Predicting the Hydration of Ground Granulated Blast Furnace Slag and Recycled Glass Blended Cements

Abstract

The use of recycled glass powder (RCGP) is investigated as a partial replacement for ground granulated blast furnace slag in blended CEM II/A-LL cements using thermodynamic modelling to simulate cement paste hydration at a water-to-cement (w/c) ratio of 0.5. This study allows a rapid means of examining the likely evolution of these materials over the first two to three years, allowing experimental work to focus on promising formulations. A comparison is made between the evolving solid phase and solution chemistries of four materials: a standard Portland-limestone (CEM II/A-LL), a ‘control’ blend, comprising equal quantities of CEM II/A-LL with GGBS and two novel blended cements containing RCGP. These represent 15% replacement (by mass) of GGBS by RCGP blended with either 40% or 60% CEM II/A-LL. The simulations were performed using the code HYDCEM, a cement hydration simulator, which calls on the thermodynamic model PHREEQC to sequentially simulate the evolution of the four cements. The results suggest that partial replacement of GGBS by 15% RCGP results in no significant change in system chemistry. The partial replacement of cementitious slag by waste container glass provides a route by which this material can be diverted from the landfill inventory, and the mass-balance and energy balance implications will be reported elsewhere.

1. Introduction

The global demand to combat climate change by the reduction of anthropogenic carbon dioxide is well established. The manufacture of cement is responsible for around 7% of CO₂ emissions, making it the third largest source of greenhouse gas after power generation and transport. In 1992, the World Business Council for Sustainable Development studied the cement industry [1], presenting the need for change in cement manufacture and concluded that “successful adoption of sustainable development by the cement industry will occur only if there is real synergy between sustainability and profitability”. The response from cement manufacturers has been extremely positive and that industrial synergy is now established and firmly embedded in cement production. Since that time, there have been a number of important studies [2,3,4,5] of how CO₂ reduction may be achieved through one or more of the following:

1.

Improve process efficiency through procedural change and plant modernization.

2.

Use of alternative fuels and alternative raw materials.

3.

Replace Portland cement clinker with supplementary cementitious materials (SCMs).

4.

Increase the use of high-strength concrete and so reduce the amount used in a single application.

5.

Develop alternative low-carbon binders not based on Portland cement.

6.

Carbon capture, storage and use (CCS, CCSU).

1.1. Previous Research

Research and development in each of these areas has accelerated over the last three decades, and the first two have been widely adopted across the entire cement industry. The development of alternative (non-Portland) clinkers shows great promise for the near future, but the adoption of CCS remains somewhat slower than might be hoped.

This paper reports the first application of thermodynamic modelling in the clinker–slag–limestone–glass system of blended cements and demonstrates how this approach allows the rapid comparison of blended cement formulations. In this way, subsequent experimental studies may be focused on the most promising compositions.

Vaitkevicius et al. [6] found, through XRD, that the addition of glass powder increases the dissolution rate of Portland cement with a corresponding acceleration of clinker hydration. XRD analysis undertaken by El-Tair and co-workers [7] found high silica content along with portlandite and calcite, which were found to contribute positively to the compressive strength in cement mortars cast as part of the study up to 28 days, which indicates that the reactivity of the RCGP peaked before then. However, with increasing RCGP additions, workability decreased. Similar effects on workability were found by Li et al. [8]. The replacement of glass powder in cement will increase the early compressive strength significantly and might reach 20%, after 7 days; however, there is no significant increase in compressive strength after 28 days, which means that the glass powders reactivity slows over time due to its amorphous nature.

