Selection method of sustainable product-service system scenarios to support decision-making during early design stages

ABSTRACT

During the early design stages of product-service systems (PSS), particular attention must be given to sustainability considerations to ensure the achievement of a sustainable PSS. However, there is a lack of decision-making support within existing design methods to help designers generating PSS solutions with reduced sustainability impact from the early design stages. To fill this gap, this paper elaborates on a decision support method coupling multi-criteria analysis and sustainability assessment, and considering PSS peculiarities. The aim of the method is to select PSS scenarios with the highest sustainability potential. Sustainability is assessed from multiple perspectives involving both the PSS actors’ points of view and sustainability pillars. The selection of the PSS design scenarios is supported by a multi-criteria analysis enabled by the ELECTRE I method. Furthermore, the paper elaborates on a case study around a PSS-oriented water-saving sanitary equipment intended for healthcare facilities. This provided some evidences of the applicability and relevance of the proposed method, which contributes ultimately to fostering sustainability integration within PSS design.

KEYWORDS:

• Product-Service System (PSS)
• sustainability assessment
• multi-criteria analysis
• decision-making
• early design

1 Introduction

Companies are increasingly compelled by regulations and incentivised by public institutions to integrate sustainability in their strategies. As a result, they are seeking to develop both profitable and eco-friendly offers without degrading the fulfilment of consumers’ needs. This trend motivates the transition from a traditional selling model (e.g. sales of the product) to the selling of functionality or result provided by an integrated solution of products and services, called Product-Service System (PSS) (Baines et al. 2007). It is assumed that PSS is likely to lower the environmental impact of the manufacturing industry (Goedkoop et al. 1999; Mont 2004). However, this assumption needs to be checked for each specific PSS offer during the design process (Agrawal et al. 2012). This is in line with the idea that the design process is one of the most impacting factors of developing systems with high sustainability potential (Morelli 2006; Pigosso and McAloone 2016).

Although many methods have been identified in PSS design literature, PSS engineering is still not well established in the literature (Cavalieri and Pezzotta 2012). In particular, the sustainability of a PSS is rarely taken into account during the early design steps (Vasantha et al. 2012). The engineering designers need then some support in order to integrate the sustainability as a decision criterion (as cost, quality and delay). Looking at existing assessment methods (e.g. LCA, TCO, LCC, etc.), it can be inferred that they are mostly dedicated to a single sustainable pillar (Doualle et al. 2015). Well-adapted methods and tools are thus required to address the PSS design peculiarities (Vasantha et al. 2012). Moreover, the sustainability concept needs a systemic view to better take into account the relationship between its various pillars and to consider the points of view of the different actors involved in the assessed system.

This paper elaborates on a decision support method coupling multi-criteria analysis and sustainability assessment, and considering PSS peculiarities. The aim of the method is to select PSS scenarios with the highest sustainability potential. Sustainability is assessed from multiple perspectives involving both the PSS actors and sustainability pillars. The selection of the PSS design scenarios is supported by a multi-criteria analysis enabled by the ELECTRE I method. The proposed method aims to support the decision-making process during the engineering process of a system, and more specifically PSS. The method provides valuable information and guidance to system engineers towards sustainable PSSs. The paper develops a case study concerning a PSS-oriented water-saving sanitary equipment, intended for health-care facilities.

The remainder of the paper is organised as follows: Section 2 investigates the scientific literature in order to identify the requirements for building the selection method. Section 3 reports on the formalisation of the method based on the identified requirements. The case study is presented and then discussed in Section 4. Section 5 provides a general discussion of the method and the research perspectives. Concluding remarks are given in Section 6.

2 Sustainability at PSS early design

This aim of this section is twofold; first to provide an overview of the research context including PSS early-design and sustainability assessment (§2.1), second to outline the main principles of the selection method (§2.2).

