Sustainability 11 03128 PDF

sustainability
Article
Evaluating Impacts of Overweight in Road Freight
Transportation: A Case Study in Brazil with
System Dynamics
Verônica Ghisolfi 1 , Glaydston Mattos Ribeiro 1 , Gisele de Lorena Diniz Chaves 2, * ,
Rômulo Dante Orrico Filho 1 , Ivone Catarina Simões Hoffmann 3 and Leonardo Roberto Perim 3
1 Transportation Engineering Program, Federal University of Rio de Janeiro—UFRJ, Centro de Tecnologia,
Bloco H, Cidade Universitária, Rio de Janeiro 21949-900, Brazil; ghisolfi@pet.coppe.ufrj.br (V.G.);
glaydston@pet.coppe.ufrj.br (G.M.R.); romulo@pet.coppe.ufrj.br (R.D.O.F.)
2 Department of Engineering and Technology, Federal University of Espírito Santo—UFES, Rodovia BR 101
Norte, km 60, São Mateus ES 29.932-540, Brazil
3 National Department of Transportation Infrastructure—DNIT, Brasília 70040-902, Brazil;
ivone.hoffmann@dnit.gov.br (I.C.S.H.); leonardo.perim@dnit.gov.br (L.R.P.)
* Correspondence: gisele.chaves@ufes.br; Tel.: +55-27-3312-1572

Received: 3 April 2019; Accepted: 16 May 2019; Published: 3 June 2019
Abstract: Overweight vehicles in ornamental stone transportation reduce the lifespan of the pavement
and increase accidents on Brazilian highways. This paper evaluates the relationship between different
policies for the loading of ornamental stone vehicles and its externalities, such as the impact on
costs with transportation, pavement maintenance and road accidents, using the System Dynamics
(SD) method. The results confirm the advantages of reducing operational costs with the practice of
overweight. However, as the overweight increases, the pavement maintenance and road accidents
costs also increase. Therefore, the best vehicle loading policy depends on the relative importance of
the economic and social costs involved.
Keywords: freight transportation; overweight vehicles; simulation; social impact; sustainability
1. Introduction
The sustainable growth of a nation depends on efficient transportation systems. Many authors
describe the positive relationship between the economic development of a country and its transportation
infrastructure [1–4]. The public investment in transportation infrastructure causes a positive effect on
long-term economic performance [5].
Brazil has difficulties in its road infrastructure that are caused by the imbalance between use and
investments, which affects the country in rankings such as logistics performance, as published by the
World Bank. In the edition of 2018, Brazil ranked 56th overall and 51st in the category of Infrastructure [6].
Still, in this category, the country ranked 73rd among 137 countries assessed by the World Economic
Forum, which considers infrastructure as one of the pillars for global competitiveness [7].
Nevertheless, if on one hand the wide offer of transportation infrastructure is beneficial for
economic development, on the other hand the increased demand for trips led to traffic jams, pollution
and accidents, the social costs of which are unevenly distributed throughout the society. Freight
transportation policies, specifically, impact a country’s economy, society and environment [8]. National
economic growth depends on the activity of freight transportation and its infrastructure enables and
supports economic growth [1]. Regarding the social and environmental areas, however, many impacts
may result from the development of infrastructure and of product transportation operations and these
impacts must be taken into consideration in the strategic planning of this sector [9].
Sustainability 2019, 11, 3128; doi:10.3390/su11113128 www.mdpi.com/journal/sustainability

Sustainability 2019, 11, 3128 2 of 35
One of the main problems of road freight transportation is overweight vehicles, which leads to
the poor conditions found on Brazilian highways, to a reduction in the life span of vehicles, to the
increase in fuel consumption and to a higher risk of accidents. The consequences of these problems are
perceived by the degradation of road infrastructure in terms of pavement, by the high operational
cost of transportation and by the decreased quality of the service being provided. Furthermore,
overweight has been identified as one of the main causes of accidents involving trucks in Brazil [10].
A particular example of this problem in Brazil is that of ornamental stones transportation in the state of
Espírito Santo, due to the gravity of accidents caused by the tipping of trucks loaded with marble and
granite blocks.
The interest in assessing the social costs and the externalities of transportation, especially road
transportation, has grown due to the need of more effective policies to control the negative effects
associated with the problem of overweight vehicle [11]. Despite all the existing legal regulation in
Brazil, there is still a high number of overweight vehicles traveling. The National Department of
Transportation Infrastructure (DNIT) showed that 77% of carrier vehicles traveled overweight and that
10% of excess load per axle may reduce the predicted life span of a pavement in up to 40% [12].
On the other hand, the instruments used to quantify the impacts of the externalities caused by
transportation are not satisfactory and there are controversies regarding the relationship between
transportation and its impacts and regarding the associated costs calculation [11]. The economic
impacts of transportation in different categories are presented, among which the following stand
out: road operational costs (energy and workforce expenses); infrastructure damage costs (road
maintenance costs); congestion and scarcity costs (delay costs imposed on users due to traffic
congestion); environmental costs (noise, water, soil and air pollution); and accident costs (material
damage and product losses) [13].
Therefore, in face of the impact of excess weight measuring challenge in freight transportation,
which affects road infrastructure, safety and additional consumption of resources to maintain traffic
conditions in the transportation network, the objective of this article is to assess the relationship
between excess weight in the road transportation of ornamental stones, the transportation operational
costs and the social costs with road accidents and pavement maintenance, using a System Dynamics
(SD) model.
Transportation systems are complex and, therefore, public and/or private authorities need to test,
in simulation environments, the policies and actions before they are effectively implemented in the real
world, which is open and dynamic. Thus, SD may be used to simulate complex systems involving
feedbacks in which human judgment, experience and logic are combined [14].
The remainder of the paper is organized as follows. Section 2 presents a literature review with
different approaches to the problem of overweight in road transportation, converging to models that
used SD. Section 3 shows materials and methods, the SD model structure and testing. Section 4 presents
a case study addressing the problem of ornamental stones transportation in the state of Espírito Santo
and details the input parameters and the data gathering methodology. Section 5 describes the simulated
scenarios and also presents the results. Finally, Section 6 presents the final considerations.
2. Literature Review
Overweight vehicles in road transportation lead not only to mechanical failures in the vehicles
themselves but also greatly accelerates pavement deterioration. For that reason, legislation in many
countries determines the limits for axle loading and vehicle gross weight, while public agencies try to
enforce these limits through vehicle weighing and inspection. Trucks that exceed the legal loading limits
increase the risk of traffic accidents and lead to unfair competition between transportation companies.
Many studies in the literature address and try to quantify the impact of this practice,
its consequences and possible mitigation actions, such as the development of technologies to inspect
overloading in a more efficient way. However, this fact involves many elements, which makes it
difficult to find studies that comprehend all aspects.
The problem of overloading was critical in Africa [15], where the incentives to overloading were
high and the levels of inspection are low. The author proposed load measurement via the direct
installation of load sensors on the axles to help vehicle operators safely find the most economic load
considering road conditions and legal limits. Furthermore, regulatory authorities would have a way of
checking load per axle at any moment and place, without the need for a permanent structure.
The frequent weighing of trucks slows the supply chain down and increases operational
costs, emphasizing the importance of using information and communication technologies to share
information [16]. Since South African overweight control centers do not have electronic tracking,
the lines generated at the centers force some vehicles, which may be overloaded, to find a way around
the inspection, damaging the pavement and increasing road maintenance costs. Because of that,
the use of a system to identify these vehicles through automatic number plate recognition cameras was
proposed [17]. These vehicles are considered non-compliant and the level of punishment to be applied
will have to be determined by the road agencies but the authors suggest that a maximum penalty must
be applied to discourage this behavior and to generate additional funds for road maintenance.
Weigh-in-motion (WIM) technologies are another alternative, since they allow trucks to be weighed
in traffic, without any interruption in the operations, which may contribute to a safer and more efficient
operation of trucks, helps to reduce the number of overweight trucks, contributes to efficient police
time use and may reduce traffic accidents [18].
In 2005, the Ministry of Transportation in Egypt issued new regulations to increase the legal loads
per axle for trucks. An assessment of truck factors (TF) regarding the old and new regulations using
real truck loads and traffic data was presented [19]. TF are used to determine the equivalent single
axle load (ESAL) needed for projects and pavement maintenance, which increased with the expansion
of allowed tolerance. This impact was converted into additional asphalt thickness, increasing the
pavement maintenance costs.
The impact of the new axle load tolerance for vehicles in Egypt on pavement maintenance costs
and operational costs of transportation using the HDM-4 were assessed [20]. Model under different
levels of inspection, road maintenance options and different vehicle loading policies. The results
showed that, although the increased load per axle requires a pavement of greater structural capability,
the lower number of trips per truck required to transport the same load may lead to a reduction in the
operational cost of transportation, thus causing the increase in load limit to be considered useful.
However, the costs considered [20] are not the only impacts caused by the practice of overloading
and it is necessary to consider, for example, costs with road accidents, environmental impacts and trip
time, the latter affecting all road users. Moreover, the agents responsible for each cost category are
also different. Therefore, the managers responsible for regulation and inspection should prioritize the
minimization of the most relevant costs according to the objectives of the established public policies.
Based on a case study of Mexico, a bi-level modeling approach was presented [21] to represent the
interaction between the vehicle loading practices of carriers and the decisions made by road planning
authorities, which are responsible both for maintenance and overload control. Results show that
at a lower level (reactive), decisions made by carriers regarding overload impact road maintenance
expenses, while at an upper level (preventive), the planner decides about inspection levels anticipating
carrier responses.
Another method used in the literature to assess the systems behavior regarding load policies and
excess weight in freight vehicles is System Dynamics (SD). The term system refers to a set of elements
that interact with each other and the term dynamic refers to systems which variables are constantly
changing over time. The system dynamics emphasized the multiloop, multistate, nonlinear character
of the feedback systems. The existence of multiple interacting feedbacks means it is difficult to hold
other aspects of the system constant to isolate the effect of the variable of interest. Therefore, SD is a
method that studies the changes in the behavior of a system over time to assess impacts of decision
making at a strategic level [22–25].
The system may be represented via two approaches: qualitative and quantitative [26,27].
The qualitative model is built on causal loop diagrams that represent the feedback relationships
between the variables that define the model’s behavior [28]. In the quantitative approach, the system is
detailed in the stock and flow diagram, configuring differential equations that are numerically solved
by means of dynamic simulation (numerical integration). For more details about the SD method,
the causal loop diagram, the stock and flow diagram and the model tests, please check [22,26–32].
In the context of road maintenance operations, a SD model was developed to simulate the
impact of maintenance and obsolescence of a road system and to define the relationship between
maintenance activities and traffic volume with the use of road revenue generated by traffic to maintain
the pavement’s operation level [33]. Thus, the simulation model supports the decision about the best
use of available resources.
A SD model for the simulation of road degradation and restoration processes was presented,
calibrated with data from eight road sections in Virginia (USA) and coupled to an optimization module
of maintenance operations whose approach provides a plan of best practices for road maintenance [34].
The results suggest that the shift to a less costly preventive maintenance, instead of a more expensive
(and more time-consuming) corrective maintenance, should bring benefits to the system as a whole.
However, the policy of preventive maintenance was questioned [35] and another model of the effect
of road conditions on the index of road accidents was presented. With data from the region of New
England (USA), it demonstrates that the mental model behind the policy of road improvement could,
in most cases, increase accidents due to higher speeds and better pavement conditions. The author
shows the consequences of the maximization of road repairs with resource expenses, which can be
causing more accidents.
Nevertheless, the context in which the above-mentioned studies were carried out belongs to a
road system with a developed infrastructure, in other words, with high capacity expressways whose
management focuses only on maintaining good conditions of already existing trafficability. This reality,
however, is not consistent with the conditions of a good part of the road network in developing
countries, such as Brazil, according to studies indicated by the National Confederation of Transport [36].
In an emerging country such as China, a SD model was built to analyze the regulation of the
weight of freight vehicles and to assess the impacts of their overloading [37]. According to the authors,
such regulation has significant economic consequences, such as increase in the transportation cost per
truck and reduction in the cost of construction, maintenance, convenience and safety of highways.
Another SD model was proposed to evaluate the long-term effects of weight regulation policies of
trucks in China on the sustainability of the activity [38]. The model presents the cumulative economic
cost, including the operational cost and the cost associated with transportation time, in addition to the
accumulated social cost, including pavement maintenance, pollutant emissions and accident costs,
so that the policy effects on road transportation sustainability could be evaluated.
The problem of vehicle overloading in China [38] is similar to the one found in freight transportation
on Brazilian highways. However, the model presented by the authors has some limitations, such as the
consideration of a route dedicated to transport a specific load. This means that traffic in the studied
route is composed of only one type of vehicle and that there is no variability of traffic volume along the
route. Such characteristics have an impact on important variables, such as road capacity, travel time
and, consequently, operational cost of transportation.
In fact, the application of strict loading policies in China has not progressed well due to the increase
in transportation costs and the reduction of the activity’s economic performance. Then, a SD model
was developed to perform a long-term evaluation of the alternative modal shift policies, encouraging a
greater use of more efficient modes of transportation, such as railways, with results showing that it
was an effective option to achieve sustainability in the activity, reducing social, environmental and
economic costs of freight transportation [39].
In short, with SD it is possible to understand and evaluate the behavior of the road freight
transportation system with overloading and its impacts on other parameters, such as the economic
Sustainability 2019, 11, x FOR PEER REVIEW 5 of 37
and social costs involved. However, some gaps in the existing literature have been identified such
In short, with SD it is possible to understand and evaluate the behavior of the road freight
as the lack of a model
transportation closer
system to overloading
with the real operational conditions
and its impacts on otherof Braziliansuch
parameters, highways, with diversity
as the economic
in volume and categories
and social of vehicles.
costs involved. However, someTherefore, theexisting
gaps in the literature has not
literature haveprovided models
been identified such involving
as
different
thevehicles
lack of awith
modeldifferent
closer toload volumes
the real traveling
operational along
conditions ofthe same highways,
Brazilian route which withcould be applicable
diversity in
volume
to Brazilian and categories of vehicles. Therefore, the literature has not provided models involving
reality.
In different vehicles withstones
Brazil, ornamental different
roadload volumes traveling
transportation alongthe
presents theproblem
same route which could
of overload, whichbe causes
applicable to Brazilian reality.
damage that are often not accounted for or considered in the strategic decisions of the sector. For that
In Brazil, ornamental stones road transportation presents the problem of overload, which causes
reason,damage
the transportation of accounted
that are often not ornamental stones
for or in theinBrazilian
considered the strategicsoutheast region
decisions of wasFor
the sector. chosen
that as the
object ofreason,
a casethe
study for the development
transportation of ornamentalofstones
a SD in
simulation model
the Brazilian that covers
southeast required
region was chosencharacteristics,
as the
aiming object
to helpof ina the
casedevelopment
study for theofdevelopment
more effectiveof apolicies for sustainability
SD simulation model that in transportation.
covers required
characteristics, aiming to help in the development of more effective policies for sustainability in
transportation.
3. Materials and Methods
Considering the Methods
3. Materials and limitations of the models in the literature regarding the application of SD to
problems that involve overweight vehicles,
Considering the limitations this study
of the models in theproposed a model addressing
literature regarding theofimpacts
the application SD to of this
practiceproblems
on roadthat
transportation.
involve overweight vehicles, this study proposed a model addressing the impacts of
this practice on road transportation.
3.1. Model Structure
3.1. Model Structure
The variables in the model were grouped into six interrelated models, as shown in the causal loop
diagram in The variables in the model were grouped into six interrelated models, as shown in the causal
Figure 1. To perform the simulation, each module of the model was structured into stock
loop diagram in Figure 1Error! Reference source not found.. To perform the simulation, each module
and flow diagrams presented in more details as follows.
of the model was structured into stock and flow diagrams presented in more details as follows.
Figure
Figure 1. Causal
1. Causal looploop diagramfor
diagram for overweight
overweight in in
road freight
road transportation.
freight transportation.
The first module addresses “Load Volume Forecast,” shown in Figure 2. In this module, trips
The first module addresses “Load Volume Forecast,” shown in Figure 2. In this module, trips are
are caused by the production of a given product and are carried via the road mode. Different routes
causedcanby be
theavailable
production of a given product and are carried via the road mode. Different routes
to transport the load volume forecasted, that can be called Route #1 and Route #2,
can be for
available
example. toThe
transport
objectivetheof load volume
the module forecasted,
“Load that can is,
Volume Forecast” be therefore,
called Route #1 andthe
to calculate Route #2,
for example.
volumeThe objective
of load, in tons,ofthat
thewill
module “Load Volume
be transported via each Forecast”
consideredis, therefore, to calculate the volume
route.
of load, in tons, that will be transported via each considered route.
The module presented in Figure 3 regards “Speed Flow” and its objective is to provide the travel
time in each road route. Due to the differences of each part of the road, called link, as the number of
travelling vehicles and their speed, travel time is calculated separately and, at the end, added to the
total transportation time of each road route. Figure 3 shows the structure of the simulation model for
the travel time of a single link of a given route, which can be replicated to the number of links on the
available routes.
Sustainability
Sustainability 2019, 11, 2019,
312811, x FOR PEER REVIEW 6 of 37 6 of 35
<split proportion of
freight volume #1>
freight volume
freight volume in route #1
next year of route #1 increase of freight
<SAVEPER> volume route #1 freight volume
of #1
total freight <SAVEPER>

