Additional Road User Costs due to Pavement Conditions and Maintenance Actions: Initial Approach for Portuguese Conditions B. Santos Department of Civil Engineering and Architecture, University of Beira Interior, Covilhã, Portugal
[email protected]
L. Picado-Santos Department of Civil Engineering and Architecture, IST, Technical University of Lisbon, Portugal
[email protected]
V. Cavaleiro Department of Civil Engineering and Architecture, University of Beira Interior, Covilhã, Portugal
[email protected]
ABSTRACT: A simplified Road User Costs model and input data for Portuguese trunk road networks were defined in a research work conducted in 2004-2007 for Portuguese conditions. The developed model allows the calculation of average road user costs values (2006 base year) considering the vehicle operating, accident and time costs. This paper describes the first activities conducted to refine the model in order to include specific conditions, namely, the consideration of pavement conditions and maintenance actions (work zones) influence. The study was conducted to perform the refinements of the model which allowed the identification of sensitive parameters for the specific conditions mentioned above, as well as the comparison with the sensitive parameters identified in several international models in use and the verification of the Portuguese legal framework for the road work zones and together with the specific values for the Portuguese conditions, a proposal of an improved formulation. The paper also presents model applications to Portuguese road networks under private concession and compares the average costs values to the ones obtained for specific work zones and pavement conditions scenarios. The conducted and ongoing research will provide the needed tool to integrate road user costs in Portuguese Pavement Management Systems economic analysis since nowadays no structured road user costs model is being used in Portugal. KEY WORDS: User Costs, Roads, Pavement Conditions, Work Zones, Mathematical Models.
1. INTRODUCTION Currently, the Portuguese Road Administration does not consider Road User Costs (RUC) in the evaluation process of road design, maintenance or rehabilitation; therefore the estimation of road life cycle costs does not include this important aspect.
Taking into account the described situation, a research work was defined in two Portuguese universities in order to develop a simplified RUC model to obtain road user cost as a function of the available official data. The simplified RUC model was developed in 2004-2007 and has provided the needed toll to consider these costs in the asset of management systems (Santos, 2007). This model is based on the simplifications of the HDM-4 (Bennett et al., 2001), COBA (Department for Transport et al., 2002) and JAE (Portuguese Road Administration, designated, at the time, as JAE – Junta Autónoma de Estradas) (GEPA, 1995) road user costs models, providing average cost values. The three main components of road user costs considered were vehicle operating, accident and time costs, along with a component related with tolling costs. The model was also developed aiming at simplicity, reduced data requirements (selected data is usually available), easy calibration, easy application and trustworthy results. However, the effect of pavement conditions and maintenance actions in RUC estimation are not considered in the simplified model developed. So, model refinements to include these important aspects are proposed in this paper to enable a more realistic road cost analysis, since they were not explicitly considered in the initial formulation. To sustain the refinements made in the model, the analysis to identified critical parameters, a study of the approaches taken by some international models in use and the characterization of Portuguese legal framework, were performed. These aspects have been incorporated in the formulation without neglecting the objective of obtaining a simple model with reduced data requirements.
2. INITIAL RUC FORMULATION The model for Portuguese conditions was developed taking into account several main aspects, specifically: the recognized conceptual principles, application to trunk roads, the impact of each component on the total users costs and the availability of Portuguese official information, moreover, four vehicle classes were considered, passenger car (PC), utility (U), heavy truck (HT) and heavy bus (HB). The results of these considerations leaded to a Portuguese model with the three main costs components identified above: the vehicle operating costs (VOC), including costs for fuel, tires, vehicle preventive maintenance and depreciation; the accident costs (AC), considering costs for accident (police and medical assistance by accident type) and casualty costs (fatalities, serious and slight injuries); and the value of time (VOT) for work and non-work travels. Eventually a component related to tolling costs (Toll) may also be added. The cost basis used is only related to direct costs for the user. The model developed has the following general formulation: RUC VOC AC VOT Toll
