Safety Impacts of Various Crash

Countermeasures
Xiaoduan Sun
UL Lafayette
February 19, 2013

2013 Louisiana Transportation Conference

 Outline
• Crash Countermeasures
• Converting urban undivided 4-lane roadway to
five-lane roadway
• Edgeline on narrow rural 2-lane highways
• Raised pavement markers
 Crash Countermeasures
• Eliminating all crash contributing factors

Human Vehicle Environment

Pre-Event

Event

Post-Event

Haddon Matrix- a useful framework for thinking about the

complexities of a crash
 Human
Road Factors (95%)
Environment
Factors (28%)

Keep in mind that 4% 24% 67%

everything we do must
accommodate humans’
needs and match vehicles'
4% 4%
capability

Vehicle
Factors (8%)
Based on Interactive Highway Safety Design Model:
by Harry Lum and Jerry A. Reagan
 Eliminating crash contributing factors
with crash countermeasures

Crash is not an
accident, it is
preventable. Crash
reduction can not
happen by chance

Source: Safer Roads: A Guide to Road Safety Engineering. K.W. Ogden. Ashgate
 CMF from HSM
# of Crash
Chapter in # of CMFs # of Crash
Countermeasures
HSM Part Content Proven to be Countermeasures with
with Known Safety
D effective Unknown Safety Effect
Effect
13 Roadway Segments 36 43 72
14 Intersections 24 27 84
15 Interchange 4 8 25
Special Facilities and
16 5 16 68
Geometric Situations
17 Road Network 3 16 5
Total 72 110 254
 Developing CMF for Louisiana
• While the majority of crash countermeasures
would be the same as the once used by other
states, a few countermeasures will be unique
in Louisiana
• Introducing few CMFs tailed to the unique
situation in the state is the objective of this
presentation
 Outline

• Crash Countermeasures
• Converting urban undivided 4-lane
roadway to five-lane roadway
• Edgeline on narrow rural 2-lane highways
• Raised pavement markers
Urban undivided multilane highways consistently
exhibit low safety performance in the U.S.
Representative Accident Rates by Location and Type of Road
Injury Total
Fatal Accidents Accidents Accidents
Number per Number per Number per
RURAL MVM MVM MVM
2 Lanes 0.07 0.94 2.39
4 or more lanes,
divided subtotal 0.063 0.77 2.09
Freeway 0.025 0.27 0.79
URBAN
2 Lanes 0.045 1.51 4.94
4 or more lanes,
undivided 0.04 2.12 6.65
4 or more lanes,
divided 0.027 1.65 4.86
Freeway 0.012 0.4 1.43

9
– 1,530 miles of undivided multilane roadways
under LADOTD system. 93% these roadways are in
urban and suburban areas

10
Solutions?
• Expensive solution: installing physical
separation either by barrier or by green
space (boulevard) has been the most
recommended crash countermeasure for
the problem

11
Solutions?
• Inexpensive option: with sufficient pavement
width, a four-lane undivided highway can also
be easily changed to a five-lane roadway with
the center lane for left-turns, which
expectedly reduces rear-end collisions.

12
The five-lane design alternative including a center TWLTL in
the median has, in the past 20 years, become a very common
multilane design alternative for upgrading urban arterials. This
design alternative has two through lanes of travel in each direction
and a center TWLTL to provide for left-turn maneuvers at
driveways and minor intersections. The total roadway width for
a five-lane TWLTL section on an urban arterial ranges from 48
ft to 72 ft depending on the lane widths employed.

From NCHRP 330, 1990

13
Pros and Cons of Two Options

• Physical barrier • Five lane

– Better traffic (motorized – Inexpensive with
or non-motorized) sufficient ROW
management – Not recommended for
– Expensive new road in Louisiana

14
However

• Under the current budgetary situation, the

expensive option is not financially feasible
• Going with the inexpensive but not perfect
solution to reduce the crashes has been one
option for the situation
• Several roadway segments in various LADOTD
districts have implemented this inexpensive
crash countermeasure in the past

15
 Four segments selected for the
analysis
Estimated
Control Installation
District Length (mi) # of Location
Section Year
Driveways
LA 3025 D3 828-23 1.228 2003 45 Lafayette

LA 182 D3 032-02 1 2007 50 Opelousas

LA 28 D8 074-01 0.92 2005 45 Alexandria

LA 1138 D7 810-06 1.07 1999 50 Lake Charles

16
Roadway Configuration

LA3025

17
LA 3025 (from 2012 Google Earth)

18
LA182

19
LA182 (from 2012 Google Earth)

20
 LA1138

(from 2012 Google Earth)