volume
increase of total
freight volume freight volume
of #2
freight volume in freight volume
freight volume
next year of route #2 increase of freight route #2
growth rate
volume route #2
<split proportion of
freight volume #2>
Figure 2. Load Volume Forecast Module. (Source: Own work using the VENSIM application).
Figure 2. Load Volume Forecast Module. (Source: Own work using the VENSIM application.)
The upper part of the diagram shows the traffic of the freight vehicles, called heavy vehicles,
The module presented in Figure 3 regards “Speed Flow” and its objective is to provide the travel
to differentiate
time in them from
each road those
route. Due that aredifferences
to the overloaded, in the
of each partmiddle partcalled
of the road, of thelink,
diagram. The lower
as the number of part
of the diagram
travellingshows theand
vehicles traffic
theirof lighttravel
speed, vehicles.
time is calculated separately and, at the end, added to the
Given the
total total volume
transportation of of
time vehicles,
each road it route.
is necessary
Figure 3to convert
shows it into unit
the structure of simulation
of the passengermodel
car equivalent
for
the travel time of a single link of a given route, which can be replicated
(PCE), defined by Highway Capacity Manual—HCM [40]. To calculate road capacity, HCM also to the number of links on the
providesavailable routes.
the adjustment factors, due to the presence of heavy vehicles, which reduces the traffic speed
The upper part of the diagram shows the traffic of the freight vehicles, called heavy vehicles, to
of the highway. After finding out the load of the whole network, an analysis of each road link is carried
differentiate them from those that are overloaded, in the middle part of the diagram. The lower part
out aiming to determine
of the diagram shows thethe
relationship
traffic of lightbetween
vehicles.equivalent traffic volume and the average capacity
of the link. Equation (1) [41] is applied to find out the travel time spent in the analyzed road link.
t = t0 ·(1 + α·(V/C)β (1)
where:
• t: adjusted time in which volume V may use road link;

• t0 : travel time for a free flow condition;
• V: traffic equivalent volume;
• C: road capacity; and
• α and β: calibration parameters.
Free flow time is calculated by the length of the link considered in kilometers and the operational
speed. The total traveled distance in each route is obtained by summing up the lengths of each link.
The module “Modal Split,” shown in Figure 4, uses travel time and distance, besides the
transportation operational cost in function of the distance and time for each pavement condition,
whose objective is to calculate the proportion of load volume to be transported in each route. In this
module, the modal choice refers to the choice of the route to be used, determined by the multinomial
logit function employed in the modal split model [42]. This way, the proportion of freight volume
transported through Route #1 is given by Equation (2).
 n 
X 
Split proportion o f f reight volume #1 = exp(−c1) / exp(−cn )  (2)
1
where:
• c1: generalized cost of Route #1; and

• n: number of available routes.
equivalent of heavy
AADT growth vehicles - capacity maximum
rate heavy vehicle <vehicle with
adjustment capacity
flow #1 ornamental stones
accumulated heavy #1> heavy vehicle grade adjustment
heavy vehicle vehicles in next year accumulated adjustment factor for for capacity
increase in heavy traffic #1 #1 heavy vehicles capacity
vehicle traffic #1 #1 percentage of
heavy vehicle #1 heavy vehicle average
<SAVEPER> adjustment factor for capacity #1
volume
truck loadage of equivalent of heavy
statutory limit vehicle - volume
<freight volume heavy vehicle
adjustment
of #1> #1
overweight truck loadage
percentage #1 traffic equivalent
total traffic Vvhp #1 v/c #1
volume #1
vehicle with #1
total vehicles
ornamental stones #1 hours per
<AADT growth
#1 year
rate> light vehicles
#1 grade adjustment
peak hour for volume #1 average travel
<SAVEPER> factor
accumulated time #1
light vehicle light vehicles
traffic #1 free flow time
increase in light #1 #1 α
vehicle traffic #1 accumulated light distance #1 β
vehicles in next year
light vehicle
#1
flow #1 free flow
speed
Figure3.3. Speed
Figure SpeedFlow
FlowModule.
Module. (Source:
(Source: Own
Own work
work using
using the
the VENSIM
VENSIM application).
application.)
Sustainability
Sustainability 11,11,
2019,
2019, 3128PEER REVIEW
x FOR 9 of8 37
of 35
<Time>
<SAVEPER> inflation rate
<PCI value
#1> cost growth
<INITIAL TIME>
rate
<truck loadage <PCI value
distance cost
<PCI value #1> #1>
#1
#3>
<cost growth toll #1 operational cost
operational cost time cost #1
rate> toll of unit freight per time #1
per distance #1 fine cost #1
volume #1
distance cost <distance
<truck loadage #3 route #3> fine cost #3 <total travel time
#2> <distance generalized cost in route #1>
operational cost route #1>
per distance #3 of route #1
toll #3 generalized cost dimensionless split proportion of split proportion of

average truck toll of unit freight of route #3 exponent freight volume #1 freight volume #2
loadage #3 volume #3
<distance generalized cost
route #2> of route #2 <total travel time
operational cost route #2>
per time #3 <total travel time toll #2
<truck loadage operational cost
#1> route #3>
per distance #2 fine cost #2 toll of unit freight operational cost time cost #2
volume #2 per time #2
time cost #3
<cost growth distance cost <truck loadage
rate> #2 #2> <PCI value
#2>
<PCI value <cost growth
#3> rate>
<PCI value
#2>
Figure
Figure 4. Modal
4. Modal Split
Split Module.
Module. (Source:
(Source: OwnOwn work
work using
using thethe VENSIM
application.)
Generalized cost depends on the type of vehicle and on the physical and operational conditions
of the road and is made of parts that vary according to the distance traveled, travel time and other
fixed costs, such as tolls and other fine costs [43]. Equation (3) is used to calculate generalized
operational cost.

GCv,r,p = β1 · Tv,r,p · CTv,r,p + β2 · Lr,p · CLv,r,p + β3 ·(CTv ) + β4 ·(CFv ) (3)
where:
• GCv,r,p: generalized cost generated by a vehicle v while traveling a link of category r,p (r = type of
terrain and p = pavement condition);
• Tv,r,p: time traveling through the link of category r,p by the vehicle v;
• CTv,r,p: parameter of operational cost per time unit required for vehicle v to travel a link of
category r,p;
• Lr,p: length of link of category r,p;
• CLv,r,p: parameter of operational cost per unit of distance required for vehicle v to travel a link of
category r,p;
• CTv: toll cost of the link for a vehicle of class v, which may be zero, in the absence of a toll or the
fare value, in case there is a toll station in the link;
• CFv: fine costs for driving overloaded vehicle of class v; and
• β1 , β2 , β3 and β4 : perception weights.
In the “Pavement Maintenance” module, presented in Figure 5, the Pavement Condition Index
(PCI) determines if the maintenance operations may be of corrective or restorative types. The Design
ESAL is the number of demands made by a standard axle, based on the Annual Average Daily Traffic
(AADT), on its growth rate, on the project’s life span and considering the weight limits per axle,
represented by the fleet truck factor (TF). Knowing the Design ESAL, the current life span of the
pavement is calculated in function of traffic demand. The accelerated degradation and the reduced
life span of the pavement are caused by excess weight per axle. For that reason, it was necessary to
characterize the composition of heavy vehicle traffic of the analyzed routes.
To calculate the Load Factors (LF), the AASHTO method [12] was used, considering the weight
per axle with the allowed tolerance and the excess weight percentage practiced. The reduced life
span is related to the PCI through the parameters α and β, related to pavement performance [44].
The PCI measures the quality of the pavement and accumulates the difference between the effect of
maintenance and the decrease in PCI caused by vehicle traffic.
The first maintenance condition is the current pavement quality level, that is, if the current PCI
is higher than the limit that requires the conduction of corrective maintenance, there is no need for
pavement intervention at that given moment of time. The second condition for the requesting of
interventions is the absence of any ongoing maintenance operation. Finally, the request for maintenance
interventions must respect the time interval in which no assessments of pavement conditions are
carried out. Considering these three conditions, if the PCI is within the limits for restorative and
corrective maintenance, corrective maintenance is made. If the PCI is under the limit for restorative
maintenance, this type of maintenance is made.
The stock effect of maintenance accumulates the input of such effect, which occurs when the
maintenance is requested but with a delay in the application of such effect on PCI due to the considerable
time needed to carry out the works. The variable maintenance operations are defined by the values 1
and 0.5 for restorative and corrective maintenance operations, respectively and zero if there is no need
for pavement intervention.
The fifth stock and flow module shows the “Social Cost of Freight Transportation,” which is
subdivided into pavement maintenance costs and traffic accident costs. Figure 6 shows the stock and
flow diagram for pavement maintenance costs.
wl/single tyres B #1 LF single tyres B #1
wl/twin tyres B #1 LF twin tyres B #1 axle factor

TF bus #1 bus #1
wl/single tyres LT #1 LF single tyres LT #1
wl/twin tyres LT #1 LF twin tyres LT #1 TF light truck #1 fleet vehicle AADT initial
days/year year #1
factor
wl/single tyres ST #1 LF single tyres ST #1 axle factor light
truck #1
wl/1double tandem ST #1 LF 1double tandem ST #1 Design ESAL total volume of average volume of <AADT growth
axis requests
#1 vehicles #1 vehicles #1 rate>
wl/2double tandem ST #1 LF 2double tandem ST #1
TF single
wl/1triple tandem ST #1 LF 1triple tandem ST #1 trailer #1 current life
spam #1 service life of
wl/2 triple tandem ST #1 LF 2 triple tandem ST #1 current ESAL project maintenance
#1 operations #1
overweight per axle wl/3 triple tandem ST #1 LF 3 triple tandem ST #1axle factor single limit for corrective
percentage #1 Parameter α Parameter β maintenance
trailer #1 #1
wl/single tyres MT #1 LF single tyres MT #1 #1
restorative limit for restorative <Time>
<total vehicles
wl/1double tandem1 MT #1 LF 1double tandem1 MT #1 <SAVEPER> maintenance effect maintenance
#1 0>
wl/2double tandem1 MT #1 LF 2double tandem1 MT #1 corrective <INITIAL
regional climate maintenance effect maintenance TIME>
wl/1double tandem2 MT #1 LF 1double tandem2 MT #1 factor time control of
#1 pavement evaluation
wl/2double tandem2 MT #1 LF 2double tandem2 MT #1 TF multi
trailer #1
wl/1double tandem3 MT #1 LF 1double tandem3 MT #1
effect of
PCI value #1 maintenance
wl/2double tandem3 MT #1 LF 2double tandem3 MT #1 axle factor multi #1 increase in
decrease of decrease in
trailer #1 maintenance effect
PCI #1 maintenance effect
#1
#1
<SAVEPER>
Figure 5.
Figure Pavement Maintenance
5. Pavement Maintenance Module.
Module. (Source:
(Source:Own
Ownwork
workusing
usingthe
theVENSIM
VENSIMapplication.)
application).
RM cost per <maintenance <vehicle with

km operations #1> ornamental stones
#1>
<cost growth <distance RM cost #1 pavement
<heavy vehicle
rate> route #1> maintenance cost #1>
#1
CM cost #1 CM cost per
km
<RM cost per
km>
<distance RM cost #2 pavement cumulative
<cost growth
route #2> maintenance cost pavement
rate> maintenance
CM cost #2 #2 <vehicle with yearly pavement cost
ornamental stones maintenace cost
<maintenance #2>
<CM cost per operations #2> <heavy vehicle
km> #2>
<RM cost per
km>
RM cost #3 pavement
<cost growth <distance <vehicle with
rate> maintenance cost ornamental stones
route #3>
CM cost #3 #3 #3>
<heavy vehicle
<maintenance #3>
<CM cost per
km> operations #3>
Figure 6. Social Cost of Freight Transportation Module—Pavement maintenance. (Source: Own work using the VENSIM application).
Figure 6. Social Cost of Freight Transportation Module—Pavement maintenance. (Source: Own work using the VENSIM application.)
The stock type variable “cumulative pavement maintenance cost” accumulates the yearly cost
with pavement maintenance throughout the whole simulation period. The flow variable “yearly
pavement maintenance cost” is the sum of all costs with maintenance in each of the analyzed routes.
The costs with corrective maintenance (CM) and restorative maintenance (RM) are defined by the cost
per kilometer, the distance of the route and the cost growth rate, as shown in Equations (4) and (5).
CM cost #1 = CM cost per km · distance route #1 · cost growth rate (4)
RM cost #1 = RM cost per km · distance route #1 · cost growth rate (5)
Finally, the pavement maintenance cost is calculated in function of the type of intervention to be
carried out, defined on previous module and their related costs, as shown in Equation (6).
pavement maintenance cost = IF THEN ELSE(“maintenance operations” = 0, 0,

(6)
IF THEN ELSE(“maintenance operations” = 0.5, (“CM cost” ), (“RM cost” )))
The second part of the “Social Cost of Freight Transportation” module addresses the costs
with traffic accidents, as shown in the stock and flow diagram presented in Figure 7. The variable
“cumulative traffic accident cost” accumulates the yearly costs with traffic accidents throughout the
simulation period. The flow variable “yearly traffic accident cost” is the sum of the costs with accidents
per year in the analyzed routes, which is given by Equation (7).
tra f f ic accident cost

