Investigating the Potential for Increased Energy Utilisation and Reduced CO 2 Emissions at Mo Industrial Park

: The potential for increased energy utilisation and reduced carbon footprint has been investigated for the industrial park Mo Industri Park (MIP), located at Mo i Rana, Norway. Process data has been gathered to quantify the energy ﬂows between industrial clients. The energy ﬂows have been visualised quantitatively in Sankey diagrams, while the quality of the available energy is presented in the form of a grand composite curve. High temperature ﬂue gas from ferrosilicon (FeSi) production at Elkem Rana represent the largest heat source available for utilisation. A theoretical assessment of potential applications for this energy is presented and includes: (1) electricity production; (2) local biocarbon production, where surplus heat is utilised for drying of wood chips; (3) post combustion carbon capture, where surplus heat is utilised for solvent regeneration. The results indicate that increasing the current energy recovery from 400 GWh to > 640 GWh is realistic. The increase in energy recovery can be used for reducing the carbon footprint of the industrial park. Investment in a common utility network for surplus heat may lower the threshold for establishing other energy clients at MIP. These are possibilities which may be investigated in more detail in future work.


Introduction
The industry sector is directly emitting 24% (8.5 Gt) of total global CO 2 emissions and is accountable for 37% (156 EJ) of the global energy use [1]. Increased attention has been given to how interactions among different companies, industries and sectors can contribute to lowering the industry sectors' energy use and greenhouse gas emissions [2]. Such interactions are often seen within the scope of the circular economy and are usually divided into three levels: the micro level (products, companies, consumers), meso level (eco-industrial parks) and macro level (city, region, nation and beyond) [3]. At the meso-level, eco-industrial parks have gained increased attention in the last decades, and this is also labelled industrial symbiosis [4]. Geographic proximity and collaboration can provide and enable both economic and environmental benefits for the companies through the exchange of energy, water, materials, as well as services such as permitting transportation and marketing [4]. Several recent studies have highlighted how an energy-intensive industry located in industrial parks can gain increased energy and material efficiency as well as CO 2 -reduced emissions through such symbiosis [5][6][7][8][9].
There is a vast literature in the field of industrial parks and their energy systems. Directions in research ranges from strategies and policies for development of eco-industrial parks [10,11], studies investigating how to measure the environmental performance [12,13], and optimisation of energy systems in industrial parks [14,15]. Energy recovery in industrial parks is a topic which has received interest in the academic community and a number of systematic design techniques have been developed and applied over the years [16]. Total site heat integration (TSHI) was introduced by [17]. They extend the design and targeting procedures from Pinch Technology [18] to the concept of total sites, which constitute factories which are energetically linked through a central utility system. The methodology presented is applicable both in so-called grassroots design of new factories and in retrofit situations in existing factories and industrial sites. The methodology and relevant tools are further developed by [19], who reports results from a number of case studies where the applications of the methodology have resulted in savings in both energy (30%) and capital (10%). Matsuda et al. [20] presents results from applying area-wide pinch technology on one of the largest chemical complexes in Japan. The study demonstrated significant potential for energy savings at the site. The results were reported to have initiated several projects on energy recovery, where the total annual energy saving was estimated to around 250 GWh. Hackl et al. [21] performed a total site analysis (TSA) on the industrial cluster of chemical companies located in Stenungsund in Sweden. The report published in 2010 was meant to provide a basis for future implementation of energy system integration. The following methodology was applied: • Stream data for temperature and enthalpy flows (T Start , T end , Q) and data on utility consumption was collected.

•
Steam from by-product incineration was also considered as a source for process heat.

•
The data was presented in so-called total site profiles (TSP) and total site composite curves (TSC). This analysis provided information on the site pinch temperature and was used to identify the most attractive measures to increase heat recovery.

•
The sitewide potential for co-generation and measure for reducing the external cooling demand below ambient temperature was also analysed.
Røynet et al. [6] revisited the Stenungsund chemical industry cluster and compared the current state of an industrial park with possible future configurations focusing on increased energy integration and replacing fossil feedstock with forest-based feedstock and compared the different scenarios in a lifecycle assessment. According to [6], little attention has been given to analyse how the environmental impact can be further reduced in already existing industrial parks. An investigation of how enviromental impact can be reduced in existing parks is case-specific in its nature, and the approach used in this study will therefore use a case study as a basis.
Mo Industrial Park (MIP) at Mo i Rana is one of Norway's largest industrial parks. The park hosts ferroalloy and steel industries together with several other industrial companies. The ferroalloy and steel industry is characterised as an energy intensive industry with large demands for electrical energy in particular. A large surplus of thermal energy is generated by the process and is available over a wide range of temperature levels. MIP has for a long time recognised the surplus energy as a valuable asset for the industrial park and has, together with the industrial clients, been working ambitiously with energy recovery. Currently, MIP is reporting an annual energy recovery of 400 GWh out of an estimated potential of 900 GWh (Figure 1). Their ambition is to increase the annual energy recovery to 640 GWh by further utilising available energy streams.
The aim of the current study is to gather data on the current energy use within the park and evaluate theoretically the potential for alternative utilisation of the surplus energy in applications which may be beneficial to the current industrial clients. The industry has identified the replacement of fossil carbon reduction materials with increased use of bio-based reduction material (biocarbon) and CO 2 capture and utilization or sequestration as the applications of greatest potential. The heat integration potential with future carbon capture facilities and biocarbon production facilities will therefore be explored theoretically. The aim of the current study is to gather data on the current energy use within the park and evaluate theoretically the potential for alternative utilisation of the surplus energy in applications which may be beneficial to the current industrial clients. The industry has identified the replacement of fossil carbon reduction materials with increased use of bio-based reduction material (biocarbon) and CO2 capture and utilization or sequestration as the applications of greatest potential. The heat integration potential with future carbon capture facilities and biocarbon production facilities will therefore be explored theoretically.

