Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.
OCR for page 35
35 Chapter Four FIELD DATA COLLECTION One of the major objectives of this study is to assemble a comprehensive database of traffic operations at stop- controlled intersections. A major investment of resources was made to collect data at a wide variety of intersections. This chapter describes the data collection process and Me database that were collected during this study. It includes a summary of the data requirement by venous models, sampling plan, a summary of We site characteristics, Me data reduction process, critical gap and foDow-up time estimation, capacity and delay estimation from Me field. DATA REQUIREMENTS FOR RECOMMENDED MODELS The data requirements for each of Me capacity and delay models are summarized In Table 4. SAMPLING PLAN Site and geometric considerations were established as Me basis for the sampling plan used to guide the selection of candidate sites. Since capacit,~measurements were highly desirable, the first criterion was Mat continuous queueing existing on at least one stop-con~oDed approach for at feast five minutes on a normal day. Second, areas In each of five geographic regions of the United States were i ~ e n ~ fi e ~ : S a n F r a n c i s c o / S a n ~ o s e , C a ~ i f o r n ~ a m e ~ o p o ~ i t a n area (southwest sector), Montgomery and Auburn, Alabama (southeast sectors, Troy and Rochester, New York (northeast sector), Milwaukee, Wisconsin (central . . .. sector), and Portland, Oregon (northwest sectors. A guideline of fourteen TWSC intersection sites In each geographic sector was established. within each geographic sector, guidelines were established for numbers of sites wad three different geometric configurations. These are shown In Table 5. Other site criteria that were used in screening of potential sites are listed below. Presence or absence of upstream traffic signals. Upstream traffic signals produce platooned flow that affects the capacity of a stop-controDed intersection. The study sample included some sites Mat were affected within I/2 mile of an upstream signal, so that the effects of platooning Table 4. Data Definition Intersection Geometry Data ·Nllmber of lanes per approach Diane designations ·Grades · Sight distance for non-pnor~ty vehicles Traffic Flow Data ·Traffic flow rates for all movements ·Tr~ic stream composition for all movements ·Proportion of non- platooned vehicles for each priority stream movement · Capacity for each non-prior~ty movement ·Length of peak period Gap "d Headway Data uncritical gap for each non-prior~ty stream ·Follow-up gap for each non-priority stream ·Minimum headway for platooned priority stream Raw Data Required for Model Estimation For Each Vehicle ·Event time at conflict point ·Movement direction ·Vehicle type ·Lane used Additional Data for Each Non-Priority Vehicle ·Event time at back of queue ·Event time et front of queue ·Event time departing queue Computed or Derived Data Required for Model Estimation For Each Traffic Stream ·Flow rates ·Stream composition For Each Non-Priority Stream Vehicle ·Gap event history while et front of queue · Measured or estimated capacity · Service time or delay ·Queue delay aTota1 stopped delay For inch Priority Sbeam Vehicle ·Propofion offree vehicles ·Minimum headway of platooned vehicles . on capacity could be quantified. Major street speed. The average vehicle speeds on die major street may or may not affect the gap acceptance behavior of minor street vehicles. The study sample included some sites with major street posted speed limits of at least 45 mi/hr.
OCR for page 36
Configuration #1 Geometry · Lanes on major street = 1 Lanes on minor Swat = 1 Optional major street exclusive left-stun lane Planned Number of Sites · Urban area = 3 Rural area= 1 Total sites = 4 Configuration It2 Geometry · Lanes on major sheet = 2 Lanes on minor street = 1 Optional major street exclusive left-sum lane Planned Number of Sites · Urban areas = 6 Rural areas = 1 Total sites = 7 36 . . One-way streets. One-way major streets may affect gap acceptance behavior differently than two-way major sweets. Major street median'. A median storage area or a two-way lefc-turn (TWLT) lane on the major street may result In a two-stage gap acceptance process for the minor street driver. This means that Me minor street vehicle crossing or boning left performs Me movement In two stages, using the median refuge to store in between the two stages. Table 5. Sampling Plan for Two-Way Stop Controlled (IWSC) Intersection Sites . Single lanes with no queue by-pass. Many stop- controlled intersections provide an opportunity for drivers on major street through movement to by- pass amajor street left turn queue on the right by using a paved or unpaved shoulder. This opportunity increases the capacity of the approach without the addition of a complete lane. The study sample included some sites that allow for this queue by-pass and some sites that precluded it. Configuration #3 Geometry Lanes on major sheet = 2 Lanes on minor street = 2 Optional major street exclusive lef`-tuin lane Planned Number of Sites · Urban axe" = 3 · Rural areas = 0 · Total sites = 3 A total of 68 unique sites were videotaped during 79 videotaping time periods. Overall, the sample size objective of 70 TWSC intersections was met. Tables 6 and 7 show the number of sites that were 1 1~ ,. ~ All videotaped by sector and by geometric configuration for TWSC intersections. The sampling plan requirements are also shown. In most cases, Me requirements were either met or exceeded.
