1

Safety Impact Methodology

Mikio Yanagisawa

RITA / Volpe

ITS Connected Vehicle Public Meeting

Arlington, VA

September 24-26, 2013

2

Safety Benefits – Basic Equations

Data Sources• # Target Crashes: Crash databases (GES, FARS,…)• Exposure Ratio: Field operational tests and controlled experiments• Crash Prevention Ratio: Field tests, objective tests, controlled experiments,

and computer simulation Safety Impact Methodology (SIM) tool

𝑬𝒙𝒑𝒐𝒔𝒖𝒓𝒆𝑹𝒂𝒕𝒊𝒐=𝐸𝑥𝑝𝑜𝑠𝑢𝑟𝑒𝑡𝑜𝐷𝑟𝑖𝑣𝑖𝑛𝑔𝐶𝑜𝑛𝑓𝑙𝑖𝑐𝑡𝑠𝒘𝒊𝒕𝒉𝑉 2𝑉

𝐸𝑥𝑝𝑜𝑠𝑢𝑟𝑒𝑡𝑜𝐷𝑟𝑖𝑣𝑖𝑛𝑔𝐶𝑜𝑛𝑓𝑙𝑖𝑐𝑡𝑠𝒘𝒊𝒕𝒉𝒐𝒖𝒕𝑉 2𝑉

𝑪𝒓𝒂𝒔𝒉𝑷𝒓𝒆𝒗𝒆𝒏𝒕𝒊𝒐𝒏 𝑹𝒂𝒕𝒊𝒐=h𝐶𝑟𝑎𝑠 𝑃𝑟𝑜𝑏𝑎𝑏𝑖𝑙𝑖𝑡𝑦 𝑖𝑛𝐷𝑟𝑖𝑣𝑖𝑛𝑔𝐶𝑜𝑛𝑓𝑙𝑖𝑐𝑡𝒘𝒊𝒕𝒉𝑉 2𝑉

h𝐶𝑟𝑎𝑠 𝑃𝑟𝑜𝑏𝑎𝑏𝑖𝑙𝑖𝑡𝑦 𝑖𝑛𝐷𝑟𝑖𝑣𝑖𝑛𝑔𝐶𝑜𝑛𝑓𝑙𝑖𝑐𝑡𝒘𝒊𝒕𝒉𝒐𝒖𝒕𝑉 2𝑉

𝐴𝑝𝑝𝑙𝑖𝑐𝑎𝑡𝑖𝑜𝑛𝐸𝑓𝑓𝑒𝑐𝑡𝑖𝑣𝑒𝑛𝑒𝑠𝑠=1−𝑬𝒙𝒑𝒐𝒔𝒖𝒓𝒆𝑹𝒂𝒕𝒊𝒐𝒏×𝑪𝒓𝒂𝒔𝒉𝑷𝒓𝒆𝒗𝒆𝒏𝒕𝒊𝒐𝒏 𝑹𝒂𝒕𝒊𝒐

¿ h𝐶𝑟𝑎𝑠 𝑒𝑠 𝐴𝑣𝑜𝑖𝑑𝑒𝑑=¿𝑇𝑎𝑟𝑔𝑒𝑡 h𝐶𝑟𝑎𝑠 𝑒𝑠× 𝐴𝑝𝑝𝑙𝑖𝑐𝑎𝑡𝑖𝑜𝑛𝐸𝑓𝑓𝑒𝑐𝑡𝑖𝑣𝑒𝑛𝑒𝑠𝑠

3

SIM Tool Overview

Crashes without V2V

Conflict exposure

with / without V2V

Harm from crashes

without V2V

Probability of crash

V distributions

Crashes prevented

Harm reduced

SIM Tool

Safety Impact Methodology

(SIM)

• SIM estimates effectiveness of individual scenarios

• SIM estimates are based on 100% system deployment

• SIM tool is developed for evaluation, not design

4

Rear-End Pre-Crash Simulation

Rear-End SIM

Simulation

DRIVER

Reaction Time

Braking Level

Control vs. Treatment

VEHICLE

Host Vehicle State

Remote Vehicle State

SYSTEM

Time-to-Collision of Warning

Automatic System Activation

Number of Crashes Number of Non-Crashes Impact Speed Distribution ΔV Distribution Host and Remote

Effectiveness

5

Rear-End Pre-Crash Walkthrough (LVS)

Control

Treatment

6

Probability of Crash EstimationExample: Lead Vehicle Stopped – Brake Response

{IF dstop ≤ dinitial THEN crash is prevented, ELSE crash occurs}

Run Monte Carlo simulations by drawing random numbers from tR & aH distributions and account for crashes occurred and prevented:

1. tR & aH from drivers without system assistance

2. tR & aH from drivers with system assistance

𝑑𝑖𝑛𝑖𝑡𝑖𝑎𝑙=𝑇𝑇𝐶𝑤𝑎𝑟𝑛×𝑣𝐻

𝑑𝑠𝑡𝑜𝑝=𝑣𝐻

2

2𝑎𝐻

+(𝑣𝐻×𝑡𝑅 )

Initial ConditionsvH = Host velocityTTCwarn = Time-to-collision of warning

Driver ResponsetR = Reaction time from warningaH = Host average deceleration

7

Lane Change Pre-Crash Simulation

Lane Change

SIM Simulation

DRIVER

Driver Reaction Time

Driver Counter Steer

Control vs. Treatment

VEHICLE

Host Vehicle State

Remote Vehicle State

SYSTEM

Initial Lateral Distance

Number of Crashes Number of Non-Crashes Impact Speed Distribution ΔV Distribution Host and Remote

Effectiveness

8

Lane Change Pre-Crash Walkthrough

Control

Treatment

9

Intersection Pre-Crash Simulation

Intersection SIM

Simulation

DRIVER

Driver Reaction Time

Driver Braking Level

Control vs. Treatment

VEHICLE

Host Vehicle State

Remote Vehicle State

Host Distance and Acceleration (Encroaching)

SYSTEM

Time-to-Intersection

Number of Crashes Number of Non-Crashes Impact Speed Distribution ΔV Distribution Host and Remote

Effectiveness

10

Straight Crossing Paths – Both Vehicles Moving

Control Treatment

11

Straight Crossing Paths – Stopped and Starting

Control Treatment

12

Input Parameters

Driver Reaction (Control) Link

?

Driver Reaction (Treatment)

Brake Force (Control)

Brake Force (Treatment)

Link?

Data Field Track Experiment Literature

Relationships Dependencies Flexible Variable inputs

*Example Mock DataLink

?

Link

?

13

SIM Results and Analysis

Pre-crash modeCrash CountImpact ModeTreatment TypeCrash Severity

SIM

pCrashCrash Ratio

Crash SeveritySpeed Reduction

Analysis

Results from SIM analyzed for effectiveness Applied to appropriate target population Pre-crash scenarios combined Effectiveness estimates lead to overall benefits