Modeling the development of a carbon capture and transportation infrastructure for Swedish industry

This work presents and applies a mixed integer programming (MIP) optimization model that minimizes the net present costs for CO 2 capture and storage (CCS) systems for cases with defined emissions costs and/or capture targets. The model covers capture from existing large point sources of CO 2 emissions in Sweden, liquefaction, intermediate storage and transportation using trucks to hubs on the coast, followed by ship transport to a storage location (excluding storage cost). The results show that the capture and transportation infrastructure, in terms of both the sites chosen for capture and the associated transportation setup, differs depending on whether the system is incentivized to capture biogenic or fossil CO 2 , or both. Waste-fired combined heat and power (CHP) plants are only chosen for capture at scale when biogenic capture targets and fossil emissions costs are combined, since the emissions from these sites comprise a combination of biogenic and fossil CO 2 . The value for the system in mitigating the costs from fossil CO 2 emissions exceeds the increased cost of BECCS at waste-fired CHPs compared to larger pulp mills given the fossil emissions cost development assumed in this work. Although the cost for capture and liquefaction dominates the total cost of the CCS system, it is not the only factor determining the choice of sites for capture. Proximity to transport hubs with short offshore transportation distances to the final storage location is also an important factor. For the transportation infrastructure, it is shown that the cost for ships is the main cost driver.


Introduction
Carbon capture and storage (CCS) and bioenergy with carbon capture and storage (BECCS) have been identified as key technologies to achieve deep reductions in carbon emissions from the energy system and carbon-dependent industries.Scenarios considered by the intergovernmental panel on climate change (IPCC) that limit greenhouse gas (GHG) concentrations in the atmosphere in line with the Paris Agreement typically contain high levels of (BE)CCS (Rogelj et al., 2018).In the European Union (EU), it has been proposed that the EU should work towards climate neutrality by Year 2050 (European Comission, 2020).Sweden, which has a target of net-zero GHG emissions by Year 2045, has in a recent public inquiry (SOU, 2020) proposed negative emissions in the range of 3-10 Mt/year by Year 2045, as a so-called 'supplementary measure' for offsetting residual emissions from hard-to-abate sectors and to contribute to the target of net-negative emissions after Year 2045.Thus, BECCS is important in generating net-negative emissions and in offsetting hard-to-abate fossil emissions.CCS is, therefore, important for industrial sectors in which fossil emissions are difficult to abate using other technologies, for example in the cement, chemical and petroleum refining industries (IVA, 2019).
Techno-economic analyses of carbon capture facilities are usually carried out on the plant level, with detailed investment cost estimations based on the specific plant configuration and conditions (see for example Biermann et al., 2018;Garðarsdóttir et al., 2018;Johnsson et al., 2020;Martinez Castilla et al., 2019).The specific cost for capturing CO 2 in such studies usually lies in the range of 40-100 €/tCO 2 depending mainly on the size of the emission source, the CO 2 concentration in the flue gas, the potential for utilizing residual heat for the desorption process during post-combustion capture, and the yearly operating time.Kjärstad et al. (2016) investigated potential transportation solutions for CO 2 in the Nordic region and concluded that ship transport is the most-cost-efficient and feasible mode of transportation, especially during a ramp-up phase.Knoope et al. (2015) compared investments in the pipeline and ship transport of CO 2 under uncertainty, and concluded that ships are the preferable mode when transporting low volumes over large distances.Large-scale CO 2 capture and infrastructure developments that consider capture from large European CO 2 emissions sources utilizing transnational pipelines for CO 2 transportation have been studied previously (see Kjärstad et al., 2013;Kjärstad et al., 2014;Morbee et al., 2012;d'Amore and Bezzo, 2017).Some of these works (Kjärstad et al., 2013;Kjärstad et al., 2014;Morbee et al., 2012) have considered capture from large electricity-generating plants and pipeline networks for (cross-border) transportation of CO 2 and potential storage sites on-shore and off-shore.d'Amore and Bezzo ( 2017) have investigated on-shore and off-shore pipelines, along with ships for CO 2 transportation, for large-scale CO 2 emitters in multiple sectors in Europe.d'Amore and Bezzo (2017) have reported costs in the range of 27-38 €/tCO 2 for a system that captures up to 70% of the total CO 2 emissions of the system over 20 years, and they have concluded that the capture cost dominates the total supply chain costs, with transport and sequestration costs never becoming more than 10% of the total system cost.
In the case of the EU, technological and cost developments in the electricity-generating sector and legal frameworks rendering crossborder transportation and land-based storage of CO 2 difficult have geared recent CCS developments more towards national systems, with capture primarily aimed at industrial sectors with hard-to-abate emissions, incentivized by national policy measures.Such emissions sources may be smaller than large power plants and, thus, the specific costs for capture, transport and storage may be higher, depending on the extent to which transport of CO 2 can be in the form of an integrated system for several capture projects.
This work builds upon previous techno-economic analyses of CCS systems, on both the plant and infrastructure levels, and presents an optimization model for studying the development of regional CCS systems.The model estimates the temporal and spatial distributions of optimal investments in capture technology at existing industrial sites and within the infrastructure.The proposed method is applied to study the impacts of, and relationships between, CCS and BECCS incentives, as well as the influence of the sensitivity to cost parameters in the CCS chain on the cost-optimal development of the CCS system.Since the main decarbonization strategy for the iron and steel industry is in the form of hydrogen-driven steelmaking, we also investigate how the exclusion of the iron and steel industry influences the results.Finally, we use the modeling to investigate how a presumed early implementer of BECCS influences the results.

