This study employs the experimental data performed on a lab-scale mineral carbonation of steel slags at the University of Leuven, Belgium. The study subsequently projected the labscale mass and energy consumption as well as the yields to an industrial-scale pilot plant (TRL 6) scenario based on the upscaling frameworks with a power law learning curve which incorporates the process efficiency and the economy-of-scale [27], [18], [28]. The projected industrial scenarios have a capacity equivalent to the industrial wet mineral carbonation installation in the study of Lee et al. [18]. The study assumes a mineral carbonation capacity of 61.8 tons per day, producing 22557 tons per year of carbonated steel slags.

There are 5 mineral carbonation cases in total which act as the baseline scenarios. These 5 mineral carbonation scenarios can be sub-categorized into 3 mineral carbonation of basic oxygen furnace (BOF) steel slags scenarios and 2 mineral carbonation of argon oxygen decarburization (AOD) steel slags scenarios with varying degrees of carbonation. The main difference between each scenario is, therefore, the carbonation degree which reflects how long the mineral carbonation took place for each case. BOF T1 case refers to the mineral carbonation of BOF steel slags at 2 bar for 3 hours. BOF T2 case refers to the mineral carbonation of BOF steel slags at 4 bar for 2 hours. The BOF T3 case refers to the mineral carbonation of BOF steel slags at 4 bars for 91 hours. The AOD T1 case refers to the mineral carbonation of AOD steel slags at 4 bar for 3 hours. Finally, the AOD T2 case refers to the mineral carbonation of AOD steel slags at 4 bar for 33 hours. The different absolute residence times are due to the differences in the composition of the slags which resulted in different carbonation duration needed to reach the maximal carbonation levels which are identified and reported in previous studies of the authors [15], [29]. Within this study, the carbonated slag products are assumed to be sold as either a replacement for basalt rock powder or the liming agent. In the baseline scenarios, it was assumed that 50% of the produced carbonated slags are sold as basalt rock powder replacements and the remaining 50% are sold as liming agent replacements.

This study also expands the baseline scenarios into multiple scenarios to account for the different possibilities in the carbonation process in terms of the technological/environmental aspect and the economic aspect. The first group of expanded scenarios includes the variations in the proportion of what the carbonated slag products can be sold as. The motivation behind adding this scenario analysis is that the use of basalt rock powder as the source of macronutrients in agriculture at present is limited to mostly organic farmers. This means that it would be difficult to achieve a high sales volume for the basalt rock powder. Therefore, the study also tries to project what should be the lowest proportion of the carbonated slag products to be sold as basalt rock powder possible in order to still achieve economic viability. This is defined as having an internal rate of return (IRR) of at least 7%. This leads to one exception of case BOF T3 where the baseline NPV has an IRR lower than 7%. Hence, instead of excluding BOF T3, the minimal amount of carbonated slags percentage to be sold as basalt rock powder substitute was determined to at least satisfy a 7% IRR. This calculation resulted in a percentage of 65% carbonated slags sold as basalt rock powder substitute and 35% as liming agent substitute.

Other expansions of the baseline scenarios include (1) the inclusion of the drying process to dry the carbonated slag products at the end, (2) the change of the electricity grid mix of Belgium to an average EU-27 + Switzerland mix, and finally, (3) the inclusion of both the drying process and the change of the electricity grid mix of Belgium to the EU-27 + Switzerland mix. The motivation for these scenarios is that at the moment of publication, it is still unclear whether mineral carbonation of steel slags would require a drying process or not. On the one hand, it is possible to use the carbonated slags directly from the reactor onto the fields for agricultural applications. On the other hand, wet carbonated slag products are heavier than dried products, which may prove to be problematic when considering the transportation of these products. The study of Lefebvre et al. (2018) [16] have identified that the transportation of carbonated rocks is the hotspot of the climate change impacts for the full life cycle of carbonation of rocks with more than 50% of the total climate change impacts depending on the transportation radii. Therefore, this study takes into account both possibilities with (expanded scenarios) and without the drying process (baseline). The motivation for having scenarios with a different electricity mix is twofold: firstly, to explore the possible impacts of what would happen if the mineral carbonation process is not used in Belgium, but elsewhere in Europe, and secondly, to investigate what would happen if the electricity mix has less share of renewable energy but cheaper. Finally, the motivation for the inclusion of the third scenario expansion is to complete the assessment with a complete list of scenarios that have both the drying process and a different energy mix.

