Not all sites are created equal – Exploring the impact of constraints to suitable biogas plant locations in Sweden

Biogas production from manure is attractive to support plans towards a circular economy as it allows for renewable energy production and nutrient recycling in agriculture. Finding optimal locations for biogas plants


Biogas as a response to multiple sustainability challenges
Current food and energy systems are facing inevitable shifts in order to ensure sustainability and security at global and local scales (Jurgilevich et al., 2016;Songstad et al., 2014).The planetary boundaries regarding climate change, and nitrogen (N) and phosphorus (P) cycles have been severely transgressed by human interference, and this influence has threatened the health of ecosystems and safe food production systems (Rockstrom et al., 2009).Tackling these issues will require major changes in efficiency, but also in the source material we use to produce energy and fertilizers.In fact, the European Commission action plan towards a circular economy argues for increased efforts to treat organic waste materials as resources (European Commission, 2015).Organic materials, such as household waste, manure, and sewage sludge, contain large amounts of N, P and potassium (K) which can be used to fertilize agricultural soils (Akram et al., 2019a); at the same time they are also appropriate substrates for anaerobic digestion in the biogas production process and as such can be a green energy source (Da Costa Gomez, 2013;Ma et al., 2018).Biogas plants may focus on sustainable energy production, but they can also facilitate the reuse of nutrients on to farmland as digestion increases N plant availability (Insam et al., 2015).The digestion process also conserves the organic matter, P, and K in the digestate which all contribute to healthy soils and can replace some of the need for mineral fertilizers (Jones et al., 2013;Tampio et al., 2017).In other words, biogas solutions can have a central role to play in a circular food and energy economy.
Sweden was an early adopter of biogas solutions and plans to expand production to meet various sustainability goals.Biogas through anaerobic digestion has been produced since the 1960s at sewage treatment plants (Svenska biogasföreningen 2004) and since the 1990s biogas has been increasingly used as transportation fuel.In 2020, about half of the produced biogas in Sweden (which was about 2.2 TWh) was from co-digestion plants (36 plants) followed by 33% in sewage treatment facilities (134 plants).About two-thirds of the produced biogas was upgraded and mostly used as transportation fuel (Swedish Gas Association, 2021).Although manure currently only represents 10% (about 0.2 TWh, ibid.) of the substrates use in biogas production (sludge and food waste represent the bulk of production), its use is expected to significantly increase by 2030.
Sweden's 2030 suggested goal for biogas production from manure is between 1.5 and 2.6 TWh (Norstedts Juridiks, 2019).By 2030 Sweden would like the transportation sector to be powered by fossil-free fuels, and by 2050 achieve zero net greenhouse gas (GHG) emissions (SOU, 2013).Biogas is a renewable energy carrier as methane and carbon dioxide (about 60 %vol.methane).It can be used to create electricity and/or heat, or-after upgrading to almost pure biomethane-as fuel for transportation or in industrial applications (Yang et al., 2014).It is becoming increasingly common that large production facilities both upgrade and liquify their biogas and transport it via liquid biogas trucks to decrease the distribution costs and avoids the need for a gas grid (Gustafsson et al., 2020).At the same time, careful management of the nutrients in digestate remains essential to avoid contributing to eutrophicationespecially as the Baltic Sea Action plan goals via eutrophication have not been met (HELCOM, 2018) and Sweden has set a zero eutrophication goal for both fresh and coastal water (Ek, 2013).

