Data-Based Assessment of Cement Replacement Strategies for Eco-Friendly UHPC ^†

Abstract

Ultra-high-performance concrete (UHPC) has a high mechanical performance, but its large cement content is associated with a high environmental impact. This study evaluates cement replacement strategies for eco-friendly UHPC using a dataset of 610 mixtures collected from the literature. The mixtures were analyzed using typical CO₂ emission factors, harmonized compressive strength, and grouped descriptors including the supplementary cementitious materials (SCM) replacement ratio, binder pathway, fiber content, and curing regime. The results show that increasing SCM replacement strongly reduces the binder-related CO₂ and generally lowers the total CO₂ intensity, while a high compressive strength can still be achieved at higher replacement ratios. Slag-rich and blended SCM systems showed the most promising balance between reducing CO₂ and maintaining strength.

1. Introduction

Ultra-high-performance concrete (UHPC) is increasingly used in structural and infrastructure applications due to its high mechanical performance and durability [1]. However, these advantages are commonly achieved with a high cement content, and, in some cases, energy-demanding curing regimes. As a result, UHPC can have a high embodied CO₂ that compromises its sustainability. Therefore, reducing the cement content of UHPC is an important step toward improving its environmental performance and supporting its adoption as an efficient solution for sustainable concrete structures [2].

A common approach to reduce the CO₂ burden of UHPC is the partial replacement of the cement with supplementary cementitious materials such as silica fume, fly ash, slags, metakaolin, rice husk ash, limestone powder, quartz powder, and other reactive mineral additions [3,4,5]. Previous studies have shown that these materials can reduce cement demand, improve particle packing, and maintain or enhance strength and durability. However, the reported eco-UHPC mixtures vary widely as the composition differs in binder content, fiber dosage, and replacement material, as well as the curing regimes [3,4,6,7,8,9]. Hence, the comparison of cement replacement strategies remains difficult because published UHPC mixtures are designed with different objectives. Some mixtures aim for maximum strength, others for cost reduction, local material use, fiber efficiency, or lower environmental impact. Furthermore, a comparison in terms of CO₂ emissions is complicated by regional differences, as local production processes, energy sources, and transportation distances significantly influence the embodied carbon of the constituent materials.

A broader data-based assessment is needed to further understand and quantify the effectiveness of different cement replacement strategies and separate the binder direct effect from other mixture-level influences. In this study, a dataset of 610 UHPC mixtures collected from the literature was analyzed. The Supplementary Cementitious Material (SCM) replacement ratio was used to describe the level of cement substitution, while binder-related CO₂, total embodied CO₂, compressive strength, and total CO₂ intensity were used to correlate mechanical and environmental performance. The analysis also considers the binder pathway, fiber content, and curing regimes to provide further insights and interpret the variability in the collected dataset. The aim of this work is to identify which cement replacement strategies are most effective for eco-UHPC and to evaluate the extend of cement replacement that can be achieved without compromising mechanical performance. The results are intended to support the more practical selection of low-carbon UHPC binder systems.

2. Methodology

This study investigates UHPC mixture design strategies aimed at reducing the CO₂ burden of cement through its partial replacement by SCM or other reactive binder substitutes. The materials considered are those that contribute to the binder system through hydraulic or pozzolanic reactivity such as silica fume, fly ash, ground granulated blast-furnace slag (GGBFS), metakaolin, rice husk ash, other slags, and other reactive mineral additions. Limestone powder and quartz powder are also considered in this dataset only when the original study treated them as binder-level replacements rather than inert fillers.

In this study, a total of 610 mixture records were collected and analyzed. The source studies and number of mixtures are listed in Appendix A. In order to enable consistent environmental assessment across all studies, a constituent-level CO₂ computation was applied. For each mix constituent, an emission factor was assigned based on typical values reported in the literature, as presented in Table 1. These values reflect cradle-to-gate (A1 to A3) assessment. In addition, curing was incorporated as a separate contributing factor. Reported compressive strengths were harmonized to improve comparability across studies using different specimen geometries and sizes. Standard cylinders were treated as the reference geometry, and cube and low height-to-width-ratio (<2) specimens were adjusted using a conservative factor of 0.95 in accordance with SIA 2052 [10] and ÖBV-Richtlinie “UHPC” [11], whereas values from prisms were retained with a factor of 1.00.

Table 1. CO₂ emission factors for eco-UHPC constituents and curing regimes.