Methodology
In order to assess the energy use and potential for recovery, a mapping of the current energy flows and temperature levels was required to establish a reference base case. This is a challenging task as it involves multiple industrial entities, the production can have seasonal variations with changing ambient conditions, not all states and magnitudes of process flows are measured or easily estimated, etc. In this work, the industrial entities are treated as independent units and only the known exchanges over the company battery limits were considered. The exchanges are either raw materials that represent significant energy content, utility exchanges, or products and effluents.
Data has been systematically gathered for all major entities in the industrial park in order to evaluate the current status and future potentials for energy use. This data serves as a parameter set for modelling the total energy flow and establishing a benchmark. Some prospective new processes are considered and the effect of the integration on the total energy flow is analysed.

Methodology
In order to assess the energy use and potential for recovery, a mapping of the current energy flows and temperature levels was required to establish a reference base case. This is a challenging task as it involves multiple industrial entities, the production can have seasonal variations with changing ambient conditions, not all states and magnitudes of process flows are measured or easily estimated, etc. In this work, the industrial entities are treated as independent units and only the known exchanges over the company battery limits were considered. The exchanges are either raw materials that represent significant energy content, utility exchanges, or products and effluents.
Data has been systematically gathered for all major entities in the industrial park in order to evaluate the current status and future potentials for energy use. This data serves as a parameter set for modelling the total energy flow and establishing a benchmark. Some prospective new processes are considered and the effect of the integration on the total energy flow is analysed.

Data Collection and Analysis
More than 100 different companies are located in Mo Industrial Park. The focus in this analysis is the process industry. Other actors in the park are considered relevant if they are either (1) large energy consumers or (2) interlinked with the process industry in terms of energy flows. Based on this, the actors listed in Table 1 were identified: The collected data was compiled from a combination of publicly available sources and from data provided by the respective companies. The publicly available data has been used wherever possible and complemented by calculations and approximations based on the literature. The collected energy data was confined to the geographical co-located industrial companies in MIP at the companies' battery limits, see Figure 2.  The collected data was compiled from a combination of publicly available sources and from data provided by the respective companies. The publicly available data has been used wherever possible and complemented by calculations and approximations based on the literature. The collected energy data was confined to the geographical co-located industrial companies in MIP at the companies' battery limits, see Figure 2. The annual energy input was found using official statistics for the companies which are subject to the Norwegian Environment Agency's reporting obligation and Pollutant Release and Transfer Register (PRTR) [22]. The electricity demand is given in MWh while consumption of fossil fuels, reduction agents, etc. are given in tonnes. The energy content was found by assuming calorific values for the different energy inputs as given in Table A4 in Appendix B.1.
The energy inputs were aggregated into the categories of electricity, coal/char/coke, oil products, gas and biocarbon to simplify the analysis. This energy is consumed and used in reduction processes or for heating purposes and by that converted to chemical or thermal energy. Data on the chemically bound energy and surplus heat were estimated by using open references. Energy assessments from literature were used as estimates for the local cases to allocate surplus heat to the different energy streams. Energy balances were established to reveal lacking information and ensure energy balance consistency. Inconsistencies in energy balances were assigned to energy losses, and thus close the balance. Figure 3 illustrates the energy flows across the battery limits for the selected actors (listed in Table 1) as a Sankey diagram. The main input is electricity and coal/char/coke which is used as a The annual energy input was found using official statistics for the companies which are subject to the Norwegian Environment Agency's reporting obligation and Pollutant Release and Transfer Register (PRTR) [22]. The electricity demand is given in MWh while consumption of fossil fuels, reduction agents, etc. are given in tonnes. The energy content was found by assuming calorific values for the different energy inputs as given in Table A4 in Appendix B.1.
The energy inputs were aggregated into the categories of electricity, coal/char/coke, oil products, gas and biocarbon to simplify the analysis. This energy is consumed and used in reduction processes or for heating purposes and by that converted to chemical or thermal energy. Data on the chemically bound energy and surplus heat were estimated by using open references. Energy assessments from literature were used as estimates for the local cases to allocate surplus heat to the different energy streams. Energy balances were established to reveal lacking information and ensure energy balance consistency. Inconsistencies in energy balances were assigned to energy losses, and thus close the balance. Figure 3 illustrates the energy flows across the battery limits for the selected actors (listed in Table 1) as a Sankey diagram. The main input is electricity and coal/char/coke which is used as a chemical reduction medium for metal oxides in the submerged arch furnaces (SAF) operated by Elkem and Ferroglobe. The data centre operated by BitFury is also a significant electricity consumer. In Figure 3, the most significant energy integration is carried out by: 1.
"Mo Fjernvarme" which recovers latent heat from flue gas from Elkem (125 GWh) for use as district heating 3.
In total, this sums up to 369 GWh of recovered energy. In addition to the abovementioned utilisations of surplus heat, MIP also produces electricity from turbines in the water-cooling utility systems. This is not shown in the Sankey diagram in Figure 3, as the focus in this work is the surplus heat.
While Figure 3 illustrates the size of energy flows between the various actors, it does not provide information on the possibilities for further heat integration, which among other things depends on the temperature level of the heating and cooling demands. This information can be illustrated in a so-called Grand Composite Curve (or Heat Surplus Diagram) [18,23] which shows the aggregated heating and cooling demands at each temperature level of the process streams. The Grand Composite Curve for MIP in Figure 4 shows that MIP has an energy surplus. The existence of pockets in the curve, such as in the case for temperatures less than 80 • C in the right end of the curve, indicates a possibility for heat integration between an enthalpy surplus and an enthalpy demand. Except for the low temperature end, the remainder of the curve shows no such significant pockets. There is no heating demand above the process pinch temperature (at 1200 • C). The energy available at the temperature interval 1200-600 • C is related to cooling of metal from the SAF. Although studies have been made [24], this energy is likely to be more challenging to utilise compared to the energy available in flue gas from Elkem (below 600 • C). In the Sankey diagram in Figure 3, this flue gas stream has been split into "Off gas, unavoidable heat loss" which represents the energy in the flue gas in the temperature interval from ambient to 180 • C. Due to local environmental considerations, governmental regulations are imposed on the lower limit of flue gas emissions. The energy stream to "Off gas pipe loss" (332 GWh) represents the energy content in flue gas in the temperature interval 600-180 • C which could be utilised.
Energies 2020, 13, x FOR PEER REVIEW 5 of 24 chemical reduction medium for metal oxides in the submerged arch furnaces (SAF) operated by Elkem and Ferroglobe. The data centre operated by BitFury is also a significant electricity consumer.
In total, this sums up to 369 GWh of recovered energy. In addition to the abovementioned utilisations of surplus heat, MIP also produces electricity from turbines in the water-cooling utility systems. This is not shown in the Sankey diagram in Figure 3, as the focus in this work is the surplus heat.  Table A3 in Appendix A for aggregated energy flow data across battery limits at Mo Industrial Park.
While Figure 3 illustrates the size of energy flows between the various actors, it does not provide information on the possibilities for further heat integration, which among other things depends on the temperature level of the heating and cooling demands. This information can be illustrated in a so-called Grand Composite Curve (or Heat Surplus Diagram) [18,23] which shows the aggregated  Table A3 in Appendix A for aggregated energy flow data across battery limits at Mo Industrial Park.
A significant high quality/temperature energy source available for utilisation can be identified from Figure 4. Above 200 °C, the available thermal effect is around 100 MW. Of these, about 40 MW is available as continuously produced flue gas at a temperature level of 180-600 °C. The main challenge is to identify existing or new energy consumers that can utilise the surplus energy in a sustainable and economical way.  Table A4 for detailed stream data.