Mathematical model
To study the development of a regional CCS system, we develop and apply a mixed integer linear programming (MILP) optimization model that minimizes the net present value (NPV) of costs for CO 2 capture and infrastructure systems.A complete model description that includes the equations for determining the investments, installed capacity, and costs for individual parts of the CCS chain are given in Appendix A, together with a complete nomenclature.The model is implemented in the general algebraic modeling system (GAMS) and solved using the GAMS Cplex solver, which uses a branch and cut approach to solve a series of linear programming (LP) subproblems.The model consists of 335,224 equations and 3215,998 variables and takes around 10 min to solve on an Intel Core i5 processor with 16 GB of installed physical memory and a relative error tolerance of 0.1% i.e., the proportional difference between the solution found by the solver and the best theoretical objective function.In the model, a binary variable is used to control which transport hubs are used (see Eq. ( 8), which exemplifies the use of this binary variable), and an integer variable is used to control how many ships are invested in to transport CO 2 between the transport hubs and the storage location.Fig. 1 gives an overview of the model.The costs consist of capital expenditures (CAPEX) and operating expenditures (OPEX) for the technologies in the investigated CCS chains, namely, capture, liquefaction, storage (on-site after liquefaction and at transport hubs), trucks and ships.The OPEX parameter consists of different cost items depending on which part of the CCS chain that is considered.For example, the OPEX for CO 2 capture and liquefaction consists of energy costs (dependent upon the captured CO 2 flow) and operation and maintenance costs (dependent upon the installed capacity), while the OPEX for transportation is additionally dependent upon the transportation distance.
The cost of the geologic storage is not included, since it will be the same for all cases (only one location for final storage is considered) and, thus, will only influence the total cost and not the system configuration.Site emissions are divided between multiple stacks located at the sites.Capture is performed on individual stacks and CO 2 liquefaction is performed at the site where it is captured.The system has a fixed number of point sources with the CO 2 available for capture being limited by the present emissions from the given stacks and the capture rate of the CCS technology.The boundary conditions governing the model are the mass balances, emission costs, and capture targets.
The objective function of the model is to minimize the net present value of the sum of all the annual costs for capture and liquefaction installations, transportation infrastructure and emissions costs.The objective function is described as: where c tot,NPV is the net present value of the total cost associated with capture, transportation, and emissions of CO 2 for the entire period investigated, c annual y is the annual costs, y is the years investigated, and r is the discount rate.The annual costs c annual y are calculated as:   according to Eq. (3).
where x i,j,y is the flow of CO 2 captured at site i from stack of type j and s i,j is the yearly CO 2 available for capture at site i from stack type j.The yearly flow of CO 2 captured at a given site relates to the flow out of the site according to Eq. (4).
where z i,l,y is the flow of CO 2 from site i to transport hub l.The annual CO 2 captured by the system is calculated using Eq. ( 5).where e capture,annual et,y is the annual CO 2 of emission type et (biogenic or fossil) captured by the system and m j,et is the share of emissions from stack type j that are biogenic or fossil in origin.The capture target for biogenic and fossil CO 2 must be satisfied according to Eq. ( 6). is the installed storage capacity at transport hub l in year y, u storage,hub is the pre-determined size of storage at a transport hub, and γ l,y is the binary variable determining whether or not hub l is used in year y.

CCS system and scenario analysis
The development of the CO 2 capture and transportation infrastructure is determined by the mixed integer programming (MIP) optimization model described in the previous section.The model is used for studying the potential synergies between fossil CCS and BECCS implementation and the impacts of cost variations in the CCS chain.