2.1 Overview of PSS sustainability assessment

A sustainable PSS can be defined as ‘an offer model providing an integrated mix of products and services that are together able to fulfil a particular customer demand, based on innovative interactions among the stakeholders of the value production system, where the economic and competitive interest of the providers continuously seeks environmentally and socio-ethically beneficial new solutions’ (Vezzoli et al. 2015). Accordingly, the design of PSS needs to change the way of thinking from product- to system-oriented (Manzini and Vezzoli 2003; Tran and Park 2015). Sustainability assessment should be then tailored to such a systemic view, which includes products, services, the relationship between them and actors involved to provide and/or receive them. However, there is a lack of methods and tools available to assess PSS sustainability, in particular during the early-design stages (Doualle et al. 2015; Halme et al. 2006; Omann 2007; Vezzoli et al. 2014).

Previous research works showed that PSS assessment is more established according to environmental and economic dimensions than the social one (Doualle et al. 2015; Halme, Jasch, and Scharp 2004). This means that particular attention needs to be paid to social dimension in order to balance the sustainability impact among all three dimensions. Furthermore, the intrinsic multidimensionality and inherent complexity of sustainability require adapting existing indicators (or developing new ones) in order to ensure a holistic and reliable assessment. Holistic sustainability indicators improve the readability of the assessment results for both PSS stakeholders and non-expert users by providing a unique score. This, however, adds some biases that are judgement and shortcuts, which may lead to wrong conclusions and poor decisions. Nevertheless, it is important to consider sustainability as a whole in order to avoid damaging one dimension while limiting harm made to another one (Abramovici et al. 2014; Chou, Chen, and Conley 2014; Partidário, Lambert, and Evans 2007; Vezzoli et al. 2014, 2017).

Among the few significant research works integrating sustainability into PSS design, one can state (Geum and Park 2011; Sousa-Zomer and Miguel 2017; Vezzoli et al. 2014; Tukker 2003; Lindahl, Sundin, and Sakao 2014; Doualle 2018). Geum and Park (2011) proposed a design method based on the service blueprint in order to ensure PSS sustainability. The authors assume that sustainability would be achieved if the designers follow the guidelines (Vezzoli et al. 2014) provided by the method. However, no real assessment is proposed by this method. The MEPSS project resulted in a set of guidelines and tools to integrate the sustainability in PSS design (Vezzoli et al. 2014). A sustainability design-oriented toolkit has been developed based on a set of sustainability dimensions and a qualitative assessment of the designed system. The assessment provides a single perspective of the system sustainability, which does not consider actors’ viewpoints, nor the impact of the selected PSS functions. Trevisan and Brissaud (2016) address the eco-design of PSS (Trevisan and Brissaud 2016). The sustainability assessment is part of a larger framework for PSS design. However, this framework specifically focuses on detailed-design; thus, the data and detailed life-cycle data required for sustainability evaluation are not well adapted to early-design. Additionally, some weak points still remain: the approaches maintain cost-analysis as the key tool for environmental assessment and the various points of view of actors are not explicitly considered. Complementary, Sousa-Zomer and Miguel (2017) introduced a QFD-based approach to support the allocation of PSS stakeholders’ requirements to sustainability pillars (Sousa-Zomer and Miguel 2017). This approach supports also the conceptual design phase through linking the requirements to intended PSS functions. However, the assessment of sustainability pillars remains quite general and actors’ viewpoints are not considered.

A set of key challenges could be derived from the above analysis, which adds to sustainability assessment complexity in particular during early-design stages. One can state data scarcity, life-cycle dimension, and assessment criteria heterogeneity. To optimise the integration of the sustainability into the radical changes, sustainability considerations have to be introduced as soon as possible in the design process to enable a radical change rather than local optimisation (Maussang, Zwolinski, and Brissaud 2009). Proper assessment methods are thus needed during PSS early-design, which take into consideration sustainability intrinsic complexity and multidimensionality, and PSS peculiarities, namely, function-based assessment and multi-actor perspective.

2.2 Key requirements for sustainability assessment at early-design stage

The previous subsection emphasised the importance to implement sustainability assessment at an early-design step and specific challenges induced by the complexity of PSS context. In the following paragraphs a more detailed state of the art is developed focusing on key constraints and requirements for such early-design assessment:

• A function-based assessment is adopted, to address the lack of available information identified previously (§2.2.1).

• A multi-actor perspective needs to be considered to cope with the multi-stakeholder implication of PSS context (§2.2.2).

• The concept of design scenario is used for the assessment, to represent and analyse alternative design solutions resulting from combining PSS components (§2.2.3).

