Large technical systems in shrinking municipalities – Exploring system reconfiguration of district heating in Sweden

With this paper we aim to contribute to the field of sociotechnical studies, and more specifically Large Technical Systems (LTSs) to understand what happens in mature systems and how different segments develop. We argue that there is a need to discuss LTSs more heterogeneously to fully grasp how phases of stagnation and drift play out differently. Our focus is on district heating in Swedish municipalities with a shrinking population. As such district heating is an important system for climate measures in Europe and in Sweden it has a long tradition of municipal ownership and strong development. The emphasis is on the ways different district heating systems evolve and how various segments of an LTS develop. We have used a mixed-methods approach and the empirical material in this study builds on municipal documents and annual reports of energy companies, an online survey to energy companies, and qualitative interviews with representatives from energy companies and municipalities. Our results show that private companies are overrepresented in small and shrinking municipalities, and that they develop their business models around small systems to keep them profitable. Instead, municipal owned companies face several challenges concerning capacity, requirements, and economic profitability. From an LTS-perspective, the analysis shows how different actors in the segments adopt different strategies to manage stagnation, leading to segments drifting apart.


Introduction
Research on Large Technical Systems (LTS) stretches back to the 1980´s, starting with studies on the dynamics of system establishment and development, and especially how this tend to occur in phases [1,2].In later studies, focus has been on aspects of system change [3], and in a recent article by Sovacool, Lovell, and Ting [4] they discussed how mature systems go through phases of reconfiguration, contestation, and in some cases stagnation and decline (cf.[5,6]).In most of these studies, the analytical viewpoint is to analyze systems, for example gas systems, district heating (DH), or electricity systems, on a macrolevel, as one LTS.This can be in terms of geographical scale, to analyze systems on a national or regional level [5], or specific case studies focusing on one or more systems [7].
What is often lacking in the literature are thus concepts and perspectives to understand how different segments of an LTS may develop in diverging directions.For example, the electricity market consists of many different actors, from large multi-national system owners to micro-producers, that together constitutes an LTS.Thus, we argue that it is important to develop analytical sensibility to diverging development paths within an LTS, as it shows tensions, different needs, imbalances in power, and system actors drifting apart.The emphasis in this paper is on the Swedish DH-system, a mature LTS developed in the 1940´s, for heating of dwellings and non-residential premises, with production of heat in a production plant and delivery of hot water through pipe systems underground to customers.In Sweden, DH holds 58 % of the heating market, based on energy usage [8].We consider DH as fitting to study from an LTS-perspective, due to large sunken costs, a clear technical core, and a heterogenous set of actors operating within the systems.The development pattern follows the phases often identified in the LTS-literature, and as we argue that DH is in a stagnation phase, it is suitable for studying development in various segments.
In previous studies, several challenges for DH have been pointed out.These are especially warmer winters following climate change, more energy efficient buildings, and competition from heat pumps [5,9].The challenges, and the proposed measures for addressing them, we argue, differ depending on the prerequisites, and the main differences are between large and small systems, and private or municipal ownership of the systems.Most studies on DH tend to be technical and focus on larger cities and municipalities [10][11][12][13] and, in recent years, on how DH can help to reduce greenhouse gas emissions, not only due to biomass usage, but especially in relation to evolution of bio energy carbon capture and storage (BECCS) [14][15][16].Compared to e.g.Finland [17], the Swedish DH market has gone through a decarbonization process over the last decades [18].In this paper we mainly focus on small DH systems in Swedish municipalities with shrinking populations, as they seldom get the same attention as larger systems.Still, they are equally important to analyze diverging developments of LTSs.An additional important argument to study small municipalities relates to the research field of shrinking cities [19,20], as small towns and municipalities in Sweden are losing population, and have been so for a long time.This leads to challenges in terms of aging populations, decreased opportunities for service, and scarce economic resources for the municipalities [21].In terms of infrastructure, most studies have focused on management of water and sewage systems [22][23][24] and in only one instance DH has been covered [25].
DH in Sweden was first established and developed in the late 1940´s under municipal ownership [26].The municipalities are an important part of the Swedish regulatory structure, as much powerregarding, for example, planning and land use policy, infrastructure management, child and elderly careis decentralized to municipalities.As the municipalities also have self-governing and taxation rights, this gives them a high degree of sovereignty but also considerable responsibilities [27].DH exists in 285 of Sweden's 290 municipalities [9], but with significant differences between them.Stockholm is by far the municipality with the largest population, with almost 1 million inhabitants.However, a majority of all municipalities are small (the average population size is 35,791, and the median is 16,080) [28], with small central towns where DH is normally placed.Furthermore, most of the plants in these systems are not combined heat and power plants (CHP) but use biomass and occasionally solid municipal waste as fuels, which is an important difference compared to DH systems in larger municipalities.DH is a system where economies of scale and economies of scope play a crucial role [7], but in Sweden most systems are small with limited markets.This puts extra pressure on the development and expansion of the core business, which is heat delivery to residential buildings [5,26].
Non-technical research on DH tends to focus on regulatory concerns [29][30][31], market development [5,9,26,32], environmental policy and impacts [18,33], or case studies focusing on the development of specific systems [7,[34][35][36].These studies often target the national level, or larger systems and cities and not smaller systems in municipalities with decreasing populations.By taking on a sociotechnical perspective [37,38], and to make a theoretical contribution by analyzing development in different parts of an LTS, we analyze the organizational, economic and social aspects of systems, to understand how system development is affected by, in this case, larger external trends.
Thus, the aim of the paper is twofold, the first being analytically driven and the second to make an empirical contribution to an underresearched part of the energy system.The aim is thus to analyze how different parts of an LTS develop in diverging directions, and to analyze trends and challenges for small district heating system owners in small, shrinking Swedish municipalities.
The paper will answer the following research questions: • How can we understand system development in various segments of a mature LTS, and more precisely in small energy companies?• Which challenges and opportunities have small DH-systems faced in recent years?
The paper is structured as follows; first the theoretical framework is presented, followed by a description of used methods and data; then the empirical results, and finally a concluding discussion.

