Resources, Conservation & Recycling: X 9–10 (2020) 100049
Available online 6 January 2021
2590-289X/© 2021 The Author(s). Published by Elsevier B.V. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
Optimizing transport to maximize nutrient recycling and green
energy recovery
Genevi`eve S. Metson
*,a, Roozbeh Feiz
b,*, Nils-Hassan Quttineh
c, Karin Tonderski
d aTheoretical Biology, Department of Physics, Chemistry and Biology, SwedenbEnvironmental Technology and Management, Department of Management and Engineering, Sweden cOptimization, Department of Mathematics, Sweden
dBiology, Department of Physics, Chemistry and Biology Link¨oping University, Link¨oping SE-581 83, Sweden
A R T I C L E I N F O Keywords: Manure Circular economy Biobased economy Nitrogen Phosphorus Life cycle assessment
A B S T R A C T
A circular biobased economy must be able to sustainably manage multiple resources simultaneously. Nutrient (nitrogen, phosphorus, and potassium) recycling and renewable energy production (biogas) can be compatible practices but require substantial transport of heavy organic waste. We combine a spatial optimization model and Life Cycle Assessment (LCA) to explore how Sweden could maximize its use of excreta resources. We use 10 × 10 km2 resolution data on the location of animal and human excreta and crop demand and model both optimal biogas plant locations and transport of nutrients to and from these plants. Each type of biogas plant (given 4 realistic mixes of excreta) is then evaluated for global warming potential, primary energy use and financial resource costs. Moving excreta through biogas plants, as opposed to simply reapplying on fields, to meet crop nutrient demands comes at a similar cost but the climate and primary energy savings are substantial. As much as 91% of phosphorus and 44% of nitrogen crop demand could be met via optimally transported excreta and the country would avoid about 1 450 kt of CO2-eq, save 3.6 TWh (13 000 tera-joules) of primary energy, and save 90
million euros per year. Substituting mineral fertilizers with recycled nutrients results in savings across all in-dicators, but the added energy and avoided greenhouse gas emissions associated with biogas production make a large difference in the attractiveness of nutrient recycling. Although the numeric values are theoretical, our results indicate that carefully coordinated and supported biogas production could help maximize multi-resource benefits.
1. Introduction
The UNs Sustainable Development Goals (SDGs) related to global food security, clean water and healthy aquatic ecosystems all demand that we change the way we manage natural resources, including nutri-ents (Kanter and Brownlie, 2019), in modern society (Randers et al., 2019). Although each natural resource has its own particularities and implications for SDGs, three types of changes are often examined: 1) increased resource substitution, 2) increased efficiency, and 3) increased recycling as ways to decrease resource depletion and pollution. Trans-forming organic materials from waste to resource is a central part of all three types of changes and is promoted in fields such as Industrial Ecology (Ayres and Ayres, 2002; Kapur and Graedel, 2004; Niutanen
and Korhonen, 2003).
Nitrogen (N) and phosphorus (P) in particular require a redesign of
waste management as losses to the environment have caused atmo-spheric (N) and aquatic (N and P) pollution, while production of mineral fertilizers to supply these essential plant nutrients also deplete fossil resources (Ibisch et al., 2016; Steffen et al., 2015). Sub-optimal/inefficient recycling of excreta (animal manure and human urine and feces) contributes to the large gap between input and output of N and P on agricultural land in Europe, which results in nutrient emis-sions to waterways and eutrophication of aquatic and marine ecosys-tems in Europe (van Dijk et al., 2016; Ibisch et al., 2016). One major challenge is the distance between food and feed production and con-sumption resulting in spatial concentrations of excreta and long trans-port distances to fields where the nutrients are mostly needed (e.g. Long et al., 2018; Nesme et al., 2015). Unprocessed excreta is bulky and heavy, making it expensive to transport. Previous research in Sweden, a country that has prioritized diligent nutrient management to reduce * Corresponding authors.
E-mail addresses: genevieve.metson@liu.se (G.S. Metson), roozbeh.feiz@liu.se (R. Feiz).
Contents lists available at ScienceDirect
Resources, Conservation & Recycling: X
journal homepage: www.sciencedirect.com/journal/resources-conservation-and-recycling-x
https://doi.org/10.1016/j.rcrx.2021.100049
eutrophication of the Baltic Sea (Ibisch et al., 2016), has shown that even when minimizing transport costs the fertilizer value of excreta would likely not offset costs (Akram et al., 2019a). Valuing more ben-efits from reuse, such as energy production or mitigated pollution, could be a way of addressing this profitability gap.
Excreta is often high in N, P, and potassium (K), but also organically bound energy. From a resource substitution perspective, using excreta as a source of bioenergy or biofuel instead of fossil fuels is a viable option to decarbonize electricity, heat, and transport systems, while also reducing methane emissions, a potent greenhouse gase (GHG), from manure storage systems (VanderZaag et al., 2018).
Because excreta is mostly available in water-rich forms, i.e. cow and pig slurry, combined nutrient and energy recovery through anaerobic digestion is a well adapted technology (Holm-Nielsen et al., 2009; Lantz
et al., 2013). Thus, treating excreta in a biogas unit followed by storage
and spreading of the residual biofertilizers could potentially offer a way of working towards several of the EU Circular Economy and Bioeconomy objectives (B¨orjesson and Mattiasson, 2008; De Schoenmakere et al.,
2018; Hagman and Eklund, 2016).
Still, effective recycling of excreta is challenging even in the face of supportive concepts like circular economy. It is much easier to keep economic value in recycling inorganic materials, such as minerals and metals, than organic material; the development of circular solutions for biobased products often lags behind (Carus and Dammer, 2018). Mandating that organic waste be reused does not, by definition, mean that excreta will be used in a way that maximizes the multiple resources it contains. For instance, in Italy, it was only after a policy framework that targeted GHG emissions reductions was implemented that manure, as a substrate for biogas production, was favored instead of energy crops (Bartoli et al., 2019). In addition, legal, political, and social acceptance barriers (in particular for human excreta (Harder et al., 2020)) can hinder development of reuse solutions, e.g. of nutrients. In other words, although excreta recycling through biogas plants has many theoretical benefits, exploring and quantifying known barriers while accounting for multiple benefits is necessary to move towards real-world change.
In practice, transportation remains one of the core challenges to in-crease resource recovery and reuse from excreta, not only for nutrients but also for energy. In Sweden, Ammenberg and Feiz (2017) showed that the transport of substrate to biogas plants and digestate towards farms is a prime concern for companies when establishing a new biogas plant. So, though biogas solutions have the potential to recover energy from organic waste while supporting nutrient recycling, without an explicit focus on the logistical transport challenges it is unlikely that the bio-energy and biofertilizer values of excreta will be maximized (Astill and
Shumway, 2016).
The overarching goal for this study is to assess if (and how) biogas solutions could contribute to Sweden meeting multiple sustainability goals (related to SDGs) in a circular economy framework.
In order to meet our objective, we first need to ascertain what resource potential is available in the country and ask the question:
1. How much of Sweden’s crop nutrient demands could be met and how much energy savings could be achieved through complete recycling of excreta via biogas plants?
