What's stopping the waste-treatment industry from adopting emerging circular technologies? An agent-based model revealing drivers and barriers

Highlights

• •Adoption feasibility remains a central issue in the Circular Economy.
• •Agent-based modeling is proposed to connect internal and ecosystem-level factors.
• •Analyses show the role of efficiencies, market growth, support frameworks and social pressures.
• •The adoption timing can be determined by the speculated market growth and technology improvements.
• •Supporting slow-growing technologies could result in sector-wide circular economy adoption.

Abstract

Many new circular economy technologies are gaining momentum, yet research on the complexity of adoption decisions driven by uncertainties, both at technology and ecosystem level, is lacking. In the present study, an agent-based model was developed to study factors that influence the adoption of emerging circular technologies. The case of the waste treatment industry was chosen, specifically its (non-) adoption of the so-called “Volatile Fatty Acid Platform”, a circular economy technology that facilitates both the valorization of organic waste into high-end products as well as their sale on global markets. Model results show adoption rates under 60% due to effects of subsidies, market growth, technological uncertainty and social pressure. Furthermore, the conditions were revealed under which certain parameters have the most effect. An agent-based model enabled use of a systemic approach to reveal the mechanisms of circular emerging technology innovation that are most relevant for researchers and waste treatment stakeholders.

Abbreviations, notations, nomenclatures, and symbols

${\alpha }_1$

Attention to social reasoning

${\alpha }_2$

Attention to economic reasoning

${\alpha }_3$

Attention to environmental reasoning

AR

Accepted ROI value

ABM

Agent-Basted Model(-ing)

CAS

Complex Adaptive Systems

CE

Circular Economy

CNG

Compressed Natural Gas

EP

Environmental pressure

IS

Investment subsidies

MG

Market growth

ODD

Overview, Design Concepts, and Details

OS

Operational subsidies

PHA

Polyhydroxyalkanoates

PUFA

Polyunsaturated Fatty Acids

ROI

Return on Investment

S

Subsidies

(s)LCA

Social and Environmental Life Cycle Assessments

SCO

Single-cell Oils

SP

Social pressure

TRL

Technology Readiness Level

VFA

Volatile Fatty Acids

VFAP

Volatile Fatty Acid Platform

YP

Technological yield increase probability

1. Introduction

The circular economy (CE) is proposed as a solution to overcome the environmental impact caused by the linear economy (Acerbi and Taisch, 2020; Blomsma and Brennan, 2017; Ellen MacArthur Foundation, 2015; Korhonen et al. 2018a; Stahel 2016) or even lately is argued as potential solution in times of crisis (Shams Esfandabadi et al., 2022). Many studies show the potential of the CE and governmental agencies view the CE as a necessary condition for structural change that arguably fosters sustainability (Geissdoerfer et al., 2017). Policy and support frameworks have been put into place (European Commission 2008; 2020a; Xu et al., 2021; Zhu et al., 2019) to pressure various industries to adopt circular economy practices. Nevertheless, many sectors find it difficult to function in the CE due to the complexity of overlapping supply chains and the need for new technologies to valorize waste into new input materials. This is also true for waste valorization technologies within the waste-treatment industry (Ranjbari et al., 2021; Tisserant et al. 2017; Priyadarshini and Abhilash 2020).

Waste is a key ingredient in the CE (Pires and Martinho, 2019; Stahel 2016). In many supply chains, the waste treatment sector comes at the end of the linear business model. In the CE, however, waste is supposed to become a new resource, where waste treatment plants both receive end-of-life materials and supply input material for new value chains. This makes them essential to the actors for the success of the CE (Aghbashlo et al., 2019; Ranjbari et al., 2021). But before waste treatment plants can fulfill this new role as supplier, they must not only adopt new technologies but also thoroughly reform their management approach, decision making process and business models.

This process requires a shift of focus away from lower-value products such as biogas, compost and metal scrap, to high-value products and thus new technologies that can be used as inputs in new value chains (Korhonen et al., 2018b; Snellinx et al., 2021). At the same time, their business model and strategies need to shift from the current cost-minimization model (Aghbashlo et al., 2019; Ljunggren 2000; Mohsenizadeh et al., 2020) towards a profit-making business model (Tongur and Engwall 2014). Such shifts are not trivial, especially when the profitability of many of the valorization technologies has yet to be proven on a large scale. This profitability is also influenced by both technological and non-technological features, including investment and operational subsidies, market sizes of the envisioned end products, and gate fees for the incoming waste (Velghe et al., 2021). Moreover, even though profitability, market demand, and competition are central to the decision-making process (Campbell and Hopenhayn, 2005; Esparza et al., 2020; Snellinx et al., 2021), additional influences on the decision-making process about technology investments are the desire for environmental legitimacy through regulatory bodies (Corvellec and Bramryd, 2012) and peer pressure arising within social networks (Snellinx et al., 2021).

Waste treatment plants apply the following reasoning when deciding whether to adopt emerging CE technologies. First, market formation is perceived as essential, as it is crucial to both the success of technologies within an industry (Hekkert et al., 2007) and the eventual profitability of the valorization technology (Cucchiella et al., 2016). Therefore, changes in the market such as market size and growth in different segments (Berry and Waldfogel 2010) influence the revenue model of the valorization technology. Second, competition exists between companies who all might be planning to adopt similar valorization technology and create or enter a new market. The adoption decision is therefore also dependent on a dynamic market environment (Campbell and Hopenhayn 2005). Third, the social environment and support frameworks for new technologies contribute to their legitimacy in the market (Snellinx et al., 2021). Last, the emerging valorization technology efficiencies are still in development processes and their production factors may change in the future (Wang et al., 2022). This contributes to the uncertainty in economic assessments.

