Transport Proceedings of the Institution of Civil Engineers

Refinement of a simplified road-user cost http://dx.doi.org/10.1680/tran.12.00057

model Paper 1200057
Batista dos Santos, de Picado Santos and Pissarra Received 19/06/2012 Accepted 14/02/2013
Keywords: management/mathematical modelling/roads & highways
Cavaleiro

ICE Publishing: All rights reserved

Refinement of a simplified
road-user cost model
j
1 Bertha Maria Batista dos Santos PhD j
3 Victor Manuel Pissarra Cavaleiro PhD
Assistant Professor, Department of Civil Engineering and Professor, Department of Civil Engineering and Architecture,
Architecture, University of Beira Interior, Covilha, Portugal University of Beira Interior, Covilha, Portugal
j
2 Luı́s Guilherme de Picado Santos PhD
Professor, Department of Civil Engineering, Architecture and
Georesources, Instituto Superior Técnico, Technical University of
Lisbon, Lisbon, Portugal

j
1 j
2 j
3

Two Portuguese universities (University of Beira Interior and University of Coimbra) researched and developed a
simplified road-user costs model for inclusion in a pavement management system, as well as input data for
Portuguese trunk road networks, between 2004 and 2007. The model can be used to calculate the average road-user
costs in relation to vehicle operating, accident, time and toll costs. This paper describes the main activities that led to
the model formulation and the input data, and presents a new development to simplify user costs estimation
considering changes in pavement condition and the influence of road work zones. These scenarios can then be easily
and reliably utilised within the process of road maintenance and rehabilitation, evaluating the needs and
consequences of road intervention actions for consideration in the cost–benefit and life-cycle cost analysis. This
paper also confirms the importance of these additional costs through a model application to a Portuguese road
network under private concession, comparing average costs with those for specific work zones and pavement
condition scenarios.

Notation Cmt i tyre market price for vehicle i (A/tyre)

AADT annual average daily traffic (vehicles/day) Ct i tyre cost for vehicle i (A/km)
AC accident cost ((A/km)/day)) ctoll i toll cost for vehicle i ((A/km)/vehicle)
AC j accident j cost ((A/km)/vehicle)) dCf incremental increase in fuel cost owing to
ac j accident j cost (police time cost) (A/accident) maintenance and rehabilitation (M&R) actions
ANA annual number of accidents with casualties ((A/km)/day)
(accidents/year) dVOT incremental increase in the value of time owing to
ANA j annual number of accidents j (accidents/year) M&R actions ((A/km)/day)
ANC k average number of casualties k by accident EA annual exposure to accidents in sections and
(casualties/accident) intersections (total) ((vehicles km)/year)
AR j accident j rate ((accidents/vehicle)/km)) FVOC,PSI VOC correction factor for a given PSI value
C total cracked pavement area (m2 /100 m2 ) ((A/km)/vehicle)
CC k casualty k cost ((A/km)/vehicle)) i vehicle class: i ¼ 1 for passenger car; i ¼ 2 for
cc k casualty k cost (A/casualty) utility; i ¼ 3 for heavy truck; i ¼ 4 for heavy bus
Cd i vehicle depreciation cost for vehicle i (A/km) IRI international roughness index (mm/km for PSI
Cf i fuel cost for vehicle i (A/km) calculation)
cf i fuel consumption for vehicle i (l/km) j accident class: j ¼ 1 for accident with slight injury;
Cm i maintenance cost for vehicle i (A/km) j ¼ 2 for accident with serious injury; j ¼ 3 for
Cmdt i total depreciation market price for vehicle i (less accident with fatalities
tyres) (A) k casualty class: k ¼ 1 for slight injury; k ¼ 2 for
Cmf i fuel market price (gasoline or diesel) (A/l) serious injury; k ¼ 3 for fatalities
Cmmt i total maintenance market price for vehicle i (A) kma i annual average kilometrage for vehicle i (km/year)