3
P (7)
= accident f orecast · ( (% accident level i · cost accident level i)· cost growth rate
i=1
The prediction of accidents is defined in function of the AADT, which regards to the overloaded
vehicles and the traveled distance, according to the accident prediction model proposed by the Highway
Safety Manual User Guide—HSM [45], as shown in Equation (8).
accident f orecast= overloaded vehicle · distance route·10(−6) ·exp(−0.312) ·correction f actor accident pred (8)
The correction factor for accident prediction is needed and it is indicated by the HSM in case the
analyzed routes are not under the ideal conditions indicated by the manual. Therefore, the accidents
are dependent of both limit of the vehicles and conditions of the highways.
Finally, the sixth module, “Policy Assessment,” receives the road maintenance costs, the accident
costs and the generalized cost of each route to calculate the cumulative total cost. Then it is able to
simulate scenarios with importance weights attributed to the costs (operational and social) in function
of different loading policies for freight vehicles in overload percentage. Figure 8 shows the stock and
flow diagram for this module.
The variable “cumulative transportation cost” accumulates the yearly transportation costs of
the available routes. The weighed cumulative operational cost is given by the stock of cumulative
transportation cost and the weight (relative importance) of such cost. Likewise, the weighed cumulative
social cost is given by the sum of pavement maintenance costs and accident costs, multiplied by the
weight of social costs. Finally, cumulative total cost is the sum of the weighed cumulative operational
and social costs of all routes involved.
This way, it is possible to assess the impact caused by varying overweight percentages in freight
vehicle loading on total system cost, as well as by varying the importance attributed to each type of
cost by decision makers.
Sustainability2019,
Sustainability 2019,11,
11,3128
x FOR PEER REVIEW 14 of
13 of 35
37
correction factor for

accident prediction accident with
<vehicle with victims #1
ornamental stones accident with no
#1> victims #1 accident with
death #1
<distance accident
route #1> forecast #1 traffic accident
cost #1
accident with accident with accident with no <cost growth

death cost victims cost victims cost rate>
<vehicle with
ornamental stones
#2> traffic accident cumulative traffic
cost #2 yearly traffic accident cost
accident
accident cost
forecast #2 accidente with
<correction factor for
accident prediction> death #2
accident with no
victims #2
<distance accident with <accident with
route #2> victims #2 death cost>
<vehicle with
ornamental stones
accident traffic accident <accident with
#3>
forecast #3 cost #3 victims cost>
accident with
<distance death #3 <accident with no
route #3> victims cost>
accident with <cost growth
<correction factor for
victims #3 rate>
accident prediction> accident with no
victims #3
Figure 7. Social
Figure 7. Social Cost
Cost of
of Freight
Freight Transportation
TransportationModule—Accidents.
Module—Accidents. (Source:
(Source: Own
Own work
work using
using the
the VENSIM
application.)
Finally, the sixth module, “Policy Assessment,” receives the road maintenance costs, the accident
costs and the generalized cost of each route to calculate the cumulative total cost. Then it is able to
simulate scenarios with importance weights attributed to the costs (operational and social) in function
of different
Sustainability loading
2019, 11, 3128 policies for freight vehicles in overload percentage. Figure 8 shows the14
stock
of 35 and
flow diagram for this module.
<freight volume <generalized cost

of #1> of route #1>
cumulative
transportation
yearly
cost
transportation cost
<generalized cost <freight volume

of route #2> of #2>
weight of
weight of operational cost
social cost
<cumulative traffic
accident cost> weighed weighed cumulative
cumulative social operational cost
cost
<cumulative cumulative
pavement total cost
maintenance cost>
Figure 8. Freight
Figure Transportation
8. Freight Policy
Transportation Assessment
Policy Module.
Assessment (Source:
Module. OwnOwn
(Source: workwork
usingusing
the VENSIM
the VENSIM
application).
application.)
3.2. Mathematical Formulation

The variable “cumulative transportation cost” accumulates the yearly transportation costs of the
available routes.
The next step The
of SD weighed cumulative
methodology operational
is the development of thecost is given by
mathematical the which
model, stock of cumulative
consists in
transportation
transforming costand
the stock and
flowthediagram
weightinto
(relative importance)
a system of equations,
of differential such cost.solved
Likewise, the weighed
via simulation
cumulative
by numerical social cost
integration. Theis model
given was
by the sum ofinpavement
structured Vensim PLE ® software—a
maintenance costs
freeand accident
software from costs,
multipliedVentana
the company by the Systems
weight of[46]
social costs. for
intended Finally, cumulative
personal total cost use.
and educational is theThe
sum of the weighed
mathematical
cumulative
equations and theoperational
model canand social costs
be obtained of all routes involved.
at supplementary materials.
This way, it is possible to assess the impact caused by varying overweight percentages in freight
3.3. Model
vehicleTesting
loading on total system cost, as well as by varying the importance attributed to each type of
cost by decision makers. model fits in the quantitative modeling category, since it involves
The proposed simulation
inductive formulation and model simulation focused on the comprehension of stock and flow structures
and 3.2. Mathematical
applied Formulationof quantifiable systems [32]. The tests considered most appropriate to
to the representation
this modeling category
The next are:
step of SDphysical conservation,
methodology dimensionalof
is the development consistency, integration
the mathematical error,
model, extreme
which consists
conditions tests, boundary adequacy tests, parameter assessment, basic behavior reproduction
in transforming the stock and flow diagram into a system of differential equations, solved via and
endogenous
simulationbehavior reproduction
by numerical tests. The model was structured in Vensim PLE® software—a free
integration.
During the
software fromformulation
the company of Ventana
the model, simulations
Systems were carried
[46] intended out toand
for personal verify the physical
educational use. The
conservation of theequations
mathematical model’s structure. For example,
and the model Figure 9 at
can be obtained shows the stock ofmaterials.
supplementary the PCI, in which it is
possible to see that the index decreases until receiving the effect of restorative maintenance in 2022 (100
3.3.and
points) Model Testing maintenance in 2028 (80 points). The stock does not become negative, satisfying
corrective
the expected behavior.
The proposed simulation model fits in the quantitative modeling category, since it involves
Dimensional consistency is the specification of the units of measurement of the variables throughout
inductive formulation and model simulation focused on the comprehension of stock and flow
the formulation of the model and the verification of their physical meaning using mathematical
structures and applied to the representation of quantifiable systems [32]. The tests considered most
equations [22]. To carry out this test, the command “Check Units” of the Vensim software was used.
appropriate to this modeling category are: physical conservation, dimensional consistency,
All inconsistencies were solved during the formulation of the model until the dimensional adjustment
integration error, extreme conditions tests, boundary adequacy tests, parameter assessment, basic
of all variables involved in the proposed model.
behavior reproduction and endogenous behavior reproduction tests.
Besides, some variables were replaced with internal variables such as “Time” (current simulation
time), “Initial Time” (initial simulation time) and “Saveper” (time interval in which simulation results
are saved), the units of which are defined the same as for the simulation model horizon, given in years.
This way, the communication and comprehension of the model was facilitated.
During the formulation of the model, simulations were carried out to verify the physical
conservation of the model’s structure. For example, Figure 9 shows the stock of the PCI, in which it
is possible to see that the index decreases until receiving the effect of restorative maintenance in 2022
(100 points) and corrective maintenance in 2028 (80 points). The stock does not become negative,
satisfying the expected behavior.
Figure 9. Stock
Figure of the
9. Stock Pavement
of the condition
Pavement conditionindex.
index. (Source: Ownwork
(Source: Own work using
using thethe VENSIM
application.)
For the proposedconsistency

Dimensional model, Euler’s integration of
is the specification method
the unitswas of used due to the
measurement lowvariables
of the amount of
computational
throughouttime required byofitthe
the formulation andmodel
its highandnumeric stabilityof[47].
the verification their In the integration
physical error test,
meaning using
mathematical equations [22]. To carry out this test, the command “Check
the model should be executed with an initial estimate for dt and this value should be progressively Units” of the Vensim
software
reduced wasuntil
by half used. All inconsistencies
there are no significant werechanges
solved during
to the the formulation
model’s of the
behavior. Themodel
timeuntil
stepthechosen
dimensional adjustment
1 of all variables involved in the proposed model.
for the model was of 4 of a year. This choice was made based on simulations with different values of
Besides, some variables were replaced with internal variables such as “Time” (current
time step for the variable “Total traffic #1.” A difference of 1% was considered acceptable for the total
simulation time), “Initial Time” (initial simulation time) and “Saveper” (time interval in which
traffic during one year in the considered route. This way, the chosen time step was of 0.25 year, since,
simulation results are saved), the units of which are defined the same as for the simulation model
whenhorizon,
the timegiven
step isinchanged
years. This from 0.25the
way, to 0.125, the change
communication andin comprehension
behavior is lower thanmodel
of the 1%. was
The extreme
facilitated. conditions test checks whether the model has an appropriate behavior when the
parameters are submitted to extreme values, such as zero or infinite
For the proposed model, Euler’s integration method was used due to the low amount of and may be carried out in two
ways:computational
via the directtimeinspection
requiredofby the model’s
it and equations
its high numericor via simulation
stability [47]. In the[22]. The variables
integration submitted
error test, the
model values
to extreme should be forexecuted
this test withand an
theinitial
behaviorestimate for dt and
expected thisverification
for the value shouldare be shown
progressively
in Table 1.
Afterreduced by half
conducting the until
teststhere
viaare no significant
simulation, changes
all the to the model’s
behaviors expected behavior.
under The time step chosen
the established extreme
conditions were respected, which confirms the reliability of the proposed model’s structure. of
for the model was of ¼ of a year. This choice was made based on simulations with different values
time step for the variable “Total traffic #1.” A difference of 1% was considered acceptable for the total
traffic during one year in Table
the considered
1. Variablesroute.
for This way, theconditions
the extreme chosen time step was of 0.25 year, since,
test.
when the time step is changed from 0.25 to 0.125, the change in behavior is lower than 1%.
The extreme conditions test checks
Module whether the model has
Variable Valuean appropriate behavior
Behavior when the
Expected
parameters are submitted to extreme values, such
Initial stock of Total as zero or infinite and may be carried out in two
Absence of vehicles for stone
ways: via theforecast
Load volume direct inspection freight model’s equations or0via simulation [22]. The variables
of thevolume
transportation and their effects
submitted to extreme valuesInitial
forstock
this of freight
test volume
and the #1 expected for the verification
behavior are shown in
over costs
0
Table 1. After conducting the and testsfreight volume #2 all the behaviors expected under the established
via simulation,
Speed
extreme Flow
conditions Maximum
were respected, whichcapacity 10 of the proposed
confirms the reliability Increase inmodel’s
averagestructure.
travel time
Proportion of freight volume
Modal split Toll #1 1.000
in Route #1 will be null
Overweight percentage per axle 10.000 Current life span will be null
Pavement maintenance
Project life span 0 PCI value will be null
The boundary adequacy test checks whether the limits of the model’s boundaries are adequate
for the simulation purpose. For that, it is possible to directly inspect the model’s equations, but the
diagrams of causal relationships and stock and flow facilitate the identification of exogenous constants
that should be considered variables of the model. If an additional structure has a significant impact
over the behavior or political implications, it must be included as an integral part of the model [22].
Road capacity was initially considered an input parameter external to the model. However,
since this parameter influences route travel times, it was chosen to include the calculation of road
capacity in the model, using the HCM as a reference [40]. It was also decided that the calculation of the
Design ESAL should be included, since it impacts the reduction in pavement life span and the decision
about road maintenance requests. Therefore, the limits of the model were expanded and adjusted as
needed throughout its construction and formulation considering the desired objective.
The tests of parameter assessment, basic behavior reproduction and endogenous behavior
reproduction were not carried out due to the absence of historical data about the AADT for the
analyzed road network of the case study. However, the tests conducted make it possible to conclude
that the model is capable of obtaining results and simulating scenarios.
4. Case Study
The case study approaches the transport of ornamental stones in the state of Espírito Santo (Brazil).
This model considered the characteristics related to the Brazilian reality, in which the roads are used by
various vehicle categories and all of them significantly impact travel time, pavement conditions and
operational costs of the routes used.
The state of Espírito Santo, which is in the Southeast region, shown in Figure 10, stands out in the
national scenario of ornamental stones because it has the largest reserve of marble in the country and a
large reserve of granite. The two main regions in this sector are: the southern state region and the
northwestern region, whose reference is the city of Nova Venécia, with a large production of granite
and a high concentration of businesses in both extraction and processing [48].
Figure 10. Map

Figure10. Map ofofBrazil
Braziland
andEspírito
EspíritoSanto
Santohighlighting
highlighting the
themain
maincenters
centersof
ofthe
theornamental
ornamentalstones
stones
industry in the region and the port of Vitória.
Considering that around 70% of all granite exported by Espírito Santo comes from the northwestern
Considering that around 70% of all granite exported by Espírito Santo comes from the
region of the state [49], the object of this study was the transportation of ornamental stones from Nova
northwestern region of the state [49], the object of this study was the transportation of ornamental
Venécia to the port of Vitória. The access to the productive center of the city of Nova Venécia may occur
stones from Nova Venécia to the port of Vitória. The access to the productive center of the city of
via two main road routes, through which the production of ornamental stones is transported to the
Nova Venécia may occur via two main road routes, through which the production of ornamental
port of Vitória, according to Figure 11.
stones is transported to the port of Vitória, according to Figure 11.
However, the links that comprise each road have specific characteristics that need to be considered,
such as Annual Average Daily Traffic (AADT) and road capacity, as indicated by Reference [12],
since the different vehicle categories and the volume of traffic that use the roads significantly impact
pavement conditions, trip time and the operational cost of using these routes.
To characterize the AADT, in 2014, DNIT resumed the National Plan of Traffic Count (PNCT—Plano
Nacional de Contagem de Tráfego in Portuguese), which identifies traffic behavior on federal paved
roads. Besides defining a systematic traffic count plan using permanent count devices in federal roads,
in 2016 and 2017 DNIT conducted Traffic Count and Classification surveys and Origin and Destination
surveys to collect information on vehicle flows on federal roads [50]. This way, the ornamental stones
Figure 11. Schematic representation of road routes and links.
However, the links that comprise each road have specific characteristics that need to be
considered, such as Annual Average Daily Traffic (AADT) and road capacity, as indicated by
Reference [12], since the different vehicle categories and the volume of traffic that use the roads
Figure 10. Map of Brazil and Espírito Santo highlighting the main centers of the ornamental stones
Considering that around 70% of all granite exported by Espírito Santo comes from the
northwestern region of the state [49], the object of this study was the transportation of ornamental
stones from Nova Venécia to the port of Vitória. The access to the productive center of the city of
Nova Venécia may
transportation routesoccur via two
analyzed main
in this roadwere
study routes, through
divided which
in links the production
according of ornamental
to the homogeneity of
stonesvolume,
traffic is transported to the port
as illustrated of Vitória,
by each junctionaccording
in Figureto11.
Figure 11.

