Virgin ia. agencies.) Shepard Research. (The of the author and not necessarily those of the. Highway & Transportation Research Council

Transcription

1 FLOW EVALUATION OF PAVEMENT INSET LIGHTS TRAFFIC USE DURING FOG---BEFORE PHASE FOR opinions, findings, and conclusions expressed in this report are (The of the author and not necessarily those of the those Highway & Transportation Research Council Virginia Cooperative Organization Sponsored Jointly by the Virginia (A of Highways & Transportation and Department University ef Virginia) the Cooperation with the U. S. Department of Transportation %n Highway Ndministration ederal 1975 September 76-R6 VHTRC by Frank D. Shepard Research Engineer sponsoring agencies.) Charlottesville, Virgin ia

2

3 was, therefore, the purpose of this research to investigate the It flow characteristics during fog within the system of pavement inset traffic lights installed on Interstate 64 across Afton Mountain. phase of the project was limited to the collection of before data This the inset lights installed on the main line only. It was felt that the amount for reliability of before data should be assured before embarking on an after data and phase. The time period available for collection of before data was collection since the inset lights will be placed in regular operation as soon as limited making it difficult to accurately predict the amount of data which would possible, available for analysis as a result of the variability in the frequency of fog be for the purpose of the study. adequate is the intent of this report to summarize the data accumulated to It and evaluate their reliability. Detailed analyses of each parameter observed date location on the westbound lanes of interstate 64 overafton Mountain One chosen for data collection. This location is on the level section of the was top with a slight horizontal curve to the right. There were no inter- mountain prior to the site for a distance of 7 miles (11.3 km). changes were collected using a system of tapeswitches placed in the con- Data shown in Figure i. The tapeswitches were placed on the highway using figuration methods, depending on weather conditions. First, during clear, dry conditions, two were attached to the road surface with double-faced tape.over which 6-inch they cm) tape was plac d. During adverse weather conditions, i.e., rain, fog (15 24 the tapeswitches were attached to thin (.22" or.56.mm) metal ribbons etc., across the highway and attached on the shoulder and median. Thismethod stretched the quick, safe installation of the switches during any weather condition. allowed from all the tapeswitches were recorded simultaneously on a 4- Data chart recorder with each particular switch being identified by assigning channel voltages. By knowing the distance between the tapeswitches different the road on the speed of the chart recorder, vehicular speeds and headways were determined and by measuring the distances between impulses on the chart. placements were obtained by installing different length tape- Vehicle on the right edge of the traffic lane (lane i) and noting which switch switch combinations were PURPOSE AND SCOPE will not be made, though various comments concerning the data will he included. EVALUATION OF TRAFFIC FLOW Traffic Flow Data Collection switches placed by this method proved to be the most durable and reliable Also, adverse weather conditions. during activated.

4 FLOW EVALUATION OF PAVEMENT INSET LIGHTS TRAFFIC USE DURING FOG--BEFORE PHASE FOR visibility resulting from fog presents a very hazardous condition Reduced the highway because of the inability of motorists to readily observe pavement on covering the mountain top during periods of rain. acute awareness of the fog problem on Afton Mountain led to a decision An the Virginia Department of Highways and Transportation to install, a lighting by consisting of pavement inset lights and low level illumination lights to system motorists during periods of fog. In addition to the lighting system, a series aid installation is being made on a 5.8-mile (9.4 km) section of highway The the top of Afton Mountain. Since fog often occurs on only a portion encompassing the mountain, the installation was divided into 3 sections with each section of at points observed to most often correspond with the fog patterns. Each separated will be controlled by 2 fog detectors located at or close to the endpoints section the section and. capable of detecting 5 levels of fog density. The light inten- of sity within each section will be controlled by the fog at each detector. fog guidance system will consist of unidirectional airport runway The in the pavement edge line along each side of the highway in both directions, lights a spacing of 2 ft. (61 m) for tangent sections and I ft. (3.5 m) for with sections. In addition to the white inset lights on the main line, amber curved will be installed on the off ranps, however, only on one side. Also, a lights.section oflow level illumination lights are installed on an on-ramp. small is felt that the lighting system will aid in highway delineation and It lead to a reduction of vehicle stopages within the highway and instances of thus running off the highway. However, it is not known how the system of lights vehicles affect vehicle speeds, headways, and placement. There has been some concern will the possibility that the system will present a false sense of security to some over and lead to higher speed along with greater differentials that may in- motorists by D. Frank Sbepard Research Engineer INTRODUCTION and signs, and other vehicles. Afton Mountain, which is traversed by markings 64, often is the site of such reduced visibility because of the low cloud Interstate of. variable speed signs will be installed in an attempt to regulate vehicular speeds. crease the possibility of accidents.

