EFFECT OF ROAD GEOMETRICS ON ACCIDENTS AND SAFETY

Around 2,38,000 people die in road crashes every year in South

Asian countries

The latest annual statistics indicate that over 80,000 people are killed on Indian roads

Riding a vehicle in India is by large becoming a dangerous

experience, and Indian roads like those of other Asian countries are becoming virtual death traps

Deaths per 1000 vehicles

Fatality Rates in Selected Developing Countries

WHO ARE SUFFERING?

US Thailand Sri Lanka Norway Netherlands Malaysia Japan Indonesia India Australia

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

Pedestrian

Cyclist

Two Wheeler

Four Wheeler

Other

Road users Killed in various modes of transport

Road accident Statistics of India 1970-2004

Source: Ministry for Road Transport & Highways

ROAD ACCIDENT SCENARIO OF INDIA 1970-2004

500000 450000 400000
385018 371204 351999 373671 386456 391449 405637 407497 406726 429910

No. of Accidents

350000 300000 250000 200000

153200 295131 282600 275541

325864 284646

150000 100000

114100
64463 70781 74665 76977 7991 9 81 966 7891 1 80888 84674 85998 9261 8

541 00

56278

601 3 1

60380

50000 0

24000 1 4500

70

80

90

91

92

93

94

95

96

97

98

99

00

01

02

03 20

19

19

19

19

19

19

19

19

19

19

19

19

20

20

20

Year

No. of Road Accidents

No. of persons Killed

Source: Ministry for Road Transport & Highways

20

04

ACCIDENT STATISTICS IN ANDHRA PRADESH

No. of Accidents No. of Deaths per Day per Hour 22.6 26.1 26.5 30.3 30.3 1 1 1 1 1

Year
Total 2001 2002 2003 2004 2005 28902 34133 34826 38937 38339 per Day per Hour in year 79.2 93.5 95.4 106.7 105 3 4 4 4 4 8248 9523 9679 11046 11076

Source: Ministry for Road Transport & Highways

ACCIDENT SCENARIO IN ANDHRA PRADESH

45000 40000 38937 34133 28902 34826 38339

No.of Accidents

35000 30000 25000 20000 15000 10000 5000 0 2001 8248

9523

9679

11046

11076

2002

2003 Year

2004

2005

No. of Accidents

No. of Deaths

Source: Ministry for Road Transport & Highways

OBJECTIVES

Identifying the Blackspot locations Identify road design elements that affect road safety. Identify how a variation in standards for design elements affects the safety of roads in different environments. Develop models for practitioners to determine the appropriate balance between road design standards, road safety.

BLACK SPOT IDENTIFICATION

BLACK SPOT IDENTIFICATION METHODS

Statistical methods Bio-medical engineering approach Engineering methods

Subjective assessment techniques

Empirical Bayes Method

STATISTICAL METHODS

Crash Frequency Method

Crash Density Method

Crash Rate Method Frequency-Rate Method Accident rate based on traffic flow Weighted severity index method Quantum of accident method Accident prone index Multi factor approach

1.Crash Frequency Method

This Method summarizes the number of crashes at location and the stretches having the more number of crashes are taken as accident prone stretches Advantages: o Simple to use o Doesnt require additional information beyond number and location of crashes Disadvantage : Traffic volume is not accounted

2. Crash Density Method

Crash Density = the number of crashes per mile for Highway Sections

3. Crash Rate Method

Crash rate/MEV Number of Crashes 1000000 * DEV n * 365 days/year

n: Analysis Time Period, generally taken as 5 years For links 0.6 miles or longer, the DEV is determined using the following equation:

Linklength DEV ABS 0.3

* DEV

ABS is Absolute value

4.Frequency-Rate Method
This method is a combination of the Crash Frequency and

Crash Rate Methods. Locations are first ranked by Crash Frequency and the worst locations re-ranked using Crash Rate.

The rational of combining Crash Frequency and Crash Rate is to eliminate or minimize the bias of the two individual methods

5. Accident Rate based on Traffic Flow

The accident rate per unit traffic flow for the stretch is

calculated and stretch having more accident rate is taken as accident prone stretch.
Accident Rate ( i ) Total no. of accident in year on the stretch i Total traffic in year on the stretch i

6.Quantum of accident method In the quantum of accident method consequent three years of data is considered for analysis

7.Weighted Severity Index Method

WST ( j )

W
i 1

* Ai

WSI( j )

WST( j ) * K PCU( j )

Based on the values of WSI, mean, standard deviation, the accident prone locations are identified and divided into three types. Accident prone locations of First Order WSI = Mean + 2SD Accident prone locations of Second Order Mean + 2SD > WSI Mean + 1.5 SD Accident prone locations of Third Order Mean + 1.5 SD > WSI Mean + SD

8.Accident Prone Index

Consistency Consistency means how frequently the accidents are taking place at the location. Tendency Tendency means whether the numbers of accidents at the location are increasing regularly or it is consistent or reduced. Level Level means that the magnitude of accidents in quantitative terms.