Logistical complexities of scaling up biogas solutions
To meet Sweden's biogas production goals, more biogas plants will be built but determining exactly where remains an important logistical question.Agricultural specialization and urbanization, in Sweden but also globally, have segregated manure (and the nutrients and energy it contains) from the crop dominated farms that can use more nutrients.This is true globally (Jones et al., 2013), but also within countries, for instance in Sweden (Pettersson et al., 2018) and France (Nesme et al., 2015).Organic wastes with high water content are bulky and have a low nutrient-to-mass ratio in comparison with more concentrated mineral fertilizers that are cheaper to transport per nutrient mass.The high transportation costs of organic wastes, such as liquid manure, is a significant hindrance for straightforward, efficient, and profitable nutrient recycling (Keplinger and Hauck, 2006), and of particular concern for biogas plant companies (Johansson and Nilsson, 2007;Ljungberg et al., 2013).Transport of both feedstock to biogas plants and digestate back to farmland is a large cost to biogas plants.Finding a location that can minimize transport costs is thus an important consideration.One way to tackle both high transport costs, and the potential for nutrient over-application, is to use optimization modeling to better plan how to link areas of organic waste production to areas of crop nutrient demand.Previous work in Sweden has demonstrated that indeed, investing more in nutrient recycling logistics could make Sweden less reliant of mineral fertilizers, but that transport costs may be prohibitively high given current fertilizer and transport prices (Akram et al., 2019a(Akram et al., , 2019b)).However, adding biogas plants to process manure and then redistribute the nutrients to farmland as digestate could change that equation (Metson et al., 2020).The additional benefits of energy recovery can not only add to financial profits, but contribute to greenhouse gas savings (Metson et al., 2020).
Still, a great many factors can play a role in determining the location of a new biogas plant.Issues related to the supply of biomass and demand for products and byproducts are usually considered, which imply the importance of transportation and logistics (Hiloidhari et al., 2017).Furthermore, social, economic, and institutional requirements must be taken into consideration.These factors tend to be more variable across space and time (e.g., in the Upper Rhine Region of Europe (Schumacher and Schultmann, 2017)), and can be more difficult to integrate into spatial modeling approaches.Still, the balance between incentives (e.g., tax breaks or strong support politically to get zoning and permitting done) and barriers (e.g., local resident protests) could be equality, or perhaps more, decisive that the supply and demand of materials (Jesus et al., 2021).Analyses that try to find appropriate locations given a set of desirable (or undesirable characteristics) are referred to as a "Facility Location Problem" (Tagliabue et al., 2021) or a "Site Suitability Analysis" (Akther et al., 2019;Kurka et al., 2012).In both cases, one must identify and understand what factors should be considered for site-selection of a new biogas plant in a specific context.A second step is to 'translate' such factors in to quantifiable and mappable variables (e. g., avoiding land zoned as housing).National (e.g., Italy (Tagliabue et al., 2021)) or multi-nation assessments tend to consider less factors (i.e., above supply and demand); those that consider more aspects of site-location are often focused on a smaller geographical region (e.g., region of Scotland (Kurka et al., 2012)) and/or a specific type of substrate (e.g., dairy farms in a region of the United States (Thompson et al., 2013)).
For Sweden, Feiz et al. (2022) have identified 12 key factors: four factors related to supply and demand (feedstock supply, biogas demand, digestate demand, and carbon dioxide demand); two factors related to infrastructure and synergies (available infrastructure, adjacent existing industries); three factors related to land-use and zoning (nearby housing, zoning, and historic preservation sites); and finally three factors related to the socio-political context (political strategies and goals, organizational capability, and local social acceptance).However, identification of these factors may not be practically sufficient.In certain cases, where there are many possible locations to choose from, the best possible outcome can be selected through geo-spatial and mathematical approaches which is the focus of the work described here.

Moving from scenarios to actionessential in sustainability research
The call for research on sustainability challenges to work across disciplines, and with stakeholders outside academia, is now relatively long-standing (Kates et al., 2001) but its practice continues to evolve and be at the forefront of how many of these challenges are researched (Fam et al., 2016;Norström et al., 2020).Involving stakeholders in the conception of research questions, the design of methods, and interpretation of results is viewed as a tangible way to create real-world desirable change.Academic research has developed many tools (from technologies, to datasets, to governance mechanisms) to tackle societal problems but if they are not adopted in action then there is likely a lack of nuanced knowledge and capacity surrounding the issue.Although co-production methods are diverse, and such research presents many challenges, there is a set of shared principles (Lang et al., 2012;Norström et al., 2020).These include a commitment to being specific about the context of an issue, support capacity building through active and iterative engagement, integrate different knowledge types, and finally setting clear shared goals.With regards to biogas development, which can be linked to larger sustainability issues around climate change, circular economy, and eutrophication, there is a need to continue to connect research and practice in meaningful ways (Lybaek and Kjaer, 2022).Although Sweden, and many other countries, have expressed support to increase biogas production, maximizing benefits, overcoming barriers, and accelerating change requires cooperation among actors, including with researchers.
In Metson et al. (2020)'s optimization model, only substrate supply and digestate demand were used to find potential locations for biogas plantsnot including the majority of the factors that are known affect location/site selection in Sweden (Feiz et al., 2022).Without a way to better account for how other land-uses and infrastructure availability factors may constrain the selection of locations for new biogas plants, it will be difficult to adapt optimization model results to be used in real-world contexts.One of the challenges is that specific factors, and the priority of those factors, in selecting a location for a biogas plant project can vary widely (depending on, for example the type of actors involved and the sociopolitical environment of a region).Similarly, there can be multiple planning processes for biogas plants at the same time which may cause sub-optimal plant location selection for either project if they are not aware of each other.The above-mentioned 12 factors identified by Feiz et al. (2022), although they provide a comprehensive overview of the types of considerations which influence location selection in Sweden, do not provide no specific value ranges for any one factor (as these vary by project) which can be made in to spatially-explicit datasets for systematic use.Operationalizing such localization factors into an optimization model requires either a case study approach and/or a flexible framework.One way to conceptualize many of these localization factors is as additional feasibility constraints on where an optimization model can decide to locate a plant.The work presented here is a step in that direction.

Study objectives
In summary, there is clearly a large potential for increased biogas production in Sweden to meet diverse sustainability goals, but work remains to be done to bridge the gap between theoretical potential and real-world planning.In this paper we ask: 1. Which known localization factors, in addition to substrate supply and digestate demand, most constrain potential biogas plant locations nationally and locally?
2. How does the introduction of more than one biogas plant, which compete for substrate and digestate markets, affect the optimal location of biogas plants?
In order to explore these questions, we use a real-world case study in Southern Sweden where there is interest in building new biogas plants using manure as a substrate.These research questions, and the case study, contribute to a larger methodological aim for our work: Create a robust and reproducible method to combine optimization modeling with location/site suitability processes which are usually more qualitative and case specific.

