v-.‘ A GOMPARISON OF ACCIDENT EXPERIENCE AT INTERSECTIONS WITH FLASHING AND REGULAR TRAFFIC SIGNAL CONTROL DURING LOW‘VOLUME TIME P531008 Thesis for the Degree of MSCE MICHIGAN STATE UNIVERSITY JOSEPH ANGELO MARSON 1976 ll mum; Illzllfllljllll Lm Ml 1| flfl Illfllljlflllfl tl :L M. , (a! ' M ,, D Aid 1w. 2.] . ‘ Michigan .533” UM" _\ n ""2 fit {£3 0 6 $5235 Mqfiw ABSTRACT A COMPARISON OF ACCIDENT EXPERIENCE AT INTERSECTIONS WITH FLASHING AND REGULAR TRAFFIC SIGNAL CONTROL DURING LOW-VOLUME TIME PERIODS By Joseph Angelo Marson Warrants have been developed to provide the traffic engineer with a means of determining the type of traffic control device which should be installed at an intersection. In the case of traffic signals, these warrants provide the minimum conditions under which signals may be justified. During the time periods when the signals are not justified, no warrants are provided to aid in selecting the proper traffic control strategy. This research project investigated the two means of traffic control at signalized intersections during the low- volume hours; namely, full-color and flashing signal Opera- tion. Accident, geometric, and volume data for 170 intersections was collected for these two signal Operations and a comparison was made to determine those conditions under which each signal Operation could be used to minimize the accident potential. Statistical tests were used to compare intersection stratifications in terms of volume, intersection geometry, approach speed limit, and signal interconnection. Joseph Angelo Marson This study investigated only the effect of the signal control on accidents and did not consider the effect on delay and other variables. Comparative tables have been developed and recommendations have been made based on the results of the analysis to assist in determining which signal Operation would be most efficient for a given set of conditions. A COMPARISON OF ACCIDENT EXPERIENCE AT INTERSECTIONS WITH FLASHING AND REGULAR TRAFFIC SIGNAL CONTROL DURING LOW-VOLUME TIME PERIODS BY JOseph Angelo Marson A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the Degree of MASTER OF SCIENCE Department of Civil Engineering 1976 ACKNOWLEDGMENTS The author wishes to express his appreciation to Dr. William C. Taylor, Professor and Chairman of the Department of Civil Engineering, for his guidance through- out the research program. Appreciation is also extended to those people at the Michigan Department of State Highways and Transportation for their assistance in obtaining the data necessary for this project; in particular, special thanks are extended to. Mr. Richard D. Blost and Mr. Dwight Hornbeck. The author wishes to thank Mr. Dennis Randolph of the Macomb County Road Commission for the data he provided. ii [ls‘lT TABLE OF CONTENTS LIST OF TABLES LIST OF FIGURES I. II. III. IV. V. INTRODUCTION REVIEW OF LITERATURE DATA COLLECTION A. B. C. D. Geometric Data Volume Data AcCident Data Data Reduction ANALYSIS OF DATA A. B. C. Accident Types and Severities - All Intersections Accident - Volume Correlation Accident Types and Severities - Intersection Classifications C.l. Volume Analysis C.2. Intersection Geometry Analysis C.2.a. Median and Direction Flow C.2.b. Angle of Intersection C.3. Speed Analysis C.4. Signal Interconnect Analysis SUMMARY AND CONCLUSIONS iii Page vii 10 ll 14 15 18 19 22 23 33 33 40 41 45 49 53 57 Page APPENDIX A — Accident Rates and Geometric Data 62 APPENDIX B - Volume of Traffic on Major and Minor Street Approaches by Time of Day 69 BIBLIOGRAPHY 74 iv LIST OF TABLES Table Page 1 Mean and Standard Deviation of Accident Rates for the Test and Control Group 20 2 Results of U-Test for Severity and Accident Type Classifications During Night~Time Period 24 3 Correlation Coefficients of Night-Time Accidents Versus Volume for the Test Group 29 4 Correlation Coefficients of Night—Time Accidents Versus Volume for the Control Group 31 5 Correlation Coefficients of Accident Severity Weightings Versus Volume 34 6 Results Of U-Test for Intersections with Ratio of Total Major/Total Minor Volume Less Than 2.0 36 7 Results of U-Test for Intersections with Ratio of Total Major/Total Minor Volume Between 2.0 and 4.0 37 8 Results of U-Test for Intersections with Ratio of Total Major/Total Minor Volume Greater Than 4.0 38 9 Results of U-Test for Type 1 Intersections 42 10 Results of U-Test for Type 3 Intersections 43 11 Results of U-Test for Type 4 Intersections 44 12 Results of U-Test for Intersections with Legs Which Meet at Angles Greater Than 70° 47 13 Results of U-Test for Intersections with Legs Meeting at Angles Less Than or Equal to 70° 48 14 Results of U-Test for Intersections with Major Streets Having Speeds Less Than or Equal to 40 M.P.H. Sl Table Page 15 Results of U—Test for Intersections with Major Streets Having Speeds Greater Than 40I'10POH0 52 16 Results of U-Test for Test Versus Control Group Intersections Which Are Isolated 54 17 Results of U-Test for Test Versus Control Group Intersections Which Are Interconnected 55 18 Summary of Mean Accident Rates for Variables Used in Study 60 vi Figure LIST OF FIGURES Data Collection Form for Intersection and Signal Information Data Collection Form for Accident Information Example of Computer Listing of Accident Data for Two Intersections ‘ Plot of Night Accident Rate Versus Total Night Volume for Intersections Under Flashing Signal Operation Plot of Night Accident Rate Versus Total Night Volume for Intersections Under Regular Signal Operation vii Page 12 l3 17 27 28 I . INTRODUCTION Traffic signals are usually installed at intersections to eliminate traffic conflicts, thereby increasing the effi— cienCy and safety of intersections. Improper use of these devices, however, can produce the Opposite effect. For this reason, warrants have been developed, and are continuously being upgraded, to aid the engineer in determining the prob- able effect of a sign installation and to promote uniformity in traffic signal installation practices. These warrants may be found in the "Manual of Uniform Traffic Control Devices for Streets and Highways."1 Although traffic signals may be installed when traffic conditions meet one or more of the warrants prescribed by this Manual, the use of such a device is often only essen— tial for a certain portion of the day. One example of this is the situation where signals are installed based on the warrant for minimum intersecting traffic volumes. The war- rant states that the required volumes must be present 8 hours per day. During the remaining 16 hours, the demand may not justify a traffic signal to prevent conflicts. Frequently, the signals are changed to flashing operation 1National Joint Committee on Uniform Traffic Control Devices: Manual of Uniform Traffic Control Devices for Streets and Highways, U. S. Department of Commerce, Bureau of Public Roads, Washington D. C., (June, 1971). during part of this period and serve as a two-way stop control. The primary reason for operating traffic signals in this manner is to eliminate the unnecessary delay that would be imposed upon drivers. Vehicles waiting at a sig- nalized intersection where there is little cross-street traffic are forced to wait when there may have been an opportunity to proceed through the intersection had there been a two-way stOp control device. This motive for the use of such signal Operations may be supplemented by addi— tional reasons. For example, stopping traffic when there are no conflicting vehicles might very well encourage dis- obedience of signals indications. In addition, due t0“ increased delay, drivers may also be induced to use less adequate routes in an attempt to avoid what they feel is unnecessary delay. The only argument which may exist in favor of retaining 24—hour full-color operation is that flashing operation may adversely affect the safety of the intersection. Current practices vary widely in the use of flashing traffic signals; One criterion that has been used was set forth in the 1961 edition of the "Manual of Uniform Traffic Control Devices."2 It states: When for a period of four or more consecu— tive hours any traffic volume drops to 50 percent or less of the stated volume warrants, it is 2National Joint Committee on Uniform Traffic Control Devices: Manual of Uniform Traffic Control Devices for Streets and Highways, U. S. Department of Commerce, Bureau of Public Roads, Washington D. C., (June, 1961). h) d; ‘y-AT-“Q J—‘o-sa‘n :‘g ~~~~~ f‘"_:‘.'\-"\.‘ h c ‘kqé‘;"?‘a wS-a.G~.-~.. bocu-Av -cns-D ..... ‘1 \—:—~.—-:s ..... wt..- ‘u\~~¢bu»\.\~ ff‘v fi"".3 L;f\“2‘ Cn3r3':fin :fi' -‘na “"§&:fi C V‘ CV..D\—..‘~.-vo A- :v--b-~—oo -v- e... no Cu‘u»-\n C.- ‘QA‘I‘ F v._ 3- s‘v-. £19-10“; . O O F‘n. s'b r~°r~n k~~ "“".3 ‘9‘ 4““ ‘;—~:v~ &. -: : k .- - r “ $4.18 crLL—te ““““ 0.“: 8-.