In total, this study uses 25 scenarios, with 5 baseline scenarios, 5 scenarios with differing proportions of the products sold, 5 scenarios with the drying process with Belgian electricity mix, 5 scenarios with the whole value chain with the EU-27 + Switzerland mix, and 5 scenarios with the drying and the whole value chain with the EU-27 + Switzerland mix.

The upscaled mass and energy balances of the mineral carbonation process are subsequently used for the life cycle assessment (LCA) and the life cycle costing (LCC). The upscaled inventory for the LCA calculation is shown below in Table 3. The LCA models used in this study were created with Activity Browser with the Ecoinvent 3.8 database. The LCA is performed in accordance with the ISO 14044 standard [5] and aligned with the ILCD handbook for life cycle assessment [6]. The environmental impacts of all 25 scenarios are then calculated using the Environmental Footprint 3.0 (EF 3.0) methodology. The functional unit is 1 kg of steel slag mineral carbonation at plant capacity. The system boundary is illustrated below in

Figure 1. The system boundary of both the LCA and LCC are within the red box which includes the slag and water mixing reactor, the mineral carbonation reactor and the membrane separation unit for the CO₂. The system boundary in green are the basis for the scenario analyses which included the drying steps as post-treatments. As the mineral carbonation of steel slags is considered a CCUS technology, the guidelines on the LCA and TEA from the Global CO₂ Initiative [9] are used as a reference. Both the LCA and the LCC have the same system boundary on an annual time scale. Within this study, the global warming potential (GWP), freshwater ecotoxicity (FET), human toxicity, cancer (HTC), and ionizing radiation (IR) are reported. The reasoning for these impact categories is that (1) as it is an investigation on the CCUS, the GWP must be net negative to ensure that the technology actually removes carbon dioxide even with all the mass and energy consumption it requires to operate the CCUS, (2) FET, HTC and IR are chosen because upon reviewing the EF 3.0 normalized results of all environmental impact categories, these impacts have the highest top 3 scores. The normalization of the EF 3.0 approach was performed in order to see the number of impacts an average citizen in the world would generate if the mineral carbonation of steel slags was employed. This means that the use of mineral carbonation has potentially high emissions to these impact categories. The results of the selected environmental impacts of all 25 scenarios are then reported in Table 5. For the LCC, the discounted cash flow model was used for the calculation under the same system boundary as the LCA. The LCC employed a project lifetime of 15 years with a discount rate of 7% per year. The annual revenues and costs included in the LCC calculation are displayed in Table 4. The maintenance costs are included as 10% of the revenues per annum and the drying costs are included according to the electricity price for the heating of 1 kg of steel slags either in Belgium or the average EU-27 + Switzerland price depending on the scenarios. Grinding costs are included in the price of the slags. This project is assumed to be constructed in Flanders, Belgium. Therefore, the costs related to labour, maintenance, energy, materials, possible income from the EU ETS scheme, and avoided landfill costs are based on Belgian data. The infrastructure costs for capital expenditure (CAPEX) are based on the study of Lee et al. [18] as the mineral carbonation plant setup and capacity are similar as indicated. The net present value (NPV) of all 25 cases is then calculated and reported in Table 5. As the scope of this study is on the MCDA, for a detailed life cycle assessment and life cycle costing of the mineral carbonation of steel slags with process hotspots, readers are encouraged to also check other assessed cases in published and accepted works by the authors in Watjanatepin et al. (2023) [15] and Watjanatepin et al. (2024) [29] respectively.

