A computational model to assess the impact of policy measures on traffic safety in Flanders: Theoretical concepts and application.

Similar documents
Using Rates of Change to Create a Graphical Model. LEARN ABOUT the Math. Create a speed versus time graph for Steve s walk to work.

Capacity Utilization Metrics Revisited: Delay Weighting vs Demand Weighting. Mark Hansen Chieh-Yu Hsiao University of California, Berkeley 01/29/04

Paul M. Sommers David U. Cha And Daniel P. Glatt. March 2010 MIDDLEBURY COLLEGE ECONOMICS DISCUSSION PAPER NO

Morningstar Investor Return

ANALYSIS OF RELIABILITY, MAINTENANCE AND RISK BASED INSPECTION OF PRESSURE SAFETY VALVES

Time & Distance SAKSHI If an object travels the same distance (D) with two different speeds S 1 taking different times t 1

What the Puck? an exploration of Two-Dimensional collisions

AP Physics 1 Per. Unit 2 Homework. s av

Interpreting Sinusoidal Functions

A Probabilistic Approach to Worst Case Scenarios

KINEMATICS IN ONE DIMENSION

2017 MCM/ICM Merging Area Designing Model for A Highway Toll Plaza Summary Sheet

Monte Carlo simulation modelling of aircraft dispatch with known faults

The t-test. What We Will Cover in This Section. A Research Situation

Proportional Reasoning

An Alternative Mathematical Model for Oxygen Transfer Evaluation in Clean Water

The safe ships trajectory in a restricted area

SURFACE PAVEMENT CHARACTERISTICS AND ACCIDENT RATE

AMURE PUBLICATIONS. Working Papers Series

Examining the limitations for visual anglecar following models

Bill Turnblad, Community Development Director City of Stillwater Leif Garnass, PE, PTOE, Senior Associate Joe DeVore, Traffic Engineer

Evaluation of a car-following model using systems dynamics

3.00 m. 8. At La Ronde, the free-fall ride called the Orbit" causes a 60.0 kg person to accelerate at a rate of 9.81 m/s 2 down.

Automatic air-main charging and pressure control system for compressed air supplies

KEY CONCEPTS AND PROCESS SKILLS. 1. An allele is one of the two or more forms of a gene present in a population. MATERIALS AND ADVANCE PREPARATION

Homework 2. is unbiased if. Y is consistent if. c. in real life you typically get to sample many times.

8/31/11. the distance it travelled. The slope of the tangent to a curve in the position vs time graph for a particles motion gives:

Market Timing with GEYR in Emerging Stock Market: The Evidence from Stock Exchange of Thailand

COMPARING SIMULATED ROAD SAFETY PERFORMANCE TO OBSERVED CRASH FREQUENCY AT SIGNALIZED INTERSECTIONS

Urban public transport optimization by bus ways: a neural network-based methodology

Chapter : Linear Motion 1

Stock Return Expectations in the Credit Market

PRESSURE SENSOR TECHNICAL GUIDE INTRODUCTION FEATURES OF ELECTRIC PRESSURE SENSOR. Photoelectric. Sensor. Proximity Sensor. Inductive. Sensor.

Strategic Decision Making in Portfolio Management with Goal Programming Model

Revisiting the Growth of Hong Kong, Singapore, South Korea, and Taiwan, From the Perspective of a Neoclassical Model

FORECASTING TECHNIQUES ADE 2013 Prof Antoni Espasa TOPIC 1 PART 2 TRENDS AND ACCUMULATION OF KNOWLEDGE. SEASONALITY HANDOUT

Methods for Estimating Term Structure of Interest Rates

Transit Priority Strategies for Multiple Routes Under Headway-Based Operations

Name Class Date. Step 2: Rearrange the acceleration equation to solve for final speed. a v final v initial v. final v initial v.

Instruction Manual. Rugged PCB type. 1 Terminal Block. 2 Function. 3 Series Operation and Parallel Operation. 4 Assembling and Installation Method

Review of Economics & Finance Submitted on 27/03/2017 Article ID: Mackenzie D. Wood, and Jungho Baek

Zelio Control Measurement Relays RM4L Liquid Level Relays

EXAMINING THE FEASIBILITY OF PAIRED CLOSELY-SPACED PARALLEL APPROACHES

What is a Practical (ASTM C 618) SAI--Strength Activity Index for Fly Ashes that can be used to Proportion Concretes Containing Fly Ash?

Do Competitive Advantages Lead to Higher Future Rates of Return?

The Current Account as A Dynamic Portfolio Choice Problem

Simulation Validation Methods

Constructing Absolute Return Funds with ETFs: A Dynamic Risk-Budgeting Approach. July 2008

Application of System Dynamics in Car-following Models

A Liability Tracking Portfolio for Pension Fund Management

Reliability Design Technology for Power Semiconductor Modules

Keywords: overfishing, voluntary vessel buy back programs, backward bending supply curve, offshore fisheries in Taiwan

Overview. Do white-tailed tailed and mule deer compete? Ecological Definitions (Birch 1957): Mule and white-tailed tailed deer potentially compete.

Real-time Stochastic Evacuation Models for Decision Support in Actual Emergencies

3. The amount to which $1,000 will grow in 5 years at a 6 percent annual interest rate compounded annually is

Idiosyncratic Volatility, Stock Returns and Economy Conditions: The Role of Idiosyncratic Volatility in the Australian Stock Market

San Francisco State University ECON 560 Fall Midterm Exam 2. Tuesday, October hour, 15 minutes

SPECIAL WIRE ROPES The Value Line

INSTRUCTIONS FOR USE. This file can only be used to produce a handout master:

Economics 487. Homework #4 Solution Key Portfolio Calculations and the Markowitz Algorithm

Performance Attribution for Equity Portfolios

A Study on the Powering Performance of Multi-Axes Propulsion Ships with Wing Pods

SIMULATION OF WAVE EFFECT ON SHIP HYDRODYNAMICS BY RANSE

DYNAMIC portfolio optimization is one of the important

LSU RISK ASSESSMENT FORM Please read How to Complete a Risk Assessment before completion

FHWA/IN/JTRP-2009/12. Panagiotis Ch. Anastasopoulos Fred L. Mannering John E. Haddock

CMA DiRECtions for ADMinistRAtion GRADE 6. California Modified Assessment. test Examiner and Proctor Responsibilities

TRACK PROCEDURES 2016 RACE DAY

Simulation based approach for measuring concentration risk

Avoiding Component Failure in Industrial Refrigeration Systems

CALCULATION OF EXPECTED SLIDING DISTANCE OF BREAKWATER CAISSON CONSIDERING VARIABILITY IN WAVE DIRECTION

Protecting the African Elephant: A Dynamic Bioeconomic Model of. Ivory Trade

Prepared by: Candice A. Churchwell, Senior Consultant Aimee C. Savage, Project Analyst. June 17, 2014 CALMAC ID SCE0350

Rolling ADF Tests: Detecting Rational Bubbles in Greater China Stock Markets

Corresponding Author

The Measuring System for Estimation of Power of Wind Flow Generated by Train Movement and Its Experimental Testing

Lifecycle Funds. T. Rowe Price Target Retirement Fund. Lifecycle Asset Allocation

Flexible Seasonal Closures in the Northern Prawn Fishery

Semi-Fixed-Priority Scheduling: New Priority Assignment Policy for Practical Imprecise Computation

LEWA intellidrive. The mechatronic All-in-One pump system. intelligent flexible dynamic high precision. Foto: ratiopharm

