Exploring potential diffusion pathways of biorefinery innovations—An agent-based simulation approach for facilitating shared value creation

1.1 Shared and sustainable value from the forest for circular bioeconomy transition

Companies' business strategies, models, and related production have a significant impact on the transition towards greater sustainability (Loorbach & Wijsman, 2013; Baumgartner & Rauter, 2017; Bidmon & Knab, 2018). However, achieving progress in such a transition requires that the diverse efforts of several societal actors at a variety of socio-technical levels be aligned (Avelino, 2021; Köhler et al., 2019). Business organizations thus need to interact and collaborate with numerous stakeholders in their value creation processes to benefit businesses as well as their stakeholders (Freeman, 2010). This entails the generation of shared values and a joint purpose in order to encourage the value creation process (Breuer & Lüdeke-Freund, 2017; Freudenreich et al., 2020). This process is essential in harmonizing and integrating economic, environmental, and social perspectives when attempting to promote overall sustainability.

It is believed that forest-based resources and related businesses are set to play a key role in the transition from our current fossil-based economy towards a more sustainable circular bioeconomy, in particular in forest-rich countries (e.g., Austrian Federal Ministry for Sustainability and Tourism et al., 2019; Ministry of Agriculture and Forestry, Finland, 2023; Regeringskansliet, 2022; Staffas et al., 2013). New, innovative products based on industry side streams, the cascading use of wood, and higher added value have been highlighted in several public and private policies (e.g., European Commission, 2018). The European Commission, for example, has published several strategy papers on numerous sustainability-related objectives, for example, the European Green Deal (European Commission, 2019a). The European Green Deal aims at intensifying the European Union's (EU's) climate ambitions for 2030 and 2050 and at mobilizing industry in order to achieve a cleaner and more circular economy. Other related EU strategies address issues such as bioeconomy (European Commission, 2018), carbon economy (European Commission, 2021a), and the development of biorefineries (European Commission, 2021b). Biorefining, as stated in the IEA Bioenergy Task 42 (De Jong et al., 2011), is commonly defined as the “sustainable processing of biomass into a spectrum of marketable products (food, feed, materials, chemicals) and energy (fuels, power, heat)” (De Jong et al., 2011, p. 10; for further definitions, see also Berntsson et al., 2014, p. 19). The bioeconomy strategy (European Commission, 2018) aims at increasing resource efficiency, circularity, and value creation, and places particular emphasis on the importance of biorefineries in using wastes or residues.

However, developing novel products is rarely possible unless a collaborative effort is made by companies with other firms and societal actors, such as citizens, non-governmental organizations (NGOs), and governments (Jonker & Faber, 2019; Pedersen et al., 2021; Planko & Cramer, 2021). The importance of establishing shared value creation, inter-organizational networks, and collaborative business models in the development and diffusion of new products in rapidly changing business environments has been highlighted by numerous studies, in particular by sustainability-focused business model, value creation, and network management studies (e.g., Aarikka-Stenroos et al., 2014; DiVito et al., 2021; Evans et al., 2017; Freudenreich et al., 2020; Hörisch et al., 2014; Jonker & Faber, 2019; Melander & Wallström, 2022; Möller & Svahn, 2009). In addition, open innovation approaches, based on the idea that collaboration and the sharing of knowledge among diverse stakeholders is advantageous for all actors involved (Chesbrough, 2003), have been found to support shared value creation and the promotion of sustainability (Camilleri et al., 2023, see also Del Rıo et al., 2015; Melander & Pazirandeh, 2019). Chistov et al. (2021, p. 11) introduce the open eco-innovation concept “as an umbrella term for all the activities of an organization that strategically utilize access to external resources to potentialize internal eco-innovation development or to commercialize internally developed technologies and intellectual property.” The role of a supportive multi-actor business ecosystem is seen to be particularly important in the case of radical sustainability innovations (Planko et al., 2016). Hence, business management scholars have come to the same conclusion as transition scholars, that is, that the successful (wide-scale) innovation diffusion of sustainability technologies requires the aligned and collective effort of multiple societal actors. Planko et al. (2016) point out that transition scholars call this collaborative process between firms and their stakeholders as “collective system building” and further define collective system building as the “processes and activities that firms can conduct in networks to collectively create a favorable environment for their innovative sustainability technology” (Planko et al., 2016, p. 2329).

Despite the recognition of the significance of forest-based businesses and related new products in a circular bioeconomy, the unsustainable use of forest resources and lack of forest protection remain key challenges from the perspective of many societal actors (Edwards et al., 2022; Nousiainen & Mola-Yudego, 2022). In addition, transparent and science-based information—especially in the case of novel forest products—is often lacking. For stakeholders such as companies, suppliers, customers, civil society, NGOs, and politicians, such information is not only essential to informed discussion and decision-making but also serves to facilitate constructive societal discourse and the generation of shared objectives (i.e., “joint purposes”) in the value creation process (Breuer & Lüdeke-Freund, 2017; Freudenreich et al., 2020). Conflicts and uncertainties often result from inadequate information, leading to the creation of a challenging operational environment both for the businesses and their stakeholders. Näyhä (2019, 2020) concluded that creating a shared understanding among the different actors was one of the key issues in promoting transition in the Finnish forest-based sector.

1.2 Biorefining of wood-based lignin

A major material under consideration in biorefining is forest-based kraft lignin (e.g., for lightweight materials and phenol-based aromatic chemicals; European Commission, 2019b). An estimated 40–50 million tonnes of such lignin are produced globally per year (e.g., Cline & Smith, 2017). Currently, it is largely burnt on-site in pulp mills for the recovery of process chemicals and in order to gain energy (e.g., Cline & Smith, 2017). However, a surplus amount is also said to be available, and it is this which could be used to generate value-added products, while maintaining the energy supply needed on-site in the pulp mill (Dessbesell et al., 2018b; Holladay et al., 2007). The kraft lignin is, due to its availability and chemical structure, considered to be a promising bio-based compound and is expected to play a major role in biorefinery development (e.g., Ragauskas et al., 2014). Its potential for replacing fossil-based substances and its use in the development of a wide variety of non-energy products are currently under investigation (e.g., Isikgor & Becer, 2015; Ragauskas et al., 2014). Hence, among the top 20 bio-based product innovations (European Commission, 2019b), six are based on lignin as their main feedstock (i.e., plant-fiber reinforced lignin bio-composites, high-purity lignin, lignin-based carbon nanofibers, lignin bio-oil, lignin-based phenolic resins, and bio-BTX aromatics).

