Ground Glass Pozzolan in Conventional, High, and Ultra-High Performance Concrete

Ground-glass pozzolan (G) obtained by grinding the mixed-waste glass to same fineness of cement can act as a supplementary-cementitious material (SCM), given that it is an amorphous and a pozzolanic material. The G showed promising performances in different concrete types such as conventional concrete (CC), high-performance concrete (HPC), and ultra-high performance concrete (UHPC). The current paper reports on the characteristics and performance of G in these concrete types. The use of G provides several advantages (technological, economical, and environmental). It reduces the production cost of concrete and decrease the carbon footprint of a traditional concrete structures. The rheology of fresh concrete can be improved due to the replacement of cement by non-absorptive glass particles. Strength and rigidity improvements in the concrete containing G are due to the fact that glass particles act as inclusions having a very high strength and elastic modulus that have a strengthening effect on the overall hardened matrix.


Introduction
The use of supplementary-cementitious materials (SCMs) can help cement-based material construction achieving sustainable development by reducing cement production and therefore reduce fossil-fuel consumption and greenhouse-gas emissions. Alternative locally available SCMs are favorable when traditional SCMs are not locally available or costly if transported. Concrete industry is currently seeking additional sources of SCMs to supplement fly ash, slag cement, and silica fume that are currently used in concrete. There are several million tons of recycled glass available and are widely dispersed across the globe. The groundglass pozzolan (G) obtained by grinding the mixed-waste glass to same fineness of cement can act as an SCM, especially it is amorphous and pozzolanic material. The mixed colored waste glass cannot be recycled and in the best scenarios, is stockpiled, causing obvious environmental and economic problems. Therefore, valorization of this waste glass as an SCM could provide dual advantages. Over decades, the G showed promising performance in mortar and concrete mixtures both in laboratory and in large-scale field applications with more focus on long-term performance. The purpose of this paper is to summarize material specifications for the G for use as a pozzolan as a partial replacement of portland cement in concrete. The overall performance (fresh, mechanical, durability, and life-cycle assessment) of incorporating G in cement-based materials (CC, HPC, and UHPC) in both laboratory and field applications are highlighted in this paper.

Mix design and fresh properties
This section of the paper focuses on the performance of the G in two types of concrete: CC and HPC. The CC is a concrete with water-to-binder ratio (w/b) of 0.55 or 0.42 with respective binder content of 350 and 390 kg/m 3 . The HPC is a concrete with a w/b of 0.30 with relatively higher binder content of 460 kg/m 3 . The slump value of the CC is in the range of ±110 mm and that for the HPC is about ±210 mm. The air contents of the two types of concrete varied between 4% and 8 %. These two types of concretes can contain different rates of G in partial substitution of cement. Figure 1 shows the compressive strength (f' c ) of the CC ( Fig. 1-a) and HPC ( Fig. 1-b) concrete types. Of course, the HPC presented higher f' c than CC due to the effect of its low w/b. Whether in CC or HPC, the f' c of concrete mixtures incorporating G are low at an early age but increase significantly in the long term. Similarly to the

CMSS-2017
typical pozzolanic materials, the G concretes have lower f' c (about 12%) than the control at early age (less than 91 days), but they develop greater f' c than the control (about 18%) at long-term (beyond the 91 days). The low earlyage f' c results from the dilution effect and the slow reactivity of the G. The improvement of the long-term f' c results is due the pozzolanic activity of the G [1,2].

Chloride-Ion Permeability
The results of the chloride-ion permeability of the CC at 28 and 91 days are shown in Fig. (2-a) while the results at 28 and 56 days for the HPC are presented in Fig. (2-b). The incorporation of the G considerably reduces the permeability in both types of concrete. However, the reduction in the CC is even greater with time than that in the HPC. The reduction is not noticeable enough at 28 days due to the slow reactivity of the G. I the CC, at 91 days, the G reduces the permeability by 75% or even 80% depending on the rate of replacement. The corresponding reduction in the HPC is about one third. The effect of G is particularly interesting in the CC by its significant role in the reduction of permeability. Indeed, the CC containing G with a w/b as high as 0.55 showed larger reduction in the permeability relative to that with a low w/b of 0.30. This demonstrates a greater contribution of the G in decreasing the permeability than a simple reduction in the water content or w/b. This important decrease in permeability in the presence of G results from its pozzolanic reaction, which consumes large portlandite crystals to produce additional C-S-H that fills the pores in the microstructure level [3][4][5][6]. It is the refining effect of the pores and the grains, which generates an increased tortuosity of the pores reflecting the decrease of the permeability.  Figure 3 shows the scaling resistance of the two types of concretes (CC and HPC). For all concrete mixtures containing or not G, the results of the mass losses versus the de-icing scaling cycles are lower than the BNQ 2621-905 standard limits demonstrating good resistance to scaling. However, the concrete mixtures containing G exhibited mass losses slightly greater than those of the control concrete. It is a usual trend observed for the supplementary cementitious materials that seem more susceptible to scaling [7]. The CC usually have mass losses about four times higher than those of the HPC due to the low w/b of the latter. Since the low w/b would reduce excessive bleeding likely to weaken the skin of the concrete and makes it more vulnerable to scaling.  Figure 4 shows the drying shrinkage of the two types of concretes. Whether in CC or HPCs, the mixtures containing G exhibit a drying shrinkage similar to that of the control, suggesting that the G does not enlarge the drying shrinkage of the concrete. The drying shrinkage of the HPC is very low, about one third of that of the CC, due to its low w/b that limits the evaporation of free water and correlatively the drying shrinkage. Shayan et al. [8] also reported similar trends. These results suggest that the G concrete may have less or similar deformation and lower risk of cracking due to drying shrinkage than a control.