Methods
We developed a four-stage methodology (Fig. 1) to answer both research questions.As a point of departure, we used the 12 identified factors that affect the site-location of biogas plants in Sweden (localization factors in Fig. 1, and Table 1).To answer the first research question, out of these factors we selected those that are clearly relevant from a spatial perspective and translated them into spatially-based constraints and found appropriate datasets that contained relevant information for those constraints (e.g. for the constrain regarding nearby housing we used a dataset for spatial distribution of residential areas in Sweden).Since we wanted to investigate the site-location of a real biogas plant, we collaborated with a biogas company on a specific case study to get an in-depth insight of the requirements when planning a new biogas plant and define and parameterize our suitability constraints.Due to their past development and relationships with other actors, this company had already decided to evaluate build a new biogas plant in particular municipalities (see section 1.2) and we wanted to use their case for a detailed analysis.Once the constraints were well defined (section 2.2), possible locations fulfilling the constraints could be identified in the case study area, in the Kristianstad municipality, and nationally (section 2.3.1).Then, it was possible to determine the most limiting constraint at different geographical levels (section 3.3.2).These G.S. Metson et al. levels were defined as: • national: the whole area of Sweden; • regional: a cluster of municipalities which makes up an area of the country; and • local: within on municipality or a few adjacent municipalities).
To answer the second question, we used a slightly modified version of the optimization model used in (Metson et al., 2020) to find the most suitable location(s) in the case study area and Kristianstad, which correspond to minimizing transportation costs for substrate and digestate (section 2.3.3).
We used geographical data from publicly available Swedish national databases which were then processed with the open source geographical information system software QGIS, including their Geographic Resources Analysis Support System (GRASS) plugin and PyQGIS plugin to (QGIS.org,2020).

Case study area
Skåne, the southernmost county in Sweden (Fig. 2), is an agriculturally intensive region of Sweden (Bårström et al., 2015) which has experienced an urbanization with more densely-built areas (SCB, 2018b(SCB, , 2017a)).Due to the considerable amount of livestock in the region (SCB, 2018a), Skåne also has an abundant supply of manure, and consequently a large biogas potential of approximately 450 GWh/year (Björnsson et al., 2011).In 2016, Skåne had 35 biogas plants, four of which produce more than 10 GWh biogas per year (Swedish Energy Agency, 2017).Skåne county's vision for 2030 is to be the leading biogas region in Europe (Skåne Region, 2015).As such biogas companies are eager to invest in new plants in the region.Here we collaborated with a biogas company that was interested in building a new manure-based co-digestion plant in either Sjöbo or Tomelilla municipalities within Skåne.
Like with Skåne as a whole, our case study municipalities have a large amount of manure, but it is not evenly distributed.Sjöbo and Tomelilla could produce 44 000 and 31 700 MWh/year, respectively, if all manure was used in biogas plants (Björnsson et al., 2011).Together the municipalities have an overall surplus of P and K and a deficit of N, when comparing crop nutrient needs and the amount of these nutrients in manure.Over 60% of the land area of these municipalities is agricultural land-mostly arable land cultivated for cereals (SCB, 2017b).Tomelilla has mostly clay till and till derived soils, and Sjöbo has slightly more glaciofluvial sediments.These municipalities are located in an area with high soil P concentrations and a neutral pH (Paulsson et al., 2015).Importantly, there is variation in the supply and demand of nutrients within the municipal boundaries (Fig. 3); transporting manure and processing it through biogas plants could help redistribute nutrients where they are needed (Metson et al., 2020).Neither municipality have biogas plants for the anaerobic digestion of manure, only wastewater treatment plants for digesting sewage sludge (County Administrative Board of Skåne, 2020).

Table 1
Description of factors which influence the site selection of biogas plants-excluding socio-political factors such as political strategies, organizational capacity, and local acceptance which are further described in Feiz et al. (2022) as they interact with already listed factors (e.g., avoiding near-by housing increases social acceptance, zoning can be changed to reflect political goals, or the supply of substrate us facilitated by farmer organizational capacity).

Factor Description
Feedstock (substrate) supply The distance to suppliers affects the transportation costs of the substrate.

Biogas demand
The distribution cost is affected by the distance to the biogas users.

Digestate demand
The distance to digestate users affects the transportation costs of digestate.CO 2 demand Would be beneficial, provided that it is possible to capture and use the CO 2 from biogas upgrading.Nearby housing Biogas plants cannot be placed too close to housing, due to odors and usage of roads adjacent to residential areas.

Zoning
Local governments divide areas of land into zones dedicated for different purposes.Zones intended for industries often have fewer restrictions, but other zones may prohibit biogas plants.

Adjacent existing industries
The distance to other industries can affect the chances of obtaining permits and the possibility of establishing beneficial collaborations.Available infrastructure Access to infrastructure.For instance, access to roads affects the ability to handle transports to and from the biogas plant.

Historic preservation sites
If objects found on the site might have historical value, an investigation must be executed.This investigation is both expensive and time-consuming.The biogas company we collaborated with (referred to as 'the biogas company') is a Nordic energy company with a specialization in the gas sector.They want to examine the possibility of building a new biogas plant that will produce liquid biogas on-site.They plan to do this in collaboration with another local biogas company (referred to as 'the local partner biogas company') which has an established network of farmers that could act as suppliers of manure and recipients of the digestate as a bio-fertilizer.Additionally, Kristianstad municipality has been on the map as a possible area for another plant, as it also has an abundant supply of manure.However, the implementation of a plant there is less certain.Kristianstad already has three biogas plants; one for co-digestion (manure, and municipal and industrial organic waste), one for landfill and one for digesting wastewater sludge (County Administrative Board of Skåne, 2020).