-»... ~25... ----~¢..3vess -..CO-‘ to e Signa Tarrants', which has not yet been published, Indi- cates a lack of consistency and understanding in the use of .0 c. o 1 3 c k 0. o u o . flasning traffic signa-s. C; tne -4 jurisdictions respond- ing to a questionnaire, 88 had specific criterion for the use of flashing traffic signals. Out of these 83, 26% never converted signals to flashing operation, 24% used criterion from the 1961 edition of the MCTCL, 28% used a reduction in volumes other than those set forth in the 1961 MUTCD and the remaining 22% had criterion not dependent on traffic volumes. Since one of the principle reasons for developing the Manual of Uniform Traffic Control Devices was to promote uniformity in the use of traffic control devices, this survey indicates the need to develop standards for the application of flashing traffic signals. Such standards should be based on statistical analyses that present evi- dence showing if and when applications of flashing traffic signals should occur. Before safety based warrants can be established for the utilization of flashing traffic signals, it is necessary to 3"Traffic Signal Warrants", Section 4.2: Criteria for Flashing Operations, National Cooperative Highway Research Program Project 3-20. determine the accident experience with different methods of Operation. Experience has shown that different types of accidents will occur under full-color and flashing Operation of the signals. With full-color operation, the predominant type of accident is the rear-end accident, whereas angle accidents are more common during flashing Operation. Using criteria such as cost, if the accident situation is minimized by the use of flashing traffic signals during non-warranted periods, then full utilization of this opera- tion should be made. If the opposite occurs, it would be necessary to consider accidents, delay of vehicles, and possibly other factors such as increased cost of operation, to determine if flashing Operation is desirable. The objective of this study was to compare the accident experience at intersections Operating under regular and flashing traffic signals. More specifically this study investigated the conditions under which use of flashing operation can be made, as well as those conditions where it should not be used. These guidelines or warrants, however, were based strictly on accident data and did not consider the effect of delay and other costs incurred by the public (such as the additional energy cost of operating full—color signals). II. REVIEW OF LITERATURE As evidenced by the literature search, very little work has been conducted in studying this type of signal operation. In referring to flashing operation of traffic signals during night hours, Paul C. Box supports this View by stating that "more studies are needed on this type of operation, since flashing Operation during off—peak hours is one way of re- ducing needless stOps by drivers on the heavier travelled or higher—speed route."4 The lack of attention given to this area has resulted in the varied use of flashing traffic signals as previously shown from the NCHRP 3-20 project. There was only one study with sufficient data to justi- fy the conclusions found in the published literature. A re- port titled "Accident Experience as Related to Regular and Flashing Operations of Traffic Signals" concludes that 24- hour full-color operation of traffic signals improves the accident situation which would otherwise be experienced with flashing operation.5 This conclusion is based on the results of a five month before and after study conducted in Washington, D. C. Three groups of signalized intersections were utilized in the analysis: 1) Group I contained 162 intersections which were 4Paul C. Box, "Traffic Control and Roadway Elements - Their Relationships to Highway Safety", Revised: Chapter 4, Intersections; 1970 Automotive Safety Foundation, p. 8. 5Guido Radelat, "Accident Experience as Related to Regular and Flashing Operation of Traffic Signals", District of Columbia Staff Report; D. C. Department of Highways and Traffic, (June, 1966). 'n converted from flashing to full-color Operation, 2) Group II contained 177 intersections with full-color operatiOn lo— cated in the same streets as Group I intersections and no more than 2 blocks from an intersection in Group I, and 3) control Group III contained 402 signalized intersections located near Group I intersections, but on different streets or at least 2 blocks from any intersection in Group I. Group III was used to correct the percent increase or decrease in accidents in Groups I and II by assuming that Group III intersections were far enough from the converted signals that they were independent from the accident stand- point. These corrections were made to reduce the effect of changing traffic conditions between the before and after periods in the analysis. Accident rates (such as the num- ber of accidents divided by total entering vehicles) were not used in this study. Rather, the difference in total accidents between two corresponding time periods was calcu- lated and the significance was analyzed. The use of Group III intersections was to account for the difference in volume between the before and after study period. The total number of accidents in Group I intersections drOpped from 64 to 35, a decrease Of 45.3%. The control group (Group III) experienced a drOp from 105 to 99 (-5.7%). Therefore, the adjusted percent change in Group I accidents using Group III as a control was 39.6%. Group II experi— enced a decrease in accidents from 70 to 46 (a decrease of 34.3%) which is an adjusted change of -28.6%. The "t" test was the statistical test used in this study. The accident decrease experienced by Group I intersections proved to be significant at the 90% level, concluding that the Change from flashing to full-color operation reduced the number of accidents. As might be expected, angle accidents showed the highest net reduction (65%), this reduction being significant at the 95% level. The Only other sub-group of total accidents showing a significant decrease was personal injury accidents. There was a before—after decrease from 42 to 25, which was a corrected 47.7% change. This was significant at the 90% level. The decrease in prOperty damage accidents (35.6%) was not statistically significant at the 90% level. Other types of accidents could not be statistically analyzed due to the insufficient number of cases in the other cells. One other interesting point which the study showed was the effect on accidents in Group II intersections (non— converted signals) due to the change in operations of Group I intersections. Total and angle accidents were both signif- icantly reduced by the change in Operations of the nearby signals. This is believed to indicate that traffic behavior at one intersection is not an independent event, but is affected by the Operation of other signals in nearby inter- sections. The literature search revealed differences in stopped- time delay between regular signal control and flashing Operation. Although the analogy Of two-way stop control devices and flashing operation of signals is not valid for purposes of accident comparisons, it does hold true for differences in stOpped-time delay. In this situatiOn, the same number of stops, as well as the length of each is the same under flashing control or stop sign control. There- fore, it was possible to review the literature and compare the stopped-time delay for signalized and 2-way stop con- trolled intersections. One study in this area was a series of field measure- ments by Volk6 to determine the stOpped-time delay for 2-way stOp, 4-way stop, traffic actuated signal, and fixed—time signal control strategies. The results showed that for a two-way stOp, the stOpped—time delay was 0.96 hours during the average hour, whereas the fixed-time signal was conSid— erably more with 1.67 hours of stOpped—time delay during the- average hour. ' A simulation of traffic flow was done by Bleyl7 to compare regular and flashing traffic signal Operation. He compared the delay under signal control to that under flasher control for the volumes used as warrants in the 1961 Manual of Uniform Traffic Control Devices. Utilizing these specif- ications, more delay will occur with the regular signal con-- trol than with flashing operation for volumes below the 50% levels in these warrants. 6Paul C. Box and Willard A. Alroth, "Warrants for Traffic Control Signals, Part II", Traffic Engineering, (Dec., 1967), pp. 22 - 29. 7R. L. Bleyl, "Simulation of Traffic Flow to Compare Regular and Flashing Traffic Signal Operation", Proceedings, Institute of Traffic Engineers, (1964), pp. 152 - 161. Charles N. Dale conducted a cost analysis of intersec— tion traffic controls in which a cost comparison was made of road user time cost.8 It showed that for intersections with 60% of the total ADT on the major leg, the cost due to stop— ped—time delay of traffic signal—controlled intersections ranged from 1.46 to 1.74 times that of two—way stop-control- led intersections for ADT's ranging from 20,000 down to 5,000. Of course, a varying split of total ADT could have a different effect on the stopped—time delay. As seen in the literature search, it was very clear that a substantial difference existed in stopped-time delay between regular signal control and two-way stop control. How- ever, few studies comparing accidents at intersections under regular and flashing signal Operations have been conducted. This study was intended to help clarify the relationship of accidents with these two traffic controls during the low demand hours and to determine if one signal Operation was significantly better from the accident standpoint than the other. The ultimate goal of this study was to recognize certain conditions under which each signal control may be used to minimize accident potential. 