In summary, the analysis of whether or not to adopt a particular CE technology requires more than simply a techno-economic study. A systemic approach is needed that takes technological as well as socioeconomic factors into account, both at the company level as well as at the level of the industrial sector (Gue et al., 2020). Literature is full of technological and techno-economic studies regarding the CE. Several recent studies have focused, for example, on the effects of technological complexity on the efficiency of emerging CE technologies (e.g. Nayak and Bhushan 2019). Some studies also focus on Social and Environmental Life Cycle Assessments (LCA), techno-economic assessments and multi-criteria assessments (e.g., Tomić and Schneider 2020; see the review by Ubando et al., 2020; Aghbashlo et al., 2022; Soltanian et al. 2022). The decision to adopt a new technology has been widely explored, highlighting the importance of market, government policies, social and environmental legitimacies (Fu et al., 2018; Gue et al., 2020), citizen pressures, and organization-management approaches (Garrone et al., 2018). The majority of these analyses have been made for go-to-market technologies, however, thus neglecting the technologies that are still in development. Those developing technologies, which are affected by operational factors, may result in improvements in associated production yield (Nayak and Bhushan, 2019). Existing studies also largely lack a rigorous consideration of market competition, leading to a static analysis. A more holistic and systemic analysis is needed to unpack the decision-making process within a dynamic market environment where the technology can be improved and where the technology adoption by waste treatment plants could result in a change in the competition dynamics.

The aim of this article is to add an ecosystem level to the analysis of the socioeconomic aspects of decision-making about whether to adopt emerging CE technologies. Such an ecosystem level is characterized by factors including market, competition and technological uncertainties, social pressures, and support frameworks. Competition can be observed at population level where waste treatment plants decide to adopt an emerging valorization technology and then occupy the market based on their production of a high-value end product. The following research questions are answered in this paper: “What is the effect of ecosystem-level factors, including (i) market growth, (ii) subsidies and other support frameworks, (iii) technological uncertainty, and (iv) social pressure on the adoption of emerging CE technologies in the competitive environment of the waste treatment industry?”

The case of the European organic waste treatment industry was chosen because CE strategies create pressure to shift organic waste valorization away from the current strategy of energetic valorization and toward creation and sale of higher-value products. The present study focuses on a valorization technology called VOLATILE FATTY ACID PLATFORM (VFAP) (H2020 VOLATILE 2020b). This technology, currently at Technology Readiness Level (TRL) 6 (see European Commission, 2017), is designed to process organic waste into Volatile Fatty Acids (VFAs) that can subsequently be transformed into high-end products such as polyhydroxyalkanoates (PHA), single-cell oils (SCO) and polyunsaturated fatty acids (PUFA). These products can be sold as chemical building blocks on global markets. Because all of these products are currently produced from fossil fuels, adoption of VFAP and sale of derived products would lead to a reduction of greenhouse gas emissions.

Different methods have been used to study the feasibility of adopting a new technology and the factors that enable adoption. Agent-based modeling (ABM) has been shown to successfully capture the complexity and the dynamism of the waste treatment industry. ABMs are computational tools for studying complex adaptive systems (CAS) where heterogeneous units across different levels interact and can result in a change in the state of the system (Edmonds and Meyer, 2017; Epstein 2006; Miller and Page, 2007). For the present study, an ABM was built that simulates the European waste-treatment industry and the decision-making process of whether to adopt VFAP.

The results of the VFAP case study show that several factors affect the decision to adopt a new technology: technological uncertainties, market size and growth, governmental support frameworks, and societal pressure. That study also sheds light on how the adoption process might evolve for alternative technologies designed to create high-value input products derived from waste. This article offers new insights for researchers studying the complex behavior of adaptation processes concerning the circular economy, as well as for waste treatment plant managers, stakeholders and policymakers looking for new opportunities to promote circular technology implementations.

This paper is structured as follows. First, the VFAP case is elaborated, specifically its technological innovation and associated high-end products. Second, the method used and the data are presented, followed by a step-by-step guide to the ABM. Then results from model exploration are presented in relation to the research questions. The paper ends with a discussion of the results with final conclusions.

2. Case: waste treatment industry & volatile fatty acid platform

This study relies on the case of the waste treatment industry and the challenges it faces regarding the adoption of new technologies. Waste treatment plants arguably represent the intersection of all supply chains, as all supply chains produce waste that needs to be processed. A large amount of currently underutilized waste has potential to be used in different sectors after processing (Stahel, 2016). For this reason, the European Commission (as a part of its Circular Economy Act (European Commission, 2020a, 2020b)) in the framework of the Green Deal (European Commission, 2019b, 2019a) has put programs in place to incentivize research on practices in the field of waste and the circular economy.

For the current study, nine organic waste treatment plants located in different European regions were selected. These waste treatment plants were all involved in the VOLATILE project (H2020 VOLATILE 2020b). The aim of that project was to develop an innovative technology called “VOLATILE Fatty Acid Platform” (VFAP) that would transform existing organic waste treatment plants into providers of volatile fatty acids (VFAs). These VFAs can be used for the production of high-end products (hereafter VOLATILE products) such as polyhydroxyalkanoates (PHA), single-cell oils (SCO) and polyunsaturated fatty acids (PUFA). All three products can be sold in global markets as sustainable alternatives to the currently-used fossil-based products. PHA, a building block for bioplastics, can be used as an alternative for fossil-based plastic in packaging, SCO represents an alternative for palm oil (thus reducing the deforestation associated with palm oil production), and PUFA represents an alternative for fish oil extractions and Omega3 (thus reducing unsustainable fishing and aquaculture practices). Moreover, the aim of using VOLATILE technology is to achieve higher levels of resource efficiency by creating more value with the same input, which could result in better environmental and economic outcomes (Michels et al., 2020; Velghe et al., 2021).

The waste treatment plant managers involved in this project associate the adoption of such technology with risk and the need to evaluate economic feasibility. Understandably, they perceive the risks as considerable. Their decision-making process not only involves the new technology but also implies a significant change to their current production process and business models. The waste treatment plants currently transform organic waste into heat and electricity from biogas and compost and generally use a cost-minimizing business model. Adoption of the VFAP requires alternative processing of the same organic waste in order to produce VOLATILE products. This would lead to reduced biogas production and therefore a shift in income (H2020 VOLATILE 2020a). A successful transition will require that the waste management plants enter a market that requires a profit-maximizing business model. The technology itself is only at TRL 6, however, meaning that there is still room for improvement. This introduces technological uncertainties which in turn influence the economic investment assessment process.