Dynamics of market correlations: Taxonomy and portfolio analysis

MODEL SELECTION FOR VALUE-AT-RISK: UNIVARIATE AND MULTIVARIATE APPROACHES SANG JIN LEE

WHO RIDE THE HIGH SPEED RAIL IN THE UNITED STATES THE ACELA EXPRESS CASE STUDY

ScienceDirect. Cycling Power Optimization System Using Link Models of Lower Limbs with Cleat-Shaped Biaxial Load Cells

The design of courier transportation networks with a nonlinear zero-one programming model

Reproducing laboratory-scale rip currents on a barred beach by a Boussinesq wave model

Time-Variation in Diversification Benefits of Commodity, REITs, and TIPS 1

3 (R) 1 (P) N/en

Evaluating the Performance of Forecasting Models for Portfolio Allocation Purposes with Generalized GRACH Method

XSz 8... XSz 50 Solenoid actuated fail-safe safety valve

Guidance Statement on Calculation Methodology

QUANTITATIVE FINANCE RESEARCH CENTRE. Optimal Time Series Momentum QUANTITATIVE FINANCE RESEARCH CENTRE QUANTITATIVE F INANCE RESEARCH CENTRE

Development of Urban Public Transit Network Structure Integrating Multi-Class Public Transit Lines and Transfer Hubs

Flow Switch LABO-VHZ-S

CALCULATORS: Casio: ClassPad 300 ClassPad 300 Plus ClassPad Manager TI: TI-89, TI-89 Titanium Voyage 200. The Casio ClassPad 300

Asset Allocation with Higher Order Moments and Factor Models

Dual Boost High Performances Power Factor Correction (PFC)

2. JOMON WARE ROPE STYLES

Proceedings of the ASME 28th International Conference on Ocean, Offshore and Arctic Engineering OMAE2009 May 31 - June 5, 2009, Honolulu, Hawaii

Machine Learning for Stock Selection

Transcription:

A compuaional model o assess he impac of policy measures on raffic safey in Flanders: Theoreical conceps and applicaion. RA-MOW-2009-006 B.B. Nambuusi, E. Hermans, T. Brijs Onderzoekslijn Beleidsorganisaie en monioring DIEPENBEEK, 2009. STEUNPUNT MOBILITEIT & OPENBARE WERKEN SPOOR VERKEERSVEILIGHEID

Documenbeschrijving Rappornummer: Tiel: RA-MOW-2009-006 A compuaional model o assess he impac of policy measures on raffic safey in Flanders: Theoreical conceps and applicaion. Aueur(s): B.B. Nambuusi, E. Hermans, T. Brijs Promoor: Prof. dr. Geer Wes Onderzoekslijn: Beleidsorganisaie en -monioring Parner: Universiei Hassel Aanal pagina s: 40 Projecnummer Seunpun: 7.3 Projecinhoud: Onwikkeling van een rekenmodel voor de verkeersveiligheidseffecen van maaregelen in Vlaanderen. Uigave: Seunpun Mobiliei & Openbare Werken, november 2009. Seunpun Mobiliei & Openbare Werken Weenschapspark 5 B 3590 Diepenbeek T 011 26 91 12 F 011 26 91 99 E info@seunpunmowverkeersveiligheid.be I www.seunpunmowverkeersveiligheid.be

Samenvaing Een rekenmodel voor he bepalen van de impac van beleidsmaaregelen op verkeersveiligheid in Vlaanderen: Theoreische concepen en oepassing In di rappor beschrijven we de onwikkeling van een rekenmodel da he effec van beleidsmaaregelen op verkeersveiligheid op he regionale niveau bepaal en de kosen en baen hiervan vergelijk om aldus een goede selecie van maaregelen e maken. De belangrijkse fasen in he model zijn de refereniesiuaie, de baseline prognose, de maaregel prognose, de besparingen en de kosen-baenanalyse. De refereniesiuaie beschrijf de huidige verkeer- en veiligheidpresaie. In de baseline prognose word de oekomsige verandering in verkeerpresaie (blooselling) en in he auonome risico in rekening gebrach. De maaregel prognose fase beschrijf de siuaie nada maaregelen doorgevoerd zijn. He verschil ussen de baseline en de maaregel prognose bepaal dan he aanal ongevallen of slachoffers die bespaard werden door de invoering van een bepaalde (se van) maaregel(en). In de kosen-baenanalyse o slo word nagegaan of de baen van een maaregel de kos ervan oversijgen. Deze analyse help om maaregelen me een verschillende levensduur e vergelijken en een keuze e maken ussen mogelijke maaregelses. He model heef berekking op verschillende weg- en kruispuncaegorieën in de regio, maar de illusraie in di rappor gebeur voor één bepaalde wegcaegorie (namelijk auosnelwegen). He model focus op onwikkelingen in de oekoms waarbij he effec van maaregelen oegepas op de gehele regio of op enkele locaies word gekwanificeerd. Er word een onderscheid gemaak ussen regionale en lokale maaregelen waarbij regionale maaregelen worden geach een invloed e hebben op he niveau van verkeersveiligheid in de hele regio en lokale maaregelen enkel op de locaie(s) waarop ze werden oegepas. Bij de onwikkeling van he model werden verschillende ses van maaregelen gees. Daarnaas werd rekening gehouden me een aanal belangrijke onwikkelingen waarop de eindgebruiker van he model geen invloed heef, zoals de groei in blooselling en de verandering in he auonome risico. De verandering in he auonome risico is e wijen aan he collecieve leerproces gesuurd door de oegenomen kennis van he verkeersveiligheidsprobleem, de voordurende verbeering van de presaie van he ransporsyseem, beer uigeruse voeruigen en wegen, een verbeerde educaie, en inspanningen op vlak van wegeving en handhaving. Toekomsige veranderingen in blooselling worden verva in groeiscenario s. In he rappor worden wee groeiscenario s (GS1 en GS2) besudeerd voor de periode 2005-2030. Daarenboven word de mehodologie geïllusreerd aan de hand van een gevalsudie uigewerk op auosnelwegen in Vlaanderen voor een periode van vier jaar (2003-2006). He aanal bespaarde slachoffers gedurende deze periode word bepaald rekening houdende me de algemene onwikkelingen (de groei in blooselling en de verandering in he auonome risico). De kosen-baenanalyse word enkel heoreisch behandeld en nie geïllusreerd in di rappor omda he doel van deze eerse analyse he bepalen van de effeciviei van maaregelen in ermen van he aanal bespaarde slachoffers is. He model da onwikkeld word, bied de gebruiker inzich in de impac van verschillende regionale en lokale maaregelen oegepas op een bepaald ijdsip. Di rappor beschrijf de mehodologie van he rekenmodel en de eerse illusraieve resulaen. In de oekoms zal he model verbeerd worden door onder andere de afhankelijkheid ussen maaregelen e beschouwen. Ook zal een sensiivieianalyse uigevoerd worden en zullen de mees onveilige plaasen in de regio gevisualiseerd worden door middel van he linken me een geografisch informaiesyseem. Seunpun Mobiliei en Openbare Werken 3 RA-MOW-2009-006