Source	Item	EF (kg CO2/kg)	Reference
UHPC mix	Cement	0.83	Long et al. [12], Chiaia et al. [13]
	Silica fume	0.014	Sharma et al. [14], Kathirvel et al. [15]
	Flay ash	0.005	Habert et al. [16]
	Metakaolin	0.332	Sun et al. [17]
	Rice husk	0.157	Hu et al. [18]
	Glass powder	0.023	Kathirvel et al. [15]
	Limestone	0.026	Harbert et al. [19]
	Quartz powder	0.048	Heidari et al. [20]
	Slag	0.019	Long et al. [12]
	Superplastizer	0.72	Long et al. [12], Chiaia et al. [13]
	Sand	0.024	Harbert et al. [19]
	Coarse aggregate	0.043	Harbert et al. [19]
Fibers	Steel fibers	2.68	Harbert et al. [19]
	Polyethylene fibers	1.6	Gomez [21]
	Standard	0.0	Shi et al. [5]
Curing regime	Steam curing	2.49	Turner and Collins [22]
	Autoclave curing	11.73	Shi et al. [5]

The cement replacement level was quantified using the SCM replacement ratio:

$$R_{SCM}={\displaystyle \frac{m_{SCM}}{m_{\mathrm{cement}}+m_{SCM}}}\times 100$$

(1)

where m_SCM is the total mass of reactive supplementary cementitious materials or binder replacements, and m_cement is the mass of the Portland cement. This ratio was used as the independent variable for the statistical analysis.

Two environmental indicators were used. First, the binder-related CO₂ was calculated to isolate the direct effect of cement replacement:

$$\mathrm{C}{\mathrm{O}}_{2,binder}={\displaystyle \sum _{i=1}^n} m_i\times E{F}_i$$

(2)

where m_i and EF_i are the mass and emission factor of each binder constituent. This CO_2,binder indicator includes cement and reactive binder replacements, but excludes fibers, aggregates, inert fillers, and admixtures. Second, the calculated total embodied CO₂ is used to represent the total environmental impact of the mixture:

$$\mathrm{C}{\mathrm{O}}_{2,\mathit{total}}={\displaystyle \sum _{j=1}^n} m_j\times E{F}_j$$

(3)

where m_j and EF_j refer to all mixture constituents including binder materials, aggregates, fillers, fibers, admixtures, and curing-related contributions, where available. To evaluate the combined environmental and mechanical performance, the total CO₂ intensity is defined as follows:

$$\mathit{Total} {\mathrm{CO}}_2 \mathit{intensity}={\displaystyle \frac{\mathrm{C}{\mathrm{O}}_{2,\mathit{total}}}{f_{c,\mathit{eq}}}}$$

(4)

where f_c,eq is the equivalent converted compressive strength. This indicator expresses the calculated total embodied CO₂ required to achieve one MPa of compressive strength.

3. Results and Discussions

3.1. Dataset Characterization and Classification

A statistical analysis of the collected dataset was first performed to account for the observed variability in mixture compositions and SCM usage. Supporting descriptors used to interpret the data were the binder pathway, composition, and replacement range. Binder pathways were classified according to the dominant binder-substitution pathway. The main classes were no/low SCM, silica-fume-rich systems, hybrid or blended SCM systems, fly-ash-based systems, slag-rich systems, limestone or mineral-powder systems, and metakaolin-based systems. These classes were used only as descriptors. They were not treated as independent variables because each pathway includes mixtures with different binder contents, fiber volumes, curing regimes, and mix-design objectives. The no/low SCM class mostly includes reference mixtures reported in the studies.

Figure 1a shows the representation of each binder pathway in the dataset. Silica-fume-rich systems are the most represented group, followed by hybrid/blended SCM systems, fly-ash systems, and slag-rich systems. Limestone/mineral-powder systems, no/low SCM mixtures, and metakaolin systems are less represented. This distribution is important because it shows that the statistical evidence is not balanced across binder pathways. Trends observed for silica fume, hybrid SCMs, fly ash, and slag-rich systems are supported by a larger number of mixtures, while conclusions for metakaolin and no/low SCM mixtures should be interpreted more cautiously.

Figure 1. Binder pathway representation in the dataset: (a) number of mixtures in each main binder pathway; and (b) mean binder composition in each pathway.