Then it is necessary to adapt a previously used spatial optimization model to locate biogas plants strategically and determine:
2. Where should biogas plants be located on the Swedish landscape to minimize transportation costs?
Finally, given that our previous work showed that the value of nutrient recycling does not outweigh transport costs (Akram et al.,
2019a), we use a Life Cycle Assessment (LCA) model to explore a larger
suite of sustainability priorities to consider:
3. What are the biggest contributors to added value (or costs) associ-ated with excreta recycling with and without biogas solutions? Results will provide important inputs to motivate the potentially large investments in logistics and infrastructure that are needed in areas that previously have not used bio-fertilizers (Parchomenko and Borsky, 2018; Spiegal et al., 2020).
2. Methods
Our approach was to first run a spatial optimization model to define the locations and excreta loads to biogas plants for different scenarios, and then evaluate selected model outputs using an LCA based approach. From the optimization perspective the goals were to minimize the total cost (which is determined by weight × distance) of transporting excreta as feedstock to biogas plants and then the digestate to agricultural fields. An important constraint was to limit any over-application of nutrients to these fields. From the LCA perspective, the goals were to quantify the impacts of such “full recycling” scenarios with selected economic and environmental metrics.
To answer our research questions we evaluated the outcomes of three scenarios (Fig. 1). The first scenario (BASELINE) estimated the cost of redistributing all excreta directly from supply areas to agricultural fields with a nutrient demand, without any biogas plants. In the second sce-nario (BIOGAS GENERIC), we introduced generic biogas plants which all excreta must pass through before the digestate is sent to agricultural fields. The third scenario (BIOGAS MIXES) is our main scenario, and here we considered both different animal manure types and different biogas plant types. The same data were used for all three scenarios, and each scenario was run in two versions; with and without sewage sludge. Sludge from wastewater treatment plants (WWTP) is used as a repre-sentation of human excreta (Section 2.3). The term manure is used when referring only to animal excreta in the manuscript. When referring to both manure and sludge, we use the encompassing term excreta.
In the following sections we explain how we calculated the avail-ability of excreta and crop nutrient demand as well as how the optimi-zation and LCA models used this information. We have made a number of decisions in constraining the models and scenarios to increase realism (Table SI-1). For instance, we use different biogas plant types and treated manure and sludge in separate biogas facilities as these sub-strates are rarely treated together. Still, it is important to note that the scenarios produced and evaluated are not intended to perfectly repre-sent reality, but rather a theoretical exercise to inform future research and decisions. After separate collection and treatment, the distribution of both types of excreta were considered simultaneously to meet crop demands.
2.1. Quantifying substrate availability
We calculated the amount of manure available in 10 km by 10 km grids for all of Sweden using the methods presented in Akram et al.
(2019b). We updated the data with 2016 records on the location and
number of animals (JBV, 2016). These updated data included the loca-tion of poultry farms, in addiloca-tion to cattle and pigs farms that were available in earlier records, which allowed us to more accurately esti-mate the spatially explicit availability of excreta on the landscape. Horses were excluded because of a lack of a spatially explicit inventory. Two other important methodological updates to fit the purpose of this work were the exclusion of excreta deposited on grazing and pasture lands and the estimation of total solids in manure. Grazing animals, by law in Sweden, should spend about 50 % of their time outside during the spring-summer period (SJV, 2019). Taking a conservative approach, we assumed that non-dairy cows, sheep and goats spend only half of the year in settings that would allow for excreta collection. Because of the importance of total solids (TS) content for biogas digestion, we parti-tioned available excreta for each animal type into slurry and solid
(mixed with bedding material) manure based on available national av-erages (Table SI-2).
Here we do not consider that nutrients excreted by all humans in Sweden are available for reuse as was done in previous work (Akram
et al., 2019a; 2019b); instead we assume that only the sludge generated
when treating wastewater in WWTPs could be processed through an on-site biogas plant. We only included excreta from humans living in settlements over 200 people for the same reason that we excluded grazing manure; it is more likely that the sludge from agglomerated households, rather than from individual rural houses, could be collected and processed for biogas production. In addition to updating the loca-tion and number of humans considered to match 2015 data (SCB, 2015), calculations for mass and nutrient content in human excreta were also revised from those presented in Akram et al. (2019b). In Sweden, nu-trients in human excreta are transformed in WWTP, where a large fraction of N is lost to the atmosphere due to denitrification. Instead of considering the weight and nutrient content in excreta from an indi-vidual human, we based the calculation on the weight, nutrient and total solid content of sludge generated in wastewater treatment plants in Sweden in 2012 (SCB, 2014) divided by the person-equivalent (p.e.) loading to those WWTPs. For K, the mean of the 2013 values given in
Andersson (2015) were used. Combining the data sources above resulted
in an estimate of the annual mass of dewatered and digested sewage sludge, as well as its TS, N, P and K concentrations. The amount and composition of sludge before biogas production was estimated based on the amount of dewatered and digested sludge that was reportedly pro-duced in Sweden.
2.2. Quantifying crop nutrient demand
The location and quantity of N, P, and K crop demand was also estimated using methods in Akram et al. (2019b), but with updated data from 2013 (JBV, 2013). It was not possible to use 2016 data to match the year used for manure because of changes in data collection and defini-tions for the inclusion of arable land. To be consistent with the update regarding manure availability, we also eliminated nutrient demand from grassland and pastures in our estimates of crop demand.
Finally, the nutrient supply from manure (type of manure, weight of manure, amount of N, P, and K) and sludge (weight of sludge, amount of N, P, and K) available in each 10 × 10 km2 grid are referred to as supply grids, and similarly, the crop demands (amount of N, P, and K) in each 10 × 10 km2 grid are referred to as demand grids. These gridded
data-sets, as was the case for 5 × 5 km2 grids in Akram et al. (2019b), are
inputs to the optimization modeling framework described below. 2.3. Determining potential biogas plant locations and types Manure
The first step in determining the potential locations of biogas plants was to determine the number and size of plants that could be accom-modated for at the national scale. We assumed that a biogas plant would accept between 10 000 and 100 000 tonnes of wet excreta as feedstock (referred to as supply in the following sections) and that all (collectable) manure in the country would be treated in these plants. As such we simply divided the total weight of available manure by 100 000 tonnes to get the minimum number of biogas plants needing to be located on the landscape (n = 162). We relaxed this assumption by allowing the model Fig. 1. Conceptual representation of excreta flows and
resource recovery considered across the three scenarios explored through optimization and LCA modeling. Nutrient demand is quantified for nitrogen (N), phos-phorus (P) and potassium (K). Demand is met by manure in supply areas as well as by sludge from wastewater treatment plants (WWTP). All three sce-narios come in two versions; with and without sludge. The scenarios increase in the detail of supply sources considered, as well as biogas plant specificity from no biogas (BASELINE) to different types of biogas plants (BIOGAS MIXES). In the BIOGAS MIXES scenario, manure types are: Cow Liquid (CL), Cow Solid (CS), Pig Liquid (PL), Pig Solid (PS), Poultry (PY), and Other (OT). The biogas plant types are called MixA, MixB and MixC plants, and they require different compositions of manure types, specified by constraints according to Table SI-3.
to place 20% additional plants (i.e. 194 biogas plants) in order to minimize the cost of transportation. The model can use all these plants but not their full capacities; this adjustment was made because not allowing more plants than the minimum needed for the substrate supply could have created scenarios with very high transportation distances. We then determined candidate areas for these plants by performing a clustering of all supply grids into 50 km by 50 km grids, and using weighted mean values (with respect to the amount of excreta supply) of their coordinates as location points. This clustering yielded a total of 206 positions (Fig. SI-1). The reason for using weighted means in the clus-tering was to avoid sending large quantities of excreta for unnecessarily long distances, by placing the biogas plants as close as possible to points with a lot of supply. This was done separately for animal manure and sludge.
