Assessment of Eco-Friendly Pavement Construction and Maintenance Using Multi-Recycled RAP Mixtures

: The demand for more sustainable solutions has led an ever-growing number of stakeholders to being committed to pursue the principles of sustainability in pavement management. Di ﬀ erent stakeholders have been looking for tools and methodologies to evaluate the environmental impacts of the solutions, for which the life cycle assessment (LCA) proved to be an appropriate methodology. This paper is focused on the LCA of road pavement multi-recycling based on the use of bituminous mixtures with high rates of reclaimed asphalt pavement (RAP). In order to promote the circular economy, a comparative analysis was performed on a road pavement section by taking into account di ﬀ erent scenarios, which stem from the combination of production, construction and rehabilitation activities incorporating di ﬀ erent RAP rates in new bituminous mixtures: 0% (as reference), 25%, 50%, 75% and 100%, respectively. LCA results have been expressed in terms of four damage categories: human health, ecosystem quality, climate change and resources. Results have shown that both recycled and multi-recycled bituminous mixtures lead to substantial beneﬁts in comparison with the solution employing virgin materials, hence embodying a sustainable approach. The beneﬁts grow with the increase in the RAP rate with an average decrease of 19%, 23%, 31% and 33% in all the impact categories for a 25%, 50%, 75% and 100% of RAP rate. Contributions: Conceptualization, validation, V.A.; formal analysis, D.V. and V.A.; investigation, D.V.; resources, J.N., A.C.F. and V.A.; writing—original draft preparation, D.V.; writing—review and editing, V.A., J.N. and A.C.F.; visualization, D.V.; supervision, J.N. and


Introduction
The road infrastructure is one of the most important and omnipresent assets in construction engineering. In fact, roads are not only an important part of a society's transportation network, but also a public asset in overall terms. Indeed, the transport of goods and people worldwide is mainly done by road infrastructures.
Consequently, in view of the importance of these infrastructures and the need to guarantee appropriate both short and long-term behaviors of the road networks worldwide, several tonnes of bituminous mixtures are produced every year. In 2007, 1.6 trillion tonnes of bituminous mixtures were produced worldwide [1]. Around the world, pavement construction companies are required to provide end-products complying with both the high standards, defined by road authorities, and the sustainability criteria. The life cycle assessment (LCA) is a methodology intended to assess the The LCA presented in this study was performed using the SimaPro software from PRé Sustainability in combination with the Ecoinvent database. This software-database combination is one of the most widely disseminated LCA tools, and provides the user with an interface, several comprehensive environmental information databases and various methods to perform the impact assessment [20]. The software provides a clear interface to perform a full LCA, comprising the four steps described in EN ISO 14040:2006 [17]. The Ecoinvent database is the most widely used in the construction sector. The database used was the Ecoinvent 3-allocation, cut-off by classificationunit base. The underlying philosophy of the cut-off approach is that the producer does not obtain any credit for the provision of recyclable materials, but is nonetheless fully responsible for their disposal as waste [66].

Functional Unit
The LCA functional unit creates the basis for comparing the different structure scenarios with the same utility for an equivalent function [12,67,68]. In this case, it involved a pavement unit that carried the same number of vehicles per year, over the same project analysis period (PAP). This period comprised the different periods referred to by different authors and specifications [6,9,29,67,[69][70][71][72][73]. The functional unit presented in this LCA was a real 1 km-long road section of a two-lane roadway, one in each direction, with an individual width of 3.5 m each and a total PAP of 69 years, from 1946 to 2015. The geometric characteristics, as well as the different mixtures used for each course, are presented in Figure 2. This figure also displays the overall maintenance and rehabilitation (M&R) strategy applied in reality by roadway concession holder (tasks, courses and application scheduling) and the 2 main structure scenarios considered in this LCA.
The first three phases of the pavement construction and maintenance history: initial construction, maintenance 1 and maintenance 2, were equal for both scenarios (A and B). Equal construction solutions and materials were also considered. Thus, these construction and maintenance actions were not taken into account in the study. For the remaining maintenance tasks, equal dates for the maintenance actions were assumed for both scenarios. However, these scenarios presented The LCA presented in this study was performed using the SimaPro software from PRé Sustainability in combination with the Ecoinvent database. This software-database combination is one of the most widely disseminated LCA tools, and provides the user with an interface, several comprehensive environmental information databases and various methods to perform the impact assessment [20]. The software provides a clear interface to perform a full LCA, comprising the four steps described in EN ISO 14040: 2006 [17]. The Ecoinvent database is the most widely used in the construction sector. The database used was the Ecoinvent 3-allocation, cut-off by classification-unit base. The underlying philosophy of the cut-off approach is that the producer does not obtain any credit for the provision of recyclable materials, but is nonetheless fully responsible for their disposal as waste [66].