5 Figure I. System of tapeswitches used for data collection. chart reco-rder used for data collection was placed in a vehicle The approximately 1, ft. (34.8 m) past the site, which eliminated any parked influence the parked vehicle may have had on traffic flow. fog problem considered here is caused by low cloud cover in the The areas. The fogs are relatively dense and uniform. However, variable mountainous conditions, fog banks, etc., do occur in the area where the highway enters fog cloud cover and during periods when broken clouds are prevalent as a result the should be noted that data were taken only during uniform fog con- It extending at least 5 ft. (152.4 m) to 1, ft. (34.8 m) in advance ditions 3 INTERSTATE 64, WEST LANE 2 TAPESWITCHES LANE CIRCI RECORDER JUNCTION BOX f TO Weather Conditions of clearing weather. of the site.

6 is very important that'the fog density be obtained as it influences It traffic flow characteristics. The density was determined by noting the number the visible centerline stripes on the pavement during daylight hours and the number of reflectorized shoulder delineators during hours of darkness. These distances of used th purposes of identifying relntive fog densities in the analysis are for It should be noted that vehicles visible are for distances greater than data. of shown, the amount depending on vehicle type and taillight size and intensity those illuminated). (when through the site at regular intervals. collection of traffic flow data during fog presented a situation in The one variable---fog density or sight distance--cbuld change instantaneously, which it difficult to detect and consider small changes. As a res!t sight dis- making were broken into five. categories; -4 ft. (-12.2 m), 5-8 ft. (15.2- tances m), 9-11 ft. ( m), ft. ( in), and 16-2 ft m) which encompassed the uniform fog.conditio s encountered during the ( flow classified were within these sight distances and are Traffic data The traffic flow and sight distance data were then cate- accordingly. presented volumes. A plot of the volume in terms of the time intervals associated.traffic, which data were collected is shown in Figure 2. Table 2 gives a summary of for speeds were obtained from lane 1 and lane 2 for cars, trucks, Vehicular tractor-trailers. Table 3 gives a summary of speeds for cars, tractor- and and all vehicles, along with the associated standard deviations and trailers Trucks were not included because of the low volume encountered. As volumes. car speeds were higher than tractor-trailer speeds, and those in lane expected, (passing) were higher than in lane i. In most ases speeds decreased with 2 4 o 9_$_D en s i ty F the.tapeswitches approximately 1, ft. (34.8 m) from the Since were sight distances and fog uniformity were monitored by driving recorders, data DATA SU ARY study. according to daylight or darkness, morning peak, off-peak, and evening gorized Table 1 gives the hours for which data were accumulated along with the peak. the time periods during which the total hours of data were.accumulated. Speeds distance, however, there were instances when speeds increased slightly sight a decrease.in sight distance as was the case for daylight hours, off-peak With and sight distances between 5-8 ft. ( m) and 9-11 ft. traffic, m). (