Although we considered a plant of 100 000 tonnes or less as 1 unit, when excreta was especially concentrated on the landscape it is possible that many of these units ended up being in the same location, meaning that they could in practice be considered as one larger unit.
For each potential biogas plant location, we let the model determine which type of plant would be best to place there. For our main scenario (BIOGAS MIXES), we derived three manure mixes that were likely to arise on the landscape based on manure availability. Each manure mix was translated to a biogas plant type. More specifically, the most com-mon type of plant would receive MixA, which is dominated by cow manure, followed by MixB, dominated by pig, and MixC dominated by poultry manure. The total weight and composition of these mixes was initially calculated manually to get average values (of N, P, and K con-centrations and of manure types), and then determine bounds for each component of the mix for realism in the biogas plants. The bounds given to the model (Table SI-3) are based on limitations in the carbon to N ratios (C:N) that allow for anaerobic digestion. While the settings of the digester allow it to operate on different C:N ratios, it may become problematic if it is lower than 10–15 (Carlsson and Uldal, 2009), and we chose the lowest limit to be 11 (C:N > 11). Consequently, the C:N ratios were between 11 and 14 for the resulting substrate mixes. We did not make the plant type selection based on TS, because the calculated TS values of the mixes (using data in Table SI-2) were reasonable for wet digestion plants which is typically below 15 % (Carlsson and Uldal, 2009).
Sludge
As we only use sludge from WWTP, we first considered the real location of the nine largest WWTPs in the country, and then identified candidate locations for biogas plants based human population density. For the nine fixed WWTP locations we estimated the total weight of sludge generated based on the person equivalents connected to each plant (Table 1). Next, we assumed that those plants receive wastewater from the population in the surrounding grids. In ArcGIS we used the BUFFER tool to draw a circle around the plant so that it encompassed
enough grids (based on the population) to match the estimated sludge production in that WWTP, where a grid’s population was included if it’s center point was within the buffer circle (Table 1).
The circle associated with the plant in S¨odert¨alje overlapped with the circle for the two plants in Stockholm. To deal with this issue, grids within this overlap were first assigned to feed the plants in Stockholm, and any remaining sludge-generating wastewater was then sent to the plant in S¨odert¨alje. The same issue (and solution) applies to plants in Malm¨o and Lund; grids were first assigned to the plant in Lund and second to the plants in Malm¨o.
The remaining grids with sludge supply were then distributed to “theoretical plants” with a minimum capacity of 1000 tonnes. These potential locations for smaller plants were determined by creating larger grids (as done when clustering manure supply grids to 50 × 50 km2
grids) with the weighted center point receiving all of the sludge from the smaller grids within. The final result was a map of concentrated loca-tions of where sludge could be treated (7 large plant localoca-tions + 90 smaller plant locations) that are considered to be supply nodes for human excreta nutrients to meet crop demand (along with the manure biogas plants). For all WWTPs, we set the incoming travel distance to 0 as usually wastewater would be collected via a sewer system or with trucks moving sludge from rural septic tanks to a treatment facility. 2.4. Determining optimized distances to and from biogas plants
The developed optimization models, one for each scenario, are strongly related to classical optimization problems; more specifically the transportation problem and the facility location problem, but with an additional set of constraining real-world conditions.
The optimization model for the main scenario, BIOGAS MIXES, should simultaneously minimize (the objective function consisting of) the cost of transportation (of feedstock) to biogas plants and the cost of transportation (of digestate) from biogas plants to croplands, without over-applying nutrients and respecting the biogas plant capacities. The optimization model contains 35 084 linear constraints, 618 integer variables and a total of more than 5.5 million continuous variables. It was implemented using the modeling language AMPL (Fourer et al., 2003) and solved using the state-of-the-art mixed-integer solver cplex from IBM (Cplex, 2009). Details are given in Table SI-15. Outputs from the model consist of the locations of biogas plants together with a transportation plan which specifies where (to which plant) each supply grid’s amount and type of manure (as feedstock) should be sent. The transportation plan also specifies which crop demand grid the excreta from each biogas plant (as digestate) should be sent to. It was never possible for a transportation plan to allow the sum of transports to a grid exceed the N crop demand in that grid, but the constraints for P and K were slightly relaxed as total national supply was larger than crop de-mand. Without this relaxation, there would have been no solution. The optimization model in its entirety is described in the Supplementary Information.
To keep all constraints in the optimization model linear, which is key when solving large-scale problems, it is necessary to make use of average nutrient concentrations (per tonne of manure mix for a biogas plant type). Since the nutrient concentrations of the manure sent to the indi-vidual biogas plants will differ (more or less) from the average values, the acquired solution is not necessarily feasible with respect to the constraints that make sure we do not over-apply nutrients.
But, with a solution at hand, it was straightforward to calculate the actual concentrations for each nutrient (N, P and K) in the manure sent to each biogas plant. Hence, one way to handle this issue, without considering a non-linear model, was to adjust the acquired solution with the following three-step procedure:
1. Solve the optimization problem using average nutrient concentration values, which yields a feasible solution with respect to transportation of manure to each plant, as well as biogas plant locations.
Table 1
Wastewater treatment plant (WWTP) sludge production calculated from the person equivalents (p.e.) connected in 2016/2017, and the estimate based on the population in the grids of a surrounding circle used to calculate sludge nutrient supply at each point. Values are rounded to the nearest thousand tonne.
City Sludge, p.e. connected to
WWTP Sludge, population in circled grids (thousand tonnes) (thousand tonnes)
G¨oteborg 82 83 Link¨oping 24 34 Lund 11 11 Malm¨o* 41 42 Stockholm* 116 154 S¨odert¨alje 32 32 Uppsala 18 18 *sum of two WWTPs
2. Consider the decisions on where to locate the biogas plants, and the in-flow of excreta to them, as final. (Corresponding variables are fixed, as in “no longer possible to vary”.)
3. Calculate actual concentrations of each nutrient at each biogas plant, and resolve the problem of redistributing the digestate to demand grids.
This will produce a feasible transportation of manure from each plant that respects the NPK-demands of the crops.
Nevertheless, the solution(s) produced by our optimization models for BIOGAS GENERIC and BIOGAS MIXES are believed to be of high quality, but the required adjustment described above means that the solution(s) are not guaranteed to be mathematically optimal.
Finally, the optimization model had one objective, minimize trans-port costs, and we used constraints to meet nutrient management con-cerns and an LCA approach to track a number of other important sustainability metrics. Results would likely be different if the model was in-fact multi-objective, focusing on multiple sustainable priorities at once. Still, constraints remain a viable way to account for diverse pri-orities given computational power limitations.