LA28

21
 Summary of Crashes
(3 years before and after)
Before After Percentage Change

Average Average
Crashes Crash Crashes Crash Crashes Crash Rate
Rate Rate

LA3025 358 10.05 147 4.59 -59% -54%

LA182 178 8.12 85 3.53 -52% -51%
LA28 206 7.38 99 4.09 -52% -45%
LA1138 260 16.01 167 10.63 -36% -34%

22
 Crash Frequency Crash Frequency
He He

0
50
100
150
200
250
a

20
40
60
80
100

0
ad
Le d-On Le -On
ft ft
Tu Tu
Le rn-
ft e Le rn-e
T ft
Tu
Le urn- rn
ft Le -f
Tu f ft
r Tu
No n-g rn
nC No -g
Re oll nC
Rig ar-E Re oll
ht nd a
T Rig r-E n
ht d
Rig urn
ht -h Tu
Tu rn

LA182
-h
LA3025

rn Sid Rt.
Sid Rt. -i es A

After
A

Total
Total
es w ng
le

Before
w i ngl
p e e Sid ipe (
Sid ( O es OD
es wi )
After
Total
Total

w i D) pe
Before

p( ( SD
SD )
)
Bla Bla
nk nk
Ot Ot
he he
r r
Changes by Crash Type

23
Changes by Pavement Surface Condition
LA3025 LA182

300 180
160 Before
250 Total
Before 140
Total
Crash Frequency

Crash Frequency
200 120 After
After Total
100
150 Total
80
100 60
40
50
20
0 0
Dry Wet Dry Wet
Pavement Surface Condition Pavement Surface Condition
LA28 LA1138

250 Before
160 Before Total
140 Total
Crash Frequency
200
Crash Frequency

120 After
After 150 Total
100 Total
80
100
60
40 50
20
0 0

Dry Wet Dry Wet

Pavement Surface Condition Pavement Surface Condition

24
 Changes by Time of the Day
LA 3025 LA 182
120
200
100
160 Before Before

Crash Frequency
Total Total
Crash Frequency

80
120 After After
Total 60 Total
80
40

40 20

0 0
6am- 12pm- 6pm- 12am- 6am- 12pm- 6pm- 12am-
12pm 6pm 12am 6am 12pm 6pm 12am 6am

LA 28 LA 1138
120 160

140 Before
100 Before Total
Total
Crash Frequency

Crash Frequency 120 After

80 After Total
Total 100

60 80

60
40

40
20
20

0 0
6am- 12pm- 6pm- 12am- 6am- 12pm- 6pm- 12am-
12pm 6pm 12am 6am 12pm 6pm 12am 6am
25
 Changes by Crash Severity

Crashes LA3025 LA182 LA28 LA1138

by % % % %
Severity Before After Before After Before After Before After
Change Change Change Change

Total 358 147 -58.90% 178 85 -52.30% 206 99 -51.94% 260 167 -35.77%

PDO 277 105 -62.10% 124 63 -49.20% 148 76 -48.68% 172 119 -30.81%

Injury
81 40 -50.60% 54 22 -59.30% 58 23 -60.34% 88 48 -45.45%
Crashes

Fatal 0 2 increase 0 0 0% 0 0 0% 0 0 0%

26
 Benefit/Cost Ratio
• Benefit—saving from reduced crashes
• Cost – striping LA 3025 LA 182 LA 28 LA 1138
Severity
• B/C=166! Level
PDO
Reduction Reduction Reduction
172 61 72
Reduction
53
Injury 41 32 35 40

Segment Total Benefits ($) Total Cost ($) B/C Ratio

LA 3025 2,753,868 14,100 195
LA 182 1,913,808 11,500 166
LA 28 2,110,212 10,600 199
LA 1138 2,317,488 12,300 188

27

ˆ
ˆ

CMF Results
Expected
Standard Estimated Standard
Crash
Deviation the CMF Deviation
Reduction

LA3025 175 27.62 0.45 0.051

LA182 110 20.53 0.43 0.062

LA28 111 21.28 0.47 0.062

LA1138 87 25.42 0.65 0.075

28
 What does the result mean?
A certainty in crash reduction
Estimated Standard CMF+
Roadway
CMF Deviation 3*Standard Deviation
LA3025 0.45 0.051 0.60
LA182 0.43 0.062 0.62
LA28 0.47 0.062 0.66
LA1138 0.65 0.075 0.88