The industries affected by such developments, such as the forest or chemical industry, are usually science driven and path dependent (i.e., the new technologies build upon the previous ones), and innovations are usually based on technology-focused research and development (R&D) projects, rather than on consumer markets or consumer trends (e.g., Bröring et al., 2006; Hansen, 2010). Accordingly, in the biorefinery area, research at various levels, in particular in the natural sciences and engineering fields, is often driven by the idea of replacing fossil with bio-based resources in order to generate more sustainable products on the markets (e.g., Hurmekoski et al., 2019). In order to guide investment and research activities, this often entails assessment of economic feasibility and of related (mostly environmental) impacts. Such assessment may also be required by funding agencies (e.g., Mahmud et al., 2021; Scown et al., 2021; Thomassen et al., 2019). Common examples here entail optimization/design approaches (Elaradi et al., 2021; Mansoornejad et al., 2010; Murillo-Alvarado et al., 2013; Santibañez-Aguilar et al., 2014), as well as prospective techno-economic approaches (e.g., as reviewed by: Scown et al., 2021; Lo et al., 2021) and life cycle analyses (e.g., as reviewed by Parajuli et al., 2015; Vance et al., 2022). These are usually intended to narrow the scope for potential action (e.g., by means of comparisons with alternatives or by the stipulation of benchmarks) and are applied to rather specific cases. In a similar vein, technical investigations relating to lignin applications (e.g., on chemical, process engineering, and analytical issues) have been the focus of research for many years (e.g., Wenger et al., 2020). The potential macro-scale impacts that may arise during the establishment of biorefineries within the forest-based bioeconomy were investigated by Asada et al. (2020) and Jonsson et al. (2021). In contrast, the level of knowledge concerning the market environment and innovation diffusion of lignin remains rather low, and only relatively few research papers are currently available (Lettner et al., 2020; Wenger et al., 2020). Earlier research papers on the (techno-)economic aspects of kraft lignin innovations (Wenger et al., 2020) tended to focus mainly on the internal (direct) factors influencing the success of commercialization (e.g., improving technology, performance, reducing costs, and optimizing processes), while the external (indirect) factors (e.g., substitute markets, demands, and regulatory issues) were barely taken into account. In general, approaches allowing for (techno-economic) technology evaluation and technology foresight, while simultaneously including a value creation perspective for biorefinery innovations, are largely absent.

1.3 Research aims and questions

In view of the above, the consideration of intra-organizational techno-economic aspects alone seems insufficient when developing novel forest-based products and thus a wider systemic perspective is called for, one which also takes into account the external innovation environment—including diverse needs, values and goals of the various stakeholders—as well as the potential (sustainability-related) consequences of the innovations once spread. Bennich et al. (2018) stated that as socio-ecological systems are complex and that there is a clear risk of unintended consequences and trade-offs arising in any transition towards a bioeconomy. Several authors have addressed the potential related to the modeling of innovation pathways in improving knowledge creation and communication among various bioeconomy stakeholders, and in enhancing understanding of multi-actor systemic level change processes towards sustainability (e.g., Bennich et al., 2018; Yang et al., 2022). In our study, we take such a systemic approach to knowledge and value creation in novel biorefinery production. Our aim is to model and simulate potential innovation diffusion pathways for lignin-based products using an agent-based approach. By engaging diverse actors and variables from different socio-technical levels, this approach generates information for various stakeholders in lignin biorefining. Thus, the use of a systems approach to integrate both internal and external variables helps creating a more holistic understanding of how the transition to a bioeconomy may be achieved (cf. Bennich et al., 2018; Bauer et al., 2017; Wenger et al., 2020).

The objective of the present paper is therefore to model and simulate potential innovation diffusion pathways for lignin-based products by means of an agent-based approach.

2.1 Collaborative efforts for shared value creation

Despite its various historical roots (Wagner Mainardes et al., 2011), the modern development of stakeholder theory and conceptualization was initiated by the work of Freeman (1984). A central tenet of stakeholder theory is that business organizations need to focus on the interests of all stakeholders and not merely on those of shareholders (Freeman, 1984, 2010). A vast stakeholder literature now exists, incorporating quite diverse theoretical developments, views, interpretations, and applications.

Stakeholder theory perspectives have also been applied in the context of sustainability management as well as in sustainable value creation and business model studies (e.g., Freudenreich et al., 2020; Hörisch et al., 2014; Lüdeke-Freund et al., 2020). Sustainable alternatives for traditional value creation and business models aim to solve current sustainability challenges by adapting more holistic perspectives, by considering the needs of various stakeholders, and by developing new collaborative solutions when meeting demands (Hörisch et al., 2014; Pedersen et al., 2021). According to Freudenreich et al. (2020, with reference to Bocken et al., 2013; Lüdeke-Freund & Dembek, 2017; Schaltegger et al., 2015; Stubbs & Cocklin, 2008; Upward & Jones, 2016), sustainable business model literature, in particular, highlights value creation as a process entailing various outcomes for different stakeholders. For Lüdeke-Freund et al. (2020, p.81), the cornerstones of theorizing about sustainable value creation center on the issues of (a) what is value; (b) for whom is it created; (c) how is value created; and (d) who captures the value. The framework thus emphasizes (a) a need for a stakeholder-responsive definition, which in turn necessitates identifying the fundamental needs of stakeholders. It is also essential (b) to consider the respective boundaries of the systems and stakeholder networks as well as to take account diverse levels as well as spatial and temporal aspects. Further, the framework highlights that (c) new value is created in various stakeholder relationships and collaborative value creation processes, whereby the various roles of the stakeholders are taken into account. This means it is important (d) to evaluate value capture from every stakeholder's perspective and to take relevant power relations into account (Lüdeke-Freund et al., 2020). Accordingly, Pedersen et al. (2021) emphasize the importance of cross-sector collaborations as a means of “providing voice to new stakeholder groups” (p. 1042) whose views are often neglected in more traditional sustainability approaches. This thus highlights such form of collaboration as a key enabler of transition (see also Rey-García et al., 2021).

In alignment with the key principles of sustainable, cross-sectoral value creation and collaborative business models, network management scholars emphasize a need for supportive ecosystems around new innovations supplementing single company's resources and capabilities. Thus, customers and users, distributors, suppliers, and investors are all needed in performing practical commercialization tasks, facilitating innovation diffusion and in creating new markets (Möller & Svahn, 2009; Sandberg & Aarikka-Stenroos, 2014). Oskam et al. (2020) explored the tensions existing between various actors in innovation ecosystems with sustainability goals. They suggest that when developing a sustainable business model innovation, ecosystems need to engage in a process which they call “valuing value,” that is, a process in which the aim is to search for a result that satisfies all participants and thus one which facilitates collaborative processes among cross-sectoral actors. In a similar manner, Breuer and Lüdeke-Freund (2017) emphasize the crucial role of the networks and shared values of the stakeholders involved in the innovation processes and the related management. Further, they argue that values-based innovation networks and business models can help significantly when addressing complex sustainability challenges.

All in all, solving difficult sustainability problems requires collaborative, cross-sectoral effort and resources among stakeholders, and this needs to be guided by shared values and objectives. In practice, however, establishing and managing such networks and finding solutions fulfilling the demands of multiple actors is highly challenging. The present paper attempts to provide an approach aimed at facilitating knowledge exchange and at increasing understanding among the various actors, and thus at mitigating related tensions.