2.5. LCA models
LCA goal and scope definition
The optimization models aimed to minimize transport costs, but did not explicitly calculate costs or environmental impacts. Thus we per-formed LCA modeling which combines the optimization results with selected conversion factors based on a bounded set of processes related to farm and biogas production practices.
The LCA models followed standard guidelines (ISO, 2006a; 2006b). The functional unit was the treatment of all available excreta in Sweden (tonne/year) as a comparison basis across the scenarios. As manure and sludge were considered to be the same input to the system in all sce-narios, activities related to their production were not included, but transportation activities, biogas production (including upgrading), use
of the produced biogas as transportation fuel (which is a common practice in Sweden for large-scale biogas plants), use of the digestate as biofertilzer (including storage and spreading), as well substitution ef-fects related to biogas and biofertilizers were included in the system (Fig. 2).
In all studied scenarios, we aggregated the LCA results under the following groups, where, as per its scenario definition, the BASELINE scenario excludes those processes that have to do with biogas production (Fig. 2):
1. Transportation, referring to transportation of excreta and digestate; 2. Handling, referring to storage and spreading of excreta or digestate; 3. Other processes, referring to biogas production, upgrading of the
produced biogas, and distribution of biogas;
4. Substitution of mineral fertilizers, referring to the amount of nutrients in the spread excreta or digestate that can replace mineral fertilizers while respecting the maximum demand as calculated based on spatial demands in the grids;
5. Substitution of fossil fuels, referring to the substitution of diesel fuel with the upgraded biogas; and
6. the Net, referring to sum of all of the above.
No biogas is produced in the BASELINE scenario and the excreta is transported directly to the demand grids (according to the output from the optimization model) where it is applied on fields. Similar to diges-tate, applicable storage and spreading emissions were considered for the undigested manure and sludge (Table SI-4).
The main scenario of interest (BIOGAS MIXES) had the most complexity and thus required the most assumptions. For each substrate mix type we received a gross list of biogas plants (from the optimization model), each having its own transportation plan (inbound and outbound distances), and substrate amount and mixture. We calculated the mean values of the substrate mix and transportation distances to/from each biogas plant in this gross list to obtain a representative biogas plant type corresponding to MixA, MixB, MixC, or Sludge (“model biogas Fig. 2. System overview of the LCA
model for the studied scenarios. Top: BASELINE scenario in which biogas is not produced; Bottom: BIOGAS GENERIC and BIOGAS MIXES sce-narios. In BIOGAS MIXES, the biogas plants are adjusted (mainly through different dilution levels for TS adjust-ment and different internal heat de-mand in the biogas plants) to match the received manure mixes (MixA, MixB, or MixC); while in BIOGAS GENERIC all biogas plants receive similar manure mixes. In all scenarios, the activities in the farms (other than storage and spreading) or wastewater treatment (except biogas production) are assumed to be outside the life cycle system boundary. Life cycle flows such as use of electricity or fuel use in processes are not shown in the diagram.
plants”). In other words, instead of creating a LCA model for each of the biogas plants in the gross list, we created a single model biogas plant (for each substrate mix type) and after modeling its life cycle impact, multiplied it by the number of plants in the gross list to get the full impact of the BIOGAS MIXES scenario (Tables SI-5 and SI-6). The overall impact of the BIOGAS GENERIC scenario was calculated in a similar way, but no distinction was made between the biogas plant types.
Inventory data and impact categories
We considered three life cycle impact categories and performance indicators: climate change, primary energy use , and resource cost (Cost).
Climate impact was selected as it is a common and relevant impact category for bioenergy studies and due to its importance in relation to Swedish environmental objectives (Swedish EPA, 2018). Energy use was chosen to investigate if the energy recovery options are sensible from an energy perspective (Arvidsson and Svanstr¨om, 2016), but also due to the general importance of primary energy use as a good proxy for the environmental impacts of different processes (Huijbregts et al., 2010). The resource cost was included to provide an indication of the operating costs (e.g. biogas plant and upgrading unit electricity demand and transportation) with the aim of putting the optimization part of the work, focused on transportation cost, into a wider perspective. There-fore, the cost assessment is not meant to replace a full financial analysis.
The LCA data were gathered from different sources (Tables SI-8 and SI-9). In almost all cases, climate change was based on Global Warming Potential model (GWP100) (IPCC, 2013). For some of the processes such
as Swedish electricity mix, we used information from Swedish literature, but also relied on generic LCA databases such as EcoInvent. To estimate the primary energy use, we used the inventory data and multiplied each by corresponding Primary Energy Factor (PEF). These PEFs were ob-tained from Swedish literature, but also from LCA databases using various Cumulative Energy Demand (CED) models (Frischknecht et al., 2015) to estimate the embodied non-renewable primary energy. Finally, to estimate the resource costs we used the inventory of the processes and the estimated cost of resources associated with them. The resource costs are proportional to the amount of substrates that is transported or treated, so, it is closely linked to the operating costs. However, it does not include investment costs. To get a realistic picture of the financial aspects of installing new biogas plants, site-specific, detailed, and comprehensive financial analysis is required which is beyond the scope of this study.
Models assumptions
For each model biogas plant, we created an attributional LCA model using a modeling tool developed by Feiz et al. (2020) but modified and adapted for this task. The biogas production, for relevant scenarios, was modelled based on a few assumptions regarding the operation of the biogas plant (e.g. yield and heat demand), flaring, slippage, and upgrading and compression of the biogas (Table SI-7).
To model the impacts of transportation, we assumed that excreta and digestate are transported by 30–35 tonne trucks. The trips were assumed to be one way, i.e. that return trips are allocated to other activities. All trucks were assumed to run on diesel with an average fuel consumption of 33.1 liters/100 km, and 4 liters/hour while idle (Green Truck
Part-nership, 2015; ICCT, 2017; Raaholt et al., 2011). The operating cost of
the trucks was assumed to be about 10 € /hour (cf. Johansson and
Nilsson, 2007; Raaholt et al., 2011). Each trip was assumed to include 1
hour idle time for loading and unloading. Consequently, the trans-portation cost factor we used was 0.12 € /tonne-km (for a 50 km
one-way trip this corresponds to 6 € /tonne) for transportation of excreta
and digestate with trucks, and about 0.73 € /tonne-km (for a 50 km
one-way trip this would be about 44 € /tonne) for transportation of the
produced upgraded biogas. For the two biogas scenarios, all biogas that is not used for internal heating of the plant, is upgraded to compressed fuel (upgraded to 98% methane) and is transported via trucks (in
composite cylinders) for 50 km where it is used as a substitute for diesel fuel (Table SI-8), but with 90% efficiency (Delgado and Muncrief, 2015). The other processes also required the use of assumptions. We assumed that the NPK content of manure, sludge or digestate that are applied as biofertilizers (after losses) replace mineral fertilizers that are commonly used in Sweden (Table SI-8). GWP was calculated based on the characterization factors reported by EU’s Renewable Energy Direc-tive (European Parliament, 2018), for instance 1 kg emission of CO2,
CH4, or N2O was calculated as 1, 25, and 298 kg of CO2-eq (GWP)
respectively. Primary energy use was calculated by converting the life cycle energy use for collection and conversion of excreta to biogas and use of the digestate as fertilizers. Direct energy uses were converted to primary energy use using Swedish PEFs (Table SI-9). The resource costs included the financial cost of obtaining required physical resources such as electricity and fuel (Table SI-9) and dummyTXdummy-(-transportation costs; expressed in euro (€) for the five processes
enumerated earlier. Costs related to investments, administration, and maintenance were not included.