8Charles W. Dale, "A Cost Analysis of Intersection Traffic Controls", Traffic Engineering, (May, 1966), pp. 45 -' 50. I I I . DATA COLLECTION Data for this study was obtained from two sources: the Michigan Department of State Highways and Transportation and the Macomb County Road Commission. A total of 169 intersec- tions were used in this study, with the author collecting data from the Michigan Department of State Highways and Transportation on 85 intersections with flashing Operation of traffic signals and 63 intersections with full-color Operation. Data from fourteen intersections with flashing operation and seven with full-color Operation were supplied by the Macomb County Road Commission. In the case of those intersections under the jurisdic- tion of the MDSH & T, information for two consecutive calen- dar years was collected for all but a few of the intersec- tions. In those few that remained, a one calendar-year period was considered. The study period for each intersec- tion was contained within the period 1968-1972. The determination of which two-year period to consider was based on the most recent years in which complete information was available. The fact that some intersections had only a one-year study period was due to the changes in either signal data, or available volume and accident information. Those intersections maintained by the Macomb County Road Commission had a study period of between six and eighteen months. This period fell in the interval of, January, 1971 and December, 1972.. 10 11 The MDSH & T Electrical Devices Unit retains an active file on each of approximately 1817 intersections in Michigan at which there is a signal under its control. Where vital information was not obtainable for the intersections and/or signals, these intersections were eliminated from considera— tion. Only 148 had complete information out of the nearly 500 intersections investigated. °For each of the intersections in this study, four major categories of data were collected: 1.) Signal Data 2.) Geometric Data 3.) Volume Data 4.) Accident Data The forms used for collecting this information are shown in Figures 1 and 2. The information at the tOp of Figure l was used for identification purposes as well as for the retrieval of accident data from computer files. Data collected for each signal included the installation date, flashing Operation hours (for the test group), conversion date and new hours (if the signal was converted to flashing operation or if the hours of flashing Operation changed), the date on which flashing operation was discontinued, and whether or not the signal was isolated or part of a system. All of this information was obtainable from the files. A. Geometric Data The Electrical Device Unit also maintains an up-to-date drawing of each intersection. These drawings provided a 12 FLASHING TRAFFIC SIGNAL ANALYSIS (Intersection and Signal Information) District: County: City/Village/Twp.: Control Section Mileage Point Signal Number Major: to to Minor: to to Routes nos./names: Major Minor SIGNAL DATA Signal is isolated/part of system Installation date:___:___y___ Flashing operation hours: to Conversion date: ___f___f___ New flashing hours: to Date of discontinued flashing operation: ___;___y___ INTERSECTION GEDMETRIC DATA 4-leg Intersection g 0ther(describe) 0 "T“ Intersection [3 It is in urban/rural area Major M159: 1. No. of approach thru lanes: 2. Approach speed limit: . mph mph 3. Grade: __ 1 I 4. Sight distance: ft. ft. 5. One way street? yes no yes no 6. Divided? yes no yes no 7. Right turn flare/lane? yes no yes no 8. Left turn flare/lane? yes no yes no Angle of Intersection: ' Date of intersection diagram:___f___1___ VOLUME DATA Major M1995 Date l. Volume count(during flashing hrs): ___ veh ___ veh _ _ Volume c0unt(during regular hrs): veh _ veh e——4———-e—— 2. Volume count(during flashing hrs): _w__’veh veh ~ - Volune count(during regular hrs): veh veh ——-———-——— After Installation Before conversion or after conversion Major Minor Major Minor 3. Average Daily Traffic: veh veh veh veh 4. Proportioned ADT: veh veh veh ' yeh 5. Portion of ADT during flashing hours veh veh veh __ veh 6. Portion of ADT during regular hours veh veh veh veh NOTES: FIGURE 1 DATA COLLECTION FORM FOR INTERSECTION AND SIGNAL INFORMATION l3 County__ Befo i ccident Period }— _..._.—- No. of Ac Acc. Rate No. of Ac Acc. Rate Accident Period —— M - No. of Ac cc. Rate 0. of Ac cc. Rate Accident Period —u-'-—v._———— No. of Ac Acc. Rate of Ac No. Acc. Rate Accident Period ..—_-—— No. of Ac Acc. Rate NI‘. of Ar Acc. Rate After Conversion 0R After i Accident Type Hulti Ie-veh e t rear turn end .-ve 0th. angl oth Accident Type Multiple—veh left rear turn cnd .“Ve 0th. nngl oth at Accident T e Multiple-veh left rea turn end Sing.-ve I‘M “ ° 0th. an 1 0th Accident T e Sing.-ve Multiple-veh cft rca In»... H 0 Oth tur end n I oth. Signal No. (Accident Data) Severity Severity O D m Severit O O L FIGURE 2 DATA COLLECTION FORM FOR ACCIDENT INFORMATION 14 majority of the intersection geometric data as shown in Figure 1. These geometric features were used to stratify the intersections as a basis for testing the conditions under which flashing operation proved to be most effective. A check was made to assure that the geometry of the inter- sections did not alter during the period of study of each intersection. B. Volume Data Accident rates for the intersections were calculated using the total number of vehicles entering the intersection during the hours of analysis. Volume counts were obtained for each intersection with two sets of counts used and aver- aged when available for the analysis period. These counts revealed the total approach volumes on the major and minor roads during both the hours of flashing Operation and the hours of full-color operation. For the "control" group (those on 24—hour, full-color Operation), the period from 12:00 A.M. to 6:00 A.M. was used for the comparison period. This is the period when flashing Operation of signals is most commonly used, simply because the lowest traffic volumes occur in this time period. Since the traffic counts obtained were influenced by seasonal fluctuations it was necessary to obtain the Average Daily Traffic (ADT) on the major and minor streets. ADT's were not available for the minor streets in a majority of the intersections under state control; thus, the ratio of major to minor volume counts was used to obtain the minor ADT. The final step in computing the necessary volumes need- ed for the accident rates was to determine that fraction of the ADT that occurred during the hours of flashing and full— color Operation on both the major and minor road approaches. This was accomplished by segregating the traffic counts, such that estimates of four volumes were ultimately estab- lished: '1.) Major street volume during the flashing period 2.) Major street volume during the full-color period 3.) Minor street volume during the flashing period, and 4.) Minor street volume during the full-color period C. Accident Data A computer retrieval system was used in the accident data collection procedure except in the situation where ambi- guity of the identification codes occurred, in which case individual accident reports were searched. The categories of accident types and severities used in this study are shown in Figure 2. It was found through the use of a computer accident analysis program that the most frequent types of multiple—vehicle accidents at signalized intersections were angle, left-turn, and rear-end accidents. These three types accounted for 80.5% of all signalized intersection accidents on Michigan's truckline system and were used in the subsequent analysis. One drawback in the use of a computer retrieval system is the difficulty in separating those accidents which oc- curred due to the intersection from those outside the l6 influence of the intersection. The Michigan Department of State Highway's and Transportation's system allows one to specify the distance to be included in the definition of an intersection. For this study, it was presumed that any accident happening within 100 feet of an intersection was the result of the intersection. An example of the computer search of accident data may be seen in Figure 3. In determining the severity classification of an acci- dent, the "worst" case was tabulated for each involvement. For example, an accident with two fatalities and three in— juries was tabulated as one fatal accident. The final result of the accident collection procedure was a listing of the number of accidents for each category occurring during hours of flashing Operation in the test group and full-color operation (midnight to 6:00 A.M.) in the control group for each intersection. In addition to this, the Egt§l_number of accidents for each intersection for the remainder of the 24- hour period (i.e. "daytime" period) was tabulated. This was done to provide a check of the similarity of accident distri- bution when both the test and control group were on full- color Operation. l7 owzu an: phoxoson oozo an: >-1u>un XXIX XXXXX XXXX XXXXXXXXXX XX x on ~aonan among" scam.