OCR for page 37
37 Table 6. Number of Unique TWSC Intersection Sites 1 , .......... ............ .......... ............ ............... ................ ................ ................... ... ... ............. ........ ,., ............... .. ..... ..... ....... , ..,~. ,. , ~" ' . ~:"'' ' "'a' ' ' " . ~"" . ~ .. 1 5 4 4 2 4 5 6 3 6 3 3 15 12 13 . 4 6 3 13 6 4 23 20 5 7 26 35 4 3 19 15 15 14 68 70 Table 7. Number of Videotaping Periods-TWSC Idtersechon Sites ................................................ . ,., - ............................... ~., :' : ........ - ...... - ,.,., ,. .... ................. ,.,.,. _ , , ,.,, ............... ., ,,,, . ~. ~., ~. ~.. ~.......... ~ .. .. .. . . . .. .. ....... .... ... .. 7 4 5 5 7 4 28 20 5 5 9 6 5 7 30 35 7 3 3 4 4 3 21 15 19 12 17 14 14 14 79 70 ................... .~ . . ............ Tables g and 9 show the number of sites categonzed by both geometric configuration and poster! major street speed There is nearly an equal split In Me number of sites avid major street speeds less Wan 40 mph and greater Man 40 mph. Table 8. TWSC Intersection Unique Sites: Geometric Configuration by Major Street Speed I. ; . ~.~.~ 2 3 Total 10 10 6 26 8 9 6 23 1 1 111 11111...11111 ·:-:-:-:-:-:-:-:-:-:-:-:-:-:-:-:-:-:-:-:-:-: ...................... 2 .: . ~P ~.S.: ...................... · r- Table 9. TWSC Intersection Videtaping Periods: Geometric Configuration by Major Street Speed - 2 Total 4 2 1 7 14 11 7 3' 8 11 6 i, ~ , , _ , ~ , 6 1 Tables 10 and ~ ~ show the breakdown of Me number of sites by geometric configuration and the distance to upstream signal control on the major street. Since the effects of vehicle progression on capacity are significant, it is important to have a mix of sites with varying distances to upstream signal conuo} and Bus varying degrees of progression and/or platooning. Table 10. TWSC Intersection Unique Sites: Geometric Configuration by Upstream Signal Control ............................................... ..... .... ............. .......... ~,.~.~0 ,~,.~F it' . . 2 3 Total 12 27 13 6 24 10 17 Table 11. TWSC Intersection Videotaping Periods: Geometric Configuration by Upstream Signal Control 2 3 Total 10 9 14 33 ;.;.;.;.;.;.;.;.;.;.;.;.;.;.;.;.;.;.;.;.;.;.;.;.;.;.;.;.;.;.;.;.;.;., ., .. .. ~ ......... - · ''''' 1 s 14 6 25 21
OCR for page 38
SITE CHARACTERISTICS This section contains a summary of the traffic data for Me TWSC intersection sites, see Tables 12 through 16. Please also refer to Working Paper 15 (NCHRP 3-46, 1995) for a complete reference of We physical characteristics of We intersection sites. The data is presented In terms of minimum, Table 12. TWSC Sites Traffic Characteristics SummarY-Sou~west Sector SWTOO1 SWT002 SWT003 SWT004 SWTOOS SWI006 SU'I007 SWTo08 SWT009 SWTO10 SWTO11 SUlT012 SWI013 SWI014 SWT015 SWT016 SWT017 swro~a SWIO19 maximum and average values for 5-minute data points. It is instructive to note the w~de range of flow rates, queue percentage, and delays ~at were exhibited in the field data for each site even dunag one videotape penod of approximately I.5 hours. Practitioners end researchers should be cogn~zant of this vanabilit~r at any particular site dunug a g~ven peak penod when interpreting the accuracy of modeled results against field data. 1 | SB 3 T SB T 2 | EB | WB 2 T B 1 1 EB 3 1 SB 3 l EB WB(Lanel) . WB(Lanc2) . 1 NB 2 NB SB . 1 WB 3 WB 3 EB WB0~l) WB(~2) . EB . 3 NB(La~cl) NB(L9~2) 3 NB~1) NBC1~2) ~ - 24 ~SB -~ 1 NB 1 NB _ 2 SB 01:48 01:S8 O1S4 01:48 01:S8 01 39 01:46 . . 01 S6 01:48 01 S3 00:40 01:4S 01 S2 01:46 01-22 01:41 01 Sl O1S9 01:47 . 1812 10321421 22S6 128418S7 = = 102O ~2372739 = = 1944 11761S27 - 372776 1836 11281412 864 348S42 ~ -~- ~ ~ = 1380 S28938 17S2 82812S7 = 10DO 624824 1584 996ll79 1716 7801310 = . 1668 12~1~6 3432 17282461 2124 9361S08 = . 