Tamanna and Tuladhar [9] undertook XRD on blended RCGP with OPC and found crystalline clinker phases, portlandite and C-S-H. RCGP was found to exhibit a high amorphous peak between 15° and 35° 2θ, with no clear crystalline trends observed. Similar peaks were observed by Nasiru et al. [10] and Li et al. [8]. In contrast to El-Tair et al. [7], RGCP did not show significant early-age strength with a substantial strength development at 56 days. However, beyond a 10% replacement level, the compressive, flexural and tensile strength decreased. The researchers concluded that RCGP is a suitable cement replacement up to 10%, but long-term curing and a finer particle size distribution are mandatory for the successful use of RCGP with higher replacement levels. XRD analysis by Nasiru et al. [10] on mortar samples with different RCGP additions typically found portlandite, quartz and calcite peaks at lower replacement levels. At 30% RCGP replacement, no portlandite was observed, which indicates pozzolanic behaviour. The observed presence of clinker phases revealed hydration was still ongoing, which may suggest that the addition of RCGP slows down clinker dissolution. These findings support the authors’ conclusion that the addition of glass powder improved the overall strength and chloride penetration because of observed pozzolanic behaviour in XRD, SEM and TG analysis.

The effect of recycled waste glass powder as a partial replacement of cement in concretes containing silica fume and fly ash by Ibrahim [11] found that a 5% replacement demonstrated an increase in compressive and tensile strengths of ordinary concrete, but this was lowered when added to silica fume and fly ash concretes.

Zhang et al. [12] combined glass powder with Portland cement to stabilize dredged sediment. They found that it did help stabilize the sediment, with strength increased with ongoing curing age, glass fineness and water content. It was also found to improve density. Tran et al. [13] found that the presence of 40 wt% recycled glass powder in a binder significantly mitigated ASR by reducing OH⁻ concentration and limiting alkali ingress, despite a lower strength gain. They also concluded that there is a 36% lowering of CO₂ with a 40wt% addition.

The greatest immediate CO₂ saving is in clinker replacement, using pozzolanic materials as SCM additions. These divide into two obvious groups; materials derived from nature, such as calcined clays, limestone flour and natural pozzolans, and industrial SCMs, such as blast furnace slag and combustion ashes (largely from coal—Pulverized Fuel Ash (PFA), waste and wood/plant product incineration), along with many other by-product materials. Of the former group, calcined clays have seen substantial global interest, following the development of LC³ formulations which show considerable promise (see https://lc3.ch/, accessed on 1 June 2025). Ground limestone is an important component of blended cements containing a source of reactive alumina, as this may be consumed during hydration [2]. The use of more traditional SCMs (slag and ash) continues apace, however, reflecting their desirable technical properties, familiarity and wide availability.

Despite the success of cement manufacturers in reducing the embedded CO₂ per tonne of cement, global economic growth has increased our consumption of cement, increasing the need for a dramatic reduction in greenhouse gas emissions. This need is immediate, and the role of blast furnace slag cement in achieving a step change is the subject of this paper.

Blast furnace slag is a by-product of iron-ore reduction, comprising glassy silicates which are recovered and granulated by rapidly quenching the slag with water. The granules are ground to produce cementitious slag ‘ground granulated blast furnace slag’ (GGBS), which is used to replace Portland cement by up to 70%, or a little more by mass. Concretes made with blended Portland-slag cement have a high durability, especially in marine applications, and remain in great demand. Around 330 Mt/year of cementitious slag is produced worldwide, and around 90% is used as a blending component in cements [4]. Changes in the iron and steel industry are likely to decrease the availability of slag in the medium term as the fraction of steel produced from scrap using electric arc furnaces increases, relative to primary iron production. The decision by Tata Steel to close the two remaining blast furnaces in Port Talbot, South Wales, has put some strain on GGBS supplies in the British Isles, so reliance on imported slag (with its associated ‘carbon miles’) will soon increase. Consequently, there is an urgent demand to conserve slag consumption, which has prompted the company Ecocem Ireland Ltd. to investigate blending cementitious slag with recycled glass powder (RCGP) as an industrial pozzolan, or supplementary cementitious material.

