IDENTIFYING CAUSAL FACTORS OF TRAFFIC ACCIDENTS IN SRI LANKA

Size: px

Start display at page:

Download "IDENTIFYING CAUSAL FACTORS OF TRAFFIC ACCIDENTS IN SRI LANKA"

Erika Phillips
6 years ago
Views:

1 IDENTIFYING CAUSAL FACTORS OF TRAFFIC ACCIDENTS IN SRI LANKA Amal S. Kumarage 1, C.R. Abeygoonawardena 2, and Ravindra Wijesundera 3 ABSTRACT INTRODUCTION The Traffic Police in Sri Lanka has maintained accident records for many years. These records have been used in compiling aggregate accident statistics that are released quarterly and annually. The data is usually used to derive a set of pre-determined statistics such as accidents by vehicle type, degree of injury, condition of road, location of accident, time of day etc. However, some of the interesting cross relationships between these observations such as that between condition of road and type of accident or vehicle type and degree of injury have not been investigated. OBJECTIVES This paper investigates a number of such relationships using the accident records for the year A total of nearly 45,000 records are used in this exercise of which around 2,000 refer to fatal accidents. Over 40 different items of data are available for each accident. METHOD The analysis has been carried out on a PC using the SPSS statistical package. It includes a preliminary analysis of computing percentages of the different types of accidents, under different vehicle and driver categories. The second phase of the analysis has identified vehicle types and driver categories that demonstrate a statistically higher or lower than average percentage for each of the different attributes of accidents. In the third phase of analysis, some of these significant variations have been tested against whatever other data that is available to identify the possible reasons for such variations. RESULTS The attributes tested include; the geographic location; the features of road geometry; environmental conditions and the condition of the vehicle. It also contained attributes of the resulting accident in terms of road users involved, movement and action taken by the different road users; and degree of injury. It further analyses some of these attributes with the personal attributes of the road user at fault such as age, and sex and also the vehicle type, condition and time of day. CONCLUSIONS The conclusions in the paper range from identifying the most common attributes that appear to be associated with accidents and accident intensity. It also dismisses other hypotheses that do not appear to provide correlation with accident incidence. The conclusions are formulated with the intention that law enforcement, safety education, road design and insurance rates can be suitably modified to influence further efforts in improving road safety in Sri Lanka and other countries. 1 Dr. Amal S. Kumarage, Senior Lecturer, Dept. of Civil Engineering, University of Moratuwa, Sri Lanka. amal@civil.mrt.ac.lk 2 Mr. Cammilus R. Abeygoonawardena, Deputy Inspector General (Traffic), Sri Lanka Police, Colombo, Sri Lanka 3 Mr. Ravindra Wijesundera, M.Phil. student in Transportation Engineering at the Department of Civil Engineering, University of Moratuwa.

2 INTRODUCTION Sri Lanka has a network of over 100,000 kms of motorable roads. Of these, 12,000 kms is managed by the National Government and termed A & B class roads. It is estimated that over 70% of the motorised vehicular traffic use these roads. These roads are mostly substandard two lane intercity highways originally designed in the early part of the last century. They have not been improved to modern standards for the most part. The motorised vehicle fleet is now estimated to be over one million operational vehicles. This fleet is made up of around 40% motor cycles, 30% cars & vans, 10% three wheelers and the balance 20% being made of buses, trucks & land vehicles (Kumarage, 1997). The increasing concern for road safety has required the identification of probable causes of accidents, particularly those relating to the severe accidents leading to fatalities. The accident data compiled by the Sri Lanka Police over several years has hitherto not been fully utilised to analyse probable causes in such detail. DATA AND STATISTICS The data used in this analysis is from the Accident Record compiled by the Sri Lanka Police for the first six months of the year 1997 (Traffic Police Sri Lanka, 1997). A total of 44, 576 vehicles have been reported involved in accidents during this period. The analysis of the data has been carried out using the SPSS statistical package (SPSS Inc. 1996). ANALYSIS The Sri Lanka Police fills a detailed form (Accident Form 297) for each accident that is reported and investigated. However, the form does not identify the primary causes of accidents. It does however, contain many details and features of the accident with respect to the vehicles involved the drivers and the nature and circumstance regarding the incident itself. The nature of the analysis reported in this paper, follows an attempt at identifying the contributions to road accidents, by the different vehicle types and driver characteristics. The vehicle types are identified as cars, lorries (trucks), motor cycles, private buses, State owned buses, three wheelers, light goods vehicles, land vehicles and bicycles. Drivers have been identified by age categories. In the analysis, the accident data is assumed to be normally distributed and the high accident vehicle types are identified at a 95% significance level. That is, those accidents that fall within one of the two tails in a normally distributed curve where the combined area of the two tails is equal to 0.15 (or 15%). This translates to a Z-score of 1. Hence those of the tested vehicle groups or driver groups having a Z-score greater than unity are identified as high accident rate groups (SPSS. Inc., 1996). For example, cars have been found to be more prone to accident than the other vehicle types when they are stationary (Refer Annex 1 and Table 1). The statistical approach adopted in arriving at this conclusion is explained below. 94 th Annual Sessions, Institution of Engineers, Sri Lanka Page - 2