Route #1 (link #1.1 and link #1.2, see Figure 11) does not have weight inspection via fixed scales,
However, the links that comprise each road have specific characteristics that need to be
which facilitates the practice of overloading. However, freight vehicles that use Route #1 may eventually
considered, such as Annual Average Daily Traffic (AADT) and road capacity, as indicated by
be submitted to inspection with mobile scales operated by DNIT. Moreover, Route #1 does not have
Reference [12], since the different vehicle categories and the volume of traffic that use the roads
tolls either, which influences the users’ choice for this route.
significantly impact pavement conditions, trip time and the operational cost of using these routes.
Route #2 (link
Sustainability 2019,#2.1
11, xand linkREVIEW
FOR PEER #2.2) has a fixed scale on the flow direction analyzed in19 this of 37 study,
To characterize the AADT, in 2014, DNIT resumed the National Plan of Traffic Count (PNCT—
that is, from north to south. However, inspections at the fixed scales are subject to the scheduling and
Plano Nacional Route de#2Contagem
(link #2.1 and de link
Tráfego in Portuguese),
#2.2) has a fixed scale onwhich
the flow identifies traffic behavior
direction analyzed on federal
in this study,
availability
that
of
is,
inspectors
from north
belonging
to south.
to theinspections
However,
responsible at
agency
the fixed
in the
scales are
state,
subject
that
to
is, scheduling
the
they do not and
operate
paved roads. Besides defining a systematic traffic count plan using permanent count devices in
continuously. Besides,
availability Routes #2 and #3to(link
of inspectors #3.1 and agency
link #3.2) have tolls.
federal roads, in 2016 and 2017belonging
DNIT conducted the responsible
Traffic Count in the
andstate, that is, they do not operate
Classification surveys and Origin
Therefore, it is initially
continuously. Besides, assumed
Routes #2 and that#3legal
(linkweight
#3.1 andlimits perhave
link #3.2) vehicle
tolls.are respected in Route #2 but
and Destination surveys
Therefore,
to collect
it is initially
information
assumed being
that legal
onweight
vehicle flows on
limits
federal roads [50]. This way, the
there is also the possibility of overweight practiced in thisper vehicle
route are respected
considering theininconsistency
Route #2 of
ornamental but stones
there is transportation
also the routes
possibility of analyzed
overweight in this
being study
practiced were
in divided
this route in links
consideringaccording
the to
inspection in it. Therefore, considering that overweight was identified as a recurring problem for the
the homogeneity
inconsistency of traffic volume,
of inspection in it.as illustrated
Therefore, by eachthat
considering junction in Figure
overweight 11.
was identified as a recurring
road transportation of ornamental stones in the state of Espírito Santo, this article presents a SD model
Routeproblem
#1 (link for #1.1
the road
andtransportation
link #1.2, seeofFigureornamental stones
11Error! in the state
Reference of Espírito
source Santo, this
not found.) article
does not have
to analyze the relationship
presents a SD model between
to analyze overweight
the relationship vehicle
between and the costs
overweight associated,
vehicle and the such
costs as operational
associated,
weight inspection via fixed scales, which facilitates the practice of overloading. However, freight
transportation
such ascosts and social
operational costs withcosts
transportation roadand accidents and pavement
social costs with road maintenance.
accidents and pavement
vehicles that use Route #1 may eventually be submitted to inspection with mobile scales operated by
Thus, we present the values of the specific constants in each module for the application of the
maintenance.
DNIT. Moreover, Thus, Route
we present #1 does not have
the values tolls
of the either,
specific whichininfluences
constants each module theforusers’ choice forofthis
the application the route.
proposed case study.
proposed case study.
4.1. Parameters for the Load Volume Forecast Module
4.1. Parameters for the Load Volume Forecast Module
For the first module, it is necessary to obtain the freight volume growth rate. For that, data on the
For the first module, it is necessary to obtain the freight volume growth rate. For that, data on
exports of
theornamental stones instones
exports of ornamental the municipality of Nova
in the municipality Venécia
of Nova Venéciawere
wereused
used[51]
[51] and
and ititisisknown
known that
an average
thatof
an34% of the
average of Brazilian
34% of the production of stones
Brazilian production of was destined
stones to thetoforeign
was destined market
the foreign between
market
2005 andbetween 2005
2008 [52]. and it2008
Thus, [52]. Thus,
is possible it is possible
to estimate to estimate of
the production thestones
production
in the of stones in the
municipality of Nova
municipality of Nova Venécia based on the export values
Venécia based on the export values [53], as shown in Figure 12. [53], as shown in Figure 12.
12. Estimate
FigureFigure of the
12. Estimate ornamental
of the ornamentalstones production
stones production in in Nova
Nova Venécia.
Venécia.
There is an observable growth in the production of stones in the period between 2008 and 2013
and an oscillation in the remaining periods, with an overall growth average of 8% a year. When the
year of 2017 is disregarded, since the base year for the beginning of the simulation is 2016, the growth
percentage is of 11% a year, value used in the simulation.
There is an observable growth in the production of stones in the period between 2008 and 2013
and an oscillation in the remaining periods, with an overall growth average of 8% a year. When the
year of 2017 is disregarded, since the base year for the beginning of the simulation is 2016, the growth
percentage is of 11% a year, value used in the simulation.
4.2. Parameters for the Speed Flow Module

For the “Speed Flow” module, it is necessary to find the AADT of each link of the analyzed routes,
retrieved from the PNCT project [50]. For each analyzed link, the average volume of the sub-links of
the National Traffic System were considered. Thus, the volume of light and heavy vehicles, identified
in the model’s base year of simulation (2016), is summarized in Table 2.
Table 2. Estimate of the average AADT (vehicles) of each link in 2016.
Routes Links Distance (Km) Light AADT Heavy AADT Total AADT
#1.1 123 998 365 1.363
#1
#1.2 45 1.784 609 2.394
#2.1 65 947 388 1.335
#2
#2.2 110 3.385 1.293 4.678
#3.1 38 4.956 2.289 7.245
#3
#3.2 26 6.515 2.030 8.546
Source: Authors based on Reference [50].
According to the Traffic Studies Manual [12], the traffic has relatively low variation in annual
growth rates. According to the manual, it has become common to adopt an annual growth rate of 3%,
adopted in the simulation model.
For the transportation of ornamental stones, there are several configurations of vehicle axles,
whose Total Gross Weight (TGW) limit varies from 23.1 to 77.7 tons. However, according to Resolution
nº 354/2010 [54], the vehicle adopted as the standard vehicle for the transportation of stones was
a special multitrailer of seven axles (three dual tire tandem axles and a one-tire single axle) with a
Combined Total Gross Weight of 57 tons.
Given the legal limit of vehicle loading and the volume of ornamental stones to be transported,
it is possible to find the traffic of stone transportation vehicles, which will be added to the traffic of
other vehicle categories. This aspect of the model is also different from the models found in literature
review, in which the authors consider only the traffic of specific vehicles, which is not consistent with
the Brazilian reality.
For information regarding the overweight practiced in the transportation of ornamental stones on
the analyzed routes, this study used data regarding weighings carried out by mobile scales operated by
DNIT on roads in the state of Espírito Santo between 2015 and 2017. Data showed that the percentage
of overweight in TGW varied from 0.05% to 57.20%, already considering the 5% tolerance defined in
the legislation. Overweight per axle, in turn, varied from 0.25% to 70.70%, already considering the 10%
weight tolerance per axle, as defined in current legislation. The operational speed of 80 km per hour
(Km/h) was considered for the free flow time on the links.
4.3. Parameters for the Modal Split Module

The operational cost associated with each of the analyzed routes was found based on the HDM-4
methodology [43]. The operational costs vary according to the type of vehicle, the type of guideway,
pavement condition and type of terrain. These differences lead to different travel times for the same
distances, which may affect transportation route choice via the road mode. Moreover, the operational
costs were calculated for the following combination of standard links:
• Roads (5 types): Multilane highways, wide lane (two-lane highways with paved at-grade shoulder),
federal two-lane highway, state two-lane highway and unpaved highway;
• Pavement (4 types): Good, fair, mediocre and poor, defined according to the IRI (International
Roughness Index [55]; and
• Terrain (5 types): level, slightly rolling, rolling, very rolling and mountainous, defined through
the vertical and horizontal geometry indices.
The maintenance items considered were the following: fuel, lubricants, tires, maintenance (parts),
workforce, depreciation and crew costs (freight vehicles). The first four of these items are related to the
distance traveled and the remaining ones to the time needed to travel this distance. Thus, it is possible
to arrive at a set of unit costs that vary according to the distance and a set of unit costs that vary with
time [43], as previous shown in Equation (3).
There is information about Routes #1, #2 and #3 concerning the type of lane and terrain on the
georeferenced database for project PNCT [50]. The vehicle class chosen was the multitrailer with 7
axles and the costs per distance and per time (cost used in first and second terms of Equation (3)) were
determined in function of the variation in pavement condition, as presented in Table 3.
Table 3. Operational costs per distance and time in relation to type of lane, pavement condition and
type of terrain for a multitrailer with 7 axles.
Cost per
Routes Type of Road_Condition_Terrain PCI Cost per Distance (R$/Km)
Time (R$/H)
Two-Lane_Good_Rolling 100 1.58 83.91
Two-Lane_Fair_Rolling 75 1.73 83.06
#1
Two-Lane_Mediocre_Rolling 50 1.91 79.61
Two-Lane_Poor_Rolling 25 2.10 72.93
Two-Lane_Good_Rolling 100 1.53 72.19
Two-Lane_Fair_Rolling 75 1.67 71.20
#2
Two-Lane_Mediocre_Rolling 50 1.84 67.94
Two-Lane_Poor_Rolling 25 2.03 63.05
Two-Lane_Good_Very Rolling 100 1.63 64.04
Two-Lane_Fair_Very Rolling 75 1.77 63.86
#3
Two-Lane_Mediocre_Very Rolling 50 1.94 63.28
Two-Lane_Poor_Very Rolling 25 2.12 61.18
Source: Reference [43].
Routes #1 and #2 are predominantly made of two-lane and rolling terrain and Route #3 is
predominantly made of two-lane and very rolling terrain. For each type of lane and terrain, pavement
conditions may be good, fair, mediocre and poor and they were related to the PCI with the values 100,
75, 50 and 25, respectively.
Table 3 shows that, as pavement conditions worsen, with a decrease in the PCI, the cost per
distance increases due to fuel, lubricant and tire consumption and to maintenance parts. Cost per
time, which is related to workforce, depreciation and crew costs, decreases as the PCI also decreases.
This may be explained by the fact that, when pavement conditions worsen, vehicle speed decreases
and, consequently, travel time increases. Therefore, associating fixed costs with a longer travel time,
costs per time are lower when the PCI decreases.
The third parameter of Equation (3) is the toll cost. The values shown in Table 4 refer to the toll in
each route and were retrieved from the website of the concessionaire [56].
Table 4. Toll fares in the analyzed routes for a truck of 7 axles.
Municipality Location Value (R$)

Route #2 Aracruz Highway BR 101, Km 171.7 30.00
Route #3 Serra Highway BR 101, Km 242 28.80
The last parameter of Equation (3) is the fine cost, related to the fines for driving overloaded
vehicle, for example. However, this cost was disregarded in this case study because we do not have
access to this kind of data and also because the scales are not operating continuously on the roads
under study. Besides, the perception weights (β1 , β2 , β3 and β4 ) are parameters difficult to be defined.
In our case study, we used one for all of them.
The Inflation rate determines the growth rate of operational costs in transportation, as well as the
other costs in other modules of the model, such as traffic accident and pavement maintenance costs.
The Extended Consumer Price Index—IPCA, produced by the Brazilian Institute of Geography and
Statistics, is the country’s official inflation index. In 2016, the IPCA had a growth of 6.29% [57], thus,
Sustainability2019,
this percentage 2019,11,
11,xxFOR
FOR PEER REVIEW 2222ofof37
3737
Sustainabilitywas
2019,used for the initial inflation rate.
Sustainability
Sustainability 2019, 11, PEER REVIEW
11, xx FOR
FOR PEER
PEER REVIEW
REVIEW 22
22 of
of 37
4.4.Parameters
Parameters
4.4. Parameters
4.4. for the forforthe
thePavement
Pavement Pavement Maintenance
Maintenance Module
ModuleModule
Maintenance
4.4. Parameters for the Pavement Maintenance Module
For the For
Formodulethemodule
the module
“Pavement “Pavement Maintenance,”
Maintenance,”
“Pavement Maintenance,” ititis
isisnecessary
it is necessary necessary totodetermine
to determine
necessary determine
determinethe AADT theAADT
the AADT
of the of ofthe
base thebase
base
For the module “Pavement Maintenance,” it to the AADT of the base
year usedyear
year inused
each inofeach
the of the
three three
routes. routes. Considering
Considering that that
each each
route route
is is subdivided
subdivided into into
two two links,
links, the
year used
used in in each
each of of the
the three
three routes.
routes. Considering
Considering that that each
each route
route is is subdivided
subdivided into into twotwo links,
links, the
the
calculation
the calculation of of ofthe
the the Design
Design ESALESAL usedused
the the highest
highest AADTAADT each of each route,which
which shown
is shown inTable
Table 2.The The
calculation
calculation of the Design
Design ESAL
ESAL used
used the
the highest
highest AADTofof
AADT eachroute,
of each route,which
route, whichisis showninin
is shown inTable
Table2.2. 2. The
The AADT AADT
AADT Growth
Growth
Growth rate rate
was
rate was
was obtained
obtained
obtained from from
fromthe thethe“Speed
“Speed “Speed
Flow” Flow”
module
Flow” module
module and the and
and theFleet
Fleet
the Fleetvehicle
vehicle vehicle factorwas
factor factor
was was
AADT Growth rate was obtained from the “Speed Flow” module and the Fleet vehicle factor was
obtained obtained
from the frommodelthe model
itself as itself
theas asthe
sum the sum ofthe
the Vehicle Factors ofeach
each category inthe
the scenario in
obtained
obtained from
from the
the model
model itself
itself as theofsum the of
sum Vehicle
of Factors
the Vehicle
Vehicle of eachof
Factors
Factors ofcategory in the in
each category
category scenario
in in
the scenario
scenario in
in
which
whichwhich
the
which thepercentage
percentage
the percentage
of ofoverweight
overweight
of overweight
per axle per
per is axleisisnull.
null.
axle null.
the percentage of overweight per axle is null.
For theFor For the calculation
calculation
For the calculation
of the base ofofyear’s
the base
the base
Design year’s
ESAL,
year’s Design
it wasESAL,
Design ESAL, itit was
necessary
ESAL, was
to necessarythe
differentiate
necessary toto percentage
differentiate the
differentiate the
the calculation of the base year’s Design it was necessary to differentiate the
of each percentage
category of ofeach
each category ofheavy
heavy vehicles inthethe vehicle fleet based on the samples of Traffic
percentage
percentage of heavy
of vehiclesof
each category
category in
of the
heavy vehicle
vehicles
vehiclesfleetin based
in on thefleet
the vehicle
vehicle samples
fleet basedofon
based onTraffic
the Count and
the samples
samples of
of Traffic
Traffic
Countand
Classification
Count and
surveys Classification
[50]
Classification carried surveys
out
surveys at [50]
the
[50] carried
links
carried outatat
under
out thelinks
analyze.
the linksThe under
under analyze.
following
analyze. fourThe
The followingfour
categories
following four
Count and Classification surveys [50] carried out at the links under analyze. The following four
categories
representing
categories heavy representing
vehicles wereheavy
chosenvehicles
to were
calculate chosen
the to
Design calculate
ESAL: the
buses,Design
light ESAL:
trucks, buses, light
semitrailer trucks,
categories representing
representing heavy heavy vehicles
vehicles werewere chosen
chosen to to calculate
calculate the the Design
Design ESAL:ESAL: buses,
buses, lightlight trucks,
trucks,
semitrailerand
and multitrailer.
semitrailer andmultitrailer.
multitrailer.
semitrailer and multitrailer.
Therespective
The respective
The respective representations
representations
representationsand the and
andtotalthe
the totalnumber
number
total number
of axlesofoffor
axles
axleseach for
for eachselected
selected
each selected vehicle
vehiclevehicle
category
vehicle category
category
The respective representations and the total number of axles for each selected category
are shownare shown
in Table in 5.
Table
The 5.The
The weights legally allowed for each type of axle, considering the tolerance, are
are
are shown
shown in
in Table
Table 5.weights
5. The weightslegally
weights allowed
legally
legally allowed
allowedfor each
for
for eachtypetype
each of axle,
type of
of axle, considering
axle, considering
considering thethe
thetolerance,
tolerance,
tolerance, are are
are shownshown
shown in inTable
Table
in Table
6. 6.
6.
shown in Table 6.
Table 5. Configuration
Table
Table of the representative
5.5.Configuration
Configuration
Configuration ofofthe
the categories
therepresentative
representative
representative used toused
categories
categories calculate
used DesignDesign
totocalculate
calculate
calculate Traffic.
DesignTraffic.
Traffic.
Table
Table 5.
5. Configuration of
of the representative categories
categories used
used to
to calculate Design
Design Traffic.
Traffic.
Representation
Representation
Representation Category
Category
Category Configuration of Axles
Configuration
Configuration Axles Axles
ofAxles
of Axles
Axles
Representation Category Configuration of Axles Axles
Single
Single
Single tyres
tyres
tyrestyres
on onaxle
on
front frontaxle
front axle
Bus
Bus (B)
Bus(B)
(B) Single on front axle 2 222
Bus (B) Twin
TwinTwin tyres
tyrestyres
on on
on
rear rearaxle
rear
axle axle
Twin tyres on rear axle
Lighttruck
Light truck Singletyres
Single tyreson onfront
front axle
Light truck SingleSingle
tyrestyres front axle
on axle
on front axle 2
Light truck (LT)
(LT)
(LT) Twin
Twin tyres
tyres onon rearaxle
rear axle 2 22
(LT) TwinTwintyrestyres
on rearonaxlerear axle
Singletyres
Single tyreson onfront
frontaxle axleofoftractor
tractor
Single
Single tyrestyres on front
on front axle ofaxle of tractor
tractor
Twin
Twin tyres
tyres onon both
both rear
rear axles
axles
of of
of tractor
Semitrailer
Semitrailer Twin tyreson
Twin tyres onboth
bothrearrear axles
axles of tractor
tractor
Semitrailer
Semitrailer (ST) (doubletandem)
(double
tractor (double tandem) 666
(ST)
(ST) (doubletandem)
tandem) 6
(ST) Twintyres
Twin
Twin tyres
tyres on
onon three
three
threerear rear
axles
rear axles
of of
axles oftrailer
trailer
Twin tyres on three rear axles of trailer
trailer (triple tandem)
(triple tandem)
(triple
(triple tandem)
tandem)
Single tyrestyres
Single
Single on front
tyres axle
onfront
on front ofaxle
tractor
axle ofoftractor
tractor
tractor
Single
Twin tyres
tyres
onon
on
both
front
rearrear
axleofof
axles
Multitrailer
Multitrailer Twin
Twin tyres
tyres on both axles of tractor
Multitrailer Twin on both
tyres(double rear
rear axles
bothtandem) axles ofof tractor
tractor 7
Multitrailer
(MT)
(MT)
(MT) tractor
(doubleoftandem)
7 77
(MT) Twin tyres(double
on axles tandem)
(double tandem)
trailer
Twintyres
Twin tyreson onaxles
axles ofoftrailer
trailer(double
trailer (doubletandem)
tandem)
Twin tyres(double
on axles tandem)
of (double tandem)
Source:
Source:
Source:Source: Reference
Reference
Reference
Reference [58]. [58].
[58].
[58].
To decide about the type Table