The mean binder composition of each pathway is shown in Figure 1b. No/low SCM mixtures are dominated by cement. Silica-fume-rich systems retain a high cement fraction, with silica fume acting mainly as a reactive fine addition rather than a high-volume cement replacement. Fly-ash and hybrid/blended systems show larger reductions in the cement share and more distributed binder compositions. Slag-rich systems show the largest reduction in cement fraction and the highest contribution from slag. Limestone/mineral-powder systems show a distinct composition, with a large fraction of limestone or quartz powder relative to other SCM types. These differences confirm that the same SCM replacement ratio can represent different binder chemistries and different carbon-reduction mechanisms.

3.2. Validation of CO₂ Evalution

The embodied CO₂ values reported in the original studies with the values calculated using the constituent-level method adopted in this work are compared in Figure 2. The reported dataset is substantially smaller as many studies do not provide embodied CO₂ values. In contrast, the calculated dataset covers the full compiled dataset, allowing a consistent comparison across mixtures. Despite the difference in data coverage, the reported and calculated datasets show similar trends. In both cases, embodied CO₂ decreases as the SCM replacement ratio increases. The set was divided into three representative fiber percentages, as the CO₂ emissions from the fibers play a major role in total mix emissions. The relative position of the fiber-content groups is also consistent such that mixtures with a fiber content ≥ 2.5% generally have the highest embodied CO₂, followed by mixtures with 2.0% fiber, while mixtures with a fiber content ≤ 1.5% occupy the lower range. The calculated values also reproduce the approximate value of the reported values (Figure 3c). Both datasets show embodied CO₂ values mainly within the range of approximately 200–1700 kg CO₂/m³. This agreement indicates that the adopted emission factors provide a reasonable basis for estimating embodied CO₂ where reported values are unavailable. Therefore, reported CO₂ values were used only as a validation subset, while calculated CO₂ values were used as the main dataset for regression analysis.

Figure 2. Comparison between reported and calculated embodied CO₂ as a function of SCM replacement ratio: (a) reported embodied CO₂; (b) calculated embodied CO₂; and (c) reported vs computed CO₂.

Figure 3. Relationships between SCM replacement ratio and: (a) binder-related CO₂; (b) calculated total embodied CO₂; (c) compressive strength; and (d) total CO₂ intensity.

3.3. Impact of SCM Replacement Ratio

The relationships between SCM replacement ratio and four response variables are presented in Figure 3. The results are separate for three fiber-content classes: ≤1.5%, 2.0%, and ≥2.5% The four response variables include binder-related CO₂, calculated total embodied CO₂, compressive strength, and total CO₂ intensity. The binder-related CO₂ decreases strongly with an increasing SCM replacement ratio across all fiber-content classes, as presented in Figure 3a. This confirms that cement substitution directly reduces the carbon burden of the binder system, with data reaching, in some cases, even 90% of binder-CO₂ reduction. The regression models show a relatively strong correlation factor (R²), indicating that the SCM replacement ratio is a strong predictor of binder-related CO₂ reduction. The total embodied CO₂ also decreases with an increasing SCM replacement ratio, as shown in Figure 3b, but the relationship is less direct than for binder-related CO₂. This is expected because total embodied CO₂ includes fibers and other mixture constituents in addition to the binder. The ratio between the SCM% replacement and total-CO₂ reduction is now 6.7–8.4, still very close to the ratio observed when taking into consideration only the binder replacement. The effect of the fiber content is clearly visible in mixtures with ≥2.5% fiber, showing a higher total embodied CO₂ than the lower-fiber groups. This confirms the need to separate binder-related CO₂ from total embodied CO₂ when evaluating cement replacement strategies.

The compressive strength shows only a weak dependence on the SCM replacement ratio, as presented in Figure 3c. The regression slopes are small and the R² values are low across all fiber-content classes. This indicates that SCM replacement does not control the strength variation in the compiled dataset. Strength is likely controlled by several interacting parameters, including binder composition, particle packing, curing regime, and fiber content. Importantly, high compressive strengths are still achieved at moderate and high SCM replacement ratios, showing that cement reduction can be compatible with a high mechanical performance.

The total CO₂ intensity decreases with an increasing SCM replacement ratio in all fiber-content classes, as shown in Figure 3d. This indicates that SCM replacement improves environmental efficiency per unit compressive strength. The decrease in CO₂ intensity is more pronounced than the change in compressive strength, meaning that the environmental benefit is mainly driven by CO₂ reduction without a compromise of strength loss. The high-fiber group has higher total embodied CO₂ even though most of the mixtures in this group were reported with a higher compressive strength. Overall, the results show that SCM replacement is an effective cement-reduction strategy for UHPC. It consistently lowers binder-related CO₂ and calculated total embodied CO₂ while maintaining compressive strength within the range required for UHPC. The reduction in total CO₂ intensity further confirms that cement replacement improves eco-mechanical efficiency.