Rating Of Analysis Elements For Accident Prone Index

Element of Analysis
Consistency (max. of 40 points)

Sr.No.

Accident Scenario
Number of accidents > 3 every year

Points
40

Number of accidents > 2 every year

Number of accidents > 1 every year No accident 2 times increase in 3 years

20
10 0 20 15

Tendency (max. of 20 points)

1 times increase in 3 years

No increase in 3 years
No accident Number of accidents in 3 years are 6 or >6

10
0 40 30 20 0

Level (max. of 40 points)

Number of accidents in 3 years are between 3 & 5 Number of accidents in 3 years are between 1 & 2 No accident in 3 years

9.Multi factor approach

Multi factor approach assigns weight to different accident reflecting severity, type of road user involved and accident cost information.

This has been mainly recommended for identifying black spots with higher pedestrian accidents.

ENGINEERING METHODS

Speed profile method

Safe coefficient method

Traffic conflict studies Wheel path study of vehicle Accident coefficient method

Accident Coefficient Method

In this method the relative accident proneness of a road section is obtained as a continuous product of partial accident coefficients which have been obtained from different geometrical conditions, traffic volume and others.

Relative accident coefficient of a section is obtained as: K= k1* k2* K3**k14

Classification of Locations based on Summary of Accident Coefficient Method

Summary Accident Coefficient (K) <75 75 420 420 1250 >1250

Type of Location Safe Slightly Dangerous Dangerous Very Dangerous

BIO-MEDICAL ENGINEERING APPROACH

Drivers characteristics or response at the location is taken into consideration. The bio-medical techniques are difficult to be used by organizations lacking in the necessary expertise for carrying out field studies

SUBJECTIVE ASSESSMENT TECHNIQUES

Based on the result of the safety evaluation by a group of drivers, traffic engineers, experts of traffic safety and others. Multi dimensional perceptual study of road safety is the ultimate aim of the subjective assessment methods.

In video logging, the whole road can be brought to the

laboratory and safety evaluation can be performed by group of experts.

EMPIRICAL BAYES METHOD

This method is used for identification of high crash locations. The EB method controls the randomness of crash data by using an estimate of the long-term mean number of crashes at a location. It is used for predicting crashes in the future and then ranking based on the predicted number of crashes.

Main disadvantage

Extensive data requirements.

Two sets of data are required to use the Empirical Bayes method:

GEOMETRICS DESIGN EFFECT ON ACCIDENT RATE

Cross-section Sight distance Horizontal alignment Vertical alignment Drainage Medians and barriers Curbs ,Shoulders and Grading

CROSS SECTION
Relative accident rate with roadway width
Road way width, m 4.5 5 5.5 6 6.5 7 7.5 8 9

Relative accident rate

2.2

1.7

1.4

1.3

1.1

1.05

1.0

0.9

0.8

HORIZONTAL ALIGNMENT

Accidents on horizontal curves tend to be of two main types

Running off the road and hitting an object Lost control and Rolled over

Reasons for this are

Driver entering the bend at too high a speed Driver was paying insufficient attention or because he misjudged the severity of the bend.

Accident rate per million vehicle kilometers with radii of horizontal curves
Radius of curve, m Accident rate 50 3.2 150 2.8 200 1.6 250 0.9 500 0.8 1000 0.4

Relative Accident rate relating with the radii of horizontal curves

Radius of horizontal curvature Relative accident rate 10 5.4-4.6 2.25 1.6 <=50 100-150 200-300 400-600 6001000 1.4 10002000 1.25 1 >=2000

VERTICAL ALIGNMENT

The alignment should be properly coordinated with the

Natural topography Available right-of-way Utilities Roadside development Natural and man-made drainage patterns

Relative Accident Rate in relation with Vertical Gradient

Grade, %
Relative accident rate

2
1

3
1.5

4
1.75

5
2.5

7
3

8
4

SHOULDERS

According to V.F.Babkov (1975), a vehicle stopped on a

shoulder does not affect the path of vehicles travelling along

the road only if it is at least at a distance of 2.7metres from the edge of the pavement, and does not affect their speed if this

distance is at least 1.5 meters.