1
Transport Refinement of a simplified road-user cost
model
Batista dos Santos, de Picado Santos and
Pissarra Cavaleiro

Offprint provided courtesy of www.icevirtuallibrary.com

Author copy for personal use, not for distribution

L section length (km) However, in road management, several RUC models with differ-
LM&R maintenance and rehabilitation zone length (km) ent degrees of formulation and data requirements have been used
LPSI section length with a certain present serviceability with good results around the world. Between 2004 and 2007
index (PSI) value (km) some of the most important ones were analysed to provide the
m travel purpose: m ¼ 1 for travel in work time; m ¼ 2 conceptual basis for a new simplified RUC model developed in a
for travel in non-work time doctoral thesis (Santos, 2007) and tested on the Portuguese trunk
NAW national average wage ((A/h)/person) road network (Santos et al., 2011a, 2011b).
NC k number of casualties k (casualties/year)
nt i number of tyres for vehicle i The models considered in the definition of the conceptual frame-
OR i,m occupancy rate for vehicle i and travel purpose m work of the model were: the World Bank HDM-RUE – ‘Model-
(occupants/vehicle) ling road-user and environmental effects in HDM-4’ (Bennett and
P pavement patching area (m2 /100 m2 ) Greenwood, 2004); the New Zealand vehicle operating cost
PSI present serviceability index (0–5) model (NZVOC) (Transfund, 2003); the Cost Benefit Analysis –
pi vehicle proportion for class i and AADT considered COBA (DfT et al., 2006a); the Manual ‘Techniques for manually
R mean rut depth (mm) estimating road-user costs associated with construction projects’
RUC road-user cost ((A/km)/day) used in the Texas department of transportation (TxDOT) (Daniels
RUCM&R road-user cost in maintenance and rehabilitation et al., 1999); and the cost model integrated in the former
zones ((A/km)/day) pavement management system (early 1990s) of the Portuguese
RUCPSI incremental increase or decrease in RUC owing to road administration (designated, at the time, as JAE – Junta
PSI ((A/km)/day) Autónoma de Estradas) (GEPA, 1995).
RUCtotal total road-user cost ((A/km)/day)
S total pavement disintegrated area (with potholes and The review showed, as expected, that beyond the basic methodo-
ravelling) (m2 /100 m2 ) logical approaches, there are three fundamental components of
si average operating speed for vehicle i (km/h) RUC to be considered in an RUC model: vehicle operating costs,
sM&Ri average operating speed in M&R sections, for accident costs and time costs. In general terms, this relationship
vehicle i (km/h) can be expressed as
TC m time cost for travel purpose m ((A/h)/occupant)
Toll toll cost ((A/km)/day)
tsl i tyre service life for vehicle i (km) 1: RUC ¼ VOC þ AC þ VOT
VOC vehicle operating cost ((A/km)/day)
VOC i VOC for vehicle i (A/km)
VOT value of time ((A/km)/day) These three main components, as well as a component related to
VOTi value of time for vehicle i ((A/km)/vehicle) the tolling costs, were considered in the simplified RUC model
VOTM&Ri value of time in M&R sections, for vehicle i proposed in 2007. The model was developed aiming at simplicity,
((A/km)/vehicle) reduced data requirements (selected data are usually available),
vsl i vehicle i service life (years) easy calibration, easy application and trustworthy results, provid-
ing average RUC values.

1. Introduction The main objective of this study is the refinement of the

A sustainable pavement system requires a comprehensive evalua- simplified RUC model to fully integrate, in a sustained, simple
tion framework that takes into account environmental, economic but reliable way, not only the average RUC but also the costs
and social indicators. The road-user costs are included in the social related to the effect of work zones and pavement condition.
class of indicators and can be defined as the costs experienced by
users when travelling along a certain length of a road. These costs This improvement will enable a more realistic road cost analysis,
typically include costs related to the value of the travel time spent since these scenarios are not explicitly considered in the initial
by the driver and passengers, expenses of operating the vehicle and formulation.
costs of accidents. In some systems the costs of tolls and discomfort
are also considered. The sum of all or part of these costs (usually To sustain the refinements proposed, the paper presents
the most significant) constitutes the road-user costs (RUC).
(a) a brief description of the main activities that led to the
At this time the Portuguese Road Administration does not original model formulation and input data definition
consider the RUC in the evaluation process of road design, (b) an analysis to identify the critical parameters of the model
maintenance or rehabilitation and therefore estimation of road life (c) a study of the approaches taken by some international
cycle costs in Portugal, as in many other countries, does not formulations, models and manuals in use such as ASTM,
include this important aspect. HDM-4, TXDOT, NJDOT and TRB (ASTM, 1983; Bennett

2
Transport Refinement of a simplified road-user cost
model
Batista dos Santos, de Picado Santos and
Pissarra Cavaleiro

Offprint provided courtesy of www.icevirtuallibrary.com

Author copy for personal use, not for distribution

and Greenwood, 2004; Daniels et al., 1999; NJDOT, 2001; To validate the model, the Portuguese average values of the RUC
The World Bank, 2010; Zaniewski, 1983) were computed with the proposed and reference models and the
(d ) the results and discussion of a case of application. results were compared (Santos et al., 2011a). The values obtained
confirmed that the contributions of the vehicle operation and the
value of time costs in total RUC for passenger cars (66% for
2. Road-user costs model VOC and 34% for VOT) and heavy trucks (84% for VOC and
The simplified road-user costs model proposed in Santos (2007) 16% for VOT) are similar to those obtained in the reference
was developed taking into account several main aspects: the models. In terms of accident costs, no comparison study was
recognised conceptual principles; the application to trunk roads; made because they usually comprise characteristic costs and
the impact of each component on the total user’s costs; the information from the network or section in analysis, so values for
availability of Portuguese official information; and four vehicle different countries or even different regions of the same country
classes as representative of the total traffic composition (passen- may be diverse.
ger car (PC), utility (U), heavy truck (HT) and heavy bus
(HB)). The model was also tested with real data and for a real network.
The outcome was analysed by the network managers and accord-
The model framework adopted was essentially based on the ing to them the figures obtained were very acceptable.
simplifications of the HDM-4 equations for the VOC, on the
COBA and the HDM-4 approach for the AC, and on the JAE The complete formulation of the model and the input data for
model and the HDM-4 equations for the VOT definition. Portuguese conditions, which permit the calculation of average
values of the RUC, can be found in Appendix 1.
The results of all these considerations led to a model with the
three main cost components identified above: the VOC, including 3. Sensitive parameters of the model
costs for fuel, tyres, vehicle preventive maintenance and deprecia- Several variability studies were carried out to identify the
tion; the AC, considering costs for accident, police and medical sensitive parameters of the model. Tables 1 and 2 present some
assistance for accident type and casualty costs (fatalities, serious results that show that the proposed model, as most of the existing
and slight injuries); and the VOT for work and non-work travel. ones, is mainly sensitive to changes in the average operating
Eventually, when appropriate, a component related to tolling costs speed defined for each class of vehicle and type of road and to
may also be added. consumption and fuel cost.

