road safety final
DESCRIPTION
safety ,road,pptTRANSCRIPT
ROAD SAFETY
ANALYTICS
Objectives:
Analysis of road safety data in TataSteel and to give prediction model and recommendations for reducing accidents and better management of road incidents.
Tools and Techniques-
Pareto charts Pie Charts Cluster Analysis Histograms Arena SAS PHP
Methodology Cause and Effect diagram
Pareto Charts – In general we can say that 20% of factors are responsible for 80% of the problems. For finding this we use pareto charts.
Clustering – We can group the data points in different clusters based on their location (co-ordinates) and risk score.
Distributions – We find the closest fit distribution to estimate the time between incidences(TBI).
Pie Charts – We can show contribution of factors with the help of pie charts.
Prescriptions.
Data Analysis
Injury Type Pareto
Injury
Injury Type frequency cum. percent
F/A 57 57 45.96774
NO 49 106 85.48387
LTI 15 121 97.58065
FATAL 2 123 99.19355
MTO 1 124 100
F/A NO LTI FATAL MTO0
10
20
30
40
50
60
0
20
40
60
80
100
Pareto Chart
frequency percent
injury
freq
uen
cy
Cu
mm
perc
en
t
Injury
Injury Type frequency cum. percent
NO 81 81 27.46479
F/A 39 120 84.50704
LTI 21 141 99.29577
FATAL 1 142 100
NO F/A LTI FATAL0
20406080
100
020406080100
Pareto Chart
Frequency Cummulative
injuryfr
equency
Cum
m p
erc
ent
2013-14 2012-13
Vehicle Type Pareto
Vehicle Type
Vehicle Type
Frequency prob cumm prob percentage
Two 60 0.3774 0.3774 37.7358
HV 55 0.3459 0.7233 72.3270
FW 26 0.1635 0.8868 88.6792
Cycle 18 0.1132 1.0000 100.0000
Two HV FW Cycle0
10
20
30
40
50
60
70
0
20
40
60
80
100
Pareto Chart
VEHICLE
FR
EQ
UEN
CY
CU
MM
PER
CEN
T
HV FW Two Cycle0
10
20
30
40
50
60
70
80
90
0
20
40
60
80
100
Pareto Chart
VEHICLEFR
EQ
UEN
CY
CU
MM
PER
CEN
T
Vehicle Type
Vehicle Type
Frequency prob cumm prob percentage
HV 81 0.5094 0.5094 50.9434
FW 37 0.2327 0.7421 74.2138
Two 27 0.1698 0.9119 91.1950
Cycle 14 0.0881 1.0000 100.0000
2013-14 2012-13
Control chart for no. of incidents per month
AprilJune
August
October
December
February
AprilJune
August
October
December
February
AprilJune
August
October
December
February
AprilJune
August
October
December
0
5
10
15
20
25
No. of Incidents per month UCL=17.633 LCL=4.367 Month
No.
of I
ncid
ents
Separation between 2012-13 and 2013-14 control chart
A B C D E
1 25 24 22 19 15
2 23 21 18 14 10
3 20 17 13 9 6
4 16 12 8 5 3
5 11 7 4 2 1
Risk Score CalculationConsequence
Pro
bab
ilit
y
Cluster Analysis combined for 2012-13 and 2013-14
Cluster Analysis for 2013-14Cluster Analysis for 2012-13
Plot of incidents based
on injury
Fatal
LTI
First Aid
No Injury
Plot of incidents based on property
damage risk score19 to 25
13 to 18
7 to 12
1 to 6
Plot of incidents based on both type risk score
19 to 25
13 to 18
7 to 12
1 to 6
Clustering of
incident locations
Represents cluster
Cluster Xcord Ycord count cum. percent Location
6 829.421 749.14 57 5721.5094
3East plant drop gate,LD#3 Traffic Signal,Near
LD#3 Office Turning
4 721.296 263.241 54 11141.8867
9L Town Gate, Diamond Crossing, G Blast Furncae
Ramp and crossing
8 1227.9 791.95 40 15156.9811
3HSM Gate, WRP Weigh Bridge, Canteen turning
5 342.892 194.757 37 188 70.9434 Security Office Traffic signal,Coke plant Drop Gate
7 451.419 1174.77 31 21982.6415
1Cabin#4 drop Gate,Near Merchant Mill Office
1 215 506.389 18 23789.4339
6West side peripheral Road,Near WGO
2 1109 298.471 17 25495.8490
6Slag Road Gate
3 154.455 964.545 11 265 100 Pellet Plant Turning, Near PH#3 Gate
6 4 8 5 7 1 2 30
10
20
30
40
50
60
0
20
40
60
80
100
count percent
CLUSTER NO.