Defining constraints
Our previous work has already identified a set of 12 localization factors that are important for biogas companies in Sweden when selecting plant locations (Table 1 (Feiz et al., 2022),).As mentioned previously, these factors were grouped into those related to supply and demand, infrastructure and synergies, land-use, and socio-political enablers and barriers.In order to narrow the localization factors considered here, we combined insights from 1) factors brought up by the biogas company to select a site for the case study area and 2) standardized national data availability.
The biogas company specified that the location for a biogas plant should consider 10 factors which aligned well with the types of factors identified in Table 1.These factors were: • Have a size of at least 5 ha.
• Not be too visible or close (preferably at least 500 m) to residential areas.
• Have a 500-m distance to protected areas or areas with specific purposes such as golf courses or military zones.• Have access to water, electricity, heat and roads capable of supporting heavy trucks.• Be close to farmers in association with the local biogas company, as substrate and digestate desirably will be exchanged with them.• Preferably be owned by a farmer associated with the local biogas company who is willing to sell the land.• Be in an area with sufficient quantities of manure.The biogas plant is estimated to have a substrate capacity of 300 000 tonnes of manure (slurry) per year.• Preferably be close to or in industrial areas.It may be easier to obtain a building permit in such areas and also to create opportunities for collaboration with other industries.• Be strategically placed concerning wind direction in the area, due to odor dissemination from the plant.
• Not be an already rejected site by an environmental assessment.
The localization factors above were reorganized in to subfactors to match the categories in Table 1, and then systematically considered for inclusion or exclusion (Table 2).The Available infrastructure, Zoning and Nearby housing factors were well-documented and had nation-wide geographical data available.As such these three factors were prioritized.The specific chosen factors (bold in Table 2) were translated into spatially-based suitability constraints and expressed in the following way: 1) The location must be close to a road with bearing capacity class 1 or 4, as they can manage vehicles weighing up to 64 and 74 tonnes respectively (The Swedish Transport Administration, 2020a).2) The location must have a 500-m distance to nature conservation areas, military zones, recreational areas, sports fields and airport runways.
3) The location must have a 500-m distance to buildings.
Each of the suitability constraints described above will, from now on, be referred to as 1) roads, 2) zoning, and 3) buildings (All data files used to construct these constraints are listed in Appendix A).Excluded factors were those where: a) data was not available, b) an included factor would likely account for it already, or 3) was not relevant to the case study site.An example of the later is Biogas demand since the produced biogas will be distributed internally through the biogas company.Substrate supply and Digestate demand were also considered, but as part of the optimization model to select among sites that fulfilled other constraint factors.

Data processing
The developed suitability constraints were applied to identify possible locations in all of Sweden, Sjöbo, Tomelilla and Kristianstad (Fig. 4).The most limiting constraint could then be compared between the national and case study levels.Furthermore, the potential locations in Sjöbo, Tomelilla and Kristianstad municipalities could be used as inputs to an optimization model which then was used to explore three different site selection scenarios.

Identifying possible locations
To identify possible locations, we created scripts for extracting the 1 km 2 grids that fulfilled the suitability constraints for 1) roads, 2) zoning, and 3) buildings, individually and then all three simultaneously.These 1 km 2 grids were used as our operational unit, in other words a 'location' can be interpreted as one grid.This size was suitable for our purposes as it was larger than the 5 ha needed for a new biogas plant site in the case study area, and could be overlayed as sub-grids for selection in the optimization model which uses 5 km × 5 km gridded data for substrate supply and digestate demand (Metson et al., 2020).For each constraint (1-3) a grid was considered a suitable location for a biogas plant if it contained: 1) A road with bearing capacity class 1 or 4. 2) A continuous 5-ha area with a 500-m distance to incompatible zoning areas.3) A continuous 5-ha area with a 500-m distance to buildings.
To determine if a grid fulfilled constraints 2 and 3, we created a 500m buffer zone around all building and zoning objects and calculated the area of each continuous land piece outside of the zone for all individual 1 km 2 grids, to see if they would fit a 5-ha plant.The center coordinates for the extracted grids were saved to a spreadsheet, for each constraint and all constraints simultaneously.

Determining the most limiting constraint
We determined the most restrictive suitability constraint by calculating the exclusion of grids as a percentage of all grids for each constraint (Equation ( 1)).This grid exclusion metric was calculated for all of Sweden and the case study area.Since Sweden is a long country with changing dominant land uses from South to North, we also divided the country into three parts, approximately according to the lands of Sweden (Norrland, Svealand, and Götaland) for further comparison.