Functional Unit
The LCA functional unit creates the basis for comparing the different structure scenarios with the same utility for an equivalent function [12,67,68]. In this case, it involved a pavement unit that carried the same number of vehicles per year, over the same project analysis period (PAP). This period comprised the different periods referred to by different authors and specifications [6,9,29,67,[69][70][71][72][73]. The functional unit presented in this LCA was a real 1 km-long road section of a two-lane roadway, one in each direction, with an individual width of 3.5 m each and a total PAP of 69 years, from 1946 to 2015. The geometric characteristics, as well as the different mixtures used for each course, are presented in Figure 2. This figure also displays the overall maintenance and rehabilitation (M&R) strategy applied in reality by roadway concession holder (tasks, courses and application scheduling) and the 2 main structure scenarios considered in this LCA. virgin AC, while in scenario B this course incorporated virgin AC, together with a certain rate of RAP that was obtained from the milling using a rejuvenator percentage. • Maintenance 4 implied the milling of both upper courses and their replacement by a new binder and surface course, respectively. In the case of scenario A, these courses consisted of stone mastic asphalt (SMA) mixtures with virgin materials, while in the courses in scenario B a certain RAP rate was incorporated together with virgin AC using a rejuvenator percentage. To assess the use of RAP material in new bituminous mixtures, different incorporation rates were considered in this LCA. Scenario B was broken down into 4 sub-scenarios, hereafter referred to as "LCSx", as can be seen in Table 1. Scenario A was subsequently renamed LCS0 to simplify comparison in the remaining course of the LCA. The first three phases of the pavement construction and maintenance history: initial construction, maintenance 1 and maintenance 2, were equal for both scenarios (A and B). Equal construction solutions and materials were also considered. Thus, these construction and maintenance actions were not taken into account in the study. For the remaining maintenance tasks, equal dates for the maintenance actions were assumed for both scenarios. However, these scenarios presented different solutions for the bituminous mixture to be applied. As such, the scenario A represents the real road section with the applied solutions and maintenance action performed by the roadway concession holder; whereas the scenario B is a proposed alternative scenario to be evaluated in the study. Scenario B considered alternative solutions using hot recycled bituminous mixtures assessed assuming the same geometry, traffic increasing and maintenance periods applied in real road section (scenario A), as presented in Figure 2. The maintenances 3 and 4, applied in scenario A and simulated in scenario B, were as follows: • Maintenance 3 implied the milling of the bituminous dense-graded mixture (AC) of the surface course and its replacement by a new surface course. In scenario A, this course consisted of a virgin AC, while in scenario B this course incorporated virgin AC, together with a certain rate of RAP that was obtained from the milling using a rejuvenator percentage. • Maintenance 4 implied the milling of both upper courses and their replacement by a new binder and surface course, respectively. In the case of scenario A, these courses consisted of stone mastic asphalt (SMA) mixtures with virgin materials, while in the courses in scenario B a certain RAP rate was incorporated together with virgin AC using a rejuvenator percentage.
To assess the use of RAP material in new bituminous mixtures, different incorporation rates were considered in this LCA. Scenario B was broken down into 4 sub-scenarios, hereafter referred to as "LCSx", as can be seen in Table 1. Scenario A was subsequently renamed LCS0 to simplify comparison in the remaining course of the LCA.

System Description and Boundaries
Both base and surface courses were set as the boundaries of the LCA analysis. Due to the insufficient and outdated data regarding wet mix macadam (WMM) and penetration macadam (PM) courses, these were not considered in the LCA. Given the comparative nature of this LCA and the fact that these two courses were present throughout all scenarios, this omission had no relative impact on the LCA outcome. The following boundaries were subsequently included:

•
The construction of the courses, as well as the rehabilitation and maintenance activities, were limited by the previously defined boundaries and were in conformity with the dates and course thicknesses of the real section (Scenario A).