3. Results
3.1. Pre-transportation situation
Although a large proportion of crop nutrient demand could be ful-filled by recycling manure and sludge in Sweden, without transportation it is impossible to properly utilize this potential. Circa 2015, 44% of crop N demand and all of P and K demand could be met with excreta na-tionally (Fig. SI-2). In fact, the total supply of P was equivalent to 113% and the supply of K 142% of crop nutrient demand. The majority of this excreta supply came from cows (Fig. SI-2). However, the spatial distri-bution of excreta on the landscape results in surplus and deficit areas, meaning that not all excreta nutrients could be used to meet crop de-mands (top panel Fig. 3). Without transport, between 15% (for N) and 40% (for K) of the 10 × 10 km2 grids in Sweden exhibit a nutrient
sur-plus (Fig. 3); meaning that most areas would have a nutrient deficit without the use of mineral fertilizers or excreta transport. However, these deficits are not always large; between 20% (N) and 50% (P) of the grids show an imbalance of only between +1 and -1 nutrient tonnes. In summary, without transport only 35% of N, 45% of P, and 71% of K crop demand could be fulfilled within grids (Table SI-10).
3.2. Optimization model comparison
Transporting excreta outside of 10 × 10 km2 grids would allow
Sweden to fulfill a larger proportion of its crop demand and the transport cost of doing so is different with and without biogas plants (Figs. 3, 4, Tables SI-10–SI-13). The majority of crop P demand (91%) and 44% of crop N demand could be met via transported excreta across the country. All N transported could be properly assigned to meet the demand of crops, but only 81% of the P transported would be “useful” to the grids where the excreta is shipped (Fig. 4). The redistribution of manure and sludge with and without biogas plants results in an almost identical capacity to fulfill crop nutrient demand, with a small difference for K (Table SI-10). Moving excreta through biogas plants instead of directly to agricultural fields results in 4% less K crop demand being met because 3% more of the transported excreta is overapplied (Table SI-10). Still, 77% of K crop demand is met in the BIOGAS MIXES scenario; 51% of the transported excreta would help to meet K crop demand. Overall, the optimization model smooths out extreme local areas of deficit or surplus (top vs bottom panels of Fig. 3).
The cost of transporting to and from biogas plants is, unsurprisingly, higher than sending excreta directly to crop demand grids (Fig. 5). Our optimization results demonstrate that, given the same constraints on nutrient application, the objective value (which is equivalent to trans-portation costs for comparison purposes) is approximately double the value of simply moving excreta from deficit to surplus areas. The
objective value increase is associated with a larger number of transports rather than a sizable difference in the distances traveled; the average transport distance is only 12 km longer in the BIOGAS MIXES scenario than the BASELINE scenario (44 km, Tables SI-11 and SI-13). Specifying three manure mixes (BIOGAS MIXES scenario in Fig. 5) as opposed allowing any type of manure to enter a biogas plant (BIOGAS GENERIC scenario) decreases the objective value slightly. By allowing for different manure types to “travel” separately from each grid, instead of being lumped together as one transportable unit (as in BIOGAS GENERIC), the BIOGAS MIXES model is actually relaxing a constraint making it easier to find demand grids to accommodate all the digestate.
Finally, the differences among scenarios are present both with and without considering sludge, even if objective values are slightly higher with sludge than without. Still, sludge, and the way it is used in the optimization model, contributes to Sweden’s capacity to meet crop
nutrient demands without mineral fertilizer. Among the three nutrients, the most striking difference is with regards to P, where an additional 20% of crop demand can be met by including sludge (Table SI-10). The transport cost is higher in scenarios where sludge is included as there is more material to transport, but the difference is much smaller than be-tween the costs of transporting excreta without (BASELINE) versus with biogas plants (BIOGAS GENERIC and BIOGAS MIXES).
3.3. BIOGAS MIXES scenario optimization
Based on our optimization model, the majority of biogas plants would be placed in the Southern half of Sweden and dominated by cow manure as an input (Fig. 6). The model places 105 MixA locations, which by definition are dominated by cow manure as an input. In areas where more pig manure is available, some MixB locations (39) are also Fig. 3. Nutrient balance maps for Sweden in tonnes of nutrient per 10 × 10 km2 grid. The top panel is pre-transportation circa 2015, and the bottom panel is the result of the Biogas Mixes optimization scenario. Values are expressed in tonnes of nutrient per 10 × 10 km2 grid. Blue represents a deficit (crop nutrient demand > manure+sludge) and red a surplus, where bold values along the y axis represent the cutoff values (tonnes) for each color bin. Small italic values (located between the bold bin values associated with the colors) represent the number of 10 × 10 km2 grids that have values in that bin. The left panels show the balance for nitrogen (N), the middle panels phosphorus (P), and right panels potassium (K).
used. In addition to incorporating the seven locations with large WWTPs in urban centers, the model used 90 fictive biogas plants to treat sludge. When excluding the sludge component, the model also assigned one biogas plant in the south of Sweden according to MixC, which is dominated by poultry manure. We will however focus on the scenario with sludge for the remainder of the study (see section above for justification).
The number of transports (connections between two grids) from manure supply grids to biogas plants is much larger than from biogas plants to grids to meet crop demand; however the transports to plants contribute less to the cost of transportation (Fig. 7). The BIOGAS MIXES model suggests that just under 10 000 transports would be required to move manure to all three types of biogas plants where each connection between a supply grid and a biogas plant would, on average, carry 1 622 tonnes of manure for 42 km (Table SI-13). In order to then redistribute nutrients to meet crop demand, the average transport would be longer (69 km) but also heavier (8 194 tonnes). For digested sludge, trans-portation distances would be the longest (mean 108 km), but with smaller loads (mean 813 tonnes).
Overall, the transportation of digestate from biogas plants towards agricultural fields tends to contribute most to transport costs.
3.4. LCA across scenarios
Processing excreta through biogas plants results in far more GHG emissions savings (lower climate impact) and less primary energy use
than if excreta are only used to meet crop nutrient demands (Fig. 8 and Table SI-14). Across scenarios and LCA metrics, the benefits of savings associated with mineral fertilizer substitution are almost identical, which is logical as a similar proportion of crop demand is met for all scenarios (Fig. 4).
The biogas scenarios both lead to the production of about 3.4 TWh of upgraded biogas (340 million Nm3). The differentiation of the biogas
plants in the BIOGAS MIXES scenario does lead to small changes (compared to BIOGAS GENERIC), but it is of relatively little importance for the overall LCA performance among scenarios. Biogas production, in either scenario, clearly leads to more activities and higher gross impacts (more resources needed). If we do not account for substitution effects, the two biogas scenarios have higher GHG emissions (by 20%), primary energy use (by 650%), and cost (by 200%) compared to the BASELINE. However, there are significant savings in all three areas due to the ca-pacity of biogas to substitute for fossil transportation fuels. When the net effects of the BIOGAS GENERIC and BIOGAS MIXES scenarios are considered then the GHG emissions (200%), primary energy use (150%), and the cost (300%) are all lower compared to the BASELINE (paren-thetical value represent this relative decrease).