02S24 684~ 1689 104 624819 1248 492919 2340 14881879 _ 180 216 408 168 48 276 1ao 408 84 36 372 S64 132 1S6 492 288 1S6 96 132 192 S64 600 S28 360 168 492 648 264 48 60 168 o o 48 ~2 24 12 o 24 228 12 24 240 96 24 o o 48 300 1S6 1S6 48 12 288 276 . 60 . 121 142 299 4S 13 113 46 166 41 16 108 379 42 79 340 177 89 34 37 104 1 1 434 338 339 1S2 68 1 392 438 147 2.8 12.2 1 2S 1.2 o.7 o.7 13.4 1 1.3 2.9 8.4 4.8 1 1 3 8.9 28.8 S.2 9.3 3.1 23 11.6 2~7 3.1 12.8 313 13 12.8 S3 10.1 1S.8 4.3 48.2 1 6 9.7 18.6 8.6 1 1 10.7 12S 1 1 6.7 28.2 72.2 1 9.9 36.8 6.2 13.2 34.9 1 1 ~.1 1 13.1
OCR for page 39
39 Table 13. TWSC Sites Traffic Characteristics-Southeast Sector B~:1) B(~2) WB(~1) WB(~2) B WB SBCl~I) SB~2) SIB amp) NBO~l) NBC[~nc2) NB0-anc3) WB EB(I~ncl) NB SB WB(19~1) WBO~anc2) Bowl) EBC~nc2) EB WB WB WB(~1) WB(1~2) EB . S4.2 21.2 S4 11.? 30.1 31.4 33 36 18.9 39.S 20.3 9 44.8 6.6 21.2 43.2 41.8 273 99.3 68 19.6 18.7 47.6 115 16.7 _ 3.3 26.8 l 8 2.4 183 _ 5.6 2.4 11.2 2.5 12 3.8 12.2 3.2 12.1 0 4.1 3.3 13.4 4.4 9.4 O 1.1 0 5.4 2.2 S.2 33 11.3 3.7 12.6 3.9 23.2 3.1 8.9 1S 23.S .9 16.8 3 10.7 4.1 9.5 3.8 15.6 0 5.7 4.9 12.1
OCR for page 40
40 Table 14. TVJSC Sites Traffic Characteristics Summar~,r-Nor~east Sector NEIZO1 NET202 NE=03 NE=Z04 NEIZ05 NElz06 NE=Z07 NETZ08 NETZO9 NET210 NET211 NEl-212 NEl213 NUlZ14 NE=zl5 NET216 N~17 2 1 SB 1 01 So 2 1 WB(L9nel) 1 01:44 1 WBC~2) 1 2 WB(19r~l) 01:04 WB(1~2) 2 SB Ol:S9 2 SB 0151 SB(~1) 01 OS SBC - e2) 1 SB Ol:SO 2 NB 01 :S1 1 SB 01:41 2 SB - - 01 ·SO 2 EB 01:47 3 _ EBCl~l) 01 SS EBC1~2) 1 EB 01:26 WB ~ 2 SB 01.47 3 ~NB0~l) ~ 01:47 NB(I ~2) ~ 1 EB 01:41 1 ~EB 01 :S3 4020 828 684 4368 2484 1908 1896 3984 972 21 12 1128 1740 1044 18Q0 1908 408 480 21.1 ll S 4 65 2.8 21.1 613 125 12.2 17 18.1 18.8 12.6 8.8 19.6 S.2 lO.S 14.4 16.1 12 7.8 8.4
OCR for page 41
41 Table 15. TWSC Sites Traffic Characteristics Summary-Cenhal Sector
OCR for page 42
42 Table 16. TWSC Sites Traffic Characteristics Summarv-Nor~west Recta, NWT401 Nurr402 Nwr403 NWI404 NWT405 Nwr406 Nwr407 Nwr408 NWT410 NWT411 NWT412 NWT413 NW r414 Nwr415 Nurr416 WB EB NB NB SB NB NB SB WB(I~cl) WB(Lsuc2) EB NB SB EB(~1) EBO~nc2) WB WB WB EB WB EB NB(~el) NBO Am) SB DATA REDUCTION PROCESS Four video cameras were used to record traffic operations at each of the TWSC intersections. Data were reduced from the videotapes using the Traffic Data input Program (TDIP, Boesen, Rin~esbacher, and Kyte, 1991~. TDIP is a computer program Mat records time events by observing a videotape and pressing a corresponding key on a computer for a particular event of interests For this study, the following events are important: We time that the vehicle passed through the conflict area, the time a vehicle entered a queue, and the time a vehicle reached the first position of a queue, the time a vehicle exited from 14.9 ao.2 S9.8 103.8 ~2 80.2 100 14.8 41.8 24.2 140.7 73.7 o 22.7 13.9 13 4S.6 282 94.? 33.8 44.5 303 O.1 14.S 3 179 1 12T6 1 63 T 19-1 13.4 136 11.8 22.6 10 33.1 13.S 44.? 6 10.2 4 1S.9 53 13.7 S.7 46.6 8 27.8 O O 83 15.8 3.6 6.6 5.8 99 6 ~ 14.7 6.9 14.8 IS S S2.3 1.1 16.9 S.4 13.8 3.? 1S.4 1.1 7S 1S 6.? the stop line. In addition, the vehicle type, turning movement type, lane usage are also recorded Once these data were recorded by TDIP, a data file is assembled using a spreadsheet for further data reduction Table 17 gives the format of the raw data file. Further data reduction exacted traffic flow parameters such as flow rate, delay, critical gap, follow-up time, and queue length. Table 18 illustrates the format of part of We summary data for one of the minor street approaches. These data are directly calculated from We raw data file of each site. Notice Rat critical gap cannot be obtained directly Dom We raw data file. Special procedures have to be used to estimate critical gap.