(1)
4
VOC AADT VOC i pi
(2)
3 3 AC AADT AC j CCk k 1 j1
(3)
i 1
4
VOT AADT ( VOTi pi )
(4)
i 1
4
Toll AADT ctolli pi
(5)
VOCi Cf i Ct i Cmi Cd i
(6)
1 2 TCm OR i ,m si m1
(7)
i 1
Considering by vehicle class:
VOTi
And for the set of all vehicle classes (without vehicle class disaggregation): AC j AR j ac j
(8)
3
CC k ANCk cc k AR j
(9)
j1
Where, AADT is the annual average daily traffic [vehicles/day]; AC is the accident cost [€/km/day]; ACj is the accident j cost [€/km/vehicle]; acj is the accident j cost (police time cost) [€/accident]; ANCK is the average number of k casualties by accident [casualties/accident]; ARj is the accident j rate [accidents/vehicle/km]; CCk is the casualty k cost [€/km/vehicle]; cck is the casualty k cost [€/casualty]; Cdi is the vehicle depreciation cost for vehicle i [€/km]; Cfi is the fuel cost for vehicle i [€/km]; Cmi is the maintenance cost for vehicle i [€/km]; Cti is the tire cost for vehicle i [€/km]; ctolli is the toll cost for vehicle i [€/km/vehicle]; i is the vehicle class: i=1 for Passenger Car; i=2 for Utility; i=3 for Heavy Truck; i=4 for Heavy Bus; j is the accident class: j=1 for accidents with slight injury; j=2 for accidents with serious injury; j=3 for accidents with fatalities; k is the casualty class: k=1 for slight injury; k=2 for serious injury; k=3 for fatalities; m is the travel purpose: m=1 for travel in work time; m=2 for travel in non-work time; ORi,m is the occupancy rate for vehicle i and travel purpose m [occupant/vehicle]; pi is the vehicle proportion of each class i for the AADT considered; RUC is the Road user cost [€/km/day]; si is the average operating speed for vehicle i [km/h]; TCm is the time cost for travel purpose m [€/h/occupant]; Toll is the Toll cost [€/km/day]; VOC is the Vehicle operating cost [€/km/day];
VOCi is the VOC for vehicle i [€/km]; VOT is the Value of time [€/km/day]; VOTi is the Value of time for vehicle i [€/km/vehicle]. The input model values for average situations were also defined for Portuguese conditions, using 2006 as the base year. This definition took into account the values used and recommended by the existing methodologies and, in particular, the values obtained from the Portuguese haulers association, companies and official bodies such as the police and emergency services. The complete formulation, as well as the defined input values for passenger car and heavy truck, can be consulted in papers presented at international conferences (Santos et al., 2008; Santos et al., 2007).
3. ADDITIONAL RUC DUE TO PAVEMENT CONDITIONS AND MAINTENANCE ACTIONS Additional RUC due to maintenance intervention periods (work zones) and changes in pavement conditions can easily be included in the proposed RUC formulation by considering specific parameter values defined for a certain maintenance strategy or for a particular pavement quality index. Several variability studies were carried out to identify the sensitive parameters of the model. These studies simulated a variation of 20 to 30% of the average values adopted for operating speed, fuel consumption, tire service life, vehicle service life and annual average kilometrage. The results obtained show that the proposed model, as most of the existing ones, is mainly sensitive to changes in the average operating speed defined for each vehicle class and type of road; and fuel consumption and cost. Besides being identified as critical parameters, speed and fuel consumption and cost are also the main factors in the VOT and VOC definition, and those that best reflect the additional RUC due to pavement conditions and maintenance actions on the network (work zones). Thus, the definition and consideration of correction factors that can be applied to the average values defined for these parameters is critical to forecast additional RUC in sections where maintenance actions are planned, or to compute the benefits associated with a better pavement condition. 3.1. Pavement Conditions Changes in operating speed and the consequent additional travel time due to the pavement conditions can be easily incorporated into the proposed model formulation by considering the section length within a certain pavement quality index and lowers operating speeds. However, the pavement conditions of the most important networks, as the national ones, do not reach in general such a degradation level that influences, in a significant way, the values of the normal operating speed. Moreover, pavements in good conditions allow vehicles to perform higher speeds with greater comfort and security, reducing travel time and accident costs. It also allows reductions in operating costs in terms of tires, maintenance and depreciation of the vehicle, but not necessarily in fuel. Opposite situation occurs for pavements in poor conditions. The condition of the road network pavements was integrated in the VOC calculations through a quality index representing the functional and structural state of the pavements. The