3 Observed percentage of stationary vehicle accidents for cars = 12.4 Population mean (mean for all types of vehicle for stationary accidents) = 8.4 Standard deviation for all vehicle types = 3.5 Z-score = ( )/3.5 Hence cars are considered as showing high accident rate when they are stationary. =1.13 (>1) The final part of the analysis investigates the severity of accidents between different types. In this process, the Fatality Index is used as a measure of the severity of injury in accidents. The Fatality Index is in the Sri Lankan context defined as the percentage of accident injury victims that succumb to their injuries within 30 days of the accident. ACCIDENT INCIDENCE The frequencies of different types of accidents have been reported in this section with respect to different features related to the vehicles, drivers and circumstances of the accident. Accidents by Vehicle Manoeuvre The following types of vehicle manoeuvres have been identified in the Police accident records; (a) stationery, (b) starting (from stationary position), (c) turning right, (d) turning left, (e) overtaking, (f) emerging from minor road, (g) reversing and (h) proceeding ahead. The accident rates resulting from each of these manoeuvres for each of the vehicle types is given in Table A of the Appendix. A total of 44,576-accident vehicle records were used for the analysis. However, incomplete vehicle records were eliminated from the analysis in some instances. For example, every record that has not indicated the vehicle manoeuvre is not considered in any cross tabulation that includes vehicle manoeuvre. As such, only 40,073 records have been used for this particular analysis. The value recorded for each vehicle category under each manoeuvre indicates the percentage of accidents falling into that particular vehicle manoeuvre for that vehicle type. For example, 12.4% of all cars involved in accidents have been stationary at time of accident. The comparative value for motorcycles is 3.3%. The bottom of the same column indicates that on average, 8.4% of all vehicles involved in accidents were stationary. In this light we can deduce that cars have a higher rate of accidents while being stationary, while motor cycles have a lower than average rate of accidents while kept stationary. In this analysis it is observed that 77% of the vehicles involved in accidents were proceeding ahead. These include rear end accidents, sideswipes and some head on crashes. The second most probable vehicle manoeuvre at the time of accident is being stationary (8.4%), followed by turning left or right at a combined total of 5.0%. Other significant vehicle manoeuvres causing accidents include overtaking accounting for 4.6% of vehicles and reversing involving 2.6% of all vehicles that have met with an accident. Table 1 summarises the high accident 94 th Annual Sessions, Institution of Engineers, Sri Lanka Page - 3