UHPGC classes
Based on the research conducted to develop UHPGC using waste-glass materials [9,10], four different UHPGC classes can be delimited in responding to various construction demands ( Table 1)

Material properties and mixture proportion
Type high sulfate-resistance (HS) cement formulated with a low C 3 A content was selected.. Silica fume (SF) conforming the CSA A3000 Standards "Cementitious materials compendium", and three ground-glass pozzolans ground to different sizes [G, fine G (FGP), and glass sand (GS)] with same Na 2 O eq of 13% were used in the UHPGC mix designs. Table 2 provides the properties of these materials. This mixture had more than 400 kg/m 3 of G as cement replacement. A polycarboxylate-based HRWRA with a specific gravity of 1.09 and solid content of 40% was used. The PVA used were 13 mm in length and 0.2 mm in diameter, and had a specific gravity of 1.3 and tensile strength of 400 MPa.  Table 3 presents the mixture proportions for seven UHPGC mixtures covering the various concrete Classes described in Table 1. A 2% volume fraction of steel fiber was used for UHPGC-1 to 6 mixtures, while 2.5% PVA fiber was used for UHPGC-7 mixture. As shown in Table  3, the design allowed using G in all mixes in contents varying between 222 and 403 kg/m 3 to replace cement and quartz powder, FGP ranging from 53 to 113 kg/m 3 to replace SF, and GS contents between 25% to 50% replacement of QS.  Table 4 provides the workability and the f'c after 2 days of HC (f'c-2d-HC), 28 days of NC (f'c-28d-NC), and 91 days of NC (f'c-91d-NC) of the seven UHPGC mixtures. A mini-slump flow of more than 230 mm indicates higher concrete workability. Strength results satisfy the requirements given in Table 1

CC and HPC
The G concretes (with replacement ratios between 10% and 30% by cement content) were used to cast various structure elements in Quebec-Canada between 2006 and 2011 including, interior slabs, exterior slabs (sidewalks), and structural wall elements in various environmental conditions -indoors and outdoors). The performance of samples taken from these concretes and subjected to laboratory curing are presented in references [11,12]. The G used had fineness (retained on 45 μm, %) = 0%, maximum-particle size (d max ) = 40 μm, mean-particle size (d 50 ) = 12 μm, Blaine specific surface = 382 m 2 /kg, and specific gravity = 2.54. In addition to the laboratorycured measurements, the long-term evaluation of the Gconcrete performances were conducted by testing core samples (during the concrete life, they exposed to different environmental and service conditions, consolidation, finishing, curing conditions ..etc.). The core results compared to the concrete performances that were initially sampled at the time of casting and

CMSS-2017
laboratory cured (ages of 28 and 91 days) are presented in Fig. 5. The results showed improved mechanical and durability properties compared to the reference mixtures. The incorporation of G in concrete contribute to enhancing the microstructure of concrete (densify) especially with age allowing a significant reduction in the pore system and generating higher mechanical and durability properties. The use of G generates dense C-S-H gel rim around the G particles and densifies the cementitious matrix. The mechanical strength was improved with time due to the pozzolanic activity of G and the microstructure enhancement.

UHPGC
The developed UHPGC using G was used in the construction of two footbridges (Fig. 6) at the University of Sherbrooke showing a potential for the UHPGC to be used in future large-scale projects [13,14]. The G used in the UHPGC mix design for constructing the bridges had maximum particle size of 100 µm and specific gravity of 2.60. The concrete exhibited excellent workability and rheological properties due to the zero absorption of the glass particles and optimized packing density for the entire concrete matrix. The mechanical properties were found to be excellent allowing reductions in the bridge thickness (material saving), and maintenance costs (due to the excellent durability properties of the UHPGC.

Life cycle assessment
In order to assess the environmental footprint of a new technology, or product, life cycle assessment (LCA) is a robust and promoted tool. A study carried out by Quantis for Recy-Quebec [15], demonstrated that the valorization of G into concrete shows significant environmental benefits compared to usual end-of-life scenario of landfilling. This valorization project showed even higher environmental benefits from landfilling than being reused as raw material for glass bottle, glass aggregate or glass wool. The replacement of cement by G can significantly reduce the carbon footprint of a typical UHPC, as shown in Fig.  7 [10]. The study demonstrated that the use of G can secure a low carbon footprint while maintaining high strength -which is difficult to obtain with other pozzolanic materials (fly ash, slag, and limestone powder).

Fig. 7 Relationships between embodied CO 2eq for the cement content and 28-day compressive strength under NC [10]
A LCA study by Deschamps et al. [16] compared the environmental life cycle footprint of a conventional UHPC and UHPGC in the case of the pedestrian bridge (Fig 6). For the same concrete volume and service life, results showed that the UHPGC can secure better environmental impacts on all indicators (Fig. 8). It was mainly due to the reduction in cement content with the incorporation of more G. Indeed, cement production appears as a main contributor for most of the environmental indicators of concrete life cycle. It is also important to note that, considering the same volume, UHPGC have a very high environmental impact compared to the conventional concrete. But, considering the volume needed to fulfill the same service life span, UHPGC can reach an environmental impact less important than conventional concrete.