The addition of biogas plants also changes the relative importance of processes considered in the LCA. For the BIOGAS GENERIC and BIOGAS MIXES scenarios, mineral fertilizer substitution only accounts for about one third of the savings across metrics. The other two thirds of the GHG emissions savings, for example, comes from the substitution of fossil fuels by the produced biogas. With regard to cost, the addition of biogas production only leads to a small improvement in absolute terms, but still represents a sizable improvement in relative terms (see above para-graph). The improvements in cost-efficiency of nutrient recycling related to biogas production are nevertheless important, especially if we note that they comes with notable environmental gains. Overall, biogas production reduces the net impact (the diamond shape in Fig. 8 and the last row in Table SI-14), particularly in climate impact and primary energy use.
4. Discussion
4.1. Nutrient and energy potential
Our findings on nutrient and energy availability in excreta resources across Sweden show, like other studies, that the potential for more cir-cular excreta management to replace fossil/mineral resources is large, but cannot replace all requirements under current consumption pat-terns. Our optimization model results indicate that the majority of P and K crop demands could be met through excreta recycling, but that farmers in many regions would still need to add N fertilizers or use biological N fixation to meet 54% of N crop demand (Fig. 4). This is similar to results we obtained from previous work (Akram et al., 2019b), which showed that in 2008 the majority of N and P demands could be met, and that K in excreta was in surplus of crop demand. The fact that previous work looking, at 2008, showed that 89% of P crop demand could be met by excreta recycling, but the circa 2015 results explored here show a slight surplus of excreta P compared to crop demand supports the idea that Sweden is close to a P balance of crop demand vs manure availability (Linderholm et al., 2012).
From an energy perspective, we estimated that full recycling could produce about 3.4 TWh of biomethane (340 million Nm3). This is the
same order of magnitude as another independent study which evaluated the potential for manure-based biogas plants in Sweden and estimated that about 1.8 TWh of biomethane (180 million Nm3) could realistically
be produced in the country (Scarlat et al., 2018). Even if Sweden pro-cessed all excreta through biogas plants, as we operationalize in our models, to produce upgraded biogas for transport, these 3.4 TWh would only cover approximately 4% of current transport sector energy use
(Swedish Energy Agency, 2019). However this amount of biogas would
still represent a sizable share (almost 18%) of current biofuels used for Fig. 4. Capacity to correctly transport N, P, and K in excreta and meet crop
nutrient demands in the BIOGAS MIXES scenario. Results are similar across scenarios and can be found in SI.
Fig. 5. Objective value results for optimization model scenarios. The objective
value is a representation of transportation cost because this value could be multiplied by a cost per tonne per kilometer transported. The value represents the minimum “cost” of transporting all excreta given the model constraints.
transport (19 TWh, Swedish Energy Agency, 2019), and is higher than Sweden’s 2030 goal for biogas production from manure (1.5–2.6 TWh, Norstedts Juridiks, 2019).
As Sweden works to meet both food system sustainability goals and circular economy goals, it seems likely that excreta, in addition to some organic wastes we have not considered here, will become even more important sources of energy and nutrients. Under current circumstances, manure remains one of the largest potential sources for recycled nutri-ents and reuse on agricultural lands actually has many co-benefits including soil carbon storage (Bolinder et al., 2020). Although manure is an important residual biomass for energy production, in Sweden forestry residues along with roadside vegetation and urban green area waste represent a very large energy potential (Hamelin et al., 2019), and food waste is a priority area in the circular bioeconomy (Carus and
Dammer, 2018). Accounting for these sources would represent
impor-tant additions to what was presented in this study. In fact, non-manure waste sources could become more important proportionally if nationally (and globally) people adopt more plant-based diets to meet a wide array of sustainability goals (Willett et al., 2019). Concurrently, the demand for crop nutrients would decrease because less animal feed would be required. A shift in human diets would also shift the amount and composition of food waste and sludge, which in turn may shift when and where they can be effectively recycled (Forber et al., 2020; Metson et al.,
2016a). Spatially optimizing land use (to match food, energy, and water
quality objectives), in addition to thinking about the highest value for waste, needs to be part of future scenario work (Femeena et al., 2018). Moving forward work should consider the effect of diverse in-terventions towards sustainability on the recycling of organic waste, Fig. 6. Location of different mix-type modeled biogas plants across Sweden with optimization model. Locations are indicated by black upward triangles (MixA),
black downward triangles (MixB), black and red triangles (WWTP) and a black square (MixC). The size of these icons indicates the number of plants at said locations. For WWTP locations, we differentiate between those locations that were pre-selected (red) and those the model selected (black). The top three panels indicate biogas plant locations when running the model with sludge, and the bottom panels are without sludge. The MixA plants are dominated by cow manure and are thus overlayed on a map of cow manure in grids (darker coloring indicates more manure in a 10 × 10 km2 grid). MixB plants are dominated by pig manure and are overlayed on a map of pig manure. MixC plant, dominated by poultry manure, was only selected by the model when ran without sludge; a single location was selected and it is overlayed on a map of poultry manure.
including those mentioned above, and more specific farming practice constraints. In the Swedish context paying close attention to organic production practices, and their effects on nutrient supply and demand, will be important. Swedish food actors have expressed a strong interest in expanding organic production to support a sustainable national food system (Karlsson et al., 2018). Current organic certification schemes do not accept human excreta as an input (Seufert et al., 2017), which could limit future food system circularity. The expansion of organic produc-tion could also shift which nutrient is in focus. A recent meta-analysis showed that organic farms often have a K deficit (Reimer et al., 2020); our results indicate that 73% of K crop demand could be met through recycling in the BIOGAS MIXES scenario (Fig. 4). If K is a larger concern for organic farms, an alternative model could be created to specifically optimize K application because nationally there is more K in excreta than there is crop demand (Fig. SI-2).
4.2. Realism of transport distances and plant locations
The mean transport distance associated with a biogas plant (inbound +outbound, 111 km for manure and 108 km for sludge, Table SI-13) is above common transport distances in Europe and the USA, but similar to what other studies consider necessary for sustainability. The type of substrate (and digestate) plays a role in how energy and economically
efficient transport can be (Bergh, 2013; Berglund, 2006). Survey re-spondents in Sweden have indicated that the real-world average trans-port distance for a biogas plant is around 38 km; where feedstock transport (average 29 km, up to 400 km) is longer than the distance to ship digestate to agricultural lands (average 9 km, up 35 km, Bergh, 2013). Our model results suggest the opposite: biogas plants, in order to minimize overall transport, were located closer to feedstocks and required longer transports of digestate towards demand grids (Fig. 7). The energy break-even point for processing cow manure through a biogas plant in Sweden is estimated to be 200 km (Berglund and
B¨orjesson, 2006), suggesting that constraints other than energy drive
transport practices because few real-world transports are that long. Results from the USA show similar distance ranges, where liquid and bulky manure is only moved within the ten’s of kilometers range while dryer, and more concentrated, organic waste is transported in the low hundreds of kilometers. For example, in Southern Michigan USA, 95% of manure produced in concentrated feeding operations was reapplied to fields within a 15 km radius, often in excess of crop nutrient needs (Long
et al., 2018). USA modeling studies to better manage nutrients to meet
crop needs present values much closer to the ones found in this study, between 100 and 400 km (Metson et al., 2016b; Spiegal et al., 2020). In Sweden, our previous work estimated that the average transport dis-tance to recycle all excreta nutrients circa 2008 was 123 km (5x5 km grids, Akram et al., 2019b); although adding biogas plants increases the number of transports required, it does not alter the necessity to move excreta from surplus to deficit areas on the Swedish landscape and as such distances are relatively similar.