•
Raw materials extraction needs to produce the mixtures applied in those courses.
All the transportation needs as refers to the transport of raw materials from the suppliers to the bituminous plant, and from the bituminous plant to the work-site and vice-versa. Considering the RAP system boundaries, a distinction was established between the pre-and post-processing of the RAP. A 'cut-off' allocation approach, as described in [6], implied that only the post-processing of recycled material, such as RAP should be accounted for by the system. In this case, this would imply that any environmental impact, resulting from the milling and hauling of RAP, would be excluded from the system. These processes, however, were already an inherent part of the considered M&R activities of the pavement and were, therefore, accounted for in the LCA.

Collected Data
Two types of data are essential for an LCA [74]: • Primary data, which is specific for the production, processes of the product or service. This data is obtained from the goods' producers, and the operators of processes and services, as well as their associations. • Secondary data presents generic and/or average data for the studied solution, considering the products and operations. Secondary data can be obtained from the sources of primary data, sometimes with some modifications, and from national databases, consultants and research groups.
For this study, the data was selected to be as representative of the Portuguese conditions during the PAP as possible. The sources for these data included national road authorities, active construction companies and technical experts. The data was mostly related to the inventory analysis of raw material extraction and production, fuels, construction and transportation vehicles and machines. This information was obtained primarily from the Ecoinvent database, but was modified whenever possible and appropriate to approach the Portuguese reality. The different data were combined in SimaPro to model the life cycles for the five different solutions defined as LCSx, where x is the number of the solution [75,76].

General Implementation
The life cycle inventory (LCI) stage consisted of the actual data collection and modelling of the system. First, this paragraph briefly discusses both the functionality and the modelling features of SimaPro. Secondly, it addresses the data sources, the calculations that were performed to provide significant input and the modelling of the distinct phases of each LCSx in SimaPro.

Extraction Phase
The extraction phase of materials comprised the two sub-phases as follows: production and transport of virgin materials. To calculate the required amount of virgin aggregate and virgin binder for each course, several calculations were performed. The starting point of these calculations was the primary data, obtained from a variety of sources, in combination with the characteristics of the functional unit, the length and width of the unit of pavement and the depth of each course.
First, to determine these amounts, a fixed binder-aggregate ratio of 95-5 was determined for all mixtures, except for the SMA mixtures, which have a 93.5-6.5 ratio. This implies that, regardless of the technical specifications, the majority of mixtures contains 95% aggregate and 5% binder. Furthermore, to derive the amount of virgin aggregate and binder from the total amount of material, the RAP incorporation percentage for each specific course was determined. Table 2 shows the Ecoinvent unit processes associated with the production of virgin aggregate and binder, as well as the transport of these materials by truck. Table 2. Material extraction processes and corresponding SimaPro unit processes.

Process Definition SimaPro Unit Process
Production of virgin aggregate Gravel, crushed (RoW) | production Production of virgin bituminous binder Pitch (RoW) | petroleum refinery operation Transport of aggregate and binder Transport, freight, lorry > 32 metric tonne, EURO4 (GLO) The second sub-phase of the extraction of materials involved the transport of virgin material to the bituminous mixing plant. The selected unit process for transport by truck in SimaPro, "Transport, freight, lorry > 32 metric tonnes, EURO4 (GLO), has a default unit of tonne-kilometre (tkm)". To define this amount for each course, the total mass of transported material and the distance between the production site and the mixing plant were calculated. A distance of 100 km and 150 km was considered from the quarry to the mixing plant and from the binder supplier to the mixing plant, respectively.

