The STARs evaluation tool: optimising network performance, road worker safety and road user safety during roadworks and maintenance.

Abstract

The aim of the Scoring Traffic at Roadworks (STARs) project is to optimise network availability, road worker safety and road user safety during roadworks. The objective was to develop a methodology to score roadworks schemes on these three interdependent risk areas which are usually considered in isolation, and to produce a practical tool for use particularly by contractors and contracting authorities in planning and assessing roadworks schemes and setting contractor targets. To achieve this objective, three risk equations for performance at roadworks have been developed and included in the STARs Evaluation Tool, together with a Multicriteria Solver Module to transform the scores into a “STARs” scale. This scale is used to produce an unbiased rating of individual management strategies. This paper describes the tool structure, briefly covers the theory governing the equations and illustrates use of the methodology with a case study. The STARs project is part of the ENR2011 DESIGN programme which is a cross-border funded Joint Research Programme initiated by ERA-NET Road.

Full text

N. Ni Nuaillain, R.Sarrazin, J.Wennstrom, J. Weekley / Transport Research Arena 2014, Paris
The STARs evaluation tool: optimising network performance, road
worker safety and road user safety during roadworks and maintenance
Nora Ni Nuallain
a
, Renaud Sarrazin
b
, Jonas Wennström
c
, Jill Weekley
d*
a
Trinity College Dublin, Republic of Ireland
b
BRRC, Belgium
c
VTI, Sweden
d
TRL, UK
Abstract
The aim of the Scoring Traffic at Roadworks (STARs) project is to optimise network availability, road worker
safety and road user safety during roadworks. The objective was to develop a methodology to score roadworks
schemes on these three interdependent risk areas which are usually considered in isolation, and to produce a
practical tool for use particularly by contractors and contracting authorities in planning and assessing roadworks
schemes and setting contractor targets. To achieve this objective, three risk equations for performance at
roadworks have been developed and included in the STARs Evaluation Tool, together with a Multicriteria Solver
Module to transform the scores into a “STARs” scale. This scale is used to produce an unbiased rating of
individual management strategies. This paper describes the tool structure, briefly covers the theory governing the
equations and illustrates use of the methodology with a case study. The STARs project is part of the ENR2011
DESIGN programme which is a cross-border funded Joint Research Programme initiated by ERA-NET Road.
Keywords: roadworks; maintenance; road worker safety; road user safety; network performance; risk
management
Résumé
Le but du projet Scoring Traffic at Roadworks (STARs) est d'optimiser la disponibilité du réseau routier, la
sécurité des travailleurs routiers et la sécurité des conducteurs et voyageurs pendant des travaux routiers et
d'entretien. L'objectif était de développer une méthode pour classer des travaux routiers selon ces trois domaines
interdépendants qui sont généralement considérés en séparément, et de produire un outil pratique qui sera utilisé
principalement par des entrepreneurs et par des pouvoirs adjudicateurs pour organiser et évaluer des projets
routiers et définir des objectifs pour des entreprises routières. Pour atteindre cet objectif, trois équations de risque
pour l'exécution des travaux routiers ont été développées et inclus dans l'outil d'évaluation, le STARS Evaluation
Tool, avec le Multicriteria Solver Module, transformant des notes en échelle "STARs". Cette échelle est utilisée
pour donner une note impartiale à des stratégies de gestion individuelles. Ce document explique la structure de
l'outil, couvre brièvement la théorie utilisée pour définir des équations et pour expliquer la méthode il y a une
etude de cas. Le projet STARs fait partie du programme ENR2011 DESIGN qui est financé par le programme
transfrontalier, Joint Research Project initié par ERA-NET Road.
Mots-clé: travaux routiers; entretien ; sécurité des travailleurs routiers ; sécurité des conducteurs et voyageurs ;
disponibilité du réseau routier ; gestion des risques
*
Tel.: +44 134770423; fax: +44 1344770356.
E-mail address: jweekley@trl.co.uk.