Although plant locations were designed to minimize transportation both to and from biogas plants, it is the spatial distribution of animals that seems to drive the overall distribution of plants (Fig. 6). The con-centration of biogas plants in the Southern parts of Sweden is also visible in Scarlat et al. (2018)’s analysis of biogas potential across Europe at a 1 km2 resolution. Although, there is agreement on where plants “should”
be located, individual actors are not likely to minimize transport dis-tances (and thus costs and emissions associated with transportation) for both feedstock and digestate at a national scale. In practice, choosing the location of individual biogas plants is influenced by several factors.
Our model does not optimize for one biogas plant, but for all plants nationally. In contrast, it is more likely that a biogas producer will choose a location that minimizes feedstock transport for an individual plant (given the current landscape of farms and biogas plants). As more and more plants are built, it will become harder to find a location with inexpensive transport to both feedstock supply and digestate demand areas. In other words, coordinated effort among various actors, notably national government agencies, will be necessary to minimize the transportation obstacle and take full advantage of excreta resources through biogas production.
Still, failing to account for regional or individual stakeholder needs, as well as logistical factors other than transport distance, can also impede scaling up biogas solutions. Taking full advantage of the biogas production potential in Sweden is challenging because different biogas systems are in place for different reasons; failing to account for different regional goals and societal preferences will likely hinder the expansion of biogas in the county (Olsson and Fallde, 2015). One could consider the existing network of relationships between stakeholders that ex-change organic materials along the food chain in order to identify where there are realistic and desirable leverage points for changes in man-agement. Still, physical proximity between actors is one of the key pa-rameters affecting the flow of materials (e.g., in a French regional agro-food network, Fernandez-Mena et al. (2020)), and our spatially explicit approach does indirectly account for this fact by minimizing transport costs.
In addition to the socio-economic system factors described above, there are also biophysical factors that would need to be further considered in order to increase the realism of the excreta recycling scenarios we presented. From a nutrient perspective the disconnect Fig. 7. Contribution of different types of transport to the total transportation
cost in the BIOGAS MIXES scenario. The distance, number of transports, as well as the weight of each transport all contribute to the objective value of the model and thus to cost. The top panel depicts the number of individual transport events for three different types of transports while the bottom panel shows the mean weight × distance of they transports. Note that a transport event is the number of connections between two grids and not directly translatable to the number of trucks that might be needed to link the amount of excreta in a grid to its destination.
between the timing of excreta production, crop demand, and hydro-logical losses, as well as the N:P:K ratios of excreta vs. crops can be barriers to effective resources use. Crops need larger amounts of nutri-ents in the spring, while humans and animals produce these nutrinutri-ents year-round. In fact, the collectable manure (not naturally excreted on pasture) we account for in our models is produced in the winter when there is no/little crop demand. Storage is thus necessary. Because we used total annual supply and demand values, we are assuming that such storage occurs. However we do not explicitly model how storage and transportation would need to change throughout the year to make-up for differences between supply and demand. In addition, our models ac-count for excreta stoichiometry and thus that N:P:K come as a bundle, but did not incorporate nutrient separation technologies that could overcome this fact to utilise the three nutrients separately. In the Upper Yahara watershed in Wisconsin USA, where nutrient induced harmful algal blooms are a problem, logistical modeling that accounts for pre-cipitation patterns, the need for seasonal storage, and separation tech-nologies has demonstrated that improved organic waste management can reduce water quality impairment (Hu et al., 2019). As such, factoring in the costs (and benefits) of storage, excreta application timing, and additional nutrient separation after biogas production would be a next steps to expand the model presented here.
4.3. Biggest contributors to savings
Overall, our results indicate that processing excreta through biogas plants and transporting digestate to agricultural land to meet crop de-mands is a net benefit for selected economic and environmental
indicators (Fig. 8). In our previous work, the monetary value of mineral fertilizer substitution could not cover the cost of transportation in Sweden (Akram et al., 2019b). Here, circa 2015, our results indicate that, even for the BASELINE scenario, the substitution value creates a net positive financial resource outcome (operating costs, focusing on resources that are used in the transportation activities, and production and sale of biogas and biofertilizer), not including investment costs. This more favorable picture is likely because we decreased the average cost of transport per tonne km from 0.2 € to 0.13 €. The lower value assumes
that return truck transports are not empty and thus not the financial burden of the same farmer (or biogas plant). In other words, reducing the cost of transport makes the fertilizer substitution value more likely to cover the cost of transport.
When considering the addition of biogas plants to nutrient recy-cling, it is the substitution of fossil fuels that compensates for the extra transport and handling costs, energy requirements, and GHG associ-ated with the plants. In fact, for climate change and primary energy use metrics, this additional fossil fuel substitution benefit very largely out-weighs the additional resource use compared to the BASELINE scenario. If one were to use a multi-objective optimization model, or one with additional constraints around feasibility or pollution, it is possible that the total and relative benefits associated with full excreta recycling would change. Our optimization and LCA model assumptions do not cover all expenditures associated with complete excreta recy-cling (e.g. investment costs), but our results do show there are many benefits to doing so.
Due to the relatively large scope of the paper, we have excluded uncertainty analysis. However, we have tried to choose realistic, and Fig. 8. Life cycle impacts of the studied scenarios. Top-left: climate impact, top-right: primary energy use, bottom-left: resource cost. The impacts are shown by
positive values (above the zero line), and the savings through substitution effects by negative values (below the zero line). The net impacts are shown by the diamond shapes. Actual numbers are presented in SI (Table SI-14).
sometimes even conservative, assumptions. With regards to net savings from GHG and primary energy perspectives in biogas scenarios, we think our results are robust, because of the very large margins. The benefits of biogas production from manure have been demonstrated in other studies as well (Lantz and Bj¨ornsson, 2016; Tufvesson et al., 2013). With regards to costs the margins are lower and it could be the case that in some variations, biogas production scenarios will not lead to a net benefit compared to the BASELINE (as noted with transport costs earlier). Therefore, we are drawing more cautious conclusions with regard to cost-saving.
Finally, the financial benefits of resource recovery in a circular economy are not enough to transform organic waste management sys-tems. Yes, as we show here, there are economic advantages, but many of the other benefits (e.g., climate change savings) are not monetized, and there are many social norms (e.g., surrounding human excreta reuse) that are not accounted for in current economic models. The lack of profitability of proposed solutions, even when considering multiple re-sources, has been clearly demonstrated with regards to sanitation
(Mallory et al., 2020; Trimmer and Guest, 2018). Even where there is an
unequivocal need for better waste management, transformation is challenging. It is important to acknowledge that political and social change are required to gain the full benefits of a circular biobased economy. These required changes include aspects that are often over-looked in circular economy discussions like social equity and justice (Moreau et al., 2017).