Table
Table 6.6.Weight
6. Weightlimits,
Weight
of maintenance limits,
limits, considering
considering
considering
that thetolerance
the
the tolerance
tolerance
will be performed ofof
onof 10%
10%
10%
the peraxle.
per
per axle. the limits for
axle.
pavement,
Table 6. Weight limits, considering the tolerance of 10% per axle.
corrective and restorative maintenances are
WeightAllowed
Weight defined
Allowed in function of the IRI, recommending an IRI value
per
per
Weight Allowed per Total interventions
Total (+ Tolerance10%)10%) Weight per Axle
higher or equal to 3.5 as a “trigger”Axle
for pavement
Axle Group(t)
Group Total (+
(t)maintenance (+ Tolerance
Tolerance 10%)[61]. It isWeight per
per Axle
recommended
Weight Axle
Axle Group (t)
to correlate these indexes
Single tyres with the PCI by means
6.00 of a subjective assessment,
6.60 as shown in Table 7.
6.60
Single
Single tyres
tyres 6.00
6.00 6.60
6.60 6.60
6.60
Twintyres
Twin tyres 10.00
10.00 11.00
11.00 11.00
11.00
Twin tyres 10.00 11.00 11.00
Doubletandem
Double tandem 17.00
17.00 18.70
18.70 9.35
9.35
Double tandem 17.00 18.70 9.35
Tripletandem
Triple tandem 25.50
25.50 28.05
28.05 9.35
9.35
Triple tandem 25.50 28.05 9.35
Source:Authors
Source:
Source: Authorsbased
Authors basedon
based onReferences
on References[59,60].
References [59,60].
[59,60].
Source: Authors based on References [59,60].
To decideabout
To about thetype
type ofmaintenance
maintenance thatwill
will beperformed
performed onthe
the pavement,the
the limitsfor
for
To decide
decide about the
the type of
of maintenance that
that will be
be performed on
on the pavement,
pavement, the limits
limits for
Table 6. Weight limits, considering the tolerance of 10% per axle.
Weight Allowed per

Total (+ Tolerance 10%) Weight per Axle
Axle Group (t)
Single tyres 6.00 6.60 6.60
Twin tyres 10.00 11.00 11.00
Double tandem 17.00 18.70 9.35
Triple tandem 25.50 28.05 9.35
Source: Authors based on References [59,60].
Table 7. Relationship between the pavement performance index and the irregularity indexes.
Irregularity Condition Interventions Effects

Assessment
IRI (m/Km) PCI (points) Maintenance PCI (points)
Very Good 1–1.9 100 - -
Good 1.9–2.7 80 - -
Fair 2.7–3.5 60 - -
Mediocre 3.5–4.6 40 Corrective 80
Poor >4.6 20 Restorative 100
Source: Adapted by Reference [61].
Thus, the limits for corrective and restorative maintenance interventions in relation to the PCI are
of 40 and 20 points, respectively. In case the corrective maintenance is carried out, the PCI increases to
80 points. If restorative maintenance is required, the PCI level increases to 100 points.
Finally, we do not have data costs of the pavement condition surveys but they are not carried out
every year. Thus, it was chosen to simulate the time interval between two PCI surveys, varying from 2
to 5 years and the duration of maintenance works, also varying from 2 to 5 years.
4.5. Parameters for the Social Cost of Freight Transportation Module

The “Social Cost of Freight Transportation” module is divided into pavement maintenance cost
and traffic accident costs. To determine pavement maintenance costs, it is necessary to determine
corrective maintenance cost per kilometer (CM/Km) and restorative maintenance cost per kilometer
(RM/Km). For that, DNIT published the average management costs for the road mode, as shown
in Table 8.
Table 8. Pavement maintenance costs.
Maintenance Lower Limit (R$/Km) Upper Limit (R$/Km) Average (R$/Km)

Corrective 566,577.08 1,761,708.99 1,164,000.00
Restorative 1,783,916.27 2,883,935.73 2,334,000.00
For road accident costs, the Institute of Applied Economic Research—IPEA [63] presented an
overall characterization of accidents in Brazilian federal roads, analyzing the evolution of their costs
based on data from the Federal Road Police—PRF for the base years 2007, 2010 and 2014. The CNT [64]
updated the study based on the number of accidents registered in 2015 and on the inflation for the
period, as shown in Table 9.
Table 9. Total cost and average cost per accident severity.
Severity Average Cost (R$/occurrence)

Accident with death (AD) 715,772.55
Accident with victims (AV) 99,805.21
Accident with no victims (ANV) 25,523.79
Costs with traffic accidents are associated with people, vehicle or property damage and institutional
costs. Costs associated with victims regarding hospital costs and loss of production, that is, how much
income traffic victims stop earning both during the period they are away from economic activities
and, in case of death, regarding their life expectancy. Costs associated with vehicles depend on vehicle
category (cars, motorcycles, bicycles, vans, trucks, buses and others) but are represented by removal,
property damage and freight loss. Finally, institutional costs refer to costs with care and property
damage [63].
To define the percentage of each type of accident according to severity, data from PRF [65] were
used which consider the accidents that occurred in the links of federal roads belonging to the routes
analyzed in the year 2016, as shown in Table 10.
Table 10. Number and percentage of accidents according to severity.
Number Percentage Accidents average/Km Estimate

Highway Severity Total
of Accidents (%) Federal Link State Link
AD 13 11 0.3 36 49
BR 259 0 AV 89 75 2.0 243 332
(Link #1.2) ANV 17 14 0.4 46 63
TOTAL 119 100 2.6 325 1 444
AD 15 3 0.1 9 24
BR 101 0 AV 367 82 3.3 217 584
(Link #2.2) ANV 65 15 0.6 38 103
TOTAL 447 100 4.1 264 2 711
AD 34 4 0.5 - 34
BR 101 0 AV 641 76 10.0 - 641
(Route #3) ANV 173 20 2.7 - 173
TOTAL 848 100 13.3 - 848
Note: 1 Estimate of accidents for link #1.1 and 2 Estimate for link #2.1. (Source: Reference [65].)
The estimate of the number of accidents in the links of state highways of Routes #1 (Link #1.1) and
#2 (Link #2.1) was made using the average of accidents per kilometers in federal highways, multiplied
by the length of the state highway link. Next the number of accidents of federal road links was added
to the estimated number of accidents in state road links.
The accident prediction model proposed by the HSM [45] suggests a correction factor when the
road conditions are not ideal. To find out the correction factor, the model was simulated considering a
scenario in which road and traffic conditions are ideal, that is, accident prediction was based only on
AADT and route length. Next, the real number of accidents was divided by the number of accidents
under ideal conditions to find out that the average correction factor for the accident prediction model
of each route was 8.5, that is, the real number of accidents is in average 8.5 higher than the number
predicted by the model under ideal conditions.
5. Results and Discussions

The simulation period was of 15 years, corresponding to the average pavement duration time
and is compatible with the economic analysis needs regarding the interval between 2016 and 2030.
After 15 years, in normal conditions, a surface renovation would be expected [66]. Table 11 summarizes
the proposed scenarios, the simulated variables and the impact analyzed in each scenario, which are
detailed below.
Table 11. Description of scenarios.
Scenario Simulated Variables Impact

Time control in pavement assessment
1. Reference (2, 3, 4 and 5 years) Maintenance;
Output flow delay due to maintenance Cumulative cost of pavement maintenance
(2, 3, 4 and 5 years)
Cumulative cost of transportation,
2. Strict, Moderate Overweight percentage from 0 to 100%
pavement maintenance and traffic accidents;
and Tolerant Policies (Increment of 10%)
Total cumulative cost
3. Assessment of the Overweight percentage and Weight of
Total cumulative cost
Best Policy social cost
Cumulative cost of pavement maintenance;
4. Overweight in
Overweight percentage cumulative cost of transportation;
all routes
cumulative cost of accidents
In Scenario 1 (Reference Scenario), different values were simulated for time control in pavement
assessment with intervals of 2 to 5 years between them and the delay in the effect of maintenance,
that is, the duration in years of maintenance works, also with intervals of 2 and 5 years. The analyzed
effects refer to the number of maintenance requests and over cumulative pavement maintenance costs.
Scenario 2 was also simulated and compared to the Reference Scenario. Following the notation
by References [37,38], three loading policies were adopted considering a variation of zero to 100% of
overweight: Strict Policy (no overweight admitted); Moderate Policy (admits up to 50% of overweight);
and Tolerant Policy (admits between 50% and 100% of overweight). The scenario with 100% overweight
was hypothetical for simulation purposes. However, we note that there were already registrations of
up to 70% in groups of axles and up to 57% in TGW. This implies in reducing road safety and other
studies are recommended for a deeper analysis of these conditions.
Scenario 3 simulated variations of overweight percentage in vehicles and variations of the
importance attributed to social costs simultaneously. With this, it was possible to identify the best
vehicle loading policy (overload percentage) under different levels of importance attributed to social
costs in relation to operational costs based on the minimization of the system’s total costs.
Finally, Scenario 4 addressed the practice of overweighting in all analyzed routes, regardless of
the presence or operation of scales. This scenario was simulated to assess the impact over all costs
considered in the model. The results of the proposed scenarios are presented as follows.
5.1. Reference Scenario

As presented before, the parameters simulated in the Reference Scenario are time controls in
pavement assessment, that is, the interval in years between the gathering of pavement conditions to
assess the need for maintenance and the delay in the output flow of the effect of maintenance or the
duration of maintenance works. The variables impacted are Maintenance and Cumulative cost of
pavement maintenance.
The delay in the output of the effect of maintenance regards the time spent in carrying out the
maintenance works to increase the PCI. The reference used to define this maintenance time was the
Program of Restoration and Maintenance Contracts—CREMA, which is subdivided into CREMA 1st
step, with a duration of two years and CREMA 2nd step, with a duration of five years [67].
It was assumed that the assessment of pavement conditions is not carried out when a maintenance
operation is in progress. This way, if the maintenance takes two years to be completed (delay),
the assessment may be carried out in intervals of 2, 3, 4 and 5 years. On the other hand, if the
maintenance has a delay of 3 years to be concluded, the assessment will not be carried out every 2
years, but it may be carried out in intervals of 3, 4 and 5 years. Table 12 shows the amount and type
of maintenance to be carried out and the total maintenance costs in each route after the simulation
experiments that comprehended, as already mentioned, the period between 2016 and 2030.
Table 12. Maintenance and Total cost per route for each combination of time intervals of pavement
maintenance and assessment.
Assessment Total Cost Total Cost Total Cost