N. Ni Nuaillain, R.Sarrazin, J.Wennstrom, J. Weekley / Transport Research Arena 2014, Paris 2
Nomenclature
AADT Average annual daily traffic
LMCC Lorry-mounted crash cushion
MCDA Multi-criteria decision analysis
PPE Personal protective equipment
RUS Road user safety
RWS Road worker safety
TMA Truck-mounted attenuator
TP / NP Traffic performance / Network performance
1. Introduction
ERA-NET ROAD II (ENR2) is a Coordination Action funded by the 7th Framework Programme of the EC
(www.eranetroad.org). Within the framework of ENR2 this joint research project was initiated as answer to the
call “Design – Rapid and Durable Maintenance Methods and Techniques” issued within a cross-border funded,
trans-national joint research programme. The funding National Road Administrations (NRA) in this joint
research project are Belgium, Germany, Denmark, Finland, Netherlands, Norway, Sweden, Slovenia and United
Kingdom.
The aim of the STARs project was to develop a methodology to score road works schemes on three
interdependent aspects which are normally considered in isolation: road user safety, road worker safety and
network performance. This will encourage national road authorities and their suppliers to take a holistic
approach to managing safety risk and network performance and facilitate more comprehensive ranking of
alternative management strategies. To achieve this objective, three risk equations for performance at roadworks
have been developed and included in the STARs Road Works Evaluation Tool, together with a Multicriteria
Solver Module transforming the individual and absolute scores into a “STARs” scale. This scale will be used to
produce an unbiased rating of individual management strategies.
At the time of writing, the project and the tool development are in the final stages, but the user interface for the
tool is undergoing a stakeholder consultation exercise and hence illustrations of the tool itself are not included in
this paper. This paper describes the working structure of the tool, briefly covers the theory governing the
evaluation criteria and finally demonstrates the methodology with a case study.
2. Tool methodology
This section outlines the structure of the tool, which consists of three ‘modules’: Road works project, Evaluation
tool and Multicriteria solver module, as shown in Figure 1.
2.1. Module I. Road Works Project
This module is the user input module where the user describes the project, i.e. the road work scheme to be
evaluated. Basically, data about the infrastructure, the road works, the layout and operational parameters must
be provided. A list of relevant parameters has been identified for each of the following three road work types:
•Major: Road works that are in place for long periods, where workers may be behind an approved safety
barrier and / or different equipment, layouts or techniques are used to manage traffic and safety compared to
minor works.
•Minor: Stationary (i.e. not mobile) road works that can only be carried out where conditions meet defined
criteria in the appropriate national guidance. Definitions may be given in terms of traffic flow, visibility
and/or the duration of the work.
•Mobile: Mobile and intermittent road works of limited duration carried out using, for example, vehicles and /
or mobile devices (such as TMA / LMCC) to create a safe working environment for short-term access to
specific sections of the road.
These definitions have been created to be applicable and meaningful in different countries.

N. Ni Nuaillain, R.Sarrazin, J.Wennstrom, J. Weekley / Transport Research Arena 2014, Paris
The parameters are classified following two main categories:
•The Fixed parameters are site, country and layout-specific or operational parameters the user decides to fix.
This means only one specific value is taken by the parameter. Typical fixed parameters are motorway size,
road works types, type of layout, length of works, traffic flow level, number of lanes closed, traffic
management safety standards, etc.
•The Variable parameters are operational parameters the user chooses to vary, in order to evaluate the impact
on the performance of the road works scheme. Typical variable parameters are speed compliance
management, type of delineation, width of open lanes, lateral distance between workzone and traffic, etc.
Some parameters are fixed in the STARs evaluation tool; others can be selected as fixed or variable by the user.
The legislative, strategic and operational issues specific to the country as well as the local constraints will of
course highly influence the user’s choice.
Values attributed to fixed and variable parameters together populate the ‘Alternatives’ table: a matrix defined by
modifying the variable parameters one by one and in which each alternative is described by the values associated
to each parameter.
2.2. Module II. Evaluation Tool
The second module is the core of the evaluation tool, consisting of the three evaluation criteria, i.e. the three risk
equations designed to assess the performance of the alternatives with respect to the three risk areas: road user
safety, road worker safety and network performance. Each equation works independently but all three use the
same set of parameters and feed, after n runs of the evaluation tool (n being the number of alternatives to assess),
the Evaluation Table. These equations have been developed using state of the art literature and partners’
knowledge and are briefly described in Section 3.
Essentially, the Evaluation table is the former Alternative table updated with the absolute (i.e. non-standardised)
scores for each alternative from the three criteria (see Table 1). The next step (in Module III) is then to select the
Figure
1
: Working structure of the STARs evaluation tool