Relative accident rate in relation with Shoulder width

Shoulder width, m Relative Accident rate (Ksh) 0.5 1 1.5 2 2.5 3

2.2

1.7

1.4

1.2

1.1

1.0

PAST REVIEWS

Pasupathy et al. (2000) and Davies (2000). These studies

have produced a range of multivariate models with quite

different relationships. The authors believe the reasons for these variations are that the relationship between road

geometry and crash risk differs between regions and that the
parameters characterise. that influence crash risk are difficult to

Davies (2000) looked at the relationship between road geometry and crash risk for all vehicle types. That study found significant effects due to the horizontal average curvature, difference between maximum and minimum horizontal

curvature, and the minimum advisory speed. Small effects were also found for the gradient, direction, sealed carriageway width and annual average daily travel. There are possibly effects associated with surface age, surface type, wet or dry surface, and accident type. There were no significant effects due to cross section slope or vertical curvature.

 Milliken and de Pont (2000 used data for heavy vehicle

crashes on the State Highway network in New Zealand. They estimated that heavy vehicle crash risk could be reduced by 8% per metre of widening for small increases in road width. This result is backed up by McLean (1997) who estimated a reduction in crash rate of 2% to 2.5% per 0.25 metres of widening. However, there were other predictors such as AADT that had a much stronger relationship with crash rate. These

other predictors were not independent of seal width, so it was

not possible to confidently attribute an increased crash rate to reduced seal width alone.

METHODOLOGY

Preparation of accident data format Accident data Collection from secondary sources

Tabulation and General Analysis of Accident Data Selection of Black spot Identification Method

Crash Density Method

Crash Frequency Ranking Method

Analysis and Identification of Black spots Selection of Major Blackspots Collection of Geometric features at selected Blackspot

Tabulation and General analysis of Geometric details Model Development for the Determining the Relation between Geometric Features and Safety

SHORTEST POSSIBLE RANGE GREATEST POSSIBLE RANGE MEASURING TIME Prism mode: <2sec

1.5 m 3000m

Direct Reflex mode: 3s up to 30m + 1s/10 m

DATA COLLECTION AND ANALYSIS

STUDY AREA ON RAJIV RAHADARI

Rajiv Rahadari is a State Highway passing thorough the

Rangareddy, Karimnagar and Medak districts, having a total length of 227 Km

On Rajiv Rahadari, 42 Km covered in Rangareddy, 81 Km covered in Medak and 101 Km covered in Karimnagar

districts, all with the two lane bituminous surface.

The

study

on

Rajiv

Rahadari

(passing

through

the

Rangareddy, Karimnagar and Medak districts) has been taken

to analyze accident data and identification of accident prone locations

The secondary data was obtained from the concerned police stations for the above mentioned districts.

Accident data analysis was carried out for the period of 2002-

2006

The accident prone locations were identified using Crash density method ,Crash Frequency method.

From the crash density method of analysis

Average Crash Density Critical Crash Density

1.81 3.13

BLACKSPOTS IDENTIFIED
Police Station
L.M.D.Colony Station Timmapur Siddipet urban Koheda Shameerpet Siddipet rural Bollarum Kukunurpally Gajwel Bejjanki Knr rural

No of Accidents
231 39 44 175 82 67 182 259 84 109

Average
46.2 7.8 8.8 35 16.4 13.4 36.4 51.8 16.8 21.8

Length of stretch(km)
10 2 4 20 12 10 28 40 14 30

Crash Density

4.62
3.90
2.20 1.75 1.37 1.34 1.30 1.30 1.20 0.73

Alwal

14

2.8

12

0.23

From the crash density method of analysis Average crash frequency = 29.22

Critical crash frequency = 49.78

BLACKSPOTS IDENTIFIED
S.No
1

Blackspot
Gajwel L.M.D.Colony Station Timmapur Kukunurpally Shameerpet Karimnagar rural Bejjanki Siddipet rural Bollarum Koheda Siddipet urban Alwal

Total no. of accidents

259 231 182 175 109 84 82 67 44 39 14

Crash frequency

64.75
57.75
45.50 43.75 27.25 21.00 20.50 16.75 11.00 9.75 3.50

2
3 4 5 6 7 8 9 10 11

For the study area

Critical Blackspot =LMD Colony (Timmapur)

STUDY AREA ON RAJIV RAHADARI

ANALYSIS OF CRITICAL BLACKSPOT

Geometrical details profiles were taken

blackspot location

under the above stretch using total station.