English summary The aim of his repor is o describe he developmen of a compuaional model o assess he effec of policy measures on safey a he regional level and compare heir coss and benefis in order o selec measures wih he mos beneficial cos-benefi raios. The main sages of he model include he reference siuaion, he baseline prognosis, he measure prognosis, he savings and he cos-benefi analysis. The reference siuaion describes he curren raffic performance and safey siuaion. The baseline prognosis akes he fuure change in raffic performance and auonomous risk ino accoun. The measure prognosis relaes o he siuaion afer applying measures. The number of savings is he difference beween he baseline and he measure prognosis. The cosbenefi analysis is used o deermine wheher he benefis of a measure ouweigh is coss. This analysis helps o compare measures wih a differen life span and o make a choice beween possible combinaions of measure ses. The model incorporaes several road and inersecion caegories in he region ye he illusraion in his repor relaes o only one road caegory (being highways). The model focuses on developmens in he fuure, hereby quanifying he effec of measures applied eiher a a regional or a locaional level. Regional measures are assumed o have an impac on road safey in he enire region while locaional measures have an impac on road safey a he locaion hey are applied. During he developmen of he model, several ses of measures were esed. Besides, a number of imporan developmens on which he user of he model has no influence such as he change in raffic performance and in auonomous risk are aken ino accoun. The auonomous risk change capures he collecive learning process caused by he growing knowledge of he road safey problem, he consan improvemen of he safey performance of he road ranspor sysem, beer equipped moor vehicles and roads, he improvemen of road safey educaion and, increasing legislaion and enforcemen. Fuure changes in raffic performance are represened by growh scenarios. Here, wo growh scenarios (GS1 and GS2) are sudied for he period 2005-2030. Also, a case sudy is carried ou for a period of four years (2003 2006) on highways in Flanders o illusrae he mehodology. The safey siuaion in erms of he number of saved casualies during he four years is assessed aking ino accoun general developmens i.e growh in raffic performance and auonomous risk. The cos-benefi analysis is only discussed in heory bu no illusraed in his repor since he aim of he firs analysis concerns assessing he effeciveness of measures in erms of he number of saved casualies. The developed model allows he user o ge insigh ino he impac of differen regional and locaional measures applied a cerain momens in ime. This repor describes he mehodology of he compuaional model and some firs illusraive resuls. In he fuure, he model will be improved o accoun for dependency among measures. Also, a sensiiviy analysis will be carried and he mos unsafe areas in he region will be displayed by linking i o a geographical informaion sysem. Seunpun Mobiliei en Openbare Werken 4 RA-MOW-2009-006

Table of conens 1. INTRODUCTION... 7 1.1 Objecive of he repor 8 1.2 Organizaion of he repor 8 2. DATA... 9 2.1 The curren siuaion: he road nework, road safey and raffic performance 9 2.2 Changes in he fuure: raffic performance and auonomous risk 10 2.3 Effeciveness of road safey measures from lieraure 10 2.3.1 Measures in Flanders...10 2.3.2 Measures a he inernaional level...11 2.4 Coss and benefis 11 3. METHODOLOGY... 12 3.1 The reference year 12 3.2 Baseline prognosis 13 3.2.1 Baseline prognosis for raffic performance a he regional level...13 3.2.2 Baseline prognosis for road safey quaniies a he regional level...17 3.2.3 Baseline prognosis for raffic performance a he locaional level...19 3.2.4 Baseline prognosis for road safey quaniies a he locaional level...19 3.3 Measure prognosis 20 3.3.1 Compuing effeciveness of measures...20 3.3.2 Dependence among measures...21 3.4 Savings 21 3.5 Cos-benefi analysis 21 3.5.1 Benefis...22 3.5.2 Coss...22 4. RESULTS... 25 4.1 The reference year 25 4.2 Baseline prognosis 27 4.2.1 Baseline prognosis for raffic performance...27 4.2.2 Baseline risk for injury accidens...28 4.2.3 Baseline prognosis for road safey quaniies...29 4.2.4 Growh scenarios GS1 and GS2 (2005 2030)...30 4.3 Effeciveness of measures (2003) 33 4.3.1 Applying a regional and a locaional measure simulaneously...33 4.3.2 Applying a locaional afer a regional measure...34 4.3.3 Applying a regional afer a locaional measure...34 4.3.4 Conclusion...35 Seunpun Mobiliei en Openbare Werken 5 RA-MOW-2009-006

4.4 Case sudy (2003-2006) 35 5. CONCLUSION AND FURTHER RESEARCH... 37 6. REFERENCES... 39 Seunpun Mobiliei en Openbare Werken 6 RA-MOW-2009-006

1. I N T R O D U C T I O N A his momen, Flanders does no dispose of a compuaional model o calculae he effecs of measures regarding road safey a a regional level. In he pas, he Policy Research Cenre Traffic Safey has already execued a number of sudies (e.g. De Brabander e al., 2005; Nuys, 2004) abou he effecs of measures such as roundabous and he insallaion of speed cameras a a locaional level (he level of road segmens and inersecions). However, an approach invesigaing he effec of policy measures on a broader area and in a broader conex of susainable developmen is sill lacking in Flanders. In his respec, an esimaion mehod which assiss regions o calculae he road safey effecs of boh regional and locaional measures and o aid in selecing measures resuling in he mos efficien cos-benefi raios is a valuable ool. The regional road safey explorer (RRSE) model developed by he SWOV (Reurings and Wijnen, 2008) is used as a saring poin for Flanders. In he road nework of Flanders, here are various road and inersecion caegories. The model o be developed will hus caer for he enire road nework alhough a presen he mehodology is only illusraed for he road caegory of highways. However, in he fuure more road caegories such as secondary roads and various inersecion caegories are o be inegraed. The model consiss of he following main sages: he reference siuaion, he baseline prognosis, he measure prognosis, he number of saved casualies and he cos-benefi analysis. To esimae he effeciveness of measures and heir cos-benefi raio, daa are required. These include daa in he reference period, daa in he baseline prognosis, daa in he measure prognosis and cos-benefi daa. The daa in he reference period describe he curren raffic performance and road safey siuaion per road and inersecion caegory. Generally, raffic consiss of moorized vehicles and oher road users such as pedesrians and cycliss who play an imporan role in road safey. However, due o daa scarciy and he fac ha he firs applicaion focuses on highways, hese vehicle kilomeres provide useful raffic performance informaion. In he curren model, raffic performance is defined as he number of moorized vehicle kilomeres on a paricular road caegory and he number of moorized vehicles passing hrough an inersecion caegory. The road safey siuaion is refleced by he number of injury accidens, he number of sligh casualies, he number of serious casualies and he number of faaliies per road or inersecion caegory. The baseline prognosis akes he change in raffic performance and auonomous risk ino accoun and acs as a reference o which he effeciveness of measures is compared. The auonomous risk change akes ino accoun he collecive learning process caused by he growing knowledge of he raffic safey problem, he consan improvemen of he safey performance of he road ranspor sysem, beer equipped moor vehicles and roads, improvemen of raffic safey educaion and, increasing legislaion and enforcemen. The measure prognosis relaes o he siuaion afer applying and esimaing he effeciveness of measures on road safey. The main oupus of he model are he number of saved casualies and he cos-benefi raio of he applied measures. The number of saved casualies is he difference beween he number of casualies in he baseline prognosis and in he measure prognosis. The saved casualies can be expressed in moneary values represening he benefis. The cos-benefi analysis is used o deermine wheher he benefis of a measure ouweigh is cos. This analysis helps o compare measures wih a differen life span and o make a choice beween possible combinaions of measure ses aiming o improve he level of road safey in he region. Seunpun Mobiliei en Openbare Werken 7 RA-MOW-2009-006