Theoretical framework -Reconfiguration of large technical systems
The first LTS studies by Thomas P. Hughes focused on similar characteristics in development of electricity systems, and how they developed in phases.The names of the phases varies between publications, but they include invention and development, technology transfer and expansion, reverse salient, and consolidation and momentum [1,37].An overarching understanding is how involved actors operate around a technical core, being it an electricity system or a communication technology, and in various phases meet different challenges that needs to be handled for the system to become an LTS.The challenges can be political, economic, social, organizational, or, of course, technical, but the interplay and dynamics in the various phases are crucial for the understanding of system development [2,7].
Other authors have given more attention to the later phases of system development when systems have reached momentum.The argument in LTS-studies is that when a system have reached momentum they are difficult to change due to the sunken cost, large amount of involved actors and its integration in society [1].However, some focus has been on change and reconfiguration, starting out with the book by Summerton [3], identifying that change occur incrementally, and has occurred over long time, as systems cross territorial borders and become integrated with other systems.Also, linkages between different types of systems with heterogeneous functions occur, as well as change through reorganization from monopoly systems into adjusting to principles of competition and economic profits.
A phase that has been given less attention is that of stagnation [6].When challenges for system development reaches a critical point, it might lead to a haltered expansion, which can be geographically or in terms of demand, and thus the system reaches stagnation.It does not have to lead to a decline, as it may remain in a steady state.An example is the Swedish electricity system that reached a plateau of demand in the mid-80´s, but never went into decline [39].Important factors are how the involved actors see the system's future and if there are alternative usage and value, or if the competition from other systems is too strong.In the gas market in Sweden, competition from electricity led to a decline in the early 1900s [40].A stagnation in demand does not necessarily lead to system stagnation, as it is important to take other aspects into consideration.Magnusson [5] showed how district heating demand in Sweden is stagnating, something involved actors agree on, but when considering economic as well as network development, it does not point towards a stagnation as such, at least yet, and certain younger systems are still expanding.The electricity demand stagnation signifies similar things, as no one can argue that the whole system is declining, it is rather that development happens in other aspects, such as new applications, or a decarbonization of the production.Stagnation needs to be studied in detail, across various system segments, in order to understand if several factors point towards decline, or if it is still a dynamic system development, but without the constant demand growth that was focused in earlier LTS -studies.
This theoretical background is important to understand how LTSs have been analyzed, and still are to some extent.It is fair to say that in the recent decade focus has shifted towards various forms of system reconfiguration, especially in adjacent theoretical fields that start out from a sociotechnical perspective.For example, studies regarding sociotechnical transitions and the multilevel perspective, focusing on how systems change, in particularly towards becoming decarbonized, as new technologies (niches) and landscape changes (defined as external structures that are changing slowly, e.g. oil prices, economic growth, wars, broad political coalitions, cultural and normative values, environmental problems), put pressure on the existing system structure (referred to as the regime) [41,42].
Sovacool, Lovell, and Ting [4] focus on system reconfiguration of LTSs.They analyze system development in mature LTSs in three phases: reconfiguration, contestation, and stagnation and decline.In a phase of D. Magnusson and I. Grundel reconfiguration, the system face challenges, but adaptation occur, and system control remains stable.In contestation phases, the stakeholders' understanding of systems becomes fragmented and clash, and can lead to a negative impact on the system, and decline.They propose two mechanisms that might lead to contestation: drift and crisis.
Sovacool, Lovell, and Ting [4] argue that drift was initially used to describe how incumbent actors in sociotechnical systems make efforts to maintain status quo, despite significant shifts in contextual relevance.Further, in situations of drift, change and system reconfiguration occur in smaller steps, away from public oversight, and constantly adjust to be suitable to the present context.However, it can also entail clashes between groups of actors with divergent interests that might lead to battles or contest among relevant social groups for control over the system.For example, new technologies or political agendas may contest the incumbent system configuration, such as the British telecommunications industry, where actors pushed for deregulation and restructuring [4].
In crisis phases, an LTS is affected by a major accident or external event that could lead to a rapid contestation.For instance, a natural disaster or accident that shows weaknesses in the system or a need for reconfiguration could lead to a substantial reorganization of the system [4].
Sovacool, Lovell, and Ting [4 ,p.1079] summarize the ideas of contestation as follows: To recap: contestation refers to the challenge of control over the system, where entities dispute, contest, compete, and contend some aspects of LTS functioning, placing them in stasis or jeopardy.Decline refers to when LTS exhibit stagnation or declining growth, often because control over a system is lost.
For our study, aspects of system reconfiguration are central.However, most studies tend to treat a system on a macrolevel with little nuance regarding how various segments of the system develop.Sovacool, Lovell, and Ting [4 ,p.1069] argue that LTS contain social and technical elements that are diverse, that they have "varied technical infrastructures that organize interactions among diverse actors and technologies to accomplish a variety of purposes".However, when discussing certain mature systems today, it is almost impossible to state that the whole system is moving into one certain direction.In the quote above, the main focus is on the "control" of systems, which was a major part of the early LTS-studies, and on how the system builders enrolled actors in order for the system to develop [37,38].Today, the situation is more complex.External pressures affect different actors and systems differently, and liberalization processes have led to much less of control over the system, and certain solutions to meet the challenges might not be applicable for all actors.In this paper, we argue that DH in Sweden can be considered as one LTS, as technical characteristics are similar throughout the various systems and they operate under the same legislation, but we want to analyze and understand the different dynamics of specific segments.The segments can themselves be considered LTSs, as can individual systems, as they operate under similar logics, so we want to bring attention to the importance of analyzing different segments and what it does with the analysis, much like the nesting doll metaphor used in the sociotechnical transition-literature [41].System reconfiguration may look differently between the segments.