2.2 Agent-based simulation of bio-based (eco-)innovation diffusion pathways

Traditional models of innovation diffusion are in general often based on the Bass model (Bass, 1969), and tend to focus on aggregate trends and behavior rather than on the decisions and interactions of individuals (e.g., Kiesling et al., 2012; Zhang & Vorobeychik, 2017). However, several relevant aspects, such as actor communications, networks, heterogeneity (as described by Rogers, 1983), and the inclusion of variables helpful for decision-makers (i.e., providing predictive and/or explanatory power in such models), are difficult to capture with such aggregate models (Kiesling et al., 2012; Zhang & Vorobeychik, 2017). Agent-based models (ABMs) represent one possible approach to dealing with such challenges by modeling complex emergent phenomena bottom-up (agents' micro-level interactions) and have therefore gained popularity in innovation diffusion research (e.g., Garcia, 2005; Kiesling et al., 2012; Zhang & Vorobeychik, 2017). In the realm of eco-innovations, which can be defined as innovations that (intentionally or unintentionally) “improve the environmental performance” (e.g., Carrillo-Hermosilla et al., 2010, p. 1075), technological innovations (including both preventive and curative environmental measures) are a common focus of research. The organizational, social, and institutional innovations, however, have so far received comparably little attention in this context (Rennings, 2000). While these types of innovations are said to be more challenging to coordinate in practice, targeting them in addition to the more commonly considered technological innovations may lead to higher environmental benefits overall (Machiba, 2010). In exploring innovation diffusion phenomena and underlying actor behavior, ABMs can improve explanatory power by incorporating elements from (neo-)institutional economics (e.g., bounded rationality) (e.g., Kiesling et al., 2012; Moncada et al., 2017). Also, ABMs are capable of including both the technical considerations and the social structures relevant in technology innovation diffusion (e.g., Moncada et al., 2017). This contrasts with the comparatively limited explanatory power of innovation diffusion models based on classical economics which depend on relatively restrictive assumptions, for example, on homogeneous actors with complete information (Kiesling et al., 2012; Moncada et al., 2017). In addition, agent-based simulation is considered especially appropriate where communication within a social network is important (Macy & Willer, 2002), and where the diffusion of innovations may be regarded as a communication process for increasing awareness of the innovation (Rogers, 1983). Previous studies have shown that the topology of an underlying network has a significant impact on innovation diffusion (e.g., Bohlmann et al., 2010; Kiesling et al., 2012; Garcia, 2005; commonly used network types in ABMs: Barabási & Albert, 1999; Watts & Strogatz, 1998; Erdös & Rényi, 1961). Conceptual ABMs are seen “as ideal learning tools for scientists to understand a system under a variety of conditions by simulating the interactions among agents” (Zhang & Vorobeychik, 2017, p. 3), and thus do not always aim at being descriptively accurate and/or predictive (Garcia, 2005). However, owing to problems related to the relative simplicity of agent rules (“toy models”) (Garcia & Jager, 2011; Zhang & Vorobeychik, 2017) and to predictive validity (Kiesling et al., 2012; Zhang & Vorobeychik, 2017), empirically grounded ABMs—that is, where empirical data are used for initialization, parametrization, and/or validity evaluation—are increasingly being applied to tackle real-world problems (Kiesling et al., 2012; Zhang & Vorobeychik, 2017). In particular, such empirically grounded ABMs (e.g., Garcia & Jager, 2011) may be used for the purposes of foresight, policy analysis, and decision support (e.g., by enabling a better understanding of how and why innovation diffusion pathways shape up) (e.g., Kiesling et al., 2012). The reviews by Kiesling et al. (2012) and by Zhang and Vorobeychik (2017) summarized several approaches in these areas and elaborated on the challenges and progress regarding model validation, since “all of them are, at least to some extent, speculative thought experiments until data for validation becomes available” (Kiesling et al., 2012, p. 219). Nevertheless, it is believed that empirically grounded ABMs are becoming more and more acceptable, both as a research tool (e.g., in the facilitation of theory construction) and as a decision-support tool (e.g., for managerial diagnostics and policy recommendation) (Kiesling et al., 2012).

With respect to innovation diffusion ABMs, to date, the main areas of application for ABMs have been agriculture, transportation, energy, and environmental innovation (Kiesling et al., 2012). There are, for example, agent-based approaches on specific wood markets which, while not explicitly relating to biorefinery operations, do include industrial/pulp wood consumers in some way (e.g., Holm et al., 2018a, 2018b; Kostadinov et al., 2014; Scholz et al., 2020). In the area of biorefineries and related innovation diffusion processes, most research papers deal with the topic of biofuels (e.g., Günther et al., 2011; Moncada et al., 2017; Stummer et al., 2015; van Tol et al., 2021; Yang et al., 2022). As one example, Moncada et al. (2017) placed a particular conceptual focus on institutions, and related this to the conceptual underpinnings of complex adaptive systems theory, (neo-)institutional economics (e.g., bounded rationality), and socio-technical systems, and also investigated import tariffs and international trade flows (Moncada et al., 2017; van Tol et al., 2021). These authors also noted that ABMs, in comparison to optimization approaches, were suitable for addressing socio-economic aspects, such as the relevant behavior of supply chain actors, particularly since the biofuel supply chains' economic performance “depends on the interaction of technical characteristics (technological path-ways and logistics) and social structures (institutions and actors behavior)” (Moncada et al., 2017, p. 895). In another paper, Yang et al. (2022) developed a community communication tool for miscanthus-based biofuels, which was based on multiple sources and included a range of actor groups (farmers, industry, community, government, and biofuel consumers). Based on these previous works on (agent-based) innovation diffusion modeling and innovation-related studies in the bio-based area, the present paper has aimed at developing a similar approach using a case study on kraft lignin and derived products (LPF resins and CFs).

3.1 Model overview

The main agent groups and decision mechanisms of the Biorefinery Products Innovation Diffusion (BioPID) ABM, are described in the following, taking the example of the innovation diffusion of kraft lignin-containing products, and are illustrated in Figure 1. In brief, the main actor groups (agents) in the BioPID ABM are the commodity producers (kraft lignin or crude oil-based counterpart—i.e., phenol or acrylonitrile), the (potential) processors of the bio-based commodity that can either continue to use the crude oil-based input or adopt the lignin-based input to produce (resins or CFs), and the consumers. The main change factors in the model are (1) the techno-economic potential regarding the bio-based commodity (extracted kraft lignin) producers, (2) the scenarios covering the GPs (based on oil-based commodity price projections, with and without carbon pricing, as well as lignin producer variants with regard to their number, capacity, and pricing strategies), and (3) the demand for bio-based-products (including the willingness-to-pay more/less for these) on the consumer side. The main result of a simulation run is then the amount of kraft lignin traded on the market each year as an indicator of the speed and success of the innovation diffusion process under the circumstances investigated.

Details regarding the behavior and data foundation of the respective agent groups are given in Section 9. The simulation mechanism is described thereafter, in Section 14. The two case studies included are described in Section 15. In Section 19., the scenario development is described (including variations in future oil prices and variations in lignin prices, lignin pricing strategies, and the number/capacities of potential lignin producers).

3.2 Agents

The fundamental characteristics of the four agent groups—the crude oil commodity producer, lignin producers, lignin processors (GPs), and consumers—are described in the following. Further related information and specifications (concerning the model mechanisms, scenarios, and parameter specifications) can be found in the respective sections below.

