Manage risk of sustainable product–service systems: a case-based operations research approach

Sustainable product–service systems (SusPSSs) offer an innovation-driven approach to production based on providing results or functions with minimal material use and emissions. Networks of SusPSSs partners are central to the decision-making of sustainability policies. Evaluations and assessments of network-oriented risks sources are therefore crucial to informing an industrial firm’s reorientation towards SusPSS. Traditionally, these risks beleaguer production and continue to grow in significance with complex production and innovation processes. This article presents a novel operations research application for evaluating network-oriented risks of industrial firms in pursuing SusPSSs. The model conceptualises a framework for network risk metrics and applies a fuzzy-based multi-criteria decision-making technique to evaluate levels of risk associated with reorientations to SusPSS approaches. It takes explicit account of multiple risk sources in aiding decision-making and assists in indicating strategies for improving business sustainability. In addition, it compares and ranks alternative SusPSSs as a system and on an indicator basis, which is a practical and effective decision support tool. A case study of an industrial firm is conducted to verify the effectiveness and applicability of the proposed approach in supporting firms’ decision on SusPSSs.


Introduction
Firms are increasingly acknowledging the importance of adopting sustainability policies and transforming these insights into operations that are environmentally and socially sound. For a start, directives such as the ISO 14000 series of standards have steadily gained acceptance and offered practical guidelines for environmental management systems to minimise negative environmental impacts and continuously improve sustainable production processes (Corbett and Kirsch 2001;Gerbens-Leenes et al. 2003). These standards draw upon interpretations of sustainability (e.g., Costanza and Patten 1995) for policy level implementation and have spurred the development of sustainable systems of finance and exchange (Seyfang and Longhurst 2013). Programmes such as Sustainable Product Development Network (Sus-ProNet), the non-profit Global Reporting Initiative™ and the Environmental Sustainability Index have also offered reporting guidelines and provided data on sustainability performance for increased global awareness of the need for sustainable production. These schemes promote ideas that are based on cleaner production, designs for the environment and eco-design (Roy 2000;Wang et al. 2015a). Nevertheless, global challenges including climate change, increasing population and pollution concerns about the planet Earth's ability of accept industrially generated wastes. As a result, research has intensified in recent years into exploring new ways of delivering quality products and services with efficient usage of resources and energy and minimal waste during production and consumption. A Sustainable product-service system (SusPSS) is one of the various sustainability initiatives designed to capture this development.
in the SusPSSs. The research conceptualises a framework for risk evaluation metrics and applies a fuzzy-based operations research technique to evaluate levels of risks associated with reorientations to SusPSS approaches. The research focuses on SusPSS as an avenue for fostering sustainability and offers a multi-criteria decision-making (MCDM) approach to evaluate network-oriented risks for the supply chains of SusPSSs. MCDM approaches are often employed with case studies to provide insights into how the application of proposed approaches can support rational decisions for various business applications (Wang et al. 2015a;Kumar et al. 2016;Viriyasitavat 2016;Teixeira et al. 2018). Apart from providing guidance for future SusPSS strategies, insights from the proposed model can be used to inform planning and control decisions and enhance the formulation of competitive business models that leverage communication and interactions in networks for SusPSS as discussed by authors such as Mont (2002), Briceno and Stagl (2006) and Krucken and Meroni (2006).
In the remainder of this article, we present the theoretical foundations for our research. We then describe our proposed model before applying it in a case study involving a manufacturing firm. We conclude with the theoretical and practical contributions of the research and potential future research directions.

Sustainability and product-service systems
Citing international concern for increasing production and consumption patterns, SusPSS advocators have long argued that these value propositions offer a viable avenue for transforming service economics into functional economies (Mont 2002). Service in this context "may refer to the role of the service sector in the economy, or to a company's offerings to its customers, or to the service (utility) provided by a product" (Heiskanen and Jalas 2003). In most developed economies (e.g., United States, Germany, United Kingdom and Japan), the service sector contributes most of the employment of the total labour force. In these economies, services are used to reinforce products, and alternative strategies for product use are explored (Mont 2002;Chen et al. 2012). In contrast, functional economies optimise "the use (or function) of goods and services and thus the management of existing wealth (goods, knowledge, and nature)" (Stahel 1997). Functional economies treat physical products as capital assets with a view to leveraging value-added services that efficiently use resources and enhance the life of physical products. Consequently, driven by the need to offer insights into how firms can functionally advance economies, SusPSS commentators have argued that the positioning of a SusPSS as a key contributor to sustainability offers a viable route for simultaneously enhancing the competitiveness of producers and minimising the environmental impact of production (Manzini and Vezzoli 2003;Mont et al. 2006).
Although several classifications for PSS offerings can be found in the literature (Roy 2000;Mont 2002;Heiskanen and Jalas 2003;Halme et al. 2004;Cavalieri and Pezzotta 2012), the ternary of orientations for achieving PSS value propositions proposed in Tukker (2004) has been the most widely applied. This classification, discussed in Tukker and Tischner (2006) with regard to sustainability potentials, consists of product-, use-and result-orientations of firms. The feasibility of these value propositions is dependent on business viability with regard to competitiveness in the market place, customer satisfaction that is reliant on customer education and involvement during design processes, and environmental soundness that gauges environmentally based superiority over traditional business models (Mont 2004).
SusPSSs have emphasised end-of-pipe attitudes and dematerialisation strategies to fulfil the needs of customers in more sustainable and life-cycle oriented ways. "End-of-pipe" attitudes focus on reducing post-production pollution and waste (Roy 2000), and to "dematerialise" production means to reduce material flow in production processes (Mont 2002) by implementing reuse and remanufacturing techniques and environmentally friendly technologies (Manzini and Vezzoli 2003). In addition, physical products need to be optimised to lower their environmental impact through design processes such as life-cycle oriented product design, eco-design, Design for Disassembly, Design for Recycling, and the sustainable product and/or service development (SPSD) process (Maxwell and Van der Vorst 2003;Aurich et al. 2006). These optimised designs result in integrated (or cleaner production) technologies that ecological economists tend to contrast with end-of-pipe technologies (e.g., Belis-Bergouignan et al. 2004).
However, researchers have critiqued the benefits of a SusPSS by highlighting rebound effects that offset SusPSS benefits (Manzini and Vezzoli 2003;Halme et al. 2004) and the awareness that eco-efficient services may not always be preferable compared to products (Mont 2002). Furthermore, the "real strength" of value propositions is their relevance to customer needs (Tukker and Tischner 2006), and these needs are shaped by the attitudes of customers and clients towards sustainable production and consumption (Mont 2002;Briceno and Stagl 2006). These factors reinforce how adopting a holistic view of sustainable consumption and production is required to evaluate and improve SusPSS characteristics and phenomena, such as environmental waste and systems innovation (Mont 2004;Ness 2008;Ceschin 2013). In this article, current research is enhanced through the introduction of an approach for evaluating network risk.
Although there is a common theme among scholars on the relevance of networks for SusPSSs, limited insight has been provided into what risks influence the reorientation of an industrial firm to more SusPSS approaches. This is because the majority of studies have focused on SusPSS evaluations to facilitate shifts towards dematerialisation of production and eco-efficiency mainly in terms of service thinking and the development new business models. These studies have also identified relationship and participation factors for creating awareness of sustainability potentials, overcoming barriers to SusPSS adoption and highlighting opportunities for leveraging innovation through SusPSSs. Yet, there is a need for industrial firms to apply a holistic evaluation that identifies network-oriented sources of SusPSS risk and prioritises the perceptions of these risks. Therefore, the current state of the literature necessitates an effective risk evaluation of SusPSSs with potentials for not only  Mont (2002) Idealises the motivation, elements and characteristics of SusPSSs

Conceptual and fundamental
Organisational involvement and close cooperation to implement environmental profiles Vogtländer et al. (2002) Proposes the eco-costs/value ratio (EVR) model, on the basis of the life cycle analysis (LCA) methodology for assessing the eco-efficiency of SusPSSs Conceptual and applied Influence eco-efficiency decisions through stakeholders' participation Heiskanen and Jalas (2003) Discusses and evaluates the eco-efficiency of non-material services, result-oriented services, product-based services and service approach facilitated eco-design

Conceptual and applied
Stakeholders' influence on company activities Manzini and Vezzoli (2003) Conceptualises a working framework of elements and characteristics to describe the sustainable potentials of product-service systems Empirical and fundamental Involvement of stakeholders along value chains throughout product life cycles Maxwell and Van der Vorst (2003) and Maxwell et al. (2006) Proposes a method to aid in evaluating sustainable criteria during product and service development Anttonen (2010) Evaluates the value chains of service providers and uses insights from the evaluation to generate a typology of service profiles

Empirical and fundamental
Supplier-side opportunities in view of changing supplier-customer relationships Geum and Park (2011) Evaluates the benefits of the product-service blueprint as an approach to clarify the products and services relationship

Conceptual and applied
Behaviour of actors and spatial relationships within network Conceptual studies generate or re-interpret ideas, empirical studies use experience or observation data to draw conclusions, applied studies solve specific problems and fundamental studies generalise to build theories enhancing managerial decisions for a firm's reorientation to more SusPSS approach but also for advancing the overall sustainability agenda.

A holistic framework for risk evaluation
To ensure a proper risk evaluation of SusPSSs, it is important to apply a holistic view of SusPSSs incorporating engineering characteristics, customer satisfaction and sustainability issues (Xu 2000;Lin et al. 2012;Wang and Durugbo 2013). Motivated by a holistic approach to evaluate risks of a SusPSS, literature was examined to formulate a framework detailing risk sources. It was for this purpose that a focus on supply chains was adopted. Modern supply chains represent network-oriented approaches to production that link suppliers (producers, processors, marketers and distributors) and customers for four main reasons: (a) to progressively add and accumulate value; (b) to retain competitive advantage; (c) to reduce costs of operations and (d) to improve collaboration and coordination among suppliers and between a supplier and a customer (Cooper et al. 1997;Themistocleous et al. 2004;Chen et al. 2017). This is in contrast to traditional, sequential industrial supply chains that are characterised by material flow downstream (supplier to customer) and information flow upstream (customer to supplier), with each division receiving information in sequence. The management of a constant and dynamic flow of information, material, cash, product, process and product/service value is vital to the success of PSSs especially in light of the complexity of roles/relationships and life cycle challenges (Dimitriadis and Koh 2005;Lockett et al. 2011;Durugbo and Riedel 2013;Xu 2015).
Overall, five main network-oriented sources of risks were identified: demand, supply, manufacturing, control and technology, as illustrated in Table 2. Demand risk represents unpredictable variations in the quantity, quality and timing of demand that results in excessive product inventory or loss of opportunities (Davis 1993;Wang and Durugbo 2013). Supply risk is triggered by variability and inconsistency by suppliers that lead to delayed, deficient or defective deliveries (Davis 1993;Wang and Durugbo 2013). Manufacturing risk is caused by unreliable production processes that result in volatility in process performances (Davis 1993;Chen and Paulraj 2004). Control risk refers to unpredictable and unknown variations of system controls within supply networks (Childerhouse and Towill 2004). Technology risk relates to technology changes within an industry sector and potential technology failures that disrupt business and service outages (Chen and Paulraj 2004). These widely studied risk sources plague supply chains, business environments and industrial networks (e.g., Davis 1993;Wang and Durugbo 2013;Choi et al. 2016), and they serve as a baseline in the framework for industrial partners to evaluate the network-oriented risks for the partner's reorientation to a more SusPSS approach.
Despite the increased attention on SusPSSs in the literature, few studies have focused on risk management of SusPSSs that support firms' strategic implementation of building a more sustainable product-service system. Furthermore, there is a lack of practical tools that help firms make appropriate decisions in the implementation of SusPSSs. Although various conceptual frameworks/models relating to SusPSSs have been provided, there is a lack of case-based operations research approaches supporting effective and sensible decisions on the adoption of SusPSSs. Therefore, this research aims to fill these gaps by developing a comprehensive framework for risk management of SusPSSs.  Impact of pre-process change on post-process Davis (1993), Fynes et al. (2004), Chen and Paulraj (2004), Ho et al. (2005), Bhatnagar and Sohal (2005), Wang and Durugbo (2013) and Wang et al. (2017) C 32 Impact of pre-process output on post-process performance

Research method
In this study, a decision model made up of two parts, as shown in Fig. 1, is proposed. The first part is a holistic framework that identifies network-oriented sources of risk as described in Sect. 2.3, and the second is a set of operations research approaches that analyses risk level and prioritises SusPSS strategies.

Set of operations research methods for risk analysis
Here, an integrated operations research approach combining fuzzy risk assessment, Fuzzy Technique for Order Performance by Similarity to Ideal Solution (Fuzzy TOPSIS), and information entropy is proposed to evaluate risks of SusPSSs. Fuzzy risk assessment is applied to rate the risk level of different items for alternative SusPSSs. Fuzzy set theory is often employed by operational research scholars to deal with uncertainties and subjectivities in risk assessment. Fuzzy risk assessment has the advantage of quantifying imprecise information and incorporating vagueness in the assessment. An increasing number of attempts to explore fuzzy set theory have been undertaken in the risk assessment domain in the last decade, including food safety risk (Davidson et al. 2006), environmental risk (Pan and Chen 2012), and supply chain risk . It is an effective way to deal with uncertainties inherent in the risk analysis of SusPSSs. The weights of decision criteria estimated through the entropy theory are integrated with fuzzy TOPSIS to generate a decision index value to rank alternative solutions. Fuzzy TOPSIS evolved from the original TOPSIS technique (Hwang and Yoon 1981) is then applied for risk evaluation of alternative SusPSSs. The main concept of TOPSIS is to define the positive ideal solution that has the lowest risk level of different risk sources and the negative ideal solution that has the highest risk level of different risk sources (Zhang et al. 2011; Wang et al. 2014). Fuzzy set theory is often incorporated with TOPSIS to deal with the uncertainty and imprecision inherent in the process of mapping the perceptions of experts (Krohling and Campanharo 2011), and it has been employed in areas such as logistics provider selection (Singh et al. 2018), green supplier selection and order allocation (Govindan and Sivakumar 2016), and eco product design (Wang et al. 2015a, b). In relationship to this study, Wang and Durugbo (2013) applied fuzzy TOPSIS to evaluate alternative solutions through analysing network uncertainty for industrial product-service delivery. However, the main focus of their research is centred on evaluating the uncertainty of a service network that delivers an industrial product-service system. While focusing on risks management for sustainable product-service systems, this research proposes a more effective and objective weighting method for evaluation criteria.
For most MCDM problems, the weights of the decision criteria are crucial to evaluating alternative solutions. Often weights are determined by key decision-makers. This type of weight calculation methods, e.g., Analytical Hierarchical Process (AHP), is often regarded as subjective weighting (Xu et al. 2003;Chan et al. 2013). Here, a more objective weighting method, information entropy weighting, was employed. Information theory, developed by Shannon (1948), is a measure of how much information is associated with a given state of events. It is concerned with quantification of information, which is also known as entropy approach. This method is particularly useful for assigning a weight to each criterion because it does not require an individual decision-maker to rank the criteria, and the relative weight of each criterion can be obtained using rather simple calculations (Zou et al. 2006;Erol et al. 2011;Zhang et al. 2011).

Steps for risk analysis
First, a panel of experts is organised for risk assessment of SusPSSs. For each SusPSS proposition, knowledge experts are asked to rate the probability of risk and severity of the consequence with respect to risk items using a range of linguistic expressions, as displayed in Table 3. A score is then denoted as P ik and S ik for the probability and severity ratings of risk item i rated by expert k, respectively.
Based on the risk ratings from the expert panel, a triangular fuzzy number (TFN) is assigned to the probability,P i (L P i , M P i , U P i ), and severity,S i (L S i , M S i , U S i ), of risk items with respect to different SusPSSs. Using the TFNP i (L P i , M P i , U P i ) as an example, L P i indicates the lower bound of probability rating as L P i min(P ik ); U P i indicates the up bound of probability rating as U P i max(P ik ); and M P i is the geometric mean of all the experts' risk probability rating for risk item i. It can be obtained as: In the same way, the lower bound (L S i ), geometric mean (M S i ), and the up bound (U S i ) of TFN for the severity of risk item i can be obtained. The two risk factors are then multiplied to determine its risk level. To simplify the calculation, a standard approximation for fuzzy multiplication is used as: With the TFNs of the probability and severity ratings, the risk level of risk item i with respect to SusPSS solution j can be calculated individually as: Following the sources of Zhang et al. (2011) and Wang et al. (2015a), the procedure of the fuzzy TOPSIS method can be described as follows: First, a fuzzy decision matrix,D, is first constructed based on a given set of risk sources and their associated items.
(4) whereR i j is the value of alterative j with respect to risk item i, which is further represented by TFNs. Then the normalized decision matrix is established, which allows comparison of the various risk items. An element of the normalized decision matrix is calculated as follows: where 1 ≤ i ≤ n for n risk items, and 1 ≤ j ≤ m for m SusPSS propositions. In this research, entropy approach is used to calculate the weights of evaluation criteria. The calculation of the entropies is straightforward. According to the decision matrix, we calculated the information entropy of i th criterion, defined as: in which K 1/ ln m and i 1, 2, . . . , n. n is the number evaluation items and m is the number of alternative SusPSS solutions considered in the evaluation. To avoid the insignificance of ln f i j , we stipulated: Here, R i j is the defuzzified risk value ofR i j using the Centre of Area method given as: The weight of entropy of ith criterion can then be defined as: in which 0 ≤ w i ≤ 1, and n i 1 w i 1. This method is particularly useful for assigning a weight to each risk item because it uses rather simple calculations and does not require individual decision-makers to separately rank them for weighting purposes. In other words, decision-makers do not have to collect additional data to calculate the weights. Now the weighted decision matrix is computed by multiplying the weighting derived from the entropy analysis to the normalized decision matrix as: whereṽ i j r i j × w i . Then we calculate the distances from negative and positive ideal solutions. Let A − and A + denote the fuzzy negative ideal solution (FNIS) and fuzzy positive ideal solution (FPIS), respectively. According to the weighted normalized fuzzy-decision matrix, we get: whereṽ + i andṽ − i are the fuzzy numbers with the largest and smallest generalized means, respectively. For each column i, the largest generalized mean ofṽ + i and the smallest generalized mean ofṽ − i are derived, respectively. Consequently, the FPIS (A + ) and the FNIS (A − ) are obtained. Then the distances (d + and d − ) of each alternative SusPSS from A + and A − can be calculated by the area compensation method as: Finally, the alternative SusPSSs can be ranked by their relative closeness indexes, which are determined by a combination of the difference distances d + and d − as follows: The set of alternative SusPSSs can then be ranked from the most preferred to the least preferred feasible solutions according to the corresponding index values.

Case background
The case organisation is a manufacturing company in south-eastern China. The company produces stainless steel bands as well as stainless-steel consumer products such as kettles and kitchen sinks. The company mainly produces customised stainless steel bands and supplies them to external clients that use stainless steel bands as raw material. The company also manufactures consumer products (e.g., kettles and kitchen sinks) using its own stainless steel as the main raw material. Despite rapid business expansion in the past 10 years, both the stainless-steel production and consumer goods operations are facing tough challenges due to intense competition in both domestic and overseas markets, soaring prices of raw materials, energy and labour, and regulatory requirement of energy use and pollution control. These challenges pose a significant question mark about the sustainability of its business.
In response, the company has made great efforts in the last couple of years to deliver a more sustainable industrial system. For example, in 2012, the company invested over half million US dollars in technologies to replace its diesel-powered annealing furnaces with natural gas furnaces. The change not only reduced the energy bill by 20%, but also significantly decreased its unit carbon emissions. Moreover, the management team was keen to explore new income streams from servitization since such a movement may secure future growth and lead to a more sustainable business. Such a strategic move would require the company to develop new capabilities that offer new services and solutions and supplement their original product offerings. However, similar to the implementation of new changes, the servitization strategies may cause a variety of risks in the wide supply chain network along with their potential benefits. Therefore, it was critical to evaluate and manage the associated risks before any new value propositions are implemented. The proposed operations research enabled decision model was applied to the case company with a view to providing some strategic guidance for its move towards a more SusPSS.

Data collection
The data collection for the empirical inquiry was carried out in the two stages. An expert panel was assembled that included the managing director, finance director, marketing manager, procurement manager, and three factory managers for stainless-steel production, kettle production, and kitchen sink operation. This selection of panel members is to ensure good understanding of the firm's overall strategic direction as well as the capabilities and challenges of its main operations. Furthermore, panel members have good knowledge and experience of its supply chain on both the demand and supply sides. A panel discussion was conducted to explore possible SusPSS value propositions. The discussion led to four SusPSS value propositions: intermediate services (e.g., breakdown, repair, and condition monitoring) (A1), shared utilisation services (A2), product-life extension services (A3), and demand-side management (A4) for risk evaluation. The panel members were then asked to rate the probability and severity of risk items with respect to four alternative SusPSSs. For each risk item, the panel members were required to give linguistic classification of the two risk factors. The collected data were used as input for risk evaluation, and the panel was then presented with analysis results. In the second stage, the managing director was interviewed two years after the initial SusPSS risk evaluation to find out what was eventually implemented and how the evaluation helped inform the organisation's decision-making for pursuing a SusPSS.

Case analysis
The proposed fuzzy methodology was applied to the case organisation to provide some strategic guidance for its transitions to a more SusPSS. First, linguistic risk ratings from expert panel members were used as raw data input for risk evaluation of four alternative SusPSS propositions considered in the case organisation. Through Eqs. (1)-(3), the risk level for all the risk items with respect to alternative SusPSS propositions was derived. As displayed in Table 4, the levels of risks associated to different risk sources varied considerably between various SusPSSs considered in the study. For instance, for shared utilisation services (A 2 ), the risk level was low for items in the supply risk source, i.e., complexity of critical material (C 25 ) and delivery frequency of critical material (C 28 ), but high for items in the control risk source, i.e., information accuracy (C 41 ) and organisational change through delivery of new services (C 44 ).
The derived risk levels of SusPSS propositions were then used as input for entropy analysis. The normalized weightings for risk items from entropy analysis are also described in Table 4. The risk items with more substantive differences in risk level between alternative SusPSSs were assigned high weighting. In contrast, the risk items with small magnitudes of difference in risk level were assigned low weighting. For instance, rapidness of technology change in industry (C 51 ) was given a much higher weighting compared to complexity of procurement technology for critical material (C 54 ), although the two risk items derived from the same technological risk source. Following the fuzzy TOPSIS procedures outlined in Sect. 3, the distances from the positive and negative ideal solution (d + and d − ) and the relative closeness to the ideal solution (θ j ) were calculated through Eqs. (4)-(14). Table 5 shows the final results.
The findings in Table 5 give a clear indication of which SusPSS value propositions the company should focus on to deliver a sustainable product-service system. In this case, shared utilisation services (A 2 ) top the ranking list and should be recommended. Spare capacity in the key production processes usually results in enormous waste, which has a negative impact on the firm's economic and environmental dimensions Triple Bottom Line. Consequently, to maintain business viability, shared utilisation services appealed most to the management team. The indication from follow-up discussions with the participants was that this option had more potential but less network-oriented risks compared to other SusPSS propositions. It was also noticed that similar relative closeness indexes were obtained for SusPSS propositions such as intermediate services (A 1 ) and product-life extension services (A 3 ), although one proposition may be more exposed to specific network-oriented risks than the other. It is due to the fact that these indexes take into consideration all the sources and levels of risk that each SusPSS proposition was rated. Such analysis is useful for choosing the most suitable strategy for the organisation to improve sustainability performance and enhance supply chain resilience.

Follow-up interview
Two years after the initial risk evaluation of alternative SusPSSs, an interview was conducted with the managing director who was involved in the original risk evaluation. The main purpose for the interview was to find out what was eventually implemented since then and how the risk evaluation of SusPSS propositions helped the company make strategic decision on SusPSSs. The company did move ahead with shared utilisation services as that was regarded as the most viable SusPSS solution to generate new income streams through a service-oriented business model. The managing director also acknowledged that although the decision was not a direct response to the initial risk evaluation of the alternative SusPSSs, the evaluation exercise had certainly contributed to their decision. More importantly, from a practical point of view, the network-oriented risk metrics and the MCDM methods enabled them to understand the risks Table 4 Fuzzy risk assessment and entropy analysis results

Risk sources
Risk items  associated with SusPSSs in a systematic and holistic way. The evaluation helped them be proactive in mitigating and managing risks in the implementation of SusPSSs. Furthermore, the implementation of any new business strategy requires firms to carefully assess the costs and benefits. The same rule applies to those manufacturing firms pursuing SusPSSs. The risk evaluation of SusPSSs was conducted in an effective and efficient manner that does not demand extensive resources and time. With the input from managers, who have good understanding and knowledge about the company's internal operations and external relationships with supply chain partners, the evaluation provided useful insights into the exposed risks with respect to various SusPSS options. Comparison of these options developed the firm's capability of foreseeing and responding to potential network risks and enabled managers to make important strategic and tactical decisions on SusPSSs.

Discussion and conclusions
With the increasing emphasis on end-of-pipe attitudes, dematerialisation strategies and optimised designs, it is important that industrial firms have effective management tools for understanding and analysing risks associated with delivering sustainable product-service systems (SusPSSs). However, network-oriented risk is a complex subject involving vagueness and ambiguity in decision-making. With this in mind, this article presents a case based operations research approach that supports the reorientation of industrial firms towards more SusPSSs by performing a structured analysis of network-oriented risks and evaluating different SusPSS value propositions. The proposed decision model includes two elements: (1) an outline of a network-oriented risk matrix derived from the literature and (2) a set of operations research approaches that assess risks using fuzzy risk assessment, calculate the importance weights of risk items using entropy analysis, and evaluate alternative SusPSS solutions using Fuzzy TOPSIS technique. There are several reasons that the proposed methods can be employed by the industrial organisations that want to explore SusPSSs. First, it provides a critical assessment of a combination of operations research approaches to evaluate SusPSS value propositions. Second, it seeks to take explicit account of multiple risk sources in aiding decision-making and assists in identifying strategies for improving business sustainability. Third, it compares and ranks alternative SusPSS value propositions as a system and on an indicator basis, which is a practical and effective decision support tool. Finally, via a real case of an industrial firm, the research offers useful insights into how the application of the proposed methodology can support rational decision-making processes to adopt the SusPSS approach.
This research makes three key contributions. First, it provides a review of existing approaches to the evaluation of SusPSS and the main relationship/organisational factors that cause risks for a SusPSS. Generally, these risks plague decision-making associated with service offerings, service costs, eco-efficiency potentials, social factors, interaction strategies, capabilities and partnerships. Second, it delineates criteria for evaluating the network-oriented levels of risk associated with reorientations to SusPSS approaches based on a holistic network view of SusPSS as supply chains. The main innovation about the measurement items of supply network-oriented risks is that these risks beleaguer production and continue to grow in significance with complex production and innovation processes, and evaluations and assessments of network-oriented risks are therefore crucial to inform an industrial firm's reorientation towards SusPSS. In addition to conventional supply, demand, and process related risks used in supply chain risk management (Pan and Chen 2012;Wang et al. 2017), our case analysis also demonstrates the importance of incorporating other dimensions (e.g., control and technological risks) in the evaluation. Third, there is a lack of case-based operations research approaches supporting effective and sensible decisions on the adoption of SusPSSs (Wang and Durugbo 2013;Baines and Shi 2015). This research fills the gap by proposing an effective decision model that integrates a holistic risk evaluation framework and practical modelling approaches including fuzzy risk assessment, entropy and Fuzzy TOPSIS for evaluating network-oriented risks of reorientations to more SusPSS. The research demonstrates that a case-based operations research approach and specific insights derived from our findings contribute to the SusPSS debate, highlighting factors and mechanisms that can make SusPSSs successful.
This research also provides important managerial implications. Competing through services is a critically important for many industrial firms in the future competition. However, the uncertainties embedded in the complex and unpredictable wide economic environment will have a significant impact on how the services-oriented business model can be more effectively harnessed. Similar to the implementation of any new business strategies, industrial firms have to evaluate benefits and risks before committing investments to SusPSSs. Although SusPSSs are critical for many industrial firms to achieve sustainability objectives, firms have to cautiously assess potential benefits and risks involved in their strategic move on SusPSSs. Firms are more likely to invest in strategies that can bring economic growth and service improvement without compromising social and environmental performances. It is essential to foresee and respond to potential risks associated with wide supply networks. One main benefit of implementing the proposed case-based operations research approach is that it provides a more holistic view of risks associated with SusPSSs and enables firms to more pro-actively assess and manage the risks and support their strategic decisions on SusPSSs. Nevertheless, the expert panel members' knowledge of the company and wide supply network is important for the insights' reliability of the proposed operations research approaches. Therefore, it is vitally important to assemble an expert panel that has good knowledge and understanding of firms' operations and the wide supply network contexts. While ensuring effective communication in the decision making process and fully utilization of knowledge from the expert panel, such an evaluation could provide valuable insights into the exposed risks of available SusPSSs, leading to key strategic recommendations for achieving sustainability objectives.
Despite the contribution outlined above, the present approach has its own limitations with potential directions for future research. For example, all the network-oriented risk sources and their associated items have to be accounted for and accumulated in the evaluation. In addition, users have to rate different risk items using linguistic expressions. The functionality of the methodology also depends highly on the knowledge, expertise and communication skills of the users. One future research option could be to consider data-driven techniques that use available transactional data from firms (Wang et al. 2015b). In addition, the SusPSS approach, with its focus on end-of-pipe attitudes and dematerialisation strategies driven by a sustainability agenda, is geared towards the realisation of functional economies. This agenda offers potential for the realisation of functional economies that fosters energy efficiencies and minimal waste as well as nature and environmentally friendly policies. There is therefore a need for studies that leverage a holistic view of production to shed light on sustainable values that are progressively created or destroyed and potential variations in the sources, levels and perceptions of risks during this process of resource acquisition. Such studies may enhance strategic decision-making of firms for sustainability policies. Potential future research could also examine the formation and evolution of partnerships for SusPSS based on themes such as value systems, co-creation and leadership.