System description
The modeling considers emissions from Swedish industries and energy plants and CO 2 transportation to storage in the Norwegian North Sea.The sites and potential transport hub locations included are shown in Fig. 2. Process-related industrial sites are included based on their emission of more than 100 ktCO 2 /y and being included in the Chalmers Industrial Case Study Portfolio (for more information, see Svensson et al. (2019)).Heat and power plants are included in the modeling based on their emitting at least 100 ktCO 2 /y and being included in the paper of Beiron et al. (2022).This results in 86 sites being included in the modeling: 48 heat and power plants, 29 pulp and paper mills, 3 refineries, 3 iron and steel mills, 2 cement plants, and 1 chemical plant.It should be noted that CCS might not be the only relevant mitigation option for the sites in the modeled sectors.For example, the iron and steel industry in Sweden is currently focusing on fossil-free steel production through direct reduction using hydrogen.Nevertheless, mitigation options other than CCS are not included in the model.
The captured CO 2 is transported by trucks on land and with ships offshore.The discount rate used in the modeling is set at 5%.In the modeling, capture can be implemented at individual stacks at a given site (see Fig. 1), and the investment cost for capture depends on the CO flow and concentration from a given stack.The model is limited to yearly time-steps over the period of 2020-2045, with investments and operating costs for sites with industrial processes considered with the assumption that production, and therefore emissions, are evenly distributed over 8000 operating hours each year.For the heat and power sector (which is typically only operated for part of the year), investment costs are related to the peak flow of CO 2 and are taken from Beiron et al. (2022).Yearly capture targets or costs for emitting CO 2 determine CCS implementation in the modeling.At each site, carbon capture may be installed to capture 0%-90% of the emissions from a given stack, and each site has 1-3 stacks from which capture can be performed.A complete list of all the included sites, the stacks at each site, and the site emissions is presented in Table A2 in Appendix A. The modeling considers the current Swedish industries, which are assumed to remain in place and be active until 2045 at the maintained production and emission levels.
It should be pointed out that the present work minimizes the societal cost based on the available cost data and is, therefore, not necessarily representative of the conditions faced by individual companies that implement CCS (e.g., the current price offered for transportation by one actor -Stockholm Exergiseems to be significantly higher than what is obtained from the costs available in the literature).

CO 2 capture and liquefaction
The capture rate and specific heat demand of CO 2 capture are based on absorption in an aqueous monoethanolamine (MEA)-based solvent.The costs for capture and liquefaction equipment are based on Eliasson et al. (2022) for process industries and Beiron et al. (2022) for the heat and power sector.Other technologies for capturing carbon may be more efficient for certain applications, but the MEA-based capture process represents a mature technology and is considered a benchmark technology for carbon capture.In addition, absorption is relatively easy to implement as an end-of-pipe solution for emission mitigation at exiting sites.The CAPEX estimations for CO 2 capture and liquefaction at process industries are based on the above-mentioned work by Eliasson et al. (2022).The expression depends on the CO 2 mass flow, as well as on the CO 2 concentration and is written as follows: where the values of α and β depend on the CO 2 concentration in the flue gas.The biogenic carbon shares, CO 2 concentrations, and values of α and β for each stack type in the model are presented in Table 1.The CO concentrations, taken from the Chalmers Industrial Case Study Portfolio, have either been documented in case studies at relevant sites or reported as a general value for an industry of that type.The specific CAPEX (€/tCO 2 ) used for implementation in the model is calculated based on a CO 2 flow corresponding to a capture rate of 90%.Different policy measures are simulated by the modeling for biogenic and fossil CO emissions .The model distributes emitted CO 2 as biogenic or fossil according to Eq. ( 5), i.e., it does not optimize the allocation between captured fossil CO 2 and captured biogenic CO 2 .Table 2 lists the general economic assumptions for the capture and liquefaction equipment (not site-or stack-specific).Operating costs for capture comprise the operational and maintenance costs and the cost for steam to regenerate the solvent in the reboiler.After separation, the CO is liquefied onsite in preparation for truck and ship transportation.

Intermediate storage
Intermediate storage is installed at both the site where CO 2 is liquefied and the transport hubs.The cost parameters for storage tanks are listed in Table 3. Intermediate storage at the site is designed to hold the total amount of CO 2 captured in a 24-hour period, while intermediate storage at transportation hubs is designed to match 120% of the CO 2carrying capacity of the ship, i.e., 20% more than the theoretically required storage capacity.The model chooses between 15 locations for investment in transport hubs (the hubs are marked with crosses in Fig. 2).

Transportation
Trucks transport the liquefied CO 2 between the site and the transport hub and ships transport the liquefied CO 2 from the transport hub to Kollsnes, Norway.It should be noted that liquefied CO 2 is currently classified as dangerous goods (Swedish Civil Contingencies Agency 2022), and therefore, trucks carrying CO 2 might need to avoid some routes in urban areas, which could lead to longer transportation distances.At Kollsnes, there will be an intermediate storage unit from which a pipeline will transport the CO 2 to the final storage location beneath the North Sea.As mentioned above, the costs for the intermediate storage in Kollsnes and the storage cost (including the pipeline from Kollsnes to the injection hole) are not included.Ship transport is chosen over pipeline transport based on the studies conducted by Kjärstad et al. (2016), Knoope et al. (2015), and the Norwegian CCS-project Longship (CCS Norway 2021).Ships are purchased in integer steps of a pre-determined size.The assumptions and input data required to determine case-specific costs for truck and ship transportation are presented in Table 4.The assumed diesel cost of 1.4 €/l might seem optimistic at the moment, with prices currently in the range of 2-3 €/l.However, it is uncertain whether the current price levels of transportation fuels are the new normal, or if they will return to lower levels.A sensitivity analysis is therefore carried out on the cost of transportation fuels.The calculated distance between a given site and a transport hub is shown in Table A3 in Appendix A. The distance from each hub to Kollsnes was measured using the GIS software and a terrain factor of 1.1 was used.The transport of CO 2 by trucks and ships contributes to additional CO 2 emissions, and the emissions factors used in this work are presented in Table 4.