7 Peak 5-8 (6:3 A.M.- A.M. A.M.) 8: Peak A.M. Peak 5-8 Off Peak 9-11 Off Peak Off Peak 16-2 Off Peak 9-11 P.M. Peak (6:3 A.M.-8: A.M.) A.M. Peak (6: P.M.-I: A.M.) Off Peak -4 A.M. Peak 5-8 Off of Data. Lane i Lane 2 Total Hours Volume Volume Volume , , , , , " I TABLE TOTAL. HOURS OF DATA COLLECTED WITH ASSOCIATED TRAFFIC VOLUMES Day Clear Peak (8: P.M.-4:3 P.M.) Off Peak (4:3 P.M.-6: P.M.) P.M. 1, , ,53 1, ,185 P.M. Peak Night Clear Off Peak TOTAL 48.74

8 PEAK NIGHT OFF -i ) (LANE -- p.m. PEAK DAY < " " (LA E...o i) PEAK DAY OFF i) (LANE PEAK DAY OFF 2) (LANE 4OO 3 2 " r P.M. PEAK- DAY i (LANE 2) 16-2 Figure 2. Volumes per hour for each sight distance (I ft..3 meter). TABLE 2 Period Weekday Weekend Winter-Spring Time 39.3 % of Total %

9 o[ g o

10 summary of headway data is given in Table 4, which includes mean head- A standard deviation, and the average volume per 15 minute interval for each way, It should be noted that all headways below 58 seconds are included in the lane. table. were also obtained for 1-second time intervals up to i seconds, Headways 1-second time intervals between i and 6 seconds. Plots of the percentages and vehicles with headways in the intervals 1-2, 2-3, 3-4, and 4-5 seconds are of in Figures 3 thro.ugh 6. Presented another way, Figures 7-1 show the per- shown centages of headways less than 2, 3, 4, and 5 seconds. concerning the queuing of vehicles in each lane for various sight Data and time intervals are shown in Tables 5, 6, and 7. for the off-peak distances off-peak day, and p.m. peak day, respectively. Specifically,. data on the night, with headways below 58 seconds from which queuing was obtained along with volumes number of queues per i vehicles, the average number of vehicles in each the and the average queue speed are given in the t bles. Queuing as shown in queue, tables was a direct function of the headway cutoff times (i, 2, 3, 4, and those seconds), i.e., for a 3-second cutoff time all vehicles (2 or more) with a head- 5 of 3 seconds or less was considered as a queue. Therefore, as the cutoff time way to 5 seconds the number of queues and vehicles per queue increased. increased queuing within 1-second headway intervals up to 5 seconds is Vehicle in Figures ii through 16 for various sight distances. These graphs show shown was a higher, queuing rate under restricted sight distances than for clear there however, this trend changed as the headway intervals increased to the conditions; Tires on right side (as viewed by driver) of vehicle were 3 feet (.91 m) or (i) from the edge line, and less summary of the placement data is given in Table 8 where the number of A and tractor-trailers positioned -3 ft. (-.91 m) and 3 ft. (.91 m) or more cars the edge line are shown. Figure 17 shows the percentages of vehicles posi- from from -3 ft. (-.91 m) from the edge line for the off-peak periods. As tioned in Figure 17 cars tended to drive further from the right edge line for noted distances in the 5-8 ft. ( m) range than during clear conditions; sight however, within the 9-1 ft. ( m) range this was reversed. In the Headways Vehicle Queuing there were more queues within the 1 to 2 second headway interval for all that conditions. It is also noted that for the 1 to 2 second headway interval sight 4 to 5 second headway. Vehicle Placement placement of vehicles relative to the right edge line of the traffic The was obtained for two placement categories as follows: lane (2) Tires were 3 feet (.91 m) or more from the edgeline.

11 Z o co

12 "1 1 t 1 1 oh

13 sec 2-3 sec ) ',, 1-2 OFF PEAK DAY 2) (LANE _- -,,,.,",/a Clear DISTANCE (FT.) SIGHT Figure 4. Percentages of headway in lane 2 during daylight off peak hours sec., %, sec.a " see,. -i...'"....'.