3. Data and methods

3.1. Methodological framework - a motivation to use agent-based modeling

This research takes a multilevel approach to investigating the VFAP adoption. First, the waste-treatment plants’ management decision-making process concerning emerging circular economy technologies is elaborated via the inclusion of the ecosystem-level factors described in the introduction. Second, an agent-based model (ABM) is developed to simulate an environment where heterogeneous waste-treatment plants face the same adoption problem while ecosystem-level factors influence their decision-making. Third, experiments are designed and the model is explored to understand the effect of the ecosystem-level factors and mechanism for VFAP adoption.

The feasibility of technology adoption has been studied extensively using a range of qualitative and quantitative approaches such as surveys (Choudrie and Dwivedi, 2005), cost-benefit models (Bazen and Brown, 2009), and economic choice models (Cainelli et al., 2015). These methods help to explain the effect of different factors on the feasibility and likelihood of technology adoption. However, they do not take the dynamics of the environment into account (i.e., what happens after one or few actors adopt the technology?). Another approach to consider is equation-based diffusion modeling, which can be non-parametric and probabilistic. It can identify the number of adopters based on the previous adoption conditioning of different factors, including non-rationality assumptions (Managi et al., 2014). Still, this approach is not flexible enough to address detailed levels of system dynamism, market competition, and interactions among different actors.

The current study is situated in a context of emerging technologies (i.e., high uncertainty associated with a technology at low TRL that can still improve) within a competitive market. Within this specific context, the research aim requires a more flexible and dynamic approach to project and simulate future scenarios based on different technology and ecosystem-level factors. Two methods could be used: (1) ABM or (2) System Dynamics (SD). While SD is a very powerful tool to build a dynamic environment (Dhirasasna and Sahin 2021; Sterman 2001), it excludes interaction among different agents, remaining instead at the level of system elements (factors). For this study, ABM (Edmonds and Meyer 2017; Wilensky 1999; Wilensky and Rand, 2015) was more appropriate because it is capable of simulating a dynamic evolving environment by including interactions among agents as well as feedback loops between the system status and the agents. Using ABM it is possible to explore CASs with emerging patterns (Miller and Page, 2007). The case under study, where the waste treatment industry faces the decision to adopt emerging circular economy technologies, is a CAS where heterogeneous waste treatment plants interact and each decide whether to adopt the technology. Their adoption decision has an effect on their peers (in the form of peer pressure) as well as on the market. Their decision to occupy a future market will increase competition as it narrows the available space on the market (market gap).

ABM has already been applied to different aspects of the waste treatment industry, including waste-generation-separation dynamics (de Souza, Bloemhof, and Borsato 2021), the adoption of biogas (Burg et al., 2021), as well as the adoption of anaerobic digestion (Falconer et al., 2020). To the best of our knowledge the method has not yet been applied in the context of adoption of emerging CE technologies.

3.2. Data

The dataset used for (1) model development and (2) model validation comes from the VOLATILE project (Nov 2016 - Nov 2020). The data collected included techno-economic evaluations, legal aspects, and business economic aspects for each technological path formulated in the project's deliverable documents. An additional 16 semi-structured interviews were conducted and transcribed (182 pages). Three workshops with experts were organized, and knowledge was gathered during project meetings. The analysis of all of these information sources allowed us to uncover decision-making criteria for potential VFAP adoption with the business cases of the waste treatment plants involved in the project. Different types of waste treatment plants were included, with different ownership (i.e., public vs. private), different types of waste treated, and different locations (Romania, Spain, Greece, Portugal, Belgium, and The Netherlands). Table 1 summarizes which data were used, how they were reflected in the model, and how they increased model validity.

Table 1.

Data type, description, and reflections in the model.

Data type	Description	Reflections in the model
Interviews	16 semi-structured interviews to characterize the decision-making of waste treatment plants (182 pages verbatim transcription)	Highlighted the social and economic components of decision-making, i.e., generally low risk-taking behavior in the industry, differences between companies due to type of ownership, and type of waste treated.
Expert workshops	3 workshops with different aims: 1) network conceptualization, 2) identifying adoption scenarios, and 3) discussion of results and validation	The most appropriate approach appeared to be community networks based on functions. The most plausible factors fostering the decision to adopt appear to be technological improvements, institutional supports, and economic factors. A special emphasis on the effects of social pressure and gate fees on adoption was revealed.
Techno-economic	Comprehensive parametrized techno-economic assessments showing the profitability of different VFAP adoption scenarios (for each VOLATILE product) (863 pages of deliverables)	As the main baseline for the economic reasoning of the waste treatment plants for each adoption scenario, economic parameters were included in the model.
Legal	Legal analysis showing the potential of subsidy frameworks to be introduced for the VFAP (452 pages of deliverables)	Two types of subsidies (investment and operational) were introduced as external factors that influence the waste treatment plants’ economic assessments.
Business modeling	Analysis of investment behaviors, the profitability of different scales of technology, uncertainty, gate fees, the potentiality of market growth and social movements	Three technological investment scales (small, medium, large) were included in the model to match different waste-treatment capacities, market growth, and market size as factors influencing adoption and entry strategies.

Data regarding the environmental assessments were also considered but were not reflected in the model. Results of the LCA analysis (H2020 VOLATILE 2020c) showed that all the adoption scenarios would result in a reduction in environmental impact. It is beyond the scope of this paper to include a detailed integration of the environmental analysis.

4. Agent-based model: SIM-VOLATILE

The ABM used in the current study, called “SIM-VOLATILE”, was developed as part of the VOLATILE project and built in a NetLogo 6.2 environment (Wilensky 1999). It was validated based on the data collected during the VOLATILE project (see Section 3.2 above). For extensive documentation of the model following the Overview, Design Concepts, and Details (ODD) protocol (Grimm et al. 2020; Grimm, Polhill, and Touza 2013; Müller et al. 2013), see Supplementary Materials, Section 1.

In brief, SIM-VOLATILE simulates an environment where different types of waste treatment plants coexist. The environment also includes other actors, e.g., governments, communities, and markets. In this environment, waste treatment plants face the decision of whether to adopt VFAP technology as an emerging circular economy technology. To make this assessment, the plants first evaluate the techno-economic aspects of the technology at the business level. That is, for each technological path of VFAP, they calculate the cost and revenue structure of the VFAP value under the assumption of investing all their waste inputs into the VFAP. Furthermore, they consider the availability of gate fee, subsidies, and market sizes as three external factors in their analysis, and they make a projection of Return on Investment (ROI) based on this analysis. They also look at local trends, i.e., the number of adopters in their neighborhood. For details of the model, see below.