Methods and material
The empirical study builds on mixed-method approaches, i.e. we combined survey data collection and qualitative research [43].We also used secondary data through statistical databases as basis for case selection, descriptive background and contextualization.Sovacool, Axsen, and Sorrell [43] argue that mixed methods approaches, often through a mix of quantitative surveys with qualitative interviews, is a novel design in energy research within social science, for example in order to triangulate the material and give more sophisticated answers to research questions.
The data used consists of data from both national and local levels, as we wanted to analyze different segments of DH.It contains both quantitative and qualitative data.At first, an online questionnaire was sent to 76 energy companies in Sweden, mainly to CEOs or persons in the organization in central positions, in municipalities with shrinking populations.The selection of energy companies were based on three criteria: I) The municipalities in which they owned and managed DH systems were municipalities with negative population development (i.e.shrinking municipalities cf.[19,44]), based on a definition by Wolff and Weichmann [45,p. 130], who have described three categories of shrinkage: • Continuous shrinkage: Population decline of >0.15 % p.a. under the studied time period.• Episodical shrinkage: Population decline of at least 0.15 % p.a. over the studied time period, with a stable or even growing population (> − 0.15 % p.a.) in at least one five-year period.• Temporary shrinkage: Municipalities have lost at least 0.15 % of their population p.a. in at least one five-year period, but not over the entire period.
We included all municipalities that had continuous and episodical shrinkage between 1968, which is the first year with comparable population statistics, and 2018.
II) The second criterion was based on the population development in the main town ('centralort' in Swedish), using the same criteria as above, of continuous and episodical shrinkage.III) The third criterion was based on the development of the DH delivery.As the public statistics for DH delivery are seldom comprehensive, due to lack of reporting and that certain systems have been interconnected, the longest time period that could be studied ran from 1996 and forward.However, the statistics are more comprehensive from 2009 when energy companies started to report statistics concerning DH production and economic results to the Swedish Energy Markets Inspectorate.Therefore, we took the longest time period available for all systems, and identified a trend in delivery, and divided it into three categories: decreasing delivered DH, stagnating delivered DH, and increasing delivered DH.Even though energy companies with increasing delivery could have been excluded, we decided to keep them in the sample, as it is interesting to understand the reasons for these contradictory trends.This applied to only four cases, and often the explanation was that it was a rather newly built system.
In total the questionnaire contained 36 questions including subquestions, but it focused on five overall themes: how the demand for DH has developed over time, if population decline in the municipality had effects on the DH business, if there were any effects of reduced demand for DH that had been identified (technical, economical, organizational), if there were any measures taken and/or planned for in order to develop the business, and what future outlooks they had for the development of the system.The response rate was 41 %, which was lower than expected, but much research points to response rates being low in general if they are not combined with, for example, economic incentives.Research also indicates that response rates for online surveys have gradually declined in recent decades [46].
After the survey, the larger sample of municipalities was narrowed down to conduct in-depth studies.We selected 19 municipalities across Sweden with a long-term population decline.Another important selection criterion for the sample, besides geographical spread and population loss, was to achieve a mix between municipally owned DH companies and privately owned companies.
After deciding on the smaller sample, we collected and analyzed annual reports from the companies operating in the municipalities and annual reports from the municipalities studied, as well as their comprehensive plans ('översiktsplan' in Swedish), which are documents in which the municipalities present future planning visions on a strategic level.We analyzed reports that covered the five-year period from 2014 to 2019, amounting to a total of 200 documents.We also collected and analyzed additional quantitative data from various sources, both among the sample but also at the national level.Example of additional data include long term data on heat delivery as well as data on pricing in the chosen localities (e.g. from the Swedish Energy Agency, the Swedish Energy Markets Inspectorate, Swedenergy, and the annual Nils Holgersson pricing survey).
In total 53 semi-structured interviews were conducted, out of which 36 with municipal representatives (municipal managers, urban and regional planners, technical personnel, and environmental planners), 14 with representatives from energy companies (CEOs, business area managers, technical managers), and 3 with experts in the DH sector.The interviewees were selected based on communication with the municipalities on whom would be most suited to answer the questions we were interested in.The purpose was to gain a deeper understanding of the findings from the survey and document study, but also get a better grasp of the challenges as well as opportunities faced by district heating companies operating in shrinking municipalities.The questions followed a similar structure as the survey, but was different depending on the interviewee (more DH focus with energy companies and more focused on the municipality with municipal representatives) but went into detail regarding technical characteristics, challenges in terms of organization, economy, innovation, decreasing population, cooperation with the municipality/energy company, recruitment of staff, and future outlooks.The interviews were conducted between February and December 2020, and due to the Covid-19 pandemic they were all held remotely with the help of online tools such as Zoom or Microsoft Teams.We experienced that it worked well, as we could still see body language and face expressions, unlike in telephone interviews.It also was an effective measure as it saved time and travel and meant that we could cover more municipalities than initially planned.The interviews were manually transcribed, and the analysis was conducted based on carefully reading through the transcription several times, coding the material in Atlas.ti, and identifying commonalities, as well as differences, based on the themes presented above.
The research design has several strengths, especially through the broad approach to methods and by basing the material on a large sample of the energy companies in Sweden in general, and all energy companies in shrinking municipalities.By conducting interviews in 19 municipalities all over Sweden, all of them shrinking, we argue that we have received valid and representative data to understand challenges for small energy companies, in this segment of the DH LTS.However, limitations lie in the response rate to the survey and the fact that all municipalities and DH systems are quite unique in themselves, and we have thus been careful in our conclusions.

District heating in Sweden
In this section we will present a background to the Swedish DH sector from the perspective of DH as a national LTS, as well as descriptive data of importance for our study.