Scenarios
Table 5 presents the three scenarios for capture incentives, including fossil emissions costs and/or capture targets for BECCS, as investigated in this work.Scenario 1 considers only a cost for emitting fossil CO 2 , without incentives for BECCS.The fossil emissions cost is shown in Fig.
and is based on the Net Zero Emissions by 2050 scenario in the World Energy Outlook (International Energy Agency 2021).Scenario 2 considers capture targets for biogenic CO 2 that are based on a Swedish public inquiry that proposed levels of negative emissions from different technologies, including BECCS (SOU 2020).Scenario 3 combines both of these incentives, i.e., a cost for emitting fossil CO 2 and capture targets for biogenic CO 2 .It should be noted that for Scenarios 1 and 3, the model only includes two alternatives for the fossil CO 2 emissions, i.e., investment in CCS to mitigate the emissions or paying the CO 2 emission cost, whereas mitigation options such as electrification and fuel switching are not considered.
A sensitivity analysis for some the parameters given in the rightmost column in Table 3 is performed for Scenario 3. Scenario 3 is chosen because it involves incentives for both fossil and biogenic CCS, and this is likely to be the case for Sweden in the future (EU ETS for fossil CO and proposed BECCS targets, incentivized by a reverse auctioning system, for biogenic CO 2 ).The sensitivity analysis includes the costs of investments in capture and liquefaction facilities (+50%) (Site CAPEX*1.5) and the cost for transportation fuel (+100%) (Fuel cost*2).The third (in addition to the base case) sensitivity case assumes that 50% of the reboiler steam demand can be covered by residual heat on-site, at no cost (Heat integration).The fourth sensitivity case investigates the influence on the cost of a so-called "early mover", by assuming that one actor, a bio-based heat and power plant in Stockholm, invests in the capture of 0.5 MtCO 2 /year in Year 2022 and captures at least this amount of CO 2 during the period of 2022-2045.The final sensitivity case excludes capture implementation in the iron and steel industry, since the Swedish iron and steel industry is currently aiming to produce fossil-free steel using green hydrogen (I&S excluded).
The Site CAPEX*1.5 is to reflect uncertainties in the investment costs for capture equipment, or conditions faced by industrial actors that use shorter economic lifetimes or higher discount rates for performing investment calculations.For instance, in the case of using the annuity factor method, using a discount rate of 5% and drastically reducing the economic lifetime from 25 to 12.5 years increase the annuity factor, and thus the yearly cost for CAPEX, by around 55%.The I&S excluded case is modelled since the Swedish iron and steel industry is currently envisioning a process where iron ore is reduced using direct reduction with hydrogen as a reduction agent.After the direct reduction, the produced sponge iron is melted using electric arc furnaces to produce steel.Thus, CCS is not the primary technological path considered in Swedish iron and steel manufacturing.For the early mover case, choosing Year 2022 for BECCS implementation, which is obviously unlikely, ensures that this actor is the first to implement capture, making it possible to judge if this has an impact on the subsequent development of the CCS infrastructure.

Table 6
Infrastructure costs for Scenarios 1, 2 and 3 over the entire modeled period.The specific system cost includes the cost for the entire CCS chain (capture, liquefaction, trucks and ships and intermediate storage) over the modeled period.The specific transport cost includes the costs for trucks, intermediate storage and ships over the modeled period.

System development and synergies between fossil CCS and BECCS
Fig. 4 shows how the amounts of fossil and biogenic CO 2 captured develop over time, and Figs.5-7 show the optimal system configurations in Years 2030 and 2045 for Scenarios 1-3.Table 6 gives the specific system cost (specific cost for the entire CCS chain over the modeled period in €/tCO 2 ), the specific cost for transportation infrastructure (specific cost for trucks, transport hubs and ships in €/tCO 2 over the modeled period), and the truck transportation distance and the ship transportation distance (1000*km/MtCO 2 ) for Scenarios 1-3.
In Scenario 1 (Figs.In Scenario 2 (Figs.4b and 6), biogenic emissions are captured in line with the capture targets by implementing capture at large pulp and paper mills and at one bio-fired CHP plant.The specific cost of the

Table 7
Indicator values for the CCS system for Scenario 3 and the three studied cost cases.To enable comparison between cases, all values are divided by the amount of CO 2 captured in that case.