N. Ni Nuaillain, R.Sarrazin, J.Wennstrom, J. Weekley / Transport Research Arena 2014, Paris 4
“best candidates” from the alternatives and transform the scores into comparable scales (e.g. normalised values
ranging from 0 to 1).
Table 1: Illustration of an evaluation table extract (showing alternativesand criteria)
Alternative RUS RWS NP
Alt 1 3 95 14
Alt 2 50 50 11
Alt 3 15 40 8
…. …. …. ….
Alt n 18 95 22
2.3. Module III. Multicriteria Solver Module
Solving a multicriteria problem starts with identifying the most efficient solutions of the problem (i.e. the best
candidates with respect to the constraints and the nature of the problem). The efficiency of a solution (an
alternative) is here measured on the basis of the dominance principle; meaning that if a solution is dominated by
another, it is removed from the set of solutions. Within the STARs evaluation tool, this step is achieved by
making pairwise comparisons of the evaluations obtained by the alternatives on the three criteria.
In a concrete multicriteria problem with several criteria and a large set of alternatives, most of the solutions are
efficient because of the poverty of the dominance relation. Then, to solve the multicriteria problem, it may be
necessary to use heuristic models which are complex and time-consuming problem solver. In STARs, we are
dealing with a limited set of criteria and a quantifiable design space. Therefore, the dominance relation remains
relatively strong which allows us to exclude inefficient alternatives and, to some extent, highlight the most
promising ones. The remaining alternatives constitute the set of efficient solutions.
Considering the example scores in Table 1 above illustrates this; if more efficient alternatives are characterised
by lower scores, alternative n will be excluded as it is less efficient than alternative 1 on nearly all the criteria.
However alternatives 2 and 3 will remain for further processing because they are more efficient than others on at
least one criterion. Ranking the efficient alternatives is difficult and, as usual in MCDA methods, additional
information is necessary in order to reduce the number of “incomparabilities” and then solve the problem.
Within STARs the additional information is provided by the use of utility functions to aggregate the 3
independent scores and finally rank the alternatives.
Utility functions play two important roles. Firstly they model the preferences on each criterion by expressing the
scores in a common normalised score. Within the STARs evaluation tool each utility function is characterised
by a marginal utility function that shapes a performance function for each criterion and transform rough scores
from the Evaluation table into normalised and comparable scores. These comparable scores are then further used
to attribute the ‘STARs rating’, which is assigned simply, according to:
Table 2: STARs rating categories
STARs rating Score
1
0score < 0.2
2
0.2 score < 0.4
3
0.4 score < 0.6
4
0.6 score < 0.8
5
0.8 score 1

N. Ni Nuaillain, R.Sarrazin, J.Wennstrom, J. Weekley / Transport Research Arena 2014, Paris
The application of utility-based methods was chosen not only because the concept is appropriate for ranking
problems, but also because it is user-friendly. Moreover the use of marginal utility functions allows easy
calibration of the evaluation tool to local constraints at a later stage.
The last objective of the STARs Road Works Evaluation Tool is to propose a single global star rating; using the
three individual scores for road user safety, road worker safety and network performance. Classically, utility
based methods use weights associated with each marginal utility function. Again in view of giving flexibility for
further development and customisation, it was decided to apply a simple additive form to aggregate the utility
functions and allow the user to assign the set of weights (e.g. 40% ; 40% ; 20%) to each criterion (the weights
give the relative importance of each criterion).
3. Risk Equations
This section describes some of the theory behind each of the risk equations / evaluation criteria. Each equation
was developed using a different approach, chosen based on characteristics of the risk area. All three equations
are based on variables drawn from the same set of input parameters in order to ensure they can be aligned in
order to form the evaluation module of the tool.
3.1. Network / traffic performance
Road works can have a large impact on motorway traffic flows and every year cost society large sums of money
in the form of delays for road users, as well as causing other traffic performance related issues such as increased
vehicle operation costs, emissions and noise. Road works firstly cause delays due to reduced speed limits in
work zone area. Secondly, since a lane closure will reduce the road’s capacity there are risks of queues in higher
traffic flows. This will cause extensive delays for road users but may also result in traffic diversion to the
surrounding road network if road users anticipate long travel times. Costs for traffic delays have been indicated
to have a large impact on total road user costs related to roadworks (NJDOT 2001). In this project road user
delays have been used as an indicator for the traffic performance and used as rating criteria for the NP risk score.
In the tool the delays are estimated using a mathematical model which considers road users’ delays in different
traffic flow conditions. The delays can be divided into three main components, namely free-flow delays and
queue delays and diversion delays (NJDOT 2001). In free-flow traffic conditions the delays are mainly
influenced by the work zone’s geometrical properties and traffic management strategies, queue delays
characterised by the roadwork’s capacity and diversion delays by the road network. In the risk equation only the
former two have been considered. The basic equation for modeling the network performance risk based on
delays can be formulated as:





(1)
where Traveltime
wz
is the travel time in the work zone is a function of its layout characteristics (L, M), duration
(D) and traffic flow (Q). Finally Traveltime
norm
is the travel time in normal conditions (B).
The methodology used to assess the free-flow delays and queue delays is based on a queuing theory based model
suggested by (Jiang 1999). In this model the free-flow delays are modelled based on simple kinematic
relationship for acceleration, deceleration and work zone speed. Different work zone configurations and
mitigation strategies will impact the work zone speed differently. These impacts can be expressed as a ‘typical
speed reduction value’(Benekohal, Ramezani et al. 2010) which have been used to model the work zone speed
The estimations of work zone speeds are based on a simplification that the most effective speed reduction
technique is the one that will be used to reduce the actual speed. This is due to lack of data on interaction effects
between different techniques. Therefore the following equation was used to model the work zone speed:
  
 !"
#$%&'()*)+,-./
0!"1234

(2)
where Vis the resulting work zone speed, V
signed
is the signed work zone speed, LawObedienceFactor is the
factor of normal work zone speeding and V
reduction
is a vector of all speed reduction values associated with work
zone layout.

N. Ni Nuaillain, R.Sarrazin, J.Wennstrom, J. Weekley / Transport Research Arena 2014, Paris 6
A queuing delay term is added to the free-flow delays in order to model the queuing situation. The main
parameter for the queuing model is the work zone capacity but a queue discharge rate for when the traffic flow
recovers is also needed. To estimate the work zone capacity, a regression model was used (Kim, Lovell et al.
2000). The capacity for a specific work zone is depending on its lane configuration, traffic composition, lane
design, length and work intensity. Based on this data, the queues and delays are simulated for each hour
including how queues build up over each hour by taking into account how the traffic flows are distributed during
aday. From this it is possible to estimate the total delay for a works. The models used have been verified against
both field and simulation data for both delays and work zone capacities, and reasonable results were achieved.
3.2. Road user safety
In terms of Road User Safety (RUS), the literature presents a number of models for calculating the crash rate or
frequency of crashes at work zones (e.g. Khattak et al. 2002, Li & Bai 2008, Meng et al. 2010). Studying the
different models, the main hyperparameters that influence the probability of a crash are:
•(L) length of work zone (including tapering and activity zone),
•(D) duration of works,
•(Q) traffic flow, and
•(B) base risk or road type (U) (i.e. urban or rural road)
Considering the various models, the model of Meng et al. (2010) is considered to be the most suitable for the
STARs project:
567
8
9:
7
;
<=
;
97
>
<?9@
A
;BC
(3)
where fis work zone crash frequency,
α
0
,
α
i
(i= 1, .., 4) are coefficients to be estimated,
ε
is the random error
term, and x
i
(i= 1, 2, 3) are the three explanatory hyperparameters associated with L,Dand Q, respectively. The
coefficients, which provide the contribution factors for each parameter, are estimated by performing regression
analysis on historical data. Using sample data from Meng et al. (2010) the equation becomes
?
DECFGFCH I
JGDD>DK L
JG8EFAHM
<
8E>FN
(4)
However, it is recognised that this model only includes L,D,Qand U, and therefore does not include all the
parameters that can be associated with a work zone. In particular, it does not include specific layout information.
It is therefore considered suitable to introduce an additional hyperparameter for the STARs project. This
hyperparameter should include information on the layout and traffic management of the work zone. It is
proposed to refer to this hyperparameter as a risk mitigation factor (M).
To determine M, the literature has been studied and the parameters that are considered to impact RUS are
identified. The literature shows that road users should be guided in a clear and positive manner while
approaching and travelling through work zones (MUTCD, 2009). Adequate warning, delineation and traffic
management should be provided to assist in guiding road users in advance of and through the work zone by
using proper pavement markings, signing and other devices that are effective under changing conditions
(MUTCD, 2009). In addition, safe speed limits should be set without un-necessarily or excessively reducing the
limit and frequent changes in the speed limit should be avoided. The temporary traffic control plan and
geometric layout should be designed so that the speed does not have to be reduced more than two speed limit
steps from the permanent posted limit. Finally, it has also been found that controlling speed with enforcement
measures reduces accidents (Morgan et al. 2010).
Using this information, thirteen parameters that contribute to the hyperparameter Mwere identified and
alternatives relating to each of these parameters were also defined. The values associated with each of the
thirteen parameters are referred to as M1 to M13. The values for each of the alternatives would preferably be
defined with a regression analysis using data obtained from previous work zones. However, there is currently a
lack of data with specific layout details and for a proof-of-concept, it is considered reasonable to make
assumptions based on the literature. Therefore, for STARs, Mis calculated as the product of the assigned values
of all the parameters that contribute to M. i.e.
O<P<QEE<OR
(5)