The input values for the Portuguese average situation were Besides being identified as critical parameters, speed and fuel
defined for 2006, the costs being updated in 2010. This definition consumption and cost are also the main parameters in the
took into account the values used and recommended by the definition of the vehicle operating cost and the value of time, and
existing methodologies and, in particular, the values obtained those that best reflect the main changes in the RUC due to
from the Portuguese Road Haulage Association, companies and pavement condition and maintenance actions in the network
official bodies such as ANTRAM (National Association of (work zones).
Transportation of Goods), ANTROP (National Association of
Transportation of Passengers), INE (National Institute of Statis- Thus, a careful consideration of these parameters in the RUC
tics), the police and emergency services. calculations and the definition of correction factors in the values

Road type PC HT

Operating speed: 2/3 of operating ˜VOT: % Operating speed: 2/3 of operating ˜VOT: %
km/h speed: km/h km/h speed: km/h

EN, ER 70 46.7 +50 50 33.3 +50

IC 80 53.3 +50 60 40.0 +50
IP 90 60.0 +50 80 53.3 +50
AE 120 80.0 +50 100 66.7 +50

PC, private cars; HT, heavy trucks; VOT, value of time; EN/ER, national and regional roads with two lanes (one in each direction) and ‘medium’
design standards; IP and IC, main roads (principal and complementary roads) with two lanes (one in each direction) and ‘high’ design standards;
AE, motorways with at least four lanes (two in each direction), median and ‘high’ design standards.

Table 1. Variability study for operating speed

3
Transport Refinement of a simplified road-user cost
model
Batista dos Santos, de Picado Santos and
Pissarra Cavaleiro

Offprint provided courtesy of www.icevirtuallibrary.com

Author copy for personal use, not for distribution

VOC component Parameters ˜ ˜VOC: %

PC HT

Fuel +20% cf +8.0 +15.6

Tyre +25% tsl 0.9 0.9
25% tsl +1.4 +1.5
Preventive maintenance +20% vsl +0.07 +0.01
+25% kma 0.9 0.04
25% kma +1.6 +0.06
Depreciation +20% vsl 26.0 1.3
+25% kma 27.8 2.7
25% kma +9.0 +4.4

VOC, vehicle operating cost; PC, private cars; HT, heavy trucks; cf, fuel consumption (l/km);
tsl, tyre service life (km); vsl, vehicle service life (years); kma, annual average kilometres
(km/year). The numbers in bold represent the most significant variations.

Table 2. Variability study for vehicle operating parameters

defined for an average situation are essential to forecast additional influence on RUC values in work zones is changes in operating
RUC in sections where maintenance actions are planned, or to speeds. For the refinement of the RUC model formulation, these
compute the benefits associated with a better pavement condition. changes and the consequent additional travel time (delay) were
incorporated by the consideration of work zone length, duration
4. Additional RUC due to work zones and of interventions and work zone posted speed limit. This last
pavement condition depends on the timing of restrictions (hours of the day and days
Additional RUC due to maintenance intervention periods (work of the week) and the legal framework of each country, which
zones) and changes in pavement condition can be included in the normally applies lower posted speed limits at night. The values
proposed RUC formulation by considering specific parameter adopted for work zone speed (constant values by road class and
values defined for a certain maintenance strategy or for a time with restriction) also reflect the operating characteristics of
particular pavement quality index. the traffic affected and configuration of the work zones.

4.1 Work zones Regarding fuel consumption, data collected in several studies,
The main parameters that can lead to additional RUC in work empirical models developed from this information (which usually
zones have been identified in several models and manuals in use, relate fuel consumption to the operating speed of vehicles) and
such as HDM-4, TXDOT and NJDOT (Bennett and Greenwood, mechanistic models (which relate fuel consumption to the forces
2004; Daniels et al., 1999; NJDOT, 2001), as being the decrease opposing motion, allowing application under different conditions)
of operating speed leading to traffic delays, which increases the show that maximum fuel consumption occurs for low and high
VOT, and the consequent additional fuel consumption associated speeds, and minimal fuel consumption for speeds of 40–60 km/h
with traffic congestion, increasing the VOC values. (Bennett and Greenwood, 2004; DfT et al., 2006a) (see Figure 1).

In the 2007 model, the speed values were defined for each type Taking into account the patterns of fuel consumption presented in
of road based on posted speed limit and typical Portuguese values Figure 1 and considering the Portuguese legal framework for
(for each type of vehicle), and were considered constants. For trunk roads with high posted speed limits (100–120 km/h and at
fuel consumption, average consumption by type of vehicle was least two lanes in each direction), which limits the private road
adopted. concessionaires to guarantee maximum operating speeds greater
than or equal to 2/3 of the posted limit in work zones (up to
Work zone additional accident costs are also considered in some 10 km per set) in the daytime (7.00 a.m.–9.00 p.m.), in this
approaches, such as Quadro (DfT et al., 2006b), by comparing period and situation there is actually a decrease in fuel consump-
work zone accident rates with those for normal conditions; tion.
however, rates in work zones are not commonly available, so they
are not considered in the proposed simplified model. Lower speeds, up to 1/3 of the normal posted speed limit, are
allowed in work zones during the nighttime, when the volume of
Nonetheless, from the parameters identified, the most significant traffic is usually low.