Base case
Site CAPEX* 1.   system in Scenario 2 is 83 €/tCO 2 .Scenario 2 exemplifies well how the sites chosen for capture are not selected solely based on minimizing the investment costs for capture and liquefaction installations.In Year 2030, two pulp and paper mills are chosen for capture, one on the east coast, which is the largest, and one on the west coast, which is the tenth largest of the pulp mills in the model.Since the CAPEX for capture and liquefaction is related to the size of the emission source (see Eq. ( 10)), the implementation of capture at the pulp mill on the west coast is motivated in part by its proximity to a transport hub close to the ship destination in Kollsnes, Norway.Comparing Figs. 5 and 6, there are considerable differences regarding which sites are chosen by the model for capture as well as the choice of transport hubs, since the biogenic and fossil emissions sources are differently distributed.However, there are similarities with respect to the build-up of the system in Scenarios 1 and 2, since in both cases these large investments are made over short time-spans.In Scenario 2, this is obvious due to the capture targets.In Scenario 1, this means that over a relatively narrow range of carbon prices, the capture and transportation of the majority of the CO 2 from the fossil point sources included in the model become economically preferable compared to the cost of emitting the CO 2 .
In this work we minimize the NPV since the model may choose to incur costs in the future, and therefore we need to account for the value of money over time.It should be noted that when minimizing the NPV there is a risk that investments are postponed, in particular if the discount rate is overestimated.In this work we apply a discount rate of 5% to represent the social planner perspective.To avoid postponing investments to a point where the deployment of technology required to reach climate goals is no longer feasible, incentives could contain intermediate targets to initiate timely implementation.Such intermediate targets would make the ramp up of the biogenic CO 2 capture system for Year 2045 in Scenario 2 take place over a longer time period.
Scenario 3, which entails a combination of CO 2 prices for fossil fuel emissions and targets for captured biogenic CO 2 (Figs.4c and 7), exposes important differences in the modeling results, as compared to Scenarios 1 and 2. In Year 2030, when the first biogenic capture target comes in to play, the model invests in capture on waste-to-energy plants in Scenario 3. In Scenario 1, on the other hand, only one waste-to-energy plant is chosen for capture implementation late in the modeled period and in Scenario 2 no waste-fired CHP plants are chosen for capture.This shows that for a case in which there is both a price placed on fossil CO 2 emissions and capture targets imposed on biogenic emissions, there is a synergy (low cost) for waste-to-energy plants, since the biogenic CO 2 captured from these sites is fulfilling the biogenic capture targets, while the fossil CO 2 captured reduces the overall system cost by mitigating the costs associated with fossil emissions.This effect is highlighted in Fig. 8, which shows the costs for biogenic capture, liquefaction, and on-site storage for the largest waste-fired CHP plant (all costs are allocated to the 65% of emissions that are biogenic) and pulp mill recovery boiler in the model, and the value of the mitigated fossil emissions in the case of the waste-to-energy plant.The costs are taken from Year 2030 when the first biogenic capture targets are implemented, and Scenario 3 for the waste-fired CHP, and Scenario 2 for the pulp mill.In Fig. 8, the value of mitigated fossil emissions (set to the avoided cost for emitting fossil CO 2 ) is allocated to each ton of biogenic CO 2 captured.The fossil emissions cost in 2030 makes the CHP plant roughly equivalent to the pulp mills recovery boiler, considering only the site-related costs (compare the blue bars with the orange bar in Fig. 8).For the system, this means that biogenic capture is shifted from large point sources to a larger number of smaller point sources that are emitting both biogenic and fossil CO 2 if the value of removing fossil CO 2 emissions offsets the cost difference in capturing biogenic CO 2 from the different point sources.
In Table 6, it is evident that the specific system cost for is lower if both biogenic and fossil CO 2 is captured (compare Scenario 3 with Scenarios 1 and 2).In addition, Scenario 3 gives the lowest specific cost for transportation infrastructure due to the combination of short ship transportation distance and truck transportation distance in relation to the amount of CO 2 captured (1,000 km per MtCO 2 ) (see rightmost column in Table 6).The specific transportation infrastructure cost is highest in Scenario 1 due to a larger share of the total CO 2 being transported long distances by ship from the northern-most transport hub, thereby requiring large investments in ship capacity.At the same time, the truck transportation distance is shortest in Scenario 1.This shows that the total ship transportation distance is more important than the truck transportation distance in determining the overall cost of the transportation infrastructure for the system, since the decreased costs resulting from shorter truck transportation distances are outweighed by the cost increase from longer ship transportation distances in Scenario 1.It is important to note that the results concerning transportation infrastructure costs would likely be different if pipeline transportation costs were to be included in the analysis.