4.1. Structure, initial settings, parameters

The model has two main structural levels: global and agents. The model incorporates a variety of parameters at each level. Global-level parameters are set to characterize the market, technology, institutions, and organizational networks. The market is parametrized based on its size and growth for each VFAP related product. Technological uncertainties are parametrized through yield factors and probabilities of the yield to be increased over time. Institutions are implemented by introducing parameters of gate fees, subsidies for the operations, and investments. Organizational networks are parametrized through the number of formal and informal links that exist between the companies and organizational network clusters. This reflects how an agent embedded in a network can be influenced by social and environmental movements. Fig. 1 shows the basic principles of the model and how agents perceive their environment.

Fig. 1.

Basic principles of the SIM-VOLATILE model.

Agent-level parameters are incorporated to introduce heterogeneity into the agent populations. Agents (in this case, waste treatment plants) are heterogeneous based on their environmental awareness, investments required to adopt VFAP (where likelihood is dependent on the agent's size and capacity), expected returns (how much the agent can expect from the investments, based on waste treatment capacity), probability of installing a pilot and probability of the pilot to be scaled up, probability of being an early adopter (i.e., no pilot and with direct investment), probability of adoption (each agent is assigned an adoption probability that defines its mindset and readiness for adoption; this is based on the type of waste being treated and public vs. private ownership). Altogether, the model incorporates a variety of fixed and dynamic parameters.

4.2. Dynamics

After initializing the model environment and assigning the parameters, the agents go through different procedures until they decide whether to adopt the VFAP technology with a certain product focus (PHA, SCO, PUFA).

First, the agents interact with the other agents in their area (their neighbors). All of these agents are connected via their organizational network. These interactions can lead to a change in the agents’ mindset regarding the environment (i.e., within the model their environmental decision values get updated). This interaction continues until the simulation is stopped. Agents then assess the degree of social pressure by considering the adoption trend in their neighborhood for each VFAP scenario (PHA, SCO, PUFA). In this way, for each scenario, the agents calculate a social pressure value that influences their decision-making. Next, agents assess the economic values for each VFAP investment scenario. For each scenario, the agent calculates a ROI which is based on all the technology, market, and institutional variables. Then, for each scenario, agents evaluate a general adoptability value, which is calculated by the following formula,

$$A{V}_i=\left({\alpha }_1\times S{P}_i\right)+\left({\alpha }_2\times \mathit{RO}{I}_i\right)+\left({\alpha }_3\times \mathit{EP}\right)$$ (1)

where $A{V}_i$ is the adoptability value for the investment scenario i, $S{P}_i$ is the social pressure for the investment scenario i and ${\alpha }_1$ is the parameter attention to social reasoning effect, $RO{I}_i$ is the return on investment for the investment scenario i and ${\alpha }_2$ is the parameter attention to economic reasoning effect, $EP$ is the environmental pressure and ${\alpha }_3$ is the parameter attention to environmental reasoning effect.

After the adoptability values are calculated, the waste treatment plants decide whether to adopt a VFAP. Those that choose to adopt then move on to an investment scenario. Waste treatment plants can choose an investment scenario that includes only one VFAP product (PHA, SCO, PUFA) as each product requires a differentiated path (Fig. 2). Each path is associated with a product and thus a global market. Each market is characterized by a certain size and growth, which could influence the agent's decision of whether to adopt the VFAP and ultimately produce a VOLATILE product.

Fig. 2.

The three VFAP technological paths leading to the three high-end VOLATILE products.

To prepare this decision and choose only one investment scenario, the waste treatment plant randomly selects an investment scenario (i.e., technological path), and checks (i) if its adoptability value is high enough (above the accepted ROI value (AR), a predetermined threshold value set at 0.05), (ii) if its adoption probability (a preset value depending on the function and the ownership) is met, and (iii) if there is a market gap large enough for one actor. The market size and market gap are assessed as follows: the agent calculates the size of 5% of the European market (reflecting the generalized low risk-taking behavior characteristic of the waste treatment industry) and assesses whether there is space left in the market for an agent of its size (based on potential VFAP production units). If all the conditions are met, the agent adopts the investment scenario and enters the market. Accordingly, the target market size and gaps are updated, which influences other agents’ decisions in the next rounds within the simulation environment. Fig. 3 summarizes the decision chart of a waste treatment plant as it assesses whether to adopt a VFAP scenario.

Fig. 3.

Decision chart of a waste treatment plant. A waste treatment plant starts from the environmental, social and economic assessments of the VFAP scenarios, then bases its decision to adopt the VFAP scenarios on the model's predefined threshold, a random value and the market gap.

4.3. Baseline selection and experimental design

Based on the economic analysis alone, none of the VFAP investment scenarios are profitable. The calculations included the techno-economic assessments and cost-revenue analysis based on the current speculated costs and market prices under the assumption that customers already exist for VOLATILE products. The calculations excluded potential support frameworks such as gate fees and subsidies. However, the inclusion of such factors (e.g., gate fee) could substantially influence the economic assessments made by the waste treatment plants and thus their decision-making.

Within this context, a baseline is first chosen that enables the VFAP investment scenarios to meet their break-even points. To do so, the gate fee is varied while other support frameworks as well as technological factors (e.g., yield) are excluded. This modeling decision is based on the experts’ observation that a gate fee is normally applied in the waste treatment industry. Support frameworks such as subsidies could also be applied, but they might be technology-dependent and not as straightforward as gate fees. Each investment scenario requires a different gate fee to become profitable. Based on the experts’ views and the calculated gate fees, the PUFA investment scenario is not feasible at this moment. That technology might require more research to increase its yield. For the PHA and SCO scenarios, a 45 €/ton gate fee (86% increase) is applied to enable them to break even.