3.3 Simulation mechanism

The basic mechanism of the model may now be described as follows. The bio-based commodity producers (kraft lignin producers) check with the GPs and with the goods consumers, what they would be willing to pay for the lignin. Regarding the GPs, this entails ascertaining whether their cost of goods production (i.e., of CF or resins) would be equal or lower when switching to lignin; for the consumers, the kraft lignin producers must ascertain their willingness to pay more for the bio-based variant. If this is not zero, then this price level is taken as the upper cost limit. Assuming these conditions are acceptable for the kraft lignin producer and depending on individual switching probabilities (which result from both the network and the risk affinity of the assigned adopter group), this agent may then produce the lignin at the end of the time step. In the subsequent time step, the kraft lignin producer offers this lignin on the market. The consumers can then decide whether or not they want to order the lignin-containing product from the GP. This decision depends on things such as the good's price and willingness to pay. Following this, and in a similar fashion, GPs have to decide whether or not to switch to using the bio-based lignin. Assuming the latter is found to be acceptable (this depends on lignin availability, ascertained profitability of bio-based goods production, and individual switching probabilities which result from both the network and the risk affinity of the assigned adopter group), the GPs start producing the product—that is, CF or phenol formaldehyde resin—at the end of this time step. This product is then bought in the subsequent time step by the consumer. Once a GP has switched production to bio-based feedstock use, we assume that this process will not be reversed (e.g., owing to contractual obligations, investments made etc.).

Computationally, a simulation run is implemented as follows: To initialize the model, the following agents are generated: a single crude oil producer, the kraft lignin producers, the GPs, and the goods consumers. Agents within the same agent group are connected in a network. In the present study, we use a small-world network (Watts & Strogatz, 1998). This is commonly applied in innovation diffusion research owing to its similarities with real-world social networks (e.g., Bohlmann et al., 2010; Kiesling et al., 2012), and to the fact that it shows “properties of both random graphs (small diameters) and regular lattices (high degree of clustering)” (Bohlmann et al., 2010, p. 747). Note, that the proximity of two agents need not be spatial in nature. For example, proximity may exist in terms of joint research projects, existing collaborations, or even competition.

Once the agents are set up, the simulation begins in discrete time steps. Here, the resolution chosen is 1 year. During each time step, the different agent types perform different actions. The crude oil producer produces and sells oil to all GPs who demand it. Since this producer represents the global oil market, it is modeled as a single agent with infinite capacity and the price is given by the respective oil price projections (International Energy Agency, 2021). Lignin producers exhibit similar behavior. Based on their ability/willingness to bear risk, they may enter the lignin market and sell lignin to GPs. In contrast to the oil producer, the capacity of lignin producers is finite and their prices are not uniform. Of all agents modeled, the GPs (i.e., the lignin processors) exhibit the most complex behavior. They first undertake a feasibility analysis in order to compare their cost of production using the crude oil-based commodity to the cost of production using lignin. Since base costs are the same in both variants, the relevant comparison is P_oil>(P_lig+C_add)*A_lig+C_swi.

This shows the price of oil $$$ {P}_{\mathrm{oil}} $$$, the price of kraft lignin $$$ {P}_{\mathrm{lig}} $$$, the additional cost $$$ {C}_{\mathrm{add}} $$$, the relative amount of lignin needed when compared to oil $$$ {A}_{\mathrm{lig}} $$$, and the cost of switching the production to bio-based materials $$$ {C}_{\mathrm{swi}} $$$. If the feasibility check reveals that switching to lignin is profitable, the GPs check the market in order to verify that enough lignin is available to satisfy their demand. Two factors influence the probability of adoption in the BioPID ABM. The first is related to the ability/willingness of an individual potential adopter and represents a proxy for various internally based pre-conditions (i.e., an umbrella term for individual and internal factors, such as the presence of existing patents), which given the context of the present paper, we have based on Rogers (1983, 2002). Hence, the agents were divided into five categories: innovators (2.5%), early adopters (13.5%), early majority (34%), late majority (34%), and, finally, laggards (16%) (Rogers, 1983). These adopter groups can be regarded as “ideal types”, that is, a conceptual formulation of five categories from a continuous innovativeness variable (Rogers, 1983). To define a base chance of adoption per year, we used these five categories and their respective distribution. The initial base chances of adoption (i.e., the respective probabilities) for the various categories were set as follows: innovators: 25%, early adopters: 16%, early majority: 9%, late majority: 4%, and laggards: 1%. These figures are used to represent the likelihood that producers begin with lignin production, the likelihood that GPs buy and use the lignin in their production, and the likelihood that consumers are willing to pay more for the lignin-containing product (in cases where the respective parameter is set to >0).

In addition to this rather internal, adopter-related parameter, the second factor influencing the possibility of adoption is network effects, that is, those items representing the organization's external innovation environment (e.g., engaging in research collaboration). Where network neighbors have already adopted the new technology, the base chance is multiplied by (1 + [n/10]), with n being the number of neighbors that have already adopted. This means, for example, that for an early adopter with five network neighbors already using the innovation, its own probability of adoption rises from 16% to 24%.

Once a GP switches to lignin-based production, it buys lignin from lignin producers instead of buying the crude oil-based commodity from the oil producer. In every time step, the amount of lignin traded provides a quantitative measure of how well the innovation is diffusing throughout the system. The specific cases investigated are described in detail in the following sections.

3.4 Biorefinery study cases

The respective product cases (lignin phenol formaldehyde resins, lignin-based CFs) are now described below. In both cases, there is one supplier for the respective fossil-based counterpart, and 100 consumers among which the total demand (phenol for LFP resins or CFs, depending on the case) is evenly allocated. The number of repetitions was 200 for the main results and 50 for the sensitivity analysis results. At the end of the section, the approach taken to estimate the potential emissions savings resulting from the corresponding substitutions is then briefly presented.

3.5 Scenario development

The model is used to investigate and compare the diffusion processes arising in different scenarios. Several authors have recommended that, concerning innovation in the forestry sector, greater emphasis be placed in the pulp and paper industry on developing an innovation-friendly culture, on cooperation and collaboration, and on targeting market and customer needs (e.g., Hansen, 2010; Leavengood & Bull, 2013; Näyhä & Pesonen, 2014). To date, however, innovation in this industry—which is considered to be relatively capital intensive and risk averse—tends to be rather incremental and production-oriented, with cost-reduction and/or quality improvement playing major roles (e.g., Hansen, 2010; Leavengood & Bull, 2013; Näyhä & Pesonen, 2014). The model scenarios were therefore designed in such a way that price developments (fossil-based and bio-based), pricing strategies, techno-economic considerations (bio-based), and production capacities (bio-based) play major roles. The scenarios developed are designed to reflect different possible combinations of (1) future oil prices and (2) lignin pricing strategies, including variations in number/capacities of potential lignin producers.

3.5.1 Fossil-based commodity prices

With respect to the oil price, we investigated variants based on projected oil price developments as well as variants that additionally include carbon pricing.

Different potential commodity price developments based on oil price projections are considered. For this, prices in the model develop annually until 2050 in accordance with the four crude oil price development projections published by the IEA, each with or without carbon pricing (International Energy Agency, 2021). As oil commodity prices correlate with oil prices (illustrated in Figure 2), the IEA oil price projections (International Energy Agency, 2021: price scenarios, page 101, Table 2.2) were taken as reference in order to model commodity prices. The IEA projections are related to four “scenarios,” that is, Net Zero Emissions by 2050 (NZ), Sustainable Development (SD), Announced Pledges (AP), and Stated Policies (SP), for all of which crude oil price levels were indicated for 2030 and 2050, respectively. The stated prices were converted to €/t (exchange rate of 2021; Deutsche Bundesbank, 2022; BP p.l.c., 2021) and, for the sake of simplicity, it was assumed that the annual model prices always developed linearly across the given projections, starting from 2021. The resulting changes in prices (€/t) were then transferred to the commodity cases.

The NZ and AP scenarios were then subject to further investigation. The former represents the scenario with the strongest price decrease, and the latter the scenario with approximately stable prices (AP).