frequency
CU
MM
. P
ER
CEN
T
Cluster Analysis
overlapped for
2012-13 and 2013-14
Represents cluster of 2012-13
Represents cluster of 2013-14
Clustering
based on risk scoreBelong to cluster 2
Belong to cluster 4
Belong to cluster 3
Belong to cluster 1
Distribution of Time between incidents
2013-14 2012-13
Distribution Parameters
Distribution Name: Weibull
Alpha: 3.64
Beta: 1.26
Expression: -0.5 + WEIB (3.64, 1.26)
Square Error: 0.007550
Sample Mean = 2.87
Sample Std. Dev = 2.78
Distribution Parameters
Distribution Name: Weibull
Alpha: 3.33
Beta: 1.29
Expression: -0.5 + WEIB(3.33, 1.29)
Square Error: 0.004127
Sample Mean = 2.57
Sample Std. Dev = 2.55
Distribution of Time between incidents
LTIFirst AidNo Injury
Control Chart for Time between Incidents
2013-14 2012-13
4/1/2013 5/24/2013 7/16/2013 8/24/2013 10/1/2013 10/28/2013 12/5/2013 1/19/2014 2/17/20140
2
4
6
8
10
12
14
16
CONTROL CHART FOR TBI(2013-14)
TBI Linear (TBI) UCL=11.759 LCL=0.113
D A T E
T B
I
4/1/2012 5/21/2012 6/22/2012 7/23/2012 8/19/2012 9/29/2012 11/23/2012 1/10/2013 2/18/2013 3/20/20130
2
4
6
8
10
12
14
CONTROL CHART FOR TBI(2012-13)
TBI Linear (TBI) UCL=10.468 LCL=0.112
D A T ET
B I
Distribution of Time between injury
Distribution Parameters
Distribution Name: Weibull
Alpha: 6.73
Beta: 1.13
Expression: -0.5 + WEIB(6.73, 1.13)
Square Error: 0.015753
Sample Mean = 5.92
Sample Std. Dev = 6.08
Distribution Parameters
Distribution Name: Weibull
Alpha: 5.32
Beta: 1.26
Expression: -0.5 + WEIB(5.32, 1.26)
Square Error: 0.003990
Sample Mean = 4.44
Sample Std. Dev = 4.02
2013-14 2012-13
Control Chart for Injury2013-14
2012-13
4/18/2
013
5/2/2
013
5/24/2
013
6/13/2
013
6/29/2
013
7/10/2
013
7/26/2
013
8/11/2
013
8/24/2
013
9/6/2
013
9/11/2
013
9/27/2
013
10/3/2
013
10/7/2
013
10/14/2
013
10/22/2
013
11/17/2
013
12/7/2
013
12/23/2
013
12/29/2
013
1/15/2
014
1/26/2
014
2/6/2
014
2/15/2
014
2/19/2
0140
5
10
15
20
25
CONTROL CHART FOR TBI(2013-14)
TBI Linear (TBI ) UCL=17.186 LCL=0.165
D A T E
T B
I
4/1/2
012
5/7/2
012
5/9/2
012
5/21/2
012
5/29/2
012
6/19/2
012
6/22/2
012
6/28/2
012
7/10/2
012
7/25/2
012
7/30/2
012
8/4/2
012
8/14/2
012
8/16/2
012
8/22/2
012
8/31/2
012
9/13/2
012
9/19/2
012
10/4/2
012
11/2/2
012
11/27/2
012
12/5/2
012
12/22/2
012
12/26/2
012
1/11/2
013
1/28/2
013
2/1/2
013
2/16/2
013
3/7/2
013
3/19/2
013
3/22/2
0130
5
10
15
20
25
30
CONTROL CHART FOR TBI(2012-13)
TBI Linear (TBI) UCL=24.88 LCL=0.14
D A T E
T B
I
Correlation between different Vehicles involved
HV HV-FW HV-TW HV-CYCLIST FW FW-TW FW-CYCLIST TW TW-CYCLIST CYCLIST0
5
10
15
20
25
4.75 4.31
7
15.5
8.8
4.67
19.5
10.29
7
14
8.24 7.75
6
0
9
4
00.875
0 0
Average Injury Risk score Average Property Damage Risk score
VEHICLE COMBINATION
RIS
K S
CO
RE
Correlation between different Vehicles involved
Combination
Incident count
HV 32
HV-FW 16
HV-TW 2
HV-CYCLIST 4
FW 5
FW-TW 3
FW-CYCLIST 2
TW 48
TW-CYCLIST 8
CYCLIST 4
HV26%
HV-FW13%
HV-TW2%
HV-CY-
CLIST3%
FW4%
FW-TW2%
FW-CYCLIST2%
TW39%
TW-CY-
CLIST6%
CYCLIST3%
Frequency Contribution
Contribution of different Vehicle combinations in risk
Combination
Cumulative Risk
Injurymateri
al
HV 152 263.68
HV-FW 68.96 124
HV-TW 14 12
HV-CYCLIS
T62 0
FW 44 45
FW-TW 14.01 12
FW-CYCLIS
T39 0
TW 493.92 42
TW-CYCLIS
T56 0
CYCLIST
56 0
HV15%
HV-FW7% HV-
TW1%
HV-CY-
CLIST6%
FW4%FW-TW1%FW-
CY-CLIST
4%
TW49%
TW-CY-
CLIST6%
CYCLIST6%
Injury Risk Contribution
HV53%
HV-FW25%
HV-TW2%
FW9%
FW-TW2%
TW8%
Property Damage risk Contribution
MASTER LOGIC DIAGRAM
Incident
VEHICULAR(5)
EXTERNAL(6)
PHYSICAL(1)
BEHAVIORAL(2)
SYSTEM(3)
INFRASTRUCTURE(4)
PHYSICAL(1)
OBSTRUCTION
DIVIDER SPILLAGE DROP GATE HEIGHT BARRIER ILLUMINATION
BEHAVIORAL(2)
INTENTIONAL
CARELESS DRIVING
TRAFFIC NORMS VIOLATION ALCOHOL OVERTAKING HIGH SPEED SLEEPY
NON INTENTIONAL
NOT KNOWLEDGEABLE
SYSTEM(3)
DESIGN
ZEBRA CROSSING TRAFFIC LIGHT CONVEX MIRROR SIGN BOARD
SUPERVISION
OTHERS(ROUTE SURVEY,SECURIT
Y
VEHICULAR ISSUES
(5)