When the gate fee parameter is fixed, a sensitivity analysis is run to understand the effect of population, network, and initial adopter related-parameters (see Supplementary Materials, Section 2). The results of this sensitivity analysis resulted in sufficient insight to assign a fixed initial structure in the model. In further work on identifying the baseline, other agent and global values are set to represent a low risk-taking environment (as indicated by the data collected in VOLATILE). To this end, parameters¹ are set to limit adoption, so that not all the agents adopt regardless of their perception of the adoption scenario (i.e. if it's profitable). This results in a small number of pioneers and the assumption of full attention to economic reasoning.

Altogether, the baseline simulation environment and its associated parameters are set as follows. The simulation time is set to 25 rounds, with each round representing 1 year. There are 10 communities with 10 waste treatment plants each. Each waste treatment plant is connected to a maximum of 20% of its organizational neighbors in its close vicinity. Waste treatment plant sizes are characterized as small, medium, and large, with a random variation around 30% of their average size. Small size plants are assigned a waste input of 10,000 tons/year, medium 20,000 tons/year and large, 30,000 tons/year. The probability of a waste treatment plant to have a VFAP pilot is 5% and the probability of its pilot to be scaled to the full size of the plant is 20%. Each waste treatment plant has three sets of environmental decision values. If the average environmental decision value of a waste treatment plant is above 90% of the others’ average decision values, then the waste treatment plant is considered as a potential early adopter. There are 6 kinds of waste treatment plants based on whether they are public or private, and whether they process water-based waste, organic-based waste, or a mix of both. The probability that a public waste treatment plant considers the VFAP adoption is 80%; a private one is 20%. The probability of a water-based waste-treatment plant adopting a VFAP scenario is 20%, an organic-based 60%, and a mixed-based 20%. These probabilities are set based on the interviews with the managers of the case study organizations and the fact that the current VFA yield for organic waste is considerably higher than water-based waste. Moreover, every five years there is a chance the VFAP technology could be improved, which could increase its yield factor by 10%. The market growth for the SCO is set at 1%, for PUFA, 5% and for PHA, 5% (the PHA market growth is explored in an experiment; see below). The accepted adoptability value for a VFAP adoption is set at 0.05. More details of the parameter values and settings are described in the ODD of SIM-VOLATILE (Supplementary Materials, Section 1).

The design of simulation experiments was done with a two-step approach. In the first step, parameters are identified that address markets, institutions, technology, and social movements (i.e., social pressure visible through the adoption trends). The environmental movements are excluded here, with the reasoning that if a waste treatment plant is assessing the VFAP investment scenarios, then it is already environmentally aware (this assumption is based on observation during the expert workshops). Therefore, the following model parameters are co-varied; 1) investment subsidies (IS), 2) operational subsidies (OS), 3) technological yield increase probability (YP), 4) market growth for small markets (MG_PHA), and 5) attention to social reasoning (${\alpha }_1$). For the parameter of market growth, for computational reasons, only the PHA market is considered as it has a very small market size and its growth could influence the agents’ decision-making process. Nevertheless, the decision to exclude the SCO and PUFA does not have a significant effect on the analysis, as they have a larger market size that makes them more attractive options. Knowing the chosen five parameters, four values are chosen for each as summarized in a factorial design of $4^5=1024$ runs. When assessing how many repetitions are needed, (Seri & Secchi's2017) power analysis framework for ABM experiments is applied. The result of the power analysis suggests that 20 repetitions for each configuration are required. Hence, the total runs is 20,480 resulting in 532,480 observations. Table 2 shows the parameters which were co-varied together with their ranges and descriptions.

Table 2.

Parameters selected for the experimental design in the first step: notations, ranges, values and descriptions.

Parameter	Notation	Range	Values	Description
investment subsidies	IS	[0,1]	(0, 0.15, 0.25, 0.35)	This parameter defines the degree to which the cost of investment can be covered by governmental supports.
operational subsidies	OS	[0,1]	(0, 0.15, 0.25, 0.35)	This parameter defines the maximum subsidy rate on the operational costs of agents.
technological yield increase probability	YP	[0,1]	(0, 0.2, 0.4, 0.6)	This probability value illustrates the likelihood that yield will increase over time due to technological advances or process efficiencies in scaling.
market growth (PHA)	MGPHA	[0,1]	(0.05, 0.15, 0.25, 0.35)	This value is based on qualitative and quantitative assessments of the PHA market, showing the supporting movements and market growths for the PHA related market.
attention to social reasoning	${\alpha }_1$	[0,1]	(0, 0.25, 0.5, 0.75)	This parameter defines the degree to which an agent is influenced by its neighbors that have already adopted the VFAP.

The second step, based on the sensitivity results of the first step experimental design, is to identify three specific scenarios in order to investigate how sector-wide technology adoption can be achieved given the simultaneous effect of the influential parameters. The configurations of the three scenarios are explained in Section 6 ‘Discussion’, below.

5. Results

Taking a sensitivity analysis approach (Richiardi et al., 2006), the results of the model experimentation are explored. Table 3 summarizes the correlation of different parameters with VFAP adoption rate. In order to plot the results and show how different configurations produce different outcomes, r-ggpplot2 (Hadley 2016) and r-ggtern (Hamilton and Ferry, 2018) packages are used. First, ggplot2 is used to visualize the sensitivity results of OS, IS, MG_PHA and ${\alpha }_1$ explorations (by using the ggplot2 smooth mean function while showing the confidence intervals). The second step builds on these sensitivity results by making further analyses using r-ggtern to visualize the simultaneous effect of the most influential parameters on sector-wide adoption.

Table 3.

Correlation tests summarizing the overall results of the sensitivity analysis depicting the role of different parameters.

Method	Pearson's product-moment correlation
Dependent variable:	VFAP adoption rate
	Coefficient	t-values
OS + IS	0.132	96.93
MGPHA	0.082	59.85
YP	-0.341	-264.71
${\alpha }_1$	0.147	108.15
N observations = 532,480

A complete presentation of the sensitivity analysis is provided in the supplementary materials. The following subsections show more focused results that provide insight into the role of (1) subsidies and support frameworks, (2) market and technological uncertainties, and (3) social pressure and movements in the VFAP diffusion.