To take account of carbon pricing, we used the IEA CO₂ price projections for the years 2030, 2040, and 2050 (International Energy Agency, 2021: projections for European Union (SP), Advanced economies with net zero pledges (AP, SD), and Advanced economies (NZ), page 329, Table B2), starting from 0 €/t in 2021 and then with prices always developing linearly across the given projections. For the fossil carbon pricing, the CO₂ emissions per commodity tonne were calculated in a simplified fashion using the chemical formulas of the commodities (phenol C₆H₆O, [poly-]acrylonitrile C₃H₃N) and then calculating the carbon content per tonne of commodity (phenol 76.6% m/m, [poly-]acrylonitrile 67.9245% m/m). Here, it was assumed that one molecule of CO₂ would be formed from one molecule of carbon (resulting in 2.81 t CO₂ from 1 t phenol and 2.49 t CO₂ from 1 t [poly-]acrylonitrile).

The price developments of the commodities for the model runs (eight projections—i.e., four with, and four without, carbon pricing) are illustrated in Figure 2. For further investigation, the AP scenario with carbon pricing (AP + CT) was chosen, representing the scenario with the strongest price increase.

3.5.2 Structure of kraft lignin producers (pricing strategy and potential capacity)

The variations developed for the kraft lignin producers (with variations in number of producers, capacities, and pricing strategies for separated lignin) are based on (1) kraft lignin production costs derived from TEAs, (2) a kraft lignin price as commonly assumed in the literature (see, e.g., Gosselink, 2011; Hodásová et al., 2015; Gabriel et al., 2017), whereby we assume the latter to be based primarily on basic separation expenditures and opportunity costs, and (3) minor-change assumptions, with only few producers present in the market taken as a reference.

For analyzing the lignin producer configurations, which are based on TEAs (variant), EoS were derived from 42 published TEAs (see also Krassnitzer et al., 2023), dealing either with obtaining lignin from black liquor by acid precipitation or by membrane filtration (Arkell et al., 2014; Axelsson et al., 2006; Benali et al., 2014; Culbertson et al., 2016; Davy et al., 1994; Dieste et al., 2016; Holmqvist et al., 2005; Jönsson et al., 2008; Jönsson & Wallberg, 2009; Kannangara et al., 2012; Laaksometsä et al., 2009; Lindorfer et al., 2019; Loutfi et al., 1991; McKeough et al., 2014; Olsson et al., 2006; Tomani, 2010; Uloth & Wearing, 1989). For detailed numbers, see Table A1. These 42 cases are modeled as respective agents in the model. The structure of the lignin producers as illustrated in Figure 3 represents the scenario TEA. The kraft lignin price (which equals the cost of production in this scenario) is on average 252 €/t in this lignin producer variant, and the average capacity is 44,853 t/year (in total, approximately 1.9 million tonnes per year).

In the variations where the lignin producer configuration is based on lignin market price expectations (PEX and PEXx variants), there are also 42 potential kraft lignin producers and the lignin production capacities as in the TEA variant, but the cost of lignin production (in our model this is also the selling price) is (initially) set to 300 €/t, which represents a price per tonne of kraft lignin commonly assumed in the literature (e.g., Gosselink, 2011; Hodásová et al., 2015; Gabriel et al., 2017). Price estimations of lignin in the literature are often based on lignin type and grade, and usually start at approximately 50 €/t (estimated fuel value, where the basis for comparison is often the price per energy unit of natural gas) and may reach 750 €/t and more, also depending on purification costs etc. (Gosselink, 2011). We assume in the PEX variant, that a market exists at a stable price of 300 €/t for basic quality kraft lignin, and that this can be sold to processors for further modification. In PEXx, we start with 300 €/t for kraft lignin, with the price changing in accordance with changes in crude oil prices (described in the following section). It is assumed that besides being based on basic preparation costs (separation, drying etc.), this price is also based on opportunity costs related to alternative oil-based products. The lignin producer variants capture 42 potential lignin producers and are thus believed to be reasonably representative of a potential European-level kraft lignin market.

Given the currently rather low innovation diffusion level of kraft lignin (with only four major suppliers in Europe and America) (e.g., Dessbesell et al., 2020), the variants with 42 potential lignin producers are quite optimistic. Therefore, in order to establish a fairly realistic point of reference, we analyzed lignin producer configurations based on only minor changes of the status quo (MC and MCx variants). Here, a total of four potential kraft lignin produces is considered with a total capacity of 107, 000 t of kraft lignin per year. Similar to PEX and PEXx, there is an MC variant with a stable price of 300 €/t lignin, and an additional MCx variant with kraft lignin prices following crude oil prices. These minor change variants, in terms of producer capacities and pricing characteristics, more closely resemble the potential for kraft lignin production in Austria. In order for innovation diffusion to occur in the simulation, the willingness/ability to bear risk was set as described in Section 14. This setting is likely to be more optimistic than that found in the status quo.

For each of the two study cases, 40 combinations were initially simulated. These resulted from the eight fossil-based commodity price variants and from the five variants for lignin pricing/capacity. After an initial screening of the results, the most significant scenarios were picked out for purposes of comparison and for ascertaining the major findings of the whole study (see Table 3).

TABLE 3. Screened and selected commodity producer scenarios (green check marks: considered in detail; yellow check marks: considered for discussion).

• Abbreviations: AP, announced pledges (oil price projection variant); CT, carbon pricing; Fos., oil price projections; Lig., lignin producer variants; MC, stable lignin producer variant “minor change”; MCx, lignin producer variant “minor change” with lignin price development following the oil price development; NZ, net zero emissions by 2050 (oil price projection variant); PEX, stable lignin producer variant “price expectation”; PEXx, lignin producer variant “price expectation” with lignin price development following the oil price development; SD, sustainable development (oil price projection variant); SP, stated policies (oil price projection variant); TEA, stable lignin producer variant based on techno-economic assessments.

4.1 Scenario selection

In order to highlight the key results and findings of the model runs, the four scenarios with a green check mark, shown in Table 3, are dealt with in detail in the following sections. Scenario MC3, being closest to the current status quo, was selected as a reference scenario, and TEA7 was chosen because it showed the most stable patterns of innovation diffusion; PEX3 (representing a larger number of potential producers) was selected as a counterpart mainly to MC3, and PEX7 (representing a different pricing scheme of the potential lignin producers) was chosen as a counterpart mainly to TEA7. The other selected scenarios (see yellow check marks in Table 3) were chosen to cover as wide a range as possible of the fossil-based commodity price developments and all respective lignin producer scenarios (i.e., covering number of potential adopters, their capacities, and pricing strategies). While the four main scenarios are considered in detail in the following sections, the other 11 scenarios are mainly used to illustrate the possible consequences of further variations in fossil-based and commodity-based price developments (the associated plots for those can be found in Appendix A).

Figure 4 provides an example of basic monitoring of a single model run, including the adopted kraft lignin and the crude oil commodity replaced over time, the respective adoption feasibilities, and behavior of (lignin and goods) producers, as well as the average cost comparison procedure of actors during the decision process.