“ anaonm nanana “saga" oncooo anscno ockaaa noo~q— oonoon annnan onoaoo docks" canoe" --oo cocoon oooaom nnoanu n~on~a ooaawm zomzoz proxy: oo< «noooo cows”: .oncko Amoco: kaoano Konnom ommono omkcna oaacno noooao "nanny ooannv «Kano: demon" cokoka enooko “noooo wflwmoo zomxoz proxy: uua uaeo «(gnu reo cauou wk: z—«z has z30 1uo mauoo >ao xao zaooo pus zuaz was ("<1 >20 ¢uo zao zeo xno mauou has» >ao zao xaooo >30 zaooo so; xamoo >co m30 zao zao mawoo >no mango >ao mu¢aa~ shale hoax< nose. ooae< 2:..4 za_.¢ moae< Hosea zz_.4 woo¢< za_o4 m.i»o moire ozouz uoaxa mooza OZJOZ ozn.¢ 20001 ozorz his» Joa 1n>~ amonn oznuz ozuuz org-z oio.z 524.: 024.: z:_na za_uo nooea ooae< ooaz< oztzm xoipo ozouu org-a ozonc ozooz u1>p Ju< xn>~ zd>oa Lu>ez xw>ea 1u>oa Lu>ea iu>oa 1a>ox 1u>l1 Lw>la 1w>ll Id>ll 1u>lx 1w>l1 xw>ua zu>lx xu>oz xu>lz lu>|1 wrong xu>uz Lu>lx our» ooa an; xu>oz 1d>ll xo>c1 xu>oa ro>uz io>na ru>oa ro>nz x.>ua iu>oz xw>uz ro>uz ru>oz io>uz iu>na 1u>na xu>ua 1u>oa umr» uu< an; «o .o «o «o no .0 so do no ~. to «0 do do no no 0. no no go do no no no so go «0 ~— Nu no no oo «o w“ do ~— on «o do do w. ~o do .zu»z~ xu>~xo .uonnunoxpon go «o .o ~— ~o do so no - no - no _o no co co _o no «o _o no so so no co no no no go m. «o no «0 ~— oo no No as FZuhzu xu>~zo oon.no.uoa onn.no.mo; sauna-ooxpzou «~.o bzuosuu<.u.n~n>4 .> ooo zhoo¢_o g) on; Zhuuzua e u . O .0. a 3“ . , ~H Z 0 o r! ‘ ' 3 0’ o e a 2" ~ . - 1-'— . .. .. C . . o—h—J—JFW 1": % l O 1 2 3 4 5 Accident Rate FIGURE 5 PLOT OF NIGHT ACCIDENT RATE VERSUS TOTAL NIGHT VOLUME FOR INTERSECTIONS UNDER REGULAR SIGNAL OPERATION 29 TABLE 3 CORRELATION COEFFICIENTS OF NIGHTeTIME ACCIDENTS VERSUS VOLUME FOR THE TEST GROUP 11.109 Number. 995. Volume Accident (Accident of (Number of Relat10n* Rate Rate) Accidents Acc1dent§7 Mn + Md -0.037 0.082 0.302 0.306 mn +.md 0.030 0.063 0.257 0.302 Mn + mm -0.203 -0.515 0.113 0.087 Md + md 0.011 0.033 0.320 0.355 Mn + Md + m m -0.013 -0.027 0.324 0.355 n d (MD + Md) x . (m + md) 0.017 0.054 0.269 0.333 Mn X mn -0.100 -0.276 0.084 0.075 Mn/LM -0.226 -0.480 0.062 0.111 mn/Lm -0.064 -0.200 0.107 0.120 (Mn + Md)/LM -0.047 -0.018 0.217 0.344 (“n + md)/Lm -0.002 -0.050 0.148 0.196 L°9 (Mn + Md’ -0.057 -0.121 0.347 '0.348 L09 (mn + ma) 0.074 0.154 0.267 0.330 L°9 (Mn + mn) -0.261 -0.558 0.247 0.227 Log (Mn + Md + mn + md) -0.034 —0.078 0.358 0.374 N=99 N=75 N=99 N=75 * See legend on following page 30 Legend: Mn = Major street night volume Md = Major street day volume mn = Minor street night volume md = Minor street day volume LM = Number of lanes on major street approach Lm = Number of lanes on minor street approach N = Sample size Note: Those intersections with zero accidents were removed such that the log of the number of accidents and accident rates could be derived. Thus, 75 intersections remained. 31 TABLE 4 CORRELATION COEFFICIENTS OF NIGHT-TIME ACCIDENTS VERSUS VOLUME FOR THE CONTROL GROUP Log Number Log Volume Accident (Accident of_ (Number of Relation Rate Rate) Accidents Accidents) Mn + Md 0.123 0.073 0.442 0.457 mn +'md 0.026 -0.153 0.138 -0.024 Mn + mn 0.082 —0.125 0.543 0.461 Md + m0 0.121 -0.008 0.438 0.382 Mn '1” Md 'I" mn + ma 0.121 —o.018 0.460 0.401 (Mn + Md) x . . 0.120 -0.010 0.416 0.320 (mn + md) Mn x mn ' 0.076 -0.124 0.493 0.348 Mn/LM 0.000 -0.091 0.337 0.341 mn/Lm —o.020 -0.240 -0.101 -0.101 (Mn + Md)/LM 0.028 0.020 0.279 0.325 (mn + md)/Lm -0.035 -0.255 0.013 -0.253 L°g (Mn + Md) 0.129 0.062 0.443 0.472 L09 (mn + ma) 0.030 -0.136 0.120 -0.032 L°9 (Mn + mn’ 0.114 -0.122 0.526 0.478 Log (Mn + Md + mn + md) 0.122 -0.051 0.452 0.392 N=70 N=58 N=7O N=58 32 range over which each signal control could be most effective. At this point, it was believed that a better relation- ship could be obtained if the severity of the accidents could be tested against the same volume relationships. One means of accomplishing this could be to correlate an esti— mate of the total accident cost for each of the intersec- tions against the volume. This analysis was conducted by using accident severity weightings as determined by Dr. Paul Abramson in a study of accident costs at intersec— tions. In this report, accident costs from various past studies were used as a basis for developing a quantitative measure of the accident histories of intersections. The result of the study was the following set of factors for urban intersec- tions: . Accident Type Factor Pedestrian 6.5 Right-angle 1.3 Rear-end 1.0 Left-turn 1.3 Other 1.4 Thus, to determine the accident history profile of an intersection, the factors are multiplied by the number of corresponding types of accidents and summed to give a single figure of merit for that intersection. 9Paul Abramson, "An Accident Evaluation Analysis", Transportation Research Board Record 486, (1974), p. 33. 33 For this study, the factors were applied to the number of accidents at each intersection for both the test and control group. This accident severity weighting as well as the log of the accident severity weighting was correlated with the volume relationships previously used. The results of this analysis is presented in Table 5. A comparisOn of the correlation coefficients obtained here with those in Tables 5 and 6 indicate little or no improvement in the ability to relate volume with accidents. The analysis of accidents and accident cost weightings as related to volume demonstrates that a strong linear relationship does not exist. C. Accidenthypes and Severities~Intersection Classifications In an analysis of accidents at intersections, it is advantageous to investigate various intersection character— istics to determine the relationship between these charac- teristics and accidents. This may help to identify those situations in which flashing Operation and full—color Opera- tion will be most beneficial. C.1. Volume Analysis As mentioned earlier in this report, volume data was collected for analysis, as the larger the conflicting volumes the higher the probability of two vehicles arriving simul— taneously. Thus, it was believed this information may provide an insight into the most effective signal Operation for various levels of volume. 34 TABLE 5 CORRELATION COEFFICIENTS OF ACCIDENT SEVERITY WEIGHTINGS VERSUS VOLUME Flashing Operation Regularygperation Accident Log Accident Lo Volume Cost (Accident Cost Cost (Accident Cost Relation Weighting Weighting) Weighting Weighting) Mn+Md 0.295 0.296 0.424 0.441 mn+md 0.263 0.305 0.152 0.003 Mn+mn 0.116 0.069 0.522 0.426 Md+md 0.317 0.351 0.430 0.385 Mn+Md+ mn+md 0.322 0.350 0.451 0.401 (MnfMd)X . Phxmn 0.090 0.071 0.485 0.331 hn/LM ‘0.067 0.101 0.294 0.281 mn/Lm 0.116 0.128 -0.117 -0.096 (Mn+Md)/LM 0.214 0.341 0.242 0.284 (mn+md)/Lm 0.153 0.201 -0.023 -0.235 L09 (Mn+Md) 0.342 0.339 0.429 0.452 P°9 (mn+md) 0.270 0.338 0.140 0.000 I§g+mn> 0.250 0.219 0.509 0.441 09 (Mn+Md+ 0.355 0.371 0.446 0.393 mn+md) N=99 N=75 N=69 N=58 35 This volume data was obtained for the "day" period and the "night" period for all intersections and was used to form the ratio of the major ADT and minor ADT. The inter- sections were grouped according to this ratio into the following categories: a. Intersections with (Major ADT/Minor ADT) less than 2.0. 'b. Intersections with (Major ADT/Minor ADT) between 2.0 and 4.0. c. Intersections with (Major ADT/Minor ADT) greater than 4.0. The U-test was again used as the test for investigating any significant differences in accident rates. The results of these tests may be seen in Tables 6 through 8. Table 6 indicates the results of the U-test for those intersections with a volume ratio less than 2.0. There were 21 and 30 intersections in the test and control group respectively. A check of the daytime accident rates indi— cated that the intersections were of similar characteristics and could be tested in the "night" period.