of injury crashes would be expected to be greater. In addition, absolute speed is not the sole factor that would affect crash severity. Other factors such as relative speed, angle of impact, vehicle crashworthiness, occupant restraint usage, and occupant characteristics will also impact the severity of the crash; however, in the absence of an equation specific to LRV crashes, the result gives some idea of the magnitude of increase in injury crash rates. Other Potential Impacts As noted above, other factors ranging from weather to fuel costs can impact collision rates. One notable impact to crash rates is the impact of changes in traffic volumes. In addition to the traffic volumes obtained from the City of San Jose, the research team observed traffic volumes for the left and through movements on North First St. from the field video data for each of the data collection periods. These data are relevant when comparing the frequency of âbefore and 1 European Road Safety Observatory, Speed and Injury Severity, http://www.erso.eu/knowledge/content/20_speed/speed_and_injury_severity.htm, as of September 15, 2009. 2 Nilsson, G. The effects of speed limits on traffic crashes in Sweden. In: Proceedings of the international symposium on the effects of speed limits on traffic crashes and fuel consumption, Dublin. Organization for Economy, Co-operation, and Development (OECD), Paris, 1982.
TCRP Web-Only Document 53 83 afterâ risky behaviors. It is possible that if traffic volumes are higher, the frequency of a particular type of conflict may also be higher. North First St. is characterized as a technology corridor, and traffic volumes could fluctuate substantially as the number of workers (i.e., commuters) in technology sector vacillates. It was noted that during the data collection period in June 2009, several of the adjacent office buildings were vacant, several of which had more than 1,000 parking spaces unoccupied. In fact, comparing the traffic volumes that were obtained from the City of San Jose in 2008 and 2009 to those that were retrieved from the June 2009 video observations, it appears that the traffic volumes have decreased at the Brokaw Rd. intersection; traffic volumes have generally increased at the Charcot Ave. intersection during the afternoon peak hour and decreased slightly during the morning peak hour; and traffic volumes have stayed within about 10-12 percent at the Trimble Rd. intersection. Traffic data from the City of San Jose and from the observation counts are summarized in Table 23. Table 23. Summary of Traffic Volume Data Movement: North First St. at: Counts from City of San Jose AM (PM) Peak Hour* Observation Counts AM (PM) Peak Hour* Ratio: Video Observation Traffic Count to City Traffic Count AM (PM) Peak Hour* Date 9/23/08 6/24/09 NB Left at Brokaw Rd. 106 (53) 95 (47) 87% (89%) NB Through at Brokaw Rd. 660 (330) 524 (243) 79% (74%) Date 3/24/09 6/24/09 NB Left at Charcot Ave. 71 (58) 68 (71) 96% (122%) NB Through at Charcot Ave. 447 (354) 523 (373) 117% (105%) SB Left at Charcot Ave. 37 (93) 30 (98) 81% (105%) SB Through at Charcot Ave. 305 (601) 294 (639) 96% (106%) Date 9/23/08 6/24/09 NB Left at Trimble Rd. 201 (193) 176 (187) 88% (97%) NB Through at Trimble Rd. 619 (388) 651 (403) 105% (104%) * AM Peak Hour: 8:00 AM â 9:00 AM, PM Peak Hour: 5:00 PM â 6:00 PM. CONCLUSIONS In conclusion, as would be expected, there is a potential for increased risks associated with the 1- year test of increased LRV speeds, although these increased risks are considered to be relatively minor given the historical safety-related data and the field observational data. These risks include: ï· Potential for increase in number of crashesâThe historical near-miss incident data show that there were approximately four near-miss incidents per year where the LRV operator applied the maximum brake. It is possible that if the LRV had been operating at 40 mph (as opposed to 35 mph), one or more of these near-miss incidents could have resulted in a
TCRP Web-Only Document 53 84 crash. Thus, there may be a potential for an increase of up to four crashes in the 1-year test period. ï· Potential for increase in rate of minor injury crashesâHistorical data show that 3 of the 11 LRV-related crashes resulted in minor injuries at the 3 study intersections in the past 3 years. With trains operating at 40 mph, the proportion of minor injury crashes is anticipated to increase by at least 67 percent to at least 1.23 during the 1-year test period. ï· Potential for increase in crash severityâWhile no specific functions or models were identified to express crash severity as a function of LRV speed, the laws of physics suggest that even a slight increase in LRV speed could have severe outcomes in the event of a crash. However, speed alone cannot be the sole predictor of the severity of a crash. Other factors such as relative speed, angle of impact, and driver characteristics (such as age) will also affect crash severity. In addition, the improvements being made to the intersections by the VTA prior to the increase in train speeds are expected to counteract some of these potential increased risks in a number of ways. The improvements are expected to: ï· Make left-turning drivers more aware of the arrival of a train; ï· Give left-turning drivers more opportunity to see a train approaching the intersection; ï· Provide more positive guidance to left-turning drivers; ï· Provide visual separation between the left-turn pocket and the trackway; ï· Decrease driver risky behaviors; ï· Improve driver compliance with traffic control devices; and overall, ï· Reduce the number of LRV-motor vehicle crashes.