5.1. Subsidies and support frameworks

IS and OS are the parameters that indicate the support frameworks in the waste treatment industry. These two parameters were varied while also varying the other parameters to see where and when their effect is visible. Fig. 4 summarizes the effect of support frameworks. The results presented in Fig. 4 illustrate how the inclusion of IS and OS has a generally positive effect on the VFAP adoption rate.

Fig. 4.

Effect of support frameworks - IS & OS parameters.

The two introduced support frameworks have effects in different situations. The OS parameter does not have major effects when the IS parameter is low (IS < 0.25). That is, when IS is low, increasing OS will not significantly influence the ROI and thus the VFAP adoptability. Similarly, the effect of IS is most visible when it is relatively high (IS > 0.15). Based on such results, it is evident that support frameworks could foster the adoption of VFAP technology, as the inclusion of both IS and OS reduces the total and fixed costs. When the costs are generally low, more waste treatment plants will be able to meet the profitability thresholds, and thus become more likely to reach the overall adoption threshold. Even so, not all the waste treatment plants will choose to adopt the VFAP technology, for two main reasons: (1) the different plant sizes influence the degree to which the support frameworks enable the feasibility of adoption, and (2) when some adoption has already occurred and the market is thus perceived to be occupied, the remaining agents’ desire for risk-taking will be more subdued. The effect of subsidies and policies can be interpreted as positive as supported by literature; however, the other model parameters show that the effects might not be perceived as straight forward. The effect of subsidies is generally context dependent (Cainelli et al., 2015; Xu et al., 2021).

5.2. Market growth

Market growth and change in the market segments for all the VOLATILE products have been observed (H2020 VOLATILE 2017). An increasing number of consumers prefer to use new, eco-friendly products in comparison to traditional fossil-based ones. For PHA, companies are responding to recent societal movements and increased awareness of plastic waste by moving towards biodegradable packaging (Gutiérrez Taño, Hernández Méndez, and Díaz-Armas 2021). For SCO, societal trends against use of palm oil produced through deforestation are incentivizing the production of SCO from alternative sources (Teng, Khong, and Che Ha Shasha 2020). Similarly, for PUFA, the trends against omega3 extracted from fish oil, linked to overfishing and unsustainable aquaculture practices, could be a driver to draw attention to alternative sources of PUFA such as microalgae (Toppe 2013).

This growth in consumer demand points to the profitability of a VFAP investment and is thus likely to motivate the waste treatment plants to consider adopting VFAP. This insight is implemented in the decision-making process of the waste treatment plants in the simulation model. The results of the effect of market growth for PHA (MG_PHA) are shown in Fig. 5.

Fig. 5.

Effect of market growth in the PHA segment (increasing MG_PHA).

When MG_PHA is increased, a larger market gap is perceived by the waste treatment plants, thus triggering risk-taking behaviors and in turn facilitating more adoption. Moreover, since the increase in MG_PHA is included in every simulated year (i.e., steady growth over time), its effect on adoption also follows a gradual function, e.g., a slow increase in the number of adopters over time. Similar to the findings of Gue et al. (2020), the results show that the market can incentivize more actors to adopt the technology. However, its effect is still limited to the model configurations and the agents’ specific conditions. That is, the agents still need to believe the technology is feasible, which may not be addressed by market gap alone.

5.3. Technological uncertainties and trajectory

Technology efficiency is an important factor in adoption decision making. Efficiency can influence the use of resources, scaling productions, and overall, cost-revenue structure (Nayak and Bhushan 2019). In the context of emerging technologies, yield factors are uncertain as many technologies are still at a low TRL. If potential improvements materialize, the technologies may show a different trajectory. Such uncertainties, including the potential improvements in emerging technologies, are implemented in the model. The effect of potential improvements is explored through the variation of YP while co-varying other parameters as indicated in Table 2. Fig. 6 demonstrates the effect of an increase in YP, i.e., what would happen under an increased probability of VFAP technology improvement. The results provide interesting insights into how different YPs influence the VFAP adoption rate.

Fig. 6.

Effect of improvements in technological efficiencies (increasing YP).

YP has a dual effect: the VFAP adoption rate at the lower YP ranges (<0.2) is higher then drops when YP rises (>0.2). This can be explained as follows. When the YP increases slightly, it positively influences the ROI, leading to a higher rate of adoption. However, when the YP increases even more, this positive influence on ROI leads to an increased amount of market space being taken up by the early adopters (e.g., via larger production units), and crowds out other potential adopters. Timing of adoption therefore becomes an important factor that is dependent on the technological trajectory.

This shows how technological efficiency is one of the most important factors influencing technology adoption as well as the need to monitor it (Nayak and Bhushan 2019). Higher yields allow the early adopters to occupy the market with large production units, which limits the market gap/size for risk-averse companies. For the case of VFAP, it took only 5 years for this technology to evolve from laboratory scale (TRL4) to pilot level (TRL6), a significant improvement in a relatively short time.

5.4. Social pressure and movements

Social recognition, trends and movements are non-economic factors that influence manager decision-making (Gue et al., 2020; Snellinx et al., 2021). Examples are social movements in the market segments (i.e., change in customer preferences toward more alternative products), social media, recommendations by the government, and the behavior of other managers in the network (neighboring connections and peer pressure). To operationalize these factors, the model implemented social pressure as peer effects. If a waste treatment plant adopts a VFAP, then the model assumes that the manager has sensed the social trends and pressures leaning toward CE technologies. In this line of reasoning, an adopter, who has sensed the environment and adopted a VFAP as a response, will influence the neighbors and expose them to the social trends and movements. Moreover, one could say that the manager who adopts is likely to have a higher level of trust in neighbors who also plan to adopt. Fig. 7 shows the effect ${\alpha }_1$ on the adoption of the VFAP. In general ${\alpha }_1$ has a positive effect on the adoption of VFAP.

Fig. 7.

Effect of social pressure (increasing ${\alpha }_1$).

The effect of ${\alpha }_1$ varies, however. It appears that the difference between ${\alpha }_1=0$ and ${\alpha }_1=0.25$ is more significant than higher levels. That means, after a certain degree of social pressure, it becomes less important despite being sustained, indicating a plateau effect.