1.1 Objecive of he repor This research aims a developing a model for Flanders o assess he effecs of measures on road safey a he regional level and help in fuure decision making. The model enables comparing he coss and benefis of differen measure ses so ha he bes ses can be seleced and applied. This repor focuses on he assessmen of he effeciveness of measures. In he following secions, he mehodology of he compuaional model is described and a firs illusraion shown o indicae he value of he model. In addiion, he advanages and challenges encounered during he developmen of he model are discussed and aspecs for furher improvemen of he model highlighed. 1.2 Organizaion of he repor In Secion 2 he daa are described and he mehodology is explained in Secion 3. Resuls from applying he mehodology o a se of highway segmens in Flanders are presened in Secion 4 while he Conclusion and furher research are he subjecs of Secions 5. Seunpun Mobiliei en Openbare Werken 8 RA-MOW-2009-006

2. D A T A In his secion, he daa required for he compuaional model are described. For each daa caegory, an overview of he required daa is firs given afer which he acual daa used o illusrae he model are described. Firs, daa describing he curren siuaion of he road nework, road safey and raffic performance are presened, followed by daa reflecing changes in raffic performance and auonomous risk over ime. Thirdly, scienific resuls on he effeciveness of measures found in lieraure are given. Finally, daa required o carry ou a cos-benefi analysis described. 2.1 The curren siuaion: he road nework, road safey and raffic performance The road nework is caegorized ino various caegories of road segmens and inersecions. The road segmen caegories include highways, primary roads, secondary roads and residenial roads. These are furher classified ino rural and urban roads. The inersecion caegories comprise hree-arm, four-arm, signalized and unsignalized inersecions which are also grouped ino rural and urban inersecions. The road segmens or inersecions in each caegory consis of specific characerisics. Examples of characerisics of road segmens include: speed limi, curve radius, shoulder widh, lane widh, number of lanes, median, ec. Among he characerisics of inersecions are sigh disance from sop line, inersecion angle, number of approach roads, approach speed, number of lanes, lane widh and raffic conrol for pedesrians and cycliss (Vog and Bared, 1998b). For he momen, he model focuses on one road caegory (highways). In he fuure, oher road caegories as well as differen inersecion caegories will be inegraed in he model. The road safey daa describe he road safey siuaion in erms of he number of injury accidens, sligh casualies, serious casualies and faaliies per road segmen and inersecion in he reference year. I should be noed ha he number of accidens and casualies is in realiy higher han shown in official saisics because no all accidens are repored and regisered by he auhoriies (Elvik and Vaa, 2004). To beer reflec realiy, underreporing facors of he various levels of severiy are used. These are facors by which a regisered road safey quaniy is muliplied in order o obain a beer approximaion of he road safey quaniies. Hence, daa on underreporing facors are required for he number of injury accidens, sligh casualies, serious casualies and faaliies per road segmen and inersecion caegory. Underreporing facors are available in lieraure bu no broken down o road and inersecion caegory. Hence, 1.75, 1.90, 1.30 and 1.05 (Elvik and Mysen, 1999) for injury accidens, sligh casualies, serious casualies and faaliies respecively are uilized. The mehodology is illusraed using 2002 daa of 99 highway segmens in Flanders. The raffic performance or he number of kilomeres driven in he enire region on highways is obained by summing up he number of kilomeres on all road segmens. The number of kilomeres on a road segmen is calculaed by muliplying he lengh of a segmen and he number of vehicles passing by ha segmen (beween 6:00am and 10:00pm) and he oal on all segmens serves as he regional raffic performance. The number of kilomeres and vehicles are expressed in housands. Table 1 shows an example of raffic performance on four road segmens (H1, H11, H12 and H13). The figures in he las column would describe he raffic performance in he enire region if all he 99 segmens were presened insead of he four. Seunpun Mobiliei en Openbare Werken 9 RA-MOW-2009-006

Table 1: Traffic performance on H1, H11, H12 and H13 in 2002 Traffic performance (TP) Road H1 Road H11 Road H12 Road H13 Toal of four segmens Lengh km (L) 3.2000 13.0000 4.6000 1.9000 L1 (in 1000s) = L/1000 0.0032 0.0130 0.0046 0.0019 Number of vehicles (V) 14,700 81,600 86,200 97,400 V1 (in 1000s) = V/1000 14.7000 81.6000 86.2 97.4000 TP (in 1000s km) = L1*V1 0.0470 1.0608 0.3965 0.1851 1.6894 2.2 Changes in he fuure: raffic performance and auonomous risk These are wo aspecs which change over ime and affec he level of road safey; hey are aken ino accoun in he model. The daa used o compue he growh in raffic performance on highways in Flanders relae o he period 1985 2006 and are obained from he FOD MV (2008) and he Federal Plan Bureau (Federaal Planbureau, 2008). The auonomous risk refers o resuls from measures aken in he pas bu sill having an impac on he curren road safey siuaion. Furher, COST329, 2004, explains he auonomous risk as he collecive learning process caused by he growing knowledge of he raffic safey problem, he consan improvemen of he safey performance of he road ranspor sysem, beer equipped moor vehicles and roads, improvemen of raffic safey educaion and, increasing legislaion and enforcemen. The effec of he auonomous risk will be quanified using ime series daa of he oal number of casualies. 2.3 Effeciveness of road safey measures from lieraure To calculae he impac of a se of measures on road safey, he effeciveness is required wih respec o injury accidens, sligh casualies, serious casualies and faaliies. Below, some measures esed in Flanders and a he inernaional level are lised. In brackes are he severiy levels on which he impac of he measure has been assessed for Flanders. However, here are severiy levels for which he effeciveness of measures are no invesigaed in lieraure and herefore need o be approximaed. 2.3.1 Measures in Flanders Roundabous (IA, SLC & SCs) - De Brabander e al., (2005) Invisible and visible speed conrols on roads wih speed limis from 80 120 km/h (IA) - Van Geir (2006) Speed limi of 30 km/hr in school zones (IA) Dreesen and Princen (2005) Fully proeced lef urn signals (IA & FATs) Dreesen and Nuys (2006) Parly proeced lef urn signals (FATs) - Dreesen and Nuys (2006) Proeced lef urn on an inersecion wih a permissive lef urn (IA) - Dreesen, (2005) Auomaic speed cameras (FATs) - Nuys, (2004) IAs=Injury accidens, SLC=Sligh casualies, SCs=Serious casualies, FATs=Faaliies Seunpun Mobiliei en Openbare Werken 10 RA-MOW-2009-006