4.2 Minor-change scenario (MC3)

The assumptions of the MC3 reflect a situation close to the current one whereby technical lignin is used as a feedstock. This means, the scenario assumes a small number of potential providers (four) who would offer lignin after taking account of (mainly) the opportunity costs, that is, accounting for the basic lignin separation costs and not using it as a fuel. Furthermore, in order to allow innovation diffusion to be observed, it is assumed that certain structures also exist within the innovation environment (e.g., in terms of innovativeness and investment opportunity) that would favor the adoption of the innovation. Simulation runs indicate that, under these conditions, diffusion occurs slowly and only reaches a relatively limited market share in both application fields. In total, the use of lignin in this scenario remains clearly below the technical potential. Here, diffusion in CF applications is more dynamic, while phenols for LPF resin application require the use of a larger total amount of lignin. As shown in Table 4 and in Figure 5, lignin would replace up to 47% (by mass) of the fossil-based counterpart (acrylonitrile) in the assumed lignin CF market by 2050, but only up to 5% of the phenol in the potential, lignin-containing, phenolic resin market. In total numbers, given the respective assumptions, about 9,000 t/year of CFs would be produced in this scenario, and assuming a total share of 15–20% (by mass) phenols in the final LPF resins, this would mean that approximately 200,000–300,000 t/year of resins would contain a share of lignin-derived phenolics. The higher relative market penetration in the case of CFs reflects the different market structure prevailing, for example, with smaller total capacity, a different GP structure, and higher estimated value added. As a consequence, innovation diffusion in this field of application is also less influenced by increasing additional processing costs (see Table A2). In contrast, the amount of lignin needed to cover 5% of the phenolics in the resin market is likely to be almost twice that needed to reach 47% of the CF market. Regarding the lignin producers, on average, two lignin producers are sufficient to satisfy respective lignin demands, while due to the different market structure in the phenol and CF cases, about three and eight GPs, respectively, would enter the market. The simulation indicates that the rate of innovation diffusion in both application fields is lower in the period 2040 to 2050 than in 2030 to 2040.

These results of the MC3 are used as a reference for interpreting other scenario results.

4.3 Scale-up scenarios (PEX3 and PEX7)

The simulations of these scenarios reveal that assuming a larger number (42) of potential suppliers, as compared to the MC3, results in a much larger market share being reached (illustrated in Figure 6). Simulations indicate that 67% (in the LPF resin case) to 74% (in the CF case) of the target application market share is acquired by 2050 (Table 4). In terms of adopted lignin, and when assuming low to medium additional cost levels, the variation in oil prices, for example, due to the introduction of carbon pricing (PEX3 versus PEX7), exerts only a minor influence on the innovation diffusion process. There is, however, a marked difference between the PEX3 and the PEX7 scenarios in the case of high additional cost levels. Low crude oil prices together with high additional costs make lignin use economically unfeasible and result in no innovation diffusion. In contrast, high oil prices seem to offset the impact of high additional costs over time and, thus, allow more actors to switch to bio-based production. Up to an average of 22 (in the phenol case) and 6 (in the CF case) lignin producers sell their lignin in these scenarios, resulting in up to 19 phenol or 9 CF producers that switch production. In the simulation variants where lignin prices move in accordance with the oil prices over time (see Appendix A, scenario rPEXx3/cPEXx3), the results in the roughly constant oil price scenario (AP3) are similar to their constant lignin price counterpart variants. However, in the falling and rising oil price variants, the respective results of the lignin price-adjusted scenarios become more similar to each other since the oil price fluctuations are offset to a large extent (e.g., in the case of higher additional cost variants and rising oil prices, innovation diffusion of lignin no longer occurs when its price also increases over time).

4.4 Individual pricing scenario (TEA7)

Assuming that the pricing strategy of potential lignin suppliers is not driven solely by opportunity cost considerations but also by individual firm strategy and site-specific production costs, when compared to the PEX7-scenario, similar results to PEX7 are observed, in particular, for low and medium additional costs (illustrated in Figure 7 and Table 4). This indicates a relatively saturated diffusion path. This pattern, however, changes under high additional costs. A small number of lignin producers providing lignin at relatively low prices enables early diffusion, even under high additional costs, or at least under high uncertainty regarding such costs. In particular, regarding the application in CFs, the production cost-oriented pricing favors early diffusion of lignin even under high additional cost assumptions. This can be explained by the much smaller overall market size. In such a market, only a small number of lignin producers, supplying at the lowest prices, are sufficient for effective diffusion. In the case of phenol resin applications, this pattern changes. In the larger overall market, the early diffusion under high additional costs is slower compared to that in the low and medium additional cost scenarios. However, the effect levels out by 2050 as a result of the rising oil prices. Thus, from an aggregate perspective, the combination of individual lignin prices that do not increase over time, with rising crude oil prices, can be regarded as the least risky scenarios in terms of overall lignin innovation diffusion. In this diverse structure of heterogeneous lignin suppliers and processors, there is more scope for finding respective “matches” that are economically feasible and thus, theoretically, also better for coping with any additional costs potentially incurred in the lignin processing. Regarding the lignin producers, however, the average number of lignin producers that provide lignin is lower with individual pricing (15 and 4 producers for phenols and CFs, respectively) than in the PEX scenarios, because the larger producers profit more from EoS. Furthermore, like in the PEX scenarios, up to 19 (phenol) and 9 (CF) lignin processors switch production in the individual pricing scenario.

In several of the scenarios involving falling crude oil prices (see Figures A5 and A6; NZ1 scenarios), no matter which lignin pricing strategy is followed, switching is initially feasible, but then the lignin use gradually becomes unprofitable as a result of increased cost pressure, and thus, from the perspective of GPs, this would be a misinvestment.

4.5 Potential emission savings

The potential emission savings were estimated for two selected scenarios: MC3, which is closest to the status quo, and TEA7, which showed the most stable innovation diffusion patterns. Figure 8 depicts the potential emission savings gained by replacing fossil raw materials (phenols and acrylonitrile) with extracted lignin. The dashed line in each graph represents the potential upper limit of emission savings, based on the theoretical maximum amount of fossil raw materials replaced when the predicted respective markets are fully saturated (1,000,000 t and 20,000 t in the phenol and CF cases, respectively). The potential emission savings differ in the two cases. On the one hand, potential emission savings are influenced by the respective innovation diffusion rates prevailing in the scenarios investigated. On the other hand, emission savings also vary as a result of differences in lignin to commodity substitution proportions, in market potential, and in the respective impacts of the commodities replaced (phenols [RER, market for phenol] and acrylonitriles [GLO, market for acrylonitrile]). Here, the impacts resulting from the processes required for the respective raw lignin modifications have not been considered due to the lack of relevant data concerning mass and energy balances. Life-cycle phases such as the use and end-of-life were also not considered.

Nevertheless, due to the overall forecasted market size, the potential for emission savings appears vastly higher in the phenolic resin case. Although the production of acrylonitrile from fossil feedstocks has a larger carbon footprint than that of phenol production (3.62 kg CO₂-Eq/kg compared to 2.94 kg CO₂-Eq/kg), the replacement of fossil-based feedstocks in the phenolic resin market may lead to higher overall emission savings. By contrast, due to the small market size and relatively rapid saturation in the CF case, a relatively large part of the emission savings potential is already exploited in the scenario with minimal changes (cMC3), compared to the savings in the stable, high diffusion scenario (cTEA7).