Modelling
A simplistic modelling approach has been applied to describe the overall mass and energy flows between the industrial processes included in this work. The purpose has been to describe qualitatively the main design tradeoffs which are being investigated. A complete description and modelling of all industrial actors within MIP is outside the scope of this work. With respect to cost estimation, the focus has been on cost related to electricity consumption and cost related to the CO2 emissions. Other energy input costs, such as the cost for biocarbon and the coal it replaces, have been omitted in this study. Capital expenses related to proposed design changes have not been included.

Silicon and Ferrosilicon Production
In the following, an overview will be given of the Silicon (Si) and Ferrosilicon (FeSi) processes. This is based on Schei et al. [25] and Tangstad [26], and further details may be found there. Elkem Rana produces FeSi in two submerged arc furnaces with a total electrical power of above 100 MW.
The Si and FeSi processes are both performed in open and semi-closed furnaces, meaning that air is supplied to the charge top of the furnace to oxidize CO(g) and SiO(g) according to Equations (1) and (2). The electrical power is supplied through three Søderberg electrodes of typically 1.4-1.9 m in diameter.  Table A4 for detailed stream data.
A significant high quality/temperature energy source available for utilisation can be identified from Figure 4. Above 200 • C, the available thermal effect is around 100 MW. Of these, about 40 MW is available as continuously produced flue gas at a temperature level of 180-600 • C. The main challenge is to identify existing or new energy consumers that can utilise the surplus energy in a sustainable and economical way.

Modelling
A simplistic modelling approach has been applied to describe the overall mass and energy flows between the industrial processes included in this work. The purpose has been to describe qualitatively the main design tradeoffs which are being investigated. A complete description and modelling of all industrial actors within MIP is outside the scope of this work. With respect to cost estimation, the focus has been on cost related to electricity consumption and cost related to the CO 2 emissions. Other energy input costs, such as the cost for biocarbon and the coal it replaces, have been omitted in this study. Capital expenses related to proposed design changes have not been included.

Silicon and Ferrosilicon Production
In the following, an overview will be given of the Silicon (Si) and Ferrosilicon (FeSi) processes. This is based on Schei et al. [25] and Tangstad [26], and further details may be found there. Elkem Rana produces FeSi in two submerged arc furnaces with a total electrical power of above 100 MW.
The Si and FeSi processes are both performed in open and semi-closed furnaces, meaning that air is supplied to the charge top of the furnace to oxidize CO(g) and SiO(g) according to Equations (1) and (2). The electrical power is supplied through three Søderberg electrodes of typically 1.4-1.9 m in diameter.
The furnaces may be charged by batch charging, but, from an environmental point of view, it is beneficial with a semi-continuous charging [27]. This is due to an increased flue-gas temperature that destroys polycyclic aromatic hydrocarbons (PAH) and dioxins. The main components of the charge Energies 2020, 13, 4627 7 of 23 mix are quartz, carbon material (charcoal and/or coal), in some cases wood chips, and for FeSi, an iron source such as iron ore pellets. Electric energy is supplied through three electrodes that are submerged in the charge. The furnace process may roughly be divided into a hot zone and a colder zone, where the latter is the upper part of the furnace and the former is the lower part of the furnace centred around a crater where an electric arc dissipates energy. The overall reaction of the furnace may be given as in Equation (3). Equation (4) describes the reaction taking place in the hot part of the furnace, where Si(l) and SiO(g) is formed. Further, SiO(g) from the hot zone will be a loss of Si units if not either reacted with carbon (Equation (5)) or condensed (Equation (6) or (7)) in the colder zone of the furnace.
In the later years it is common at Norwegian Si and FeSi plants to have energy recovery. At Elkem Rana there is currently a boiler installed to recover heat for the district heating system at Mo i Rana, the largest based on recovered energy in Norway.
The flue gas composition and flow rate from a FeSi SAF has been described based on parameters outlined in Table 2 and represent typical average flue gas composition from a FeSi furnace. The composition will vary some based on operating conditions and charge composition. Typically, the moisture content increases during the winter due to ice and snow in the raw materials. It represents a simplistic modelling approach which can be representative for a "generic" quality of FeSi. A description of the energy flow around the furnace is given here, which will give a qualitatively correct representation. The composition of the flue gas from the charge top is specified and is based on information from the project Staged Combustion for Energy Recovery in Ferroalloy industry (SCORE) [28,29]. The flue gas flow rate is assumed to scale linearly with the furnace electrical power. The same applies for mass flows of ore, carbon and metal produced [30][31][32]. The mass and energy balances for combustion of furnace flue gas with air and recirculated flue gas are calculated using the open source physical properties library [33], which is based on the Shomate enthalpy correlations from the National Institute of Standards and Technology (NIST) Chemistry WebBook [34,35]. Recirculated flue gas is used to meet the maximum inlet temperature (750 • C) to the waste heat recovery unit (WHRU) in cases where the amount of excess combustion air (ingress air) alone would not be sufficient to cool down the flue gas before the waste heat recovery unit. The simulation of the SAF performance was based on the parameters given in Table 2. The waste heat recovery unit was simulated with design parameters as given in Table 3. For external energy clients, it was assumed that the thermal energy was taken out as saturated steam. The flue gas from the SAF was assumed to have a maximum temperature of 750 • C. The physical properties of water and steam were calculated using the open source implementation of the international-standard IAPWS-IF97 steam tables [36]. Table 3. Design parameters for waste heat recovery unit with steam turbine for electricity production.