Delay Maint. * Maint. * Maint. *
Pavement Maintenance Maintenance Maintenance
(Years) Route #1 Route #2 Route #3
(Years) #1 (R$) #2 (R$) #3 (R$)
2 R and 1
2 4R 1,658,018,037 2,427,611 4R 182,418,339
C
2 3 4R 1,010,624,854 2R 1,272,970 4R 84,030,351
1 R and 2
4 3R 878,864,580 2,480,086 3R 74,492,566
C
5 2R 388,020,604 2R 1,575,210 2R 28,275,918
1 R and 1
3 2R 251,825,934 788,948 2R 17,472,238
C
3
4 3R 876,486,084 2R 1,703,138 3R 74,492,573
1 R and 1
5 2R 387,335,132 1,522,382 2R 28,275,918
C
4 2R 724,094,260 2R 1,703,138 2R 64,026,375
4
5 2R 385,919,132 1R 480,876 2R 28,275,918
5 5 1R 43,413,788 1R 480,876 1R 2,747,492
Note: * In this column, the letter “R” represents “Restorative” maintenance and the letter “C,”
“Corrective” maintenance.
Costs vary according to three characteristics: maintenance type; route length; and percentage of
stone transportation vehicles in the analyzed route. For example, for a delay of 2 years in the effect of
maintenance and an interval also of 2 years between pavement assessments, the following are required:
4 restorative maintenances in Route #1; 2 restorative maintenances and 1 corrective maintenance in
Route #2; and 4 restorative maintenances in Route #3. Although Routes #1 and #3 receive the same
amount of restorative maintenances, the cost is different due to the difference in route length and the
percentage of stone transportation vehicles in relation to the total of heavy vehicles.
Furthermore, a route may also present different costs with the same amount of a same type
of maintenance, as occurred with Route #1 under a delay of 2 years and intervals of 2 and 3 years
between pavement assessments. In both cases, Route #1 required 4 restorative maintenances and,
even so, the total maintenance cost in the period was different due to the different percentages of stone
transportation vehicles that contribute to pavement degradation and its maintenance cost.
By analyzing the same intervals of pavement assessment, as maintenance delay increases,
the number of maintenance requirements and, consequently, its cost decreases. The minimum pavement
maintenance cost in all routes happens with the biggest interval of time between pavement assessments
and the delay of the effect of maintenance. However, managers should decide about the trade-off
between maintenance cost reduction and pavement performance index, which measures trafficability
conditions in the road network. The bigger the time interval between pavement assessments, the lower
the maintenance costs but also the worse the pavement conditions, as shown in Figure 13.
Figure 13 shows two extremes of time intervals simulated for pavement assessment and carrying
out of maintenance in Route #1. In the first case (Figure 13a), the intervals are of two years with the
conduction of four restorative maintenances and, in the second case (Figure 13b), the intervals are of
five years for both variables with the conduction of only one restorative maintenance. Table 12 makes
it clear that the difference in maintenance costs is of 97%, however, PCI remains below the limit for
restorative maintenance under the established traffic conditions. In 2026, for example, PCI is near to
zero, which means that traffic may occur but in extremely poor conditions.
maintenance cost in all routes happens with the biggest interval of time between pavement
assessments and the delay of the effect of maintenance. However, managers should decide about the
trade-off between maintenance cost reduction and pavement performance index, which measures
trafficability conditions in the road network. The bigger the time interval between pavement
assessments,
Sustainability the
2019, 11,lower
3128 the maintenance costs but also the worse the pavement conditions, as shown
25 of 35
in Figure 13.
(a) (b)
Sustainability
Figure2019, 11, x FOR PEER REVIEW 27 of 37
13. Value of PCI in Route #1 for an interval of two years (a) and of five years (b) in pavement
assessment and
andthethe duration
duration of maintenance
of maintenance works.works.
(Source:(Source:
Own workOwn work
using using the
the VENSIM VENSIM
application).
Due to the great possibility of combining scenarios with simulated parameters, the remaining
application.)
scenarios
Due used
to thefixed
greatvalues based of
possibility oncombining
the analysesscenarios
of the reference scenario. parameters,
with simulated For the delaythe in the output
remaining
of theFigure
effect
scenarios of
used maintenance
fixed values or duration
based on theof maintenance
analyses of works,
the the
reference reference
scenario.
13 shows two extremes of time intervals simulated for pavement assessment and carrying used
For was
the CREMA
delay in 1st
the
Step,
output
out with a duration
of the effect of
of maintenance of two
inmaintenance years
Route #1. In or [67]. The
theduration
first case time
of(Figure interval
maintenance to gather
works,
13Error! pavement
the reference
Reference sourceused conditions
was CREMA
not found.a), was
the
considered
1st Step, with
intervals areto of
be three
a two years.
duration
years of Thus,
twothe
with the
years base
[67].scenario
conduction The wasinterval
of time
four simulatedto with
restorative gather these parameters
pavement
maintenances in and
the asecond
and,conditions fixed
was
overweight
considered
case (Figureto percentage
13Error! aiming
be threeReference
years. tosource
Thus, enable later
the base
not comparisons
scenario
found.b), was with theare
the simulated
intervals remaining
with of these scenarios
parameters
five years for
for both anddifferent
a fixed
variables
loading
overweight
with policies.
the conduction Figure
percentage 14
of aiming
only oneshows the
torestorativecumulative
enable latermaintenance. costs
comparisonsTable with
with thetransportation
remaining
12 makes (operational
scenarios
it clear that the cost),
fordifference
different
pavement
loading
in maintenance
policies.
maintenance costs and
Figureis14 traffic
ofshows
97%, theaccidents
cumulative
however, (social
PCI costscosts),
remains with as well
theaslimit
transportation
below the total costs forcost),
(operational
for restorative the Reference
pavement
maintenance
Scenario.
maintenance
under and traffic
the established accidents
traffic (social In
conditions. costs),
2026,asfor
well as the total
example, PCI costs
is near fortothe Reference
zero, Scenario.
which means that
traffic may occur but in extremely poor conditions.
Figure 14.
Figure Cumulativeeconomic,
14. Cumulative economic,social
socialand
andtotal
totalcosts
costsin
inthe
the base
base scenario.
scenario.
ItIt can
can be
be seen
seen that
that the
the operational
operational cost
cost with
with transportation
transportation is is more
more expressive
expressive than
than the
the sum
sum of
of
social costs
social costs with
with pavement
pavement maintenance
maintenance andand traffic
traffic accidents,
accidents, representing
representing 63%63% of
of the
the cumulative
cumulative
total costs in the period between 2016 and 2030. Nevertheless, it is worth highlighting
total costs in the period between 2016 and 2030. Nevertheless, it is worth highlighting that social that social
costs
costs are undersized, since only pavement maintenance and traffic accident costs
are undersized, since only pavement maintenance and traffic accident costs were considered. Other were considered.
Other important
important costs werecostsnot
were not considered,
considered, amongamong
them, them, the delay
the delay of vehicle
of vehicle trips trips
due to due to accidents
accidents and
and road blocks stands out, since it has several consequences for those involved,
road blocks stands out, since it has several consequences for those involved, such as the need for such as the need
for companies
companies to keepto keep security
security stocks
stocks andrisk
and the the ofrisk of degradation
degradation and environmental
and environmental pollution
pollution with
accidents involving vehicles with hazardous loads. The next scenarios show the impact ofof the
with accidents involving vehicles with hazardous loads. The next scenarios show the impact the
percentage of overweight over system
percentage of overweight over system costs. costs.
5.2. Strict, Moderate and Tolerant Policies

In the Strict Policy scenario, it was assumed that overload, both in TGW and on axles, is null,
that is, the legal weight limits would be followed by freight carriers. In this case, the costs are shown
in Figure 15.
5.2. Strict, Moderate and Tolerant Policies

In the Strict Policy scenario, it was assumed that overload, both in TGW and on axles, is null,
that is, the legal weight limits would be followed by freight carriers. In this case, the costs are shown
in Figure 15.
Figure 15. Cumulative economic, social and total costs in the Strict Policy scenario.
The
The legal
legal imposition
imposition regarding
regarding weight
weight limits
limits in
in freight
freight vehicle
vehicle loading
loading increases
increases thethe operational
operational
cost The legal imposition regarding weight limits in freight vehicle loading increases the operational
cost of
of transportation,
transportation, once once thisthis cost
cost isis subdivided
subdivided by by the
the load
load transported,
transported, preventing
preventing the the economy
economy
cost
of of transportation,
scale achieved with once this costThus,
overloading. is subdivided
despite by legal
the the loadloadtransported,
limit of the preventing
vehicle is the economy
respected, the
of scale achieved with overloading. Thus, despite the legal load limit of the vehicle is respected,
of scale
operational achieved with
cost cost overloading.
of transportation Thus,
increases despite
because the legal
a larger load
numberlimit of the vehicle
of vehicles is respected,
are required the
to carry
the operational of transportation increases because a larger number of vehicles are required to
operational
the load. cost of transportation increases because a larger number of vehicles are required to carry
carry the load.
the load.
On
On thethe other
other hand,
hand, social
social costs
costs with
with maintenance
maintenance and and accidents
accidents are are reduced
reduced in in relation
relation to to the
the
On
Reference the other hand, social costs with maintenance and accidents are reduced in relation to the
ReferenceScenario,
Scenario,asasshownshowninin Figure
Figure 16.16.
In the maintenance
In the maintenance of theofpavement,
the pavement,the reduction
the reductionof costs of
Reference
occurs in Scenario,
this as shown
scenario because, in Figure
despite 16.the
In the maintenance
increase in the of the pavement,
number of axles the reduction
passages, due oftocosts
the
costs occurs in this scenario because, despite the increase in the number of axles passages, due to the
occurs ininthis
increase the scenario
number of because,
of trips,ititdespite
canbe the increasethat
beconcluded
concluded in the
the weight
number ofthe axles passages, due to the
increase in the number trips, can that the weight of of axles
the axles is more
is more relevant
relevant in
in the
increase
the in the
deterioration number
of the of trips,
roads. it can
Likewise, be concluded
accidents that the weight of the axles is more relevant in
deterioration of the roads. Likewise, accidents areare influenced
influenced by bythethe number
number of vehicles
of vehicles traveling
traveling on
the
on deterioration of the roads. Likewise, accidents are influenced by the number of vehicles traveling
thethe highway,
highway, by by
thethe excess
excess weight
weight andand
by by
thethe traffic
traffic conditions
conditions of of
thethe highways,
highways, which
which areare reduced
reduced by
on
by the highway, by the excess weight and by the traffic conditions of the highways, which are reduced
thethe practice
practice of overweighed
of overweighed vehicles.
vehicles.
by the practice of overweighed vehicles.
Figure 16. Comparison of costs between base scenario and strict policy.
The operational cost with transportation increases by 45%, costs with maintenance and traffic
The operational
accidents reduce by cost
41%with
andtransportation increases
12%, respectively and by 45%,
total costs
cost with maintenance
increases by 18%. Itand traffic
is worth
accidents reduce
highlighting by 41%should
that studies and 12%, respectively
be carried andtototal
out aiming cost the
measure increases
numberbyof18%. It is caused
accidents worth
highlighting that studies should be carried out aiming to measure the number of accidents caused
The operational cost with transportation increases by 45%, costs with maintenance and traffic
accidents reduce by 41% and 12%, respectively and total cost increases by 18%. It is worth highlighting
that studies should be carried out aiming to measure the number of accidents caused exclusively by
the vehicles
Sustainability that
2019, 11,transport
x FOR PEER ornamental
REVIEW stones, considering that the number of accidents in the model 29 of 37is
estimated by the percentage of volume of these vehicles and the total number of accidents in the road,
accidents in the road, without
without differentiation differentiation
of the type of the type
of vehicle involved. of vehicle
Moreover, it is involved. Moreover,
also necessary it is also
to attribute then
necessary
importance to attribute
to each type thenofimportance
cost in ordertoto
each typeon
decide of the
costbest
in order to decide on the best scenario.
scenario.
In
Inthe
theModerate
ModeratePolicy,
Policy,the
thepercentage
percentageofofoverload
overloadvaries
variesfrom
from10%10%toto50%50%and
andininthe
theTolerant
Tolerant
Policy,
Policy,ititvaries
variesfrom
from60% 60%toto100%.
100%.InInthis scenario
this overweight
scenario overweight was gradually
was graduallyincreased, from
increased, 10%10%
from to
100%, in order
to 100%, to separately
in order assess
to separately the the
assess behavior of each
behavior costcost
of each andand
the the
totaltotal
system cost,cost,
system as shown
as shown in
Figure 17.
in Figure 17.
Figure 17. Comparison of the costs with the Moderate and Tolerant Policies.
The costs presented regard the cumulative cost at the end of the simulation period, that is,
The costs presented regard the cumulative cost at the end of the simulation period, that is, they
they are the total costs between 2016 and 2030. It can be seen that, as the overweight percentage
are the total costs between 2016 and 2030. It can be seen that, as the overweight percentage increases,
increases, the operational cost of transportation decreases, that is, the unitary cost per tons of load
the operational cost of transportation decreases, that is, the unitary cost per tons of load carried,
carried, which confirms the economy of scale achieved by the practice of overweight in transportation.
which confirms the economy of scale achieved by the practice of overweight in transportation. The
The minimum transportation cost occurs with an overweight percentage of 100%.
minimum transportation cost occurs with an overweight percentage of 100%.
On the other hand, the minimum social costs happen with an overweight percentage of 10% (but
On the other hand, the minimum social costs happen with an overweight percentage of 10% (but
they are even lower under the Strict Policy) and increase as overweight increases, as shown in Figure 18.
they are even lower under the Strict Policy) and increase as overweight increases, as shown in Figure
While the reduction in total system cost may reach 23% between the Strict Policy and the Tolerant
18.
Policy, the variation in social costs with the increase in overweight is not so significant. In the case of
pavement maintenance costs, that happens because there are limiting factors to carry out maintenance
works, such as the limits of pavement condition [61] in function of the IRI, the time intervals between
the survey of such conditions, which may take from two to five years, as well as the duration of
maintenance works, that is, the maintenance cost increases within these limits and not exclusively in
function of the overweight being practiced.
Therefore, although pavement performance decreases as the overweight practiced by the
transportation of ornamental stones increases, an overweight of up to 60% only in these vehicles would
not lead to a significant increase in maintenance costs, considering the limiting factors presented and
the case study in which only ornamental stone vehicles travel with excess weight, that is, the excess
weight of the other freight vehicles that travel along the road was not assessed in this study. In the case
of traffic accidents, deeper studies should be carried out regarding the occurrence of accidents and the
cause exclusively related to overweight in ornamental stone vehicles.
Figure 18. Social costs with the Moderate and Tolerant Policies.
The costs presented regard the cumulative cost at the end of the simulation period, that is, they
are the total costs between 2016 and 2030. It can be seen that, as the overweight percentage increases,
the operational cost of transportation decreases, that is, the unitary cost per tons of load carried,
which confirms the economy of scale achieved by the practice of overweight in transportation. The
minimum transportation cost occurs with an overweight percentage of 100%.
However,
On the otherdecision makers
hand, the minimummustsocial
consider the importance
costs happen of each cost
with an overweight category of
percentage for society.
10% (but
Therefore, the next scenario assesses the costs based both on the variation in overweight and
they are even lower under the Strict Policy) and increase as overweight increases, as shown in Figure on the
importance of social costs in relation to the operational cost of transportation.
18.
5.3. Assessment of the Best Policy
The variation in the relative importance of the social cost in relation to the operational cost must
be considered by decision makers as a criterion to assess the best policies to be adopted based on the
minimization of total costs. Table 13 shows total system costs, considering the importance of social cost,
ranging from 0 (not important) to 100 (extremely important) for each loading policy (strict, moderate
and tolerant).
Table 13. Cumulative total cost considering a variation in overweight percentage and the importance
of social cost in relation to the operational cost (R$ millions).
Equally
Importance of Social Cost Not Important Little Important Very Important Extremely Important
Important
Weight of social cost 0 10 20 30 40 50 60 70 80 90 100
Strict Policy 0% 62.4 58.0 53.6 49.1 44.7 40.2 35.8 31.4 26.9 22.5 18.0
10% 58.9 54.9 50.8 46.9 42.8 38.9 34.8 30.8 26.8 22.8 18.8
20% 54.6 51.2 47.7 44.2 40.7 37.3 33.8 30.4 26.9 23.4 19.9
Moderate Policy 30% 50.9 47.8 44.8 41.7 38.6 35.5 32.5 29.4 26.3 23.2 20.1
40% 47.6 44.9 42.2 39.5 36.7 34.0 31.2 28.5 25.8 23.0 20.3
50% 44.8 42.4 39.9 37.5 35.0 32.6 30.1 27.7 25.2 22.8 20.3
60% 42.4 40.2 38.0 35.8 33.6 31.4 29.2 27.0 24.8 22.5 20.3
70% 40.3 38.5 36.7 34.9 33.1 31.4 29.6 27.8 26.0 24.2 22.5
Tolerant Policy 80% 38.4 37.3 36.1 35.0 33.8 32.7 31.6 30.4 29.3 28.2 27.0
90% 36.6 35.7 34.7 33.7 32.8 31.8 30.9 29.9 28.9 28.0 27.0
100% 35.0 34.2 33.4 32.6 31.8 31.0 30.2 29.4 28.6 27.8 27.0
Note: The highlighted cells indicate the lowest values of the respective columns.
The results in Table 13 show that, for the case in which social costs have lower importance than
the operational cost or they are equally important, the best alternative is 100% overweight to reduce
the system’s total cost. For a relative importance of social costs ranging from 60% to 80%, total costs
are minimum with freight vehicles being 60% overweighed. Only in the scenario in which social costs
are extremely important in comparison to the operational costs, the minimum total cost is identified in
the strict policy, in which the legal weight limits are respected.
It is up to decision makers, again, to attribute due importance to each analyzed cost category.
Nevertheless, considering that pavement deterioration in the analyzed road links affects the country’s
logistics infrastructure and the competitiveness of the national production and that costs with accidents
are assumed to be null in an ideal scenario, social costs should be considered extremely important to
the interests of the society as a whole and not only to a reduced group of private companies.
5.4. Overweight in All Routes