A holistic approach to sustainable organic waste management is needed, and our results point to the fact that coordinated national level actions are likely required.
5. Conclusions
Our findings suggest that policies and practices that support coor-dinated logistical planning for nutrient recycling and biogas production can help maximize benefits at the national scale. Although the N:P:K amounts, and ratios, from excreta cannot fulfill all of Sweden’s crop needs, moving these some of nutrient resources across the country can help meet crop nutrient demands, and in some locations reduce the risk of nutrient-induced pollution. The added benefits of biogas production, as opposed to only reusing excreta for its nutrient value, could outweigh the additional costs that come with transporting substrate to, and digestate from, biogas plants. The most important benefits of adding biogas plants is green energy production and its associated GHG emis-sions reduction. These results are underpinned by an a national opti-mization model focus on transport; without coordinated action among diverse stakeholders, including government, farmers, and biogas pro-ducers, the full value of the sustainability benefits estimated here will not be realized. In order to move from theory to action, bridging the gap between national assessments like this one, to local constraints and objectives will be needed. One way forward would be to work with distributed case study areas across Sweden and use locally adapted cost and benefit values, including investment costs and higher temporal and spatial resolutions for selected variables.
CRediT authorship contribution statement
Genevi`eve S. Metson: Conceptualization, Methodology, Validation, Visualization, Supervision. Roozbeh Feiz: Conceptualization, Method-ology, Visualization. Nils-Hassan Quttineh: Conceptualization, Meth-odology, Visualization. Karin Tonderski: Conceptualization, Methodology, Validation, Funding acquisition.
Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Acknowledgements
We would like to thank Uno Wennergren for helping seed the ideas of this work as well as securing funding. We thank Usman Akram for his work compiling many of the datasets used, and Anton Sundblad for independently validating the processes by which these datasets were compiled. This work was supported by the Swedish Council for Sus-tainable Development (Formas-942-2016-69) and the Biogas Research Center (BRC) at Link¨oping University.
Supplementary material
Supplementary material associated with this article can be found, in the online version, at 10.1016/j.rcrx.2021.100049
References
Akram, U., Quttineh, N.-H., Wennergren, U., Tonderski, K., Metson, G.S., 2019. Enhancing nutrient recycling from excreta to meet crop nutrient needs in sweden a spatial analysis. Sci. Rep. 9 (10264) https://doi.org/10.1038/s41598-019-46706-7. Akram, U., Quttineh, N.-H., Wennergren, U., Tonderski, K., Metson, G.S., 2019.
Optimizing nutrient recycling from excreta in sweden and Pakistan: higher spatial resolution makes transportation more attractive. Front. Sustain. Food Syst. 3 (50)
https://doi.org/10.3389/fsufs.2019.00050.
Ammenberg, J., Feiz, R., 2017. Assessment of feedstocks for biogas production, part II results for strategic decision making. Resour. Conserv. Recycl. 122, 388–404.
Andersson, P.-G., 2015. F¨altf¨ors¨ok med kommunalt avloppsslam från Malm¨o och Lund under åren 1981–2014. Technical Report Rapportserie 17. Hushållningss¨allskapet. Arvidsson, R., Svanstr¨om, M., 2016. A framework for energy use indicators and their
reporting in life cycle assessment. Integr. Environ. Assess. Manag. 12 (3), 429–436.
https://doi.org/10.1002/ieam.1735.
Astill, G. M., Shumway, C. R., 2016. Profits from pollutants: economic feasibility of integrated anaerobic digester and nutrient management systems 184, 353–362. 10.1016/j.jenvman.2016.10.012.
Ayres, R.U., Ayres, L., 2002. A Handbook of Industrial Ecology. Edward Elgar Pub., Cheltenham, UK; Northampton, MA.
Bartoli, A., Hamelin, L., Rozakis, S., Borze¸cka, M., Brand˜ao, M., 2019. Coupling economic and GHG emission accounting models to evaluate the sustainability of biogas policies. Renew. Sustain. Energy Rev. 106, 133–148. https://doi.org/10.1016/j. rser.2019.02.031.
Bergh, A., 2013. Factors for Success in Swedish Biogas. A Study Based on the Swedish Agricultural Sector. Swedish University of Agricultural Sciences, Uppsala, Sweden. Master’s thesis.
Berglund, M., 2006. Biogas production from a systems analytical perspective. Environmental and Energy Systems Studies, Lund university.ISBN: 9789188360809 Berglund, M., B¨orjesson, P., 2006. Assessment of energy performance in the life-cycle of biogas production. Biomass Bioenergy 30 (3), 254–266. https://doi.org/10.1016/j. biombioe.2005.11.011.
Bolinder, M.A., Crotty, F., Elsen, A., Frac, M., Kism´anyoky, T., Lipiec, J., Tits, M., T´oth, Z., K¨atterer, T., 2020. The effect of crop residues, cover crops, manures and nitrogen fertilization on soil organic carbon changes in agroecosystems: a synthesis of reviews. Mitigation Adapt. Strateg. Glob.Change 25, 929–952. https://doi.org/ 10.1007/s11027-020-09916-3.
B¨orjesson, P., Mattiasson, B., 2008. Biogas as a resource-efficient vehicle fuel. Trends Biotechnol. 26 (1), 7–13. https://doi.org/10.1016/j.tibtech.2007.09.007.
Carlsson, M., Uldal, M., 2009. Substrathandbok f¨or Biogasproduktion. Technical Report. Svenskt Gastekniskt Center.
Carus, M., Dammer, L., 2018. The circular bioeconomy—concepts, opportunities, and limitations. Ind. Biotechnol. 14 (2), 83–91. https://doi.org/10.1089/
ind.2018.29121.mca.
Cplex, I.I., 2009. V12.1: user’s manual for cplex. Int. Bus. Mach. Corp. 46 (53), 157.
De Schoenmakere, M., Hoogeveen, Y., Gillabe, J., Abdullah, S., 2018. The Circular Economy and the Bioeconomy: Partners in Sustainability. Technical Report EEA Report No 8/2018. European Environmental Agency.
Delgado, O., Muncrief, R., 2015. Assessment of Heavy-Duty Natural Gas Vehicle Emissions: Implications and Policy Recommendations. Technical Report. ICCT.
Tufvesson, L., Lantz, M., Bj¨ornsson, L., 2013. Milj¨onytta och samh¨allsekonomiskt v¨arde vid produktion av biogas från g¨odsel. Technical Report, 86. Environmental and Energy Systems Studies, Faculty of Engineering, Lund University.
van Dijk, K.C., Lesschen, J.P., Oenema, O., 2016. Phosphorus flows and balances of the European Union Member States. Science of The Total Environment 542, 1078–1093. ISBN: 0048-9697
European Parliament, 2018. Directive on the promotion of the use of energy from renewable sources (recast).
Feiz, R., Johansson, M., Lindkvist, E., Moestedt, J., Påledal, S.N., Svensson, N., 2020. Key performance indicators for biogas production—methodological insights on the life- cycle analysis of biogas production from source-separated food waste. Energy 200.
https://doi.org/10.1016/j.energy.2020.117462.
Femeena, P.V., Sudheer, K.P., Cibin, R., Chaubey, I., 2018. Spatial optimization of cropping pattern for sustainable food and biofuel production with minimal