Design Parameter Value
Flue gas inlet-outlet temperature (

Carbon Capture
Conventional amine-based absorption technology for post combustion CO 2 capture has been considered in this work. At MIP there are several potential flue gas sources which are candidates for a carbon capture facility, where either use or storage of CO 2 is possible. For absorption-based CO 2 capture, the main energy requirements are linked to thermal regeneration of the amine solvent which takes place in a stripper column typically operating at a reboiler temperature of around 120-130 • C. Once the solvent system has been selected, the thermal energy requirements are defined by the specific reboiler duty (SRD) (GJ/t CO 2 ), which is a function of the CO 2 concentration in the flue gas [37] and of the capture rate.
The specific reboiler duty for state-of-the-art absorption technology has been demonstrated by using 30 wt% mono ethanol amine (MEA) as solvent at the Norwegian test facility TCM [38]. Gorset et al. [39] reports that the specific reboiler duty of 3.8 MJ/t CO 2 captured was achieved with MEA as a solvent operating at 87% capture rate on a flue gas with 3.7 vol% CO 2 .
The influence on energy consumption and the concentration and capture rate has been investigated theoretically by [40]. They derived the following theoretical relation for the ideal work of mixing of CO 2 as a function of flue gas concentration and CO 2 capture rate: By normalising with the results reported by for MEA, we get a rough estimate for the SRD as a function capture rate and flue gas composition as follows: Figure 5 shows the estimated specific reboiler duty for a range of conditions when estimated using Equation (9). It is expected that the estimated SRD is too low for high concentrations of CO 2 . However, the qualitative trend can be expected to be correct and the accuracy to be sufficient at this conceptual level of design. Knudsen et al. [41] tested three different solvents (20 wt% MEA, CESAR1 and CESAR2) on flue gas from a coal power plant (13 vol% CO 2 ). The best solvent (CESAR1) achieved an SRD value close to the 90% capture line in Figure 5.

Biocarbon Production
The modelling framework and assumptions made regarding biocarbon production was to a large extent based on the work by Olszewski et al. [42]. They present a techno economical evaluation of biocarbon production processes, with focus on Norwegian conditions and conclude that there are economic benefits by supplying logwood rather than wood chips to the end user of biocarbon. We therefore consider production of biocarbon at the ferroalloy plant where surplus heat may be utilised for the energy intensive drying process of virgin biomass. The composition of produced biocarbon, pyrolysis gas and pyrolysis oil is estimated from empirical correlations presented in [43].

Key Performance Indicators
The key performance indicators used in this study are focusing on CO2 footprint, net electricity consumption or costs related to both factors. The key performance indicators in this study are limited to a per company level, or on the battery limits of the production facility. The performance of the whole park was not assessed.
1. CO2 footprint as kg CO2 equivalents per kg metal produced (kgCO2/kg Me) 2. Net electricity consumption as kWh(el) per kg metal produced (kWh/kg Me). Net electricity consumption is electricity input to SAF minus electricity produced in the waste heat recovery steam cycle 3. Electricity and CO2 emission costs per kg metal produced (EUR/kg Me).
The parameters related to the key performance indicators are summarized in Table 4. Table 4. Economic parameters related to key performance indicators. Base case figures are given in bold.

Parameter Range
Electricity price from grid 0.03-0.1 EUR/kWh CO2 emission price 5-30 EUR/t CO2 CO2 emissions from electricity production 25-440 g CO2-eq/kWh Figure 5. Estimated specific reboiler duty for a generic solvent as a function of CO 2 capture rate and flue gas concentration. Based on theoretical work of mixing [40] and normalised with results for 30 wt% MEA [39]. The performance of the CESAR1 solvent (90% capture rate, 13 vol% CO 2 ) [41].

Biocarbon Production
The modelling framework and assumptions made regarding biocarbon production was to a large extent based on the work by Olszewski et al. [42]. They present a techno economical evaluation of biocarbon production processes, with focus on Norwegian conditions and conclude that there are economic benefits by supplying logwood rather than wood chips to the end user of biocarbon. We therefore consider production of biocarbon at the ferroalloy plant where surplus heat may be utilised for the energy intensive drying process of virgin biomass. The composition of produced biocarbon, pyrolysis gas and pyrolysis oil is estimated from empirical correlations presented in [43].