4
Transport Refinement of a simplified road-user cost
model
Batista dos Santos, de Picado Santos and
Pissarra Cavaleiro

Offprint provided courtesy of www.icevirtuallibrary.com

Author copy for personal use, not for distribution

Fuel IRC empirical 1993

in fuel consumption models simulations such as those presented
16 Fuel COBA empirical 2002 above.
Fuel consumption: litres/100 km

14 Fuel HDM mechanical Portuguese

conditions 2006 When traffic diversions are needed, changes in operating costs
12
and travel time should be considered in the same way as
10 described above.
8
6 The refinements proposed to consider the effects of the works
zones programmed in the RUC calculations are presented in
4
Equations 2–5.
2
0 2: RUCM&R ¼ dCf þ dVOT
0 20 40 60 80 100 120 140 160
Vehicle speed: km/h
(a)
X
4
16 dCf ¼ AADT (0:2 Cf i pi )
i¼1
14
1
for sM&Ri < si and ER, EN
Fuel consumption: litres/100 km

3: 3
12

10
X
4
8 dVOT ¼ AADT ðVOTM&Ri pi Þ  VOT
4: i¼1

4 Fuel IRC empirical 1993 X

2
1
VOTM&Ri ¼ ðTCm ORi,m Þ
2
Fuel COBA empirical 2002 5: sM&Ri m¼1
Fuel HDM mechanical Portuguese
conditions 2006
0
0 20 40 60 80 100 120 140 160 This refinement was tested in order to evaluate, for several work
Vehicle speed: km/h
(b)
zone scenarios and Portuguese road classes, the influence of work
zones on unit RUC values (A/km), by vehicle (Teixeira, 2011).
Figure 1. Fuel consumption against vehicle speed for: The road classes considered were motorways (with at least four
(a) passenger cars; and (b) heavy trucks lanes, median and ‘high’ design standards; AE), principal (IP)
and complementary (IC) main roads (with two lanes and ‘high’
design standards) and national (EN) and regional (ER) roads
(with two lanes and ‘medium’ design standards).
Taking into account the scenarios described, additional RUC in
work zones due to changes in fuel consumption were considered Some results can be observed in Figures 2 and 3. In these figures
for Portuguese national and regional roads, and eventually in the baseline scenario (range of 0%) represents the average unit
main roads with two lanes, operating at lower speeds (up to 1/3 values of RUC for each vehicle class calculated with the initial
of the posted speed limit). In such cases, in the daytime, there is model formulation, and includes vehicle operating, travel time
a high probability of frequent stops, resulting in an increased fuel and toll costs.
consumption associated with the movement at very low speeds.
For that reason, in this scenario, an increase of 20% in fuel Through combined analysis of the charts and results and consid-
consumption was considered. ering a typical traffic distribution of 80% for PC, 10% for U,
9% for HT and 1% for HB (based on traffic data provided by
The choice of this additional consumption value is based on road concessionaires and Portuguese average values), it can be
representative PC vehicle manufacturer information that points to found that reducing the speed to 2/3 of the normal operation
urban fuel consumption values 20–30% higher than the combined speed without additional fuel consumption, during the daytime,
values (used for the definition of an average situation) and, for an led to an additional cost of around 10% for motorways and 17–
HT, the additional values of 30–40% that are commonly obtained 20% for the remaining network. For a speed reduction to 1/3 of

5
Transport Refinement of a simplified road-user cost
model
Batista dos Santos, de Picado Santos and
Pissarra Cavaleiro

Offprint provided courtesy of www.icevirtuallibrary.com

Author copy for personal use, not for distribution

80·0
70·0 (III)
60·0 (IV)
50·0 (III) (III)
40·0
ΔRUC: %

30·0 (IV)
(IV) (III) (I)
20·0 (I)
(I)
10·0 (I) (II)
(II) (II) (II) (IV)
0
⫺10·0
⫺20·0
⫺30·0 PC U HT HB
Day: 2/3 speed, without additional fuel consumption, with toll (I)

Day: 2/3 speed, without additional fuel consumption, without toll (II)

Night: 1/3 speed, without additional fuel consumption, with toll (III)

Night: 1/3 speed, without additional fuel consumption, without toll (IV)

Figure 2. RUC variations on motorways for different scenarios

and vehicle classes (2010 unit values)

100·0 (II)
90·0 (II)
80·0 (II)
70·0
ΔRUC: %

60·0
50·0 (II)
40·0
30·0
(I) (I)
20·0 (I)

10·0 (I)

0
PC U HT HB

Day: 2/3 speed, without additional consumption of fuel (I)

Night: 1/3 speed, additional consumption of fuel (II)

Figure 3. RUC variations on principal main roads for different

scenarios and vehicle classes (2010 unit values)

the normal operation speed, without additional fuel consumption 4.2 Pavement condition
on motorways but considering an additional fuel consumption of Changes in operating speed and the consequent additional travel
20% on the remaining network, at nighttime, the additional user time due to the pavement conditions can be incorporated into the
cost rises to 42% (motorways) and 74–85% (for the remaining proposed model formulation through consideration of the section
network). A combined analysis considering that road works take length within a certain pavement quality index and lower operat-
place during the day and night periods, with a day/night ing speeds.
distribution of traffic equal to 85%/15%, leads to additional
costs of about 15% for highways and 25–30% for the remaining However, the pavement conditions of the most important net-
network. works, such as the national ones, do not in general reach a level

6
Transport Refinement of a simplified road-user cost
model
Batista dos Santos, de Picado Santos and
Pissarra Cavaleiro

Offprint provided courtesy of www.icevirtuallibrary.com

Author copy for personal use, not for distribution

of degradation that significantly influences the normal operating PSI IRI: m/km Correction factors for VOC
speed. On the other hand, it is also known that pavements in good
condition allow vehicles to maintain higher speeds with greater 0 4.25 1.15
comfort and security, reducing travel time and accident costs. 2.0 3.50 1.05
They also allow reductions in operating costs in terms of tyres, 3.5 2.00 1.00
maintenance and depreciation of the vehicle, but not necessarily 4.7 0.50 0.95
in fuel. The converse occurs for pavements in poor conditions. 5.0 – 0.95