index adopted in the proposed model was the Present Serviceability Index (PSI), which ranges from 0 (for a pavement in a poor state) to 5 (for a new pavement). Equation (10) presents a PSI formulation developed for Portuguese trunk roads (PicadoSantos et al., 2006). This formulation is currently used by the Portuguese Road Administration and was adopted in the proposed RUC model in order to obtain PSI values to use in the calculation of additional VOC due to pavement conditions. The changes in VOC as a function of PSI (or IRI – International Roughness Index) have been obtained from expressions developed by applying regression analysis to real data, resulting in several formulations such as those presented in HDM-4 (World Bank, 2007), TRB (TRB, 1983), ASTM (ASTM, 1983) and by Picado-Santos et al. (Picado-Santos et al., 2006) for Portuguese conditions. The study of these formulations, as well as an analysis of recent data on Portuguese trunk road pavements condition (Picado-Santos and Pereira, 2009) and average user cost (Santos, 2007) was used to develop an equation that reflects the change of VOC as a function of PSI. This analysis allowed the definition of Portuguese average and extreme values for the relation VOC – PSI. These values, presented in Table 1, were used to calibrate the Portuguese mathematical model (see equation (12)). Table 1: PSI versus RUC values for specific Portuguese scenarios PSI
IRI (m/km)
Correction factors for VOC
0 2.0 3.5 4.7 5.0
4.25 3.50 2.00 0.50 -
1.15 1.05 1.00 0.95 0.95
The refinements proposed to include the consideration of pavement conditions in RUC calculations are presented in the equations (10), (11) and (12). RUC ⁄
VOC
(10)
VOC
(
)
(11) (12)
Where, RUCPSI is the incremental increase or decrease in RUC owing to [€/km/day]; PSI is the present serviceability index [0-5]; FVOC,PSI is the VOC correction factor for a certain PSI value; IRI is the international roughness index [mm/km]; R is the mean rut depth [mm]; C is the total cracked pavement area [m2/100m2]; S is the total pavement disintegrated area (with potholes and raveling) [m2/100m2]; P is the pavement patching area [m2/100m2]. Figure 1 shows the curves obtained from the VOC correction factors recommended by TRB-ASTM, a Portuguese model (UC-UM) (Picado-Santos et al., 2006) and the proposed one (RUC PT). The TRB study only presents a correction factor for fuel consumption in USA
road networks. This study was completed by ASTM for non-fuel VOC components (engine oil, tires, maintenance and repair, and depreciation expense). Still, the correction factors presented in Figure 1 for TRB-ASTM approach were obtained by applying the Portuguese VOC values (Santos, 2007) to the disaggregated correction factors for fuel and non-fuel components. By the analysis of figure 1 it is possible to conclude that the trend of the proposed mathematical model is similar to the ones studied, even if the correction factors for worst pavements are smaller. This can be justified by considering technological advances in vehicles and the type of road network (trunk road), where low PSI values are not common.
VOC Correction Factor (%)
125,00 120,00 115,00 110,00 UC-UM 2006
105,00
TRB-ASTM 1983 (PC)
100,00
TRB-ASTM 1983 (HT)
95,00
RUC PT 2006
90,00 0
1
2
3
4
5
PSI
Figure 1: VOC correction factors for different PSI values Sensitivity analyses considering changes in VOC with IRI were also carried out, since most of the existing formulations use IRI as the main parameter for compute PSI in trunk road networks. In Figure 2 are presented the curves obtained from HDM-4 and JAE RUC when applying: the Portuguese 2006 VOC values to PC and HV in HDM-4 and JAE model; HDM-4 default values for T7 (AADT=6500) and T8 (AADT=20000) traffic classes; and the values proposed (RUC PT) in this study.
VOC Correction Factor (%)
130 125 120 115
HDM-4 T7 2007
110
HDM-4 T8 2007
105
HDM-4 2006 (PC)
100
HDM-4 2006 (HV)
95
RUC JAE 1995
90
RUC PT 2006
85 0
2
4
6
8
IRI (m/km)
Figure 2: VOC correction factors with IRI