Key Performance Indicators
The key performance indicators used in this study are focusing on CO 2 footprint, net electricity consumption or costs related to both factors. The key performance indicators in this study are limited to a per company level, or on the battery limits of the production facility. The performance of the whole park was not assessed.
Net electricity consumption as kWh(el) per kg metal produced (kWh/kg Me). Net electricity consumption is electricity input to SAF minus electricity produced in the waste heat recovery steam cycle 3.
Electricity and CO 2 emission costs per kg metal produced (EUR/kg Me).
The parameters related to the key performance indicators are summarized in Table 4. The carbon footprint is analysed on two levels or scopes. Scope 1 include the direct CO 2 emissions from the combustion of fuels from the stationary processes within the industrial park. Scope 2 emissions include the indirect CO 2 emissions related to the consumed electricity in the industrial park.
Biogenic CO 2 emissions are regarded as CO 2 neutral in the CO 2 footprint. The biogenic CO 2 emissions will not contribute towards the CO 2 emission cost, in line with the regulatory framework EU Emissions Trading System (EU ETS).
The carbon emissions related to the consumption of electricity depend on regional attributes. The regional average production mix, along with imports from neighbouring regions, will yield different carbon intensities of CO 2 /kWh depending on the regional scope. Several on-line statistics are available which can provide a snapshot of the CO 2 footprint per kWh electricity produced or consumed in a given region. The average carbon intensity for different regional scopes are Norwegian (30.5 g CO 2 -eq/kWh, low-voltage mix) [44], Nordic electricity mix (130 g CO 2 -eq/kWh) [45], and European (446 g CO 2 -eq/kWh) [44]. An approximation for the Norwegian high voltage electricity mix was used as a baseline at 25 g CO 2 -eq/kWh (the high voltage mix has a lower carbon intensity because of lower losses).
The CO 2 price in Europe has since 2008 and 2019 varied from 30 to 5 EUR/t CO 2 . The highest prices were obtained prior to the financial crisis in 2008 and again in 2019 where prices approached 30 EUR/t CO 2 . In the period after the financial crisis in 2008 and up until 2017, the CO 2 prices were 10 ± 5 EUR/t CO 2 [46]. A CO 2 price of 30 EUR/t was used as the baseline.
The price of electricity for non-household consumers varies in the range from 0.15 EUR/kWh (Germany) to about 0.07 EUR/kWh (Finland). Norway is listed with a price around 0.09 EUR/kWh [47]. These electricity prices for Norway are significantly higher than what Statistics Norway presents as electricity prices for energy intensive industry (0.3 NOK/kWh which is about 0.03 EUR/kWh) [48]. An electricity price at 0.03 EUR/kWh was used as baseline. Figure 6 shows the fraction of the total electricity price which represents the CO 2 costs as a function of CO 2 generation caused by electricity production. For the Norwegian market with a low specific CO 2 production per kWh electricity, the CO 2 costs represent a small part of the total costs. However, if a European electricity mix is considered with almost 10 times higher specific CO 2 production per kWh, the cost of CO 2 becomes a significant fraction of the overall electricity cost.
Energies 2020, 13, x FOR PEER REVIEW 10 of 24 The carbon footprint is analysed on two levels or scopes. Scope 1 include the direct CO2 emissions from the combustion of fuels from the stationary processes within the industrial park. Scope 2 emissions include the indirect CO2 emissions related to the consumed electricity in the industrial park.
Biogenic CO2 emissions are regarded as CO2 neutral in the CO2 footprint. The biogenic CO2 emissions will not contribute towards the CO2 emission cost, in line with the regulatory framework EU Emissions Trading System (EU ETS).
The carbon emissions related to the consumption of electricity depend on regional attributes. The regional average production mix, along with imports from neighbouring regions, will yield different carbon intensities of CO2/kWh depending on the regional scope. Several on-line statistics are available which can provide a snapshot of the CO2 footprint per kWh electricity produced or consumed in a given region. The average carbon intensity for different regional scopes are Norwegian (30.5 g CO2-eq/kWh, low-voltage mix) [44], Nordic electricity mix (130 g CO2-eq/kWh) [45], and European (446 g CO2-eq/kWh) [44]. An approximation for the Norwegian high voltage electricity mix was used as a baseline at 25 g CO2-eq/kWh (the high voltage mix has a lower carbon intensity because of lower losses).
The CO2 price in Europe has since 2008 and 2019 varied from 30 to 5 EUR/t CO2. The highest prices were obtained prior to the financial crisis in 2008 and again in 2019 where prices approached 30 EUR/t CO2. In the period after the financial crisis in 2008 and up until 2017, the CO2 prices were 10 5 EUR/t CO2 [46]. A CO2 price of 30 EUR/t was used as the baseline. The price of electricity for non-household consumers varies in the range from 0.15 EUR/kWh (Germany) to about 0.07 EUR/kWh (Finland). Norway is listed with a price around 0.09 EUR/kWh [47]. These electricity prices for Norway are significantly higher than what Statistics Norway presents as electricity prices for energy intensive industry (0.3 NOK/kWh which is about 0.03 EUR/kWh) [48]. An electricity price at 0.03 EUR/kWh was used as baseline. Figure 6 shows the fraction of the total electricity price which represents the CO2 costs as a function of CO2 generation caused by electricity production. For the Norwegian market with a low specific CO2 production per kWh electricity, the CO2 costs represent a small part of the total costs. However, if a European electricity mix is considered with almost 10 times higher specific CO2 production per kWh, the cost of CO2 becomes a significant fraction of the overall electricity cost. Comparisons made on an energy basis have received criticism in that an energy quantity does not reflect the energy potential or thermodynamic "value" of a material [49]. For electricity where losses for voltage (potential) transformations is for the most case negligible, the electrical energy can Comparisons made on an energy basis have received criticism in that an energy quantity does not reflect the energy potential or thermodynamic "value" of a material [49]. For electricity where losses for voltage (potential) transformations is for the most case negligible, the electrical energy can be used directly in comparisons, irrespective of voltage. To quantify energy for mass flows in heat exchange or involving heat of reactions and phase transformations, some reference must be made to the thermal, mechanical and chemical potential. A simplified analysis based on ambient reference can be made using the concept of exergy or potential useful work extraction [50]. The exergy accounts for the difference in value for work production of a given energy quantity available for two different temperatures. Still, this analysis suffers from the same problem of attributing a per unit product value when having multiple products and quantities.
For ease of comparison and for comprehensive communication of results, this work has focused on comparing the effect on the change in net electricity consumption when including CO 2 capture and biocarbon production, while attributing a quota cost for the emission of CO 2 .