4.6 Sensitivity analyses

In order to validate the model and gain insight into how sensitive the results are with respect to specific input parameters, a one-factor-at-a-time sensitivity analysis was performed. As a starting point for this, we chose the TEA7 scenario, as this showed the most promising and stable results. For each parameter sweep, we only modified one parameter, keeping all others constant, and observed the outcome of the innovation diffusion process. The latter is quantified in terms of the amount of lignin traded in the year 2050. The results of this analysis are presented in Figure 9 (LPF resin) and Figure 10 (CF).

Concerning the additional cost, most of the time the actual value of the additional cost has no significant effect on the diffusion process. However, the diffusion process is hindered once the additional cost is large enough to make the overall production cost, when using lignin, comparable to that when using crude oil. Depending on the investigated scenario and product, this break-even point or “deterministic feasibility” is found at different values. Since the transformation expenditures (depreciation) represent only a tiny fraction in the overall production cost structure (see red bar in Figure 4)—when calculated per tonne of lignin and depreciated over 10 years—they have no significant influence on diffusion in our model, unless values are increased to (probably) unrealistic levels. However, in practice, this parameter would not only include expenditures related to the investment but also switching costs of various kinds, and thus, the maximum value used for such costs was € 100,000,000.

To analyze the influence of the number of potential lignin producers, we initialized the model in the TEA scenario, yet we generated more or fewer lignin producer agents by randomly removing or duplicating those from the baseline scenario. This also changed the overall amount of lignin available on the market. This analysis provides different results for the two products investigated. While the CF case is not much influenced by a reduced amount of available lignin (because the projected market for this product is saturated relatively fast), the LPF resin case is nearly linearly dependent on the number of lignin producers, until saturation is reached at a point of roughly 40 producers. We performed a similar analysis for the number of GPs. Here, we kept the overall production capacity constant but varied conditions to reflect whether capacity is generated by a few large or by many smaller companies. Again, the LPF resin case shows much more sensitivity, especially for a low number of GPs.

Finally, the influence of the innovativeness of the firms was investigated. When initializing the simulation, each firm was assigned a probability of moving into the most innovative adopter group (see Section 14 for details). This resulted in an increase in the amount of traded lignin for both the products investigated. Note that the amount of traded lignin surpasses that reached when compared to the situation where there are no additional costs.

5.1 Discussion of key results

The major results and their interpretation are given in Table 5, and the related implications are discussed in the subsequent sub-section.

5.2 Theoretical and practical implications

Novel forest-based businesses are important in the transition towards a more sustainable circular bioeconomy. The successful diffusion of innovations—such as the substitution of kraft lignin-based products for their fossil-based counterparts—is not possible unless the stakeholders involved engage in some form of collective effort and pursuit of shared goals. This includes actors directly connected to the lignin-based supply chains as well as wider societal groups. There is a potential of goal conflict (Table 5, items 5, 7, and 9), for example, in terms of how individual actors or groups of actors might maximize benefits (e.g., with respect to various pricing mechanisms) versus how the overall innovation diffusion process might be most effective (which in turn could benefit the environment as well as actors). Agent-based simulation of innovation diffusion pathways for kraft lignin products using BioPID not only introduces multi-faceted situations and mechanisms behind potential diffusion pathways but also acts as a tool enabling diverse societal stakeholders to engage in informed knowledge exchange and communication (cf., e.g., Macy & Willer, 2002; Yang et al., 2022). In other words, information provided by BioPID can promote dialogue between different actors, thus helping them to set common goals (or joint purposes, see Breuer & Lüdeke-Freund, 2017; Freudenreich et al., 2020).

Despite wide-scale agreement on the need to abandon fossil-based production, barriers to change still exist in the form of social and political resistance, inflexible planning policies, general reluctance, etc. (e.g., Béfort, 2020). The creation of a suitable political environment and power structure is imperative in supporting any transition towards a non-fossil society (e.g., Köhler et al., 2019; Meadowcroft, 2011). In the context of bioeconomy businesses, coherent, long-term policies play a crucial role (Kelleher et al., 2019). Political factors are also important drivers in the case of lignin production and processing, where the limited number of lignin suppliers on the market, as well as the high level of uncertainty related to the additional costs of lignin processing, represent major challenges for innovation diffusion. The results described above show that the interplay of several different variables is decisive for the successful, “lower-risk” diffusion of bio-based innovation (Table 5, items 3, 4, 8, and 10). Once again, this stresses the need for coherent, long-term policies. Often, dominating incumbent actors in the regime level aim to prevent the (market) entry of new business entrants into regime (Geels, 2010). In such a case, the bio-based materials would have to compete with their fossil-based counterparts in the petro-chemical industry. Despite the relatively blurred boundaries between niche and regime players in the case of several bioeconomy applications (see, e.g., Hermans, 2018), the pulp and paper companies may be regarded as mature regime-level actors, and the processing of lignin to (partly) replace crude oil-based products would require niche-level innovations and their scaling up to more large-scale production. However, this is likely to entail overcoming a range of challenges related to lock-ins and path dependencies of the mature forest-based companies (e.g., Markard et al., 2012). Overall, public policies should thus provide a stronger normative directionality (see Köhler et al., 2019) and create a more compelling environment in order to enable easier entrance and participation of various actors in the markets and greater market functionality (e.g., avoidance of monopolies and improved security of lignin supply). BioPID can provide policy makers in all societal levels involved in bioeconomy development with better understanding of the lignin-based innovation system, and thus promote better informed and coherent decision-making in the transition towards non-fossil societies.

Further, promoting the diffusion of lignin-based products requires a collaborative effort among niche companies: they need to collaborate horizontally with other niche actors as well as vertically with the regime actors providing raw material. The results of the BioPID approach indicate that both a limited number of lignin suppliers and GPs (phenols) on the market (Table 5, items 1 and 11) as well as a lower level of innovativeness among respective actors (Table 5, item 12) pose challenges to innovation adoption and subsequent diffusion. While the setting of shared goals with suppliers and producers would promote collaboration, the establishment of relationships with traditional and dominant regime actors (i.e., lignin producers) with rigid organizational cultures still remains a challenge (e.g., Kuhmonen et al., in press). Collaborative efforts and potential joint strategies among niche actors can provide a way forward in these relationships. The crucial importance of collaboration has also been highlighted by many management studies on cross-sectoral value creation, networks, business ecosystems, and open innovations. Melander and Wallström (2022), for example, highlighted the importance of horizontal collaboration in finding innovative, more environmentally friendly solutions. At the same time, they highlighted that environmental and economic incentives as well as trust between the companies are a prerequisite for establishing collaborative relationships. However, niche companies are often very reluctant to share proprietary knowledge on their business models and technologies, since their competitiveness is based on these unique solutions and this may weaken their competitive position and increase the risk of them being exploited by other firms (e.g., Kuhmonen et al., in press; Melander & Pazirandeh, 2019). Nevertheless, areas of potential collaboration still exist, for example, in lobbying campaigns, and in learning processes relating to their (shared) suppliers, customers, investors, or infrastructure. They can also jointly promote suitable legislation and regulative networks. In the case of lignin-based value creation, this could mean, for example, establishing common infrastructure for material characterization, or that knowledge gained throughout research projects—which often receive funding from the public budget—is made available to potential adopters in order to reduce information asymmetries and increase innovativeness. In addition to relationships between the companies, supportive ecosystems around new innovations include actors such as customers and end users as well as investors (see, e.g., Möller & Svahn, 2009; Sandberg & Aarikka-Stenroos, 2014). All these actors can utilize the information created by BioPID when making their (more conscious) purchasing or investing decisions. Altogether, in the realm of innovation in the bio-based field, major challenges exist in relation to organizational culture, cooperation, and collaboration (e.g., Bröring et al., 2020; Golembiewski et al., 2015; Hansen, 2010; Leavengood & Bull, 2013; Näyhä & Pesonen, 2014). Companies in the forest-based sector are often reluctant to participate in open networks and open innovating (D'Amato et al., 2020; Näyhä, 2021). The knowledge provided by BioPID may serve to reduce mistrust among different companies and stakeholders and thus help to create a more secure basis for collaboration.