To consider these scenarios, the condition of the pavements was PSI, present serviceability index; IRI, international roughness index;
VOC, vehicle operating cost.
integrated into the initial RUC formulation only in the vehicle
operating costs calculations through a quality index representing
Table 3. VOC correction factors, PSI and IRI for Portuguese model
the functional and structural state of the pavements. The index
calibration
adopted was the ‘present serviceability index’ (PSI), which ranges
from 0 (for a pavement in a poor state) to 5 (for a new
pavement).
current average state of Portuguese trunk road pavements, which
Equation 6 presents the PSI formulation adopted in the refine- according to the latest available data presents a PSI of about 3.5
ment of the model. This equation was developed by Picado- (Trindade and Horta, 2009), and the average vehicle operating
Santos et al. (2006) for the Portuguese trunk roads in order to cost obtained by Santos (2007) for the same period. This scenario
reflect the condition of the national road network and was used constitutes the baseline with a correction factor of 1.
by the Portuguese Road Administration during the first decade of
this century. The formulation is derived from the version used in Given that the Portuguese road administration adopts a PSI equal
the pavement management system of the State of Nevada to or less than 2 as an indication of the need to intervene in the
(Sebaaly et al., 1996) and that developed during the AASHO pavement quality of the network, a correction factor of 1.05 was
road test (HRB, 1962). Similar equations developed for and defined for this scenario (based on the approaches studied). For
representing other realities, can be incorporated in the model as extreme scenarios a gain of 5% was considered for new
well. pavements and an increase of 15% for badly degraded pavements.

:
PSI ¼ 5e0 0002598IRI=2  0:002139R2 The analysis of the existing approaches and the characteristic
Portuguese values presented in Table 3 permitted the definition of
0:5
6:  0:03ðC þ S þ PÞ a new Portuguese VOC–PSI relation and thus the consideration
of this additional cost in determining the RUC, through Equations
7 and 8.
The changes in VOC as a function of PSI (or IRI, the
international roughness index) have been treated in several stud- 7: RUCPSI ¼ VOC ðF VOC,PSI  1Þ
ies by applying the regression analysis to real data, resulting in
several formulations such as those presented by HDM-4 (The
World Bank, 2010), TRB (Zaniewski, 1983), ASTM (ASTM,
1983) and by Picado-Santos et al. for Portuguese conditions F VOC,PSI ¼  0:0006 PSI3 þ 0:0072 PSI2
(Picado-Santos et al., 2006).
8:  0:0612 PSI þ 1:1498
The study of these formulations, as well as an analysis of recent
data on Portuguese trunk road pavements condition (Picado-
Santos and Pereira, 2009) and average user cost (Santos, 2007), Figure 4 shows the curves obtained from the VOC correction
was used to develop a mathematical model that reflects the factors recommended by TRB-ASTM, the Portuguese model
change of VOC as a function of PSI for Portuguese conditions. (UC-UM) (Picado-Santos et al., 2006) and the proposed model
Table 3 shows the set of correction factors that support the (RUC PT).
proposed formulation (for PSI and IRI).
The aggregated correction factors presented in Figure 4 for the
The range of PSI and IRI values was chosen to ensure that the TRB-ASTM approach were obtained by applying the Portuguese
most extreme conceivable circumstances were examined as well VOC component distribution (for fuel, engine oil, tyres, main-
as the usual expected pavement conditions in operation. The range tenance and repair, and depreciation cost) obtained from data
examined was beyond values normally expected in Portugal. collected in 2006 (Santos, 2007) to the disaggregated correction
factors recommended in the TRB-ASTM study. A similar proce-
In the range of values, ‘reference’ situation corresponds to the dure was adopted for the distribution of traffic.

7
Transport Refinement of a simplified road-user cost
model
Batista dos Santos, de Picado Santos and
Pissarra Cavaleiro

Offprint provided courtesy of www.icevirtuallibrary.com

Author copy for personal use, not for distribution
VOC correction factor: %

1·3 JAE90 (aggregate) HDM-4 calibrate (aggregate)

VOC correction factor: %

1·2 1·3 HDM-4 default RUC_PT (aggregate)
(aggregate)
1·1 1·2

1·0 1·1
UC_UM (aggregate)
0·9 RUC_PT (aggregate) 1·0
TRB_ASTM (aggregate)
0·8 0·9
0 1 2 3 4 5
0·8
PSI
0 1 2 3 4 5
IRI: m/km
Figure 4. VOC correction factors for different PSI values and
models Figure 5. VOC correction factors for different IRI values and
models

From analysis of Figure 4 it can be concluded that the trend of

the proposed mathematical model is similar to those others
studied, even if the correction factors for the worst pavements are degradation (such as rutting, cracking, potholes, ravelling and
smaller. This can be justified by considering technological ad- patching).
vances in vehicles and the type of road network data used (trunk
road), where low PSI values are not common. Some results obtained by applying the proposed refinement to
motorways and principal main roads are shown in Figures 6 and
Analyses of changes in VOC with IRI were also carried out, since 7, where the RUC variation due to the pavement condition (PSI)
most of the existing formulations use IRI as the main parameter is compared with the average values obtained from the initial
for the computed PSI and VOC variation in trunk road networks. model formulation. The average RUC constitutes the reference
This analysis is presented in Figure 5, where it is possible to scenario and corresponds to a variation of 0%.
observe the curves obtained from HDM-4 and JAE RUC values
when applying the models to typical Portuguese VOC component From the analysis of Figures 6 and 7 it is possible to observe that
and traffic distribution. A curve representing the quadratic equa- for the PSI indicative of need for intervention on the quality of
tion obtained from the IRI values considered for Portuguese pavements (PSI ¼ 2), the additional cost of RUC is about 3% (for
calibration is also included. all vehicle classes).