Probabilty Distribution Probability Distribution

LA1138 LA3025

0.9989

0.65 0.88 0.45

CMF Value CMF Value

29
Probability Distribution Probability Distribution
Results Discussion
• The crash reduction by the re-striping/lane
conversion projects is striking and the
estimated CMF is impressive (crash
countermeasures, as listed in the first edition
of the HSM, seldom yield CMF values smaller
than 0.5)
• The estimated CMF and standard deviation on
all roadway segments indicate a certainty that
a re-striping project reduces crashes.

30
Results Discussion
• Reductions are consistent cross crash category
• It is a very cost-effective crash countermeasure
• Demonstrating the need for flexibility in selecting
the best safety improvement project under the
existing constraints (financial or otherwise).
• If and when funds do become available and
sufficient right-of-way (ROW) can be obtained,
these two 5-lane roadway segments can be
converted to a boulevard roadway type, a
concept very much promoted today in urban and
suburban areas in Louisiana

31
 Sustainable crash reduction
LA3025
3 years before
150
Crash Frequency

3 years after 3 years after after

100

50

0
2000 2001 2002 2004 2005 2006 2008 2009 2010
Year

32
3 years before 3 years after after
3 years after

Hurricane Rita

3 years after after

3 years before

3 years after

33
 CMF as a function of AADT

AADT vs. Estimated CMF

y = 3E-09x 2 - 0.0001x + 1.8028
0.9 R2 = 0.996
0.8
0.7
0.6
CMF

0.5
0.4
0.3
0.2
10,000 15,000 20,000 25,000 30,000
AADT

34
Due to the huge success of the lane-conversion
project, more segments from LADOTD District 3 have
been recently re-striped:

• LA 14-Bypass in Abbeville
• LA 14 in Abbeville
• US 190 in Eunice
• LA 93 in Sunset
• LA 14 in New Iberia

35
 Acknowledgement

• Mr. Nick Fruge from District 3

• Ms. Bridget Webster from District 8
• Mr. Jason Roberson from District 4
• Mr. Tyson Thevis from District 7
 Outline
• Project background
• Converting urban undivided 4-lane roadway
to five-lane roadway
• Raised pavement markers
• Edgeline on narrow rural 2-lane highways
• Raised pavement markers (delineation
purpose)
The need to have Louisiana CMF on Raised
Pavement Markers (RPM)
CMF from the HSM
Setting Traffic Volume Crash Type
(Road Type) (AADT) (Severity) CMF Std. Error
≤ 20,000 Nightime 1.13 0.2
Rural 20,001-60,000 All Types 0.94 0.3
(Four-lane Freeways) >60,000 (All Severities) 0.67 0.3
Ref: Bahar, G., C. Mollett, B. Persaud, C. Lyon, A. Smiley, T. Smahel, and H. McGee. National
Cooperative Highway Research Report 518: Safety Evaluation of Permanent Raised
Pavement Markers. NCHRP, Transportation Research Board, National Research Council,
Washington, DC, 2004.

Should the state continue the practice?

Developing CMF for RPM

• Data
– Annual RPM and striping ratings
– Crash
• Analysis
– By setting (urban vs. rural)
– By time (nighttime vs. daytime)
 Ratings
• Three condition ratings:
– ‘G’ as Good
– ‘P’ as Poor
– ‘F’ as Fair
• Rating ‘C’ as Construction
2002 2003 2004 2005 2006 2007 2008 2009 2010
Control Section
Section Length

450-91 2.54 G G P G G F F F P
450-92 1.36 F F G G G F F F P
450-93 3.40 F F G G G F F F P
450-94 1.17 F F G G G F F F P
450-95 0.13 F F G G G F F F P
450-96 0.38 F F G G G F F F P
 Summary

Number of Segments in Each Rating Group in Nine years

Freeway
GG GF GP FG FF FP PG PF PP

Rural 606 85 171 63 110 140 75 31 285

Urban 1,028 189 280 156 214 266 141 88 734

Total 1,634 274 451 219 324 406 216 119 1,019
Average Crash Rate by Combined Ratings on
Rural freeways