OCR for page 43
43 Table 17. Raw Database Format for I WSC Intersections 00:10:04 , ~ ~ ~, ~ ~ ~ . B' .... ,,,, . .... ...... , , .,,,, , .,, ... . ................................ ~~ t: ........ ........ ............ ............... . ~- 00:10:04 WBTH 1 1 00:10:08 WITH 2 1 00:10:09 NBRT 1 2 00 09 32 00:10:13 EBRT 1 1 00:10:14 WBLT 1 1 00:10:13 00:10:16 WBLT 1 1 00:10:14 00:10:23 EBRT 1 1 00:11:02 WBLT 1 1 00:11:00 00:11:04 EBTH 1 1 00:11:05 EBTH 1 1 00:11:06 EBBS 1 1 00:11:07 EBTH 1 1 00:11:10 EBTH 1 1 .;;;;;;;;; ;;;;;; ;.;;;;;.;;;; :::::::::::::::::::~:~:::::::::: ..................... I 00:10:08 00:10:13 00:10:14 00:10:13 00:10:14 00:11:00 00:11:00 Notes: PassTime is the time the vehicle passed the conflict point, LaneUse is the lane number the vehicle used; VehType is the vehicle type; EnterQ is the time the vehicle entered the queue; FirstQ is the time the vehicle reached the first position in the queue; E~tQ is the time the vehicle existed from the stop line. EnterQ, FirstQ, and ExitQ are recorded only for minor stream vehicles. Table 18. Example of Summary Data Format Collected for Each Intersection - ., , ... ,, ,., , ,,,. , .... , ., ...... ,.,. .,., , , _ _ ~ . ,, ~,.~,,,, , ~ ' . ~' .~ . ~, , . ~., ~ 1 216 0 184 400 33.8 7.8 41.6 3.9 2.6 100 4.6 0 2 200 0 176 376 25.0 7.3 32.3 3.4 2.3 80 3.3 0 3 216 0 172 388 32.3 7.7 40.1 3.4 2.6 98 4.7 0 4 188 0 108 296 20.6 8.9 29.S 2.9 2.2 79 2.3 0 5 176 0 104 280 4S.1 11.8 S6.9 3.6 3.S 95 5.3 0 6 212 0 88 300 64.8 10.4 75.3 3.6 2.6 100 5.8 1 7 2Q8 0 152 360 40.9 8.9 49.8 3.9 2.9 100 S.2 0 Notes: Ti is the time period; Ad, VTH, and ~ are the traffic volumes for left tum, through, and right turn movements; AD, SD, and TD are delays in queue, in server (stop line), and total; Tt and To are average follow-up time and move-up time; %Q is the percent time where the approach has a continuous queue; AL is the average queue length; °/~;ITK, ALTO, and °/`iMTR are the percentages of heavy bucks, light bucks, and motorcycles; °/ciLT, PITH, and SORT are the percentages of leR tum, through, and right turn movements. ...~Ho~TE S o 2 3 1 1 o .................. ! = -. ' .. ~ ! ..~., ~, .,~,0,~,n, ', :,Y,.,O,W,. , o o o 1 3 o o S4 S3 S6 64 63 71 S8 o o o o o o o 46 47 44 36 37 29 42
OCR for page 44
44 CRITICAL GAP AND FOLLOW-I]P TIME ESTIMATION Defining Critical Gap Events To use the maximum likelihood method to estimate the critical gap, the accepted gap and the maximum rejected gap of each vehicle must be known. These data need to be extracted Dom the raw data file prepared in the data reduction process. It Is important to correctly define the gap events that reflect a driver's behavior at TWSC intersections. The passage time of any vehicle that conflicts directly with the subject vehicle can be defined as a begin gap event. If it is a "lag", the begin gap event is the arrival time of the subject vehicle at the stop line (f~rst-in-queue timed. The end gap/lag event must be a major street vehicle that conDicts wad the subject vehicle. The end gapAag event was defined as such because only the major street vehicles were observed to affect the minor sheet driver - specifically, what actually determines a driver's decision whether to enter He intersection is when the next major street vehicle wall active at the intersection. Vehicles on He opposing minor street generally do not exhibit clear priority over He subject approach, no matter what kind of movement they are making. Rather, drivers were observed to enter the intersection on a first-come-first-serve basis. Three examples to illustrate the importance of correctly defining He key events are shown in Table 19 and In Figures 19 Trough 25. In Example I, the subject approach vehicle (SBET) reached He stop fine at 00:10:48. It rejected several gaps before accepting the final gap. The rejected gaps include EBTH(a lag), EBTH-WBET, WBET-EBTH. The accepted gap was NBTH-EBTH, which equaled 4 seconds. Note that the vehicle NBTH is included in defining He begin gap event (~e accepted gap), but would not be included in defining He end gap event. Figures 19 through 25 illustrate He gap events discussed in this example. Ensample 2, anEBTH vehicle (00:10:58) was defined as an end-gap vehicle rawer Han the SETH (00:10:58), because only vehicles once major street can be defined as an end-gap vehicle. Example 3 shows a situation where a "lag" exists.