* The results for the various accident classifications indicated a number of significant differences, with the left-turn accidents in the flashing group being significantly greater than the regular group. The control group (regular operation) had a signifi- cantly greater accident rate in the prOperty damage, rear-end, *This check implies only that the "total" night accidents could be compared and does not necessarily indicate that the other individual accident classifications are statistically the same in the "daylight" period, although this assumption was made. 36 TABLE 6 RESULTS OF U-TEST FOR INTERSECTIONS WITH RATIO OF TOTAL MAJOR/TOTAL MINOR VOLUME LESS THAN 2.0 Mean Accident Rate Classification Normalized Test Control g£_Accident ngtatistic §522p_ §£922_ Significant* Total Day ~1.05 2.27 1.78 No Total Night -1.45 2.91 2.36 No Severity: Fatal -1.36 0.11 0.00 No . Injury -1.11 1.88 0.81 No P.D.O. -2.20 0.92 1.55 Yes Multiple Vehicle: Left-turn -1.86 0.78 0.27 Yes Rear—end -2.47 0.28 0.83 Yes Angle -0.70 1.33 0.73 No Single Vehicle: Ran—off-road -l.92 0.07 0.26 Yes No. Intersections in Test Group = 21 No. Intersections in Control Group = 39 * 90% level of confidence 37 TABLE 7 RESULTS OF U-TEST FOR INTERSECTIONS WITH RATIO OF TOTAL HAJOR/TOTAL MINOR VOLUME BETWEEN 2.0 AND 4.0 Mean Accident Rate Classification Normalized Test Control of Accident ngtatistic Grggp_ EEEEB. Significant* Total Day -0.84 1.67 1.53 No Total Night -1.19 3.20 2.21 No Severity: Fatal --- 0.00 0.00 ~- _ Injury -0.11 1.16 1.00 No P.D.O. —1.26 2.03 1.24 No Multiple Vehicle: Left-turn -0.64 0.29 0.20 No Rear—end -l.07 0.63 0.94 No Angle -1.78 1.46 0.45 Yes Single Vehicle: Ran-off-road ~0.l4 0.39 0.35 No No. Intersections No. Intersections * 90% level of confidence in Test Group = 42 in Control Group = 23 38 TABLE 8 RESULTS OF U-TEST FOR INTERSECTIONS WITH RATIO OF TOTAL MAJOR/TOTAL MINOR VOLUME GREATER THAN 4.0 Mean Accident Rate Classification Normalized Test Control Qf_Accident ngtatistic §£93p_ §£922_ Significant* Total Day —0.55 1.73 1.70 No Total Night —l.68 2.22 3.31 Yes Severity: Fatal -0.67 0.04 0.00 No Injury —1.09 0.89 1.00 No P.D.O. —l.73 1.40 2.35 Yes Multiple Vehicle: Left-turn -0.93 0.30 0.05 No Rear-end -2.37 0.62 1.63 Yes Angle -0.38 0.71 0.50 No Single Vehicle: Ran-off-road ~0.12 0.24 0.28 No No. Intersections in Test Group = 36 No. Intersections in Control Group * 90% level of confidence 39 and ran-off—road classifications. Although there are mixed results related to the type of accident, the accident severity index favors flashing Operation. The results of the test for intersections with a volume ratio between 2.0 and 4.0 are shown in Table 7. A compari— son could again be made since there was no significant difference in the "daytime" accident rates of the test and control groups. Although all of the mean accident rates of the test group were greater than the control group (except the rear-end rate), only the angle accident rates proved to be significant. The remaining group of intersections - those with volume ratios greater than 4.0 — were tested for significant differences.” As it may be seen from Table 8, the "day" period showed no difference, implying that a comparison could be made for the "night" period. The control group accident rates proved to be significantly greater than those of the test group in the categories of total night accidents, P.D.O. accidents, and rear-end accidents. There were no test group rates that were significantly greater than the control group rates. The angle accidents in the test group did not appear significant, which is probably due to the small level of conflict between minor and major street traffic. As a result of these tests on intersection accident rates for varying ratios of major street volume to minor street volume, it appeared that fOr large ratios (greater than 4.0) flasher Operation had significantly fewer accidents. 40 The intersections with volume ratios less than 2.0 indicated significant differences which were beneficial to both signal Operations. Due to this split in significant differences, the desirable type of Operation was not immediately discern- able. It is important to note that consideration was not given to the magnitudes of the volumes, which may have proved to be more effective in delineating the efficient uses of each type of Operation. It was felt that this type of analysis would be more biased, due to the fact that it is a common practice of having signals at intersections with large volumes under regular operation and those with low cross street volumes under flashing Operation. An analysis on volumes rather than volume ratios would have been appropriate had this been a before/after analysis or had there not been standards used in determining which inter- sections would operate under the two signal options. C.2. Intersection Geometry Analysis The next variable to be tested for its effect on accidents was the intersection geometry. Aside from traffic volume, the physical configuration of an intersection probably has the greatest influence on the accident potential. Two geometric considerations were used in this analysis: 1) The effect of a one or two-way street with and without medians and 2) the effect of the angle of intersection. C.2.a. Median and Direction Flow The intersection types used in this analysis are as follows: 1.) Four—leg intersections where one or both of the roads are a one-way street. 2.) "T" intersections, where both streets are two-way. 3.) Four~leg intersections where both streets are two-way undivided. 4.) Four—leg intersections where both streets are two-way and one or both are divided. As before, the Wilcoxon rank-sum test was used to compare the differences between flashing operation, and regular Operation of the signals. Tables 9 through 11 indicate the results of the tests. The Type 1 (refer to Table 9) intersections showed only one accident classification which was significantly different between the test and control group. The "rear-end" accident rate for intersections with regular Operation was greater than those with flash— ing operation. Each group tested had a sample size of 12 intersections. The intersections with flash- ing operation had an average main street volume of 19,700 vehicles per day as Opposed to intersections with regular operation, which had an average of 21,000 vehicles per day. It should also be noted that the control group had a mean accident rate greater than the test group for angle accidents, 41 42 TABLE 9 RESULTS OF U-TEST FOR TYPE 1 INTERSECTIONS Mean Accident Rate Classification Normalized Test Control gf_Accident ngtatistic Erggp_ E5222. Significant* Total Day -1.328 1.47 1.90 No Total Night —0.751 2.60 2.95 No. Severity: Fatal --~ 0.00 0.00 -- Injury -1.443 0.40 1.09 No 9.0.0. -0.289 2.20 1.86 N6 Multiple Vehicle: Left-turn —0.260 0.58 0.35 No Rear-end -2.194 0.17 0.64 Yes Angle —0.924 0.64 0.92 No Single Vehicle: Ran-off-road —O.289 1.06 0.68 No No. Intersections in Test Group = 12 No. Intersections in Control Group = 12 * 90% level of confidence 43 TABLE 10 RESULTS OF U-TEST FOR TYPE 3 INTERSECTIONS Mean Accident Rate Classification Normalized Test Control gquccident ngtatistic Gregg §£222_ Significant* Total Day —0.417 1.84 1.65 No Total Night —0.701 2.83 2.16 No Severity: Fatal -—- 0.00 0.00 -- Injury -0.180 1.07 0.84 No p.0.0. -0.794 1.79 1.35 s. Multiple Vehicle: Left-turn‘ -1.854 0.22 0.26 Yes Rear-end -l.009 0.73 0.88 No Angle -1.531 1.23 0.53 No Single Vehicle: Ran-off—road -0.279 0.08 0.12 No No. Intersections in Test Group = 61 No. Intersections in Control Group = 31 * 90% level of confidence 44 TABLE 11 RESULTS OF U-TEST FOR TYPE 4 INTERSECTIONS Mean Accident Rate Classification Normalized Test Control gf_Accident g-Statistic Grggp_ §£222_ Significant* Total Day -0.026 1.98 1.71 No Total Night -0.182 1.79 2.01 No Severity: Fatal -l.168 0.32 0.00 No Injury -0.260 0.66 0.72 No 9.0.0. -0.675 0.81 1.30 so Multiple Vehicle: Left-turn —0.130 0.13 0.12 No Rear—end -0.701 0.39 0.95 No Angle -0.493 0.85 0.53 No Single Vehicle: Ran-off-road —0.130 0.18 0.20 No No. Intersections in Test Group = 11 No. Intersections in Control Group = 15 * 90% level of confidence 45 which is the opposite of what has occurred in other test results. The sample sizes for Type 2 intersections were 11 and 4 for the test and control groups, respec— tively. These intersections did not have similar mean accident rates for the daytime period and, therefore, could not be compared for the two signal 'operations. The most common type of intersection is the Type 3 intersection (both streets two—way undivided) and the test results for these inter— sections is shown in Table 10. Only one accident classification was significantly different. Left— turn accidents in the control group occurred more- frequently than those in the test group. The last type of intersection which was investigated were those with at least one divided street (see Table 11). There were no accident rates having significant differences for Type 4 intersec- tions. As with the previous intersection types, it does not appear as though the accident rate compari- sons justify the use of one signal operation over the other solely on the basis of the intersection geometry. C.2.b. Angle of Intersection The angle of the intersection was analyzed separately from the other physical characteristics because it has a major influence in the accident potential of an intersection. This is due to the 46 sight restriction or inconvenience it places upon drivers when attempting