Rural and 24 hours Rural and night hours

0.9 0.817 0.25
0.8 0.760
0.196
0.666 0.2
0.7

Avg. Crash Rate

Avg. Crash Rate

0.159 0.163
0.6
0.15
0.5
0.4 0.1
0.3
0.2 0.05
0.1
0 0
GG FF PP GG FF PP

Striping and RPM rating Striping and RPM rating

23% increase 23% increase

Average Crash Rate by Combined Ratings on
Urban freeways

Urban and 24 hours Urban and Night hours

2.5 0.5
2.113 2.077
2.005 0.406
0.384 0.369
2 0.4

Avg. Crash Rate

Avg. Crash Rate

1.5 0.3

1 0.2

0.5 0.1

0 0
GG FF PP GG FF PP

Striping and RPM rating Striping and RPM rating

 Average crash rate by single rating on rural
freeways
Rural and 24 hours Rural and 24 hours
0.9 0.9
0.8 0.760 0.8
0.724 0.692 0.706
0.675 0.7 0.658
0.7

Avg. Crash Rate

Avg. Crash Rate

0.6 0.6
0.5 0.5

0.4 0.4

0.3 0.3
0.2
0.2
0.1
0.1
0
Striping 0
G F P
G F P

Striping rating RPM rating RPM

Rural and night hours Rural and night hours
0.25 0.25

0.2 0.180 0.178 0.2

0.161 0.165 0.168
Avg. Crash Rate

Avg. Crash Rate

0.152
0.15 0.15

0.1 0.1

0.05 0.05

0 0
G F P G F P

Striping rating RPM rating

 Results of Statistical Test
(Average Crash Rate between Good and Poor)
t-test for Equality of Means

Roadway Crash
Feature 95% Confidence Interval of
Type Rate at
Mean Std. Error the Difference
t df
Difference Difference
Lower Upper
AADT ≤ 20,000
Rural RPM Night -1.781 489 -0.033 0.018 -0.069 0.003
Rural RPM 24 Hrs -1.101 489 -0.065 0.059 -0.181 0.051
Rural RPM+Striping Night -2.603 309 -0.063 0.024 -0.110 -0.015
Rural RPM+Striping 24 Hrs -2.591 309 -0.212 0.082 -0.373 -0.051
20,000≤AADT ≤ 60,000
Rural RPM Night -2.665 816 -0.038 0.014 -0.066 -0.010
Rural RPM 24 Hrs -3.249 816 -0.142 0.044 -0.228 -0.056
Rural RPM+Striping Night -2.285 492 -0.047 0.020 -0.087 -0.007
Rural RPM+Striping 24 Hrs -2.840 492 -0.168 0.059 -0.284 -0.052
AADT ≤ 60,000
Rural RPM Night -2.128 1339 -0.025 0.012 -0.049 -0.002
Rural RPM 24 Hrs -2.573 1339 -0.102 0.040 -0.180 -0.024
Rural RPM+Striping Night -2.800 889 -0.045 0.016 -0.077 -0.013
Rural RPM+Striping 24 Hrs -3.504 889 -0.186 0.053 -0.289 -0.082
 Highway Crash
Feature Rating N Mean CMF

CMF Type
AADT≤ 20,000
Hour

291 0.139
Rural RPM Night Good 0.81

Development Rural RPM 24 Hrs

Poor
Good
200
291
0.172
0.635 0.91
Poor 200 0.7
Rural RPM+Striping Night Good 225 0.138 0.69
• Crash rate is used for the Poor 86 0.201
analysis Rural RPM+Striping 24 Hrs Good 225 0.644 0.75
Poor 86 0.856
• Only “Good’’ ratings and “Poor” 20,000 ≤ AADT≤ 60,000
ratings are considered Rural RPM Night Good 436 0.141 0.79
• Nine years data is used for both Poor 382 0.179
Rural RPM 24 Hrs Good 436 0.596 0.81
ratings Poor 382 0.738
Rural RPM+Striping Night Good 329 0.148 0.76
Poor 165 0.195
Rural RPM+Striping 24 Hrs Good 329 0.602 0.78
Poor 165 0.77
AADT≤ 60,000
Rural RPM Night Good 745 0.153 0.86
Poor 596 0.178
Rural RPM 24 Hrs Good 745 0.655 0.87
Poor 596 0.757
Rural RPM+Striping Night Good 606 0.155 0.78
Poor 285 0.2
Rural RPM+Striping 24 Hrs Good 606 0.655 0.78
Poor 285 0.841
 Results Discussion
• RPM does offer safety benefit to the state
rural freeways based on all analysis methods
• Because of combined effects of two ratings, it
is hard, if not impossible, to accurately
estimate CMF for RPM
• It is conservative to say CMF on RPM is about
0.90
• No safety benefit of RPM is detected on urban
freeways
 Outline
• Project background
• Converting urban undivided 4-lane roadway
to five-lane roadway
• Raised pavement markers
• Edgeline on narrow rural 2-lane highways
 Edgeline Requirement