OCR for page 45
45 Table 19. Examples to Illustrate Begin and End Gap Events for Maximum Likelihood Method for Critical Gap Computations l Example 1: *Sou~bound is ~e subject approach PassT'me Movement EnterQ FirstQ Ex~tQ 00:10:50 EBTH 00:10:51 WBLT 00:10:51 00:10:51 00:10:51 00:10:52 EBTH Be~ Gap 00:10:54 NBTH 00:09:31 00:10:48 00:10:54 00:10:55 SBLT 00:08:42 00:10:48. 00:10:54 End Gap 00:10:58 EBTH 00:10:59 WBTH 00:11:00 EBRT 00:11:02 EBRT 00:11:03 EBTH Example 2: *The subject approach is northbound. PassTime Movement EnterQ FirstQ ExitQ 00:10:50 EBTH 00:10:51 WBTH 00:10:51 WBLT 00:10:25 00:10:26 00:10:51 00:10:52 WBTH Begin Gap 00:10:55 SBTH 00:09:31 00:10:30 00:10:54 00:10:55 NBLT 00:08:42 00:10:24 00:10:54 00:10:58 SBTH End Gap 00:10:58 EBTH 00:11:00 EBRT 00:1 1:02 EBRT 00:11:03 EBTH Example 3: *The subject approach is nor~bound. PassTime Movement EnterQ FirstQ ExitQ 00:09:45 WBTH 00:09:50 WBTH Begin Lag 00:10:55 NBLT 00:10:24 00:10:24 00:10:54 00:10:58 SBTH End Lag 00:10:58 EBTH 00:11:00 EBRT 00:1 1:02 EBRT 00:1 1:03 EBTH
OCR for page 46
46 0) ~ ·cl MILT e=_ ~ be ~Ail;_ ·1 10:41. ~ MILT Be_ a IRAQI. SALT nor ~ W ~ ~1 1: ~ 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 rT 10:48 lO:S3 10:SS 11:03 10:48 10:S3 10:S8 11:03 Figure 19. Gap Event Example Figure 21. Gap Event Example ~131U yam ·' 10~. "I.T Sect · 4; (~) ~ or_ e lo:SZ. BALI ~1 e IIC i ~ ;~ ~ 1 1 1 1 1 1 1 1 1 ~1 1 1 i ~1 ~ 1 1 ~ ~ 1 1 ~ 1 1 1 1 1 ~ rT 10:48 lO:S3 10:58 11:03 10:48 lO:S3 lO:SS 11:03 .... . . ~ Figure 20. Gap Event Example Figure 22. Gap Event Example
OCR for page 47
47 (I == ~ ~ 10:54. TO ~ Cot ~ ~ ~ ~ gap of SBLT . 1 (- 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 10:48 10:S3 10:S8 11:03 Figure 23. Gap Event Example (6) SBLT pow at lO:SS, a¢coptod ~ IMP Of ~ coconde. NBTH def m" the beam gap Ad ElBTH defin" the "d gap . Hi' rl I I I I I I I I I I I I I I 10:48 lO:S3 lO:S8 11:03 Figure 24. Gap Event Example (7) BB1.H pesece at tO:SS. defying to cad gap event . ;~ ~ 1\; ~ 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 10:48 lO:S3 lO:S8 11:03 Figure 25. Gap Event Example Computer software developed during this research project was used to extract these gap event data. First, the software generates a Its file in which all the gaps rejected and accepted by each minor street vehicle are listed, along with some ofthe attributes associated with each gap event. These attributes include the minor street vehicle type, turning movement type, and lane usage; the begin gap vehicle's turning movement, vehicle type, and lane usage; the end gap vehicle type, turning movement, and lane usage; Me accepted and rejected gap size; the maximum rejected gap; the number of rejected gaps; the number of rejected gaps Mat are larger Man the accepted gap size, and the minor street vehicle's delay. The software further processes1he data file and extracts all of the accepted and the max~m~nn rejected gaps based on selected criteria, such as~rningmovement,vehicleWpe, and delay. An input file is then generated for a computerized maximum likelihood estimation program developed by Troutbeck (1992~. The software provides a report of the mean critical gap, the standard deviation, and the number of observations. The software processes the critical gap estimation for a specified period at a site or across different sites where similar intersection attributes exist.
OCR for page 48
48 Table 20. Example of Exbachug Follow-Up Time Data ...................... ; ...,, e.......... 00:39:31 00:39:34 00:39:53 00:39:58 00:39:58 l 00:39:59 00:40:00 00:40:04 00:40:06 00:40:09 00:40:14 00:40:17 00:40:20 00:40:24 00:40:25 00:40:27 00:40:30 00:40:35 00:40:38 00:40:41 00:40:43 00:40:46 00:40:47 00:40:50 00:40:55 00:41 :00 00:41 :01 00:41:02 ..... ~ ^ - Knot EACH NBLT EB7H W8rH EBTH EBTH WBTH WITH EBTH NBLT NBLT EBTH NBLT EBTH WBrH EBTH NBLT EBTH NBLT NBLT WBTH WBTH NBLT NBLT NBLT WBrH FRTU 00:39:50 00:39:58 00:39:50 00:40:10 00:40:08 00:40:12 00:40:06 00:40:09 00:40:30 00:40:38 00:40:36 00:40:40 00:40:26 00:40:28 00:40:51 ~- 00:40:41 00:40:47 00:40:51 00:40:46 00:40:48 00:40:54 Measurement of Follow-up Time Unlike He critical gap, the follow-up time is measured directly from the collected field data. The follow-up time can be also considered to be the saturation headway of Be minor street vehicles. A follow-up time is observed only under the following conditions: · He following vehicle has been queued, i.e., when the vehicle arrives at the intersection, there is already at least one vehicle in front of it; · botihof~e vehicles (i.e., He lead vehicle and the following vehicle) use the same gap in the conflicting streams. When a follow-up hme is observed, it is calculated based On the ex~t-queue times of the two vehicles using the same gap. The software was used to extract the follow-up times from the raw data file. Table 20 illustrates a portion of a sample data set where the follow-up times are observed and exacted. The minor sweet approach is in the northbound direction in this example.