Sensitivity analysis
Table 7 summarizes the results of the sensitivity analysis, i.e., applying the parameter variation given in the rightmost column in Table 5 for Scenario 3. The increased investment cost for capture (Site CAPEX*1.5)results in the largest increase in specific system cost and, thereby, a reduction of the total amount of CO 2 captured (less fossil carbon is captured, whereas the biogenic capture targets are still valid).The truck and ship fuel usage levels increase relative to the base case, so as to focus the capture installations to larger point sources and, thereby, lower the specific CAPEX for capture and liquefaction.Increasing fuel cost (Fuel cost*2) decreases the truck fuel use, as well as the emissions intensity from the truck and ship infrastructure relative to the base case.The lower level of truck fuel usage is achieved by implementing capture at slightly smaller stacks that are located closer to coastal transport hubs.The ship fuel usage level is not noticeably affected by increasing fuel costs, indicating that the use of ships in the system is not sensitive to their fuel-related costs.Heat integration significantly reduces the specific system cost (and increases the amount of CO 2 captured), although it results in higher truck fuel use due to the increased demand for CO 2 transportation when capture is implemented at additional sources.Excluding the iron and steel industry from the analysis (I&S excluded) while maintaining the fossil CO 2 cost and target on biogenic capture results in a system with a slightly higher specific cost than the base case, which is due to several factors.First, implementing capture at a large iron and steel plant entails a lower cost compared to implementation at the other plants in this study.Second, two of the iron and steel plants included in this work have good conditions for transportation infrastructure.These two plants, Luleå and Oxelösund (cf.Fig. 2), are located at potential transport hub locations, which are used in all the sensitivity cases, except for the I&S excluded case, which is reflected in the increased use of truck fuel for this case.The high level of ship fuel usage in the I&S excluded case indicates that there is less-utilization of the purchased ships (which are bought in integer steps of a fixed size; cf.Section 3.1.3and Table 4) compared to the other cases, giving higher ship fuel use per transported tonne of CO 2 .
The early mover case results in a slightly higher specific system cost due to the model choosing a system that needs to capture biogenic CO 2 before the targets imposed on BECCS come in to play.The ship and truck fuel intensities are higher than in the base case, indicating that with respect to infrastructure considerations, this solution is not optimal from a societal perspective.Fig. 9 compares the use of transport hubs over time, between the base case and the early mover case.In the early mover case, the hub in Stockholm is used from Year 2022 onwards, compared to the base case, where it is used from Year 2030.The first implementation of fossil emissions capture in the early mover case is at the iron and steel plant in Oxelösund in Year 2023, capturing around 30% of the total emissions from this site and transporting it to the hub in Stockholm.In Year 2024, with the implementation of the hubs in Oxelösund, Lysekil, Luleå and Slite, which are mostly used for transporting CO 2 from large fossil emitters, full capture is installed at the steel plant, and the CO 2 is instead transported to Oxelösund, similar to the base case.In other words, this rather large change in initial conditions has only a minor impact, limited in time to 1 year, on how the fossil capture system is developed.The transport hub in Helsingborg is used from Year 2030 in the early mover case (Year 2045 in the base case) and receives captured CO 2 from waste-fired CHP plants in southern Sweden.Although the early mover case results in a slightly higher societal cost, the increase in specific system cost is only around 2% compared to the base case, while the increase in the cumulative amount of biogenic CO 2 is around 10%.Fig. 10, a and b show the cost structure of the system.Both the CAPEX and OPEX are dominated by capture and liquefaction, so the cost is most sensitive to the Site CAPEX*1.5 case and the Heat integration case.The cost structure of the truck transportation stands out in that the OPEX is significantly higher than the CAPEX.The ship transportation CAPEX is similar in magnitude to the OPEX, and all in all, significantly higher than the truck costs, which makes the transportation infrastructure moresensitive to ship-related cost uncertainties.Note that the OPEX for capture and liquefaction is high compared to what has been proposed in previous studies (Garðarsdóttir et al., 2018;Johnsson et al., 2020).The OPEX is sensitive to the energy cost and the integration into the plant energy system (Biermann et al., 2018;Eliasson et al., 2022).Heat integration with the existing site energy system is investigated in only one sensitivity case in this work and should be investigated further.
Fig. 11 shows the incremental annualized CAPEX (M€/year), i.e., the size of additional investments taken each year, and the cumulative CO 2 capture (Mtonne) for the sensitivity analysis of Scenario 3. The Site CAPEX*1.5 case delays the break-even point between emitting and capturing fossil CO 2 , and thus the first investment in CCS equipment, by 2 years compared to the base case, which leads to a lower cumulative level of CO 2 captured over the modeled period.Conversely, the Heat integration case brings the break-even point, and the first investments, forward by 2 years, resulting in the highest level of cumulative capture and the lowest annual system cost over the period.The early mover case does not differ notably from the base case in terms of the cumulative level of CO 2 .However, it differs in terms of how much biogenic versus fossil CO 2 is captured and overall requires higher investments in relation to the CO 2 captured.The I&S excluded case, which removes the option for the model to invest in capture at three large fossil sources, expectedly results in lower cumulative CO 2 capture and lower levels of early investments.The Fuel costs*2 case delays a significant part of the investments by 3 years, albeit not to the extent seen in the Site CAPEX*1.5 case.Although the sensitivity cases shift investments in time, the fossil point sources tend to send CO 2 to the same transport hubs once capture is implemented (apart from the iron and steel mill in Oxelösund sending CO 2 to Stockholm in the early mover case in Year 2023).The only minor difference relates to whether the chemical manufacturing plant on the west coast makes use of the transport hub in Lysekil or Göteborg or both.This shows that the configuration of the fossil CCS system is rather robust to the investigated sensitivity cases.