Social pressure is a complex measure, and it can be seen in different interconnected ways. It can be a movement in society by consumers and citizens revealed in the form of subsidies, directives and peer technology adoption behaviours (Dhirasasna and Sahin 2021; Strang and Meyer 1993). While it is difficult to disentangle its effect from other factors, it is still important to consider it separately, as it can generate dynamics of S-curve adoption diffusion.

6. Discussion

This study contributes insight into the complex interplay of market factors and socioeconomic assessments of emerging CE technologies. Building on the extensive literature on techno-economic and (social) life cycle assessments (see Nayak and Bhushan 2019; Tomić and Schneider 2020 and the review by Ubando et al., 2020), a data-driven dynamic model (SIM-VOLATILE, an ABM) was developed to study the economic assessments in a more stochastic and dynamic manner in comparison to linear models with fixed assumptions. Exploring the ABM resulted in identifying pathways of emerging CE technology adoption that are built upon support schemes, market elements, and social pressure. To assist in the systematic discussion of the novelty of these research findings, Fig. 8² shows an integrative visualization of the simultaneous effect of ecosystem level factors. This figure illustrates the space where YP, S and MG_PHA are effective. The effect of ${\alpha }_1$ is excluded, as the embedded social pressure and market growth might be seen as more overlapping in comparison to the other factors.

Fig. 8.

Market formation conditions for emerging uncertain technologies: simultaneous effect of support frameworks (S), technological yield increase (YP), Market growth for PHA (MG_PHA).

Fig. 8 highlights the significant effect of YP as compared to MG_PHA and S. The technological efficiency and improvement in yield is the main adoption factor. As explained in Section 5.3 above, YP has a dual effect in which lower improvements can results in higher adoption rates. If the technology improvement occurs rapidly, then early adopters will occupy the market and increase the competition. Moreover, as explained in Section 5.1 above, the effect of S is constant, supporting the adoption rate in all configurations. Similarly, the effect of MG_PHA is generally positive; however, it is bounded to the S and YP ranges. In other words, when market growth is attractive but there are not enough subsidies, and when the technology is at a low TRL, then the adoption rate will be lower. Such results demonstrate that there are some configurations more appropriate to sector-wide CE adoption over the simulation time steps. To better illustrate how different configurations could support a sector-wide adoption, three scenarios (A, B, C; shown in Fig. 8) are selected (results shown in Table 4).

Table 4.

Three scenarios differentiated by the level of support under different market growth and technological improvement probability conditions.

Id	Scenario	YP	S*	MG	Adoption rate	Time to 95% of maximum	relative diff within steps
A	Support of slowing growing technology	0.2	0.4	0.1	0.461	23	0.186
B	Support of fast growing technology	0.4	0.4	0.1	0.224	11	0.131
C	Strong market growth & fast growing technology	0.4	0.0	0.2	0.273	12	0.301
* S stands for support as a combined measure for innovation on subsidies and operational subsidies. Therefore, S = IS + OS.

Scenario A represents the configuration where support frameworks are assigned to a slow-growing technology in a slow-growing market. In this scenario, adoption can be expected to reach a very high level but will take a long time. Scenario B extends Scenario A by assuming that support frameworks are given to a fast-growing technology in a slow-growing market. In this configuration, adoption will reach a moderately low rate in a short time. Scenario C considers a configuration where both the technology and the market grow quickly without any support frameworks. In this scenario, the adoption rate reaches a moderately low level in a short time, with a higher adoption level than in scenario B. Such results show the conditionality of support framework effects, which could come from different contextual factors (De Clercq et al. 2017; Managi et al., 2014).

6.1. Support schemes and market formation

As the results of SIM-VOLATILE have shown, subsidies are not the only most important element in a waste treatment sector with well-functioning markets (i.e., large market size), technology and networks. When potential profits or social pressure are high, innovations will happen regardless of the level of subsidies. However, investment and/or operational subsidies can represent an important incitement to adopt an innovation when the innovation system is less developed and when a certain level of subsidies are applied (see results of Scenario A, Table 4 above). Drawing the industry's attention to and subsequently installing an emerging technology offers the chance to developed it further and to become tailored to specific local needs. This generates deeper insight into market opportunities and could lead to social pressure among waste treatment companies. This is particularly relevant for emerging circular technologies, which are typically accompanied by many uncertainties.

The necessity of subsidies becomes obvious when looking at the example of CNG and bioCNG. Several waste companies own CNG trucks, own CNG gas stations, and produce biogas from waste. However, widely available biogas is not converted to bioCNG, despite the profitability of the bioCNG business case, simply because CNG converted from natural gas is still cheaper than bioCNG. Existing support frameworks dedicated to CNG also divert attention from the transition to alternative sources (Decorte et al., 2020).

Market formation and growth are crucial to innovation adoption throughout the waste treatment industry. Waste treatment plant managers are aware that adoption of a VFAP will mean entering global markets, which puts them at a disadvantage position with regard to producers already occupying global markets. Waste treatment plants adopting a VFAP face high fixed as well as variable costs, which increases risk in either mature or declining markets (Tongur and Engwall 2014). An important remark on the functioning of a CE is the need for products to be economically competitive with their traditionally produced counterparts (Korhonen et al., 2018a). In a growing market segment, however, the waste treatment plants can explore and secure market segments while optimizing their processes as well as the adopted VFAP technology. Even so, large markets are often more competitive and require high-quality products to ensure profitability (Campbell and Hopenhayn 2005). Without market growth, the timing of the adoption can increase the benefits for an individual company. For the sector as a whole, however, it means that early adopters can flood the market, which may inhibit further innovation adoption as shown in Table 4, Scenarios B and C (see Zink and Geyer 2017), and as discussed above.