National development of DH
DH use in Sweden has increased significantly since 1970.This is a result of a strong development in municipalities with a focus on DH in oil reduction plans, and climate impact reductions [26].During the Swedish modernization process following WW2, focus was especially put on social reforms and the construction of the Swedish welfare state, in which infrastructure investments and development were integral parts.Goals of modern buildings and homes were substantiated with running water, kitchens of a certain standard, as well as central heating, which often meant DH [34].At this time, it can analytically be easier to understand DH as a national LTS.The Swedish national organization for DH started in 1949 as a way to coordinate and mainstream the implementation of DH in the largest municipalities [47].In this phase, technological development and technological transfer phases were apparent from a macro-perspective, but all municipalities adopted and adjusted the technologies to fit their city's needs, meaning that municipal LTSs with their own individual logics were developed.
In Sweden, small systems utilize heating plants, and not combined heat and power (CHP), even though the reason for mobilizing actors in the 1940´s was a focus on CHP [47].On a national level, 46 % of produced heat comes from biomass, 21 % from municipal solid waste, industrial excess heat 8 %, and around 2 % fossil fuels (oil or coal) and the rest is a mixture of sources (e.g.heat pumps, and different forms of gas).In most municipalities there is only one DH network (with a few exceptions), often with several different plants.There are differences between all systems, as large systems might have a CHP and a mixture of waste, biomass, and some smaller oil burned plants to be used during peak demand.In smaller systems there is generally one biomass plant, and one oil plant for peak hours.The oil plants are usually remaining since the times of building the system, when main fuels were oil [26], and it is safe to say that system reconfiguration has occurred by de-carbonizing the production in practically all systems.
Usually, DH is delivered to >90 % of multi-dwelling buildings in Swedish municipalities, 60 % to non-residential buildings, and only to a smaller share to one-or two-dwelling buildings (around 10-20 %), depending on expansion strategies.Multi-dwelling and non-residential buildings have historically been, and still are, the most important market segments, due to economy of scale, as one connection gives a high heat load compared to single houses that require a more fine-grained grid and thus high investment costs.During the earliest development phases of technological transfer and expansion (especially during the 1950's to 1970´s), municipalities first installed DH in municipal buildings, like city halls and schools, and in the municipal housing companies newly built houses, in order to both expand the energy company and housing company in a coordinated fashion [7,26,34].
Business models among energy companies are to a large extent similar, with the same company owning production facilities and network, with a few exceptions, also selling the heat directly to the customers.However, in multi-dwelling buildings the fee for DH is often included in the rent, which tends to make the actual heating system invisible for its customers.Other customers, like in one-or-twodwelling buildings or non-residential, sign a deal with the DH company directly, and it is basically up to the company to decide which new customers to connect and not [26,34].In later sections we discuss how ownership might matter for strategies and how ownership has changed.

Deliveries and economic results
The trend in recent years shows a stagnation in terms of demand on a national level, as usage has been leveling off due to warmer winters, competition from heat pumps, saturated markets, and new market conditions [5].Fig. 1 shows the development of DH usage in TWh between 1970 and 2020.Usage is defined as the actual usage of heating, which should be distinguished from production, as heat losses are subtracted when presenting usage, and it does thus more clearly reflect the actual usage and the fluctuations that occur due to outdoor temperature and energy efficiency measures.After an increase in usage until early 2000s, the trend shows a leveling off, although with pikes during cold years, like 2010.However, this macro perspective only gives one part of the picture.
An important aspect of the development of DH in the Swedish context is to understand the differences between different types of DH companies.A clear pattern is that the oldest systems in Sweden are in the largest municipalities (measured by population).The correlation between starting year and population is − 0.44 -meaning the larger population, the older the system.Fig. 2 shows a graph of the pattern, discounting the five largest and smallest municipalities in terms of population, in order to make the pattern clear (for example Stockholm, with 977,000 inhabitants, distorts the graph, but the year of system establishment was between 1953 and 1961 for the five largest and between 1971 and 2002 for the five smallest).Some trends are clear: all D. Magnusson and I. Grundel municipalities with over 80,000 inhabitants started DH systems before 1990, and a significant number of systems also started after 1980 [7].
There is no significant difference in the development of DH-usage between large and small systems over the last years.Larger municipalities, in terms of population, have older systems, see Fig. 2, and have had a positive population development (see Table 1), while the opposite is in general true for the municipalities with small populations (see Table 2).DH-deliveries always fluctuate depending on the temperature, but a general trend is that most of them have a negative trend (see Tables 1  and 2).
Based on the knowledge of Swedish shrinking municipalities, and on knowledge of different years of starting system based on municipal sizes, we have continued the analysis to distinguish between shrinking (i.e.population-wise small) municipalities, and non-shrinking, with specific focus on the former in order to understand system development dynamics in these LTS-segments.Table 1 shows the economic results in the 25 largest municipalities (except Lund since it was not possible to find statistics for Lund since 2009 and Nacka as it is not possible to distinguish statistics for the DH-system in Nacka as it is owned by Stockholms Exergi).All of them have shown strong economic results and most of  Table 2 shows the results for the 25 municipalities with the smallest population (except Storfors where it is not possible to find comparable statistics over the studied years).The economic results are generally positive, but in many cases rather modest, of around one to three million SEK, and the few that give economic yield to the owner are private owned ones.This shows that there are differences in terms of economic results among small and large systems, and that ownership matters.

Ownership changes
An important trend in the Swedish DH-sector is changed ownership, which is part of a larger system reconfiguration [3,51], that has led to new segments within the LTS.This is mainly a result of ownership restructuring or diversification, driven by liberalization in the mid-1990s.Consequently, many municipally owned energy companies, including their DH assets, were sold during the 1990s and 2000s, resulting in a change from 95 % municipal ownership in the district heating sector in the 1980s to a more diverse ownership structure including businesses that were privately owned, jointly owned by more than one municipality, jointly owned by private interests and