Optimization
We used a variant of the optimization model presented in Metson et al. (2020) to determine the most suitable biogas plant location among the suitable sites in the case study area.Briefly, the model is a "Facility Location Problem" (FLP) with demanding side constraints.FLPs are solved to find locations with minimized transportation costs concerning, for instance, outgoing goods to customers (Liu, 2009).Transport costs are influenced by the distance and weight of transport events among grids.In addition to minimizing weight times distance, our particular optimization model places additional constraints to both increase realism and avoid the over-application of nutrients to agricultural lands.
The center coordinates of the 1 km 2 grids that fulfilled all three constraints simultaneously were set as possible biogas plant locations.For each such possible location, all pairwise distances to the supply nodes were calculated using the center coordinates of 5 km × 5 km grids.Supply nodes were grids with manure (which by definition contains N, P, and K).Similarly, demand nodes accepting the digestate leaving biogas plants were based on crop nutrient needs.Each 5 km × 5 km grid could both have a supply and demand of nutrients, and thus, be a supply node and demand node simultaneously.
Three additional model constraints ensured that: 1) the biogas plant received exactly as much substrate as its capacity (300 000 tonnes/year), 2) the biogas plant sent out almost nearly as much digestate as it received substrate, and 3) the amount of digestate sent to the demand grids matched crop nutrient needs.
However, to obtain a feasible solution, the third constraint had to be relaxed because in the studied region there was more P and K supply in manure than crop demand.A limitation of the model was that it did not consider whether the 5 km × 5 km grids had a surplus or a deficit of the nutrients before receiving digestate.When applied at the national scale (as it was designed to) all non-pasture manure was transported to a biogas plant somewhere on the landscape.As such, regardless of if a grid had a surplus or deficit of a nutrient before applying the model, we could assume that the entirety of the crop nutrient demand could be met with digestate from a biogas plant if it was available.When applying the model locally, with only a few plants to place on the landscape, it was possible for the model to send digestate to grids that already had surpluses because manure in said grid was not sent away.We focus our results and discussion on the location selection aspects of the model and not on the specific transport distances or nutrient balances that result from the model application; but more information on these aspects is available at the national level (Metson et al., 2020) and case study level

Table 2
The localization factors in prioritized order and separated into subfactors.The green-marked factors were top-priorities and further studied, where one or more of the subfactors were selected (bolded) and either turned in to a suitability constraint or used directly in the optimization model.The yellowmarked factors not prioritized, although they were still considered feasible.The factors marked in red were not considered (The Swedish National Land Survey, 2020; The Swedish Transport Administration, 2020b).(Lindegaard and Ranggård, 2020).
The optimization model was run three times according to three scenarios 2 (Fig. 4).In Scenarios A and B, only one biogas plant was allowed to be placed at a time.More specifically, scenario A used the center coordinates of suitable location 1 km 2 grids within Sjöbo and Tomelilla municipalities.Scenario B used the center coordinates of grids in Kristianstad municipality.And finally, for Scenario C, the model was adjusted to place two biogas plants instead of one at a time to see if the plants were placed differently than in Scenario A and B. In other words, we wanted to see if the placing of a plant in the case study area should account for a future plant in Kristianstad.

Suitability constraints on biogas plant locations
At the national level, 10% of all 1 km 2 grids fulfilled the three suitability constraints simultaneously.Access to roads clearly had a major impact since the pattern of the locations resembled a road network (Fig. 5 A).Indeed, constraint 1 (roads) was the most limiting constraint as it removed significantly more 1 km 2 grids (79%) than constraint 2 and 3 (zoning and buildings), which were roughly equally limiting (Fig. 5 B).
Across all three regions (lands) of Sweden (Fig. 6 A), constraint 1 (roads) remained the most limiting suitability constraint (Fig. 6 B).However, the relative importance of the two other spatial factors varied.In Norrland, constraint 2 (zoning) was more restrictive than constraint 3 (buildings), while in Svealand and Götaland it was the opposite.Buildings being the least restrictive in Norrland is logical as the region has a lower population density than the rest of Sweden.When comparing Götaland and Svealand, the buildings constraint was more restrictive in Götaland.
In the Sjöbo and Tomelilla area, constraint 3 (buildings) was most limiting, which contrasts to Sweden as a whole (Fig. 7).It removed almost twice as many grids (70%) as suitability constraint 1 (roads, 35%) and 2 (incompatible zoning, 14%).The road network in the municipalities mostly consisted of roads with a bearing capacity class 1 and 4, which consequently left a large number of available locations when considering only constraint 1 (Fig. 7 A).Similarly, for constraint 2 (Fig. 7 B), the majority of the grids were identified as suitable as a result of the relatively limited extent of nature preserves, military zones, and recreational areas in the case study area.When applying constraint 3 (Fig. 7 C), less suitable locations were remaining, quite expectedly, due to the number of residential buildings spread over the area.Only 105 1 km 2 grids fulfilled the three constraints simultaneously (Fig. 7 D).Within the municipal border of Kristianstad, north of Sjöbo and Tomelilla, 187 locations fulfilled all three constraints simultaneously.