to cross a street. ' The intersections in this study were placed into two categories — those intersections which meet at angles greater than 70° and those which meet at angles 70° or less. The U-test was con- ducted for both the test and control groups to determine if any differences in the night—time accident rates occurred as a result of the angle. The results are shown in Tables 12 and 13. There were 85 and 56 intersections in the test and control group respectively, which met at angles greater than 70°. Table 12 shows that a significant difference occurred in the left-turn and rearfend categories. The left—turn accident classification showed that the accident rate for signals under regular operation was less than that for flashing operation. Flashing signal operation proved to be more favorable in the rear—end category. No other accident classification showed a significant difference between the two signal operations. As such, neither Operation was considered favorable for the angle tested. Table 13 shows the results for the test and control group with extreme intersecting angles. There were 11 intersections in the test group (flashing operation) and 13 in the control group. None of the accident classifications showed a 47 TABLE 12 RESULTS OF U-TEST FOR INTERSECTIONS WITH LEGS WHICH MEET AT ANGLES GREATER THAN 70° Mean Accident Rate Classification Normalized Test Control of Accident g—Statistic Grggp' EEQEB. Significant* Total Day —0.634 1.79 1.74 No Total Night —0.l33 2.85 2.37 No Severity: Fatal —1.152 0.02 0.00 No Injury —0.978 1.28 0.93 No P.D.O. —0.285 1.59 1.45 No Multiple Vehicle: Left-turn -2.002 0.40 0.27 Yes Rear-end -2.322 0.56 0.87 Yes Angle —0.232 1.16 0.65 No Single Vehicle: Ran—off-road -0.021 0.30 0.24 No No. Intersections in Test Group = 85 No. Intersections in Control Group = 56 * 90% level of confidence 48 TABLE 13 RESULTS OF U-TEST FOR INTERSECTIONS WITH LEGS MEETING AT ANGLES LESS THAN OR EQUAL TO 70° Mean Accident Rate Classification Test Control 9f_Accident ngtatistic Group EEEEE Total Day 44.5 2.03 1.33 Total Night 63 3.02 2.17 Severity: Fatal 65 0.21 0.00 Injury 60 1.03 0.44 2.0.0. " 68 1.78 1.72 Multiple Vehicle: Left-turn 57 0.52 Rear-end 58 0.66 Angle 38 1.43 Single Vehicle: Ran—off-road 52 0.08 No. Intersections in Test Group = 11 No. Intersections in Control Group = * 90% level of confidence 13 0.04 1.09 0.31 0.54 Significant* NO No No No NO NO NO NO NO 49 significant difference between the test and control group, although in every classification except "single vehicle ran-off-road" the intersection under flashing operation had a higher mean accident rate than those under regular Operation. It should be noted that this was also true of the daytime accident rate, thus indicating that these intersections are 'more accident prone for reasons other than the angle of the intersecting streets. C.3. Speed Analysis The speed of a vehicle approaching an intersection not only affects the drivers' ability to avoid a possible con- flict, but also influences the decision made by cross street traffic. Such is the case of two—way stOp control, in which the stopped vehicles must decide if the gap in traffic is acceptable fer a safe crossing. For this reason, a study of the speed at the intersections was conducted. Since it was infeasible to obtain the 85th percentile approach speed for traffic at the intersections under study, the posted speed limits were obtained for both the major and minor streets. For the intersections under MDSH & T control, this data was gathered from the sign inventory division. The speed limit was obtained for all but a few of the major streets (since the major streets are under MDSH & T control) but data was available for only half of the minor streets. The major street speed was the variable used in this analysis since it logically would have the greatest effect on the accident potential. It was decided to arbitrarily 50 segregate the approach speeds into categories. After view— ing the speeds for the intersections under study, the following two categories were chosen: 1) those intersec— tions with a major street approach speed less than or equal to 40 mph, and 2) those greater than 40 mph. The results of the U-test which was conducted are given in Tables 14 and 15. Only the left-turn accident category proved to be significant for speeds less than or equal to 40 mph (Table 14). The test group had a higher mean accident rate in this category which did not appear to be considerable (0.27 for the test group as opposed to 0.26 for the control group). It did not appear as though one signal operation was favorable over another for speeds under 40 mph. Intersections with speeds greater than 40 mph also show— ed only one category of a significantly greater accident rate are shown in Table 15. Rear—end accidents at intersec- tions under regular signal operation had a considerably higher mean accident rate than did the test group (1.26 and 0.29, respectively). It appears for this reason, rear-end accident reduction could be attained by flashing signal operations. It was suspected, however, that this signifi- cant difference was due in part to other independent variables such as the volume (since volume has a great effect on rear-end accidents). Thus, without analyzing a combination of variables in this particular situation, it was difficult to conclude that one signal operation was more effective than the other. 51 TABLE 14 RESULTS OF U-TEST FOR INTERSECTIONS WITH MAJOR STREETS HAVING SPEEDS LESS THAN OR EQUAL TO 40 M.P.H. Mean Accident Rate Classification Normalized Test Control §§:Accident g-Statistic Group. EEEEE. Significant* Total Day —0.92 1.63 1.68 No Total Night -0.65 2.93 2.27 No Severity: Fatal -0.88 0.01 0.00 No Injury -0.04 1.04 0.75 No P.D.O. -0.49 1.93 1.54 No Multiple Vehicle: Left—turn —1.89 0.27 0.26 Yes Rear-end -1.27 0.66 0.86 No Angle -l.52 1.26 0.54 No Single Vehicle: Ran-off-road -0.52 0.29 0.32 No No. Intersections No. Intersections * 90% level of confidence in Test Group = 57 in Control Group = 44 52 TABLE 15 FCTIOXS WIT: MAJOR STREETS RESULTS OF U-TEST FOR INTEPSL " TER THAN 40 M.P.H. HAVING SPEEDS GRLA Mean Accident Rate Classification Normalized Test Control gf_Accident ngtatistiE Egggp’ Gregg Significant* Total Day -0.72 2.09 1.75 No Total Night -1.08 2.37 2.45 No Severity: Fatal -1.02 0.08 0.00 No Injury -l.31 1.56 1.23 No P.D.O. -l.47 0.73 1.23 No Multiple Vehicle: Left—turn -0.19 0.64 0.16 No Rear-end —2.89 0.29 1.26 Yes Angle -0.42 1.00 0.61 No Single Vehicle: Ran-off-road -0.72 0.16 0.03 No No. Intersections in Test Group = 35 No. Intersections in Control Group = 18 * 90% level of confidence C.4. Signal Interconnect Analysis The Washington D. C. study which was discussed in the literature review revealed the effect on accidents due to the change in signal operation at nearby intersections. The "total" and "angle" accident figures were significantly reduced by the change in operation of nearby signals, lead- ing to the hypothesis that traffic behavior at one intersec- tion'is affected by signal Operation at other intersections. In the data collection process of this study, the intersections were classified into two groups: 1) those which had signals interconnected with nearby signals and 2) those which were isolated intersections. Once again the U-test was used to determine the effect this variable had on the accident experience at the intersections studied. The first test was made on the isolated intersections as shown in Table 16.. There were 14 intersections in the test group and 9 in the control group. Although the mean accident rates for all classifications (except rear-end accidents) in the test group were greater than those in the control group, none proved to be significantly greater. Table 17 shows the results of the U-test for those in- tersections with signals which are interconnected. There were 79 and 60 intersections in the test and control group respectively. It should be noted that the mean accident rates for both groups were similar to one another, contrary to what was found in the isolated intersection analysis. The "rear—end" accident category was the only one found to have a significant difference between the two groups of data. 53 54 TABLE 16 RESULTS OF U-TEST FOR TEST VERSUS CONTROL GROUP INTERSECTIONS WHICH ARE ISOLATED Mean Accident Rate Classification Test Control gf_Accident g—Statistic Group, EEEEB Significant* Total Day 50.0 2.52 1.75 No Total Night 48.0 4.25 1.52 No Severity: Fatal 58.5 0.16 0.00 No Injury 57.0 3.15 0.94 No P.D.O. 54.5 1.08 0.59 No Multiple Vehicle: Left-turn 60.0 1.43 0.12 No Rear-end 59.5 0.67 0.94 No Angle 51.0 1.57 0.23 No Single Vehicle: Ran-off-road 58.5 0.18 0.00 No No. Intersections in Test Group = 14 No. Intersections in Control Group = 9 * 90% level of confidence Note: Since most of these intersections had zero accidents, the same ranking occurred in the U-Test. 55 TABLE 17 RESULTS OF U-TEST FOR TEST VERSUS CONTROL GROUP INTERSECTIONS WHICH ARE INTERCONNECTED Mean Accident Rate