2.3.2 Measures a he inernaional level A he inernaional level, he handbook of road safey measures, (Elvik and Vaa, 2004) provides useful informaion. Below is a lis of caegories of measures described in he handbook and he number of measures belonging o he caegories. Road design and road furniure (19 measures) Traffic conrol (17 measures) Driver raining and regulaion of professional drivers (19 measures) Vehicle design and proecive devices (14 measures) Road mainenance (4 measures) Police enforcemen and sancions (9 measures) Public educaion and informaion (3 measures) In his repor, hree relevan measures for highways are sudied: speed reducion (130 o 110, 130 o 120 and 120 o 110km/hr) on highways, auomaic warning of queues wih variable signs and saionary speed enforcemen. The effeciveness of hese measures as given in Elvik and Vaa (2004) is used in he model. Speed reducion and auomaic warning of queues wih variable signs reduce he number of injury accidens by -0.14 (-0.20,-0.07) and -0.14 (-0.22,-0.08) respecively. Saionary speed enforcemen reduces he number of injury accidens by -0.06 (-0.09,-0.04). The effeciveness of hese measures on he oher severiy levels is no provided in lieraure and an approximaion is made. 2.4 Coss and benefis In order o calculae he oal annual cos of measures, he cos of he applied measures mus be known. These coss include he cos per measure per kilomere per year for road segmens and he cos per measure per piece per year for inersecions. The cos informaion will be obained by consuling expers. In addiion, he value in euros of an injury acciden, a sligh casualy, a serious casualy and a faaliy are required o calculae he annual benefis of he applied measures in erms of he casualies prevened. Excep for sligh casualies, 20,943, 725,512 and 2,004,799 euros are he values of an injury acciden, a serious casualy and a faaliy respecively (De Brabander and Vereeck, 2007). The value of a sligh casualy is o be esimaed from oher inernaional sources. Finally, a discoun rae is required o express he coss and benefis in erms of he nominal value of he reference year and a discoun rae of 4% (Transumo, 2008) will be uilized. The cos-benefi analysis is no illusraed in his repor since he aim of he firs analysis concerns assessing he effeciveness of measures in erms of he number of casualies saved. Seunpun Mobiliei en Openbare Werken 11 RA-MOW-2009-006

3. M E T H ODO L O G Y This secion describes he srucure of he model and how i calculaes he effecs of measures on road safey and cos-benefi raios. The formulae used are obained from Reurings and Wijnen, (2008). Figure 1 presens he various sages of he model afer which each sage is explained in deail. Secion 3.1 elaboraes on he reference year. Secion 3.2 discusses he baseline prognosis focussing on he change in raffic performance and auonomous risk. Nex, secions 3.3 and 3.4 respecively presen he measure prognosis and he savings. Finally, secion 3.5 gives an overview of he cosbenefi analysis. Reference year Baseline prognosis Measure prognosis Cos-benefi analysis Savings Figure 1: Sages of he model 3.1 The reference year The firs sage describes he raffic and road safey siuaion in he reference year. The raffic siuaion consiss of raffic performance i.e. he annual number of kilomeres ravelled by moorized vehicles per road caegory and he annual number of moorized vehicles hrough an inersecion caegory. The road safey siuaion comprises he number of injury accidens, sligh casualies, serious casualies and faaliies per road segmen and inersecion caegory. Underreporing of road safey quaniies is aken ino accoun by muliplying he regisered quaniies by an underreporing facor per severiy level. Underreporing is lower for some severiy levels (Reurings e al., 2007). Moreover, differen caegories of road segmens and inersecions migh have differen underreporing levels. In his repor, i is assumed ha all road segmens and inersecions have he same underreporing facor as he road or inersecion caegory o which hey belong. The compuaion is illusraed for injury accidens (IA s ). IAs c IAs c, regisered * fc, acciden wih IAs c = incremened number of injury accidens on caegory c IAs c, regisered = regisered number of injury accidens on caegory c fc, acciden = underreporing facor for he number of injury accidens on caegory c The oher severiy levels are incremened in a similar manner. Apar from he number of injury accidens, sligh casualies, serious casualies and faaliies, he road safey siuaion in he reference year is refleced by various indicaors. Seunpun Mobiliei en Openbare Werken 12 RA-MOW-2009-006

The mos imporan ones are: he acciden risk r ), he number of casualies per injury acciden ) ( 0, c N, he number of sligh casualies per oal casualies ( c N,c 1, he number of serious casualies per oal casualies N ) and he number of faaliies per 100 ( 2, c casualies N ) per road segmen or inersecion caegory. The acciden indicaors are ( 3, c compued as follows: IAs c rc, TPc N C c 0,. c, IAs c N SLC c 1, c, Cc N SCs FATs N wih c c 2, c and 3, c * 100 Cc Cc IAs c = incremened number of injury accidens on all infrasrucures belonging o road segmen or inersecion caegory c in he reference year TP c C c = raffic performance on all infrasrucures belonging o road segmen or inersecion caegory c in he reference year = number of casualies on all infrasrucures belonging o road segmen or inersecion caegory c in he reference year SLC c = number of sligh casualies on all infrasrucures belonging o road segmen or inersecion caegory c in he reference year SCs c = number of serious casualies on all infrasrucures belonging o road segmen or inersecion caegory c in he reference year FATs c = number of faaliies on all infrasrucures belonging o road segmen or inersecion caegory c in he reference year 3.2 Baseline prognosis Having described he reference siuaion, he baseline prognoses can be calculaed. The baseline prognosis conains he number of injury accidens, sligh casualies, serious casualies and faaliies before he impac of measures is aken ino accoun. To deermine he baseline prognosis, wo developmens on which he user of he model has no influence are aken ino accoun. They include he change in raffic performance and in auonomous risk. A his sage, he baseline prognosis concerning raffic performance and road safey quaniies is calculaed. We disinguish beween baseline prognoses a he regional level and a he locaional level. 3.2.1 Baseline prognosis for raffic performance a he regional level Traffic performance increases each year and his resuls ino uncerainiy abou he fuure. Thus, differen scenarios represening specific growh raes in raffic performance are considered. Every growh scenario consiss of a se of raffic performance growh facors relaing o a paricular year and infrasrucure caegory. The baseline raffic performance for infrasrucure caegory c in year due o growh scenario A (TP A, c, ) is given by: Seunpun Mobiliei en Openbare Werken 13 RA-MOW-2009-006

TP * A, c, g A, c,1 * g A, c,2 *...* g A, c, TPc wih g A c, TP c, = growh facor for raffic performance in year on road segmen or inersecion caegory c due o growh scenario A = raffic performance on road segmen or inersecion caegory c in he reference year Flemish daa from 1985-2006 (FOD MV, 2008), are used o predic he yearly growh in vehicle kilomeres for he period 2007-2030. A long period is seleced because he model is developed wih a long ime frame o predic fuure values unil 2030. The non-linear logisic growh curve model proposed by Oppe (1989) is uilized for he predicion. This curve is characerized by a slow growh a he sar, rapid growh wih ime and limiing growh afer some ime. The seleced model is of he form below as esimaed by CurveExper 1.3 sofware (Daniel Hyams). y i i *exp * year 1 1 2 0 i wih y i 0 1 2 i = prediced number of vehicle kilomeres driven in year i = limiing value of vehicle kilomeres driven = number of imes he iniial number of driven vehicle kilomeres (year i=0) mus grow o reach he limiing value = rae of growh in he number of driven vehicle kilomeres as he curve approaches he limiing value = error erm The correlaion coefficien and he sandard error esimae are used o assess he goodness of fi of he model. The model wih he larges magniude of he correlaion coefficien and he smalles sandard error esimae is considered bes. In his case he magniude of he correlaion coefficien and he sandard error esimae of he seleced model are 0.9525 and 686.7576 respecively. The model is hen fied in SAS 9.1 and Table 2 conains he resuls obained from he non-linear logisic growh curve model. The parameer esimaes obained are used o predic he number of vehicle kilomeres for he period 1985-2030 (Table 3). Table 2: Parameer esimaes (sandard errors) and Confidence inervals Parameer Esimae (sandard errors) Confidence inervals 23487 (960.4900) (21495, 25479) 0 1.4676 (0.0753) (1.3115, 1.6238) 1 0.1123 (0.0117) (0.0881, 0.1366) 2 Var ( i ) 441.1100 (66.4987) (303.2000, 579.0200) The comparison of he observed and he prediced daa is presened in Figure 2. I can be seen ha he evoluion in he prediced and he observed kilomeres is close. In oher words, mos of he variabiliy in he daa is capured by he predicion alhough here appears lile aleraion a some poins. Seunpun Mobiliei en Openbare Werken 14 RA-MOW-2009-006