For trunk roads and current values of IRI (up to 3.5 m/km), most 5. Model applications
of the models tested produce similar results. Higher values of The RUC formulation and input model values proposed were
IRI, such as those considered in HDM-4, are not frequent in applied to two Portuguese motorway networks under concession
Portuguese motorway networks under concession. with good results: Scutvias (A23) and Aenor (A7, A11).

When interpreting Figures 4 and 5, it is necessary to be aware 5.1 Presentation of results

that the index adopted in the proposed model for the definition of Tables 4 and 5 include the data provided by the private road
the VOC correction factor is PSI, in which the calculation concessionaire Scutvias, in order to perform the RUC calcula-
depends not only on IRI but also on the pavement’s superficial tions, and some application results.

12·0
10·0
8·0
6·0
ΔRUC: %

4·0
2·0
0
⫺2·0
⫺4·0
PC U HT HB
⫺6·0
PSI ⫽ 0 PSI ⫽ 2 PSI ⫽ 3·5 PSI ⫽ 4·7 PSI ⫽ 5

Figure 6. RUC variations on motorways for different scenarios of

pavement condition and vehicle classes (2010 unit values)

8
Transport Refinement of a simplified road-user cost
model
Batista dos Santos, de Picado Santos and
Pissarra Cavaleiro

Offprint provided courtesy of www.icevirtuallibrary.com

Author copy for personal use, not for distribution

14·0
12·0
10·0
8·0
ΔRUC: %

6·0
4·0
2·0
0
⫺2·0
⫺4·0
⫺6·0 PC U HT HB

PSI ⫽ 0 PSI ⫽ 2 PSI ⫽ 3·5 PSI ⫽ 4·7 PSI ⫽ 5

Figure 7. RUC variations on principal main roads for different

scenarios of pavement condition and vehicle classes (2010 unit values)

5.2 Discussion of results

Data Scutvias (A23) 2010
For average RUC values and comparing the results obtained from
Network length (km) 177.5 the reference models (Bennett and Greenwood, 2004; DfT et al.,
Total AADT 10 574 2006a) and the proposed model (Santos, 2007), the main role of
pi a PC 0.7972 the vehicle operation costs (approximately 60%) and the value of
U 0.0648 time (approximately 20%) in total RUC is confirmed. Therefore,
HT 0.1289 knowing that the VOC is mostly influenced by pavement condi-
HB 0.0091 tion and the VOT by maintenance interventions (on a temporary
Accidents With slight injury 59 basis), a decrease in RUC can be expected when a pavement care
With serious injury 10 maintenance programme is applied, and at the same time a
With fatalities 2 reduction in the number of accidents.
Casualties Slight injury 88
Serious injury 15 It can be seen that toll costs also make a significant contribution
Fatalities 2 to RUC in a proportion similar to the value of time and that,
Approximate toll cost (A/km) 0.20 (virtual tollb ) despite the small contribution of accident costs in the results
obtained for the networks analysed, this component will be more
a
Information processed from the traffic data provided by the road significant in low to medium design standard roads. For that
concessionaire. reason they must also be considered in the calculations.
b
The approximate toll cost values provided by Scutvias correspond to
a uniform rate for all vehicle classes for 2006.
PC, private cars; U, utility vehicles; HT, heavy trucks; HB, heavy buses. To validate the refinements proposed, the model was tested in a
maintenance (work zone) and pavement condition scenario, for a
Table 4. Data provided by road concessionaire Scutvias for 2010 unit section 1 km long, a speed reduction to 2/3 of the posted limit
(to 80 km/h), PSI equal to 2.0 and without deviations, according to

Scutvias (A23) Scutvias (A23)

Average values Work zone (1 km) PSI ¼ 2.0 m/km

Costs RUC: (A/km)/day RUC: % RUC: (A/km)/day RUC: %

VOC 2613.90 58.41 2743.55 54.70

AC 90.72 2.03 90.72 1.81
VOT 822.30 18.37 1233.46 24.59
Toll 948.28 21.19 948.28 18.91
RUC 4475.21 100.00 5016.01 +12.08

The value in bold represents the additional RUC.

Table 5. Portuguese RUC model application results (2010 values)