Table 1
DH in the 25 largest municipalities in terms of population (except Lund and Nacka).We show two different yearly differences (2009-2019 and 2009-2020) as 2020 was a warmer year compared to previous five, and with two comparisons, it gives a fairer picture of usage.Source: Swedish Energy Markets Inspectorate [50].municipalities, state-owned and cooperatively owned [9].The shares of delivered heating can be seen in Table 3. Municipal ownership, which includes ownership by one or more municipalities, is still the largest kind with 63 %, while the private actors, account for a rather small share of the delivered heat (13 %).Joint ownership, which can be shared between municipalities and private or state owners, have a large share (17 %).
There is a difference when analyzing the number of municipalities, in the largest while private actors are overrepresented in smaller municipalities.This is a trend that can be seen among the small, often northern municipalitiesthese municipalities have sold their energy companies to private actors [9,49], and these are shrinking municipalities [21].It shows also how there are, arguably, system segments among smaller systems, between municipally owned and non-municipally owned systems.It can also be concluded that price, which is defined as the amount SEK paid for by the consumer, for infrastructure service in general, in DH especially, is higher in smaller municipalities.The average price for DH, according to the Nils Holgersson annual pricing survey [52], shows that the average annual price per m 2 was 160 SEK (median 171 SEK).The highest price, in the small municipality Ydre was 210 SEK and 110 SEK in Luleå (with excess heat from the steel mill in SSAB).Among the 50 largest municipalities, in terms of population, the average price per m 2 was 164 SEK and among the smallest populations wise, 178 SEK.
Taken into account that private ownership is over-represented in municipalities with low population, the price differences between ownership categories is important.Among the 50 municipalities with the highest price, 56 % have private ownership and 42 % municipal.Among the municipalities with the lowest price, 94 % are municipal and 0 % private (the rest consists of shared ownership or state-owned).In Table 4 the DH prices among the categories are presented.Although, with a large span between min and max, the municipal energy companies have a lower price than the other owner categories [52].
We can conclude this section by arguing that treating DH as one national LTS, without going into depth in various segments, this do not show the whole picture of how system reconfiguration (mainly privatizations and changed business logics from self-cost price to market pricing) have created large differences between individual systems.The stagnation trend is also visible among many companies, considering DH usage (not yet economically), especially among large and private energy companies.

System reconfiguration among small DH systems
In this section we present the findings from the survey and interviews.The themes are based on challenges and opportunities for energy companies, with the former going into depth regarding profitability and ownership, fuel prices and competence.These are themes inductively identified from the material, that also gives an understanding of LTS aspects of small systems, compared to large, in the light of processes of shrinkage.

Shrinking processes not an issue in DH?
Decreasing population among small municipalities is a pressing issue, following urbanization processes and younger people moving to large cities, something that also relates to landscape changes, i.e. slowly changing external, structural trends, as discussed in the transition literature [41], and might lead to needs for system reconfiguration in many infrastructure systems.Similarly, the research field of shrinking cities and municipalities has focused on the effects of deindustrialization on population and infrastructure development [44].
In our study, population decline was not considered a major factor yet by the involved actors.The results from the survey shows that 32 % think that decreased populations can be a larger factor in the future, and the quote below signals how several think about the matter:

If you do not take these things into account in the long term, you will have problems, but in the short term it is not a big issue. The important thing is that in the long run, the starting point is that we sell less heat and thus that the next investment should aim for a slightly lower volume at all times. (Representative, energy company A).
This was a surprising result, given that all studied energy companies are operating in municipalities that are shrinking and that other studies have shown how infrastructure has been affected by these matters.In water and sewage systems, technical problems of over-sized infrastructure with sedimentation, higher temperatures and bacterial growth has been identified, as well as considerable organizational and economic challenges [22,24].The explanations might be that the DH-systems generally are located in the central town and in most municipalities the population decrease has not been as strong as in the whole municipality, even though the central town usually shows population decline as well [28].It was clear in the interviews that population decline was not in the mind of the energy companies in the same way as the municipal representatives [cf.21].The latter category often took on a broader perspective and reflected on how decline effects other sectors in the municipalities and how the problems may multiply, while the energy companies had their focus on energy and little else.We also argue that small changes in population do not affect them as direct as it does for the municipality economically.Compared to for example water and waste, over-dimension of DH is not a matter in the same way as demolitions has only been limited so far.However, every demolition is an economic loss for the DH company, and in the 90´s, there were a large wave of demolitions supported by state subsidies [53].

Challenges for small DH systems
The most pressing matters are rather other aspects, which has been concluded in previous studies [cf.5]: warmer winters, more energy efficient buildings, and competition from heat pumps.All interviewed actors state that these are matters they think a lot about, and even though population decline is a factor, these are much more pressing.Treating DH as a sociotechnical system, it is obvious that technical challenges are minor compared to organizational and various external factors.All together, they point towards stagnation, but actor strategies and dynamics within the segments show that reconfiguration occurs, but  in different ways.

Competition and customer service
The only factor that the energy companies can try to control is competition from heat pumps, a market segment that is increasing globally [54].Competition has been present in all development phases of DH, and based on political decisions and market dynamics, the pendulum has been swinging back and forward, and in recent years, heat pumps have gained momentum.In our study it was clear that the energy companies experienced that as a problem, and something difficult to manage, mainly because of difficulties to keep current DH-customers.
We definitely notice it, we do.We have some who disconnect, but it is mainly residential customers.But it is by no means a mass exodus.It's in (town X) we see it more, but it's probably a good local heat pump company that can speak for itself, I guess, and gives good-looking offers.In these cases, we have a difficulties matching heat pumps, especially with low electricity prices and low interest rates, then it is incredibly difficult.

(Representative Energy company A).
Another matter was the ability to focus on customer service, something several expressed as difficult to prioritize.

There has been zero and no communication with our customers from this business. It is also a challenge for the future that we must have better communication with customers. (CEO Energy Company B).
There are several reasons to this, with lack of time as the main reason, something we will discuss in the section below, but also that it "kind of works as it is", meaning that the problems have been small so far.However, and as the quote above shows, there might be an urgency to do things better, especially due to the competition from other heat sources, but with fast changing electrical prices, the pendulum might swing back towards DH.
Another aspect that might impact the interest in communication with customers relates to whether the companies see potential in expansion.Many of the interviewed energy companies expressed that the grid was rather well developed and saw little potential in further expansion, as there is a saturation [cf.6].In many of the central towns, there were limited development of new buildings, especially among multi-dwelling buildings, and among single houses the potential is already limited due to low heat density.
The grid itself can be said to have expanded as far as possible, there are certain areas where there are villas with direct electricity that were built in the 80s.So, it's a big cost for customers there to convert to district heating.Otherwise, the network has expanded as much as possible.It is densifications that might be relevant, and there are a few each year that connects to the grid.(

CEO Energy company C).
To summarize, competition from heat pumps is a challenge as is the lack of time for customer communication, which can be seen as factors for a possible non-expansion.This is a rather new situation, and something that earlier LTS-studies did not focus on, as most studies focused on growth aspects [4].However, to manage without growth require focus on the existing businesses and customers, and for small energy companies it can be challenging to develop more than absolute core business of production and distribution.