For trunk roads and current values of IRI (up to 3.5m/km) most of the models presents similar results. Higher values of IRI, as the ones considered in HDM-4, are not frequent in Portuguese freeways networks under concession. Thus, it was considered a significant increase in the user cost for an IRI equal to 4.25 when compared with the models studied. It must be considered in the interpretation of Figure 2 that the index adopted in the proposed model for the definition of the VOC correction factor was PSI, which calculation depends not only on IRI, but also on pavement superficial degradations (such as rutting, cracking, potholes, raveling and patching). Moreover, the pavement management system adopted by the Portuguese Road Administration considers that an intervention is justified when PSI is equal to or less than 2.0, corresponding to an IRI of about 3.5m/km. 3.2. Work Zones The main parameters that can lead to additional RUC in work zones have been identified, in several models in use, as being the decrease of operating speed, which increases the VOT, and the consequent additional fuel consumption, increasing the VOC values. However, of the two parameters, the most significant influence on RUC values occurs due to changes in operating speeds. These changes and the consequent additional travel time can be incorporated into the proposed model formulation by the consideration of section length with work zones and lower operating speeds. Regarding fuel consumption, data collected, empirical models developed from that information (which usually relate fuel consumption to the operating speed of vehicles) and mechanistic models for fuel consumption (which relates this consumption to the forces opposing motion and allowing applications under different conditions), show that maximum fuel consumption occurs for low and high speeds, and minimal fuel consumption for speeds of 40 - 60km/h (Bennett et al, 2001) (Department for Transport et al., 2002). Thus, for trunk roads with high operating speeds (with at least two lanes in each direction), and considering the Portuguese legal framework which limits the road private concessionaires to guaranty operating speeds greater than or equal to 2/3 of normal operating speed in work zones (up to 10km per set) at day time (7h-21h), there is actually a decrease in fuel consumption. Lower speeds, up to 1/3 of normal operating speed, are allowed in work zones during the night time. In such cases there is a high probability of frequent stops, resulting in an increased fuel consumption associated with the movement at very low speeds. These cases are more common in two lane roads (one in each direction). Taking into account the scenarios described above, additional RUC in work zones due to changes in fuel consumption were only considered in Portuguese National and Regional Roads with two lanes (one in each direction) operating at lower speeds (up to 1/3 of normal operating speed), and in these cases an increase of 20% in fuel consumption was considered. The choice of this value took into account PC representative vehicle manufacturer information that points to urban fuel consumption values of 20-30% higher than the combined ones and, for HT, the additional values of 30-40% that are commonly obtained in fuel consumption models simulations. When traffic diversions are needed, changes in operating costs and travel time should be considered in the same manner as described above. The refinements proposed including the aspects mentioned above are presented in the following equations: (13)
∑
(
) for ∑ ⁄
(14)
(
) ∑
(
(15) )
(16)
Where, dCf is the incremental increase in fuel cost owing to M&R actions [€/km/day]; dVOT is the incremental increase in the value of time owing to M&R actions [€/km/day]; ENs are the National Roads with two lanes (one in each direction) and “medium” design standards; ERs are the Regional Roads with two lanes (one in each direction) and “medium” design standards; RUCM&R is the road user cost in maintenance and rehabilitation zones [€/km/day]; sM&Ri is the average operating speed in M&R sections, for vehicle i [km/h]; VOTM&Ri is the value of time in M&R sections for vehicle i [€/km/vehicle]. 4. MODEL APPLICATIONS The RUC formulation and input model values proposed were applied in two Portuguese freeways (multilane) networks under concession (Scutvias and Aenor) with good results. Table 2 includes the data provided by the private road concessions for RUC calculations. Some applications results are shown in Table 3. Table 2: Data provided by road concessions for 2006 Data
Scutvias (A23)
Aenor (A7, A11)
Network length (km) Total AADT pi (1) PC U HT HB Accidents With slight injury With serious injury With fatalities Casualties Slight injury Serious injury Fatalities Approximate toll cost (€/km)
177.5 10290 0.7987 0.0628 0.1295 0.0090 68 6 2 89 9 2 0.20 (virtual toll(2))
165.4 7769 0.8255 0.1337 0.0308 0.0010 55 9 0 89 9 0 0.07 (PC) 0.18 (HT)
Note to Table 2: (1) Information processed. (2) The approximate toll cost values provided by Scutvias correspond to a uniform rate for all vehicle classes.
For the two RUC applications some differences can be found in the average values presented for the VOC and the toll cost. This is caused essentially by the AADT values found for each freeway and by the different toll cost and system of toll charging. Virtual toll charging is adopted by Scutvias (the toll is paid by the taxpayer) and real charging by Aenor (the toll is paid directly by the users).