Most suitable location in case study area
When only placing one biogas plant in Sjöbo and Tomelilla (Scenario A) the most suitable location of the 105 options was positioned South of the area and within Tomelilla (black square Fig. 8).Substrate was collected from the 5 km × 5 km grids around the most suitable location, with only one completely outside the two municipalities (Figs. 8 and  9A).Digestate was sent to more, and slightly further-away, grids to meet crop demand compared to substrate collection (Fig. 9 A).The pattern looks reasonable considering that the optimization model aimed to minimize transportation costs, and consequently, distances.When considered alone, a new plant in Kristianstad would be located at the Southern edge of the municipality, right on the border with Tomelilla (Scenario B, Fig. 9 B).Substrate would be supplied from grids in all three municipalities.
When both areas were optimized simultaneously, Scenario C, the availability of supply and demand for nutrients were interacting as the biogas plants were placed differently than in Scenario A and B. The most suitable location in Tomelilla was located further south (Scenario A vs C, Fig. 9) and in Kristianstad it was located further north (Scenario B vs C, Fig. 9).The 5 km × 5 km grids where substrate was collected and digestate was distributed also changed when accounting for plants individually versus together.The supply and demand grids used in Tomelilla shifted south and those for the plant in Kristianstad shifted north to meet requirements and minimize transport costs. 2 The thesis work explored an additional set of three scenarios that only minimize transport distances for substrate supply as the was most interested in these results (Lindegaard and Ranggård, 2020).The work also explicitly compares the transport costs and distances among all six scenarios.G.S. Metson et al.

Reality-check: do our results make sense?
First, we answer our research questions, and contextualize them with other study results.
Nationally, access to roads was the most important localization factor constraining locations for biogas plants, but multiple factors are important to consider (question 1).Although we did not quantify all factors identified in Feiz et al. (2022) across Sweden, or all local factors mentioned by the biogas company, the three we did use are highly relevant across contexts.Avoiding incompatible land-uses, which includes residential housing, but also other land-use zoning classifications, is a concern for many organic waste management industries across countries (Scarlat et al., 2018;Thompson et al., 2013).In particular, residents often express that bad odors and increased transportation in Fig. 5. Suitable biogas plant locations across Sweden given selected constraints.Represented as A) the 1 km 2 grids (green) that fulfilled all three constraints in Sweden and as B) a comparison of how many grids each constraint excluded where the green bars illustrate the percentage of removal.Fig. 6.Suitable biogas plant locations across among Swedish lands given selected constraints.Represented as A) the delineations of Götaland, Svealand, and Norrland used for our comparisons (three shades of green) which are contrasted with the black lines which represent the official political borders of the lands.And as B) a comparison of how many percent of the grids were excluded by each constraint in the different parts of Sweden, illustrated by green bars.The dark green color is representative for Norrland (top), the mid-green color for Svealand (middle), and the light green color for Götaland (bottom).
the area as major concerns (Lantz et al., 2007).Our buffer of 500 m for a plant location from incompatible zoning may have been a conservative estimate from a European perspective; in a study of biogas plant acceptance in Germany, France, and Switzerland a distance of 3.1 km was necessary to get majority approval (Schumacher and Schultmann, 2017).Although concerns about odor are often a main concern, the socio-political context and perceived benefits of biogas can decrease concerns -Switzerland for instance showed higher levels of acceptance than Germany and France (Schumacher and Schultmann, 2017), a country which has high acceptance for existing biogas plants and positive experiences with the capacity of plants to manage issues (Soland et al., 2013).Although we could not account for all the nuanced contributors to potential political or social opposition to a plant (which can often be a deal-breaker to selecting a location (Feiz et al., 2022;Ketzer, 2020)) using existing land-use maps and a spatial buffer around buildings gave us a systematic way of avoiding problematic areas.Of course, more detailed information would be needed for each particular project, but our systematic method does not preclude adding additional factors (or editing factors such as increasing the distance to housing to account for odor concerns) to further focus efforts on suitable sites.
The suitability constraints we quantified had different impacts on how they restricted possible locations regionally.Incompatible land-use zoning was the second most restricting constraint in Norrland, while the constraint for remaining far from housing (buildings) had more impact (as second restricting) in Svealand and Götaland (Fig. 6 B).This disparity is probably an effect of the geographical differences regarding population density, where, for example, Götaland and Svealand have more densely-built areas than Norrland; making the buildings constraint more restrictive.There was even a large constraint impact difference between the case study area and Götaland (which includes Skåne).The roads constraint was almost half as restrictive in the case study area than in Götaland, and vice versa for buildings (Fig. 6 B).
Yes, placing more than one plant at a time does affect the optimal locations for plants (question 2).Simultaneously placing a plant in the Sjöbo/Tomelilla area and in Kristianstad municipality forced each further apart than if they were placed independently (Fig. 9).This result shows that biogas plants can have an impact on the location selection of other plants.This impact is in line with the fact that the competition for substrate materials increases in areas with a high density of biogas plants.Also, it might be difficult to find suitable areas for digestate distribution that contributes to sustainable nutrient recycling if new plant locations are not coordinated (Wellinger et al., 2013).In our case, there is a deficit of all three nutrients (N, P and K), in the south-western part of Skåne (Metson et al., 2020).Hence, when the optimization model considered digestate distribution, it seems reasonable that the plant in Kristianstad was located far south to be closer to these nutrient deficit areas.Therefore, the plants in Tomelilla and Kristianstad were situated close to each other and the availability of substrate and need for nutrients were interacting.