Conclusions
A MIP optimization model is developed and applied to study the development of CCS infrastructure systems that include capture, liquefaction, and transportation applied to fossil and biogenic (BECCS) emissions sources in Sweden.
The results show that the CCS system configuration differs according to the incentive offeredapplying a cost only for fossil emissions, a cost only for BECCS targets, or a combination of these.The system configurations differ both in terms of the sites chosen for capture and the transportation infrastructure used.Waste-fired heat and power plants in this work are assumed to emit 65% biogenic CO 2 and 35% fossil CO 2. Therefore, combining the cost for fossil fuel emissions and targets for biogenic CO 2 capture has a strong effect on the waste-to-energy sector, in which captured emissions both assist in reaching the capture targets for BECCS and decrease the total system cost by mitigating fossil-derived emissions.This means that the value gained by the system in mitigating costs from fossil emissions by capturing CO 2 at waste-fired CHP plants outweighs the cost benefit of implementing capture at larger pulp mills, which would otherwise have a lower specific CCS cost (€/tCO 2 ).Based on these results, it is important to investigate further the impacts of different policy schemes to motivate BECCS and fossil CCS in combination, to be used as the basis for designing policies that would incentivize BECCS and fossil CCS.With the fossil emissions cost in the model (based on the Net Zero Emissions by 2050 scenario in the World Energy Outlook) and the cost data and assumptions used for CCS equipment, fossil capture is ramped up rapidly.In principle, this outcome is similar to that achieved when setting capture targets in terms of the MtCO 2 captured in a specific year.The configuration of the infrastructure that transports the CO 2 from the large fossil point sources in the model, i.e., using hubs to which the captured CO 2 is sent, is resilient to the changes investigated in the sensitivity analysis.However, the implementation is shifted in time when changes are made to the cost structure.
Although the results show that the on-site equipment items (for capture and liquefaction) dominate the cost structure, the design of the CCS system is not based solely on minimizing the capture and liquefaction costs.Proximity to transport hubs, especially those that require short-distance ship transportation routes to reach the final storage location, is also an important factor in deciding where capture is to be implemented.The cost for transportation infrastructure (trucks, ships and transport hubs) is, for the most part, affected by ship transportation, such that the infrastructure costs show a positive correlation with increased ship transportation distances.These results highlight the S. Karlsson et al. importance of considering the costs for CO 2 transportation in technoeconomic evaluations of CCS.Finally, it should be pointed out that the present work minimizes the societal cost based on the available cost data and is, therefore, not necessarily representative of the conditions faced by individual companies implementing CCS (e.g., the price offered for transportation at present seems to be significantly higher than the costs listed in the literature).

Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Table A2 shows the sites and stacks included in the modeling and the biogenic and fossil CO 2 emissions from the sites.

Table A1
Nomenclature for the model.

Fig. 1 .
Fig. 1.Flowsheet overview of the model.The black dashed line indicates the parts of the CCS chain included in the cost minimization modeling.The red dashed lines include the parts of the CCS chain that are located at the site i∈I.

Fig. 2 .
Fig. 2. Industrial sites, locations for transportation hubs and ship routes with investment possibilities for the model.The transport hub location names are given in the figure.The sizes of the industrial sites correspond to the total emissions from the sites, with the smallest site emitting around 100 ktCO 2 /y and the largest site emitting around 3300 ktCO 2 /y.

Fig. 3 .
Fig. 3.The cost development trajectory for fossil CO 2 emissions, as given by the Net Zero Emissions by 2050 (NZE) scenario in the World Energy Outlook (International Energy Agency 2021) using a € to USD conversion factor of 0.88 €/USD and linear interpolation between decades.

Fig. 4 .
Fig. 4. Amounts of fossil and biogenic CO 2 captured, respectively in Scenarios 1 (a), Scenario 2 (b), and Scenario 3 (c).The sharp increase in the level of biogenic CO captured in Year 2045 is due to the drastic increase in the modeled capture target for Year 2045 (if the time period of the study would have been extended beyond Year 2045, this would be a new plateau similar to the one seen between 2030 and 2044).