System reconfiguration through ownership changes
Demand and usage affect the profitability, but are not the only factors.In our survey, we asked the energy companies about the development of their profitability.The results show some differences, as 30 % answered that it increased and 27 that it decreased, see Fig. 3. Interestingly, there was a difference based on ownership.Among the ones that answered "increased" half were private companies and half municipal, but among those that answered "decreased", all were municipally owned.
There are some explanations to this, but going back to the discussion on pricing, municipal companies often keep lower prices and also have pride in this and often try to lower the prices.
We actually lowered the district heating rate from last January this year by 3 %.This is not so common; it is more common to raise prices every year.It is usually the constant whining you hear, that the price is raised and DH becomes more and more expensive.We have been on the same price level for a couple of years and now we took the chance to lower it.

Just to give something back to the customers and maybe make them a bit happier. (CEO Energy Company C).
This means, however, that the margins for small municipal companies are small.Low prices as a strategy, with often increasing costs and new requirements, means that they might be vulnerable for large changes.One of the energy companies expressed how they were affected by the need to invest in a new filter in the production, and that changed their result from positive to negative.
What might drive prices and needs to be taken into consideration is the required return on investment.Most of the energy companies stated in the survey that they expect the required return on investment to Fig. 3. Question from the survey concerning economic profitability.

D. Magnusson and I. Grundel
remain the same, although there are a few that expect them to increase.Out of these, most are municipally owned, which matters in terms of being owned by municipalities with a negative economic development and population decline.

The municipality has businesses that are bleeding economically and then the municipality thinks that they should take money from the [energy] company. (CEO Energy Company D).
This is an interesting contrast to the fact that the energy companies did not consider population decline as a major factor.Even though the connection between economic and population decline was clear for the municipalities it was still not considered a major challenge.
Among the private actors, a difference can be if the ownership changes.The private owners have a high required return on investment.There are municipal ones with high requirements as well, but these are often the larger ones.Among the private actors the strategies differ, if they have an exit strategy, or if there is a long-time perspective.In many of the private companies the owners are pension funds that see infrastructure as a stable, long-term investment, even though they expect a return of investment about 5-7 % per year.

There is a difference when looking at venture capital companies. Then you see that the ownership of the company is maybe 5 or 10 years. You make a transformation of the company and then sell it, and what they add is transformation knowledge. But in infrastructure operations this is not the case, then it is the capital you contribute. The time horizon is not set based on how fast you want to sell the company, but how fast the investors want their money back. (CEO Energy company E).
The municipal companies argue that required return on investment is a matter of communication with the board of directors, which consists of municipal politicians.With a good dialogue, the politicians understand the needs of the company and the margins for economic yield.That might be a challenging task, to establish a board of directors with competence that can have an impact on the company to support the CEO.They also seem to go beyond "only" looking for what's best for the company in term of economy, but are also taking a broader societal responsibility.

No, there have been no historical ones [required economic yield]. They have not picked anything directly from the company. First and foremost, they see it as a societal benefit, that we exist and that we can deliver environmentally friendly district heating. I have no requirements that I should make a lot of profit and that the municipality should be able to withdraw a lot of money from the company. (CEO Energy company F).
To summarize, ownership have an impact on prices and the views of return of investment.Given that privatizations are over-represented among shrinking municipalities, this goes back to aspects of different LTS segments and varying logics.For private actors, in the market-based pricing, DH can be a great investment if societal values are not considered as important as for municipal actors.This shows a clear divide, between private companies in shrinking municipalities, making a good profit, compared to municipal ones, making none or limited profits.

Factors connected to DH production
Other factors driving profitability are efficiency in production and distribution.Many of the energy companies work with measures to optimize the plants and the distribution continuously, but several express that they do not have enough time to do so.45 % replied that they had changed strategies for operation and maintenance in recent years, meaning that the majority had not.Lack of time, limited staff, and economic restraints were stated as main obstacles.Those that had time, express that it had been beneficial, but you need resources in order to do so.
We have sold very little heat this year because it has been very warm.And it affects the result.And then I see that the places I am responsible for have actually gone better than budget despite small volumes, because we have worked much with fuel mix and optimization.Of course, this costs in the form of consulting hours and people who sit and optimize, but it has clearly paid off.So, I have not had negative results this year in these four places, thanks to the optimization of production.

(Representative Energy company G).
To lower return temperatures is one main focus, but it requires quite some work among the customers, and in many cases, the plants are getting old and require attention.The ones stating that they have changed the routines mainly pointed towards maintenance in order to prevent rather than being one step behind all the time.
Among the energy companies that have replied "decreased profitability" in the survey, all but one answered that fuel prices is one of the most important factors that influenced profit.All energy companies responded that they had changed strategies in fuel purchases, and the written responses pointed towards more local purchases and partnerships and trying to be more flexible.The interviews gave more insights in the matter.
The main aspect is that prices have increased drastically in recent years.Companies in northern Sweden argued that competition with pulp industries, and the lack of taking out logging residues or forest residues (usually treetops, branches, stumps, in Swedish: GROT) from the forest forces prices upwards.There is also a matter of being small, as interviewed energy companies, except the private ones, are small actors with comparably small needs.This means that they do not even get offers in procurements, and that contracts are short-term.
Technical aspects are thus more connected to organizational matters than purely technical "nuts and bolt" ones.In the next section we will look closer into this matter.