14 > i ; ].2-15 Clear DISTANCE (ft.) SIGHT 5. Percentages of headways in lane 1 during nighttime off- Figure 1-2 sec. 3-4 sec (ft.) il)istance SIGHT HEADWAYS (LANF 1) -,4 -r peak hours (i foot.3 meter). 25 " 4-5 sec. 6. Percentages of headways in lane 2 during nighttime Figure hours (i foot.3 meter). off-peak

15 DAY, OFF-PEAK- LANE i. I second Clear DISTANCE (ft.) SIGHT 7. Percentages of headways less than time shown (i foot.3 meter). Figure DAY, OFF-PEAK- LANE 2 4 seconds i 3 Clear DISTANCE (ft.) SIGHT ' o-o1 second Figure 8. Percentages of headways less than time shown (i foot.3 meter). 13 4O 3O 5 seconds- 4 seconds 2 seconds i 2 seconds 4 3O 2 seconds i seconds oooo o,8,6ooo olo

16 NIGHT, OFF-PEAK LANE I i second Clear 9. Percentages of headways less than time shown Figure foot.3 meter)..(i 14 5O 4 seconds seconds 3 seconds 2O i seconds SIGHT DISTANCE (ft.)

17 NIGHT, OFF-PEAK- LANE 2 seconds --.-_._ 5 seconds []' seconds -' -- O 3 seconds i second A Clear SIGHT DISTANCE (ft.) i. Percentages of headways less tha6 time shown Figure foot.3 meter). (I

18 r 16 OD.

19 17 o. r o o r r- oo u' O co o c. O O

20 o oz> -

21 o o D O OFF-PEAK LANE I Sight Conditions DAYLIGHT, O Clear I INTERVAL (sec.) TIME DAYLIGHT, OFF-PEAK LANE 2 Sight Conditions O Clear / \ 5-8 feet " J "-----O 5-8 feet O _ ) 9-ii feet // Z feet // i ii. Vehicle queuing for headway intervals shown Figure foot.3 meter). (i 2 15 \ O 9-11 feet / feet Z i..." I ferval (sec.) TIME 12. Vehicle queuing for headway intervals shown Figure foot.3 meter). (i

22 o o. 2 o o NIGHT, OFF-PEAK- LANE i 1, I INTERVAL (sec.) TIME 13. Vehicle queuing for headway intervals shown (i foot =. Figure meter) Vehicle queuing for headway intervals shown (i foot Figure meter) i,- -'''....' ft. //.-" ft. 5-8 lear ft NIGHT, OFF-PEAK LANE 2 i TIME INTERVAL (sec.)

23 P.M. PEAK LANE I Sight Condition DAYLIGHT, O Clear i INTERVAL (sec.) TIME P.M. PEAK- LANE 2. '._ O Clear // X "".X\ O O9-11 ft. //." f t. "'%.. /...' i INTERVAL (sec.) TIME. 16. Vehicle Queuing for headway intervals shown (i foot Figure meter) ft..t. l f l- I o Vehicle Queuing for headway intervals shown (i foot Figure.15. meter) I. DAYLIGHT, S ght Condition //// \ o

24 O oh o o.h o.h o o o C.).r'l Ha-)

25 = 5 25 CL ,SIGHT DISTANCE (ft.) 17. Placement of vehicles from -3 feet from edge line for off-peak Figure periods.