3.4. Mechanical and Environmental Performance by Binder Pathway

The primary regression analysis presented above shows the overall relationship between the SCM replacement ratio, CO₂ reduction, compressive strength, and total CO₂ intensity. However, the scatter in the data also indicates that effective low-carbon UHPC design also depends on the main binder pathway.

Figure 4a summarizes the SCM replacement range covered by each binder pathway. No/low SCM mixtures are concentrated near zero replacement. Silica-fume-rich systems are mainly located at low to moderate replacement levels, with a median near 20%. Fly-ash and hybrid/blended systems cover wider replacement ranges, extending into moderate and high replacement levels. Slag-rich systems show the highest replacement levels, with a median close to 70% and an upper range approaching 95%. Limestone/mineral-powder systems occupy a narrower moderate replacement range, while metakaolin is represented by very few mixtures.

Figure 4. Eco-performance efficiency by binder pathway: (a) SCM replacement range by pathway; and (b) binder-related CO₂ emissions by pathway.

Figure 4b shows binder-related CO₂ as a function of the SCM replacement ratio for each binder pathway. The same general pattern appears across the descriptor classes where the binder-related CO₂ decreases as the SCM replacement increases. This confirms that the main trend observed in the main matrix is not limited to one binder type. However, the pathway-level distribution provides additional interpretation. Silica-fume-rich mixtures are mostly located at lower replacement ratios and, therefore, often retain relatively high binder-related CO₂. Fly-ash and hybrid systems occupy the intermediate replacement and CO₂ ranges. Slag-rich systems extend to the highest replacement ratios and reach the lowest binder-related CO₂ values. Limestone/mineral-powder systems are clustered in a narrower region and generally show moderate binder-related CO₂. These results show that the strongest binder-level carbon reductions are associated with high-volume replacement pathways, particularly slag-rich and some blended systems.

3.5. Impact of Curing Regime

The curing-regime descriptor provides an additional explanation for the scatter observed in the strength and CO₂-intensity results. Figure 5 shows that thermally accelerated curing regimes, especially autoclave or pressure-based curing, are associated with a higher mean compressive strength than standard curing. However, they can also increase the calculated total CO₂ when curing-related energy is included. Steam, hot-water, and heat/oven curing occupy intermediate positions whereas the standard curing shows the lowest compressive strength and environmental impact. This indicates that curing can improve mechanical performance but may reduce part of the environmental benefit as the additional energy demand is high. Therefore, the curing regime should be treated as a key modifier when interpreting cement replacement strategies.

Figure 5. Curing regime effect on mechanical and environmental outcomes.

4. Conclusions

In this study, a data-based assessment of cement replacement strategies for eco-UHPC is presented. The following conclusions can be drawn:

• SCM replacement is an effective way to reduce the carbon impact of UHPC. Across the collected dataset, higher SCM replacement ratios led to a clear reduction in binder-related CO₂ and a general reduction in total embodied CO₂. The effect was strongest at the binder level, confirming that cement reduction directly lowers the main source of emissions in UHPC mixtures.
• The SCM replacement ratio does not negatively affect strength. High compressive strengths were still achieved at moderate and high replacement levels, showing that cement reduction can be compatible with UHPC performance. The scatter in the data indicates that the binder composition, packing, fiber dosage, and curing regime must also be considered when designing eco-UHPC.
• Slag-rich and blended SCM pathways showed the best overall potential. These systems generally provided a lower CO₂ intensity while maintaining high strength levels. The results also show that a high fiber content and energy-intensive curing can increase the total CO₂, so eco-UHPC design should evaluate the full mixture and curing regime in addition to the cement replacement ratio.

Other performance indicators critical to UHPC applications, such as durability and tensile behavior, were not consistently reported and were therefore excluded from the analysis. Consequently, the findings are limited to the relationship between SCM replacement, compressive strength, and embodied CO₂. In addition, the environmental assessment considered only production-stage embodied CO₂. While this enables a consistent comparison of mix designs, overall sustainability also depends on factors such as durability, maintenance, service life, and material savings. Future studies should therefore investigate the influence of SCM replacement on long-term performance and incorporate life-cycle and service-life assessments to provide a more comprehensive evaluation of eco-UHPC.