BRAKE FAILURE
INDICATOR LIGHT TYRE BURST
FAIL SAFE BRAKE
FAILURE
STEERING JAMMED
HAND BRAKE NOT WORKING
EXTERNAL (6)
STREET DOGS
Contribution of Organizational
factors and Use of proactive data(FY-
14)
Issues Count cumulative count
Physical
Divider 2
34Spillage (material/water) 13
Environmental (Illumination) 11
Height Barrier 2
Drop Gate 6
Behavioral
Overtaking 28
192
Alcoholic 1Sleepy 0
Not knowledgeable 11Traffic norms violation 5
Careless Driving 107High Speed 40
SystemNo Zebra Crossing 1
32Traffic Light 11
Convex Mirror 3Others (supervision, route survey) 6
No Sign Board 11
InfrastructureBad Road Condition /intersection issue 6
53level crossing 17
Narrow Road (Congestion) 13Blind Turn 6
No separate cyclist/pedestrian pathway 11
Vehicular Issues
Brake Failure 4
16Steering problem 4
Indicator Light Failure 0Others (No reverse mirror) 6
Tyre Burst 2
External Issues Street dogs 6 6
Histogram of various Factor Contributions
Div
ider
Spill
age
(mat
eria
l/w
ater
)
Envi
ronm
enta
l (Il
lum
inati
on)
Hei
ght B
arri
er
Dro
p G
ate
Ove
rtak
ing
Alco
holic
Slee
py
Not
kno
wle
dgab
le
Traffi
c no
rms
viol
ation
Care
less
Dri
ving
Hig
h Sp
eed
No
Zebr
a Cr
ossi
ng
Traffi
c Li
ght
Conv
ex M
irro
r
Oth
ers
(sup
ervi
sion
, rou
te s
urve
y)
No
Sign
Boa
rd
Bad
Road
Con
ditio
n /i
nter
secti
on is
sue
leve
l cro
ssin
g
Nar
row
Roa
d (C
onge
stion
)
Blin
d Tu
rn
No
sepa
rate
cyc
list/
pade
stri
an p
athw
ay
Brak
e Fa
ilure
Stee
ring
pro
blem
Indi
cato
r Li
ght F
ailu
re
Oth
ers
(No
reve
rse
mir
ror)
Tyre
Bur
st
Stre
et d
ogs
Physical Behavioral System Infrastructure Vehicular Issues Ex-ter-nal Is-
sues
0
20
40
60
80
100
120
2 13 11 2 6 28 1 0 11 5 107 40 1 11 3 6 11 6 17 13 6 11 4 4 0 6 2 6
Frequency
PHYSICAL12%
BEHAV-IORAL48%
SYSTEM12%
IN-FRAS-TRUC-TURE18%
VEHICULAR ISSUES7%
EXTERNAL3%
Factor Contribution Based on Occurrence
Physical 28
Behavioral 109
System 27
Infrastructure 42
Vehicular Issues 16
External Issues 6
Factor Contribution Based on Frequency
Physical 34
Behavioral 192
System 32
Infrastructure 53
Vehicular Issues 16
External Issues 6
Physical10%
Behav-ioral58%
System10%
Infra-structure16%
Vehicular Issues5%
External Issues2%
Sub factor Contribution in Physical Factor
Count cumulative count
Physical
Divider 2
34
Spillage (material/water
)13
Environmental (Illumination)
11
Height Barrier 2
Drop Gate 6
Divider6%
Spillage (mate-rial/wa-
ter)38%
Envi-ron-
mental (Illu-mina-tion)32%
Height Barrier
6%
Drop Gate18%
PHYSICAL
Sub factor Contribution in Behavioral Factor
Count cumulative count
Behavioral
Overtaking 28
192
Alcoholic 1
Sleepy 0
Not knowledgeable 11
Traffic norms violation 5
Careless Driving
107
High Speed 40
Over-taking
15%
Al-co-
holic1%
Not knowledgabl
e6%
Traffic norms viola-tion3%
Careless Driving56%
High Speed21%
BEHAVIORAL
Sub factor Contribution in System Factor
Coun
tcumulative count
System
No Zebra Crossing 1
32
Traffic Light 11
Convex Mirror 3
Others (supervision, route survey)
6
No Sign Board 11
No Zebra Crossing3%
Traffic Light 34%
Convex Mirror
9%
Others (supervision, route survey)
19%
No Sign
Board34%
SYSTEM
Sub factor Contribution in Infrastructure Factor
Countcumulative count
Infrastructure
Bad Road Condition
/intersection issue6
53
level crossing 17
Narrow Road (Congestion) 13
Blind Turn 6
No separate cyclist/pedestrian
pathway11
Bad Road Condition /intersec-
tion issue11%
level cross-
ing32%
Narrow Road (Con-gestion)
25%
Blind Turn11%
No sepa-rate
cyclist/pades-trian path-way21%
INFRASTRUCTURE
Sub factor Contribution in Vehicular Issues Factor
Count cumulative count
Vehicula
r Issues
Brake Failure 4
16
Steering problem 4
Indicator Light Failure 0
Others (No reverse camera)
6
Tyre Burst 2
Brake Failure
25%
Steering problem
25%Others (No reverse mirror)
38%
Tyre Burst13%
VEHICULAR ISSUES
Factor Contribution in total Injury Score
PROCEDURE: Assumption : Each contributing sub factor is equally responsible for any
incident.