Potential New Energy Clients
The energy analysis for Mo Industrial Park shows that there is a significant energy surplus within the park. In the temperature range of 600-200 • C, surplus energy is accessible in the form of hot flue gas from the Elkem production. The default utilisation of this surplus heat would be waste of heat recovery with electricity production, which is currently installed at several ferroalloy production sites in Norway. However, electricity production gives an inherently low utilisation factor (typically <30% of the thermal energy is converted to electricity) and has high capital costs associated with waste heat recovery boilers for super-heated steam production and high-pressure steam turbines. Figure 7 illustrates a number of alternative processes which may utilise surplus thermal heat directly as saturated steam from a more inexpensive waste heat recovery system. be used directly in comparisons, irrespective of voltage. To quantify energy for mass flows in heat exchange or involving heat of reactions and phase transformations, some reference must be made to the thermal, mechanical and chemical potential. A simplified analysis based on ambient reference can be made using the concept of exergy or potential useful work extraction [50]. The exergy accounts for the difference in value for work production of a given energy quantity available for two different temperatures. Still, this analysis suffers from the same problem of attributing a per unit product value when having multiple products and quantities. For ease of comparison and for comprehensive communication of results, this work has focused on comparing the effect on the change in net electricity consumption when including CO2 capture and biocarbon production, while attributing a quota cost for the emission of CO2.

Potential New Energy Clients
The energy analysis for Mo Industrial Park shows that there is a significant energy surplus within the park. In the temperature range of 600-200 °C, surplus energy is accessible in the form of hot flue gas from the Elkem production. The default utilisation of this surplus heat would be waste of heat recovery with electricity production, which is currently installed at several ferroalloy production sites in Norway. However, electricity production gives an inherently low utilisation factor (typically <30% of the thermal energy is converted to electricity) and has high capital costs associated with waste heat recovery boilers for super-heated steam production and high-pressure steam turbines. Figure 7 illustrates a number of alternative processes which may utilise surplus thermal heat directly as saturated steam from a more inexpensive waste heat recovery system. In the following theoretical evaluations, we have focused on two energy demanding production activities which are closely related to improving the sustainability of ferroalloy production processes by reducing the carbon footprint in the processes. Both biocarbon production and post combustion carbon capture represents value chains which are common to both ferro alloy producers at MIP. Both production processes are energy intensive operations which are able to utilise surplus heat in the temperature range around 200 °C. The energy consumption for carbon capture is linked to the CO2 concentration in the flue gas. A simple evaluation of the effect of using recirculated flue gas for flue gas temperature control is therefore included. In the following theoretical evaluations, we have focused on two energy demanding production activities which are closely related to improving the sustainability of ferroalloy production processes by reducing the carbon footprint in the processes. Both biocarbon production and post combustion carbon capture represents value chains which are common to both ferro alloy producers at MIP. Both production processes are energy intensive operations which are able to utilise surplus heat in the temperature range around 200 • C. The energy consumption for carbon capture is linked to the CO 2 Energies 2020, 13, 4627 12 of 23 concentration in the flue gas. A simple evaluation of the effect of using recirculated flue gas for flue gas temperature control is therefore included.
In terms of robustness and energy integration between the two ferroalloy producers at MIP, local biocarbon production and carbon capture makes sense as the value chains are interconnected.

Ferrosilicon Production
The base case for ferrosilicon production was chosen as a 40 MW furnace with a waste heat recovery unit for electricity production. The furnace performance was estimated based on the parameters outlined in Table 2.

Semi-Closed Furnace with Recirculated Flue Gas
Currently, most silicon and ferrosillicon production in Norway is carried out using semi-closed furnace technology, where ingress air is used for controlled combustion of the charge gas being generated in the SAF. In order to maintain a stable flue gas temperature (650-750 • C) towards the waste heat recovery boiler, a high degree of ingress air is required. This leads to a dilute flue gas with high O 2 surplus and low CO 2 concentration. The large amounts of ingress air also lead to significant stack loss, since ingress air is heated from ambient temperature to the stack temperature of around 150 • C. Thus, the semi-closed furnace technology limits the amount of energy which can be recovered from the flue gas.
The ferrosilicon industry has been working on establishing a semi-closed furnace technology where the combustion of the charge flue gas takes place at a lower oxygen concentration. The recirculated flue gas (from downstream the waste heat recovery system) is used for temperature control in the flue gas entering the waste heat recovery boiler. A semi-closed furnace technology with recirculated flue gas will enable recovering more energy in addition to the possibility of avoiding peak temperatures in the combustion zone. Avoiding peak temperatures will also reduce the NO x formation from the process (Wittgens et al. [28]). Figure 8a illustrates a base case scenario for a generic ferrosilicon production, with a SAF integrated with a waste heat recovery boiler producing super-heated steam for electricity production in a Rankin steam cycle. The process flow diagram indicates how recirculated flue gas may be used for temperature control in the combustion zone and upstream of the waste heat recovery boiler. In a semi-closed furnace, the amount of ingress air can be limited to what is required for complete combustion of the flue gas. For such cases, recirculated flue gas can be used to control the flue gas temperature at the inlet of the waste heat recovery boiler below the target 750 • C. Figure 8b illustrates how reduced ingress air (reduced O 2 content in the flue gas) affects both the stack loss and leads to reduced net electricity consumption (5% reduction) as more thermal energy is available for steam production. The stack loss is reduced from close to 7 MW down to about 1 MW when ingress air for cooling the flue gas is replaced by recirculated flue gas. The carbon dioxide concentration in the flue gas increases from 4 to 14 vol% as we move towards a semi-closed furnace with recirculation of flue gas operating at 7 vol% O 2 . An increase in CO 2 concentration will have a positive impact on the energy consumption in any downstream CO 2 capture processes.

Biocarbon Production
Replacement of fossil carbon (typically coal, char or coke) with more CO 2 -neutral biocarbon (charcoal and some woodchips) has been an important research area over the last decades [42,51]. Local production of biocarbon from virgin biomass offers several obvious integration opportunities in terms of energy recovery. The main motivation for introducing biocarbon is to reduce the carbon footprint of the silicon production. The drying process of the biomass prior to pyrolysis/biocarbon production can operate using waste heat from the ferrosilicon production process and other surplus heat sources within MIP. Thus, enabling efficient use of surplus energy with a less capital intensive and more thermodynamically efficient process than electricity production. Although we focus here on the ferrosilicon production, the use of biocarbon is also relevant for the ferromanganese production in Mo industrial park.