Figure 1:

System Boundary of the LCA and LCC

Prices related to the slags are referred from the US Geological Survey report [30]. Water prices are referred from the EurEau report [31]. Energy prices from the EuroStat datasets [32]. Labour costs referred from the Belgian Federal Government values [33]. The estimation of the maintenance cost is based on the studies of Komnen et al. [34]. The price of the liming agents is referred from CDN Environment report [35]. The possible income from avoided carbon emissions are estimated from the carbon prices published by TradingEconomics [36]. Finally, the avoided landfilling costs are estimated from the report of CEWEP [37].

To perform the MCDA, the AHP approach is used. The AHP approach followed within this study was in line with the multi-attributional decision-making guidelines of the Global CO₂ Initiative for the environmental and economic assessment of CCUS [10]. The objective of this MCDA is “What is the preference order of the mineral carbonation of steel slag scenarios?”. By answering this question, it will be possible to identify the optimal tradeoff between the environmental and economic performance amongst all steel slags mineral carbonation scenarios. And subsequently, a decision could be made on which scenarios to choose and further investigate for process improvements and upscaling. The AHP approach first calculated the pairwise matrix of all the scenario attributes. Firstly, the main criteria are weighted. In this study, the two main criteria are (1) environmental performance and (2) economic performance. Both are given equal weight to designate equal importance for both the environmental and economic impacts at 0.5 and 0.5. The main motivation for assigning equal weighting for both the environmental and economic criteria is twofold. Firstly, in order to justify the definition of CCUS of mineral carbonation, the GWP must be net negative or at least neutral. Therefore, the environmental impact must be well quantified. Secondly, the economic performance is equally important since the CCUS technology is neither impactful nor useful if the costs to operate such a CCUS technology is too high. In that case, alternative options of the technology or even paying for the emission allowances might be cheaper than investing in an expensive technology that is not economically viable. This explains the motivation behind this weighting. This equal weighting was also used in the guidelines document of the Global CO₂ Initiative [10] where at 50% allocation are weighted for both the environmental and economic impacts as a baseline guidance for CCUS technology assessments. Similarly, the MCDA study of Falsafi [19] also used equal weighting (0.5–0.5) for the allocation between environmental and economic impacts. Subsequently, the study then defines the 5 attributes to be optimized which are GWP, FET, HTC, IR, and NPV with a discount rate of 7%. The NPV is a sub-criteria under the economic performance while the rest of the attributes are sub-criteria to the environmental performance. To assign the weight to the environmental performance sub-criteria, the normalized EF 3.0 results of the environmental impacts of mineral carbonation are consulted based on previous studies [15], [29] which found that the most impactful environmental category when deploying mineral carbonation is IR, FET and HTC in descending order. With this information in combination with the definition of a CCUS that must have a net negative GWP and economic impacts, the global priority weight is then assigned as 0.325 for the GWP, 0.05 for the FET, 0.025 for the HTC, 0.1 for the IR, and 0.5 for the NPV based on the AHP pairwise matrix calculation. The pairwise matrix calculation for the determination of the aforementioned weighting values assigned are tabulated in the Supplementary Materials (Table 1). The global priority weight is also given in Figure 2. To calculate the local priority vector, 5 pairwise matrix calculations are performed for each attribute. Each matrix consists of a 25 by 25 matrix to account for all the 25 scenarios. Each attribute has an order of preference quantitatively defined in order to have an objective analysis. The calculation of the pairwise matrices entail the comparison of each of the raw environmental and economic assessment results from the LCA and LCC (in Table 3) respectively to determine the performance of each indicator and each scenario. Subsequently, the local priority vector is then calculated based on the normalization of each attribute’s pairwise matrix for all 25 scenarios. All 5 normalized pairwise matrices for all attributes are displayed in the Supplementary Materials - particularly Supplementary Materials Table 2 to 6. Finally, to calculate the global priority score for each attribute for each scenario, the global priority weights are then multiplied by the local priority vectors. The final preference scores are then the summation of all global priority scores of each attribute (GWP, FET, HTC. IR and NPV) for each of the 25 scenarios. The final preference score results are tabulated in Supplementary Materials - Table 7. Based on the final preference scores, it is then possible to rank the scenarios from highest to lowest scores and indicate the optimal scenario. This ranking is illustrated in Figure 3. In addition to this, the contribution analysis of all 25 scenarios is also performed to interpret which attributes contribute most to each score for each scenario. This contribution analysis is displayed in Figure 4.