For every incident, Score is divided by no. of sub factors to get score/sub factor.
Finding no. of sub factors of each factor in each incident.
Multiply no. of sub factors to score/sub factor to get contributing score of each factor.
Add all incident’s score for each factor.
Factor Contribution in Total Injury Score
FactorsInjury Risk
Score
Physical 87.85
Behavioral 558.12
System 126.53
Infrastructure 130.50
Vehicular Issues
62.33
External Issues
25.67
physical9%
Behavioral56%
System13%
Infra-structure
13%
Vehicular Issues6%
External Issues
3%
Contribution in total risk(Injury)
Factor Contribution in Total Property Damage
FactorsProperty
damage Risk score
Physical 53.32
Behavioral 280.53
System 69.05Infrastructur
e28.27
Vehicular Issues
75.33
External Issues
0.50
physical11%
Behav-ioral55%
System14%
Infra-structure6%
Vehicu-lar Is-sues15%
External Issues0%
Contribution in total risk(Property Damage)
Comparison of contribution based on Frequency and Total
Risk
Physical10%
Behav-ioral58%
System10%
Infra-structure16%
Vehicular Issues5%
External Issues2%
FACTOR CONTRIBUTION BASED ON FREQUENCY
phys-ical9%
Behavioral56%
System13%
Infra-structure13%
Vehicular Issues6%
Ex-ternal Issues
3%
FACTOR CONTRIBUTION IN TOTAL RISK(INJURY)
Comparison of contribution based on Frequency and Total
Risk
Physical10%
Behav-ioral58%
System10%
Infra-structure16%
Vehicular Issues5%
External Issues2%
FACTOR CONTRIBUTION BASED ON FREQUENCY
physical11%
Behav-ioral55%
System14%
Infra-structure6%
Vehicu-lar Is-sues15%
External Issues0%
FACTOR CONTRIBUTION IN TOTAL RISK(PROPERTY DAM-
AGE)
Correlation between different organizational
measures and number of incidents
Effect of Speed violations on Total incidents due to high speed
MonthSpeed
Violations
No. of incidents(High speed)
April 334 2May 414 2June 326 0July 410 0
August 296 0September 321 1
October 330 14November 311 1December 231 7
January 215 5February 126 4
March 114 3
April
May
June Ju
ly
Augus
t
Sept
embe
r
Octob
er
Novem
ber
Decem
ber
Janu
ary
Febr
uary
Mar
ch0
50
100
150
200
250
300
350
400
450
334
414
326
410
296321 330
311
231215
126 114
2 2 0 0 0 1 14 1 7 5 4 3
Speed violations vs No. of in-cidents due to High Speed
Speed Violations No. of incidents(Highspeed)
MONTH
FR
EQ
UEN
CY
Effect of Speed violations on Total incidents
Effect of Heavy vehicle inspection on Total incidents due to vehicular issues
MONTHSNo. of
inspections
No. of incidents
due to vehicular
issuesApril 434 1
May 362 2
June 403 0
July 368 0
August 308 2
September 330 0
October 334 2
November 417 4
December 395 2
January 528 0
February 421 1
March 288 2
April
May
June Ju
ly
Augus
t
Sept
embe
r
Octob
er
Novem
ber
Decem
ber
Janu
ary
Febr
uary
Mar
ch0
100
200
300
400
500
600
1 2 0 0 2 0 2 4 2 0 1 2
434
362403
368
308 330 334
417 395
528
421
288
No. of HV inspection vs No. of incidents due to vehicular issues
No. of incidents due to vehicular issues No. of inspections
MONTH
FR
EQ
UEN
CY
Effect of HV inspection on Total incidents
Effect of R-SAP on Total incidents due to careless driving
MonthR-SAP
Conducted
No. of Incidents due
to careless driving
April 106 5
May 98 8
June 116 8
July 101 6
August 89 9
September 92 11
October 110 18
November 102 8
December 115 10
January 104 8
Apr'13 May'13 June'13 July'13 Aug'13 Sep'13 Oct'13 Nov'13 Dec'13 Jan'140
20
40
60
80
100
120
140
10698
116
101
89 92
110102
115
104
5 8 8 6 9 1118
8 10 8
R-SAP conducted Vs No. of incidents due to Careless driving
R-SAP Conducted Careless Driving
MONTH
FR
EQ
UEN
CY
Effect of R-SAP on Total incidents
Y MONTH
Speed Violation
Heavy Vehicle Checking
RSAP
6 1 334 434 10610 2 414 362 989 3 326 403 1167 4 410 368 10111 5 296 308 8911 6 321 330 9218 7 330 334 1109 8 311 417 10213 9 231 395 1158 10 215 528 10413 11 126 421 1038 12 114 288 98
Analysis of Variance
Source DF Sum of Squares