Classification Normalized Te§t_ Control of Accident g—Statistic Erggp’ EEEEE Significant* Total Day —0.54 1.70 1.67 No Total Night —0.31 2.73 2.59 No Severity: Fatal ~1.24 0.02 0.00 No Injury -0.75 0.96 0.89 No P.D.O. -0.71 1.77 1.71 No Multiple Vehicle: Left-turn -1.37 0.24 0.24 No Rear-end —2.61 0.57 0.98 Yes Angle —0.98 1.17 0.68 No Single Vehicle: Ran-off-road -0.74 0.30 0.34 No No. Intersections No. Intersections * 90% level of confidence in Test Group = 79 in Control Group = 60 56 This may be due to the platooning effect which occurs with syncronized traffic signals. The vehicles are grouped together when approaching the intersections; in the control group this proved to be hazardous, whereas in the test group it presented no problems. Therefore, there may be some justification for operating signals in the flashing manner where necessary if they are a part of a system. V. SUMMARY AND CONCLUSIONS This study was conducted to compare the night—time accident experience at signalized intersections under full- color and flashing operation, as a basis for identifying those conditions under which each signal Operation could be used to minimize the accident potential. The study involved the analysis of accidents at 99 intersections with flashing traffic signal control and 70 intersections with full—color signal control. The mean accident rate for the test group and control group during the daylight hours (hours of full-color opera— tion) was found to be 1.82 and 1.69 accidents per million vehicles, respectively. During the night-time hours, the accident rates were 2.78 and 2.42 for the test and control group. Neither of these differences proved to be signifi- cant at the 90% level of confidence. The data was then stratified into various intersection classifications based on volume, geometry and traffic control features. A correlation of total night accidents versus total entering volume showed no significant correla- tion for either the test or control group. A similar corre- lation analysis was conducted using accident cost factors in place of total accidents in an attempt to find if a better correlation could be obtained. This analysis showed little or no improvement in correlating the two variables. A test was then made to determine if the volume ratio (Major ADT/Minor ADT) had a differential effect on the two 57 58 accident rates. The data was stratified into three volume ratio classifications: less than 2.0, 2.0 - 4.0, and greater than 4.0. Only those intersections with a volume ratio greater than 4.0 showed all significant differences in favor of one signal operation. For these intersections, the test group had a significantly lower accident rate in the total night, P.D.O., and rear—end accident categories. The next variable tested was the intersection geometry. The Type 1 intersections (four—leg intersections where one or both of the roads are a one-way street) showed a significant difference in only the rear-end accident classification. The control group had a mean accident rate of 0.64 accidents per million vehicles as opposed to a rate of 0.17 for the test group. For Type 3 intersections (four-leg intersections where both streets are two-way undivided) the control group had a significantly greater left-turn accident rate than the test group (0.26 and 0.22, respectively). The other geometric consideration-~angle of intersection--did not favor either signal Operation. A speed analysis was conducted to determine the effect on the accident rate of speed limits less than or greater than 40 mph. The control group had a significantly higher rear- end accident rate than the test group for those main streets with speed limits greater than 40 mph. It was hypothesized that the difference was due to l) the difficulty in stopping at higher speeds, and 2) the average volume in the control group being higher than the test group. 59 The final analysis of the data considered the effect of interconnected signals. Those intersections which were isolated showed no significant differences between the test and control group for any of the accident classifications. The majority of the intersections were interconnected (n = 79, 60 for the test and control group, respectively) and only the rear-end accident classification showed a signifi— cant'difference. The control group had a mean accident rate of 0.98 and the test group had a mean rate of 0.57. A summary of the results of the analysis are presented in Table 18. The four accident rate classifications which were most important in this analysis (total night, rear-end, angle, and P.D.O. accident rates) are shown for both the test and control group. This table indicates those situations under which either the test or control group accident rate proved to be significantly different than the other group. The results of the analysis do not define a clear advantage of one signal operation over the other. They do, however, indicate certain situations under which one Opera- tion may reduce the potential for certain types and severities of accidents. It is recommended that one use the results of the analysis and the accident history of an intersection to determine if it would be advantageous to utilize flashing signal operation for the night-time period. For example, accidents may be reduced at an intersection with a volume ratio greater than 4.0 and a high incidence of rear-end 60 .oumu usocwoom Hmuomno naucmowmwcmwm sues macho mou00fiocfi “so "ouoz m~.H mn.0 Ho.0 00.H aom.a mm.0 mo.~ 0m.~ own unfinq em.H mm.H em.e e~.H em.o ee.e s~.m mm.~ sew woman mm.0 o0.H no.0 0m.H v0.0 no.0 mm.H mm.v oopmHOmH Emvmhm H0.a >0.H oo.0 hH.H «om.0 no.0 mm.m o>.~ unmm mv.a mm.H no.0 0H.H «no.0 om.0 no.~ mo.m 05A mamc< ~>.H oh.H Hm.0 m¢.H 00.H oo.0 0H.~ m0.m 00w 0m.H Ho.0 no.0 no.0 no.0 no.0 H0.N m>.H v mane mm.H mo.a mm.0 m~.H oo.0 no.0 oH.~ mo.~ m cowuomm IuoucH oo.H 0~.~ No.0 oo.0 «oo.0 0H.0 mm.m 0o.m H emm.m 00.H 0m.0 H>.0 «mo.H no.0 «Hm.m NN.N 0.0A em.a me.~ mv.e ses.a sm.o no.0 H~.N o~.m o.ene.~ cease oEsHO> «mm.a «0.0 m>.0 om.H «no.0 om.0 om.m H0.N 0.Nv em.a hm.a Ho.0 oH.H aom.0 no.0 mv.m o>.N moonuoomnousH Had Honusoo poms Houucoo umoa Houucoo vows Honucou umoa doom ucoonoom oumm undefined oumm ucoowooa oumm unopened manonum> omofimo huuomoum mamse osmiumom unmwz Hence MQDBm ZH DmmD mmqdem¢> mom mmadm BZNQHUUé z¢m2 ho MfidZZDm ma mammfi 61 accidents by changing to flashing operation. One should then monitor the accident situation to assure that the number of angle accidents did not increase significantly. This should lead to a reduction in the total night, rear-end, and P.D.O. accident rates for that intersection, as indicated in Table 18. Similarly, this procedure could be applied to other intersection variables and accident types. APPENDIX A Accident Rates and Geometric Data Note: 1.) 2.) 3.5 4.) 5.) 6.) 7.) 8.) 9.) 10.) 11.) 12.) l3.) 14.) 15.) 16.) 62 ACCIDENT RATES AND GEOMETRIC DATA EXPLANATION OF TABLE CATEGORIES Accident rates are expressed as number of accidents per million vehicles entering intersection Intersection identification number Night-time Night-time Night-time Night~time Night-time Night-time Nightfitime Night—time Night-time Night-time single-vehicle ran-off-road single-vehicle other multiple-vehicle left—turn multiple—vehicle rear—end multiple—vehicle angle multiple—vehicle other property damage injury fatal total Day-time total System: 0 1 means signal is isolated means signal is interconnected Number of Approaches: 3 means intersection has three legs 4 means intersection has four legs 0 means intersection has other than three or four legs Major Speed: Speed limit (mph) on major street approaches Minor Speed: Speed limit (mph) on minor street approaches 17.) 18.) 19.) 20.) 21.) Note: 22.) 23.) 24.) 25.) Major # Dir.: Minor # Dir.: Major Minor Inter. Angle: Div.: DiV.: 0 1 means means means means means means means means 63 major major major major major major major major street street street street street street street street is is is is is is is one—way two-way one-way two-way divided undivided divided undivided Angle (in degrees) of intersection of major and minor street Number of night accidents is that occurring in a 2 year period. For those few which were not, the number of accidents was proportioned to represent a 2 year period. Left—turn: Rear-end: Angle: Other: Night-time multiple-vehicle left-turn Night-time Night-time Night-time other multiple—vehicle rear-end multiple—vehicle angle 64 (.HO mmmZDZ 0 0 0 0 00 H H N N 00 0 0 0H.