The blackspot locations under LMD colony station are given in table.

The total length of stretch considered (Alugunur to Timmapur) for analysis is 6.43km.

BLACKSPOTS UNDER ALUGUNUR TO TIMMAPUR STRETCH

BLACK SPOT
ALGUNOOR EENADU OFFICE LMD COLONY MAHATMA NAGAR ST ANTHONY SCHOOL

CRASH RATE
16 0.25 9 0.25 0.25

THIMMAPUR

Analysis and model development

.xls

CR =

-11.99 + 18.8 614(CM) + 6.01(USW) - 15.0266(PSW) 0.0153(RAD) - 6.5864(SE) - 0.0908(HSD)+7.4118(TR) 0.0213(VSD) R2 = 0.835
99% significance value 1.884 1.554 1.628 1.615 0.818 2.427 0.819 6.87

Variable
CM USW PSW RAD SE HSD TR VSD

T-test value
-1.74 -1.32 -1.4 1.38 -0.28 2.38 -0.38 6.61

LOCATION
ALGUNOOR EENADU OFFICE LMD COLONY MAHATMA NAGAR ST ANTHONY SCHOOL THIMMAPUR

ACTUAL VAUES OF CRASH RATE

14.36 0.41 3.27 0.32 0.57 8.15

PREDICTED VALUES OF CRASH RATE

14.71 0.76 3.62 0.67 0.92 8.50

Paved Shoulder Width Crash Rate Relative Accident Rate

3 6.17 1.00

2.5 13.68 2.22

2 21.20 3.43

1.5 28.71 4.65

1 36.22 5.87

Radius
Crash Rate Relative Accident Rate

1000
0.87 1

500
8.52 9.75

250
12.35 14.12

200
13.11 15.00

150
13.88 15.87

100
14.64 16.75

50
15.41 17.62

Super Elevation
Crash Rate Relative Accident Rate

7 4.82 1

6 11.41 2.37

5 17.99 3.73

4 24.58 5.10

3 31.17 6.46

2 37.75 7.83

1 44.34 9.20

Horizontal Sight Distance

Crash Rate Relative Accident Rate Total Rise Crash Rate Relative Accident Rate Vertical Sight Distance Crash Rate Relative Accident Rate

200 3.15 1

150 7.69 2.44

125 9.96 3.16

75 14.50 4.61

50 16.77 5.33

8 31.81 1

7 24.49 0.77

6 17.08 0.54

5 9.67 0.30

4 2.26 0.07

250 10.53 1.00

200 11.59 1.10

150 12.66 1.20

100 13.72 1.30

50 14.79 1.40

CONCLUSIONS

Along the horizontal curves, radius and sight distances are insufficient for the vehicles which are moving at greater than the design speed which is 60kmph.

Most of the accidents were occurred near the junction where there is no proper sight distance for the vehicles coming from the minor road to merge or diverge from major road. This problem can be overcome by providing the service roads for the minor road vehicles so that they can merge or diverge into major road traffic.

The number of culverts along the section are more. At these locations
the carriage way width was drastically decreased and because of improper signboards the drivers are unable to judge the situation.
continued

Even the road geometrics were designed well accidents were

occurring due to some deficiency in sight distance and mostly due to

the more number of cross roads joining the major road and more number of culverts along the road and because of no cautionary sign boards regarding the cross roads and culverts for alerting the vehicular drivers of major road.

LIMITATIONS OF STUDY

Blackspots were identified based on the total number of accidents in

the given stretch, traffic volumes were not considered so it is not

possible to relate accident rate with traffic flowing along the road.

If the accident data is precise i.e. time and date of accident and gap between two accidents occurred in a section is known then there will be a possibility of fitting a better binomial or Poisson distribution which is not possible with the present available data.

The model developed can be used for predicting the future crash rate value, but accident rate does not solely depend on geometric features but also the number of cross details, number of culverts and traffic regulations maintained along road.

SCOPE FOR FURTHER WORK

Blackspots can be identified by the other methods which consider the

traffic volume.

Accident cost analysis can be carried out to find the appropriate balance between road safety benefits and costs.

Road safety audit can be done to get the further details like encroachments, position of signboards, etc., and other improvements can be stated by auditing along the road.

THANK YOU