Mean Square
F Value Pr > F
Model 3 26.757535 8.919178 0.78 0.538
Error 8 91.492465 11.436558
Corrected Total 11 118.25
Model Fit Statistics
R-Square 0.226 Adj R-Sq. -0.0639
Analysis of Maximum Likelihood Estimates
Parameter DF EstimateStandard
ErrorF Value Pr > |t|
Intercept 1 0.4805 13.2828 0.34 0.745
HVC 1 -0.0235 0.0173 -1.36 0.212
RSAP 1 0.1562 0.1376 1.14 0.289
SV 1 -0.00463 0.0106 -0.44 0.674
Forecasting using Decomposition
Method
Month Forecast Actual Sum
Square Error
September 12.62 11.00 2.61
October 17.48 18.00 0.27
November 7.00 9.00 3.98
December 11.03 13.00 3.88
January 7.39 8.00 0.37
February 13.41 13.00 0.17
March 8.60 8.00 0.36
Validation for number of Incidents
MSE 1.67
Model Building Data: April2011-Aug2013 Validation Data: Sep2013-March2014
Forecast for number of incidentsMonth Polynomial Actual FY-15April 7.23 10May 10.61 10June 14.49 8July 10.00
August 13.53 September 6.77
October 10.29 November 7.52 December 11.46
January 8.39 February 7.87
March 11.51
Quarter wise
0 2 4 6 8 10 1202468
101214
Injury per month
2011-12 2012-13 2013-14
Month
No.
of I
njur
ies
Validation for number of injury
Month Forecast (2013-14)
Actual 2013-14 SS
September 7.74 9.00 1.5952
October 10.08 13.00 8.5452
November -0.75 3.00 14.065
December 4.92 8.00 9.4715
January 5.09 7.00 3.6291
February 6.27 8.00 3.0020
March -0.39 1.00 1.9424
MSE 6.0357
Model Building Data: April2011-Aug2013 Validation Data: Sep2013-March2014
Forecast for No. of Injury type IncidentsMonth Polynomial Actual FY-15
April 4.65 4May 8.83 4June 9.60 5July 8.64
August 9.73 September 7.39
October 7.61 November 5.96 December 10.47
January 8.17 February 9.36
March 8.46 Forecast for First Aid
Forecast for LTI
Prescriptions: In 2013-14, 6 incidents happened due to street dogs which mostly resulted in LTI.
Inside a steel plant, presence of street dogs should not be tolerated and street actions should be taken to eliminate this hazard.
In some incidents, illumination was also a major contributing factor. (E.g. Near pellet plant level crossing, illumination is poor as well as road condition is also poor.) These types of roads should be identified and proper action should be taken)
Over 3 years, there is an increasing trend of incidents near Diamond crossing; even then there is no traffic signal at the crossing. It should be implemented as soon as possible. Also at major crossings (Diamond, LD#2, LD#3, Cabin#4), most near misses are either due to crossing with high speed or carelessness of cyclists/pedestrians. To reduce this no. ,there must be a speed limit (say 20kmph) at all crossings and if possible separate pathways for cyclists/pedestrians.
In some areas, proper sign boards are not present at roads. This should be immediately identified and implemented.
Vehicle failure (mainly steering jammed and brake failure) may lead to a very serious accident inside plant. So these types of incidents should have separate investigation to take preventive actions.
In case of property damages, more than 50% incidents happened due to dashing in which a major contribution is, while reversing a heavy vehicle. So if we strictly implement the availability of reverse camera, these types of incidents can be reduced.
As far as behavioral related issues are concerned, careless driving and high speed are long term issues. But sleepiness and alcohol must not be tolerated. In such cases heavy penalty should be imposed to avoid any future serious accident.
Spillage prone roads should be identified (like near sinter and pellet plant) and there should be warning about spillage to avoid any skidding/slipping.
With the help of regression it is evident that heavy vehicle inspection and speed violation checking are helping in reducing road accidents but RSAP is not up to the mark. More needs to be done in that area.