Production Phase
The mixture production phase addresses the environmental impacts related to the production of the different mixtures considered in the system. It was assumed that all the mixtures were produced at a conventional heavy fuel oil (HFO) fired batch mix plant. To account for variations in composition, mixing temperature, moisture content of aggregates and initial temperature of raw materials, of the several types of mixtures, the thermal energy (TE), required to produce the different bituminous mixtures, was determined according to Equation (1) [67]. where, • TE is the thermal energy (MJ/tonmixture) necessary to produce one ton of bituminous mixture; Recycling 2020, 5, 17

of 21
• m i is the mass of aggregates of fraction i; • C i is the specific heat capacity coefficient of aggregate fraction i; • M is the total number of aggregate fractions; • t mix is the mixing temperature of a bituminous mixture; • t 0 is the ambient temperature; • m bit is the mass of bitumen; • C bit is the specific heat capacity coefficient of bitumen; • W i is the water content of aggregates of fraction i; • C w is the specific heat capacity coefficient of water; • L v is the latent heat required to evaporate water; • C vap is the specific heat capacity coefficient of water vapour; • CL is the casing losses factor.
The casing losses (CL) is defined as the thermal energy that is lost by heating plant iron, instead of being used to heat the mixture components [21]. This factor was considered the same for all mixtures presented in this study, in accordance with the findings presented by Santos et al. (2018).
The different parameter values were based on literature and on average values in real practice [21,67,77]. Table 3 shows the values of the parameters used to calculate the thermal energy. For the mixtures, the standardized values were applied according to EN 12697-35 [77]. The actual environmental impact, resulting from the production of the bituminous mixtures, was modelled by the SimaPro unit process "heat production, heavy fuel oil, at industrial furnace 1 MW | heat, district or industrial, other than natural gas | cut-off, U" from the Ecoinvent database.
In addition to the environmental damage, resulting from the heating of the mixtures, the mixture production phase of courses incorporating RAP mixtures, also partly accounted for the impact of RAP processing. The RAP processing sub-phase was divided into pre-processing and post-processing. The pre-processing of RAP was therefore attributed to the construction and M&R phases, while the post-processing was attributed to the mixture production phase. The post-processing of RAP consisted of four activities (crushing, stacking, conveying and screening), and was considered to have a combined capacity of 184 tonnes per hour [6].
To account for the environmental impact of RAP production in SimaPro, the four main RAP production activities, mentioned above (crushing, stacking, conveying and screening), were modelled in SimaPro by different unit processes, as Table 4 shows. As this unit process is generally expressed in hours (hr), the total amount of hours required for producing each course was thus used as input for both activities in SimaPro.

Construction and M&R Phases
The construction and M&R phases have considerable environmental impacts on the system that need to be assessed. These impacts namely result from the machine operations involved in the construction of the courses and from the transportation of bituminous mixtures and milled material. The data concerning the activities and machine efficiencies were based on the literature review and on consultations to contractors and road experts.
Following the methodology used by other authors [67], machine operations were modelled in SimaPro using two different unit processes from the database, depending on the power of the machinery used. To describe the light machine operations, the SimaPro unit process 'Machine operation, diesel, ≥18.64 kW and <74.57 kW, high load factor' was used. The heavy machine operations were modelled by the unit process 'Machine operation, diesel, ≥74.57 kW, high load factor'. For the transportation by truck, the 'Transport, freight, lorry > 32 metric tonnes, EURO4 | cut-off, U' was adopted.

Work-Zone Traffic Management and Use Phases
The work-zone traffic management phase accounts for the differential fuel cost and emissions released by on-road vehicles, due to congestions generated during M&R actions, in comparison to those generated during normal road operation [67]. Considering the nature of the analysis and the fact of the study evaluating a real section of road, due to the lack of realistic data, as refers to traffic management during maintenance phases, the environmental impacts of this phase were not included.
The use phase takes account of the environmental impacts, which result from the interaction of the pavement with the vehicles, on the environment and humans throughout its PAP [67]. Other factors have been considered during the usage phase of the pavement, namely: pavement-vehicle interaction, traffic flow, leachate, carbonization and lighting [67].
In this study, however, the usage phase was not taken into account, although the contribution of this phase to the overall environmental impact of the pavement life cycle is potentially significant (or even dominant) [78]. Again, the reason for this omission lies on the lack of well-documented information regarding the case-study. However, such omission did not have significant impacts, due to the comparative basis of the analysis performed.

End-of-Life Phase
The end-of-life (EOL) phase of a pavement includes the destination of the pavement after its PAP. There are two main possible destinations for a given pavement: (1) remaining in place, or, (2) removal [67]. In this study, the pavement was assumed to remain in place and to undergo maintenance 4. All environmental impacts of this phase were therefore considered in the extraction of materials, in the mixture production, in construction, and in M&R phase.

Life Cycle Impact Assessment
The third step of the LCA, the life cycle impact assessment (LCIA), addresses the relationship between the inventory and the exterior; what kind of effect the system has on humans and the environment. It consisted of the following steps: classification, characterization and optional normalization, grouping, and weighting [17]. The midpoint impact category and endpoint normalization results were calculated by applying the life cycle impact method IMPACT 2002+. IMPACT 2002+ is a combination of four methods: IMPACT 2002 [22], Eco-indicator 99 [79], CML [80] and IPCC [23].

LCA Global Results
To evaluate the high RAP incorporation rates, and more specifically the multi-recycling rehabilitation, five rehabilitation scenarios for the functional unit were considered, which were referred to as LCS0, LCS1, LCS2, LCS3 and LCS4 (Table 5). Classification is the first step of the LCIA, where each elementary flow is assigned to a certain impact category, in consonance with the substances' potential for this category. In the second step (characterization), the individual emissions from the elementary flows that contribute to a single impact category are summed up. Firstly, however, the emissions are converted into indicators, using factors calculated by the IMPACT 2002+ model, in order to account for their relative contribution to a certain impact category. Table 5 shows the total results for each scenario.  Figure 3 shows the relative decrease in the characterization indicator for each scenario compared to the baseline scenario LCS0. The total average decrease for LCS1, LCS2, LCS3 and LCS4, was 20%, 28%, 36% and 40%, respectively. As can be concluded from this figure, the use of RAP and multi-recycling led to a considerable decrease in the environmental impact across all characterization factors. Moreover, this decrease was proportionate to the rate of RAP considered in the scenario; all indicators decreased as the rate of RAP increased. For the LCS4 scenario, the biggest decrease can be observed in the mineral extraction category (46%), while the global warming category displays the smallest decrease (31%).
The second part of the LCIA consisted of a damage assessment. According to ISO standards, this step is optional for an LCIA and is strongly similar to the characterization one. In the damage assessment, however, the categories were defined from an endpoint approach, in contrast to the midpoint approach used in the characterization step. Each damage category was thus compiled from several impact categories. The damage categories provided by the IMPACT 2002+ method, together with their units, are presented in Table 6. This table also includes the different impact categories assigned to each damage category. Recycling 2020, 5, x FOR PEER REVIEW 11 of 22 The second part of the LCIA consisted of a damage assessment. According to ISO standards, this step is optional for an LCIA and is strongly similar to the characterization one. In the damage assessment, however, the categories were defined from an endpoint approach, in contrast to the midpoint approach used in the characterization step. Each damage category was thus compiled from several impact categories. The damage categories provided by the IMPACT 2002+ method, together with their units, are presented in Table 6. This table also includes the different impact categories assigned to each damage category.  Table 7 shows the total damage assessment results for each scenario. As can be expected from the characterization results, a similar decrease in impact can be observed in all damage categories for scenarios comprising RAP. Furthermore, as Figure 4 shows, this decrease is again proportional to the percentage of RAP included in the mixtures. The biggest damage category decrease was observed for resources (43% for LCS4), and the smallest for climate change (31% for LCS4).   Table 7 shows the total damage assessment results for each scenario. As can be expected from the characterization results, a similar decrease in impact can be observed in all damage categories for scenarios comprising RAP. Furthermore, as Figure 4 shows, this decrease is again proportional to the percentage of RAP included in the mixtures. The biggest damage category decrease was observed for resources (43% for LCS4), and the smallest for climate change (31% for LCS4).   The third step, normalization, was also optional according to the ISO standards. Nonetheless, it can provide valuable insight about the extent of an impact category result with a specific reference. To normalize the damage factors, each impact per unit of emission was divided by the total impact of all substances of the specific category for which characterization factors exist, per person per year (for Europe) [81]. These numbers are ratios and normalization, and therefore, resolve the mismatch of units (see Table 8). The normalization results provided a similar output to the damage assessment. Figure 5 shows the relative decrease in the normalization factor for each scenario, compared to the baseline scenario. The normalization results show great resemblance with the damage assessment outcome; the indicators of each category decrease as the RAP rate increases. For LCS4, the biggest difference was again observed in the resources category (43%) and the smallest in the climate change category (31%). The third step, normalization, was also optional according to the ISO standards. Nonetheless, it can provide valuable insight about the extent of an impact category result with a specific reference. To normalize the damage factors, each impact per unit of emission was divided by the total impact of all substances of the specific category for which characterization factors exist, per person per year (for Europe) [81]. These numbers are ratios and normalization, and therefore, resolve the mismatch of units (see Table 8). The normalization results provided a similar output to the damage assessment. Figure 5 shows the relative decrease in the normalization factor for each scenario, compared to the baseline scenario. The normalization results show great resemblance with the damage assessment outcome; the indicators of each category decrease as the RAP rate increases. For LCS4, the biggest difference was again observed in the resources category (43%) and the smallest in the climate change category (31%). A linear correlation between the RAP incorporation rate and the environmental impact can be observed. This seemed to indicate that, theoretically, when constructing or maintaining flexible pavement courses, the solutions with high RAP recycling rates should be taken into consideration.
The results from these three steps can be explained by the following aspects: (i) replacing virgin material for RAP resulted in an obvious reduction in virgin aggregate and binder production; (ii) the contribution of RAP production process were insignificant when compared with the aggregate and binder production process; (iii) multi-recycling, or in this case study, the reuse of RAP during the 4th maintenance resulted in a second reduction in virgin materials, in comparison with the baseline A linear correlation between the RAP incorporation rate and the environmental impact can be observed. This seemed to indicate that, theoretically, when constructing or maintaining flexible pavement courses, the solutions with high RAP recycling rates should be taken into consideration.
The results from these three steps can be explained by the following aspects: (i) replacing virgin material for RAP resulted in an obvious reduction in virgin aggregate and binder production; (ii) the contribution of RAP production process were insignificant when compared with the aggregate and binder production process; (iii) multi-recycling, or in this case study, the reuse of RAP during the 4th maintenance resulted in a second reduction in virgin materials, in comparison with the baseline scenario; (iv) due to the reduction in virgin material, there was also a reduction in virgin material transport and, hence, in the overall transport impact; (v) the SMA mixtures considered, in the baseline scenario, had a higher mixing temperature than the corresponding RAP mixtures in the alternative scenarios, thus leading to a reduction in mixture production emissions and energy.

Assessment of RAP's Impacts on Mixture Production
This section only evaluates the impacts of the phases directly related with the mixture production, since RAP recycling has only direct impacts in terms of raw materials needs, RAP treatments, and additional transport distance. The following topics are intended to show the impacts of the use of this material on the overall process associated with the production of mixtures.
The LCA results from the five scenarios considered in this paper were divided by the following life cycle phases: virgin material extraction (consisting of virgin aggregate production and virgin binder production), mixture production, RAP production and transport. This makes it possible to assess the contribution of each phase to the overall environmental impact of the life cycle of each scenario; the characterization results for each category being divided by each life cycle phase. These contribution results for LCS0, LCS1, LCS2, LCS3 and LCS4 are shown in Figures 6-10, respectively.         From these figures, it can be concluded that the production of virgin bitumen accounts for the biggest environmental impact. It has the biggest contribution to 12, 11, 11, 9 and 10 impact categories for LCS0, LCS1, LCS2, LCS3 and LCS4, respectively. This result was in line with the results obtained in other studies [11,78,[82][83][84]. Transport and mixture production had the biggest contribution to only one, two or three categories, depending on the scenario. The production of virgin aggregate accounted for the biggest contribution to only one impact category, mineral extraction, but across all scenarios. The environmental impact of RAP production, even in LCS4, never contributed the most  From these figures, it can be concluded that the production of virgin bitumen accounts for the biggest environmental impact. It has the biggest contribution to 12, 11, 11, 9 and 10 impact categories for LCS0, LCS1, LCS2, LCS3 and LCS4, respectively. This result was in line with the results obtained in other studies [11,78,[82][83][84]. Transport and mixture production had the biggest contribution to only one, two or three categories, depending on the scenario. The production of virgin aggregate accounted for the biggest contribution to only one impact category, mineral extraction, but across all scenarios. The environmental impact of RAP production, even in LCS4, never contributed the most to any impact category. From these figures, it can be concluded that the production of virgin bitumen accounts for the biggest environmental impact. It has the biggest contribution to 12, 11, 11, 9 and 10 impact categories for LCS0, LCS1, LCS2, LCS3 and LCS4, respectively. This result was in line with the results obtained in other studies [11,78,[82][83][84]. Transport and mixture production had the biggest contribution to only one, two or three categories, depending on the scenario. The production of virgin aggregate accounted for the biggest contribution to only one impact category, mineral extraction, but across all scenarios. The environmental impact of RAP production, even in LCS4, never contributed the most to any impact category.
For each rehabilitation scenario, the relative influence of the different life cycle phases was assessed in terms of the different damage categories presented in LCA global results. This provides a more comprehensible and intuitive approach to the relative contribution of each life cycle phase to the overall environmental impact; the mentioned results for each impact category are again combined into damage factors. The results obtained for each damage category are presented in Figures 11-14.                Similar conclusions can be drawn from these figures in comparison with the ones showing the contribution to all impact categories. The contribution of RAP production is insignificant for all damage categories, across all scenarios. Even the relative contribution of 2% in LCS4 (100% of RAP incorporation) to human health, the largest impact observed, is still very small. On average, the production of bitumen delivers the greatest contribution across all damage categories except for Similar conclusions can be drawn from these figures in comparison with the ones showing the contribution to all impact categories. The contribution of RAP production is insignificant for all damage categories, across all scenarios. Even the relative contribution of 2% in LCS4 (100% of RAP incorporation) to human health, the largest impact observed, is still very small. On average, the production of bitumen delivers the greatest contribution across all damage categories except for climate change, in which the mixture production has the biggest impact. For the resource damage category, in particular, the contribution is almost 80% in the LCS0.

Conclusions
The main purpose of this paper was to evaluate the impact of high RAP incorporation and multi-recycling of bituminous mixtures in flexible pavement management. Such an impact was assessed from an environmental perspective by performing an LCA. The functional unit of this case study consisted of a flexible pavement of a road section with 1-km length and 7 m width, throughout a 69-year performance analysis period. The case study compared five maintenance and rehabilitation scenarios, referred to as 'LCSx', for this section, each with a different RAP incorporation rate in the bituminous mixtures: 0%, 25%, 50%, 75% and 100%, respectively.
The case study LCA findings were the subject of the LCIA and included characterization, damage assessment, normalization and single score results. The characterization step demonstrated the environmental impact of the five LCS life cycles across 15 impact categories. Throughout all categories, without exception, a general decrease in impacts can be observed in proportion to the RAP rate used.
The most important conclusions that can be drawn from this study are the following: • The incorporation of RAP in bituminous mixtures, and especially considering its multi-recycling, had lesser environmental impacts than the use of solutions with only virgin materials.

•
The most significant impact factors considered in this case study were human health, natural resources and climate change or global warming. The LCA case study demonstrated that across all factors, RAP incorporation is beneficial and significantly reduces the impact. • Moreover, a linear correlation was established between the RAP incorporation rate and the environmental impact. This suggested that, theoretically, when constructing or maintaining flexible pavement courses, a 100% RAP incorporation (or the complete recycling and multi-recycling of materials) should be pursued.
Although these results may seem promising, it should be considered that this case study represented certain practices, and hence it should be adapted for other construction, maintenance and rehabilitation scenarios, and for different geographic regions or types of pavements.
It should be pointed out that the data for this study was obtained from multiple active companies and experts to provide the most possible accurate and comprehensive input for the case study.
In conclusion, the case study of a road section rehabilitation has demonstrated a clear decrease in the environmental impact when RAP was reused, either once or multiple times, during M&R phases. The multi-recycling of flexible pavement should, therefore, be favored over using virgin materials in new bituminous mixtures. Besides the study on the environmental benefits of RAP multi-recycling, significant worldwide research has been carried out to address the mechanical and durability advantages of this technology with promising results.
Bearing in mind that decision-making in pavement management is a fairly difficult approach, the LCA approach should be integrated toward a multi-purpose optimization framework and should include factors such as: social dimension, structural objectives, constraints and life cycle cost analyses.
In this way, it should be performed to calculate the associated costs. Furthermore, it should be noted that the choice for the software tools to perform an LCA, as well as for the database and the impact assessment method has a distinctive influence on results. Therefore, further work should be done with a view to cross-examine the same functional unit across different tools.