4 rate vehicle categories in order of the most common type of vehicle manoeuvre at time of accident. Table 1: Vehicle Types with High Accident Rates by Vehicle Manoeuvre Vehicle Manoeuvre % of all vehicles involved in accidents Vehicle types with high Accident Rates Common Factor Identified Going Ahead 77.1 Motor Cycles, three wheelers & Small sized vehicles, bicycles sudden manoeuvres Stationery 8.4 Cars Medium sized vehicles Overtaking 4.6 Private & State buses Larger sized vehicles with competitive driving habits Turning Right 3.7 Land vehicles & bicycles Slow acceleration vehicles, no electrical signals Reversing 2.6 Lorries Poor visibility Emerging from Minor Road 1.4 Motor cycles & land vehicles Turning Left 1.3 State buses Starting from stationary position 0.9 Private buses & land vehicles Small vehicles attempting to share right of way Competitive driving habits/no rear view mirror Accidents by Driver Action The Police records indicate a number of driver actions associated with each accident. These are, (a) lost control; (b) wrong side; (c) failure to signal; (d) high speed; (e) avoiding an accident; (f) collision with street furniture; (g) skidded and (h) no unusual action. These are given in Table B of Appendix. It can be seen that no unusual action has been recorded in 59% of the cases. This probably is a proof that present accident investigation system does not positively identify the primary causes of accidents. Over 32% indicate lost control, while skidding is reported in 2.8% of the cases. Wrong side is reported in 2% of vehicles, while high speed is recorded only in 1.2% of the cases. Avoiding an accident, failure to signal and collision with street furniture are recorded in less than 1% of the vehicles. The High Accident Rate vehicle categories by driver action are given in Table 2. This classification appears somewhat ambiguous in the case of some driver actions. For example, high speed is attributed to only 1.2% of drivers involved in accidents. However, collision with street furniture, skidding, losing control could also be interpreted as being speed related. These if added up would increase speed related accidents to 36.8% 4. Table 2: Vehicle Types with High Accident Rates by Driver Action Driver Action % of vehicles involved in Vehicle types with high accident incidence Common Factor Identified 4 If mechanical defect related vehicle cases are reduced, even then speed related vehicle cases make up 32%. Driver defects such as intoxication, poor eyesight, fatigue may also come under lost control. 94 th Annual Sessions, Institution of Engineers, Sri Lanka Page - 4

5 accidents Lost Control 32.2 Three Wheelers Unstable design Skidded 2.8 Lorries & Land Vehicles Tyre & Load problems Wrong Side of Road 2.0 Bicycles Short trip vehicles High Speed 1.2 Private Buses & Lorries Heavy Vehicles needing longer breaking distance Avoiding an Accident 1.0 Lorries Failure to Signal 0.8 Bicycles No signalling facility Collision w/street furniture 0.6 Lorries Greater Night Driving Accidents by Driver Condition The Police records identify (a) intoxication, (b) defective eyesight; (c) fatigue; and (d) other physical defects as condition of driver at time of action. However, only 4.3% of drivers involved in accidents had any of the above reported. That is, 95.6% of all vehicles involved in accidents did not report an unusual driver condition. In the other cases, driver condition was reported as normal. The analysis of driver s condition reported by vehicle type is given in Table C of the appendix. Intoxication was reported in 1.4% of drivers involved in accidents, while fatigue is reported in 0.4% of cases. Defective eyesight is 0.1%. Table 3, gives the vehicle wise identification of high accident rate categories based on driver condition. Driver Condition Table 3: Vehicle Types with High Accident Rates by Driver Condition % of all drivers involved in accidents Vehicle types with High Accident Rates Common Factor Identified Intoxication 1.4 Land vehicles Defective Eyesight 0.1 Bicycles & Land vehicles Fatigued 0.4 Bicycles Manually powered No driver license required/predominant rural use Accidents by Vehicle Defects Vehicle defects have been classified as those relating to the improper function of (a) brakes; (b) tires; (c) steering; (d) lights and (e) load defects. The analysis given as Table D in the appendix indicates that 96.2% of vehicles involved in accidents did not report a vehicle with any one of the above defects. With respect to defects recorded, 2.1% of vehicles involved in accidents were relating to brake defects, 0.9% due to tyre defects, 0.4% pertaining to load defects, and 0.4% due to steering and light defects. The vehicle wise analysis of high accident rate vehicles due to vehicle defect recorded is given in Table 4. In this analysis it can be seen that land vehicles have the highest incidence of accidents due to vehicle defects. This is possibly due to the absence of any vehicle fitness certification requirement. Land vehicles are predominantly used in rural areas for agricultural purposes. These are mostly two wheel hand tractors. 94 th Annual Sessions, Institution of Engineers, Sri Lanka Page - 5