Benefits and limitations of our approach
Second, given the results above, we explore how relevant our approach can be beyond the case study area.From a spatial perspective, there are two general approaches the problem of site-location for biogas plants (1) suitability analysis (e.g., Dhaka Bangladesh (Akther et al., 2019)) and (2) optimization analysis (e.g., Guangdong China (Shi et al., 2008)).Our approach combines both approaches which we view as a strength, especially for increasing its usefulness across contexts.However, the need to standardize the suitability approach (akin to our localization factors) in to a set of systematic constraints means that one loses some of the nuance that can go in to a more qualitative multi-criteria analysis.
The methodology presented can be applied anywhere in Sweden without having to take political boundaries into account.In our case study considering municipal boundaries was necessary, but other energy companies might have different entry points which are less geographically restricted and which makes our datasets particularly appealing.Other studies have developed GIS methodologies for determining potential biomasses and locations for biogas plants but usually focus on high-resolution data for specific regions (e.g., in Finland (Höhn et al., 2014), in addition to others mentioned in the introduction).Consequently, the geographical applicability of their methods might be limited.Although we presented a specific case to motivate our selection of constraints and optimization results, the approach is systematic with script automation and data retrieval from national databases.As such it is replicable and relevant across the whole country.Our approach might also be useful in other countries, depending on dataset accessibility, since the spatial factors we accounted for are relevant to plant site selection in many contexts including manure management through biogas across Europe (Scarlat et al., 2018) to municipal organic waste in emerging Asian countries (Pandyaswargo et al., 2019).Details can easily be customized in the scripts, such as plant size, bearing capacity classes and minimum distances from buildings and zoning areas.
Still, as one moves from a national perspective to a local one, the priority of localization factors (either as suitability constraints or enablers) affecting site selection will change (Feiz et al., 2022) which puts limitations on how realistic the outputs of our models can be.Even within our own study we could see how the relative importance of constraint factors shifted when looking at the case study area.Both Thompson et al. (2013) and Sliz-Szkliniarz and Vogt (2012), who conducted spatial analyses to identify suitable biogas plant locations on a local scale in the USA and Poland, respectively, highlighted the importance of proximity to an existing electricity distribution infrastructure.Although the presence of roads, which we did account for, increases the likelihood that other types of infrastructure are present in the vicinity, it is not guaranteed.Similarly, the 500 m buffer we used as the constraint for avoiding housing areas should minimize issues with smells and un-desirable views.However, this chosen distance might be insufficient in some areas, as the spreading of odors is highly dependent on wind direction and views depend on local topography and existing buildings and vegetation (as hinted at in other European regions earlier).Many of the factors we did not consider become even more relevant at spatial scales below 1 km 2 when trying to find the exact parcel of land for building.
Substrate supply and digestate demand are factors that operate across multiple geographical levels/resolutions, which makes them particularly difficult to account for appropriately.One of the advantages of our national optimization approach is that it can be applied either before or after a particular site has been selected.In this paper we used the optimization to select among possible locations, which means the most suitable location is in part determined by supply and demand.If a location was already pre-selected, one could still use this optimization framework to determine which areas would result in the smallest transport costs to source manure as a substrate and distribute digestate to meet crop demands.Still, the 5 km × 5 km resolution of these datasets do not account for the potentially large number, and diversity of, farmers in a grid and how they view collaboration with a biogas plant.Fig. 8.The optimization result for Sjöbo/Tomelilla placing one biogas plant on the landscape.The orange 5 km × 5 km grids represent where substrate was collected, and the black 1 km 2 grid is the chosen most suitable biogas plant location.The green 1 km 2 grids represent all of locations within the municipality borders that were considered suitable according to all three constraints from which the optimization model could choose from to place the biogas plant.
Farmers have a crucial role regarding the implementation and profitability of a manure-based biogas plant (Scarlat et al., 2018), and it is common that farmers are unaware about the biogas potential of agricultural residues, and that the existing policy measures to change this ignorance are ineffective (Lönnqvist et al., 2015).It is also possible that farmers are uninterested in collaborating as they already reuse their manure.Perhaps more farmers could be encouraged to participate in the biogas production process by highlighting the enhanced fertilization qualities (Lantz et al., 2007), lower greenhouse gas emissions (Insam et al., 2015;Möller, 2015), and reduced odor (Orzi et al., 2018) that manure obtains during the anaerobic digestion process.Our work has limitations as it does not investigate the processes by which more support or incentives can facilitate biogas solutions.
Here we focused on manure substrate, but parts of our approach could be adapted to biogas plants looking for other substrates.For instance, crop residues or bioenergy crops could be expressed in a Fig. 9.The most suitable locations (black 1 km 2 grids circled with red) for biogas plants in A) Sjöbo/Tomelilla, B) Kristianstad, C) Sjöbo/Tomelilla and Kristianstad simultaneously.Substrate was collected from the orange 5 km × 5 km grids (light orange for Sjöbo/Tomelilla, dark orange for Kristianstad) and digestate was distributed to the blue 5 km × 5 km grids (filled for Sjöbo/Tomelilla, crossed for Kristianstad).
similar way as the manure availability grids.For those plants looking at municipal food waste, wastewater, or slaughter waste, site selection is in most cases much more constrained.Although some of the localization factors we mapped as part of answering question one would be relevant, the fact that these feedstocks are usually already fixed in space likely makes substrate supply the largest constraint.Still, pre-selected locations for diverse substrates could be integrated into the optimization model to ensure that the distribution of digestate of multiple biogas plants do not compete (e.g., sewage sludge and manure (Metson et al., 2020)).Finally, the optimization model component of our work could also be expanded to account for aspects related to the pricing and delivery of energy from such plants.