N. Ni Nuaillain, R.Sarrazin, J.Wennstrom, J. Weekley / Transport Research Arena 2014, Paris
The equation used in the STARs tool is therefore:
?S
DECFGFCH I
JGDD>DK L
JG8EFAHM
<
8E>FN
T<
(6)
According to Rouphail et al. (1988) a constant rate of 0.49 accidents / km-day is appropriate. Therefore, for short
term road works (minor and mobile), this constant rate multiplied by Mis used.
UEVW<<<
(7)
3.3. Road worker safety
The risk score for road worker safety is calculated using the following equation:
XYZX[\Z)+X
]

]
9X
^

^

,(8)
where B is the ‘base risk’ and is a factor representing the inherent risk of the works location, and

C

F
,(9)
where

C
_ O `'[a+b'
OEP c+a'[a+b'
(10)

F
d O 'e
OEP (fa
OEg c+(fa
(11)
R
A
is the risk score for traffic management and mobile elements of the road works. For major and minor road
works this will be the setting out and clearing away phases; for mobile works it is likely to cover the whole
works. R
B
is the risk score for the static elements of the road works. For mobile works this term is likely to be
equal to zero. The equations are:
X
]
 
h
i
C
9
>

C
9
>

H
jk
,
(12)
X
^
 
h
i
F

F
9
A

A
k
,
(13)
where C represents the time per lane required for laying traffic management per carriageway crossing, D
1
/ D
2
/
D
3
/ D
4
are the length of time (hours) for setting out works / carrying out works / maintenance / clearing away
works respectively, and L
1
/L
2
/ L
3
are the number of workers required for setting out and clearing away / in
closure / for maintenance. L
4
is the number of carriageway crossing required and

h

L
l
L
m
L
n
o
L
p
L
q
D>
r9
L
l
L
n
o
L
p
L
q
K8
rO
>

(14)
The square-bracketed term in both R
A
and R
B
represents the exposure of workers carrying out the traffic
management or road works respectively. Q
*
represents the risk caused by the vehicles passing a worker, where
Q
1
is the hourly flow, Q
2
is the speed limit through the works zone, Q
3
is the speed compliance factor, (=0.9 if
high enforcement, 1.0 if low enforcement or 1.1 if no enforcement), Q
4
is the percentage of HGVs in traffic flow
and Q
5
/ Q
6
is the number of lanes closed / open respectively.
Thus, the term Q
2
Q
3
represents the ‘actual speed’ of vehicles passing the workers, and the two terms represent
the difference in consequence of a worker being hit by an HGV compared to a car (with different speed
thresholds for each).
The remaining two terms in the equation are M
A
and M
B
which represent the risk mitigation measures relevant to
the traffic management and mobile elements, and the static elements respectively. M
A
and M
B
are factors taking a
value between 0 and 1, depending on the mitigation measures that are being used.
These factors (here denoted M) will be calculated as follows
:
S:
;
:
s

sBC
t
;BC
T:
;
t
;BC
uv
(15)