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00H0N0000 0 0 0 0 00 H H N N 00 0 H 00.N 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 eHH0N0000 0 0 0 0 00 H H N N 00 0 0 eH.N NN.N 00.0 00.0 00.0 N0.N 00.0 00.0 00.0 00.0 00.0 000000000 0 0 0 0 00 H H N N 00 00 e H 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 0000N0000 H N 0 s 00 H 0 N N 00 00 0 H 0N.H 00.0 00.0 00.H 00.0 00.0 00.0 00.0 00.0 00.0 00.0 000000000 H 0 0 N 00 H H N N 00 00 0 H 00.H 00.N 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00H0H0000 N 0 0 0 00 H H N N 0N 00 0 H 0H.H 00.H 00.0 00.H 00.0 00.0 00.H 00.0 00.0 00.0 00.0 0000N0000 N 0 0 H 00 H H N H 0N 00 0 H H0.0 00.0 00.0 00.H 00.H 00.0 N0.H 00.0 00.0 00.0 00.H 0000H0000 0 H 0 0 00 H H H H 00 00 0 H 00.0 s0.H 00.0 00.H 00.0 00.0 00.0 N0.0 00.0 00.0 Nm.0 00N0H000H 0 0 0 H 00 H H N H 0N 00 0 H 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 0N00H000H 0 H H H 00 H H N H 0N 00 0 H HH.H 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 eH00H0000 0 0 0 0 00 H H N N 0N 0e 0 H 00.H 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00H000000 0 0 0 0 00 H 0 N N 00 00 0 H 00.H 00.0 00.0 0H.H 00.N 0H.H 00.N 00.0 00.1 00.0 00.0 00000000, H H 0 H 00 H H N H 0N 00 0 H 00.0 00.H 00.0 00.0 00.H 00.0 00.0 00.0 00.0 00.0 00.0 00H0N0000 N N H H 00 H H N H 0N 00 0 H 00.N 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 0000Ns000 0 H 0 0 00 H H N N 0N 00 H H 00.H NH.H 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 000000000 N N H 0 00 H H N N 00 00 0 H 00.H N0.N 00.0 N0.0 00.N 00.0 00.H 00.H Nm.0 00.0 00.0 00H0N0000 0 N 0 0 00 H H N N 00 00 0 H 00.0 00.0 00.0 00.0 00.0 00.0 N0.0 00.H 00.0 00.0 00.0 aNHON0000 0 H 0 s 00 H 0 N N 00 00 0 H NN.H H0.H H0.0 00.0 H0.0 00.0 00.0 HN.0 00.0 HN.0 00.0 0000N0000 N 0 H 0 00 H H N N 0N 00 0 H 00.H 00.H 00.0 0N.0 00.0 00.0 00.0 00.0 NN.0 NN.0 00.0 0000HH000 0 0 0 N 00 H H H N 00 0 H 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00N0H0HH0 0 0 0 0 00 H H N N 00 0 H H0.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 0000000HH m 0 m m 00 H H N N 00 0 H 00HH 00n0 00H0 00.H 00.H 00.0 N0.N 00.0 00.0 00.0 00.0 0N00000H0 H 00 H H N N 00 s H 00 0 0H s 00 0 0..H 00.H 00.0 00.H 00.H 00.0 00.0 00.0 0H00000H0 H N H 0 00 H H N N 00 0 H 0N.H 00.0 00.0 00.0 00.0 00.0 00.H 00.0 00.H 00.0 00.0 0000N00H0 0 H 0 0 00 H H N N 00 0 H 00.H 00.0 00.0 00.0 00.0 00.0 00.0 0N.H 00.0 00.0 00.0 0000000H0 0 H 0 0 00 H H N N 00 0 H HN.H HN.N 00.0 00.0 HN.N 0H.H 00.0 0H.H 00.0 00.0 00.0 0000000HH 0 H 0 0 00 H H N N 00 0 H 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00000000: H H 0 0 00 H H N N 00 0 H 00.H m0.N 00.0 00.H 00.0 00.0 00.H 00.H 00.0 00.0 00.0 0H00N00H0 0 0 0 s 00 0 0 N N 00 0 H 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 0N00H00H0 0 H 0 0 00 H H N N 00 0 H -0.H 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00H0H00HH 0 H 0 0 00 H H N N 00 0 H 00.0 00.0 00.0 00.0 00.0 00.0 00.0 N0.H 00.0 00.0 N0.H 0000H00H0 H H 0 0 00 H H N N 00 00 0 H N0.H 00.0 00.0 00.0 00.0 00.H N0.H N0.H 00.0 00.0 00.0 0H00H00H0 0 0 0 N 00 H 0 N N 00 0 0 HN.N 00.0 00.0 00.N 00.0 00.0 00.0 00.0 00.0 00.0 00.N 0000000H0 0 0 0 0 00 H H N N 0N 00 H 0 00.0 00.0 00.0 0H.0 00.N 00.0 00.0 00.0 00.0 00.0 00.0 0N00000H0 0 0 0 0 00 H H N N 00 0 H 0H.H Nm.H 00.0 00.0 0H.H 00.0 00.0 00.0 00.0 00.0 00.0 00N0NH003 . . a .240 020 2000 0400 000692 E R R D D v.0 L . . I I E E ammo o o w m 0 w. w 0 0 W. m H. .2090 .2090 H4050 .020 Home «5000 0302.0 02mm .503 00:08 25H 0 H .< I E_ T M D D an an S S C . m m m m R R R m 60“ m R m m 0020.0 00.50:?» mHoHHHNS 33:0 0 00 .. .. s m .0 r. O n. T >000m>mm oHaquoz mqosz mszwaHoos mszneoon 00002 meme azmoHoo< 655 H 0 0 0 00 H H N N 0N 0 H N0.H 0.0H 00.0 N0.0 00.N 00.0 00.0 00.0 00.0 00.N 00.0 uHO0NNOH0 N 0 0 0 00 H H N N 0 H 00.0 00.H 00.0 00.0 00.H 00.0 00.H 00.0 00.0 00.0 00.0 0000H0000 N N N 0 00 H H N N 0 H 00.N 00.0 00.0 Nn.H 00.0 N0.H N0.H N0.H N0.H 00.0 00.0 000000000 0 0 0 0 00 H H H H 0 H 00.0 00.0 00.0 00.N 0H.N 00.0 00.0 00.0 00.0 00.0 00.H 0N00H0000 N 0 0 H 00 H H N H 0 H 00.H 00.0 00.0 00.0 00.0 00.0 00.H 00.0 00.0 00.0 00.N 00H0N0000 n H H H 00 H H N H 0 H 00.N 0H.0 00.0 .00.0 0H.0 00.0 00.0 N0.0 N0.0 00.H 00.0 0N00N0000 0 H 0 0 00 H H N N 0 H 00.0 0N.H 00.0 00.H 00.H 00.H 00.0 00.H 00.0 00.0 00.0 0000H0000 H 0 0 0 00 H 0 N N 00 0 0 00.H 00.N 0N.N 00.0 00.0 00.0 0N.N 00.0 00.0 00.0 00.0 0000HN000 H 0 0 0 00 H H N N 0N 0 H 0N.0 00.H 00.0 00.0 00.H 00.0 00.H 00.0 00.0 00.0 00.0 0HO0H0000 H H 0 0 00 H H N N 00 0 H 00.H NN.N 00.0 00.H 00.H 00.0 00.0 00.0 00.0 00.0 00.0 00N0N0000 H 0. 0 H 00 H H N N 00 0 H N0.H 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00N0N0000 H 0 0 0 00 H H N N 0N 00 0 H N0.H N0.H 00.0 00.0 N0.H 00.0 N0.H 00.0 00.0 00.0 00.0 0HH0N0000 H 0 0 H 00 H H N N 00 0 H 0H.H 00.H 00.0 00.0 00.H 00.0 00.H 00.0 00.0 00.0 00.0 00HON0000 0 N 0 N 00 H H N N 00 0 H NH.H 00.H 00.0 n0.H 00.H 00.0 00.0 n0.H 00.0 00.0 n0.H .0H0NH000 H 0 0 0 00 H H N N 00 0 H 00.N N0.H 00.0 00.0 00.H 00.0 N0.N 00.0 00.0 00.0 00.0 0000N0000 H 0 0 H 00 H 0 H N 00 00 0 H 00.H 00.N 00.0 00.H 00.H 00.0 00.H 00.0 00.0 00.0 N0.H 0000N000H N 0 0 H 00 H 0 H N 00 00 0 H H0.0 00.H 00.0 00.H 00.0 00.0 00.H 00.0 00.0 00.0 00.0 «000N0000 N 0 0 0 00 H H N N 00 H H 00.N 00.0 00.0 00.H N0.H 00.H N0.H 00.0 00.0 00.. 00.0 0000N000H N 0 o 0 00 H H N N 00 0 H N0.H 00.0 00.0 00.H 00.0 NN.0 00.0 00.0 00.0 00.0 NN.0 r000NH000 0 n H N 00 H H N N 00 0 H 00.0 00.0 00.0 00.0 0 .H 00.0 00.0 00.0 0H.H 00.0 00.0 0H00N0000 0 0 0 0 00 H H N N 00 0 H 0H.n 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 0000H0000 N 0 H 0 00 H H N N 00 0 H 00.H 00.0 00.0 00.N 00.H 00.0 00.N 00.0 00.H 00.0 00.0 0N00H0000 0 H 0 0 00 H H N N 00 0 H NN.0 00.H 00.0 00.0 00.H 00.0 00.0 00.H 00.0 00.0 00.0 NH00H0000 0 0 0 N 00 H H N N 0N 00 0 H 0N.H 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 uHO0NH000 0 0 0 H 00 H H H N 00 0 H 00.N 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 p00nHNnNH 0 0 H H 00 H H H N 00 0 H H0.H N0.0 00.0 00.0 N0.0 00.0 00.0 00.0 00.0 00.0 Hn.0 uH00HN000 0 0 0 H 00 H H H N 00 0 H nN.H 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 uN00NH000 H H 0 H 00 H H N N 00 00 0 H 00.H 00.0 00.0 00.0 00.0 00.0 00.H 00.H 00.0 00.0 00.0 u000NH000 H 0 0 0 00 H H N N 00 0 H 00.H 00.0 00.0 0N.N 00.N 00.0 00.N 00.0 00.0 00.0 00.N 0N00NH000 0 N 0 0 00 H H N N 0N 00 0 H HN.N 00.0 00.0 0N.H 00.N 00.0 00.0 00.N 00.0 0N.H 00.0 0HO0NH000 0 0 0 H 00 H H N N 00 0 H 0N.N 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00H0HH000 0 0 0 H 00 H H N N 00 0 H 00.N 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 0000HH000 N H 0 0 00 H H N N 00 0 H 0H.H 00.0 00.0 0N.N 00.0 00.0 00.0 0N.N 00.0 00.0 00.0 0000HH000 0 0 0 0 00 H H N N 00 0 n 00.H N.HH 00.0 NN.0 00.0 00.0 00.N 00.N 00.0 00.0 00.0 0HH00000H H 0 0 0 00 H 0 N N 00 0 H 00.0 00.H 00.0 00.0 0N.H 00.0 00. 00.0 00.0 00.0 00.0 0000N0000 0 H 0 N 00 H H N N 00 0 H 00.N 0H.N 00.0 00.0 0H.N H0.0 00.0 H0.0 00.0 00.0 H0.0 00NON0000 m . . m m m w .zmm mHmHeqnz mHosz o A u L I M m M m M m .... s -23 mezmoHuoa mzHa-amqu 0:002 m0 mNmZDZ wag BZQDHUU‘ 66 a...au¢-.u 1.4-d. n o n N H H N N HH HH H o HH.N no.o no.o no.n nn.o no.o no.o oo.o oo.o no.n no.n uHHnnnnnH n n n H no H H N N mm mm H o NH.H no.o no.o nn.n no.o nn.n no.o Ho.o no.o NH.H no.o NHHnnnnnH n n o H oo H H N N HH HH H o Hm.N oo.o oo.o no.o no.n no.o no.o no.o nn.n no.o no.n uNHnnnnoo n n n H oo H H N N HH oH H o NH.H no.o oo.o oo.o oo.o no.o no.o no.n no.o oo.n nn.n uHHnnonoH H n n H oo H H N N HN oH H o Ho.H Ho.H onoo oooo Hn.H nnno Ho.H nnon nnnn ooon nnnn mannnnoH n n n H H n N N HH HH H HH.N oo.o oo.o nn.n no.o no.o no.o nn.n no.o no.o nn.n uonnnnnoH o H o N on H o N N nH om H H Ho.H NN.o noon NN.n onno onoo oonn NN.n nono nnnn nono unnnnnnoH n n o H oH H n N N HH mm H n HN.H no.o no.o on.o no.o no.n no.o no.o no.o no.o nn.n uNnnnnNNH N o N .N on H n N N oH mm H o NH.H N.HH Hooo N.HH noon nnoo H.HH nono H.HH noon nHHn umnoonnno n n o H HN H H N N oH oH H NH.H no.o Ho.o oo.n no.o oo.o Ho.o oo.o no.o no.o no.o umnnnnonH n n n H oo H H N N NH HH H HHHN oo.o Ho.o oo.H oo.o on.o oo.o no.o no.o oo.o on.o uHonnnnoo n o. n H on H H N N oH HH H NH.o no.o oo.o oo.o oo.o oo.o oo.o no.o no.o on.o oo.n annnonoH n n n H on H H N N HH HH H HH.o no.o oo.o oo.o oo.o oo.o oo.o oo.o oo.o oo.o oH.H UNoonoooo n n _n N H H N N mm mm H NH.o no.o Ho.o on.n no.o no.o oo.o oo.o oo.o Ho.o oo.o uHoooonoH n o o H HN H H N N HH H H oo.H oH.H oo.o oo.o oH.H oH.H oo.o oo.o oo.o oo.o oo.o umHnNNoHN N o H H on H H N N oH H H Ho.N Ho.H oH.o No.H oo.H oo.o mo.H no.o No.H oo.o oo.o uHHoNNoHN N o N H oo H H N N HH H H No.H HN.N oo.o Ho.o oH.H Ho.o Ho.o oo.o Hm.o oo.o oo.H UNNNNHoHN n n o H oo H H N N om H H HN.N HN.o oo.o on.o HN.o HN.o oo.o oo.o oo.o no.o no.o HNNoNNnHN n H o H oo H H N N HH H H HN.H HH.N oo.o No.H No.H oH.H oo.o HH.o oo.o oo.o oo.n uoHnNNoHN o o o H oo H H N N HN H H NH.H HN.H oo.o NH.H NN.o HN.H oo.o oo.o oo.o oo.o no.o uoHoNHoHN N N on H oo H H N N HH H H NH.N NH.H oo.o Ho.H HN.H oo.o NH.N NH.N oo.o HN.H oo.o “HNnNHoHN H H H n oo H H N N HH o H HH.H oH.H oo.o oH.H oo.o oo.o oo.o oo.o oo.o oo.n oo.o uNHoHHnHN H o o H oo H H N N oH H H oo.H mH.H no.o HH.H oo.o oo.o HH.H oo.o oo.o oo.o oo.H uHHanoHo n N o H oo H n N N oH H H HN.H HN.N oo.o Ho.o NH.H Ho.o no.o NH.H oo.o no.o no.o annHHHHH H o H H oo H n N N oH H H oH.o HH.N oo.o HN.N no.o oo.o NH.H no.o NH.H on.o on.o annHHHHo o o n H Ho H o N N HH H H HH.H oH.H oo.o oo.o oH.H oo.o no.o no.o Ho.o no.o oH.H uNnnHHHHH o H o H HH H H N N H o NH.H HN.H oo.o oo.H HN.H no.o oo.o HN.H no.o oo.o Hn.n uNnnHNnHH m . . m m nu D .240 ozm zmso o