Fig. 8 .
Fig. 8. Costs of biogenic capture and liquefaction, including on-site storage, for the largest waste-fired CHP plant (taken from Scenario 3) and pulp mill recovery boiler (taken from Scenario 2) in the modeling for Year 2030.The value of mitigated fossil emissions (set to the avoided cost for emitting fossil CO 2 ) is allocated to each ton of biogenic CO 2 captured.
4a and 5), capture of fossil emissions from the emission sources included in this work are driven by the emission cost from Year 2024 onwards.Additional investments are then made in Years 2026 and 2028, as well as later in the period, as the emissions cost keeps increasing.This indicates that with the modeled cost development, CCS at large industrial sites can become cost efficient in the near-term, and policies aimed at ramping up development might be motivated.The specific cost of the system over the entire modeled period in Scenario 1 is 84 €/tCO 2 .The biogenic emissions captured in Scenario 1 represent the share of biogenic emissions in the combined stack from the cement plants and, in Years 2044 and 2045, also from a waste-fired CHP plant.

Fig. 9 .
Fig. 9. Timeline of the use of transport hubs in Scenario 3 and the base case (blue), and the early mover case (orange) where a bio-fired CHP plant in Stockholm implements capture in Year 2022.

Fig. 10 .
Fig. 10.Cost structure of specific CAPEX (a) and OPEX (b) for Scenario 3 including the sensitivity analysis.

Fig. 11 .
Fig. 11.Incremental annualized CAPEX (bars) and cumulative captured emissions (lines) over time for Scenario 3 and the studied cost cases from Year 2020-2045.

Y
Time-steps in years, Y ∈ [yearstart, ⋯, year end ] I Sites included in the model I ∈ [site1, …, siten] J Stack type J ∈ [stack type1, …, stack typen] L Coastal transport hubs L ∈ [hub1, …, hubn] ET CO 2 emission type ET ∈ [biogenic, fossil ] xi,j,y Flow of CO 2 captured at site i ∈ I from a stack of type j ∈ J in year y ∈ Y [tCO 2 ] zi,l,y Flow of CO 2 between site i ∈ I and hub l ∈ L in year y ∈ Y [tCO 2 ] γ l,y Binary variable determining whether hub l ∈ L is used in year y ∈ Y [γ l,y ∈ 0, 1] ctot,NPV Total net present value of system [M€] c annual OPEX for ships transporting CO 2 between hub l ∈ L and the final storage location in year y ∈ Y [M€] e capture,total et Total CO 2 of type et ∈ ET captured [tCO 2 ] type et ∈ ET captured in year y ∈ Y [tCO 2 ] e em,total et Total CO 2 of type et ∈ ET emitted by the system [tCO 2 ] type et ∈ ET emitted by the system in year y ∈ Y [tCO 2 ] Investment in CO 2 capture capacity at site i ∈ I on stack of stack type j ∈ J in year y ∈ Y [tCO 2 ] Investment in liquefaction capacity at site i ∈ I in year y ∈ Y [tCO 2 ] Investment in on-site storage capacity at site i ∈ I in year y ∈ Y [tCO 2 ] Investment in CO 2 storage capacity at transport hub l ∈ L in year y ∈ Y [tCO 2 ] Investment in truck transport capacity between site i ∈ I and hub l ∈ L in year y ∈ Y [tCO 2 ] Investment in CO 2 ship transport capacity at hub l ∈ L in year y ∈ Y [tCO 2 ] Installed capture capacity at site i ∈ I on stack of stack type j ∈ J in year y ∈ Y [tCO 2 ] Installed liquefaction capacity at site i ∈ I in year y ∈ Y [tCO 2 ] Installed on-site storage capacity at site i ∈ I in year y ∈ Y [tCO 2 ] Installed storage capacity at transport hub l ∈ L in year y ∈ Y [tCO 2 ] Installed truck transport capacity between site i ∈ I and hub l ∈ L in year y ∈ Y [tCO 2 ] Number of ships installed to transport CO 2 from hub l ∈ L in year y ∈ Y [b ship l,y ∈ [0, 1, 2…n

Table 1
Shares of biogenic carbon, CO 2 concentrations, and values of α and β (according to Eq. (10)) for each stack considered in the model.

Table 2
Economic assumptions made for the capture and liquefaction equipment.

Table 4
Input data and assumptions for the truck and ship transportation calculations performed in the modeling.

Table 5
Policy scenarios used for the modeling.
Yearly supply of CO 2 emissions available for capture on site i ∈ I from stack type j ∈ J [tCO 2 ] mj,et The ratio of emission type et to total CO 2 emissions from stack of type j ∈ J LT Lifetime of equipment [years] e target et,y Emissions capture target for emission type et ∈ ET in year y ∈ Y [tCO 2 ] e CO2 i,et Yearly emissions from a given site i ∈ I of type et ∈ ET [tCO 2 ] c CO2 et,y Cost for emitting CO 2 of type et ∈ ET in year y ∈ Y [M€/tCO 2 ] S. Karlsson et al.

Table A2
Included sites and stacks for the modeling performed in this work.The site ID abbreviations are as follows: PP, Pulp and paper; Ce, Cement; R, Refinery; IS, Iron and steel; C, Chemical; HP, Heat and power.