The purpose of this scenario is to assess again the behavior of the costs considered in the model
under the condition of overweight being practiced in all possible routes used to transport ornamental
stones production, regardless of the existence of scales and their use for inspection and considering the
absence of a toll station in Route #1.
This scenario is likely to occur in practice because, despite the existence of the weighing scale
on Route #2, it does not work all the time. For that, the variable “Overload Percentage” was added
to vehicle loading in Route #2. Table 14 shows the operational costs with transportation, pavement
maintenance and traffic accidents, as well as the cumulative total cost for each percentage of overweight
practiced in both possible routes for the transportation of ornamental stones.
Table 14. Costs in the scenario in which all routes have overloading (R$).
Cost of Cost of Cost of
Scenario Overweight Total
Transportation Pavement Maintenance Traffic Accidents
Strict Policy 0% 62,444,356.00 8,072,306.00 9,956,531.00 80,473,193.00
10% 56,880,892.00 6,418,579.00 9,999,498.00 73,298,969.00
20% 51,983,908.00 6,862,883.50 9,936,235.00 68,783,026.50
Moderate
Policy 30% 48,008,948.00 6,113,306.50 9,944,422.00 64,066,676.50
40% 44,772,580.00 7,871,248.50 10,003,806.00 62,647,634.50
50% 41,712,944.00 8,553,867.00 9,969,701.00 60,236,512.00
60% 39,496,832.00 8,552,256.00 10,067,161.00 58,116,249.00
Tolerant 70% 37,418,580.00 11,413,970.00 10,135,188.00 58,967,738.00
Policy 80% 35,393,508.00 16,106,194.00 10,125,914.00 61,625,616.00
90% 33,539,002.00 16,100,504.00 10,124,556.00 59,764,062.00
100% 31,890,532.00 16,098,772.00 10,132,575.00 58,121,879.00
Note: the highlighted cells indicate the lowest values of the respective columns.
As the overweight percentage equally increases in both routes, there is a more equalized proportion
of distributed load volume and, different than what happened in the previous scenarios, the costs with
pavement maintenance and traffic accidents initially decrease, since these impacts do not overload only
one route. Next, with 30% and 20% of overweight, maintenance and accident costs increase, respectively.
Total cost is still minimized under the Tolerant Policy, with an overweight of 60% per vehicle,
since the operational cost is more expressive than the social costs, in case the relative importance of
each cost is not considered.
The relative importance of social costs in relation to the operational cost was also simulated for
the case in which both routes have overweighed ornamental stones vehicles. The results once again
show that, the higher the importance of social costs due to pavement maintenance and traffic accidents
in relation to the operational cost of transportation, the lower the percentage of overweight should be
in order to minimize total costs.
The difference in costs between the scenarios with overweighting only in Route #1 and with
overweighting in Routes #1 and #2 varies from 6% to 15% according to the weight of the social cost,
that is, economically it is better for the overweighed ornamental stone vehicles to use two routes
instead of only one, since the damage caused by overweight in only one route is higher for the system
under the established conditions.
That happens because, with overweighed ornamental stone vehicles only in Route #1,
the proportion of load volume in this route is in average 90%, causing most ornamental stone
vehicles to use this route, thus punishing it in relation to Route #2. Now if overweighting is practiced
in both routes, the proportion of load volume in Route #1 drops to 48% in average, therefore balancing
the damage caused by overweight in both routes. Nevertheless, it is known that, in practice, freight
vehicles choose Route #1 due to its lower operational cost and due to the absence of tolls and that
overweighed vehicles also choose this route due to the absence of scales and the lower probability of
inspections in comparison to Route #2.
6. Conclusions
This study presented an SD model that enables the assessment of the relationship between
overweight in road transportation of ornamental stones and the costs associated with the operational
cost of transportation and to social costs due to traffic accidents and pavement maintenance. Moreover,
weights were attributed to the operational and social costs to assess the best policies in each case.
A literature review was carried out and showed some limitations, such as the consideration of a
roadway dedicated to the transportation of a specific load; this view simplifies the problem in relation
to reality, since it does not consider the traffic of other vehicles and its impacts on important variables
such as road capacity, travel time and operational cost of transportation.
The model proposed regarding the case study showed the importance of observing the time
interval between the assessments of pavement conditions and the duration of maintenance works,
which significantly impact pavement maintenance cost. However, these time intervals do not take into
consideration the delay between making a decision about the need for maintenance and the effective
beginning of road works, including biddings and the hiring of third-party providers to carry out the
maintenance service. This delay may further affect the pavement performance indicator, so it is valid
to analyze, in the future, the impact caused by this delay.
The results confirm the advantage of the entrepreneurial strategy of overweighting ornamental
stones transportation vehicles to reduce the unit operational cost of transportation. Nevertheless, as the
overweight percentage increases, the social costs involved also increase. This way, there is a trade-off
between the reduction of operational cost and social costs and they must be measured according to
their respective attributed importance.
The practice of overweighting per axle increases productivity and profit in the transportation
industry, since it decreases the unitary transport cost per tons of load; on the other hand, it brings
disadvantages to society, such as the fast deterioration of pavements, an increase in road maintenance
costs and reduced road safety due to the increase in accident rates, as seen in the analysis performed in
this study.
Despite the scenario with 100% overweight was hypothetical for simulation purposes, we note that
there were already registrations of up to 70% in groups of axles and up to 57% in TGW of ornamental
stones vehicles. This implies in reducing road safety and other studies are recommended for a deeper
analysis of these conditions.
Also, a more detailed cost-benefit analysis is indicated because the use of overweighed vehicles
reduces the number of trips but increase the unitary costs of the trips (generalized costs) due the
lower average speed and fine costs, for example. Moreover, the overweighed vehicles need more
maintenance due to the accelerated wear of the vehicle parts but the fleet used is smaller and therefore
the number of maintenances is also smaller [21]. This kind of analysis was not carried out in this study
due to lack of available data but it is also indicated for future researches.
The deterioration of Brazilian highways has a direct impact on logistics costs and, consequently,
on the competitiveness of national products in a globalized economy. Furthermore, the lack or low
quality of infrastructure discourages new investments. Regarding traffic accidents, this study only
uses measurable costs, that is, the costs of accidents analyzed here are underestimated, since no studies
have been found that consider the cost with the loss of lives, for example. The literature mentions the
difficulty of measuring the loss of lives beyond the economic loss. Besides, it was not possible to find
data that attempted to estimate or suggest a value to this loss.
Also, there are other costs associated with these negative externalities that were not measured in
this study, such as indirect costs related to trip delays in the occasions in which the road are blocked
due to accidents, leading companies to keep safety stocks (even those companies that depend on
daily supply of inputs) and the personal losses of various road users. Other costs that were not
measured in this study regard environmental impacts caused when these accidents involve vehicles
transporting hazardous materials, risking the environmental integrity of the ecosystem around the
event. Therefore, social impacts should be better studied and measured for a more precise assessment
of the real costs involved.
For the scenario in which social costs are deemed extremely important, results show that a strict
policy should be adopted regarding the tolerance of overweighting in carrier vehicles. However,
an ever higher legally admitted tolerance has been identified in Brazil regarding the weight transmitted
per axle, which contributes to the accelerated deterioration of pavements and other associated damage.
Even so, violations of weight limits are frequent on national roadways. That is why it is necessary
to invest in more efficient policies and actions such as uninterrupted inspection, driver awareness,
mandatory declaration of freight weight in the invoice and fines equivalent to the profit achieved
through the practice of overweighting.
In this context, this study contributes to the literature in that it attempts to complement the
knowledge in the area, supporting strategic planning and decision making concerning the development
and implementation of policies that regulate freight transportation in the Brazilian road transportation
system, assisting in the understanding of the system’s behavior, in the quantification of the costs
involved based on the percentage of overweighting being practiced but especially on the importance
attributed to each cost category.
Within the same scope, some factors may be analyzed in more detail if addressed by new studies.
Examples of that would be: the inclusion of a module considering the environmental impacts caused
by the practice of overweighting in freight vehicles, such as increased fuel consumption, costs with air
pollutant and greenhouse gas emissions; the analysis of other social impacts for a better estimate of the
actual costs indirectly incurred by the society due to the practice of overweighting; the inclusion of
other vehicle categories with overweighting that travel along the analyzed route; and the comparison
of costs in a scenario of logistic integration with the use of more efficient modes of transportation,
such as railways and waterways.
The proposed model contributes to the understanding of the dynamic behavior of the ornamental
stones transportation system under different vehicle loading policies, with results that support decision
making in favor of sustainability in transportation.
Supplementary Materials: The following are available online at http://www.mdpi.com/2071-1050/11/11/3128/s1,