1.2. Recycled Glass Powder as a Supplementary Cementitious Material

In principle, glass is 100% recyclable, and every tonne of glass re-melted saves ~580 kg of carbon dioxide emissions [14], as well as decreasing the energy needed to make glass and reducing reliance on virgin raw materials. Recycling rates across Europe vary; for example, in the British Isles, 84% of glass is recycled in Ireland [15] and 74% is recycled in the UK [16]. Waste glass cullet is commonly recycled, but its reuse in producing new glass is limited by its composition. Recycled cullet not recovered for glass container manufacturing (due to unsuitable composition, sizing, contamination, or market conditions) is often disposed of in landfill, which attracts Landfill Tax, adding to the cost of disposal. Container glass is uneconomic to decolourize, which means coloured glass is unlikely to be recycled by the glass industry, making its use as an SCM particularly attractive. The recycling of container glass cullet contributes to national “circular economy” aspirations, while at the same time eliminates all costs associated with landfill.

The incorporation of glass in cements is not a new idea and has been the subject of study for some decades [17,18,19], and of course natural glasses have been used in cements since antiquity. There are two important factors to consider in using glass as an SCM, due to the potential occurrence of alkali–silica reactions (ASRs) in concrete. Alkalis in concrete mixes can be a cause for concern. High alkali loads in concrete can trigger deleterious expansion if reactive silicate aggregates are present. The first factor is the alkali metal content and the second is particle size.

Alkali metals are introduced as metal carbonate flux, used to lower the melting temperature of the glass and for container glass and comprise around 13% Na₂O by mass of the whole [20]. The alkali load in concrete is proportional to the binder content and the alkali level of the binder, expressed as equivalent sodium oxide content, where:

%Na₂O equivalent = Na₂O + 0.658%K₂O (by mass)

(1)

A suitable specification for recycled glass cullet suitable for “flux/binders” can be found in BS PAS 101 “Recovered container glass Specification for quality and guidance for good practice in collection” See Table A1 of ref [21]. Practically in cements, the alkali content of the binder is limited as a precaution against alkali silica reaction (ASR) at 0.6% (Equation (1)). In most countries, alkalis are limited to the equivalent Na₂O content per cubic meter of concrete at 1.8 to 4.5 kg/m³, depending on the reactivity classification of the aggregate used and on how variability in the binder alkali level is reported by the cement manufacturer. Alternatively, a ‘low alkali’ binder, defined as one that does not exceed 0.6% Na₂Oeq, may be used if aggregate reactivity is high or unknown.

The second consideration is the particle size of the glass. The suppression of alkali silica reaction is also addressed by fine grinding of the glass (to a few microns), as has been demonstrated elsewhere [19], and it has been found that the addition of glass powder can enhance ASR resistance if the particle size is below 300 microns [20]. This introduces another dimension to using glass powder in blended cement, that of grinding energy. For a comminution process to grind a material to increasingly smaller sizes, commensurate amounts of energy are required. This is not a linear relationship, however, but is governed by the defect density of the material to be processed. In the case of glasses, comminution to a few microns is relatively straightforward, but further reduction is increasingly energy intensive. To summarize, there seems to be no universal rule to be drawn from the literature relating to the absolute particle size of the glass or to the hydration kinetics of such glass-blended cement, and no convincing minimum particle size has been identified which will universally avoid ASR reactions. Nonetheless, the literature suggests particles below 40 µm will react completely in the cement environment without detriment.

1.3. Hydration of Blended Recycled Glass Powder in CEM II/A-LL Slag Cements

It is proposed that a new blended cement formulation may be produced by incorporating recycled glass powder, blended with ground granulated blast furnace slag to produce a novel industrial pozzolan. This in turn would be blended with ‘common cements’—in this case, Portland-limestone cement to produce a modified CEM II/A-LL type binder. CEM II/A-LL was chosen as the most tightly constrained limestone cement composition in order to limit chemical variations in this study and in ongoing experiments.

In terms of cement classification, this is a quaternary cement. The objective is to conserve cementitious slag by its partial replacement with recycled glass powder. This study examines the likely consequence of doing so by the thermodynamic modelling of both phase evolution and solution chemistry. This mixed RCGP–GGBS powder (known here as the ‘Development Blend’) comprises 85% GGBS and 15% recycled glass powder, RCGP.