6 Table 4: Vehicle Types with High Accident Rate by Vehicle Defects Vehicle Defect % of all vehicles involved in accidents Vehicle types with high accident incidences Common Factor Identified Brake Defects 2.1 Land vehicles, all buses & Heavy vehicles lorries Tyre defects 0.9 Lorries Poor maintenance & Overloading Load defects 0.4 Land Vehicles Difficult terrain Steering defects 0.2 Land vehicles Vehicle designs/no fitness certification Light defects 0.2 Bicycles Generally no lighting. ACCIDENT SEVERITY In this section, the severity of accidents has been reported with respect to vehicle types, driver action and other circumstances of the accident. Severity of Accident and Driver Condition In this analysis, the driver condition has been analysed with the severity of the accident. The severity of accident has been classified as (a) fatal, (b) grievous, (c) non-grievous and (d) noinjury. The driver condition is classified as discussed earlier. Table 5 shows that accidents involving intoxicated drivers are more severe. The same trend is seen for accidents where the driver condition is listed as fatigued. Table 5: Severity of Accident by Driver Condition Accident Type Intoxicated Fatigued Other No Total Defects Defects Fatal 9.3% 12.3% 3.4% 74.9% 100% Grievous 5.4% 2.8% 3.8% 88.0% 100% Non-Grievous 1.7% 0.5% 2.4% 95.3% 100% No Injury 1.2% 0.1% 2.5% 96.1% 100% The analysis clearly indicates that intoxication and fatigue in particular are especially contributory to an increase in the severity of accidents. It also points out that 25% of fatal accidents are caused by unacceptable driver conditions, whereas in no-injury accidents, this is only 3.8%. Driver s Condition and Age 94 th Annual Sessions, Institution of Engineers, Sri Lanka Page - 6

7 The analysis of driver s age with condition indicates that intoxication is found in all age categories. Eyesight and other physical defects are also found in all ages. Only in the case of fatigue is it significantly higher in the older age categories. Fatality Index and Age of Victim The fatality index is computed as the percentage of fatalities among all accident victims who sustain personal injury. This when plotted against the different age categories as shown in Figure 1 those victims aged between years has the lowest index of around 6%. Those above 60 years have the highest fatality index of over 17%. Figure 1 Fatality Index by Age Categories of Victims Fatality Index (%) < > Age Group (years) Fatality Index and Action of Victim Police Records pertaining to the casualties among pedestrians and passengers (other than drivers and riders) identify a number of actions during which the accident occurred. It can be seen from Table 6 that 42% of the above casualties are while travelling, boarding or alighting from a vehicle. A further 25% are while crossing a road and 27% while walking alongside or by the edge of the road. The highest percentage of accident victims by the categories identified therein shows that passengers travelling inside a vehicle are most vulnerable to accidents. However, a relatively low fatality index suggests that vulnerability to death in the event of an accident is lower for them. 94 th Annual Sessions, Institution of Engineers, Sri Lanka Page - 7

8 Table 6: Fatality Index by Action of Victims Road User Passenger Pedestrian Action of Victim Percentage of Accident Injuries Percentage of Accident Fatalities Fatality Index (%) Boarding a Vehicle 1% 1% 9.4% Alighting from a Vehicle 1% 2% 17.2% Inside a Vehicle 37% 27% 6.9% Falling from a Vehicle 2% 4% 21.1% Shoulder / edge of road 21% 26% 11.5% On road 6% 7% 10.5% Crossing but not on legal crossing 24% 27% 10.1% Crossing on legal crossing 1% 1% 13.3% Other actions/not stated 8% 5% 5.8% Total 100% 100% 9.1% Table 6, shows that falling or alighting from a vehicle, though having a low percentage of accidents (3%) have a higher percentage (6%) of fatalities. Crossing a road away from a pedestrian crossing has a higher percentage (24%) of accident injuries, but a lower percentage of fatalities (10.2%), whereas accidents caused on pedestrian crossings have a lower percentage (1%) but has a relatively higher fatality index (13.3%). Possible reasons for high fatality index are that the pedestrians seem to be less vigilant while using a pedestrian crossing since it is their right-of-way, whereas a significant proportion drivers do not take respect that right as they do not reduce speed when nearing a pedestrian crossing that is in use. It should also be noted that some of the pedestrian crossings are positioned close to or within bends or at the crest of a hill, such that visibility is reduced, making them more vulnerable passage for pedestrians. PRIMARY CAUSES FOR FATAL ACCIDENTS The Police data does not enable the direct identification of the primary cause for an accident. This is a present weakness in the reporting format. However, by a case wise identification of all possible defects categories as (a) vehicle defects, (b) driver condition and (c) road condition, it is possible to identify the most probable single cause for accidents. This has been carried out for all fatal accidents in the data set and the results are given in Table 7. It can be seen from the analysis, that vehicle defects cause 7.4% of all fatal accidents. Speed related fatal accidents make up 27.7%. Travelling on the wrong side make up a further 6.8%, while failure to signal is also significant at 2.3%. Road conditions account for 1.8% of fatal accidents. Aggressive driver behaviour classified as being forced off the road and trying to avoid an accident makes up another 5.8%. The above reasons make up for 51.8% of all fatal accidents. The manner in which accidents are recorded does not enable the identification of other primary causes such as driver defects and intoxication. It is also noted that road condition is not properly assessed in recording. 94 th Annual Sessions, Institution of Engineers, Sri Lanka Page - 8