Thank You
Rohit RajSonu Kumar
Backup Slides
Incidents/Accidents
on Road(124)
Incidents/Accidents
on Road(124)
Cause and Effect diagram for Road IncidentsCause and Effect diagram for Road Incidents
External Factors(17)External Factors(17) Human Factors(83)Human Factors(83)
Material /Water Spillage
Sudden appearance of Street Dogs/animals
Stones and other obstructions
Driver mentally stressed or tired Unskilled in
driving/driving a new Vehicle
Lack of sleep, carelessness (taking risks, overtaking)
Road conditions(15)Road conditions(15)
Sharp and blind turns
Slope (increasing or decreasing)
Uneven or damaged road
Steering jammed
Fail safe brake, blinker, wiper, hand brake not working
Tyre bursted
Vehicle Breakdown (9)Vehicle Breakdown (9)
Low Visibility due to fog or night time
Environmental Factors (N/A) Environmental Factors (N/A)
Rainy or windy conditions
(seasonality)
Cause type Pareto
Real Causes
CAUSES Frequency cum. percent
Human Factors 83 8366.935483
9
External Factor 17 10080.645161
3
Road Condition 15 11592.741935
5Vehicle
Breakdown9 124 100
Human Factors
External Factor
Road Condition
Vehicle Breakdown
0
10
20
30
40
50
60
70
80
90
0
20
40
60
80
100
Pareto Chart freque...
CAUSE
FR
EQ
UEN
CY
CU
MM
PER
CEN
T
Real Causes
CAUSES Frequency cum. percent
Human Factors 105 105 75
Road Condition 14 119 85
Vehicle Breakdown 12 131 93.5714286
External Factor 9 140 100
0
20
40
60
80
100
120
0
20
40
60
80
100
Pareto Chart Freque...
CAUSE
FR
EQ
UEN
CY
CU
MM
PER
CEN
T
2013-14 2012-13
Month type ParetoMonth Frequency Cum. Freq. percent
April,13 6 6 0.04878
May,13 10 16 0.130081
June,13 9 25 0.203252
July,13 7 32 0.260163
August,13 11 43 0.349593September,1
311 54 0.439024
October,13 18 72 0.585366
November,13 9 81 0.658537
December,13 13 94 0.764228
January,14 8 102 0.829268
February,14 13 115 0.934959
March,14 8 123 1
April,
13
May,1
3
June
,13
July
,13
Augus
t,13
Sept
embe
r,13
Octob
er,1
3
Novem
ber,1
3
Decem
ber,1
3
Janu
ary,
14
Febr
uary
,14
March
,14
0
4
8
12
16
Frequency
0 2 4 6 8 10 12 140
2
4
6
8
10
12
14
16
18
20
Month
Fre
qu
en
cy
2013-14
Month type ParetoMonth Frequency Cum. Freq. percent
April,12 8 8 5.633803
May,12 11 19 13.38028
June,12 21 40 28.16901
July,12 9 49 34.50704
August,12 21 70 49.29577
September,12 6 76 53.52113
October,12 8 84 59.15493
November,12 9 93 65.49296
December,12 10 103 72.53521
January,13 11 114 80.28169
February,13 10 124 87.32394
March,13 18 142 100
April
,12
May
,12
June
,12
July,1
2
Augu
st,1
2
Sept
embe
r,12
Octob
er,1
2
Novem
ber,1
2
Decem
ber,1
2
Janu
ary,
13
Febr
uary
,13
Mar
ch,1
30
5
10
15
20
25
811
21
9
21
68 9 10 11 10
18
No. of incidents
0 2 4 6 8 10 12 140
5
10
15
20
25Scatter Plot Frequency
2012-13
Month wise comparison
0 2 4 6 8 10 12 140
5
10
15
20
25
frequency 13 frequency 14
Cluster Analysis for 2013-2014
Plot of incidents based
on injury
Fatal
LTI
First Aid
No Injury
Plot of incidents based on property
damage risk score19 to 25
13 to 18
7 to 12
1 to 6
Plot of incidents based on both type risk score
19 to 25
13 to 18
7 to 12
1 to 6
Clustering of
incident location
sRepresents
cluster
6 4 8 5 7 2 1 30
5
10
15
20
25
30
35
0
20
40
60
80
100
120
frequency cum %
Cluster NO.
Fre
quency
Cum
m.