Internal organizational challenges
Fuel purchases are a major factor for the energy companies, small or large, and usually lands on the CEO.A conclusion from the interviews is the vast number of tasks that lands on the CEO in small energy companies.In larger organizations there is a wider range of staff with expertise in certain fields, but in smaller organizations, everything from time consuming fuel purchases, to reporting statistics to authorities, strategic planning and customer care lands on the CEO.It makes the organizations rather vulnerable in case of sick leave or persons changing jobs.The focus on system builders in LTS-studies [1] comes to mind, but here it is rather a matter of system managing than building.
Possibilities to recruit staff, see Fig. 4, was identified as a major challenge.First of all, there is a lack of staff in general in the energy business, especially with increasing demands of special skills.For example, all personnel that are operating the DH plant need a certification, which is costly and demanding for small energy companies, but there is also a lack of education that matches the demands perfectly.

Too few people are getting the proper education in that industry. You must be arithmetic, and you must understand the combustion process. Then there is an interplay with the staff on exactly what happens depending on fuel used. Not many district heating managers know what happens to different fuel mixtures. (CEO Energy company D).
This points to the complexities at hand, and that the CEO cannot be expected to know everything in detail, there is a need for specific knowledge in the process.
There are differences between large and small companies, as the larger ones do not express any problems.They argue that the energy sector is considered interesting among younger persons, due to interest in climate change matters.It seems like larger companies can offer more complex and innovative positions, that makes them attractive, but that smaller ones cannot make those offers.It is also a line of work that requires night work a few times per month for safety reasons.
In the small, northern municipalities, they stress that there is a general lack of people and that it is a competition between industries, for D. Magnusson and I. Grundel example mining, for persons that are "generally skilled" in industrial work.These persons do not have particular training in the energy field but can work in production in several branches.This means that they are sought for in various industries, and that the salary becomes an important factor.
The strategy to handle this has in many cases been to educate the staff internally, as long as they are "good workers" and seem to want to remain in the company.
We have moved away from requiring formal specific training, we are instead looking for good people who can be trained internally.There are no trained operating technicians, you see instead that you have someone suitable internally, "a good person", who has a decent background and who can learn the job.For example, in [town X], he who is an operations technician will retire and there has been a trainee throughout the year.(Representative Energy Company G).
In the survey 35 % of the respondents point to the fact that they have "no time" or "no resources" to innovate (see Fig. 5), due to limited capacities and time.The ones that answer yes mainly focus on some aspects of digitalization or changed routines for maintenance, which is quite different compared to companies that can focus on for example BECCS [15,16].
To summarize, lack of time and competence leads to lesser possibilities for specialization and innovation.It shows a difference between the LTS segments, as small municipal companies have limited staff and trouble recruiting, while larger, and private ones, can take advantage of economy of scale and thus work with specialization.

Opportunities through cooperation
The main opportunities for the energy companies to handle challenges lies in various forms of cooperation, in or between companies.
The ownership changes points towards specific business models among private actors and shows how inner dynamics in the various LTSsegments adjust to challenges.The large municipalities remain municipally owned to a large extent, while the smaller often have been privatized.This is also a process of consolidation.Magnusson [9] showed how the three large energy companies Fortum, E.ON and Vattenfall sold most of their small grids around 2008-2009, and how three new actors entered the market: Neova (now Nevel), Värmevärden, and Solör.Over the following years, owners of Värmevärden and Solör have acquired more grids and also larger private energy companies with operation in several municipalities (Värmevärden bought Adven in 2020 with grids in 10 municipalities and Solör bought Vasa värme with grids in 6 municipalities).
Their business model points to advantages with owning grids in several places and with remote steering where possible, and staff that can operate in several municipalities.It makes the overhead costs lower and there are possibilities to buy fuel in larger amounts as well as having expertise in the organization.Another private actor argued that they have specific competence in running small plants, that's their key business model: These companies are very bad at handling smaller plants.Most of the time, I would say 99 %, they always lose money on their smaller plants because they are too big and ignore them.For those companies, the smaller boilers are more of a headache than a plant.However, we are specialists in smaller boilers because we have a small organization and we do not have much overhead compared to these large companies, and we know how to handle that.That is our business model, and perhaps our strength.(Representative Energy company H).
For the municipal energy companies, there are instead other possibilities at hand.In a few of the interviewed energy companies, the CEO splits the time between the energy company and the municipal housing company, and they can cooperate with personnel for maintenance work.
However, cooperation between different energy companies seems to be more limited.Often, lack of time is stated as the main reason, and they express that they would like to cooperate more, for example through knowledge extent.Cooperation is always promoted as favorable [20,55], but it is important to remember that in certain cases, especially in northern Sweden, the distances are so large that the possibilities for physical cooperation is limited.
In the interviews, one of the advantages put forward is the fact that small organizations can, if managed correctly, make fast decisions, communicate fast, and run with low overhead costs.Several state that they can have closer contacts with the customers, which gives opportunities to understand each other.
In order to manage a stagnation, which the national level points towards, acquisitions of smaller plants and developing the business model have been a strategy in the segment of private actors and leads to opportunities for specializations and economy of scale.The logic is thus different compared to small municipal companies.

Conclusion and policy implications
In this paper we have studied the development of DH systems in Sweden, with a focus on small, shrinking municipalities and we analyzed how we can understand the development from an LTSperspective.In this section we will answer our research questions and suggest policy recommendations.

