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 (SusProNet), 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;. 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.
By definition, a SusPSS, also known as a 'sustainable service' (Heiskanen and Jalas, 2003;Halme et al., 2004) or an 'eco-efficient product-service system' (Manzini and Vezzoli, 2003;Ceschin, 2013), is a Product-Service System (PSS), i.e., an innovation-driven approach to production that shifts business tenets from delivering physical products only to delivering integrated product-service offerings that meet the needs of clients and customers (Durugbo and Riedel, 2013;Rondini et al., 2017). In contrast to other forms of PSSs such as technical PSSs or industrial PSSs (Aurich et al., 2006), a SusPSS focuses on reorienting current unsustainable trends in production and consumption practices (Manzini and Vezzoli, 2003). The SusPSS approach pursues this target through outcomes in the form stakeholder value, which researchers often characterise as benefits for consumer/citizen, company and government groups in terms of the Triple Bottom Line, i.e., the environmental, economic and social dimensions of sustainability (Vogtländer et al., 2002;Mont, 2002;Maxwell and Van der Vorst, 2003;Tukker, 2004;Vezzoli et al., 2015). In the SusPSS approach, firms are encouraged to form partnerships with stakeholders to strategize the provision of results or functions and creatively generate ideas that reduce the environmental impact of companies by factors between 4 and 20 (Roy, 2000;Schmidt-Bleek, 2008). Scholars and practitioners widely acknowledged that risks associated with SusPSSs such as service offerings, service costs, eco-efficiency potentials, social factors, interaction strategies, capabilities and partnerships play a key role in how sustainability policies are adopted and how partners are chosen to deliver the sustainable product-service mix (see, for instance, Krucken and Meroni, 2006;Durugbo and Riedel, 2013;Choi et al., 2016). Nevertheless, there has been little consideration in the SusPSS literature on approaches to manage risks of an industrial firm in a SusPSS. 3 This article proposes a real-case based operations research approach for evaluating network-oriented risk of industrial firms in SusPSSs. Risk is used in this context as "the chance, in quantitative terms, of a defined hazard occurring" (the Royal Society, 1992). Risks are highly random in nature and are caused by different forms of uncertainty existing 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 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. 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).
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 postproduction 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 endof-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.

Conceptual and fundamental
Organisations collaborating to leverage levels of creativity 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), Maxwell et al., (2006) Proposes a method to aid in evaluating sustainable criteria during product and service development

Conceptual and applied
The level of control by companies over main life cycle stages Halme et al., (2004) Proposes a set of indicators to evaluate sustainability of services directed to households

Conceptual and fundamental
Institutional arrangements of delivering services directed to households Mont (2004) Uses a SusPSS framework of products, service, infrastructure networks, business viability, customer satisfaction and environmental trustworthiness in case studies. Combines interviews, survey and literature sources to develop and assess the SusPSS evaluation framework

Empirical and applied
Business-to-consumer relationships to manage unsustainable consumption patterns Tukker (2004) Evaluates market and sustainability potentials for product-service offerings using value creation and sustainability models

Conceptual and fundamental
Improve customer loyalty through relationship development with clients Briceno and Stagl (2006) Evaluates social and humanistic factors within SusPSS

Conceptual and applied
Relationship building for new shared norms, attitudes and social frameworks that support transitions to more sustainable consumption patterns Krucken and Meroni (2006)

Conceptual and applied
Communicative structure for business management and stakeholder empowerment Evans et al., (2007) Evaluates environmental, economic and social performance of solution-oriented partnerships

Empirical and applied
Fundamental change in the relationship between stakeholders 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 suppliercustomer 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 Proposes a concept of sustainable productservice efficiency to explore the relationship between product-service value and the sustainability impact

Conceptual and fundamental
Socio-economic issues for the evaluation of SusPSSs Mylan (2015) Applies the sociology of consumption and practice theory to improve the understanding of processes that influence the diffusion and uptake of SPSS

Conceptual and fundamental
Demand-side view of the diffusion of SusPSS innovations Proposes a hybrid model integrating Discrete Event Simulation with Agent-Based Modelling

Conceptual and applied
Dynamic features of customer behaviours, process requirements and sustainability assessment *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.
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 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; (a) to retain competitive advantage; (b) to reduce costs of operations and (c) to improve collaboration and coordination among suppliers and between a supplier and a customer (Cooper et   1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57  58  59  60  61  62  63  64  65 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 Pulraj, 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. Impact of on-time delivery C20 Delay of critical material delivery

C31
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;C32 Impact of pre-process output on post-process performance 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.

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

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 and Chan, 2013;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 . 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;Chen 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). 1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57  58  59  60  61  62  63  64  65 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 and for the probability and severity ratings of risk item i rated by expert k, respectively.   where ̃ 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:

Steps for risk analysis
where 1 ≤ ≤ for n risk items, and 1 ≤ ≤ 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 = 1 ln ⁄ and = 1, 2, ⋯ , . 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 , we stipulated: Here, is the defuzzified risk value of ̃ using the Centre of Area method given as: The weight of entropy of i th criterion can then be defined as: in which 0 ≤ ≤ 1, and ∑ = 1 =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: where ̃=̃× . Then we calculate the distances from negative and positive ideal solutions. Let Aand 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: where ̃+ and ̃−are the fuzzy numbers with the largest and smallest generalized means, respectively. For each column i, the largest generalized mean of ̃+ and the smallest generalized mean of ̃− are derived, respectively. Consequently, the FPIS (A + ) and the FNIS ( − ) are obtained. Then the distances (d + and d -) of each alternative SusPSS from A + and − can be calculated by the area compensation method as: [( 1 − 1 ) 2 + ( 2 − 2 ) 2 + ( 3 − 3 ) 2 ] Finally, the alternative SusPSSs can be ranked by their relative closeness indexes, which are determined by a combination of the difference distances d + and das 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 13 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 equations (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 (A2), the risk level was low for items in the supply risk source, i.e., complexity of critical material (C25) and delivery frequency of critical material (C28), but high for items in the control risk source, i.e., information accuracy (C41) and organisational change through delivery of new services (C44).

C3 Manufacturing (and process) risk
C31 Impact of pre-process change on post-process 6.0 6.1 4.1 7.3 0.027 C32 Impact of pre-process output on post-process performance 6.2 5.3 5.3 8.9 0.043 C33 Degree of a product decomposable to simpler components 6.1 5.9 8.9 6.9 0.025 C34 Degree of product modularization 4.5 5.9 8.6 6.5 0.042 C35 Redesign frequency 6.6 7.3 5.  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 (A2) 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 1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57  58  59  60  61  62  63 64 65 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 (A1) and product-life extension services (A3), although one proposition may be more exposed to specific networkoriented 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 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 1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57  58  59  60  61  62  63  64  65 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 servicesoriented 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.