Thermodynamic modelling was used to compare the hydration chemistry of four formulations:

(1)

A control sample comprising 50% GGBS and 50% CEM II/A-LL (all quantities by mass) but without RCGP.

(2)

A cement paste containing 60% CEM II/A-LL combined with 40% of the RCGP–GGBS development blend.

(3)

A cement paste containing 40% CEM II/A-LL combined with 60% of the RCGP–GGBS development blend.

(4)

‘Standard’ Portland-Limestone cement (CEM II/A-LL) containing 10.48% CaCO₃ (measured and supplied by Ecocem France, Dunkirk works).

The control mix is typical of that used in one of the national markets served by the cement manufacturer Ecocem.

2. Materials and Methods

2.1. Hydration Modelling Using HYDCEM

HYDCEM is an innovative, composite modelling code which allows users to simulate hydration in a range of cement types and pozzolans, which may or may not interact with chemically aggressive solutions. Developed originally by Holmes et al. in 2019 [22], it has been developed extensively and continues to evolve [23,24,25,26]. The user is required to write input parameters into a graphical user interface (Figure 1), which are used in subsequent PHREEQC simulations [27,28]. PHREEQC is the chemical simulator underlying HYDCEM and is a widely used and well-established geochemical modelling code. PHREEQC calls a database of thermodynamic constants [29] developed specifically for the simulation of cementitious systems.

As PHREEQC is a C-based computer language, it was decided in 2020 to re-write HYDCEM in C# [26] so coupling and sequential calculations may be more computationally effective [23]. C# is a multi-purpose object-orientated programming (OOP) language developed by Microsoft as part of the .NET Framework. Moreover, C# is a high-level language that is widely used for desktop and web applications and is popular for game development. Using OOP simplifies programming as it optimizes the code allowing users to continually develop it, so methods can be used in any instance of the object of that class and can be re-used multiple times. More details are published in references [23,26,30], but the consequence of recoding the algorithm is a huge improvement in numerical efficiency and hence run-times. Since its inception, HYDCEM has been updated and now includes the option to model RCGP blended cements by specifying the proportion of cement, glass powder and GGBS, as shown in Figure 1.

The chemical simulator (PHREEQC) is a public domain code (available for the USGS website at: https://www.usgs.gov/software/phreeqc-version-3, accessed 1 June 2025) and has been in continuous development since 1980. HYDCEM is a code used with PHREEQC and is available from the authors. It will shortly be available from a dedicated website, until when, requests to use it are welcomed by its authors.

2.2. Solid Solutions

Practically, HYDCEM invokes functions in PHREEQC, and two of these are important to consider when comparing predictions in blended cements. The first of these are solid solutions where several compounds in the cement hydrates form non-ideal solid solutions with each other. The hydrogarnets, for example, show partial solid solution between the hydrogrossular (C₃AH₆) typical of Portland cements and a siliceous hydrogarnet (C₃AS_0.24H_5.52) seen as a single phase. More siliceous hydrogarnets, as found in blended slag and fly ash cements, span a miscibility gap, where two hydrogarnets co-exist between the range C₃AS_0.5H_5.0 and C₃A_0.76H_4.48. As the system becomes more siliceous than this, a single hydrogarnet phase with variable composition becomes stable. Two phases with simple stoichiometries (C₃ASH₄ and C₃AS₂H₂) are known in this range. The solid solution series end member (grossular garnet, C₃AS₃) cannot be formed at atmospheric pressure but is seen as a natural, metamorphic mineral. In addition, a solid solution series exists between andradite garnet (C₃FS₃) and hydro-andradite (C₃FH₆), but there is little evidence of extensive solid solutions with aluminium bearing hydrogarnets at atmospheric pressure and temperature.

The major solid solution in cement hydrates is, of course, the gel phase. At high Ca:Si ratios, (>1.5), C-S-H is a relatively simple stoichiometry, forming a solid which dissolves incongruently on leaching, such that the calcium-rich structural units are more soluble in water than are the silicate units. As the Ca:Si ratio of the gel falls (such as in blended Portland-slag or Portland-fly ash cements), the gel phase incorporates increasing amounts of aluminium, and the term ‘C-A-S-H’ is widely used to describe these compounds. Conceptually, the C-S-H ‘line’ on the triangular Ca-Al-Si hydrate phase diagram becomes the C-A-S-H ‘triangle’ broadening as the Ca:Si ratio decreases.