N. Ni Nuaillain, R.Sarrazin, J.Wennstrom, J. Weekley / Transport Research Arena 2014, Paris 8
where
w--
is the effectiveness score associated with a particular risk mitigation measure, N is the total number
of possible risk mitigation measures and n is the number of risk mitigation measures being used. Therefore as the
user selects each measure that is present at the particular roadworks being assessed, the mitigation factor
reduces, and the overall RWS risk score reduces. The starting value for M (i.e. before any measures have been
selected) will be the summation of all the possible risk reduction measures. Table 3 lists both general risk
mitigation measures that are considered in the tool and also for risk mitigation used for the static elements only.
Table 3: Risk mitigation measures considered in the tool.
Risk mitigation measures used for both static and mobile elements Measures used for static elements only
Workforce training Lateral distance between work zone
and traffic Delineation
Workforce PPE standards Use of TMA/LMCC Use of vehicle restraint systems
Traffic management vehicle conspicuity
standards Use of lookout systems Use of physical traffic management
Works design legislation / standards Lane closure mechanism Control through carriageway design
Safety management system requirements Signage levels Width of open lanes
Each mitigation measure must therefore be assigned a potential effectiveness score. The method chosen is to
score each mitigation measure (or type of measure) according to a five-level hierarchy of controls based on that
recommended by the UK’s Health and Safety Executive (HSE), which in turn is based on the general principles
of risk prevention set out in Article 6(2) of European Council Directive 89/391/EEC. This hierarchy consists of
five grades of effectiveness of hazard control: 1 - Elimination (the most effective), 2 – Substitution, 3 –
Engineering controls, 4 – Administrative controls, and 5 – Personal protective equipment (the least effective).
Each mitigation measure is therefore assigned a maximum value, based on its position within this hierarchy. The
options associated with each measure are then assigned values between one and this maximum according to their
effectiveness in achieving the full potential of that mitigation measure. For example: ‘Use of vehicle restraint
systems’ is assigned a maximum value of 20; this is a relative score based on the potential effectiveness of
vehicle restraint systems compared to other mitigation measures. The options the user can select for this
parameter are: ‘None’, ‘Low performance’, and ‘High performance’. These options are assigned the values 20, 5,
and 1 respectively. This would mean that if a high performance vehicle restraint system is selected, the full
potential of that measure is realised, and the risk score is reduced.
4. Case study: Evaluation of risk mitigation measures
A case study is presented below in order to illustrate how the developed tool can be used for selection of
roadwork risk mitigation strategies. The case study is for a roadworks carried-out on a two-lane motorway in the
‘fast’ lane where the hardshoulder is used for traffic flow meaning two lanes remain open during works. The
AADT of the road was assumed to be 60 000 vehicles/day, 20 % heavy goods vehicles and the roadworks to
have a duration of 20 days. The tool was used to generate different risk mitigation alternatives and evaluate
against the risk equations using the Multicriteria Solver Module.
Figure 2: Illustration of the roadwork layout used in the case study.
Based on the roadwork layout the following risk mitigation alternatives were identified as feasible options to be
considered for the scheme:
•Speed compliance management (Nothing or Vehicle activated signs or Spot enforcement)
•Delineation (Panels or Temporary barrier)
•Lane closure mechanism (Tapers or Both Tapers and IPV)

N. Ni Nuaillain, R.Sarrazin, J.Wennstrom, J. Weekley / Transport Research Arena 2014, Paris
•Use of vehicle restraint systems (None or Low performance)
•Use physical traffic management (None or Transverse pavement marking)
•Use of TMA/LMCC (None or One)
If all possible combinations of risk mitigation strategies are considered there are in total 96 possible solutions.
Using the Multicriteria Solver Module it was possible to reduce this number by eliminating the dominated
alternatives. By doing this, three alternatives remained as candidates for being the preferable solution. These are
shown in table 4 together with their respective risk mitigation strategy.
Table 4: The three dominant alternatives’ respective risk mitigation strategies.
Speed compliance
management Delineation Lane closure
mechanism Use of vehicle
restraint systems
Use of physical
traffic
management
Use of
TMA/LMCC
Alt 62 Vehicle activated
signs Temporary barrier Both Low performance None 1
Alt 64 Vehicle activated
signs Temporary barrier Both Low performance Transverse
pavement marking 1
Alt 96 Spot enforcement
(police presence, spot
speed cameras) Temporary barrier Both Low performance Transverse
pavement marking 1
The non-dominated solutions, their respective utilities and consequent STARs ratings are presented in table 5.
The dominance property means that every alternative has at least one risk utility that is higher than the two other
alternatives. Using a uniform weight distribution between the risk equations (33%) it is noted that all alternatives
earn the same number of STARs since their utilities all are in the range 0.4 to 0.6. For delineation, lane closure
mechanism, use of vehicle restraint systems and TMA/LMCC all have a preferable measure since the three
alternatives all use the same. If vehicle activated signs are used then transverse pavement marking gives a higher
utility. Conversely spot enforcement together with transverse pavement marking yield a higher utility.
Alternative no. 96 has therefore the highest utility and would be the preferable alternative. From the utilities it
can also be observed that the utilities of the RWS and RUS risk equations are more sensitive than the NP risk
equation. In this case it can be attributed to that the traffic flow is high and therefore queue delays are the
predominant delay component, which is reflected in the low score.
Table 54: Utilities and ratings of the three dominant alternatives.
Utility RWS Utility RUS Utility NP Utility ALL *RWS *RUS *NP *ALL
Alternative 62 0.7657 0.4210 0.3995 0.5287 4 3 2 3
Alternative 64 0.8173 0.4682 0.3994 0.5616 5 3 2 3
Alternative 96 0.8356 0.4884 0.3994 0.5744 5 3 2 3
Summary
This paper has described the structure of the STARs evaluation tool for assessing the road user safety risk, the
road worker safety risk and the network performance impact of a road works scheme. Data describing the
scheme are input by the user and a decision is made as to which of these parameter values are fixed for the
roadworks in question, and which are variable in order to assess the impact of different management strategies.
Risk scores are then calculated for road user safety, road worker safety and network performance for each set of
parameter alternatives. Finally these scores are normalised and the alternatives ranked using multicriteria
decision analysis with utility function methods to produce unified ‘STARs rankings’. This paper also briefly
describes the theory behind the three risk equations used to produce the three scores. The methodology has also
been applied to a case study to illustrate its usage when evaluating different schemes for a motorway roadwork.
Using the Multicriteria Solver Module it was possible to identify three roadwork schemes that were dominant
over the other alternatives.