9
Transport Refinement of a simplified road-user cost
model
Batista dos Santos, de Picado Santos and
Pissarra Cavaleiro

Offprint provided courtesy of www.icevirtuallibrary.com

Author copy for personal use, not for distribution

the Portuguese legal framework described above for a main road user costs associated with work zones may increase significantly
network with work zones operating during the daytime. and are higher than those due to pavement condition. The work
zone additional costs depend mainly on day/night operating speed
The total RUC obtained from this scenario considers the delay and traffic distribution and pavement condition must be main-
costs (VOT changes) and additional non-fuel components cost, tained above PSI ¼ 2 to minimise additional RUC.
resulting in an increase of 12% compared with the average values
of RUC. This additional cost can be disaggregated into about 9% Although the additional user cost due to work zones is established
due to changes in VOT and 3% due to the pavement condition as the most significant factor, it is recommended that the effect of
and demonstrates the importance of taking work zones and pavement condition also be incorporated in the analysis. The
pavement condition into account in RUC calculations. inclusion of this effect is justified in order to identify, from a user
cost perspective, the best time to perform the maintenance and
6. Conclusion rehabilitation works to improve the pavement condition.
The fact that user costs do not debit agency budgets as agency
costs do, combined with uncertainty regarding some values (value As a result, a simplified but trustworthy model such as the one
of time, effects of agency activities on accident rates), may developed for the Portuguese trunk road network can help
incline transportation decision makers to give less credence to decision makers to include user costs in life-cycle cost analysis
user costs than to their own agency cost, restricting their ability (LCCA). This will permit a more accurate technical and econom-
to find the lowest total cost solutions. ic planning of the maintenance and operation actions and thereby
achieve optimal solutions.
Conversely many road-user cost models with strong concep-
tual frameworks have been developed in the past and are in
Appendix 1
use. Nevertheless, many countries or regions that wish to
consider RUC in their road life-cycle costs analysis lack the A1.1 Road-user cost model
means to obtain and update all the data required for these A1.1.1 Model formulation
models.
RUCtotal ¼ ðRUC 3 LÞ þ ðRUCM&R 3 LM&R Þ
The simplified road-user cost model proposed solves the problem
of the amount of information needed to calibrate models such as 9: þ ðRUCPSI 3 LPSI Þ
HDM-4, and allows easy application for different scenarios. With
the refinements proposed to include the effect of pavement
condition and work zones, the average user costs as well as those
associated with maintenance and rehabilitation intervention peri- 10: RUC ¼ VOC þ AC þ VOT þ Toll
ods can be considered over the life of the infrastructure using
simple models.

The pavement condition was included by defining correction X

4

factors to be applied to the vehicle operating costs (non-fuel VOC ¼ AADT ðVOCi pi Þ
11: i¼1
components). These factors reflect a variation of the average
RUC between 3% (for PSI ¼ 5) and 9% (for PSI ¼ 0), with an
additional cost of about 3% for PSI ¼ 2 (level for the need of
intervention in the pavement). 0 1
X
3 X
3
AC ¼ AADT @ ACj þ CCk A
The effect of work zones was incorporated by setting lower 12: j¼1 k¼1
average speeds along the work sections (2/3 of the normal speed
for work ongoing during the daytime and 1/3 for nighttime) and
an additional fuel consumption of 20% for roads with ‘medium’
design standards operating at lower speeds (up to 1/3 of the X
4
normal posted speed limit). When applied to the Portuguese road VOT ¼ AADT ðVOTi pi Þ
network, for daytime works, these considerations result in an 13: i¼1
additional cost of approximately 10% for motorways and 17–
20% for a two-lane type of network. For nighttime works, an
additional cost of 42% for motorways and 74–85% for a two-lane
type of network was obtained. X
4
Toll ¼ AADT ðctolli pi Þ
14: i¼1
The scenarios tested show that in most cases of intervention the

10
Transport Refinement of a simplified road-user cost
model
Batista dos Santos, de Picado Santos and
Pissarra Cavaleiro

Offprint provided courtesy of www.icevirtuallibrary.com

Author copy for personal use, not for distribution

Considered by vehicle class 17: Cti ¼ ðnti Cmti Þ=tsli

15: VOCi ¼ Cf i þ Cti þ Cmi þ Cdi

18: Cmi ¼ Cmmti =ðvsli kmai Þ

16: Cf i ¼ cf i Cmf i 19: Cdi ¼ Cmdti =ðvsli kmai Þ

Data PC HT

Representative vehicle Car medium Truck articulated

Average operating speed (km/h) EN/ER IC IP AE EN/ER IC IP AE

70 80 90 120 50 60 80 100
Vehicle service life (year) 10 12
Annual average kilometrage (km/year) 20 000 85 000
Occupancy rate (occupants/vehicle) 2 (1 work driver + 1 non-work 1 (1 work driver)
passenger)

PC, private cars; HT, heavy trucks; EN/ER, national and regional roads with two lanes (one in each direction) and ‘medium’ design standards; IP
and IC, main roads (principal and complementary roads) with two lanes (one in each direction) and ‘high’ design standards; AE, motorways with
at least four lanes (two in each direction), median and ‘high’ design standards.

Table 6. Passenger car and heavy truck general input data values

Data PC HT

Fuel Gasoline: 5.9 litres/100 km Diesel: 44.0 litres/100 km

Diesel: 4.8 litres/100 km
Market price (31/12/2010): Market price (31/12/2010):
Gasoline 95: A1.485/litre Diesel: A1.270/litre
Diesel: A1.270/litre
Tyres nt ¼ 4 tyres/vehicle nt ¼ 12 tyres/vehicle
tsl ¼ 40 000 km tsl ¼ 200 000 km
Market price: A75/tyre Market price: A490/tyre
Preventive maintenance A1626/10 years A28 920/12 years
Depreciation A17 720/10 years A87 135/12 years
Value of time A6.93/h (work time) A9.61/h (work time)

Accident costs: A/accident Accident type With light injuries With serious injuries With fatalities
Police assistance 57.5 160.0 250.0
Medical assistance 18.0 103.50 103.5

Casualties costs (A/casualty) Light injuries: A40 000/casualty

Serious injuries: A90 000/casualty
Fatalities: A500 000/casualty

Toll cost A0.075/km A0.185/km

PC, passenger cars; HT, heavy trucks; nt, number of tyres per vehicle; tsl, tyre service life.