perc
ent
Cluster Means Cluster Xcord Ycord count cum percent Important locations
6 826.25 722.84 32 32 26.0163 LD 2 traffic signal, LD 3 traffic signal, level crossing near LD 3
4 709.9 254.6 30 62 50.4065 Diamond Crossing, L Town Gate, G Blast furnace Turning
8 1225.3 791.5 18 80 65.0407 HSM Gate, WRP weighbridge
5 356.56 177.44 16 96 78.0488 Security Office traffic Signal, West Plant level Crossing
7 456.22 1205.2 9 105 85.3659 Cabin 4
2 1064.1 224 8 113 91.8699 Slag Road Gate
1 127.2 492.2 5 118 95.935 West Peripheral Road
3 131.2 935.8 5 123 100 Near Power House 3 Gate
Clustering
based on risk scoreBelong to cluster 2
Belong to cluster 4
Belong to cluster 3
Belong to cluster 1
Cluster Analysis for 2012-2013
Plot of incidents based
on injury
Fatal
LTI
First Aid
No Injury
Plot of incidents based on property
damage risk score19 to 25
13 to 18
7 to 12
1 to 6
Plot of incidents based on both type risk score
19 to 25
13 to 18
7 to 12
1 to 6
Clustering of
incident location
sRepresents
cluster
Cluster
Xcord Ycord count cum. percent Location
2 805.423 387.308 26 26 18.30986 LD#2 Traffic Signal, G Blast furnace turning
4 389.962 199.692 26 52 36.61972 Security Office Traffic Signal,coke plant drop gate
6 366.077 1121.96 26 78 54.92958 cabin#4 drop gate, near merchant mill
7 1229.57 779.13 23 101 71.12676 HSM Gate, WRP Weigh Bridge
5 850.118 916.647 17 118 83.09859 LD#2 Traffic Signal, CRM Island
3 164.222 512.778 9 127 89.43662 west side peripheral road
1 517.375 578.25 8 135 95.07042 East plant drop gate
8 1155.71 315.143 7 142 100 slag road gate
Clustering
based on risk scoreBelong to cluster 3
Belong to cluster 1
Belong to cluster 2
Belong to cluster 4
2013-14 2012-13
Distribution of Time between LTIs
Distribution Parameters
Distribution Name: Weibull Alpha: 16.8
Beta: 1.21
Expression: -0.5 + WEIB(16.8, 1.21)
Square Error: 0.047390
Distribution Parameters
Distribution Name: Weibull
Alpha: 23
Beta: 1.34
Expression: -0.5 + WEIB(23, 1.34)
Square Error: 0.070247
Sample Mean = 20.8Sample Std. Dev = 15.8
Sample Mean = 15.3Sample Std. Dev = 12.2
CONTROL CHART FOR LTI
2013-14 2012-13
4/19/2
013
6/11/2
013
8/3/2
013
8/11/2
013
9/2/2
013
10/1/2
013
10/1/2
013
10/7/2
013
10/17/2
013
11/6/2
013
11/17/2
013
12/4/2
013
12/23/2
013
1/9/2
014
2/5/2
0140
10
20
30
40
50
60
70
80
CONTROL CHART FOR TBI(2013-14)
TBI Linear (TBI) UCL=69.28 LCL=0.88
D A T E
T B
I
4/1/2
012
4/29/2
012
5/14/2
012
5/21/2
012
6/21/2
012
6/22/2
012
6/30/2
012
7/30/2
012
8/4/2
012
8/9/2
012
8/16/2
012
8/19/2
012
8/22/2
012
9/17/2
012
10/4/2
012
11/2/2
012
12/1/2
012
12/17/2
012
12/25/2
012
2/1/2
013
2/1/2
0130
10
20
30
40
50
60
CONTROL CHART FOR TBI(2012-13)
TBI Linear (TBI) UCL=56.97 LCL=0.45
D A T E
T B
I
Distribution of Time between First Aid
Distribution Parameters
Distribution Name: Weibull
Alpha: 6.88
Beta: 1.27
Expression: -0.5 + WEIB(6.88, 1.27)
Square Error: 0.006406
2013-14 2012-13
Sample Mean = 8.39Sample Std. Dev = 7.43
Sample Mean : 5.88Sample Std. Dev : 5.32
Distribution Parameters
Distribution Name: Weibull
Alpha: 9.32
Beta: 1.15
Expression: -0.5 + WEIB(9.32, 1.15)
Square Error: 0.019446
CONTROL CHART FOR FIRST AID2013-14
2012-13
4/18/2
013
5/21/2
013
6/13/2
013
6/29/2
013
7/10/2
013
7/26/2
013
8/20/2
013
9/6/2
013
9/11/2
013
9/27/2
013
10/7/2
013
10/14/2
013
10/31/2
013
12/17/2
013
12/23/2
013
1/15/2
014
1/26/2
014
2/7/2
014
2/17/2
0140
5
10
15
20
25
30
CONTROL CHART FOR TBI(2013-14)
TBI Linear (TBI) UCL=22.02 LCL=0.22
D A T E
T B
I
5/7/2
012
5/9/2
012
5/29/2
012
6/19/2
012
7/10/2
012
7/25/2
012
8/2/2
012
8/16/2
012
8/31/2
012
9/13/2
012
9/26/2
012
11/10/2
012
12/5/2
012
12/26/2
012
1/11/2
013
1/28/2
013
2/16/2
013
3/7/2
013
3/19/2
013
3/22/2
0130
5
10
15
20
25
30
35
40
CONTROL CHART FOR TBI(2012-13)
TBI Linear (TBI) UCL=33.68 LCL=0.21
D A T E
T B
I
Distribution of Time between No Injury
2013-14 2012-13
Distribution Parameters
Distribution Name: Weibull
Alpha: 5.18
Beta: 1.1
Expression: -0.5 + WEIB(5.18, 1.1)
Square Error: 0.005930
Sample Mean = 4.47
Sample Std. Dev = 4.87
Distribution Parameters
Distribution Name: Weibull
Alpha: 8.76
Beta: 1.24
Expression: -0.5 + WEIB(8.76, 1.24)
Square Error: 0.015624
Sample Mean = 7.66Sample Std. Dev = 6.64
CONTROL CHART FOR NO INJURY2013-14
2012-13
4/5/2
012
4/13/2
012
5/7/2
012
5/26/2
012
6/12/2
012
6/15/2
012
6/26/2
012