Figure 2:

Global priority weight for the MCDA

Figure 3:

Total final preference scores from the MCDA

Figure 4:

Percentage contribution of each attribute score for each carbonation scenario

Table 5 below summarizes the results of the LCA and the LCC for selected parameters, namely the GWP, FET, HTC and IR for the environmental impacts and the NPV for the economic performance for all 25 scenarios of mineral carbonation. The results of these parameters are used as a criterion in the MCDA.

In essence, the environmental impacts for the baseline cases showed that the BOF T3 case is the most net negative case of the mineral carbonation of steel slags. This is because in a wet carbonation process, the longer the carbonation time in the reactor, the higher the amount of carbon dioxide that is sequestered on the surface of the steel slags, resulting in a larger net negative impact. This trend is observed for both the BOF and AOD slags mineral carbonation where the more net negative results are observed the longer the mineral carbonation takes place. When it comes to the other impacts, particularly the FET and HTC, there is a further decrease in both impact categories with the longer duration of the carbonation due to the fact that more carbon dioxide is being absorbed into the slags, thus requiring fewer treatments of the emissions due to the avoided impacts of reusing the emitted gases in a different application. In terms of the IR, the trend is reversed, meaning that the shorter the mineral carbonation duration, the lower the IR results become. This is due to the fact that the shorter the mineral carbonation duration, the lower the energy is consumed in the process, hence, lowering the IR impacts.

With respect to the NPV, the results depend on the optimal tradeoff between the costs related to the consumption of the materials and energy and the revenues generated from the carbonated slags in agriculture. The results showed that the highest NPV case is BOF T2 and AOD T1 respectively. Based on the previously published and accepted study of Watjanatepin et al. (2023) [15] and Watjanatepin et al. (2024) [29] the hotspot in the LCC is the energy costs of the mineral carbonation. Therefore, this explains why the longer duration of mineral carbonation resulted in less NPV. A scenario analysis was also performed on the LCC to investigate the lowest percentage of the carbonated slag products to be sold as basalt rock powder to still achieve the internal rate of return (IRR) of 7%. The motivation behind this is that in reality, it might be quite difficult to reach a high proportion of revenues from selling the carbonated slags as basalt rock powder since the market for basalt rock powder is quite limited to organic farming.

Comparison of drying and no-drying scenarios revealed a negative environmental impact of drying. This was expected since an additional step has been added to the value chain of the mineral carbonation of steel slags. Similarly, the NPV was also lower when the drying step was added. As a future study aspect, the drying step will become particularly important as it is related to the potential environmental impact savings from the reduced mass of the slags to be transported during the use phase of the carbonated slags. This is because dried carbonated slags would weigh less and therefore result in lower environmental impacts from the transportation to the site of application. The LCA study by Lefebvre et al. (2018) on the mineral carbonation of rocks in agriculture have indicated the transportation step as the hotspot in climate change, with transportation accounting for up to 50% of the total life cycle impacts for the GWP depending on the transportation radii [16]. At the present moment in this study, the use phase has not been assessed yet but the impact measurement campaigns are ongoing. The environmental impacts are also higher when the average EU-27 + Switzerland mix is used instead of the Belgian electricity mix. This could be explained by the higher shares of renewable energy in the Belgian electricity mix than the average EU-27 + Switzerland mix. However, the NPV is higher for the cases with the average EU-27 + Switzerland mix because the electricity prices are lower than the prices in the Belgian market. Therefore, this is also one potential tradeoff aspect that can be further explored: higher electricity prices would result in lower NPV but potentially lower environmental impacts in this scenario. This also means that the location of where the mineral carbonation technology is to be adopted within the steel manufacturing value chain could directly influence the environmental and economic impacts. Another interesting aspect of the BOF T3 scenario is that the IRR is dependent on the discount rate that was used for the calculation of the LCC. Up to this point in the research project, the discount rate of 3%, 7% (baseline used and reported in this study) and 15% has been calculated. The trend observed is that only the discount rate of 3% would make the BOF T3 scenario economically viable, but a discount rate of 3% often means that this has to be a public project that is subsidized by authorities. Nonetheless, this would result in the highest carbon sequestration and be the best case environmentally. The influence of the discount rates on the IRR of the best environmental scenario could be worth investigating further as a point of future studies.