Given both the different innovation diffusion patterns arising from the two product categories studied (Table 5, items 2, 6, 9, and 11), as well as the high number of potential lignin applications currently being researched (e.g., Wenger et al., 2020), it is important not to neglect the need for generating balanced product portfolios while considering issues at different levels of analysis (e.g., firm-level, market structures in socio-technical system, and macro-level consequences). This is a challenge, particularly from the perspective of the companies involved and their strategic management and decision-making. Given such a context, we believe that BioPID can provide managers with more relevant information for decision-making on their product portfolios.

By enhancing the availability and flow of information among actors in diverse interactions, and thus serving to increase transparency and decrease mistrust among various actors, the BioPID has value for diverse actors in the operating environment of companies. In sum, therefore, by reducing levels of uncertainty, we believe that BioPID can make the fundamental needs of diverse stakeholders more visible in value creation processes (see Lüdeke-Freund et al., 2020, p. 81).

5.3 Limitations and outlook

As acknowledged by several authors (e.g., Kiesling et al., 2012; Zhang & Vorobeychik, 2017), the (predictive) validation of prospective agent-based innovation diffusion simulation remains challenging for several reasons (e.g., lack of appropriate data). This is also the case for the BioPID approach (an exploratory tool rather than a predictive tool). Particular strong variations exist in the literature regarding the expected additional costs and, in particular, with respect to CF (cf., e.g., referenced sources in Section 16). This is at least partly related to the lower TRL as compared to that of phenolic resins (e.g., European Commission, 2019b), resulting in higher associated levels of design uncertainty, and, possibly, of higher design freedom (eco-design paradox; e.g., Genus & Stirling, 2018; Poudelet et al., 2012). In general, it is assumed that the quality of the bio-feedstock is such that it can be used in the respective products and markets, which implies that current technological challenges—for example, concerning lignin characterization, constant homogeneity, etc. (e.g., Holladay et al., 2007; Ragauskas et al., 2014)—can be adequately addressed in the future, which may appear rather optimistic at this point. Several scientific sources point to the relevance of “learning effects” in technology development (e.g., Daugaard et al., 2014; Lieberman, 1984; Thomassen et al., 2020). These are expected to occur as cumulative capacity rises, and in addition to the benefits of EoS, contribute to a lowering of unit production costs. Currently, due to the lack of relevant data available, the quantification of “learning effects” remains a difficult task. Overall, to address (some of the) prevailing uncertainties—both with regard to the data and potential future preconditions—several scenarios were introduced (e.g., with regard to raw material prices and pricing strategies), broad ranges and combinations of prices and costs were tested in a range of simulation runs (including indication of standard errors), and sensitivity analyses were conducted on a range of parameters.

In the BioPID model, once basic criteria had been fulfilled, the lignin was adopted as soon as a lignin producer and a GP entered a contractual agreement, which probably is a shortcoming in our current assumptions (e.g., no minimum number of lignin suppliers required on the market and no possibility to switch back). Related expenses only became a significant decision criterion when they were relatively high. Thus, further adaptations of the model could be made in regard to the actual switching to lignin (e.g., introducing entry barriers or by refining estimations concerning the extent or nature of expenditures). The perspectives of various affected actor groups (e.g., company representatives, societal actors, and land owners)—including their different perceptions, needs, and (partly conflicting) goals—need to be better understood and ecological aspects need to be considered more strongly for the biorefinery innovation diffusion pathways to actually contribute to sustainable development (e.g., Dieken et al., 2021; Mustalahti, 2018; Näyhä, 2019; Tan et al., 2019). Closer examination of such issues could include analyzing the various preconditions faced by individual actors (ability/willingness) and network structures representing the embedding of actors in (external) social structures (e.g., Bohlmann et al., 2010; Kiesling et al., 2012). The gathering of survey data (e.g., from respective actor groups) could help generate more empirical information on social structures, to refine the BioPID parameters accordingly, and thus introduce a stronger empirical reliability and validity into the model. In the current BioPID model, a major focus was placed on supply chain actors (who are equipped with a certain heterogeneity). Other actors have so far only been included implicitly: for example, in the current BioPID model, high costs of crude oil-based inputs could be interpreted as government measures (including carbon pricing), low “additional costs” for bio-based innovations as incentive measures, or an increased innovativeness (see sensitivity analysis) as the promotion of network activities and R&D projects. Other (horizontal) actor groups such as regulatory organizations thus could receive more emphasis in the future. With regard to the actor groups explicitly included, some issues that could be further explored are related to lignin market requirements from the perspectives of the potential lignin buyers (market structure such as minimum number of lignin suppliers on the market, required lignin qualities, etc.), refinement of the preconditions and behaviors of the consumer agents (e.g., willingness to pay for a bio-based counterpart product, attitudes towards partly bio-based materials; Günther et al., 2011; Stummer et al., 2015; Zwicker et al., 2023; Ruf et al., 2022), and analysis of actor networks (e.g., empirically derived, actual networks; loose/highly connected networks) and corresponding refinement of the model networks.

Regarding the sustainability-related impacts, in the present paper, calculation of possible emission savings merely served to establish a link between the BioPID model and approaches to analyzing potential sustainability consequences arising from corresponding innovation diffusion pathways. However, considerations of possible changes to previous mass and energy balances arising during production, of subsequent life-cycle stages (e.g., lightweight carbon fiber materials in vehicles could lead to reduced fuel consumption and emissions), of geographical issues (e.g., regional energy mixes), or modeling efforts to address the uncertainties of parameters, just to name a few examples, were beyond the scope of this paper and remain the subject of future research. The highlighted potential goal conflicts and trade-offs (see Table 5), including analysis of which individual actors or actor groups win or lose in different scenarios, may also be an issue worth exploring further. With regard to some issues, the BioPID ABM already allows for more extensive analyses than those conducted in the present paper. For example, at the level of the GPs, it would be possible to put the focus of analysis on individual actors, for example, the respective successful production capacities of GPs adopting lignin, as well as on the resulting price (ranges) of the final products (phenol for LPF resins and CFs) in different scenarios. Focusing on individual actors' characteristics could provide valuable insight into the potential roles and outcomes relating to individual actors in the different scenarios and could also be a starting point for better incorporating management perspectives. This could be supported by participative and collaborative approaches using BioPID as a facilitator (e.g., Yang et al., 2022). On the system level, new and radical biorefinery innovation pathways require active forms of collaboration (ideally based on shared values), multidisciplinary, multi-objective and participatory approaches, more efficient (environmental) management practices, and improved and more transparent decision-making processes (Dieken et al., 2021; Mustalahti, 2018; Näyhä, 2019; Tan et al., 2019).