26 ft. ( m) and 16-2 ft. ( m) sight distance ranges was a tendency to approach the placements observed under clear conditions. there tractor-trailers, 93% of the trailers were less than 3 feet from For right edge line during clear weather conditions, however, this figure de- the to 4% for visibility distances of 5-8 ft. ( m). As the creased distance increased from 9 to 2 ft. (27.4 to 61. m) the percentages sight accident analysis was made of all accidents occurring from December An 1972, when the highway was opened, through February 28, Only accidents 22, the 5.8-mile (9.4 km) section of highway on which the pavement inset lights within installed was considered. Table 9 is a compilation of all data from arebeing the 2.16-year survey. Accidents 64 Total Accidents 18 Injury killed 1 Persons conditions Weather Clear 26 i) Cloudy ii 2) in-depth an@lysis was made of the 9 accidents occurring during fog An and is summarized in Table i. Also comments concerning each acci- conditions on Road Ice on Ice Skid increased to those, found initially for clear conditions. Accident Analysis TABLE 9 SUMMARY OF ACCIDENT ANALYSIS Ice, snow or sleet 17 3) Rain (no fog) 7 4) Fog 9 5) 36 Daylight Darkness 28 dent are listed. TABLE i SUMMARY OF FOG RELATED ACCIDENTS Total Accidents Accidents Injury Persons Killed Daylight Darkness.5

27 stopping on (2) Vehicle turning from center crossover and struck by vehicle. Vehicle stalled on icy bridge hit by two other vehicles. A short time later (5) vehicle skidded across ice onto vehicle involved in previous accident another (6) Vehicle hit truck when attempting to pass. Vehicle ran into patch of fog and driver lost control. Charged with reckless (7) driving. was difficult to determine how many accidents were a direct result of It however, many of the incidents were obviously caused by restricted sight con- fog, numerous variables associated with this project, coupled with problems The in data collection and the time frame in which it was possible to collect encountered have limited the quantity of data that have been collected for analysis. theo.data, the sufficiency of the data, in terms of quantity, that have been Therefore, for this before phase is questioned. It should also be noted that accumulated the data anticipated and thought possible to collect under the work plan although be inadequate for the purpose of deciding the success or failure of the pro- would.lighting system it would add to the data base concerning traffic flow during posed noted earlier, data were broken down in various Gategories depending As sight conditions, time of day and certain vehicle characteristics. The short on periods associated with the a.m. and p.m. peak periods along with the infre- time of fog with various sight conditions limited the available data in some quency whereas in other categories there are substantially more data. An categories, will be made to determine the reliability of the available data in con- attempt Vehicle stopped in traffic lane after being lost in fog and was struck by (!) vehicle. Driver was 84 years old (daylight); charged with unlawful another highway. Vehicle stopped on roadway and hit by.two others. Road covered with ice; (3) 7 years old. driver (4) Vehicle went out of control on ramp. Driver charged with reckless driving. (3: a.m.). Vehicle skidded (8.) icy bridge. on Vehicle skidded on icy bridge. (9) ditions. RELIABILITY OF DATA conditions for various types of delineation, in this case pavement inset lights, fog which very little data are available. for an after project phase, using statistical estimates along with engineering sidering judgments.