In conclusion, the GWP results showed that all the mineral carbonation treatments can serve according to the definition of a CCUS where there is a net negative GWP after the treatment due to the carbon sequestration that outweighs the energy and mass consumed during the carbonation process. Similarly, the FET and HTC results also decrease along with the increased amount of sequestered carbon dioxide due to the avoided impacts of the emission treatments. Inversely for the IR, the IR impacts actually decrease, if there is less sequestration due to the shorter mineral carbonation duration. Finally, the NPV results displayed a reflective optimal point where a tradeoff has to be established between the costs related to the mineral carbonation process (and in particular, the energy costs) with the revenues generated from the carbonated slags products, carbon avoidance revenues according to the European Emission Trading Scheme (EU-ETS) and the avoided landfilling costs of the slags.

The results from the MCDA are illustrated below in Figure 3. It should be noted that the EU-28 mix shown in the figure refers to the average EU 27 + Switzerland energy mix. The figure shows the final total preference scores for each of the 25 scenarios assessed in order to ascertain and rank the performance of the scenarios based on the 5 criteria: GWP, FET, HTC, IR and NPV. As a recap, the aim of this paper, besides assessing the environmental and economic performance of the different scenarios presented in the previous section, is to identify optimal scenario(s) based on the empirical data of steel slags mineral carbonation and on projected scenarios. Across all projected scenarios, the case of BOF T2 with the average EU-27 + Switzerland mix ranked the highest (Figure 3), followed by AOD T1 with the average EU-27 + Switzerland mix, BOF T3 with the 65% of carbonated slags product sold as basalt rock powder, baseline BOF T2 and baseline BOF T3 respectively.

With respect to the empirical cases, if a scenario has to be chosen to be implemented based on only the five baseline cases that has been performed experimentally, then the BOF T2 scenario would be recommended based on this finding. The results also show that the total preference scores decreased when the drying step was added. This is due to the fact that the drying step contributed negatively to increase the environmental impacts and to decrease the overall NPV for the economic performance. It is also generally observed that the cases with the use of the average EU-27 + Switzerland mix resulted in higher total preference scores than the baseline cases because of the higher NPV as a consequence of the lower electricity prices than the average Belgian market mix prices. It is also very interesting to indicate that one limitation of the AHP approach (but also in general for any MCDA) is that the result can vary and is very much dependent on the weighting chosen by the practitioner. In this particular study, an equal 50:50 weight was given to both the environmental and economic criteria. However, the ranking could change completely if this weight ratio is modified which will consequently give a different priority to certain criteria. Therefore, it is advised as a future work possibility to investigate further by conducting a sensitivity analysis on the MCDA study in order to investigate the impacts of the weighting choices on the final preference weighting score. Regardless of this, the current study already gives a preliminary indication of the optimal scenario to be employed in an industrial steel manufacturing value chain to steer towards a circular economy. BOF T2 with the average EU-27 + Switzerland mix seems to be the most optimal scenario when both the environmental and economic performance are taken into account. This may indicate that there has to be support from the Belgian authorities to give subsidies to industries wanting to adopt mineral carbonation as a CCUS in order to have a better performing BOF T2 (or other baseline scenario of interest) to increase the economic performance. This also refers to supportive actions such as lowering the prices of electricity or increasing the revenue streams like the avoided landfilling costs (local authority) or EU-ETS allowance price (European authority) to augment the economic viability.