System development in various segments of mature LTSs
In former LTS-studies, most focus has been on the earlier phases of system development.In this study we have shown how the DH-system in Sweden has developed in recent years, together with a theoretical understanding of mature LTSs.In several studies on LTS, the analytical starting point is to see LTS as one system, although with a sensitivity for diversity and that different sociotechnical systems have specific characteristics, [1,4,5].We argue that this leads to a too broad analytical lens to fully understand internal dynamics within an LTS.In this also lies discussions on system control and lack thereof, or contestation of consensus and how this can lead to processes of reconfiguration.
In this paper, we have given specific attention to DH-systems in small municipalities, to show the heterogenous development within the DH-sector in Sweden, and how different segments develop differently.Arguably, the system as a whole is in a phase of stagnation, with stagnating heat loads due to external factors (warmer winters, energy efficiency, competing heating systems, and saturated markets), but there is a large difference in terms of urgency, challenges, solutions, and resources when going deeper into various segments of the DH-sector.It is not possible to talk about system control of DH from a national LTSperspective, the companies can only control their own individual system/s.
By analyzing the DH-sector from the lens of segments, interesting dynamics become visible, and by dividing the material based on shrinking municipalities, the analysis has shown differences between small and large companies, but also based on ownership.Analysis of other LTSs might require other ways of identifying segments, but for our purpose, this was a fruitful method.
In general, the trend is that a vast majority of the DH-systems have decreasing demands for heat, but newer systems are the exception as they are still expanding.However, the economic result differs significantly.The large companies generate strong economic results and economic yield, while the smaller struggle in a larger sense.We argue that this has to do with a process of drift in the sector, as there were more similarities before the liberalization processes in the 1990´s started when almost all systems were municipally owned, and all were run on costbased pricing.The liberalization led to privatizations, but also substantial investments in combined heat and power, mostly with biomass but also with waste incineration, in the large systems.The smaller municipal systems are, arguably, still run more on similar principles as before 1996, while private companies in small municipalities have different business models, often higher prices, and high required economic yields.All in all, this leads to a heterogenization of the LTS, where it is not possible to think about system control, consensus, or even contestation; it is rather one for each own.
The drift has led to heterogenous segments, and to conclude the paper we argue that there are three major ownership segments: large municipal owners, small municipal owners, and private owners.Challenges are very different, and the private and large municipal owners share more similarities than small and large municipal owners, as these have large economic yields, a different organization structure with more specialized competence, and thus more resources for development.
DH is a suitable case for studying later phases of system development, also in relation to shrinking aspects, but certain characteristics are specific, especially the market pricing, which gives actors opportunities to make a large profit.Compared to water and sewage, which is regulated in Sweden, this is a large difference, but to fully grasp how different mature LTS develop, further studies are needed.

Challenges and opportunities for small DH systems
Several of the trends mentioned in the paper pose major challenges to the small municipalities and DH systems, and together they create a pressing situation.Fuel prices, limited resources, small organizations, limited possibilities for requirement, and slim economic margins are severe challenges.Population decline has not yet been identified as a major problem by the energy companies, but it is a challenge for the municipalities as such.Fewer inhabitants mean lower tax revenues, and this puts pressure on the municipal services.The capacity of municipalities to handle their mandatory tasks is considered an increasing challenge, and against this background it is understandable that it is attractive to sell their energy companies, both to receive economic capital to use elsewhere and because energy companies are not considered core businesses, and political competence to evaluate and develop them is often lacking.
The results show that private actors are overrepresented in the smaller municipalities.We also know that prices among private companies in general are higher, and that the companies have a high required rate of return.For the municipalities this means, in the end, higher DH-prices, both for the private customers and for the municipal housing companies and municipal buildings.Together with high prices for other infrastructure services, it means higher costs for inhabitants.
The largest opportunities for development lie in various forms of cooperation.Privatization is obviously not cooperation in itself, but to be able to pool resources and remote-control plants, as well as share costs for overhead and competence in several systems it is an advantage.For the municipal systems, we see some forms of cooperation, such as a shared CEO-position or cooperation with the municipal housing company.However, a factor that limits possibilities for cooperation between energy companies in northern Sweden is large distances.It is, for example, not possible to share personnel between systems if they are 150 km apart.

Policy implications and recommendations
Policy implications from the study point towards an importance for national policy makers to acknowledge the differences and consider the different prerequisites for DH-companies in Sweden.It would be reasonable to consider if there should be differences in certain legislations or to offer support in various forms, as these systems still are very important for the municipalities.The strong trend of electrification, together with higher shares of intermittent renewable electricity sources, have also led to a debate about high electricity prices in Sweden during the winter.DH plays an important role in transferring some of the demand for energy for heating away from the electricity grid, just as was argued in the late 1990s and 2000s when Sweden closed the Barsebäck Nuclear Power Plant.Subsidies were then specifically placed on DH in order to get households and industries to convert from electricity heating and oil heating to DH [56].With more fluctuating electricity prices, and probably higher on a yearly average, as well as limited capacity at certain times, DH and CHP will continue to play an important role, but policy support is important.
Even though the energy companies do not consider shrinking processes to be a pressing issue at the moment, there is a clear risk that the challenges will increase in the future as very few factors point towards a change in urbanization processes.Demolitions of buildings cannot be ruled out, and the impact on energy companies, both technically and economically, can be substantial.We can assume that the effects will not be as strong as in water and sewage, but an awareness from policy makers as well as municipalities and energy companies on infrastructure effects from continuous shrinking processes.

Declaration of competing interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Table 2
[50]n the 25 smallest municipalities in terms of population (except Storfors).We show two different yearly differences(2009-2019 and 2009-2020)as 2020 was a warmer year compared to previous five, and with two comparisons, it gives a fairer picture of usage.Source: Swedish Energy Markets Inspectorate[50].
the different ownership categories operate in.Municipal actors own DHsystems in 168 of the 285 municipalities that have DH (26 %) while private actors operate in 74 municipalities(26 %).When comparing the share between heat and municipalities, it means that private actors are overrepresented among smaller DH systems.That pattern is confirmed when analyzing the DH-ownership in the largest and smallest municipalities (measured by population).Compared to both delivered heating and share of the municipalities, municipal owners are overrepresented
Magnusson and I. Grundel Interviewee: We distinguish quite clearly between what is local operation and what is the engineering forces in the whole.Interviewer: How do you organize it?There are some geographical clusters but there are still relatively large distances?Can you organize it in regions?Interviewee: The local operation is on site, and they do not go between the plants.The engineering forces have no problem traveling between different places in Sweden.It does not matter much.They are simply a bit scattered, but they are working on projects.It's geographically close enough.(CEO Energy company E). D.