1985 1989 1993 1997 2001 2005 2009 2013 2017 2021 2025 2029 Vehicle kilomeres (in millions) 25000 20000 15000 10000 Observed vehicle kilomeres (in millions) Prediced vehicle kilomeres (in millions) 5000 0 Year Figure 2: Comparison of observed and prediced raffic performance daa Here, we consider wo opions for predicing he growh in he number of vehicle kilomeres. The firs opion is o use he parameer esimaes in Table 2 o capure he growh in percenage per year. These growh facors represen he firs growh scenario (GS1) (Table 3). The second opion sars from he prediced growh in he number of vehicle kilomeres in Belgium beween 2005 and 2030 as deermined by he Federal Plan Bureau (FPB) (Federaal Planbureau). Yearly growh facors for he second growh scenario (GS2) are compued by spreading he prediced growh in vehicle kilomeres beween 2005 and 2030 by he FPB over differen years using he same rae as deermined by he logisic growh model raher han a more unrealisic consan rae. The formula used o compue he second se of growh facors is given below. Growh rae in prediced km in year i * Overallgrowh in prediced km FPB (2005-2030) Overallgrowh in prediced km (2005-2030) wih i = 2005, 2006,, 2030 The overall growh in prediced km (2005-2030) is 12.93% obained as he percenage difference beween he prediced kilomeres for he period 2005 2030 (Table 3) while he overall growh in prediced km FPB (2005-2030) is 22%. Table 3 conains he wo ses of growh facors represening GS1 and GS2. Table 3: Prediced growh in vehicle kilomeres 1985-2030 Year Observed km (in millions) Growh facors in he observed km Prediced km (in millions) Growh in prediced km (in millions) Growh facors GS1 Growh facors GS2 1985 9,630 10,160.0300 1986 10,320 1.0717 10,811.6879 651.6579 1.0641 1987 11,240 1.0891 11,469.1724 657.4845 1.0608 1988 12,690 1.1290 12,128.3821 659.2097 1.0575 1989 14,260 1.1237 12,785.1725 656.7905 1.0542 1990 13,600 0.9537 13,435.4598 650.2872 1.0509 1991 14,100 1.0368 14,075.3208 639.8611 1.0476 1992 14,480 1.0270 14,701.0859 625.7650 1.0445 1993 14,950 1.0325 15,309.4175 608.3316 1.0414 Seunpun Mobiliei en Openbare Werken 15 RA-MOW-2009-006

1994 15,790 1.0562 15,897.3738 587.9563 1.0384 1995 16,380 1.0374 16,462.4535 565.0797 1.0355 1996 16,750 1.0226 17,002.6216 540.1681 1.0328 1997 16,860 1.0066 17,516.3173 513.6957 1.0302 1998 17,930 1.0635 18,002.4449 486.1276 1.0278 1999 18,850 1.0513 18,460.3506 457.9058 1.0254 2000 19,290 1.0233 18,889.7886 429.4380 1.0233 2001 19,360 1.0036 19,290.8780 401.0894 1.0212 2002 19,680 1.0165 19,664.0560 373.1779 1.0193 2003 19,800 1.0061 20,010.0275 345.9715 1.0176 2004 20,270 1.0237 20,329.7163 319.6887 1.0160 2005 20,460 1.0094 20,624.2166 294.5004 1.0145 1.0246 2006 21,210 1.0367 20,894.7497 270.5331 1.0131 1.0223 2007 21,142.6235 247.8738 1.0119 1.0202 2008 21,369.1976 226.5741 1.0107 1.0182 2009 21,575.8537 206.6561 1.0097 1.0165 2010 21,763.9702 188.1165 1.0087 1.0148 2011 21,934.9021 170.9320 1.0079 1.0134 2012 22,089.9654 155.0633 1.0071 1.0120 2013 22,230.4244 140.4590 1.0064 1.0108 2014 22,357.4837 127.0593 1.0057 1.0097 2015 22,472.2817 114.7980 1.0051 1.0087 2016 22,575.8877 103.6060 1.0046 1.0078 2017 22,669.3002 93.4124 1.0041 1.0070 2018 22,753.4467 84.1465 1.0037 1.0063 2019 22,829.1852 75.7385 1.0033 1.0057 2020 22,897.3065 68.1213 1.0030 1.0051 2021 22,958.5366 61.2302 1.0027 1.0045 2022 23,013.5405 55.0039 1.0024 1.0041 2023 23,062.9251 49.3847 1.0021 1.0037 2024 23,107.2437 44.3185 1.0019 1.0033 2025 23,146.9989 39.7552 1.0017 1.0029 2026 23,182.6472 35.6482 1.0015 1.0026 2027 23,214.6018 31.9546 1.0014 1.0023 Seunpun Mobiliei en Openbare Werken 16 RA-MOW-2009-006

2028 23,243.2367 28.6349 1.0012 1.0021 2029 23,268.8897 25.6530 1.0011 1.0019 2030 23,291.8657 22.9760 1.0010 1.0017 3.2.2 Baseline prognosis for road safey quaniies a he regional level This secion presens formulae for he baseline prognoses of he road safey quaniies a he regional level. The compuaions for injury accidens are given firs, followed by hose for oal casualies, faaliies, serious casualies and lasly sligh casualies. 3.2.2.1 Baseline prognosis for he number of injury accidens The baseline prognosis for he number of injury accidens is obained by muliplying a caegory s raffic performance and is baseline risk. The baseline risk is deermined by he risk in he reference year and he auonomous risk change. The baseline risk br c, on caegory c in year afer he reference year is given by: br c, = f * c, 1 * f c,2... fc, rc wih f c, = modificaion facor for he auonomous risk on caegory c in year r c = acciden risk for caegory c in he reference year The baseline prognosis for he number of injury accidens, b_ias A,c, in year on caegory c due o growh scenario A is given by: b _ IAs A, c, = br c, * TPA, c, wih br c, = baseline risk in year for caegory c TP, A c, = raffic performance in year on caegory c due o growh scenario A (secion 3.2.1 ) The modificaion facor for he auonomous risk is esimaed using he exensive ime series daa available on casualies in Belgium (1970-2006). The exponenial funcion recommended in lieraure (Van den Bossche e al., 2005; COST329, 2004; Oppe, 1989) is fied o he daa. The exponenially decreasing rend over ime can be explained in erms of a collecive learning process (COST329, 2004), caused by he ever-increasing knowledge of he raffic safey problem and he consan improvemen of he safey performance of he road ranspor sysem. For example, cars and roads become beer equipped, raffic safey educaion improves and legislaion and enforcemen effors have also increased over ime. All hese measures resul in a decreased risk over ime and he exponenial funcion has been proposed o describe his rend. The equaion for he exponenial funcion is bx y ae (Ahlfors, 1953). Here, x is he value of he independen variable (years), while y is he value of he dependen variable (road safey risk). The number e (approximaely 2.7182) is he base of naural logarihm. Figure 3 shows a decreasing rae of -0.0449. The same rae is assumed for each year. Therefore, 0.9551 is used as he yearly modificaion facor for he auonomous risk in he model. Seunpun Mobiliei en Openbare Werken 17 RA-MOW-2009-006