Table 7. Passenger car and heavy truck input data costs (2010
values)

11
Transport Refinement of a simplified road-user cost
model
Batista dos Santos, de Picado Santos and
Pissarra Cavaleiro

Offprint provided courtesy of www.icevirtuallibrary.com

Author copy for personal use, not for distribution

1X 2 Regional Development/Northern Ireland Economic

VOTi ¼ ðTCm ORi,m Þ Assessment of Road Schemes) (2006a) The COBA Manual.
20: si m¼1
In Design Manual for Roads and Bridges, vol. 13.
Department for Transport, London, UK.
DfT et al. (Department for Transport, Scottish Executive
21: TCm¼1 ¼ NAW Development Department, Welsh Assembly Government/
Llywodraeth Cynulliad Cymru and The Department for
Regional Development/Northern Ireland Economic
Assessment of Road Schemes) (2006b) The QUADRO
22: TCm¼2 ¼ 0:25 NAW Manual. In Design Manual for Roads and Bridges, vol. 14.
Department for Transport, London, UK.
GEPA (Gestão de Pavimentos) (1995) Sistema de Gestão da
And for the set of all vehicle classes (without vehicle class Conservação. Sistema de Custos dos Utentes (Maintenance
disaggregation) Management System. Road User Costs System). Junta
Autónoma de Estradas, Lisbon, Portugal, Contrato 4915 (in
Portuguese).
23: ACj ¼ ARj acj
HRB (Highway Research Board) (1962) The AASHO Road Test,
Report 5, Pavement Research. Highway Research Board
Special Report 61. Transportation Research Board,
X
3 Washington, DC, USA.
CCk ¼ ðANCk cck Þ ARj NJDOT (New Jersey Department of Transportation) (2001) Road
24: j¼1 User Cost Manual. Department of Transportation, Trenton,
NJ, USA.
Picado-Santos L and Pereira P (2009) Relatório de
Caracterização do Estado do Pavimento, IC1 – Costa de
25: EA ¼ 365 AADT L Prata (Pavement Characterization Report, IC1 – Costa de
Prata). Report AENOR No. 1. ACIV, DCE, University of
Coimbra, Coimbra, Portugal (restricted access imposed by
26: ARj ¼ ANAj =EA contractor) (in Portuguese).
Picado-Santos L, Ferreira A and Pereira P (2006) Estruturação de
um Sistema de Gestão de Pavimentos para uma Rede
Rodoviária de Carácter Nacional (Structuring a pavement
27: ANCk ¼ NCk =ANA management system for a national road network). CEC –
Revista Engenharia Civil (26): 45–59 (in Portuguese).
Santos B (2007) Modelação dos Custos dos Utentes na Gestão da
Estrada (Modelling Road User Costs in Road Management).
A1.1.2 Input data for Portuguese conditions PhD thesis, University of Beira Interior, Covihã, Portugal (in
Tables 6 and 7 present the input data defined for passenger cars Portuguese).
and heavy trucks. A similar process was used for the utility and Santos B, Picado-Santos L and Cavaleiro V (2011a) A simplified
heavy bus values. road user costs model for Portuguese highways.
Transportation Research Record 2225: 3–10.
REFERENCES Santos B, Picado-Santos L and Cavaleiro V (2011b) Vehicle
ASTM (1983) ASTM-STP 884: Measuring road roughness and its operating, accident and user time costs in pavement
effects on user costs and comfort. ASTM International, West management systems: approach for Portuguese conditions.
Conshohocken, PA, USA, pp. 127–142. Journal of Civil Engineering and Architecture 5(8): 723–731.
Bennett C and Greenwood I (2004) Modelling Road User and Sebaaly P, Hand A, Epps J and Bosch C (1996) Nevada’s approach
Environmental Effects in HDM-4. PIARC, Paris, France, vol. to pavement management. Transportation Research Record
7, The Highway Development and Management Series. 1524: 109–117.
Daniels G, Ellis D and Stockton W (1999) Techniques for Teixeira J (2011) Análise do Efeito dos Troços em Obras nos
Manually Estimating Road User Costs Associated with Custos Suportados pelos Utentes das Estradas (Analysis of
Construction Projects. Texas A&M Transportation Institute, the Effects of Work Zones on Road User Costs). Master’s
College Station, TX, USA. thesis, University of Beira Interior, Covihã, Portugal (in
DfT et al. (Department for Transport, Scottish Executive Portuguese).
Development Department, Welsh Assembly Government/ The World Bank (2010) HDM-4 Road User Costs Model, version
Llywodraeth Cynulliad Cymru and The Department for 2.00. The World Bank, Washington, DC, USA.

12
Transport Refinement of a simplified road-user cost
model
Batista dos Santos, de Picado Santos and
Pissarra Cavaleiro

Offprint provided courtesy of www.icevirtuallibrary.com

Author copy for personal use, not for distribution

Transfund (2003) Using the NZVOC Model to Prepare PEM S. A.). Proceedings of the XV CILA – Congresso Ibero-
Vehicle Operating Costs. Data Collection, Motueka, New Latino Americano do Asfalto, Lisbon, Portugal (in
Zealand. Portuguese).
Trindade M and Horta CO (2009) Sistema de gestão de Zaniewski J (1983) Fuel Consumption Related to Roadway
pavimentos da EP – estradas de Portugal, S. A. (The Characteristics (Discussion and Closure). Transportation
pavement management system of the EP – roads of Portugal, Research Board, Washington, DC, USA.

WHAT DO YOU THINK?

To discuss this paper, please email up to 500 words to the
editor at [email protected]. Your contribution will be
forwarded to the author(s) for a reply and, if considered
appropriate by the editorial panel, will be published as a
discussion in a future issue of the journal.
Proceedings journals rely entirely on contributions sent in
by civil engineering professionals, academics and students.
Papers should be 2000–5000 words long (briefing papers
should be 1000–2000 words long), with adequate illustra-
tions and references. You can submit your paper online via
www.icevirtuallibrary.com/content/journals, where you
will also find detailed author guidelines.

13