Table S1: Mathematical Equations of the SD model.
Author Contributions: V.G., G.d.L.D.C. and G.M.R. conceived and designed the system dynamics model of this
work; V.G., G.M.R. and I.C.S.H. collected and analyzed the data; V.G. wrote the paper; and, finally, G.d.L.D.C.,
G.M.R., R.D.O.F., I.C.S.H. and L.R.P. revised the paper. All authors have read and approved the final manuscript.
Funding: This research was funded by Coordination of Improvement of Higher Level Personnel (CAPES),
grant number 1587576, by National Council for Scientific and Technological Development (CNPq), grant number
307835/2017-0 and 308411/2018-8 and by Carlos Chagas Filho Research Support Foundation (FAPERJ), grant number
203.231/2017.
Conflicts of Interest: The authors declare no conflict of interest. The funders had no role in the design of the
study; in the collection, analyses or interpretation of data; in the writing of the manuscript or in the decision to
publish the results.
References
1. Diaz, R.; Behr, J.G.; Ng, M. Quantifying the economic and demographic impact of transportation infrastructure
investments: A simulation study. Simulation 2016, 92, 377–393. [CrossRef]
2. Hong, J.; Chu, Z.; Wang, Q. Transport infrastructure and regional economic growth: evidence from China.
Transportation 2011, 38, 737–752. [CrossRef]
3. Ozbay, K.; Ozmen-Ertekin, D.; Berechman, J. Contribution of transportation investments to county output.
Transp. Policy 2007, 14, 317–329. [CrossRef]
4. Weisbrod, G. Models to predict the economic development impact of transportation projects: Historical
experience and new applications. Ann. Reg. Sci. 2008, 42, 519–543. [CrossRef]
5. Lakshmanan, T.R. The broader economic consequences of transport infrastructure investments.
J. Transp. Geogr. 2011, 19, 1–12. [CrossRef]
6. The World Bank. Connecting to Compete 2016: Trade Logistics in the Global Economy. The Logistics Performance
Index and its Indicators; The World Bank: Washington, DC, USA, 2016.
7. World Economic Forum. The Global Competitiveness Report 2017–2018. 2017. Available online:
http://www3.weforum.org/docs/GCR2017-2018/05FullReport/TheGlobalCompetitivenessReport2017%
E2%80%932018.pdf (accessed on 11 November 2018).
8. Schade, B.; Schade, W. Evaluating Economic Feasibility and Technical Progress of Environmentally Sustainable
Transport Scenarios by a Backcasting Approach with ESCOT. Transp. Rev. 2005, 25, 647–668. [CrossRef]
9. PwC—PricewaterhouseCoopers. Meeting the 2050 Freight Challenge, Sydney. 2009. Available online:
http://infrastructure.org.au/wp-content/uploads/2017/06/Freight-2050-Final-Multi-Colour.pdf (accessed on 7
August 2017).
10. Blower, D.; Woodrooffe, J. Survey of the Status of Truck Safety: Brazil, China, Australia and The United States;
Transportation Research Institute, The University of Michigan, UMTRI: Ann Arbor, MI, USA, 2012.
11. Castro, N. Mensuração de externalidades do transporte de carga brasileiro. J. Transp. Lit. 2013, 7, 163–181.
(In Portuguese) [CrossRef]
12. DNIT—Departamento Nacional de Infraestrutura de Transportes (National Department of Transportation
Infrastructure). Manual de Estudos de Tráfego; Instituto de Pesquisas Rodoviárias, Publicação 723; DNIT:
Rio de Janeiro, Brazil, 2006. (In Portuguese)
13. Litman, T.A. Transportation Cost and Benefit Analysis: Techniques, Estimates and Implications, 2nd ed.; Victoria
Transport Policy Institute: Victoria, BC, Canada, 2009.
14. Manataki, I.E.; Zografos, K.G. A generic system dynamics based tool for airport terminal performance
analysis. Transp. Res. Part C Emerg. Technol. 2009, 17, 428–443. [CrossRef]
15. Foster, J.D.G. Measurement of Central or Offset Axle Load by Axle-Mounted Strain Gauges. Strain 2003,
39, 21–26. [CrossRef]
16. Hoffman, A.J.; Coning, A. An Intelligent Freight Corridor Overload Control System. In Proceedings of
the International Conference on Intelligent Transportation Systems, Qingdao, China, 8–11 October 2014;
Volume 17, pp. 1732–1739.
17. Coning, A.; Hoffman, A.J. Overload Control Centre Bypass Detection System (Extended Abstract).
In Proceedings of the International Conference on Intelligent Transportation Systems, Qingdao, China,
8–11 October 2014; Volume 17, pp. 1904–1905. [CrossRef]
18. Jacob, B.; de La Beaumelle, V.F. Improving truck safety: potential of weigh-in-motion technology. IATSS Res.
2010, 34, 9–15. [CrossRef]
19. Osman, O.; Ghazolly, M.E.; Mousa, R.M. Impact of Increasing Legal Axle Loads on Truck Factor in Egypt.
In Proceedings of the International Gulf Conference on Roads, Doha, Qatar, 10–13 November 2008; Volume 4,
pp. 1–10. [CrossRef]
20. Mousa, R.; Ghazolly, M.E.; Osman, O. The Impact of Increasing Legal Truck Axle Loads on the Egyptian
Roads. In Proceedings of the International Conference on Applications of Advanced Technologies in
Transportation, Athen, Greece, 27–31 May 2008; Volume 10, pp. 1–14. [CrossRef]
21. Moreno-Quintero, E.; Fowkes, T.; Watling, D. Modelling planner–carrier interactions in road freight transport:
Optimisation of road maintenance costs via overloading control. Transp. Res. E Logist. Transp. Rev. 2013,
50, 68–83. [CrossRef]
22. Sterman, J. Business Dynamics: System Thinking and Modelling for a Complex World, 1st ed.; McGraw-Hill:
New York, NY, USA, 2000; ISBN 978-0072389159.
23. Alasad, R.; Motawa, I. Dynamic demand risk assessment for toll road projects. Constr. Manag. Econ. 2016,
33, 799–817. [CrossRef]
24. Li, Y.; Taylor, T.R.B. Modeling the Impact of Design Rework on Transportation Infrastructure Construction
Project Performance. J. Constr. Eng. Manag. 2014, 140, 1–8. [CrossRef]
25. Yuliawati, E.; Hadiwardoyo, S.P.; Susantono, B.; Tjahjono, T. System Dynamics Model for Airport
Characterization in Hub-and-Spoke Networks. IJTech 2015, 6, 733–742. [CrossRef]
26. Nabavi, E.; Daniell, K.A.; Najafi, H. Boundary matters: the potential of system dynamics to support
sustainability? J. Clean. Prod. 2017, 140, 312–323. [CrossRef]
27. Qudrat-Ullah, H.; Seong, B.S. How to do structural validity of a system dynamics type simulation model:
the case of an energy policy model. Energy Policy 2010, 38, 2216–2224. [CrossRef]
28. Morecroft, J.D. Strategic Modelling and Business Dynamics: A Feedback Systems Approach, 2nd ed.; Wiley:
Hoboken, NJ, USA, 2015; ISBN 978-1118844687.
29. Forrester, J.W.; Senge, P.M. Tests for building confidence in system dynamics models. In System Dynamics;
Legasto, A.A., Jr., Forrester, J.W., Lyneis, J.M., Eds.; TIMS Studies in the Management Sciences; North Holland
Publishing Company: Cambridge, MA, USA, 1980; pp. 209–228.
30. Karnopp, D.C.; Margolis, D.L.; Rosenberg, R.C. System Dynamics: Modeling, Simulation and Control of
Mechatronic Systems, 5th ed.; John Wiley & Sons: New York, NY, USA, 2012; ISBN 978-0470889084.
31. Lobontiu, N. System Dynamics for Engineering Students: Concepts and Applications, 2nd ed.; Academic Press:
Cambridge, MA, USA, 2017; ISBN 8-0128045596.
32. Zagonel, A.A.; Corbet, T.F. Levels of Confidence in System Dynamics Modeling: A Pragmatic Approach to
Assessment of Dynamic Models. In Proceedings of the International Conference of the System Dynamics
Society, Nijmegen, The Netherlands, 23–27 July 2006; Volume 24, pp. 1–19.
33. Chasey, A.D.; Garza, J.M.; Drew, D.R. Using Simulation to Understand the Impact of Deferred Maintenance.
Comput.-Aided Civ. Infrastruct. Eng. 2002, 17, 269–279. [CrossRef]
34. Fallah-Fini, S.; Rahmandad, H.; Triantis, K.; Garza, J.M. Optimizing highway maintenance operations:
dynamic considerations. Syst. Dyn. Rev. 2010, 26, 216–238. [CrossRef]
35. Friedman, S. Is counter-productive policy creating serious consequences? The case of highway maintenance.
Syst. Dyn. Rev. 2006, 22, 371–394. [CrossRef]
36. CNT—Confederação Nacional do Transporte (National Confederation of Transport). Pesquisa CNT de
Rodovias, Relatório Gerencial, 21st ed.; CNT: Brasília, Brazil, 2017. (In Portuguese)
37. Hang, W.; Li, X. Application of system dynamics for evaluating truck weight regulations. Transp. Policy 2010,
17, 240–250. [CrossRef]
38. Liu, P.; Mu, D. Evaluating Sustainability of Truck Weight Regulations: A System Dynamics View. J. Ind.
Eng. Manag. 2015, 8, 1711–1730. [CrossRef]
39. Liu, P.; Mu, D.; Gong, D. Eliminating overload trucking via a modal shift to achieve intercity freight
sustainability: A system dynamics approach. Sustainability 2017, 9, 398. [CrossRef]
40. TRB—Transportation Research Board. Highway Capacity Manual 2010; National Research Council:
Washington, DC, USA, 2010; ISBN 978-0-309-16077-3.
41. BPR—Bureau of Public Roads. Traffic Assignment Manual; Urban Planning Division, US Department of
Commerce: Washington, DC, USA, 1964.
42. Ortúzar, J.D.; Willumsen, L.G. Modelling Transport, 3rd ed.; John Wiley & Sons: New York, NY, USA, 2001;
ISBN 9780471861102.
43. MTPA—Ministério dos Transportes Portos e Aviação Civil (Ministry of Transport Ports and Civil Aviation);
UFRJ—Universidade Federal do Rio de Janeiro (Federal University of Rio de Janeiro). Centro de Integração
Logística—Desenvolvimento da Metodologia de Localização de CILs. Revisão e ajustes na base de dados Georreferenciada
do PNLT para Aplicação da Metodologia de localização de CILs; Etapa 5, Apêndice I. Tomo II; DNIT: Brasília,
Brazil, 2016. (In Portuguese)
44. Sun, L.; Liu, X. General deterioration equation for pavement performance. J. Tongji Univ. 1995, 23, 512–518.
(In Chinese)
45. AASHTO—American Association of State Highway and Transportation Officials; FHWA—Federal Highway
Administration. Highway Safety Manual: User Guide; The National Academies: Washington, DC, USA, 2014.
46. Ventana Systems, Inc. Vensim Simulation Software. 2018. Available online: http://www.vensim.com
(accessed on 15 June 2018).
47. Volino, P.; Magnenat-Thalmann, N. Comparing Efficiency of Integration Methods for Cloth Simulation.
In Proceedings of the Computer Graphics International, Hong Kong, China, 3–6 July 2001; pp. 265–272.
[CrossRef]
48. Castro, N.F.; Marcon, D.B.; Freire, L.C.; Lima, E.F.; Almeida, P.F. Impacto do APL de Rochas Ornamentais do
Espírito Santo nas Comunidades. In Recursos Minerais & Sustentabilidade Territorial: Arranjos Produtivos Locais;
Fernandes, F.R.C., Enríquez, M.A.R.S., Alamino, R.C.J., Eds.; CETEM/MCTI: Rio de Janeiro, Brazil, 2011;
Volume 2, pp. 139–176. ISBN 978-85-61121-84-6.
49. Menezes, R.G.; Sampaio, P.R.A. Rochas Ornamentais no Noroeste do Estado do Espírito Santo; CPRM: Rio de
Janeiro, Brazil, 2012. (In Portuguese)
Infrastructure). Nota Técnica Nº 002/2017/DE. Detalhamento Técnico-Científico da Primeira Versão da
Metodologia Estabelecida para Estimativa de Volume Médio Diário Annual—VMDA em toda a Malha
Rodoviária Pavimentada Federal. 2017. Available online: http://servicos.dnit.gov.br/dadospnct/Modelagem
(accessed on 10 December 2017). (In Portuguese)
51. Ministry of Mines and Energy (Ministério de Minas e Energia). Rochas Ornamentais e de Revestimento. Relatório
Técnico; Secretaria de Geologia, Mineração e Transformação Mineral: Brasília, Brazil, 2009. (In Portuguese)
52. Chiodi Filho, C. Balanço das Exportações e Importações Brasileiras de Rochas Ornamentais em 2008; ABIROCHAS:
São Paulo, Brazil, 2009. (In Portuguese)
53. Alice Web. Exportação Municípios 1997–2017. 2017. Available online: http://aliceweb.mdic.gov.br/ (accessed
on 10 December 2017). (In Portuguese)
54. CONTRAN—Conselho Nacional de Trânsito. Resolução nº 354 de 24 de junho de 2010. Estabelece requisitos de
segurança para o transporte de blocos e chapas serradas de rochas ornamentais; Diário Oficial da União: Brasília,
Brazil, 2010. (In Portuguese)
55. Park, K.; Thomas, N.E.; Wayne Lee, K. Applicability of the International Roughness Index as a Predictor of
Asphalt Pavement Condition. J. Transp. Eng. 2007, 133, 706–709. [CrossRef]
56. ECO–101. Mapa Interativo—Pedágios. 2016. Available online: http://www.eco101.com.br/Mapa-Interativo.
(accessed on 25 December 2017). (In Portuguese).
57. IBGE—Instituto Brasileiro de Geografia e Estatística (Brazilian Institute of Geography and Statistics). Séries
Históricas. 2017. Available online: https://ww2.ibge.gov.br/home/estatistica/indicadores/precos/inpc_ipca/
defaultseriesHist.shtm (accessed on 25 December 2017). (In Portuguese)
Infrastructure). Quadro de Fabricantes de Veículos. Diretoria de infraestrutura rodoviária, Coordenação geral de
operações rodoviárias; DNIT: Brasília, Brazil, 2012. (In Portuguese)
59. CONTRAN—Conselho Nacional de Trânsito. Resolução nº 210 de 13 de novembro de 2006. Estabelece os limites
de peso e dimensões para veículos que transitem por vias terrestres e dá outras providências; Diário Oficial da União:
Brasília, Brazil, 2006. (In Portuguese)
60. CONTRAN—Conselho Nacional de Trânsito. Resolução nº 258 de 30 de novembro de 2007. Regulamenta os artigos
231, X e 323 do Código Trânsito Brasileiro, fixa metodologia de aferição de peso de veículos, estabelece percentuais de
tolerância e dá outras providências; Diário Oficial da União: Brasília, Brazil, 2007. (In Portuguese)
61. DNIT – Departamento Nacional de Infraestrutura de Transportes (National Department of Transportation
Infrastructure). Manual de Restauração de Pavimentos Asfálticos, 2nd ed.; DNIT: Rio de Janeiro, Brazil, 2006.
(In Portuguese)
Infrastructure). Custos Médios Gerenciais. 2017. Available online: http://www.dnit.gov.br/custos-e-
pagamentos/custo-medio-gerencial/ANEXOIXCUSTOMDIOGERENCIALNOVEMBRO2016.pdf (accessed
on 10 December 2017). (In Portuguese)
63. IPEA—Instituto de Pesquisa Econômica Aplicada (Institute of Applied Economic Research). Acidentes de
trânsito nas rodovias federais brasileiras: Caracterização, tendências e custos para a sociedade; Relatório de pesquisa,
Ipea: Brasília, Brazil, 2015. (In Portuguese)
64. CNT—Confederação Nacional do Transporte (National Confederation of Transport). Pesquisa CNT de
Rodovias, Relatório Gerencial, 20th ed.; CNT: Brasília, Brazil, 2016. (In Portuguese)
65. PRF—Polícia Rodoviária Federal (Federal Road Police). Acidentes. 2017. Available online: https://www.prf.
gov.br/portal/dados-abertos/acidentes. (accessed on 9 January 2018). (In Portuguese)
66. Zhang, W.; Cai, C.S. Fatigue Reliability Assessment for Existing Bridges Considering Vehicle Speed and
Road Surface Conditions. J. Bridge Eng. 2012, 17, 443–453. [CrossRef]
Infrastructure). Terminologias rodoviárias usualmente utilizadas; DNIT: Brasília, Brazil, 2007. (In Portuguese)
© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).

Sustainability 11 03128 PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Sustainability 11 03128 PDF

Uploaded by

Copyright:

Available Formats

sustainability

Keywords: freight transportation; overweight vehicles; simulation; social impact; sustainability

Sustainability 2019, 11, 3128; doi:10.3390/su11113128 www.mdpi.com/journal/sustainability

Sustainability 2019, 11, x FOR PEER REVIEW 5 of 37

total freight <SAVEPER>

t = t0 ·(1 + α·(V/C)β (1)

• t: adjusted time in which volume V may use road link;

• c1: generalized cost of Route #1; and

toll #3 generalized cost dimensionless split proportion of split proportion of

wl/single tyres B #1 LF single tyres B #1

wl/twin tyres B #1 LF twin tyres B #1 axle factor

RM cost per <maintenance <vehicle with

CM cost #1 = CM cost per km · distance route #1 · cost growth rate (4)

RM cost #1 = RM cost per km · distance route #1 · cost growth rate (5)

pavement maintenance cost = IF THEN ELSE(“maintenance operations” = 0, 0,

tra f f ic accident cost

correction factor for

accident with accident with accident with no <cost growth

<freight volume <generalized cost

<generalized cost <freight volume

3.2. Mathematical Formulation

For the proposedconsistency

Figure 10. Map

Figure 11. Schematic representation of road routes and links.

Figure 11. Schematic representation of road routes and links.

4.2. Parameters for the Speed Flow Module

Table 2. Estimate of the average AADT (vehicles) of each link in 2016.

4.3. Parameters for the Modal Split Module

Table 4. Toll fares in the analyzed routes for a truck of 7 axles.

Municipality Location Value (R$)

To decide about the type Table

Table 6. Weight limits, considering the tolerance of 10% per axle.

Weight Allowed per

Irregularity Condition Interventions Effects

4.5. Parameters for the Social Cost of Freight Transportation Module

Table 8. Pavement maintenance costs.

Maintenance Lower Limit (R$/Km) Upper Limit (R$/Km) Average (R$/Km)

Table 9. Total cost and average cost per accident severity.

Severity Average Cost (R$/occurrence)

Table 10. Number and percentage of accidents according to severity.

Number Percentage Accidents average/Km Estimate

5. Results and Discussions

Table 11. Description of scenarios.

Scenario Simulated Variables Impact

5.1. Reference Scenario

Assessment Total Cost Total Cost Total Cost

5.2. Strict, Moderate and Tolerant Policies

5.2. Strict, Moderate and Tolerant Policies

5.4. Overweight in All Routes

Supplementary Materials: The following are available online at http://www.mdpi.com/2071-1050/11/11/3128/s1,

You might also like