To further elucidate and aid in the interpretation of the MCDA results, a contribution analysis of the total preference score composition was performed (Figure 4). It should be noted here that the orders shown in Figure 4 are not a direct match to the order in Figure 3. The order of Figure 4 is based on the five groups of sub-scenarios ranging from BOF T1 to AOD T2. It should also be noted that the EU-28 mix shown in the figure refers to the average EU 27 + Switzerland energy mix. Regardless, this does not influence the results. The contribution analysis can be interpreted in groups of 5 scenarios since the trends are repeated amongst the 5 treatment types – BOF T1-T3 and AOD T1-T2. For BOF T1, T2 and AOD T1, the total preference score composition shows that the NPV is actually the highest contributor to the final score. Among these three cases, it is very interesting to observe that the IR scores and the GWP scores are inversely proportional, as explained in the paragraphs above. As a recap, the longer the mineral carbonation, the more carbon is sequestered onto the slags, resulting in higher GWP scores, but the longer carbonation duration results in higher energy consumption which consequently lowers the IR score. This trend is exactly observed between BOF T1, T2 and AOD T1 in this figure. When it comes to BOF T3 and AOD T2, a similar trend is observed where the total preference score of both cases is dominated by the GWP scores due to the high carbon sequestration of the two cases. And since the IR score is inversely proportional, it is clearly seen that the IR score contribution is fairly minimal for both BOF T3 and AOD T2. Moreover, since these two cases have lower absolute NPV scores than BOF T1, T2 and AOD T1, the contribution by NPV is also lower.

As stated above, the main limitation of an MCDA study relates to the weighting since the final ranking score as well as the contribution of the scores will vary according to the weight given to the environmental and economic criteria [10]. To address this uncertainty, it is possible, and advisable as a future study, to perform a sensitivity analysis by varying the weighting of the criteria in order to investigate the impacts of the choice of practitioner. Aligning with the recommendations in the studies of McCord [10] and Falsafi [19], other allocation of the weightings would be interesting to further investigate in order to visualize the impacts on the final MCDA results if other weightings are used (for instance, 0.3 to 0.7 or 0.4 to 0.6 or other ratios). In addition to this, it is also recommended to perform a sensitivity analysis on the NPVs by varying the discount rate. The motivation behind this suggestion is that the contribution of the NPV score is the highest in many carbonation cases, and therefore, this warrants further investigation into what would impact on the total score if the NPV also changes [24]. Alternatively, it might also be of interest to perform a survey with steel-making industries to assess the weight that they would give to decide on the adoption of mineral carbonation as a CCUS. Finally, it would also be of interest to add more sub-criteria to the assessment, for instance, the inclusion of all impact categories of EF 3.0 or other economic parameters. As this study is a part of an ongoing research project, the effects of the carbonation level on the slags in agriculture is still being investigated. Once these impacts have been quantified, it would certainly be interesting to include scenario analyses based on these findings. Additionally, the investigation on whether an extended or a shorter duration of mineral carbonation process should be performed could also be investigated further. Another interesting aspect that is worth further investigation is how would the LCA results change if the investment is directed towards a different CCUS technology? It could be very interesting to perform a consequential LCA to quantify the impacts related to the change in the demand or on the investment of the current system under investigation. Nonetheless, even with the current results, it can be hypothesized that if the IRR of mineral carbonation is lower, economic performance is would also be lower. Therefore, the cases become less preferred, especially if the weighting is given more on the economic performance and not at 50:50. If the IRR is lower, the case should at least ensure that it can have better environmental performance (for instance, capture a lot of C0₂ ) to at least overtake the spot with a better preference score. Thus, the relationship between the IRR and the environmental impacts is certainly an attractive future prospects to explore deeper.However, despite this limitation, the current results obtained are already a very useful preliminary indication for technology developers to ascertain the optimal scenario and the direction of development. Furthermore, the results are also very relevant for technology adopters since this indicates which scenario could fit into the value chain best [25]. Based on the MCDA results presented in this study, the authors are now using these results to plot the Pareto front and to subsequently, calculate the optimal carbonation duration for the process optimization which will become the topic of a future publication. To this end, the MCDA results do indicate a ranking of possible environmentally sustainable and economically viable scenario pathways to valorize both waste streams of the steel manufacturing value chains, namely the steel slags and the waste carbon dioxide in the flue gases.