9 Table 7: Identified Primary Cause for Driver Fatality Accidents % Total % 1.Vehicle Defects Brakes 5.1 Tyres 1.1 Steering Light Driver Action Speeding 1.5 Skidded 1.5 Lost Control Wrong Side 6.8 Failed to Signal Driver Behaviour Following Avoiding Action 1.5 Forced off road Road Condition Over Precipice 0.4 Other Unusual Features Total Percentage of Cases Explained 51.8 CONCLUSIONS The paper concludes that the Police accident data can be used for a better understanding of the causes of accidents. According to the accident data analysed for the first six months of the year 1997, a number of vehicle types have been identified for been more contributory to accidents than others. Driver defects and vehicle defects have also been identified according to vehicle types. The fatality index or the vulnerability to death in the event of personal injury has also been analysed. Finally, fatal accidents have been analysed to determine a possible primary cause. In this respect, speed related accidents appear to be the most contributory. Vehicle defects, driving on the wrong side and aggressive driving are all identified as being significant causes of fatal accidents. These results could be used in (a) improvements to Traffic Police enforcement strategies; (b) priority in safety related road designs and road signage and (c) in improving road user awareness programs. ACKNOWLEDGEMENTS The authors gratefully acknowledge the efforts in compiling the data used in this analysis by Chief Inspector Sanders and Nishanthi Costa of the Police Information Division. REFERENCES Kumarage, A.S., (1997). Estimation of Operational Vehicle Fleet 1996 Update, University of Moratuwa, Sri Lanka. SPSS Inc., (1996), SPSS: Release 7.5.1, U.S.A. Traffic Police (1997) Road Traffic Accident Statistics, Sri Lanka. University of Moratuwa, (1999). Transport Statistics Database, Sri Lanka. 94 th Annual Sessions, Institution of Engineers, Sri Lanka Page - 9

10 Appendix- Table A: Percentage of Vehicles Involved in Accidents by Vehicle Type and Vehicle Manoeuvre Percentage Total Stationary Starting Turning right Turning left Overtaking Emerged from Minor Road Reversing Going ahead Total % Cases Car ,149 Lorry ,001 Motor Cycles ,310 Private Bus ,717 State Bus ,547 3-Wheeler ,525 Light Vehicle ,524 Land Vehicle Bicycle ,817 Total ,073 Appendix -Table B: Percentage of Vehicles Involved in Accidents by Vehicle Type and Driver Action Percentage Total Vehicle Type Lost Control Wrong Side Fail to Signal High Speed Avoid Accident Collide with street Furniture Skidded No. Unusual Action Total % Cases Car ,451 Lorry ,296 Motor Cycles ,452 Private Bus ,858 State Bus ,839 3-Wheeler ,627 Light Vehicle ,976 Land Vehicle Bicycle ,930 Total , th Annual Sessions, Institution of Engineers, Sri Lanka Page - 10

11 Appendix -Table C: Percentage of Drivers Involved in Accidents by Vehicle Type & Driver Condition Vehicle Type Percentage Total Intoxicated Defective Eye sight Fatigued Other defectives Normal Total % Cases Car ,398 Lorry ,180 Motor Cycle ,353 Private Bus ,843 State Bus ,830 3-Wheeler ,572 Light Vehicle ,881 Land Vehicle Bicycle ,927 Total ,486 Appendix - Table D: Percentage of Vehicles Involved in Accidents by Vehicle Type & Vehicle Defects Vehicle Type Percentage Total Load Brakes Tyres Steering Lights No Total % Cases Defect defect Car ,396 Lorry ,175 Motor Cycle ,359 Private Bus ,836 State Bus ,830 3-Wheeler ,581 Light Vehicle ,859 Land Vehicle Bicycle ,912 Total , th Annual Sessions, Institution of Engineers, Sri Lanka Page - 11

Excessive speed as a contributory factor to personal injury road accidents

Excessive speed as a contributory factor to personal injury road accidents Jonathan Mosedale and Andrew Purdy, Transport Statistics: Road Safety, Department for Transport Summary This report analyses contributory