6/27/2
012
7/12/2
012
7/23/2
012
8/14/2
012
8/25/2
012
8/30/2
012
9/29/2
012
10/12/2
012
10/27/2
012
11/17/2
012
11/29/2
012
12/10/2
012
1/10/2
013
1/18/2
013
2/1/2
013
2/22/2
013
3/6/2
013
3/14/2
013
3/16/2
013
3/25/2
0130
5
10
15
20
25
30
CONTROL CHART FOR TBI(2012-13)
TBI Linear (TBI) UCL LCL
D A T E
T B
I4/1
/2013
4/19/2
013
5/4/2
013
5/14/2
013
5/27/2
013
6/26/2
013
7/23/2
013
8/16/2
013
8/22/2
013
9/11/2
013
10/1/2
013
10/28/2
013
11/2/2
013
11/10/2
013
11/24/2
013
12/1/2
013
12/5/2
013
12/30/2
013
2/1/2
014
2/6/2
014
2/18/2
014
3/9/2
014
3/11/2
014
3/19/2
0140
5
10
15
20
25
30
35
CONTROL CHART FOR TBI(2013-14)
TBI Linear (TBI) UCL LCL
D A T E
T B
I
Effect of Speed violations on Total incidents
MonthSpeed
Violations
No. of incidents(over
all)
April 334 6
May 414 10
June 326 9
July 410 7
August 296 11
September 321 11
October 330 18
November 311 9
December 231 13
January 215 8
February 126 13
March 114 8
0
50
100
150
200
250
300
350
400
450
334
414
326
410
296321 330
311
231215
126 114
6 10 9 7 11 11 18 9 13 8 13 8
Speed violations vs Road Incidents
Speed Violations No. of incidents(overall) MONTH
FR
EQ
UEN
CY
Effect of Heavy vehicle inspection on Total incidents
MONTHSNo. of
inspectionsTotal no. of incidents
April 434 6
May 362 10
June 403 9
July 368 7
August 308 11
September 330 11
October 334 18
November 417 9
December 395 13
January 528 8
February 421 13
March 288 8
1 2 3 4 5 6 7 8 9 10 11 120
100
200
300
400
500
600
434
362
403368
308330 334
417395
528
421
288
6 10 9 7 11 11 18 9 13 8 13 8
No. of HV inspection vs Total no. of incidents
No. of inspections Total no. of incidents
MONTH
FR
EQ
UEN
CY
Effect of R-SAP on Total incidents
MonthR-SAP
ConductedNo. of
Incidents
April 106 6
May 98 10
June 116 9
July 101 7
August 89 11
September 92 11
October 110 18
November 102 9
December 115 13
January 104 8
Apr'13 May'13 June'13 July'13 Aug'13 Sep'13 Oct'13 Nov'13 Dec'13 Jan'140
20
40
60
80
100
120
140
10698
116
101
89 92
110102
115
104
6 10 9 7 11 1118
913
8
R-SAP conducted Vs No. of incidents
R-SAP Conducted No. of Incidents
MONTH
FR
EQ
UEN
CY
USING DECOMPOSITION METHOD (Quarter wise)
Forecast for No. Of IncidentsPolynomial Quarter Actual
29.7 Q1 2828.3 Q2 26.9 Q3 23.4 Q4 22.2 Q1 19.8 Q2 17.5 Q3 13.0 Q4
Forecast for First AidMonth Polynomial Actual FY-15
April 2.64 2May 7.02 4June 7.08 5July 8.11
August 7.58 September 5.88
October 5.92 November 4.75 December 8.64
January 7.93 February 7.95
March 7.82
Forecast for LTIMonth Polynomial Actual FY-15
April 2.13 2May 1.52 0June 2.42 0July 1.00
August 2.53 September 1.72
October 1.86 November 1.38 December 2.01
January 0.42 February 1.33
March 0.95
Usefulness to Tata Steel Here with the help of Cause and Effect diagram we can
identify all the possible causes responsible for the incidents taking place here related to human, vehicle, road, environmental and external factors.
Next with the help of Pareto Chart we can identify major factors responsible for the occurrence of incidents. Using this we were able to identify major type of vehicles, type of injury, type of causes, months which constitutes maximum (here 80%) of incidents.
Moreover, We can also use Pareto chart for identifying highly frequent incident occurring areas which were identified by Cluster Analysis, explained in next slide.
Cluster Analysis :
This is a mathematical tool for grouping a set of objects in such a way that objects in the same group are more similar and within cluster sum of squares is minimum.
This can be used in Tata Steel to group all incidents in a few clusters identified by their co-ordinates on a map of size (width: 1332px, height: 1447px) .
Also, these clusters can be utilised as input for predicting various injury type incidents in a particular area. Here we have used k-means clustering and have taken k =8.
Probability Distribution:
We have fit time between occurrences of incidents both frequency wise and injury type wise (No injury, First aid, LTI).
Using this fitting, we will be able to say what is the pattern of time between occurrences as well it's parameters
can be used in prediction.
E.g.- We have found that TBO follows Weibull distribution with parameters alpha = 1.26 and beta =3.64. So we can say that a particular incident will occur with a particular probability after a certain number of days.