28 estimate of the sample ize required to detect prescribed differences An averages was used in determining the adequacy of the speed data. It should between noted that it was assumed that the standard deviation was the same in the before be after phase. By assuming a difference in the average before and after speeds and sample size needed for various levels.of confidence may be estimated. Table ii the the sample sizes required for statistical significance for i, 2, 3, 4 and 5 gives to Table ii for conditions of darkness during the a.m. peak, Referring speed differentials in the 3-4 mph ( meter/sec.) range would have to average found between the before and after phase to obtain valid comparisons. However, be look at the data acquired within the off-peak indicates that speed differences a the 2-3 mph ( meter/sec.) range would be sufficient for statistical sig- in Table ii shows a 4-5 mph ( meter/sec.) average speed differential nificance. for reliable comparison for the a.m. daylight peak, however, the off-peak necessary p.m. peaks would require only 1-2 mph (.4-.9 meter/seco) differences. and same procedure was used to determine the sample size needed for sig- The headway results as was used for the speeds. For data taken during off- nificant periods of darkness (Table 4), after data would have to reflect mean headway peak in the 4 to 6-second range for results which would have high confidence differences associated with statistical analysis. Before data taken during darkness for levels a.m. peak periods is insufficient for any statistical analysis because of the the time periods (Table 4) during daylight hours, average headway differ- For would have to be in the 2 to 3-second range for lane i and 3 to 5-second ences reference to Figures 3 and 5, there seems to be definite trends in With percentages of headway within lane i for each sight condition in the i to 2- the 2 to 3-second and 3 to 4-second time intervals. The plots in Figures 4 second, 6 for lane 2 were more erratic because of the smaller value as compared to and i. Also, Figures 7 through i show definite trends in the percentages of lane less than i, 2, 3, 4, and 5 seconds. headways data for nighttime off-peak, daylight off-peak, and p.m. peak hours Only included because of the small amount of queues associated with the low volume were data collected within the other time periods..referring to Figures ii through of definite trends emerge when comparing the queuing rates for various time inter- 16, for the different s±ght distances..the queuing rates are generally higher in vais 1 to 2-second time. interval for all fog sigh t conditions than thosefound_for the 26 Speeds differences in the before and after average speeds for.5 and(l- B ) mph.8,.9 and.95. Only those sample sizes ace included which are within the.7, size obtained in the before phase, as statistical significance could not be sample with the data if the sample size required was larger than those shown. associated Headways 8mall sampleosizes range for lane 2. Vehicle _Q clear conditions.

29 IIII

30 to Table 8, it is seen that most of the vehicle placement data Referring obtained during the daylight hours with that gathered during the off-peak were the highest. The table shows that placement data (off-peak, daylight hours) being obtained from 518 cars during clear sight conditions, 463 cars for sight dis- were of 5-8 ft. ( m), 68 cars in the 9-11 ft. ( m) range, tances cars in the ft. ( m) range, and 55 cars in the 16-2 ft. 1,174 ( m) range. would be difficult to develop a before-after accident analysis as It years of after data would be desirable, and an after phase would be completed 2 felt that an in-depth analysis of accidents in an after phase would be bene- is as a means of determining any influence of the guidance system ficial the on it was the intent of this report to summarize the data accumulated As date as the before phase of the project and evaluate their reliability, con- to Speed data collected during the off-peak periods, during both daylight and i. conditions, along with data collected during the p.m. daylight peak darkness sufficient for statistical analysis with high levels of confidence seems.5, i- B.9-.95). ( Although statistical analyses would be difficult for the majority of headway 2. collected because of the high standard deviations, definite trends data are when plotting headways as a function of sight conditions for the apparent and daytime off-peak periods..night- Sufficient data concerning vehicle queuing were collected for darkness off- 3. and daylight off-peak and p.m. peak periods only. Plots of queuing peak as function of time intervals show definite tendencies when comparing queuing a clear conditions as opposed to limited fog sight distances. during Vehicle placement data appear to be satisfactory only for.the daylight off- 4. hours. peak A valid before-after accident analysis is questionable because of the limited 5. related data; however, it is felt that fog in-depth investigation of any an related accidents in an after phase would be desirable to determine any fog interrelationships between the guidance system and the accidents. possible 28 Placement Accident Analysis tothat time. Also, the number of fog related accidents within the 2 year prior period would probably be inadequate for statistical analysis. However, it after accident experience. CONCLUSIONS clusions will be limited to these aspects of the project.

31 phase study, however, certain limitations are recommended as follows: All data would be obtained from the one location considered in the before (i) phase. Data would be collected only for off-peak periods during both daylight and (2) e.g., between the hours of 8:3 a.m.-4:3 p.m. and 6:3 p.m.- darkness, 29 RECOMI ENDATIONS on the data collected and the analysis made, it is felt that Based information is available for consideration to be given for an after sufficient 12: p.m., with only weekdays being considered. One year should be allowed for collection of data during the after phase for (3) guidance system involving the pavement inset lights only. the

32