Biocarbon Production
Replacement of fossil carbon (typically coal, char or coke) with more CO2-neutral biocarbon (charcoal and some woodchips) has been an important research area over the last decades [42,51]. Local production of biocarbon from virgin biomass offers several obvious integration opportunities in terms of energy recovery. The main motivation for introducing biocarbon is to reduce the carbon footprint of the silicon production. The drying process of the biomass prior to pyrolysis/biocarbon production can operate using waste heat from the ferrosilicon production process and other surplus heat sources within MIP. Thus, enabling efficient use of surplus energy with a less capital intensive and more thermodynamically efficient process than electricity production. Although we focus here on the ferrosilicon production, the use of biocarbon is also relevant for the ferromanganese production in Mo industrial park. Figure 9a shows the base case ferrosilicon process integrated with biocarbon production. Steam from the waste heat recovery boiler is used for the drying of virgin biomass prior to a pyrolysis process for production of biocarbon. By-products from the biocarbon production process are pyrolysis oils and pyrolysis gas. The latter is assumed to be used directly as a heat source and added as hot (combusted) flue gas to the waste heat recovery boiler. The pyrolysis oil has not been included in the energy or CO2 balance. One may envisage upgrading the pyrolysis oils to more valuable chemicals or fuels as indicated in Figure 7. However, a pragmatic approach has recently been proposed [52] where the pyrolysis oil may be used as binder for the biocarbon material, in which case it would enter the SAF together with the biocarbon and contribute to the waste heat energy production. Figure 9b shows the impact of the biocarbon production (given as the bio to fossil carbon ratio) on electricity production, thermal energy consumption for virgin biomass drying and the thermal energy contribution from pyrolysis gas. The increase in net electricity consumption is moderate ( 5%) at 70% bio to fossil shares. This is due to the thermal contribution from burning pyrolysis gas generated in the biocarbon production. This makes up for more than 50% of the energy consumed by the drying of virgin biomass.  Figure 9a shows the base case ferrosilicon process integrated with biocarbon production. Steam from the waste heat recovery boiler is used for the drying of virgin biomass prior to a pyrolysis process for production of biocarbon. By-products from the biocarbon production process are pyrolysis oils and pyrolysis gas. The latter is assumed to be used directly as a heat source and added as hot (combusted) flue gas to the waste heat recovery boiler. The pyrolysis oil has not been included in the energy or CO 2 balance. One may envisage upgrading the pyrolysis oils to more valuable chemicals or fuels as indicated in Figure 7. However, a pragmatic approach has recently been proposed [52] where the pyrolysis oil may be used as binder for the biocarbon material, in which case it would enter the SAF together with the biocarbon and contribute to the waste heat energy production.   Tables 2 and 3 for design parameters.

CO2 Capture
Carbon capture and storage (CCS) is an alternative or supplement to biocarbon production in order for the ferroalloy production to reduce its carbon footprint. Conventional CO2 capture is carried out using absorption technology where the regeneration of the solvent (usually amine based) takes place in a stripping column which is operated on thermal energy at around 200 °C. This energy demand is well suited for integration with the steam production in a waste heat recovery unit. Figure 10a shows the base case FeSi process integrated with an absorption-based CO2 capture processing unit. The solvent regeneration in the stripper column is operated using steam from the waste heat recovery unit. As with all CO2 capture technologies, the specific energy demand is linked to the CO2 concentration in the flue gas through the ideal work of separation (see Figure 5 and [40]).  Figure 9b shows the impact of the biocarbon production (given as the bio to fossil carbon ratio) on electricity production, thermal energy consumption for virgin biomass drying and the thermal energy contribution from pyrolysis gas. The increase in net electricity consumption is moderate (≈5%) at 70% bio to fossil shares. This is due to the thermal contribution from burning pyrolysis gas generated in the biocarbon production. This makes up for more than 50% of the energy consumed by the drying of virgin biomass.

CO 2 Capture
Carbon capture and storage (CCS) is an alternative or supplement to biocarbon production in order for the ferroalloy production to reduce its carbon footprint. Conventional CO 2 capture is carried out using absorption technology where the regeneration of the solvent (usually amine based) takes place in a stripping column which is operated on thermal energy at around 200 • C. This energy demand is well suited for integration with the steam production in a waste heat recovery unit. Figure 10a shows the base case FeSi process integrated with an absorption-based CO 2 capture processing unit. The solvent regeneration in the stripper column is operated using steam from the waste heat recovery unit. As with all CO 2 capture technologies, the specific energy demand is linked to the CO 2 concentration in the flue gas through the ideal work of separation (see Figure 5 and [40]). In general, the specific energy demand (MJ/kg CO 2 captured) is reduced with increasing CO 2 concentration. As already demonstrated for the base case (see Figure 8b), a semi-closed SAF will deliver a flue gas with significantly higher concentration (12-14 vol% CO 2 ) as opposed to around 4 vol% CO 2 in a SAF operating with excess ingress air for temperature control. It is therefore relevant to investigate the impact of both CO 2 capture rate and the oxygen concentration in the flue gas leaving the SAF flue gas combustion zone.