Usefulness of results
Finally, we frame our findings in terms of their applicability moving forward.
The maps of suitability constraints presented here can help biogas actors target locations that warrant further exploration with additional local datasets.The constraints we accounted for narrowed the biogas company's focus down to 105 possible locations to further explore in the case study area.They can now more easily identify and start a dialogue with the right stakeholders, such as farmers who are close to these locations.In other words, even though we did not include the factor 'land owned by farmers already collaborating with the local biogas company', our approach allows companies to more selectively look for relevant data sources and collaborations to account for said 'missing' factor.The municipality's building committee is often one of the most important stakeholders in the end, since they will be granting the building permit and often control detailed zoning plans (The Swedish National Board of Housing, Building and Planning, 2018); entering a dialogue with them with multiple location options could be beneficial to get the process moving.The fact that our approach does not have to narrow to only one suitable location increases the likelihood that one of the candidate locations could fulfill all local constraints (including environmental assessment and building approval) than if the method could only show one optimal site.
Our optimization results show that selecting biogas plant sites in isolation will likely result in sub-optimal transportation when many plants come to operate close to one another.This result may be of most immediate interest to companies, like the one that we collaborated with, who operate regionally or nationally and plan to build multiple plants over time.However, minimizing transport costs and maximizing nutrient reuse from manure also has large societal benefits which could then warrant governments and other organizations to want to participate in facilitating good biogas plant location selection.Our results in Scenario C demonstrated that to minimize over-application of N, P, and K, the two biogas plants would target different areas to reuse digestate.Minimizing nutrient over-application or greenhouse gas emissions via transport might not be top priorities for an individual biogas company, but they are still an important actor for stimulating nutrient redistribution and minimizing pollution.A systematic collaboration with different stakeholders, from governmental down to individual businesses is necessary to be able to effectively increase biogas production (Bourdin and Nadou, 2020;Hengeveld et al., 2016;Westerman and Bicudo, 2005) in a way that helps Sweden meet its Reduced Climate Impact and Zero Eutrophication environmental objectives (The Swedish Environmental Protection Agency, 2019).

Conclusions
This work provides insights to the challenge of finding suitable locations that both public and private biogas actors face when planning new biogas plants.We have developed a methodology that is based on spatially explicit datasets for suitability constraints concerning 1) sufficiently large and durable roads, 2) keeping a large enough distance from housing developments, and 3) other potentially conflicting landuse zoning.This approach identified 105 suitable locations for a new biogas plant in Sweden's Sjöbo and Tomelilla municipalities.Although our methodology focused on a case study area, it is based on national datasets and the impact of the suitability constraints varied at local, regional, and national levels.Although the availability of roads was most constraining at the national level (removing 79% of grids), the further South one looks, the more other constraint types become important (roads remove 62% of grids in Southern Sweden and 35% in our case study area).By involving a stakeholder in defining the research questions and scope of the methodological approach we were able to ensure the saliency of our results and contribute to real-world acceleration towards sustainability.
The datasets compiled and analyzed here represent an advancement for moving from theory to practice; given the pressing nature of climate change and eutrophication mitigation in Sweden, and globally, such work is a step forward.Our approach can be applied anywhere in Sweden without having to consider political boundaries.Energy companies can use our results as groundwork to communicate with stakeholders and explore appropriate options that balance multiple considerations for selecting a location for a biogas plant.For instance, a biogas company could more easily identify farmers in a region they may want to partner with as our approach narrows down the number of suitable locations and potential substrate collection and digestate distribution areas.
In addition to our research's usefulness to individual actors, it makes a contribution in our understanding of the need for coordinated action, and provides a systematic tool to facilitate such cooperation (given the model and datasets are standardized).Since other plants might affect the most suitable location of an intended biogas plant, long-term planning, both within a company and across companies, will be crucial for biogas actors.In future studies, it would be interesting to incorporate more spatially explicit suitability constraints such as land ownership or access to infrastructure and do so in diverse case study areas.

Fig. 1 .
Fig. 1.Methodological overview.The colors separate the activities into the four main stages.

Fig. 2 .
Fig. 2. Geographical scope of the study.A) Sweden with the region of Skåne outlined in the south of Sweden and B) A close-up of Skåne with its 33 municipalities (thin black boarders).Three of them -Sjöbo, Tomelilla, and Kristianstad -are in focus in this work (bold black boarders).

Fig. 3 .
Fig. 3. Nutrient balance maps of Sjöbo and Tomelilla municipalities for A) nitrogen B) phosphorus, and C) potassium expressed as kg of nutrients per hectare (based on Metson et al. (2020) which primarily uses data from 2013 to 2016, and which accounts for fertilizer recommendations based on soil type and crops grown in each grid).The balance is an expression of the difference between the supply of nutrients as manure and crop demand in each 5 km × 5 km grid.Positive values (red range) denote a surplus of the nutrient compared to crop demand, while negative values (blue range) denote a deficit of the nutrient.

Fig. 4 .
Fig. 4. Overview of the three last stages in the methodology.The colors correspond to the stages in Fig. 1.

Fig. 7 .
Fig. 7. Suitable1 km 2 locations in Sjöbo and Tomelilla municipalities.Where A) represents suitable grids according to constraint 1 for roads, B) represents suitable grids according to constraint 2 for zoning, C) represents suitable grids according to constraint 3 for buildings, and D) represents suitable grids according to all constraints simultaneously.