Comparing Simulated Safety Performance to Observed Crash Occurrence - - PDF document

Comparing Simulated Safety Performance to Observed Crash Occurrence

Flávio Cunto, Ph.D Universidade Federal do Ceará Frank Saccomanno University of Waterloo

Outline Outline

• Motivation
• Objectives
• Methodology
• Major results
• Concluding remarks

2

Motivation Motivation

• 1. The need for a more mechanistic analysis of risky

situations

• 2. Traffic conflict data collection methods
• 3. Advances on microscopic traffic simulation models
• 4. Is there a relative validity between simulated

measures of safety and crashes?

• 5. Linking Surrogate Measures of Safety to crashes

would improve TC paradigm?

3

Objetive Objetive

This paper describes an objective approach for linking simulated safety performance expressed in terms of the crash potential index (CPI/veh) to observed crash occurrence, with the primary focus being on rear-end crashes.

4

Microscopic Model for Safety Assessment Microscopic Model for Safety Assessment

5

Vehicle Interactions Vehicle Interactions

Rear-end: Angled:

6

Crash Potential Index (CPI) Crash Potential Index (CPI)

7

MADR

f(x)

DRAC car, dry pavement i,t

f(x)

truck, dry pavement

Microscopic Platform Microscopic Platform -

• VISSIM

VISSIM

8

Microscopic Platform Microscopic Platform -

• VISSIM

VISSIM

9

Linking CPI to Crash Occurrences Linking CPI to Crash Occurrences

• Need very detailed traffic information
• Test assertions:

– Test 1: CPI over time (5min prior the crash) – Test 2: CPIcrashx CPInon-crash (5min) – Test 3: CPI1hx Crash rate (aggregated)

10

Test 1 Test 1

11

Test 2 Test 2

12

Test 3 Test 3

13

Concluding Remarks Concluding Remarks

• Calibration/validation procedure for safety

measures yield best estimates of simulation inputs

• Tests for linking SP to observed crashes

support assertion crashes occur when CPI is high

• Microscopic approach proved to be useful in

providing an objective measure of safety in practical applications

14

Future Work Future Work

Better vehicle tracking data better models and more accurate analysis More research regarding MADR more inclusive model Need to consider different road-users Incorporation of safety into existing traffic simulation models

15 16

Acknowledgments:

The National Council for Scientific and Technological Development

Thank you!

Flávio Cunto flaviocunto@det.ufc.br

17