Toal casualy's risk 40 35 30 25 20 15 10 5 y = 7E+39e -0.0449x R 2 = 0.9737 0 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010 Year Figure 3: Toal casualy s risk in Belgium: 1970-2006 3.2.2.2 Baseline prognosis for he number of casualies The baseline prognosis for he number of casualies, b_c A,c, in year on caegory c due o growh scenario A is given by: b C A, c, _ = b _ IAs A, c, * N0, c wih b _ IAs A, c, = baseline prognosis for he number of injury accidens in year on caegory c due o growh scenario A N 0,c = number of casualies per injury acciden on caegory c in he reference year (secion 3.1 ) 3.2.2.3 Baseline prognosis for he number of faaliies The baseline prognosis for he number of faaliies, b_fats A,c, in year on caegory c due o growh scenario A is given by: b FATs A, c, N 3, c _ = * b _ C A, c, wih b C A, c, 100 _ = baseline prognosis for he number of casualies in year on caegory c due o growh scenario A N 3,c = number of faaliies per 100 casualies on caegory c in he reference year (secion 3.1 ) 3.2.2.4 Baseline prognosis for he number of serious casualies, b_scs A,c, The baseline prognosis for he number of serious casualies, b_scs A,c, in year on caegory c due o growh scenario A is given by: b SCs A, c, _ = N 2, c * b _ C A, c, wih b C A, c, _ = baseline prognosis for he number of casualies in year on caegory c due o growh scenario A Seunpun Mobiliei en Openbare Werken 18 RA-MOW-2009-006

N 2,c = number of serious casualies per oal casualies on caegory c in he reference year (Secion 3.1 ) 3.2.2.5 Baseline prognosis for he number of sligh casualies, b_slc A,c, The baseline prognosis for he number of sligh casualies, b_slc A,c, in year on caegory c due o growh scenario A is given by: b _ SLCA, c, b _ CA, c, b _ FATs A, c, b _ SCs A, c, b C A, c, wih _ = baseline prognosis for he number of casualies in year on caegory c b FATs A, c, due o growh scenario A _ = baseline prognosis for he number of faaliies in year on caegory c b SCs A, c, due o growh scenario A _ = baseline prognosis for he number of serious casualies in year on caegory c due o growh scenario A 3.2.3 Baseline prognosis for raffic performance a he locaional level To assess he effeciveness of locaional measures on a locaion belonging o caegory c in year, he number of injury accidens and casualies ha would occur a ha locaion before applying he locaional measures is required. These road safey quaniies are he baseline prognosis a he locaion and are deermined by he oal injury accidens and casualies on he caegory o which he locaion belongs and he raffic performance a he locaion. To compue he raffic performance a he locaion, i is assumed ha locaions have he same growh facor for raffic performance as he road or inersecion caegory o which hey belong. The raffic performance of a locaion, TP A,loc,, in year due o growh scenario A, is calculaed as follows. TP * A, loc, g A, c,1 *...* g A, c, TPloc wih TP loc = raffic performance of locaion, loc in he reference year g A c,, = growh facor for he raffic performance in year on caegory c due o growh scenario A (Table 3) 3.2.4 Baseline prognosis for road safey quaniies a he locaional level These are he number of injury accidens, sligh casualies, serious casualies and faaliies a a paricular locaion before a measure is applied. They are deermined by he oal number of road safey quaniies on he caegory o which he locaion belongs and he raffic performance a he locaion. This is illusraed for injury accidens (he oher severiy levels are compued analogously). The number of injury accidens in year a locaion, loc, belonging o caegory c due o growh scenario A before applying a measure is given by: IAs IAs TP wih A, loc, A, c, A, loc, * TPA, c, TP A, c, = raffic performance in year on caegory c due o growh scenario A IAs, A c TP,, = number of injury accidens in year on caegory c due o growh scenario A A loc, = raffic performance a locaion, loc, in year due o growh scenario A. Seunpun Mobiliei en Openbare Werken 19 RA-MOW-2009-006

By replacing he number of injury accidens wih sligh casualies, serious casualies and faaliies, he baseline prognosis for oher severiy levels are compued. 3.3 Measure prognosis In his phase of he model, he impac of measures is calculaed on he number of injury accidens, sligh casualies, serious casualies and faaliies. The final selecion of measures incorporaed in he model will be decided on in consulaion wih policy makers. Here, he mehod used o calculae he impac of one or several measures on road safey is described. For each measure, he modificaion facor per severiy level is required. The modificaion facor is he expeced proporion of he raffic safey quaniies remaining afer he measure is applied. 3.3.1 Compuing effeciveness of measures Two ypes of measures need o be disinguished wih respec o his regional: regional and locaional measures. Firs, he regional measures are described afer which he locaional ones are explained. 3.3.1.1 Regional measures Regional measures have an impac on road safey in he enire region. From he baseline prognosis, b_ias A,c,, b_slc, b_scs A,c, and b_fats A,c, denoe he number of injury accidens, sligh casualies, serious casualies and faaliies respecively. Assume P regional measures are applied in year on caegory c wih modificaion facors given by V IAs,p, V SLC,P, V SCs,p and V FATs,p. The remaining road safey quaniies in year on caegory c due o growh scenario A afer applying P measures are obained by muliplying he modificaion facors and he road safey quaniies in he baseline. This is illusraed by he following equaions: IAs, A, c = b _ IAs A, c, * VIAs,1 *...* VIAs, P SLC, SCs, A, c = b _ SLCA, c, * VSLC,1 *...* VSLC, P A, c = b _ SCs A, c, * VSCs,1 *...* VSCs, P FATs, A, c = b _ FATs A. c * VFATs,1 *...* VFATs, P 3.3.1.2 Locaional measures One of he objecives of he model is o assess he effeciveness of locaional measures on safey. However, no all measures can be applied on all road or inersecion caegories, for example a roundabou on a highway. Furher, cerain measures can only be implemened a locaions as i may be very expensive o apply hem in he enire region. Such measures are ermed locaional measures and only have effeciveness a he locaion hey are applied. The effeciveness of locaional measures on road safey is obained as follows. Suppose K locaional measures are applied a a paricular locaion, loc, wih modificaion facors given by V IAs,K, V SLC,K, V SCs,K and V FATs,K for injury accidens, sligh casualies, serious casualies and faaliies respecively. The remaining number of injury accidens (IAs A,loc, ), sligh casualies (SLC A,loc, ), serious casualies (SCs A,loc, ) and faaliies (FATs A,loc, ) a a locaion, loc, in year due o growh scenario A afer applying K locaional measures is given by he following equaions: IAs A, loc, = IAs A, loc, * VIAs,1 *...* VIAs, K SLC A, loc, = SLC A, loc, * VSLC,1 *...* VSLC, K SCs A, loc, = SCs A, loc, * VSCs,1 *...* VSCs, K FATs A, loc, = FATs A. loc, * VFATs,1 *...* VFATs, K Seunpun Mobiliei en Openbare Werken 20 RA-MOW-2009-006