N. Ni Nuaillain, R.Sarrazin, J.Wennstrom, J. Weekley / Transport Research Arena 2014, Paris 10
The STARs evaluation tool is designed to benefit two user groups in particular. For roadworks contractors it can
help plan works to maximise safety and minimise congestion, by calculating the impact on the three risk areas of
varying the traffic management options that are available. For maintenance contracting authorities it can help
specify the level of performance required for a scheme and set appropriate contractual limits for the impacts of
road maintenance and construction works on traffic.
Acknowledgements
The STARs project is part of the ENR2011 DESIGN programme of work which is a cross-border funded Joint
Research Programme initiated by ERA-NET Road.
References
Benekohal, R. F., H. Ramezani, et al. (2010). Queue and user's costs in highway work zones. Springfield, IL,
USA, Illinois Center for Transportation.
Fishburn, P. C., (1970). Utility Theory for Decision Making. Wiley, New York.
Griffith, A. & M. Lynde (2002). Assessing Public Inconvenience in Highway Work Zones, Oregon Department
of Transportation, Federal Highway Administration: 104.
Jiang, Y. (1999). "A Model for Estimating Excess User Costs at Highway Work Zones." Transportation
Research Record: Journal of the Transportation Research Board 1657(-1): 31-41.
Khattak, A.J., Khattak, A.J., Council, F.M. (2002). Effects of work zone presence on injury and non-injury
crashes. Accident Analysis and Prevention 34 (1), 19–29.
Kim, T., D. J. Lovell, et al. (2000). A New Methodology to Estimate Capacity for Freeway Work Zones.
Transportation Research Board Annual Meeting.
Li, Y., Bai, Y. (2008). Development of crash-severity-index models for the measurement of work zone risk
levels. Accident Analysis and Prevention 40 (5), 1724–1731.
Meng, Q., Weng, J., Qu, X. (2010). A probabilistic quantitative risk assessment model for the long-term work
zone crashes. Accident Analysis and Prevention 42 (6), 1866–1877.
MUTCD, (2009). Manual on Uniform Traffic Control Devices, Federal Highway Administration (FHWA),
mutcd.fhwa.dot.gov/pdfs/2009/part6.pdf
Morgan, J.F., Duley, A.R. and Hancock, P.A., (2010). Drivers responses to differing urban work zone
configurations. Accident Analysis and Prevention 42, 978-985.
NJDOT (2001). Road User Cost Manual. New Jersey Departemnt of Transportation (NJDOT).
Rouphail, N., Zhao, M., Yang, S., Fazio, J., (1988). Comparative study of short-and long-term urban freeway
work zones. Transportation Research Record 1163, TRB, National Research Council, Washington, DC.
Roy, B., (1985). Méthodologie Multicritere d’Aide a la Decision. Economica, Paris.