Tradeoffs between Biocarbon Production and CO2 Capture
Exploring the integration between waste heat recovery for electricity production, biocarbon production and CO2 capture is illustrated in Figure 11a. In the parametric studies illustrated in Figure  11b-d, the key performance indicators are presented as contour lines with biocarbon fraction as the abscissa and CO2 capture rate as the ordinate. Figure 11b shows that carbon neutral metal production can be achieved with a biocarbon fraction above 55% and 90% capture rate. The associated increased net electricity demand is a 12% increase compared to the lower left corner of Figure 11c. In terms of the estimated electricity and CO2 related operating costs in Figure 11d, there is an apparent 40% reduction when moving towards the carbon neutral upper right corner. It is important to notice that the cost figures do not include capital costs associated with any of the proposed measures for increased energy recovery through electricity production, carbon capture or biocarbon production. The net electricity consumption in Figure 10b is affected by both the capture rate and the oxygen concentration in the flue gas leaving the SAF combustion zone. This is because the energy demand in the CO 2 capture unit is linked to the CO 2 concentration in the flue gas (see Figure 5). The higher the CO 2 concentration in the flue gas, the less energy has to be provided by the waste heat recovery unit and more electricity can be produced. This is accomplished using a semi-closed SAF operating with low surplus oxygen content and therefore a high CO 2 concentration in the flue gas. According to Figure 10b, a semi-closed furnace with recirculated flue gas for temperature control can achieve almost the same net electricity consumption with 90% carbon capture as a semi-closed furnace operating with ingress air for temperature control and without carbon capture.

Tradeoffs between Biocarbon Production and CO 2 Capture
Exploring the integration between waste heat recovery for electricity production, biocarbon production and CO 2 capture is illustrated in Figure 11a. In the parametric studies illustrated in Figure 11b-d, the key performance indicators are presented as contour lines with biocarbon fraction as the abscissa and CO 2 capture rate as the ordinate. Figure 11b shows that carbon neutral metal production can be achieved with a biocarbon fraction above 55% and 90% capture rate. The associated increased net electricity demand is a 12% increase compared to the lower left corner of Figure 11c.
In terms of the estimated electricity and CO 2 related operating costs in Figure 11d, there is an apparent 40% reduction when moving towards the carbon neutral upper right corner. It is important to notice that the cost figures do not include capital costs associated with any of the proposed measures for increased energy recovery through electricity production, carbon capture or biocarbon production.   Tables 2 and  3 for design parameters.

Conclusions
The potential for increased energy utilisation and reduced carbon footprint at the Norwegian industry cluster Mo Industri Park has been investigated. Process data on energy flows between the industrial clients in the park show that there is a significant potential for increasing the annual energy recovery within the park. The goal formulated by MIP of increasing the current annual energy recovery of 400 GWh to up to 640 GWh seems technically realistic. This can be accomplished by introducing carbon emission mitigating technologies like carbon capture from flue gas sources within the park and local biocarbon production as a replacement for fossil fuel-based reduction materials in the ferrosilicon production. Both of these measures can, together with conventional electricity production from waste heat recovery units, meet the stated target of energy recovery and at the same time provide close to carbon neutral ferrosilicon production.
The calculations indicate that a semi-closed SAF with recirculated flue gas for temperature control will achieve a 5% lower net electricity consumption compared to a similar SAF using ingress air for temperature control. Further, implementation of recirculation of flue gas will increase the CO2 concentration and, as a result, decrease the energy demand related to carbon capture. A further improvement of the energy recovery may be possible if the waste recovery system would be rebuilt to handle higher temperatures, however, this may increase scaling and material wear.
The viability of the biocarbon production will be dependent on the market conditions and the biocarbon price. Establishment of the energy cost for biocarbon from a possible future local biocarbon production should be evaluated in future works to get a more complete picture of the energy input costs. Figure 11. FeSi production with integrated CO 2 capture and biocarbon production. See Tables 2 and 3 for design parameters.

Conclusions
The potential for increased energy utilisation and reduced carbon footprint at the Norwegian industry cluster Mo Industri Park has been investigated. Process data on energy flows between the industrial clients in the park show that there is a significant potential for increasing the annual energy recovery within the park. The goal formulated by MIP of increasing the current annual energy recovery of 400 GWh to up to 640 GWh seems technically realistic. This can be accomplished by introducing carbon emission mitigating technologies like carbon capture from flue gas sources within the park and local biocarbon production as a replacement for fossil fuel-based reduction materials in the ferrosilicon production. Both of these measures can, together with conventional electricity production from waste heat recovery units, meet the stated target of energy recovery and at the same time provide close to carbon neutral ferrosilicon production.
The calculations indicate that a semi-closed SAF with recirculated flue gas for temperature control will achieve a 5% lower net electricity consumption compared to a similar SAF using ingress air for temperature control. Further, implementation of recirculation of flue gas will increase the CO 2 concentration and, as a result, decrease the energy demand related to carbon capture. A further improvement of the energy recovery may be possible if the waste recovery system would be rebuilt to handle higher temperatures, however, this may increase scaling and material wear.
The viability of the biocarbon production will be dependent on the market conditions and the biocarbon price. Establishment of the energy cost for biocarbon from a possible future local biocarbon production should be evaluated in future works to get a more complete picture of the energy input costs.
The current work does not consider that capital costs of implementing the proposed measures and is focused primarily on the ferrosilicon production within the park. This is because the major surplus heat source was associated with flue gas from the ferrosilicon production. However, the proposed measures for energy recovery and carbon emission mitigation are also applicable to the ferromanganese production and to the steel production. In order to carry out a more extensive optimisation of surplus heat utilisation within the park, the utility systems for energy distribution should be considered. Funding: This publication has been funded by HighEFF-Centre for an Energy Efficient and Competitive Industry for the Future, an eight-year research centre under the FME-scheme (Centre for Environment-friendly Energy Research, 257632/E20). The authors gratefully acknowledge the financial support from the Research Council of Norway and user partners of HighEFF.

Conflicts of Interest:
The authors declare no conflict of interest.

Abbreviations
The following abbreviations are used in this manuscript:  Tables A1 and A2 show stream data which has been the basis for the grand composite curves for current base case at MIP and for a revised situation with biocarbon production and carbon capture integrated at MIP. Table A1. Base case stream data as basis for Grand Composite curve in Figure 4.

Process Stream
Tstart ( • C) Tend ( • C) kW/C MW GWh   Table A3. Aggregated data on energy flows between companies within MIP. This table is the basis for construction of Sankey diagrams in Figure 3 (excluding the bottom three rows).