OCR for page 49
Irene 2 - ~ Inane 1 - > lee 1 `\ > Ire 2 71 ~ Figure 26. musha~cion of Conflicting Movements at Multi-Lane Sites Field Estimation of Critical Gap and Follow-up Time The critical gap and follow-up lime were measured for each site based on the procedures discussed previously. The measurements were collected Tom a total of 59 videotaped penods (53 unique intersections). The critical gap or follow-up lime at some sites could not be measured because of a I~m~ted data sample or because of unusual geometry at the site. The measurements and associated data are listed In Appendix IT of Working Paper 16, and include the mean values of Me critical gap and follow-up time, the standard deviation, and We number of observations for each site. Two critical gap measurement results for sites with multi-lanes on the major street were also given, and denoted as AR Lanes and Cony: Lane respectively. At sites with multi-lanes, "conflicting" movement vehicles may travel in more than one lane. Therefore, not all the vehicles of a "conflicting movement" have physical conflict potential with We minor street vehicle. Figure 26 illustrates such a situation. In Figure 26, the major street has two through lanes ~ each direction. For the minor street right An, only vehicles ~ lane 2 from the left side have physical conflict potential wad the minor street right turn vehicle. For the minor street left turn, all Me vehicles from the left side have physical conflict potential with the minor street vehicle; however, those vehicles in Lane ~ from the right side have the greatest physical conDict potential with the minor street left turn vehicle. · . . . - . . - ~.. 49 Issues arise regarding how to define the gap events at multi-lane sites. Two ways to deal with this were investigated. The first method was to include ah the vehicles In a conflicting movement to define the gap events, no matter which lane the vehicle travels in. The second method was to include only those vehicles ~ the physically conflicting lanes while defining the gap event. In the first memos it does not matter which lane the conflicting vehicles are in, including vehicles traveling parallel to each other, therefore, this method results in a large number of smaller rejected gaps and accepted gaps. Under this method, the critical gap is smaller Man that obtained using the second method. Using the second memos Is more rational, because the gap event confronted by the minor sheet driver is much clearer. Vehicles in the non-conflicting lane also have some elect on the minor street Diver; however, this effect has been taken into account implicitly by the set of accepted and rejected gaps used to obtain the critical gap estimate. For example, by excluding the vehicles in the non-conflicting lane, the accepted gaps and rejected gaps calibrated from Me raw data file are usually larger when compared to those at a s~ngle-lane site. As a result, the critical gap would be larger than that at a single-lane site. The following analysis and calculations are based on We values calculated using the second memos. The same issues of defining gap events are relevant at a multilane site to detemiiIiing the "weight' to assign to mayor street vehicles approaching trom me right when considering the parameter vp in the capacity formula for a minor street left turn movement (if there are N lanes from He right, then the total through flow approaching from Me rift may be divided by the number of lanes N). The critical gap and follow-up time measurement results were first analyzed according to movement type, intersection geometry type, major street speed, and geographic sector. The mean values of each group were calculated. Although this approach may provide a generalized sense of what range of values We critical gap and follow-up time have under venous U.S. conditions, it may have also obscured some of Me factors that may cause large variations in Me cntical gap and follow-up time. To address these issues, regression analysis was conducted to investigate these favors and develop generalized equations for critical gap and follow-up time calculations.
OCR for page 50
50 FIELD CAPACITY ESTIMATION Two capacity terms have been used during this project: field capacity and model capacity. Field capacity is defined as the true capacity measured in the field and that seines as the basis for testing different capacity models. Mode} capacity is the capacity predicted by a theoretical mode! based on inputs of certain traffic and intersection characteristics. Unless the field capacity is measured correctly as the basis for capacity mode! testing, the mode! testing process cannot be conducted successfully. The field capacity of a minor stream or approach can be measured directly in the field under a continuous queue condition. Under a continuous queue condition, the departure flow rate is unequivocally t e actual field capacity of a minor stream movement or a minor street approach. However, at most sites, it is typically difficult to obsene a continuous queue in the field for a sufficient length of time to estimate capacity. Therefore, the number of usable data points usually limits the mode} testing process. To obtain enough data points of field capacity, shorter lateral data must be used, say 1-m~nute lateral data. The HCM regards the 15-minute peak period as the time for which average capacity should be stated, however, it is rare to find this condition at unsignalized intersections. For example, in urban areas, drivers may divert to alternative routes if they experience high delay at unsignalized intersections. Further, many intersections are signalized long before reaching this condition. One obvious disadvantage of using I-minute interval data is He large variation In flows between consecutive short lime intervals. For example, if one vehicle on the mirror stream passes through the intersection, the resultant capacity would be 60 veh/hr. Thus, with I-minute interval field data, the capacity can only be reported to an accuracy of 60 vph. The data collected dunng the NCHRP 3-46 project showed that, at many of He intersections, even a I-minute continuous queue situation may be rare. It was found that a high number of continuous queuing data points were concentrated at a small number of intersections. For example, among all He I-minute intervals where a continuous queue was found,onlyI4 intersections