2.3. Kinetics

The second feature of PHREEQC to be used in HYDCEM is that of kinetically constrained reactions. If the reactants (clinker minerals) are brought into equilibrium with water, the simulation would be of complete hydration—a state not seen in real cements. PHREEQC has the option to prevent complete reaction by imposing kinetic constraints on the reaction, so that only a certain fraction of the clinker minerals hydrate at a particular time step. HYDCEM uses this feature, invoking the Parrot and Killoh kinetic model [31]. However, when kinetic reactions of several minerals and their interactions with solid solutions are performed together, run times can be impractically long. To overcome this, the C-A-S-H phase chemistry is not modelled explicitly as a non-ideal solid solution but is ‘discretised’ as a sequence of single phases of fixed stoichiometry, read as additions to the thermodynamic database. This uses features of C# programming, which markedly speed up the simulations, allowing complex hydration predictions to be undertaken in a few tens of seconds.

2.4. Clinker and Binder Dissolution

In the model described here, the dissolution kinetics of the cement phase is simulated using the Parrot and Killoh methodology [31]. As mentioned above, while this method was developed for OPC/plain cement, it is assumed that it is still applicable for the higher limestone content in the CEM II/A-LL cement as part of the blended binder here.

The degree of reaction (DOR) of the GGBS and RCGP is slower than cement, so an empirical non-linear regression equation, employing four-parameter logistic (4PL), is used in HYDCEM, as suggested by Kulik et al. [32], using Equation (2). Based on [32], b is taken as 0.7, c is 85.1, d = 60 and g = 1.

$$DOR\left(\%\right)=d+{\displaystyle \frac{\left(a-d\right)}{1+{\left(\frac{t}{c}\right)}^b}}$$

(2)

The DOR and change in volume for a 51 g GGBS and 9 g RCGP blend is shown in Figure 2.

2.5. Hydration and Thermodynamic Considerations

Holmes et al. [33] described how oxide compositions are converted into molar quantities of the clinker constituents, proportioning the molar amounts of K₂O, Na₂O, MgO and SO₃ in C₃S, C₂S, C₃A, and C₄AF. The remaining clinker phases (lime, calcite, gypsum, periclase, arcanite and thenardite) are provided in the first time-step/solution and allowed to reach equilibrium with the pore solution to contribute to the formation of hydrate solids. Consideration is also paid to the initial oversaturation of C-A-S-H, gypsum, portlandite, syngenite and ettringite in the first 12hrs of hydration. An oversaturation factor (=0.15n) is added to the input for these phases to obtain improved predictions of the pore solution data, where n is the number of charged species involved in the hydration reaction (n = 2, 3, 4.95, 5, 3 and 15 for gypsum, portlandite, C-A-S-H, syngenite, brucite and ettringite, respectively [34]). Finally, to accurately predict the changing pH and pore solution chemistries, alkali (K and Na) distribution between the pore solution and the C-A-S-H previously developed by Shaji et al. [35] and shown in Figure 3 were added.

The thermodynamic input for the slag and glass powder were provided through phase equations derived from their oxide content. This method was developed by Shaji et al. [36] to describe the glass phase of fly-ash in addition to the other crystalline phases. This provides a comprehensive way to include the glass phase, which will provide accurate predictions of phase assemblages, pore solution chemistry and pH. For the slag and recycled glass powder used here, it was assumed they constitute 100% amorphous glass with no crystalline phases with a 50:50 split between active and inactive glass. The reaction stoichiometries shown are based on a 1 kg = 1 mole stoichiometry, which makes it convenient to add into the required molar quantities of the PHREEQC input file. The thermodynamic phase equations for the slag and recycled glass powder used in the analysis, following the methodology described in [36], are provided in

Table 1. Slag and RCGP glass phases equations for the reactive and unreactive phase.

