Out of Context Curves: Version 2 Risk Ranking of State Highway Curves for Skid Resistance Monitoring

1
Out of Context Curves Version 2Risk Ranking of
State Highway Curves for Skid Resistance
Monitoring and Treatment
Presented by Colin Brodie Principal Safety
Advisor (HNO) New Zealand Transport Agency
2
Acknowledgements

• CO-AUTHORS
• Peter Cenek Research Manager, Opus Central
  Laboratories
• Dr Robert Davis Consulting Statistician
• Dr Fergus Tate MWH Technical Development Leader
• CURVE SELECTION ANALYSIS
• Dr Robert Henderson Research Engineer, Opus
  Central Laboratories

3
Topics covered in presentation

• Background
• Identification of Curves
• Derivation of Crash Risk Model
• Trial application in Hawkes Bay and Rotorua
  Networks

4
Problem Backgound

• 1/3 of all SH rural crashes occur on bends
• Equates to over 1100pa
• 48 on wet roads (gt 500pa)
• Transit NZs skid resistance policy only targets
  increased skid resistance on curves lt 250mR
• Majority of the more severe crashes are occurring
  on moderate to easy radii curves

5
The Evidence
6
Australian Experience
Peak of injury crashes 300m Radius Peak of
fatal crashes 400m Radius
7
Present Situation T10 Specification

• 5 Categories
• Curves lt 250mR (Cat2) IL 0.5, TL 0.4
• Curves gt 250mR (Cat4) are event free,
  no geometric constraint (IL 0.4, TL 0.3)
• Cat 4 (gt250mR) curves are being managed to TL
  which is a low Skid Resistance (Too Late)

8
Solution Risk Ranking of Curves

• Determine the crash risk of each curve
  individually
• SCRIM survey HSD geometry data used to determine
  approach and curve speeds
• Difference in speeds , known as OoCC factor, used
  to rank curves as low, medium and high risk (OoCC
  version 1 2007)
• Refinement is to use crash prediction modelling
  (OoCC version 2)

9
Curve Identification Procedure

• 10m HSD segments used from 2008/09 SCRIM survey
• 30m rolling averages identified
• Curve starts and ends
• - 30m ave gt 800mR
• Curve radius and speed
• - Tightest 30m Ave
• - All curves below 500mR
• Approach speed
• - Ave of 500m 10m segment speeds in both
  directions
• RPs and co-ordinates identified for start, end
  and length of curve

10
Curve Identification Pictorial
11
Data for Crash Risk Model Derivation

• Crash Covered Period 1997 to 2002
• Total of 95440 curves (6 years, each side of
  road)
• 3244 crashes (all reported injury, including
  fatals)

12
Histogram of curve length
6000
5000
4000
3000
2000
1000
0
100
200
300
400
500
600
700
800
900
1000
13
Histogram of curve radius increasing dirn.
14
Histogram of curve speed increasing dirn.
2500
2000
1500
1000
500
0
20
30
40
50
60
70
80
90
100
110
15
Histogram of approach speed increasing dirn.
14000
12000
10000
8000
6000
4000
2000
0
50
60
70
80
90
100
110
16
Histogram of OOCC effect increasing dirn.
10000
8000
6000
4000
2000
0
0
10
20
30
40
50
60
70
80
17
Histogram of SCRIM coefficientcurvature(m) lt 250
5000
4000
3000
2000
1000
0
0.3
0.4
0.5
0.6
0.7
18
Histogram of SCRIM coefficient250 curvature
(m) lt 500
4000
3500
3000
2500
2000
1500
1000
500
0
0.3
0.4
0.5
0.6
0.7
19
Crash rate versuslength
25
20
15
Crash rate per curve
10
5
0
0
100
200
300
400
500
Length
20
Crash rate versusskid resistance
25
20
15
Crash rate per curve
10
5
0
0.3
0.35
0.4
0.45
0.5
0.55
0.6
0.65
0.7
Scrim coefficient
21
Crash rate versusdifference between approach and
curve speeds
25
20
15
Crash rate per curve
10
5
0
0
10
20
30
40
50
Difference between approach and curve speeds
(km/hr)
22
Crash rate versusADT
25
20
15
Crash rate per curve
10
5
0
100
1000
10000
ADT
23
Trial Application of Model

• Crash rate per 108 veh entering curve
• Low Risk, personal risk lt7
• Medium Risk, .7 personal risk 14
• High risk, personal risk gt14

24
Trial application of model
Low Risk
High Risk
25
Trial application implications

• 18 km of curves lt 250mR would have IL decreased
  from 0.5 to 0.4
• 38 km of curves lt 250mR would have IL increased
  from 0.5 to 0.55
• 39 km of curves gt 250mR would have IL increased
  from 0.4 to either 0.5 or 0.55
• Cost Implications!!!

26
The Future

• Incorporate approach gradient in model
• Establish most appropriate form for curve length
  in model (linear or log-linear)
• Assign an appropriate IL to each curve based on
  RISK
• Determine cost and economic effects if new
  procedure
• Hard wire curves into RAMM with IL
• Possible national roll out in 2009/10

27
Thank You
Colin Brodie Principal Safety Advisor (HNO) New
Zealand Transport Agency PO Box 430 Tauranga New
Zealand Tel [phone] Email
colin.brodie_at_nzta.govt.nz