had more thanI0 such minutes of intervals, and they combine to account for 86% of the total data points of field capacity. The advantage of using the departure flow as the field capacity under a continuous queue condition is that it reflects He true capacity of He intersection. This is especially important when the true capacity cannot be obtained through other methods. For example, where the minor street approach allows right turn sneakers, the capacity usually Increases to some extent as compared to when no sneakers are allowed. Kyte (1992) used He following equation to estimate the field capacity of unsignalized intersections which are undersaturated (i.e. no continuous queue): 3600 t5 tmv where (10~ en Is the field capacity for the minor stream or minor sweet approach, vph t5 iS He average service delay of vehicles once Hey arrive at the stop line, see to is the average move-up time Tom second position to reaching He stop line, see The service delay is measured for a specified interval and averaged for all the minor stream vehicles that passed the intersection during the interval. Service delay is He delay that occurs at the first position of the queue (the stop line). It is the duration Tom when a vehicle reaches the stop line until it exits the stop line. The measurement of service delay does not require a continuous queue condition. This information is available in the macroscopic database established for each site. The move-up time is the amount of time from when the previous vehicle exits the stop line until the subsequent queued vehicle reaches the stop line. It
OCR for page 51
51 is meas~ed In a manner similar to the service delay. The measurement of move-up time requires that at feast two vehicles are present so that the subsequent vehicle is queued. Therefore, a move-up time may not tee measured for each vehicle. An average move- up time measured for ad the time Centrals was usually used In the above equation. It should be noted that move-up time is different from follow-up time as used in the capacity models. Move-up time Is part offoDow-up time. FoDow-up time consists of move-up time and hesitation time, the tninimum time a driver has to spend at the stop line. This method is based on the assumption that We service delay for each minor stream vehicle is a randomly distributed variable Mat is affected by Me composition and volume of the major conflicting streams as well as by the gap acceptance process. The sum of service time and move-up time is a variable that reflects Me average time that each minor stream vehicle occupies the server facility, here defined as the stop line. Based on the concept of queuing theory, the capacity of a facility is Me inverse of the senice time plus move-up time. The advantage of the application of this method is Mat it does not require a continuous queue condition to estimate capacity; therefore, the number of data points that can be used for the mode! testing is significantly increased. In addition, this method gives the capacity of a single sewer (right turn sneaker is not counted), which is consistent and comparable with the results given by various theoretical capacity models. While this method gives identical results to measuring departure flow under continuous queue conditions, it is not known and is Impossible to determine whether this method is valid when a continuous queue does not exist. It is possible that drivers may be more relaxed under low flow conditions with higher resene capacity (critical gap estimation methods cany the same caveats). Another potentially serious problem is that, in many cases, the exact instant of the beginning and the end of afiIst-in-queue senice time is not easy to detect. The absorption capacity method is another field capacity estimation method that was considered by Me project team. This method is based on the same concept as the theoretical gap acceptance models. Knowing or assigning the critical gap to, the follow- up time tf, and the major stream gap distnbution, the number of vehicles using each gap can then be calculated using the equation: q = O forG``ctc Gi-tc +1 forGet If ~ c (106) where As Me number of minor stream vehicles that can use gap Gi (integer value), and Gi is Me major stream gap size, sec. The capacity can then be calculated by aggregating this number of vehicles over all gaps based on a certain time period, as shown in Equation 107. 1 n T if where T is the length of time period. (10~ It Is assumed that each Giver has a constant critical gap and follow-up time, and each major stream movement is 1leated in an identical manner, i.e. each major movement has the same effect on the minor stream driver. Equation 106 assumes a stepwise fimction for the number of vehicles using each gap. Similar to Equation 105, this method does not require a continuous queue condition to estimate the field capacitor; The drawback of this method, compared to the procedures mentioned before, is Mat to and tf must be defined for the application. Comparisons were made between Equation 105 and Me absorption method. The first comparison of these two methods was conducted for continuous queue situations where true capacity is known. The capacity estimated by these two methods was compared with the departure flow rate, which is the true capacity. Another comparison was conducted for non-continuous queue situations. Previous studies have shown that both Harders model and the Siegloch/Troutbeck model can give good capacity
OCR for page 52