Mix No.	Active/Inert Glass Phase
2 (50% slag)	(SiO2)6.0653(Al2O3)0.9651(Fe2O3)0.0494(CaO)7.8955(MgO)1.9639(SO3)0.0102(K2O)0.0478(Na2O)0.033 = + 6.0653 SiO2 + 1.9302 AlO2− + 0.0988 FeO2− + 7.8955 Ca+2 + 1.9639 Mg+2 + 0.0102 SO4−2 + 0.0956 K+ + 0.066 Na+ − 17.831 H+ + 8.9155 H2O; log_k = ±999, Vm = 401.6064 g/cm3
3 (34% slag)	(SiO2)6.0702(Al2O3)0.9659(Fe2O3)0.0494(CaO)7.902(MgO)1.9655(K2O)0.0479(Na2O)0.0331 = + 6.0702 SiO2 + 1.9318 AlO2− + 0.0988 FeO2− + 7.902 Ca+2 + 1.9655 Mg+2 + 0.0958 K+ + 0.0662 Na+ − 17.8664 H+ + 8.9332 H2O; log_k = ±999, Vm = 401.6064 g/cm3
3 (6% RCGP)	(SiO2)12.2273(Al2O3)0.1307(Fe2O3)0.0227(CaO)2.8507(MgO)0.4286(K2O)0.0566(Na2O)1.0634 = + 12.2273 SiO2 + 0.2614 AlO2− + 0.0454 FeO2− + 2.8507 Ca+2 + 0.4286 Mg+2 + 0.1132 K+ + 2.1268 Na+ − 8.4918 H+ + 4.2459 H2O; log_k = ±999, Vm = 401.6064 g/cm3
3 (51% slag)	(SiO2)6.0702(Al2O3)0.9659(Fe2O3)0.0494(CaO)7.902(MgO)1.9655(K2O)0.0479(Na2O)0.0331 = + 6.0702 SiO2 + 1.9318 AlO2− + 0.0988 FeO2− + 7.902 Ca+2 + 1.9655 Mg+2 + 0.0958 K+ + 0.0662 Na+ - 17.8664 H+ + 8.9332 H2O; log_k = ±999, Vm = 401.6064 g/cm3
3 (9% RCGP)	(SiO2)12.2273(Al2O3)0.1307(Fe2O3)0.0227(CaO)2.8507(MgO)0.4286(K2O)0.0566(Na2O)1.0634 = + 12.2273 SiO2 + 0.2614 AlO2− + 0.0454 FeO2− + 2.8507 Ca+2 + 0.4286 Mg+2 + 0.1132 K+ + 2.1268 Na+ − 8.4918 H+ + 4.2459 H2O; log_k = ±999, Vm = 401.6064 g/cm3

.

The hydration of four paste formulations was performed by HYDCEM.

Table 2. Relative oxide composition of the materials and of the formulations simulated (shaded).

Composition of Blending Components (Mass%)				Composition of Paste Formulations
Oxide	RCGP	GGBS	CEM II A/L	Control	40% Blend	60% Blend
SiO2	68.89	35.59	17.98	26.79	27.02	29.28
CaO	14.99	43.24	62.45	52.86	53.07	50.73
MgO	1.62	7.73	2.57	5.15	4.267	4.69
Al2O3	1.25	9.61	4.14	6.88	5.83	6.25
Fe2O3	0.34	0.77	2.85	1.81	1.99	1.78
TiO2	0.08	0.66	0.33	0.50	0.43	0.45
Mn3O4	0.04	0.15	0.24	0.20	0.20	0.19
Na2O	6.18	0.2	0.24	0.22	0.58	0.67
K2O	0.5	0.44	0.53	0.49	0.50	0.49
CaCO3	0	0	10.48	5.24	6.29	5.24
Total	93.89	98.39	101.81	100.14	100.18	99.77

shows the relative oxide composition of the blending components and of the four blends used in the simulations (

Table 3. Mix details.