Previous Updated
Current LaDOTD
MUTCD MUTCD
Policy (1994)2
(1994) (2000)1
Road
No Requirement 20-ft or Wider 22-ft or Wider
Width

Greater than
AADT No Requirement No Requirement
3,000
 Research Investigation
• The 2007 study on 10 segments of narrow rural
2-lane highways proved that:
– With the edge line, vehicles tend to move away from
the road edge; thus, the risk of having a running-off-
roadway crash is likely to be reduced
– The implementation of edge lines is likely to reduce
the head-on and sideswipe collisions at night because
of the reduced number of vehicles crossing the
centerline in the nighttime.
• The impact of edge line on crashes is also
investigated on the selected segments from all
LaDOTD districts
 after
before

Control Section Highway Number Suggestion

(District 3) Log from and to Mile post (Log Mile)

823-27 0087 Starting at milepost 4.0 for 3

0-1.89 miles (0.25 mile before the
control section)
• 389-01
after
before

Control Section Highway Number Suggestion

(D3) Log from and to Milepost (Log Mile)

389-01 0098 Starting at milepost 27 for 6 miles

2.59-7.15 (log-mile 2 for 6 miles)
 after
before

Control Section Highway Number Suggestion

(D4) Log from and to Milepost (Log Mile)

048-02 0169 Starting at milepost 22 for 5

4.72-8.29 miles (log mile 4.5 for 5 miles)
• Crash data analysis
• Three years before and three years after

2005 2006 2007 2008 2009 2010 2011

Before Installation Year After
 Total Crashes
2005 2006 2007 2009 2010 2011
DOTD
Total Total Total Total Total Total
District
2 23 34 24 19 8 17
3 86 68 67 81 85 68
4 12 16 8 12 5 6
5 84 74 85 90 99 72
7 21 30 14 10 14 17
8 16 13 15 10 14 10
58 5 3 4 2 4 1
61 32 36 17 15 15 20
62 85 103 83 71 62 63
Total 345 346 295 290 299 263
 Results
Before 3 Years- After 3 Years
Naïve B-A Method 2 Improved Prediction
DOTD
Reduction in Index of Reduction Index of
District
(No. of CSECT) Crashes Effectiveness in Crashes Effectiveness

2 (1) 4 0.58 10 0.38

3 (9) -13 1.05 -17 1.07
4 (2) 13 0.62 18 0.54
5 (4) -18 1.07 1 0.99
7 (2) 24 0.62 14 0.72
8 (2) 9 0.77 13 0.69
58 (1) 5 0.54 2 0.69
61 (3) 35 0.58 44 0.52
62 (4) 75 0.72 108 0.64
 Results

Before 3 Years- After 3 Years

Improved Prediction Method

Estimated
Expected Stdev. CMF Stdev.
Reduction

194 48 0.81 0.041

Result Interpretation

  2

0.701 0.868 1.0278

 Benefit-Cost Analysis
2004-2007 2009-2011 Including Loss
Severity Safety
(Before (After Change of
Type Benefit ($)
Years) Years) Quality of Life

Fatal 12 13 -1 4,376,304 -4,376,304

Injury 424 341 83 71,139 5,904,537

PDO 550 498 52 3,292 171,184

Total Benefit 1,699,417

Cost ($0.15 per

Benefit B-C Ratio
foot)
All Control
86,835 1,699,417 19.57
Sections
 Putting together
• Our Analysis 0.81

Safety Trend for the -5.6%

Narrow Rural 2-lane

Final Estimated CMF 0.87

HSM
CMF for Rural Two Lane

Source: Highway Safety Manual (1st Edition), Vol. 3, 2010

 Results Discussion
• Implementing edge line is most likely to
reduce number of crashes based on our crash
analysis
• The expected reduction is estimated 13%
 Summary

A very effective short-term

crash countermeasure for urban
undivided 4-lane roadway

Reducing crashes on rural freeway

Results in lower crash rate

Thank You and Questions?