From the results obtained we can confirm the main role of the VOC (approximately 50%) and the VOT (approximately 20%) in total RUC. Therefore, knowing that the VOC is mostly influenced by pavement condition and the VOT by maintenance interventions, in a temporary basis, a decrease in RUC is expected when a care pavement maintenance program is applied, reducing at the same time the number of accidents. Moreover, it is also possible to conclude that toll costs have also a significant contribution in RUC (approximately 20%), and still, despite the small contribution of the AC component in the results obtained for the analyzed networks, must be also considered in the calculations, since it has a more significant impact in low-medium design standard roads. Additional RUC due to a maintenance/rehabilitation and pavement condition scenario in a section with 1km long, a speed reduction to 2/3 of normal operating speed (to 80km/h), PSI equal to 2,0 and no deviations was tested (see table 3). This scenario takes into account the Portuguese legal framework described above for main road network with work zones operating during day time. Table 3: Portuguese RUC model application results (2006 values) Scutvias (A23) Average values Costs VOC AC VOT Toll RUC
Scutvias (A23) Work Zone PSI=2,0
Aenor (A7 and A11) Average values
Aenor (A7 and A11) Work Zone PSI=2,0
RUC RUC RUC RUC RUC RUC RUC RUC (€/km/day) (%) (€/km/day) (%) (€/km/day) (%) (€/km/day) (%) 2.267 € 83 € 703 € 742 € 3.795 €
60% 2% 19% 19% 100%
2.384 € 83 € 1.055 € 742 € 4.264 €
56% 2% 25% 17% +12%
1.352 € 73 € 505 € 637 € 2.567 €
53% 3% 19% 25% 100%
1.421 € 73 € 758 € 637 € 2.889 €
49% 3% 26% 22% +12%
In this scenario, because fuel consumptions associated with high speeds, as the ones practiced in freeways, are high, the occurrence of lower speeds allowed by law for work zones does not increase fuel consumption, thus it was not considered in the analysis. The total RUC obtained considers only additional time costs and non-fuel components costs, resulting in an increase of 12% when compared to the average values of RUC.
5. CONCLUSIONS Many road user costs models with strong conceptual frameworks have been developed in the past and are in use, however, many countries or regions that wish to consider RUC in their road life cycle costs analysis lack the means to obtain and update all the data required for the application of these models. The RUC model defined for the Portuguese highway network fills this gap and can be used as a tool in road management systems as, currently, no RUC model is being used by the Portuguese road administration. The simplified formulation presented in this paper allows an easy, fast and reliable integration of these costs in realities that, until now, were not considered. The performed simulations clearly show the importance of the consideration of additional RUC due to the occurrence of maintenance actions. For pavements condition similar results were found, however, with less influence when compared to work zones additional cost.
The inclusion of these aspects on RUC calculations, through the definition of correction factors to be applied on the sensitive parameters of the model, will increase the accuracy of the RUC values (for total service life) to be used in asset management systems. Thus, it will be possible to identify the best strategic programs and reach optimal solutions with good benefits/cost relations, contributing to the sustainability of the infrastructures through an extended transport costs consideration (construction, maintenance and user costs). REFERENCES ASTM, 1983. Measuring Road Roughness and Its Effects on User Costs and Comfort. ASTM-STP 884, p. 127-142. Bennett, C., Greenwood, I., 2001. Modelling road user and environmental effects in HDM-4, Volume 7. The Highway Development and Management Series, PIARC, Paris, France. Department for Transport, Scottish Executive Development Department, Welsh Assembly Government / Llywodraeth Cynulliad Cymru, The Department for Regional Development / Northern Ireland, 2002. Design Manual for Roads and Bridges, Volume 13: Economic Assessment of Road Schemes, Section 1: The COBA Manual. UK. GEPA – Gestão de Pavimentos, Lda., 1995. Sistema de Gestão da Conservação. Sistema de Custos dos Utentes. JAE – Contrato 4915, Lisboa. Picado-Santos, L., Ferreira, A., Pereira, P., 2006. Estruturação de um Sistema de Gestão de Pavimentos para uma Rede Rodoviária de Carácter Nacional. CEC – Revista Engenharia Civil, N.º 26, p. 45-59. Picado-Santos, L., Pereira, P., 2009. Relatório de Caracterização do Estado do Pavimento, IC1- Costa de Prata (Pavement Characterization Report, IC1- Costa de Prata). ACIV, DCE - University of Coimbra, Report AENOR nº1 (confidential report until released by contractor). Santos, B., 2007. Modelação dos Custos dos Utentes na Gestão da Estrada (in portuguese). PhD Thesis, Covilhã, Portugal. Santos, B., Picado-Santos, L., Cavaleiro, V., 2007. Vehicle Operating, Accident and User Time Costs in Pavement Management Systems: Approach for Portuguese Conditions. The Fifth International Conference on Maintenance and Rehabilitation of Pavements and Technological Control, Utah, USA. Santos, B., Picado-Santos, L., Cavaleiro, V., 2008. A Road User Costs Model for Portuguese Trunk Roads. Proceedings of the 3rd European Pavement and Asset Management Conference, CD edition –1027, Portugal. The World Bank, 2007. HDM-4 Road User Costs Model. Version 1.20, Washington D. C. Zaniewski, J., 1983. Fuel consumption related to roadway characteristics (Discussion and closure). TRB.