6.2. Guidance and legitimacy

Beyond the purely technological and economic factors, waste treatment plants are looking for guidance in their search for potential innovations (Hekkert et al., 2007). Pressure to make the right choice is extremely high as investment in one technology results in sunk costs and potential lock-ins for further innovation (Seto et al., 2016). This creates an opportunity for institutions to join social or environmental movements. The CE can be a powerful framework to identify a certain emerging technology as legitimate (Jacobsson and Lauber 2006), especially as it is on the agenda of many governments (Gutberlet et al., 2020). For organic waste treatment, recovery of materials (and more specifically producing VFAP derived end products) is a step in the right direction according to the CE action plan (European Commission 2020b) and Waste Framework Directive (DIR 2008–98). Results of SIM-VOLATILE show that agents perceive trends and learn through interacting with others in their network. The results show that if market growth and prospective profits are low, this social pressure factor can be an important instigator that leads to higher adoption levels. This element of the model is particularly relevant for innovation questions regarding the omnipresent circular economy, and it shows a significant effect on adoption decisions as well as the gravity of proper guidance (Bergek et al., 2008; Hekkert et al., 2007; Tigabu et al., 2015).

6.3. Practical implications

In the adoption feasibility assessments, the results emphasize the interaction of technological improvements and market growth in addition to potential support frameworks. Subsidies can support profitability at any time, but under conditions of a preferred technological improvement and market growth, the early adopter has the possibility of dominating the future market. With an eye on the potential for competition, waste treatment plants may benefit from deploying an entrepreneurial approach to navigate those CE technologies in which the produced high-end product are accompanied by growth in yield and market size. Hence, the timing of the adoption will become an important factor for competition.

For policymakers, a sector-wide adoption (if this is the goal) can be achieved by supporting slow growing technologies that are currently in the development process. This will support smaller actors (i.e., late adopters) to be able to adopt the new technology and enter the market. Support frameworks always foster adoption to a degree, but their effectiveness depends on contextual factors, such as technology, market maturity, firm-specific factors and organizational culture. It is important to remember that the waste-treatment industry can be generally perceived as a traditional industry compared to other industries, with little experience in selling high-end products in competitive global markets. Support frameworks can therefore be instrumental in the adoption of CE emerging technologies. Policy should also consider the status of the market and the technology, as an exclusive focus on direct subsidies could result in less effective innovation uptakes (Xu et al., 2021).

6.4. Limitations and future research

Limitations of this study can be found in the case chosen and in the modeling approach. First, this study was performed using only one type of CE valorization technology in the context of only one industry (waste treatment). Several alternative technologies are likely to be available that a waste treatment plant could also consider adopting. Inclusion of additional technologies in the model may flatten the adoption curves while adding delays. The current model relies on data to generically conceptualize the waste treatment industry as a whole, ignoring national or regional differences; adding this type of specificity would improve the model. In case of emerging technology, such as VFAP under study here, detailed parameterization of the model could result in a non-realistic model with false predictive power. Moreover, ABM is only one tool with inherent limitations such as model dependency on the initial conditions (Manzo 2014). Scholars may also benefit by using other approaches such as network analysis and simulations.

Furthermore, within this study parameters were identified to show the effect of different technological and ecosystem-level factors; those parameters were treated as distinct concepts. One may argue that the introduced factors overlap as concepts. For example, the growth in the market can be linked to social pressure. Similarly, the existence of support frameworks can be seen as dynamic responses to social pressures. Although the model does acknowledge the connection between such factors, each factor would benefit from separate attention. Such specificity could be implemented in the model and in the decision-making process of whether to adopt the technology. The present study combines all of these factors to shed light on how to assess the decision-making process regarding the adoption of emerging CE technologies. In future research, overlapping and uncertainty could be reduced by using a participatory approach to co-develop a dynamic ABM.

Last, this study rests on the assumption that the proposed CE technology is superior to business-as-usual in terms of environmental outcomes, as indicated by environmental assessments. This can be seen as a limitation, as the focus lies mostly on the socioeconomic side of CE adoption. Future studies should contest this assumption and deploy a more consequential approach in the adoption by coupling ABM (or similar type dynamic methods) with LCAs while checking for impacts on both resources and outcomes. That future research may shed light on possible rebound effects in the CE and help to reconfigure adoption strategies.

7. Conclusions

The effect of ecosystem factors on emerging circular economy technology adoption can be characterized in three ways. First, when market growth and technological efficiency are low, subsidies and support frameworks for emerging technologies can be a very effective instrument to positively influence ROI and thus influence decisions to adopt the new technology. Second, envisioned market growth and technological efficiency were revealed as important factors. Technological efficiencies highly influence the ROI calculations and market growth motivates managers to take a risk and enter a growing market. Last, this study showed that social pressure influences decision-making, highlighting the role of non-economic factors in the decision-making processes.

Regarding competition dynamics, the effect of innovation adoptions at ecosystem level is most significant. Adoption decisions can cause market saturation in a context of high technological efficiency, resulting in voluminous production and small market growth rates. At firm level, correct timing of technology adoption can lead to profitability, while eventually inhibiting sector-wide technology adoption at the ecosystem level. Furthermore, adoption decisions affect trends and social pressure. Therefore, even a single technology adoption can also instigate a new social pressure. The decision to adopt also creates an opportunity to explore markets and further develop technological efficiency more tailored to local needs.

Clear guidance for waste treatment companies and adequate support frameworks for the waste treatment sector are shown here to be important for mitigating the effect of uncertainties regarding emerging circular economy technology adoption. A combination of favorable market situations, technological performance, social pressure and support frameworks can aid the sector-wide implementation of circular economy technologies. Moreover, policymakers may benefit from an approach that combines firm-level and ecosystem-level factors by understanding which changes in the environment can be instrumental for the adoption of emerging circular technologies and to what extent. The results also shed light on the potential evolution of the adoption process for other technologies that create high-value input products derived from waste. As such, this article may be of value for researchers wishing to expand their insights on the complex behavior of adaptation processes concerning the circular economy, as well as stakeholders, waste treatment plant managers and policymakers seeking ways to promote circular technologies.

CRediT authorship contribution statement

Siavash Farahbakhsh: Conceptualization, Methodology, Software, Validation, Formal analysis, Writing – original draft. Stien Snellinx: Conceptualization, Methodology, Software, Validation, Resources, Formal analysis, Writing – review & editing. Anouk Mertens: Conceptualization, Methodology, Resources, Writing – review & editing, Funding acquisition. Edward Belderbos: Resources, Validation, Formal analysis, Writing – review & editing. Liselot Bourgeois: Conceptualization, Resources. Jef Van Meensel: Supervision, Writing – review & editing.

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.

Data availability

• The data that has been used is confidential.