Mix No.	Proportion of Each Powder (%)
	CEM II/A-LL	GGBS	RCGP
1	100
2 (control)	50	50
3	60	34	6
4	40	51	9

). A water-to-binder (w/b) ratio of 0.5 was used throughout, and the simulations were made on a thermodynamically closed system (no mass loss or gain).

3. Results

HYDCEM Predictions

Figure 4 show the developing solid phase chemistry of the 100% CEM II/A-LL pastes hydrating over 1000 days at 20 °C. Similar predictions of the solid phase evolution of the three blended cements are shown below, and not only are they remarkably similar in phase quantities but the rate at which the phases dissolve and precipitate are also comparable. Mix 2, or the ‘control’ sample (50% slag + 50% cement, Figure 5), has been used as a reference blend on numerous studies by Ecocem in the past. In this case, the cement component is the CEM II/A-LL formulation used in this study.

Comparing the final (1000 day) phase quantities with each other shows the total hydrate abundance for the two experimental glass-slag pastes (60% CEM II/A-LL + 40% GGBS/RCGP (Mix 3, Figure 6) and 40% CEM II/A-LL + 60% GGBS/RCGP (Mix 4, Figure 7) to be slightly greater and slightly less than that of the control paste (respectively), as shown in Figure 7 and Figure 8. Regarding the solution compositions predicted by HYDCEM at 1000 days, again little difference is seen between the control (Mix 2) and the simulated slag-glass blends with CEM II/A-LL. In each case, the solution chemistry is dominated by sodium and potassium, calcium, silicate, carbonate and sulphate as the next most abundant dissolved ions. For each ion in solution, the variation in predicted concentration between each blended cement is not significant.

Replotting this data into stacked columns (Figure 8) and normalised to 100 mol% (Figure 9) makes comparison of the results much easier and shows there is little difference between the blended cements. This shows that hydrate compositions in mixes 3 and 4 span that of the control sample (1:1 clinker:slag, without glass) and suggests that GGBS and RCGP will be consumed and react with portlandite from the clinker to form C(A)SH gel.

4. Discussion

This study is entirely theoretical, providing a rapid comparison between blended slag cements, with and without container glass powder. It was undertaken to examine the likely consequence of RCGP inclusion in blended cements and to guide ongoing experimental studies.

The use of recycled glass powder as a partial replacement for cementitious slag is predicted to result in a hydrate assemblage comprising the same phases as for a conventional slag cement, within the limits reported in this study. This is unsurprising, given the similarity of the bulk oxide composition of the blended materials. Differences are seen between the slag-free, limestone-bearing cement system and the blended slag cement: The gel phase for 100% CEM II/A-LL and the pure cement system shows a high Ca:Si ratio of 1.65 and a lower ratio (1.33) for the aluminous C-A-S-H gels predicted for each of the blended slag cements. This is an expected result. Similarly, the predicted hydrogarnet in the pure cement assemblage contains some iron, whereas, in the blended slag–glass–cement system, the hydrogarnet is siliceous. These small variations in hydrate composition indicate that blended RCGPG–GBFS–limestone clinkers are all very similar within the compositional limits considered here. Overall, this suggests that RCGP may be used with confidence in slag–limestone cement formulations.

It should be highlighted that the findings above are based on assumptions around cement, slag and recycled glass dissolution predictions which need to be validated by experimental data, which will form part of future studies.

5. Conclusions

Decarbonisation and circularity are among the top priorities in modern civil engineering. To achieve these sustainability goals, the search for and use of alternative cement-based materials and understanding their reaction mechanisms during cement hydration is critical.

The simulations here indicate that the partial replacement of cementitious slag (GGBS) by recycled container glass powder (RCGP), up to at least the 15% level investigated, is technically viable. The RCGP–GGBS development blend will hydrate similarly to other slag cements to produce an essentially similar assemblage of hydrate phases.

These results set the case for inclusion of the blended cement product into BS 8500 2:2015+A2:2019 Concrete [Clause 4] [37] as a Type II addition and additionally provide valuable information towards compliance with recently published BSI Flex 350 [38].