The mehodology uilized o compue he effeciveness of measures (secion 3.3.1 ) is based on he assumpion ha he effec of each measure on road safey is independen of oher measures. Neverheless, his assumpion is likely o be incorrec in some cases. One would expec dependence (ineracions) beween some combinaions of measures, for example, he combinaion of alcohol and drugs on road safey (Van Vlierden and Lammar, 2007). Such ineracions are no aken ino accoun by his mehodology and in he following secion we describe he various forms of ineracions o be invesigaed in fuure. 3.3.2 Dependence among measures Though no performed in his firs analysis, here are possible ineracion effecs beween measures when implemened simulaneously. For example, he combinaion of humps and rumble srips reduce he number of injury accidens by 0.27(-0.30,-0.24) (Elvik and Vaa, 2004). These ineracions are rarely modelled and only he individual effec of he measures is assumed. The hypohesis is ha measures affec road safey independenly. To beer reflec realiy, ineracions will be aken ino accoun in a fuure version of his model. In paricular, aenion will be paid o: synergism, (he oal effec exceeding he sum of he individual effecs), subsiuion (he oal effec being less han he sum of he individual effecs) and addiiviy (he oal effec equal o he sum of he individual effecs). 3.4 Savings When he effeciveness of he regional and locaional measures is calculaed, he number of saved IAs, SLC, SCs and FATs can be deermined. The number of saved quaniies resuls from he difference beween he baseline and he measure prognoses. For example, he saved number of IAs on caegory c in a specific year due o growh scenario A equals he difference beween he number of IAs according o he baseline prognosis and he measure prognosis in ha year. If BP-IAs A,c, and MP-IAs A,c, denoe he number of IAs in he baseline prognosis and measure prognosis respecively, hen he corresponding number of saved IAs, S-IAs A,c, is obained as follows: S IAs A, c, BP _ IAs A, c, MP _ IAs A, c, The saved number of sligh casualies, serious casualies and faaliies is calculaed using he same procedure. 3.5 Cos-benefi analysis In order o deermine he benefis of he applied regional and locaional measures, he value of an injury acciden, a sligh casualy, a serious casualy and a faaliy mus be expressed in euros. A cos-benefi analysis is one of he ools used o assess he possible measures by comparing heir profiabiliy. This analysis allows he comparison of measures wih a differen life span and jusifies he choice beween possible combinaions of measures. Here, we focus on wo crieria namely: ne cash value (NCV) and cos-benefi raio (CBR). The NCV is he difference beween he cash value of benefis and coss. A se of measures is profiable when he NCV is posiive. However, When he NCV is used; large measure ses (wih large coss and benefis) are given preference as hey have higher NCV han smaller measure ses. The cos-benefi raio is he raio beween he cash value of coss and benefis. I is possible o have negaive benefis in a cos-benefi analysis. The negaive benefis should no be added o he coss bu should be deduced from he posiive benefis. Adding negaive benefis o he coss would wrongfully lead o a lower CBR. This is one of he disadvanages of he CBR, (Janssen, 2005). Anoher limiaion of he CBR is ha all effecs need o be expressed in moneary erms. On he oher hand, a cos-benefi analysis has he advanage of aking ino accoun boh he inended effecs such as he consrucion coss and side effecs for Seunpun Mobiliei en Openbare Werken 21 RA-MOW-2009-006

example noise nuisance (Vlakveld e al., 2005). Furher, he CBR is useful for comparing he profiabiliy of measures wih varying coss (Janssen, 2005). The NPV and he BCR rank projecs according o heir ne benefi. Thus, i remains a pracical ool for he comparison of several possible measures (Ampe e al., 2008). 3.5.1 Benefis The benefis B c, in year on caegory c, are he producs of he number of saved quaniies and he values in euros corresponding o a paricular road safey quaniy. Le W IAs, W SLC, W SCs and W FATs denoe he moneary value of an injury acciden, a sligh casualy, a serious casualy and a faaliy respecively. Then, B c, is given by: B c, S IAs A, c, * WIAs S SLCA, c, * WSLC S SCs A, c, * WSCs S FATs A, c, The oal sum of benefis over all caegories, C, in year is obained by summing he benefis on all caegories in ha year i.e. 3.5.2 Coss C B C, = B c,. C 1 The oal cos per year is deermined by summing he cos of regional and locaional measures. Measure coss on road caegories are calculaed per kilomere while hose of inersecion caegories are esimaed per piece. Informaion wih respec o coss of measures will be obained from auhoriies in Flanders. In he following secions an explanaion of how he coss of regional and locaional measures are calculaed is given. The calculaions for regional coss are presened firs and hose of locaional coss will follow. 3.5.2.1 Coss of a regional measures Suppose a se m conains j regional measures i.e. m=1, 2,, j and he cos of measure j in year denoed by R j,. If hese measures are applied in year, on caegory c, he oal cos R c, of he applied regional measure se on caegory c in year equals he oal cos of all measures as follows: Error! Objecs canno be creaed from ediing field codes.= j j R j, 1 The oal cos, R, of all regional measures applied in year on all caegories C equals: C R = R c,. c 1 * W FATs 3.5.2.2 Coss of a locaional measures Le a se m conain n locaional measures i.e. m=1, 2,, n ha are inroduced a locaions belonging o caegory c in year. Le L,n, be he cos of locaional measure n applied in year. The oal cos L c, of he applied locaional measures on caegory c in year hen equals: n L c, = Ln, lm wih 1 l m = lengh of he road segmen if he measure is applied on a road segmen l m = 1 if he measure is applied on an inersecion The oal cos, L, of all locaional measures applied in year on all caegories C equals: Seunpun Mobiliei en Openbare Werken 22 RA-MOW-2009-006

C L = L c c 1,.The overall cos in year is deermined by summing he cos of regional and locaional measures (R +L ). The cash value of coss (CVC) The CVC is obained by expressing he oal coss in year of all he applied measures in erms of he nominal value of he reference year using a discoun rae R as below: CVC 1 (1 R) R L wih R L R = discoun rae for he enire calculaion period = cos of locaional measures in year = cos of regional measures in year = las year of he calculaion period The cash value of benefis (CVB) The CVB is calculaed by expressing he oal benefis over all caegories C in year in erms of he nominal value of he reference year using a discoun rae R as follows. C, CVB = C 1, 1 B C, (1 R) The ne cash value (NCV) The NCV is he difference beween he cash value of coss (CVC) and he cash value of benefis (CVB) and is compued by he following expression. NCV = CVB CVC The cos-benefi raio (CBR) This indicaes how much higher he coss are compared o he benefis and i is compued as: CBR = CVC / CVB Apar from he cos-benefi analysis, here are oher evaluaion ools. These are briefly described wih heir advanages and disadvanages. The firs mehod used for measure assessmen is he cos-effeciveness analysis (CEA). This mehod is ofen used o deermine he effeciveness of measures. This is done using a raio of saved vicims versus he cos aached o he measure. This mehod only includes he inended effecs (safey effecs in his case) and he coss incurred o aain hese effecs. However, in order o make policy decisions i is required o have insigh ino all relevan social effecs and no only he inended ones (Vlakveld e al., 2005). Anoher ool is he cos uiliy analysis (CUA). This is comparable o he CEA. The only difference is ha he CUA uses qualiy adjused life years as a basic concep. While in he CEA couning is done in amoun of life years, he CUA uses a weighing for he life years in erms of qualiy of life. This ool akes ino accoun he severiy of an injury and he disabiliy going wih he injury. One barrier o his ool is ha he kind of medical daa required is no always available (Ampe e al., 2008). Incase of muliple decisions, mehods ha handle several crieria are required. One of he opions is he muli-crieria analysis or muli-crieria decision analysis (MCDA). A Seunpun Mobiliei en Openbare Werken 23 RA-MOW-2009-006