52 estimates given good estimates of Me cntical gap and foRow-up time. As the hue capacity cannot be measured directly under non-continuous queued conditions, We capacity predicted by either Harders' mode} or Me SiegJoch/Troutbeck mode! was used as the basis of comparing the two field capacity estimation methods. Table 21 and Table 22 show the comparison between these two methods. Table 21. Statistical Capacity Comparison Results Under Continuous Queueing Conditions , it' A ................. . ................................................................................ ''. ~s ' :'' . ~ .................... , . ,. ~,,,. ~....................................... Constant Sty Error of Y R Squared No. Observations X Coefficient Std Error Coe~c~ent o 18 0.92 3S 1.0 0.01 o 64 0.52 3S 0.96 0 03 . Based ontihe s~icaI results of these comparisons Me following conclusions were reached: . . . The departure flow under continuous queuing reflects He true capacity. However, the difficulty of finding a continuous queue in the field for a substantial period at most sites usually limits the ability of this memos to test capacity models over a wide range of flows. Equation 105 does not require a continuous queue condition. Under continuous queue conditions, it gives identical results to the departure flow. The difficulty of measuring move-up time precisely for each vehicle and the use"' of an average ''~move-up time value for the whole period may cause some variation in the results; however, the variation does not have a significant effect on the final results because the move-up time is small relative to the senice time. The results also show a good correlation between the theoretical mode} results and Kyte's method (Equation 105) under non- continuous queue conditions. The absorption method can be applied as long as major street volume or gap data are available. It is not necessary to measure the minor street delay or flow to apply this method except to estimate the critical gap and follow-up time. When minor street traffic data such as flow rate during · · . continuous queuing or average service delay are available, the absorption method does not show any advantages over Equation 105. The absorption method presents a problem when different Wing movements exist on~emajor street. Minor skeam drivers behave very differently when Hey face different combinations of burning movements. For example, if the oncoming vehicle is a right turn vehicle on the major street, the minor street driver would accept a smaller gap, which would not be reflected by this method. Equation 105 was therefore recommended as He basis for comparison of He mode! testing tasks in the absence of continuous queue data. Whenever possible, He departure flows during continuous queuing should also be used for mode} testing, especiaDywhen~ng sites wad unusual traffic and geometry characteristics, such as right turn sneaker conditions. If necessary, He absorption method can be used for conducting special tests where He other methods cannot be applied. Table 22. Statistical Results of Canacitv Companion IJnder Non~Cnntinil~ll~ t~llelle~lline ~nnAifinn~ l :':':': :':':':':':':':':':':': :':':':':':':':':':':':':::':::'::'::::::':':':':':':':':':' :':':':':':':':':':':': :':::::::::::::::::::: ':::::::::::::::::':':':':':':':':':': :':::: ':':':: :':':':':':':': :':::::::::::::':::'::::::: ::::::::::::::::::::: ::: :~:~:::::::::::::::::::::::::::::: ::::: ::::::: .:~:~:~:~:~:~:~:~:~:~:~:~:~:~:::~:~:~:~:~:~:::::::::::::: : :::::::::::::::: ::::....:::: :::::::::: :::: .:~:..:.:.:~...:,:~:.:.:.:.:.,:.:~:~:2:::::: ::::::::::::::':::':::: ::'::: :::::::::::: ::: ::::::::: :':::':::':'::':':':::' ::::: :: :::::::::::::::::::::::::::::::::::::::::::::::::::::::: :':: :':::::::::: '::::::::::::: ::::::::::::::::::::::: :::::::::::::::::::::::::::::::: ::::::::::: :::::::: a::::::::::::: Constant O O Stir KIT. Y 41 40 R Squared 0.81 0.81 No. Obs. 33 33 X Coefficient 0.98 0.96 Stir Err. Coefficient 0.02 0.01 _ _ . . o S3 0.79 33 0.99 0 02 o SS 0.77 33 0.97
OCR for page 53
53 FIELD DELAY ESTIMATION The vehicle delay was measured from the field data based on specified time intervals of 5 minutes and 15 minutes. The Traffic Data Input Program (Boesen, Kyte, and Rin~isbacher, 1991) was used to extract the traffic information from the videotapes and a raw data file was assembled us~ngtheinformation extracted. (The format of the raw date file was given in Table 17~. The following information is included in the raw data file: . . . . . . . passage time of each vehicle movement direction vehicle type lane usage time a vehicle enters queue time a vehicle reaches the first position in We queue time the vehicle exits the queue. The data are sorted in an ascending order based on Me passage time of each vehicle through the intersection. The individual delay experienced by each vehicle can be calculated from the enter-queue time, f~rst-in-queue time, and the exit-queue time. The difference between the f~rst-in-queue time and the enter-queue time is called "queue delay", and the difference between the ex~t-queue time and first-'n-queue time is called "service delay?' or "wont delay". The "total delay?' is the sum of queue delay and service delay. Since the delay mode! predicts average delay experienced by the minor stream vehicles during a given period, the field delay should also be measured based ontme intervals and averaged for all of Me minor stream . .. .. ve nc es. For this study, delay data were collected from the raw data file by averaging all of the individual delays experienced by all of Me minor stream vehicles Mat departed the intersection during a given interval. For example, if 4 vehicles departed Me intersection during Me same 5-minute time interval, and the individual vehicle delays for these 4 vehicles were 10, 5, 5, and g seconds, respectively, the mean average total delay would be 7 seconds [~10+5~5+g)/4 = 7 seconds]. Delay measured In this way is somewhat inconsistent USA Me delay predicted by the delay models. The delay models usually predict average delay for Dose vehicles that arrived during a given interval. The two melons would give the same results if all of Me vehicles arrived and departed the intersection during the same interval. Working Paper (NCHRP 3-46, 1995) indicated that . when average delay is less than 60 sec/veh, there is no significant difference between the two methods if a 5-minute or longer interval is used, as the interval increases, the error between the two methods is reduced; and if a 15-minute interval is used, an average of 94°/O of the total vehicles would arrive and depart during the same interval. The difference between the two delay measurement methods is insignificant for 5 minute and 15 minute periods of aggregation. Delay was measured in this manner (delay data sorted according to the departure time of the vehicle) for this study to keep it consistent with other traffic parameter measurements such as flow rate and queue length. Flow measured in this manner during continuous queue conditions yields the field capacity for a given interval. The delay modeltesting process is discussed later in this report. A 15-minute interval delay data was used to reduce the vanance.
OCR for page 54
54 . . , I. ~
Representative terms from entire chapter: