Experimental and numerical investigation of using waste glass aggregates in asphalt pavement to mitigate urban heat islands

In this study, the experimental and numerical effects of using waste glass as aggregates of asphalt pavement are evaluated. The main reason for using this waste material as aggregates of hot mix asphalt (HMA) is to alleviate an environmental problem associated with asphalt pavements called urban heat islands. This phenomenon can increase the temperature in urban areas compared to their suburbs. Regarding the experimental part, two different HMA mixtures containing 100% limestone aggregates (HMAL) and 100% glass aggregates (HMAG) are made in this study. An experimental setup is used to simulate the solar radiations on top of HMA specimens. As a result, thermal parameters, including thermal conductivity, thermal diffusivity, and specific heat capacity, are measured and calculated using the heat transfer equations and the heat transfer test. These results are then used to develop finite element models for two different pavement structures with different asphalt concrete layers (one of them with HMAL and the other with HMAG). Furthermore, the air temperature data, extracted by TRNSYS software for Bechar city in Algeria, are used for modeling. The surface temperature, first and second temperatures in the asphalt pavement are obtained. The results revealed that using this waste aggregate increased the surface temperature during the day, which can make it susceptible to rutting. However, it reduced the surface temperature at night. More importantly, the HMAL absorbs 34% and released 47% more heat than HMAG during days and nights. Hence, the HMAG performance can mitigate the UHI effects. Moreover, using this waste material as aggregates in HMA can introduce a recycling method with low costs.


Introduction
Urban heat island (UHI) is a phenomenon for an urban area where the temperature is higher in comparison with its surrounding and rural areas. Previous investigations revealed that compared to the surrounding urban areas, the temperature could be up to 12 ℃ higher in the center of urban regions (Voogt 2002). The UHI is not a new phenomenon, and it has been observed in most cities since 1820 in regions with different latitudes and climates. Compared to the surrounding cities, the average and minimum temperatures are higher in cities reducing the length of cold seasons and cold temperatures in cities (Gaston 2010). The UHI temperature difference is attributed to many reasons, including city structures, low-albedo asphalt pavements, deprivation or lack of vegetation and green areas, and the emitted heat from the transportation system. The local geographical and meteorological parameters, including temperatures, wind, and humidity, can also affect the UHI effects, and this phenomenon can be seen in most cities. This temperature increment in urban areas can cause serious health problems, and during the heatwave of summer in 2003 in Europe, 70,000 people died (Robine et al. 2008;Hirsch et al. 2021). This phenomenon has grabbed the researcher's attention for 20 years, revealing more information on the mechanism of the UHI. They reach several breakthroughs, two of which are of great importance; Firstly, the temperature difference is noticeable and important at night, and the second discovery is the intensification of the UHI in the summer with urban sprawl and population growth of cities (Alberti 2008). The UHI increases the temperature, resulting in impaired water, air quality, and air pollution due to the higher demand for air conditioners. These environmental issues have negative impacts on human health. The intense effect of these issues was observed in 2003, when many people died during the heatwave in cities of France, Italy, and Germany (Robine et al. 2008).
Besides, this type of paving is responsible for a significant amount of the UHI, and in previous studies, the maximum UHI in Nanjing, Shanghai, Beijing, and Hong Kong was approximately 6, 7, 10, and 10.5 °C, respectively (Zeng et al. 2009;Memon et al. 2009). As a result of this dark surface, solar radiation undergoes storage and absorption, especially during the summer months. Consequently, the UHI is linked to climate change (Nakayama and Fujita 2010;Founda 2011). According to a study conducted in several urban areas in the United States, pavement accounts for 29-39% of these surfaces (Akbari and Rose 2008). According to recent studies, the adverse effects of UHI are increasing worldwide; some vital actions need to be taken to limit these effects (Yang et al. 2015).
Asphalt pavements have a substantial impact on the air quality of cities because of high thermal inertia, low albedo and reflectivity, and their extension all over urban areas, which exacerbate the UHI effects. In this study, high-thermal resistant asphalt pavement is made of waste glass aggregates to reduce the heat accumulation in the pavement, contributing to the mitigation of the UHI effects. For this goal, the control hot mix asphalt containing 100% limestone aggregates (HMAL) is compared with hot mix asphalt with 100% glass aggregates (HMAG). Therefore, the thermal properties and thermal conduction of HMAL and HMAG specimens are evaluated, and the effects of using glass aggregates on the UHI are investigated. The experimental results are then used to develop finite element models by which the heat transfer in two different asphalt pavement structures is investigated. Finally, the effects of using these pavement structures on the UHI are discussed.

Literature review
Different research has been conducted to mitigate the UHI effects associated with the pavements, all of which used three main principles: raising the reflectivity of pavements, using the cooling potential of water inside pavements, and using materials with different thermal conductivities to change the heat transfer rate inside the asphalt pavement (Shamsaei et al. 2022). Developing cool pavements have addressed the problems associated with asphalt pavements' hot surfaces. Hence, various pavements with reflective surfaces, thermochromic asphalt pavements, evaporative pavements, pavements containing phase change materials (PCMs), high conductive pavements, and heat-harvesting pavements were evaluated to accomplish this goal (Qin 2015). Considering the various approaches, the ability of the asphalt pavement to transfer heat and the amount of solar radiation that it absorbs will determine the reduction in surface temperature . Regarding asphalt pavement's thermal properties, thermodynamic features and energy transfer can be important factors for the UHI. The thermodynamic features of asphalt pavement are affected by its thermal diffusivity and heat capacity, and energy transfer refers to the movement of energy between the pavement courses and the atmosphere (Stempihar et al. 2012). Pavement surface temperatures are affected by changes in thermal conductivity as well. Consequently, heat can pass from the pavement layers into the soil. (Chu et al. 2020). Therefore, asphalt pavements' thermal behavior plays a critical role in reducing UHI.
Many approaches were tested to alleviate the UHI effects of pavements, one of which is increasing the albedo of pavement surfaces. Albedo is a unitless and non-dimensional parameter indicating the amount of reflected solar energy from a surface. In this method, more solar energy is reflected from pavements' surfaces, reducing the heat absorption by radiation (Mammeri et al. 2015). The other method can be using insulating materials. The waste glass was one of these materials used as aggregates in pavements, and thermal conductivity tests were applied to this material in two studies. The thermal conductivity of waste glass aggregates was half of the limestone aggregates. The results indicated that the insulating behavior of waste glass could decline the heat transfer and heat entering in summer and frost penetration in winter (Berraha et al. 2018;Vaillancourt et al. 2019). In freezing and thawing conditions, waste glass aggregates showed insulating and capillary barrier potential (Picard 2018).
Altering asphalt concrete's thermal conductivity is another practical way to alleviate the UHI effect. Thermal conductivity is closely related to asphalt mixture materials. Asphalt concrete's thermal properties have been altered by the addition of aggregates, fillers, air voids, and binder additives. In asphalt pavement mixtures, aggregates account for most of the weight (90-95%) or volume (75-85%). Other components include fillers and binders. Hence, it is undeniable that materials play a major role in asphalt pavement mixture's thermal performance (Cong et al. 2014;Yinfei et al. 2014;Hu and Yu 2016;Vo et al. 2017;Petkova-Slipets and Zlateva 2018). According to recent studies, the thermal conductivity of asphalt pavement was decreased by using some materials. Using this technique reduces the temperature at the bottom of asphalt concrete and increases the surface temperature of the pavement (Doulos et al. 2004). Materials such as ceramic and expanded polypropylene beads, fly ash cenospheres, and lightweight aggregates covered with resin (bauxite, shale ceramsite, volcanic rock, pottery sand, and diatomite) were used to decline thermal conductivity (Mallick et al. 2004;Huang et al. 2009;González-Corrochano et al. 2011;Feng et al. 2013;Ren et al. 2014;Pancar 2016;Wang et al. 2018Wang et al. , 2019Deng et al. 2019;Yinfei et al. 2020).
Asphalt pavements with thermal resistance have several advantages, among which reducing the overall temperature of the pavement is the most important (Yinfei et al. 2014(Yinfei et al. , 2016. According to a study, 30% of the heat can be absorbed in the middle of the subgrade that can be transferred to the asphalt concrete at night (Yu et al. 2015). Additionally, improving the thermal resistance of asphalt pavement will prevent the bitumen from aging due to oxygen exposure (Hou et al. 2018). The UHI effects are mitigated when the surface temperature is lower and heat cannot be readily transferred through thermal resistance asphalt pavement (Mohajerani et al. 2017). In addition, numerical modeling can be a reliable method to estimate the pavement temperature used for the investigation of the UHI (Mammeri and Lallam 2019).
Besides the experimental tests, numerical methods have been used to address the pavements' UHI effects. One of the benefits of these studies was reducing the number of tests. Indeed, Mirzanamadi et al. (2018) developed some models based on randomly distributed microstructures. The input data were the thermal conductivity of aggregates, asphalt binder, and the air void. However, when the structure was randomly distributed, the error was about 7.94% (Mirzanamadi et al. 2018). This random distribution was improved, and the asphalt mixture was considered as a composite substance that has two phases, including the discrete one (aggregates and air voids) and the continuous one (asphalt matrix). Hence, the accuracy of the method was increased after considering the asphalt mixture as a heterogeneous substance (Han et al. 2021). The random distribution was also improved to introduce a volumebased model. As a result, air voids, aggregates, and asphalt binders were defined with identical cells. Thus, the volume of these particles in the mixture was used to define the number of cells associated with them in the model. Then, a mesh network should be developed randomly to have more accurate results (Chu et al. 2020). Considering all these experimental and numerical studies, more research is needed to understand the true impacts of this type of pavement on the UHI.

Materials
The specifications of hot mix asphalt materials, including limestone and waste glass aggregates and asphalt binder, are mentioned in this section. All specimens were made at the LCMB (Laboratoire sur chaussée et matériaux bitumineux) of the École de Technologie Supérieure in Montreal, Canada.

Limestone aggregates and waste glass aggregates specifications
The aggregates' gradation and physical properties are depicted in Fig. 1 and Table 1, respectively. The standard limits for aggregate gradation are based on MTMDET (2017), HMA Quebec Standard. Limestone is provided from a quarry in Saint-Philippe city and is extracted in several granular sizes by the Construction DJL company. The Tricentis non-profitable organization also supplied waste glass aggregates. A different range of glass aggregates used in this study is depicted in Fig. 2.

Asphalt binder specifications
The performance grade (PG) 70-28 binder, which is a common binder for road pavements in Montreal, was used to make asphalt concrete specimens. The properties of this type of binder are reported in Table 2.

Hot mix asphalt specifications
The asphalt concrete mixture was designed based on the HMA Quebec Standard [43]. The effective volumetric binder content of 12.2% (by mass) was chosen, for both limestone and glass, specimens as the optimal binder content. This is the volume of binder covering the aggregates without being absorbed by aggregates. The binder contents of limestone and glass specimens for tests were 5.4 and 4.9% (by mass), respectively. Firstly, limestone and glass aggregates were heated (around 180 ℃). They were then mixed with the heated binder (around 168 ℃). A ventilated oven was used to cure the loose HMA mixture for at least 30 min (less than 2 h). The mixture was then compacted by Superpave Gyratory Compactor (SGC) to make cylindrical specimens. The diameter of the specimens was 150 mm, and the target air void was 5.5%. The control specimens (HMAL) comprised 100% limestone aggregates, and the test specimens (HMAG) consisted of 100% waste glass aggregates. HMAL and HMAG had air void contents of 4.9% and 5.4%, respectively.

Experimental test to measure thermal properties
Five thermocouples were placed in different positions to measure the temperatures in different depths specimens to calculate thermal parameters. The position of thermocouples is shown in Fig. 3. Heat transfer rates in one direction (vertical) for both test and control specimens were measured. For this aim, the lateral sides of specimens are insulated with polystyrene to avoid heat exchange with the environment from these sides. They are then placed under a tungsten iodide (infrared) lamp (250 W) to simulate solar energy. As it is shown in Fig. 3, five thermocouples were placed in specimens to measure and  (2); Insulation surrounding cover made of polystyrene (3); the clamps rings for fastening (4); infrared heating lamp (5); and the temperature data record system (6) record the temperatures for 24 h. Different parts of this test method are depicted in Fig. 4.

Thermal parameters
Some significant parameters are associated with the thermal behavior of asphalt pavements and heat exchange with the environment, including thermal conductivity, specific heat capacity, density, and thermal diffusivity. The relationship between these parameters is shown in Eq. 1: where λ is thermal conductivity (W/m·K), C p is specific heat capacity (J/kg·K), ρ is density (kg/m 3 ), and α is thermal diffusivity (m 2 /s). The specific heat capacity (C p ) is measured in Eq. (2): where C p is specific heat capacity (J/kg·K), ΔE (J) is the amount of energy supplied by the lamp to increase the temperature of the specimen by ΔT (K), and m (kg) is the mass of the specimen. The energy supplied by the lamp is estimated by measuring the time needed to increase the temperature of a mass of 1 kg of water to a certain temperature. Also, the specific heat capacity of water is 4158 J/kg.K. Hence, the lamp delivers a power of 0.91 J/s to the surface of the test specimen. Moreover, ∆T is a function of two parameters: time (s) and positions of thermocouples (mm), by which the thermal diffusivity is calculated in Eq. 3: where α is thermal diffusivity (m 2 /s), t is time (s), and x is position (m + 2; m + 1; m; m-1; m + 2) in Fig. 3.

Finite element modeling
The thermal properties of HMAL and HMAG specimens were determined by laboratory tests, presented in Table 3. These properties were used as input parameters to develop a 2D numerical finite element (FE) model by the Cast3M software package. In this model, the heat transfer from the top of the surface and at different depths of asphalt concrete is considered, and it was verified with the experimental results. Moreover, the air temperature data was extracted by TRN-SYS software for energy simulation. The asphalt pavement structure consists of four layers, including the asphalt concrete with a thickness of 10 cm, a base layer whose thickness is 25 cm, a 30 cm-thickness subbase, and the subgrade layer was defined to develop the FE model, which is illustrated in Fig. 5. The numeric model is based on a finite element method using the Cast3M software. The input data are the thickness of the layers, material properties, and effective evolutions of climatic parameters. The edge effects induced by the road shoulders are supposed to be negligible so that the pavement area can be considered infinite, and the diffusion problem is 1D-type. However, the numerical model is chosen bidimensional, considering that Cast3M meshing options are restricted to 2D or 3D. The mesh is composed of stacking of 61 single eight-nodded quadrilaterals ('QUA8' Cast3M-type with quadratic interpolation), to spare computation time. Different material properties are assigned to the elements belonging to the respective layers of the pavement. The soil is extended to a depth of 20 m with elements of progressive size. An adiabatic boundary condition is imposed at

Transient conduction
Transient conduction is the thermal energy flow during which the temperature varies with time. The relationship between changing the temperature during the time with other thermal parameters is shown in Eq. 4: where ∆T is the temperature difference, α is thermal diffusivity (m 2 /s), and t is time (s).

Convection
Convection is the transfer of heat between the surface of the pavement body and the air, which is defined by Newton's law in Eq. 5: where T S is the surface temperature (K), T air is the air temperature (K), and h is the convective exchange coefficient (W/m 2 .k), which is calculated in Eq. 6: where k air = 0.027, P r (Prandtl number) = 0.7, L = 0.15 m, R e = VL∕ (V is the wind speed (m/s 2 ), and is the kinematic viscosity of the air, which is equivalent to 16.01 × 10 -6 m 2 /s).

Radiation
Radiation is another way of heat exchange between asphalt pavement and the environment. Solar radiation is partially absorbed by the pavement surface. In this study, solar radiation is considered in the modeling. For this aim, the absorption coefficient (α = 1-albedo) is assumed to be 0.9. The radiation from the surface to the environment is calculated in Eq. 7: where ε is the emissivity coefficient of the surface assumed to be 0.93, is the constant Stefan Boltzmann (5.67 × 10 -8 W/m 2 .K 4 ), T s is the surface temperature (K), and T sky is the temperature of sky radiation (K) depending on humidity, the ambient air temperature, the cloudiness factor of the sky, and the local atmospheric pressure. This parameter is calculated in Eq. 8: where 0 is the emissivity of the clear sky, C cover is the sky cloud cover factor. The implicit transient scheme (theta method), which is nonlinear to take the radiation equations into account, was used to develop numerical models by the Cast3M software package. For this aim, the boundary condition was applied on the surface, which can be a combination of the following cases: • Convective exchange with a variable air temperature in time. • Imposition of a variable heat flow of solar origin (irradiation). • Radiative exchange with an infinite environment.
The boundary condition imposed at the bottom of the soil is a zero flux. Therefore, the hottest month of the year in the studied region (July) was used for the analysis. Different thermal parameters, inclining air and sky temperature, solar flux, and the heat exchange coefficient related to the pavement, are depicted in Fig. 6.
The region tested in this model is the city of Bechar in Algeria. The maximum and minimum ambient temperatures of the studied period were 42.39 and 20.88 °C, respectively. The maximum and minimum sky temperatures were also 36.18 and −2 °C, respectively. The maximum solar flux incident on the surface over this period was 1032W/m 2 . In July, low-speed winds were observed in the city. As a result, the heat exchange coefficient was 16.74 W/m 2 K.

Results and discussion
The finite element analysis for specimens containing limestone (HMAL) and waste glass (HMAG) was performed. The base course materials were assumed the same for these two asphalt mixtures.

The surface temperature evaluation
The finite element analysis result for the surface of HMAL and HMAG specimens for the last ten days of the month is demonstrated in Fig. 7. As can be seen, the surface temperatures are higher in the early afternoon and lower late at night. Compared to HMAL specimens, the insulation property of glass avoids entering heat into the asphalt concrete. However, the accumulation of heat at the surface was increased, raising the surface temperature. The surface temperature for both HMAL and HMAG specimens (8) T sky = T air ( 0 + 0.8 1 − 0 C cover ) 0.25 and the air temperature is depicted in Fig. 7. Because of the less heat accumulation in HMAG specimens, it cooled down faster. More importantly, compared to days, the surface temperature is greater than the air temperature at night. As a result, the heat stored in the asphalt pavement is released into the air through convection at night, contributing to the UHI.

The surface and interface temperature analysis
To have a better thermal conduction comparison between HMAL and HMAG mixtures, the temperature between the interface of asphalt concrete and the base course was investigated. The surface temperatures and the temperatures in the interface of HMAL and HMAG and the base course for the last ten days of July are shown in Fig. 8. As mentioned in Sect. 3.1, a considerable fluctuation between the day and night surface temperatures is obvious. The temperature between the interface of the HMAG and the base course was decreased, and this temperature is less than the interface of the HMAL. This decrease is attributed to the depth of heat penetration declined by a cyclical fluctuation in surface temperature. The heat penetration depth of the surface temperature is directly related to the diffusivity and the thickness of materials. Besides, there is a phase shift in maximum temperatures (at the peak) between the surface and the first interface. It was also observed for the second interface. This phase shift is attributed to diffusivity. The difference in diffusivity between the HMAL layer (0.624 × 10 -6 m 2 /s) and the HMAG layer (0.383 × 10 -6 m 2 /s) caused a phase shift between the surface and the interface of the base course to be 2 h for the HMAL and 4 h for the HMAG. The phase shift between the first and the second interface of the HMAL and the HMAG is similar for both simulations because the diffusivity is the same in the base layer (0.904 × 10 -6 m 2 /s). This phase shift between the two interfaces of the base course is about 5 h. Although the diffusivity of the asphalt concrete layer is higher than the base course, the phase shift is higher in the base course. This higher phase shift is attributed to its greater thickness (25 cm for the base course and 10 cm for the asphalt concrete).
The average temperature differences during days and nights for the HMAL and the HMAG at each interface are reported in Table 4. As is seen in the table, the average maximum surface temperature during the day was observed at the HMAG (69.0 °C), and the minimum temperature at night was also observed at the HMAG (35.6 °C). The highest temperature difference was also observed at the surface of the HMAG, which is hotter during the day and cooler at night. Hence, compared to the HMAL, less heat is transferred at the first interface (asphalt concrete and the base course) of the HMAG. In stark contrast, a reversed trend was observed at this interface. Indeed, the average maximum interface temperature during the day was higher for the HMAL (55.8 °C), and the average maximum interface temperature at night was lower for the HMAL (42.3 °C). The first interface received more heat during the day for the HMAL and released more heat at night. Thus, the thermal exchange should be analyzed.

Heat exchange analysis on the paving surface and in the pavement structure
The heat flux process in asphalt pavement is of great significance. The thermal fluxes come in different ways. They can come from solar radiation to the ground (flux sol ). The other way can be the irradiation of the sky (flux Irr ), which includes the solar flux reflected by the surface. Finally, the thermal fluxes occur due to the convection between the air and the pavement surface (flux conv ). This can be compared to the conduction flux (flux cond ), which is heat transferring in the pavement. In this study, the data are taken every hour for the whole month of July. The flux sol is a basic measured data from solar irradiation. The flux conv and the flux Irr are calculated in Eqs. 5 and 7, respectively. The flux cond (W/m 2 ) is also calculated in Eq. 9.
where is thermal conductivity (W/m.K), and ∇T is the temperature gradient (K/m). The incoming and outgoing heat flows of the pavement are analyzed. The incoming flow (flux ent ) is similar to the flux sol , and the outgoing flow (flux out ) is the flux Irr . The flux conv can be either incoming or outgoing, depending on the relative temperature between the air and the pavement surface. If the pavement is warmer than the air, the flux conv is outgoing and negative. In this case study, the flux conv is always negative because the temperature at the road surface is always higher than the air temperature. Two thermal balances were made to determine if there was an accumulation or release of heat in the pavement.
(9) q cond = − ∇T Firstly, the heat balance, which is the difference between the flux ent and flux out , was calculated. If the balance is positive between the flux ent and flux out , heat is absorbed by the surface, and if the value is negative, heat is emitted from the surface. The conduction heat balance was then measured to show how much heat entered or emitted from the asphalt concrete and the base layer. The heat exchange for these layers occurs through the surface and the base course. The heat balance between the environment and the HMAG pavement surface and the conduction heat balance within the HMAG layer between the surface and the second interface is demonstrated in Fig. 9.
The blue and black curves show the heat fluxes entering and emitting from the asphalt surface, respectively. The heat balance is shown with red curves, and the green curves show the conduction heat balance between the surface and the second interface. The analysis of the curves shows that the surface heat balance during July is almost constant. The amount of released heat is a bit more than the amount of heat absorption. The slope of the linear regression line of the heat balance (red dashed line) is almost zero, which is almost -50 W/m 2 . The conductive heat balance is positive (entering heat) at the beginning of the month, so the heat was absorbed, but it decreased during the month and reached zero. The slope of the linear regression line of the conductive heat balance (green dashed line) showed a decrease in heat absorption. Because the asphalt accumulates less and less heat during the month, and the heat balance with the environment was changed a little, the heat was transferred into the base course. Hence, as is seen in this figure, the thermal flux is positive at the surface during the day, leading to heat absorption. The heat flux is negative at night, emitting heat from the surface to the atmosphere. The averages of hourly heat balance values and heat conduction balance for HMAL and HMAG during days and nights for July are reported in Table 5.
Regarding the results in Table 5, the heat balance between the surface and the atmosphere is smaller during the day than at night. The average heat absorption and heat release for the surface of HMAL during the day and at night were 199.7 W/m 2 and 203.5 W/m 2 , respectively. These values for the HMAG were also 160.8 W/m 2 and 168.9 W/m 2 , respectively. Compared to the HMAG, the HMAL absorbed and released 24 and 20 percent more heat in 24 h, respectively. This is attributed to the higher conductivity of limestone aggregates, and this higher heat release of HMAL exacerbates the UHI effects. The conduction balance indicated that the HMAL absorbed 34% and released 47% more heat than HMAG during days and nights. This higher heat release contributes to increasing the temperature of the pavement structure and maintaining higher temperatures in the pavement.
Regarding the conduction balance, more heat was absorbed by both materials during the day in comparison with their heat release at night. The heat release occurred in the direction of the surface and the base course. The positive difference between heat release and heat absorption indicates that heat was accumulated at the second interface, as shown in Fig. 10. The temperatures at the surface and at the first and second interfaces in July are depicted in this figure. The temperature was increased at the second interface between the base course and the subbase course by 14 °C. This increase was greater for the HMAL.

Discussion
The developed numerical models, based on laboratory measurements of thermal parameters of asphalt concrete mixtures with and without waste glass, showed a different thermal behavior for two pavement mixtures. As it was mentioned in the previous sections, the surface of the HMAG is warmer than the HMAL during the day. However, the reverse trend was observed at night, and the HMAG surface was cooler. Although the heat accumulated on the surface, it was not transferred into the pavement structure due to the higher thermal resistance of the HMAG.
Moreover, when the surface temperature of HMAG is higher, the modulus of the asphalt pavement surface is smaller. As a result, compared to the HMAL, it may be susceptible to rutting deformation. Therefore, a stiffer asphalt binder should be used at higher temperatures, or a stiffening additive such as lime will be added to the asphalt mixture. However, the temperature at the bottom of the HMAG layer was lower than the HMAL layer. This lower temperature brings about a higher modulus, which enhances the fatigue resistance of the HMAG. More importantly, the HMAG received less heat and remained much stiffer due to the higher thermal resistance. As a result, the HMAG is more durable under traffic loads, especially in arid regions where the daily thermal difference is significant.
The results of this work deal exclusively with the thermal behavior of a pavement structure made of conventional asphalt concrete and glass aggregates. Indeed, the pavement structure proposed here does not meet the expected performance of high-traffic pavements. On the other hand, in urban areas, where urban heat islands are observed, many paved surfaces, such as residential streets and large parking lots, do not require high-performance pavements. Besides, several parallel studies to this project are in progress, focusing on improving the physical properties and mechanical performance of high-strength asphalt mixes containing waste glass at the ETS Laboratory. From an environmental point of view, mitigation of the heat islands makes it possible to reduce the use of air conditioning which requires energy and generates greenhouse gasses. Besides, global warming which has been a serious issue can be alleviated when cities' temperatures are lower. Moreover, the recycling of post-consumer glass reduces the use of virgin aggregates whose production consumes a lot of energy. At the same time, the material has interesting thermal and mechanical properties and absorbs less bitumen.

Conclusion
In order to study the impact of bituminous pavements on urban heat islands, heat transfer tests were conducted on two asphalt concrete pavement structures. The first one had a bituminous overlay with 100% glass aggregate (HMAG), and the other was a conventional bituminous overlay with 100% limestone aggregate (HMAL). The thermal properties of these two asphalt mixtures were determined at the ÉTS bituminous materials laboratory. These experimental results and solar radiation data from an arid region of Algeria were also used to simulate the heat transfer in these two pavement structures with four layers using the finite element method.
It was concluded that the largest temperature differences between day and night occur at the surface of bituminous pavements. Compared to the HMAL mixture, a larger surface temperature difference was observed for the HMAG. The smallest temperature differences occurred at the base of the pavement, at the interface with the asphalt concrete. More importantly, the differences measured for HMAG were smaller than for HMAL. Thus, the high-thermal resistance aggregate reduced the heat absorption.
The comparison of heat balances obtained from heat exchanges between the environment and the pavement surface and between the surface and the granular base of the pavement showed that HMAL absorbs more heat during the day and emits more heat at night. This excess heat emission occurred in the direction of the base course and the pavement's surface, contributing to a temperature increase in the pavement and maintaining this high temperature inside the pavement structure. The pavement structure with the high-thermal resistance HMA reduced heat storage and heat emissions at night. Therefore, this suitable performance can mitigate the urban heat islands effects. Data availability The datasets generated during and/or analyzed during the current study are available from the corresponding author upon reasonable request.