• PSS design scenarios should be evaluated using a multi-perspective indicator system, consistently with the multi-dimensionality of sustainability (§2.2.4).

• The selection of the PSS design scenarios is supported by multi-criteria analysis, which helps cope with and discuss various stakeholder points of views (§2.2.5).

2.2.1 Function-based assessment

The early-design context is characterised by a lack of information about the system designed which induces specific difficulties for assessing sustainability. At this stage of the design process, the functions of the system are the key pieces of information available: a vision of the whole system is possible as soon as all the functions provided by the system are defined (Maussang et al. 2007; Van Ostaeyen et al. 2013). Consequently, in order to integrate sustainability assessment during the early-design, a function-oriented assessment of the system under design is required (Rondini et al. 2016). In this perspective, the assessment approach should integrate functional models covering both the service and the product dimensions. Among the recent models addressing more or less these requirements, one could mention the Function Hierarchy Modelling (Van Ostaeyen et al. 2013) and the Product Service Concept Tree (Pezzotta et al. 2018). Their common aim is to represent PSS solutions in a structured approach and to assess them starting from the customers’ needs.

2.2.2 Considering multi-actor perspective

A sustainable system could be seen as a social construction, based on ‘attraction forces’, which catalyse the participation of several partners (Morelli 2006). Therefore, the main actors involved in the PSS offer need to be identified from the very beginning of the design process. This early involvement makes possible analysing the sustainability impacts of the PSS from the various actors’ perspectives, thus leading to a co-production of the envisioned PSS solution. Moreover, stakeholder involvement helps catalysing mutual awareness among them and, consequently, improves their potential interactions. Such interactions have potentially a great impact on the performance and sustainability of the whole solution (Vezzoli et al. 2015). In this sense, stakeholder involvement helps increase their awareness of the sustainability and ease the collaboration towards more sustainable design scenarios (Peruzzini, Marilungo, and Germani 2015; Vasantha, Roy, and Corney 2016).

2.2.3 Notion of scenario at early-design

In the context of PSS, each design alternative involves a complex system including material parts, service activities as well as associated value networks. This complexity of the PSS value system induces various design alternatives, which should be analysed, filtered, etc., during the design process. In line with the PSS scientific literature (Medini and Boucher 2019; Boucher, Medini, and Fill 2016; Lelah et al. 2014) we propose to represent and formalise these alternatives using the concept of ‘design scenario’. The notion of scenario improves the quality of decision-making by generating a shared representation of the system and (Chemack, 2004). Moreover, the PSS literature highlights the need to design the actors and their interactions for PSS delivery (Morlock, Dorka, and Meier 2014). A scenario describes a combination of available information about given customer needs, functions, solutions, and network of actors. As such, a given scenario evolves throughout the PSS design process; it goes from a minor set of information about needs up to a complete picture of needs, functions, solutions and actors’ network. During the design process, multiple scenarios are developed by the designers, which represent the prospective alternatives. Thus, the improvement of the PSS sustainability potential can be based on the evaluation then selection of the scenarios, according to their respective sustainability impacts as proposed in section 3.

2.2.4 The need for multi-dimensional assessment

As emphasised in section 2.1, a holistic sustainability assessment requires an adaptation of existing indicators or development of new ones. This is a challenging perspective due to the difficulty to propose a generic and re-usable approach covering the very large scope of sustainability. Based on the scientific literature on sustainable PSS, we propose below a contribution with a classification of a large set of indicators into categories corresponding to complementary dimensions of the assessment.

MEPSS project resulted in 18 sustainability dimensions, which covered the three pillars of the sustainability dimensions (cf. Table 1), which are assumed to include the main drivers of PSS sustainability (Vezzoli et al. 2014). These dimensions will be used as a reference for the sustainability assessment for the proposed method. In order to enable sustainability assessment, the above dimensions need to be operationalised through measurable (qualitative or quantitative) indicators. Due to the different sustainability aspects covered by the dimensions, a significant number of indicators is required. A predefined list of 80 indicators is used and has been established through a literature analysis about sustainability indicators from different scientific domains: Management, Environment, Corporate Social Responsibility (Chou, Chen, and Conley 2014; GRI 2013; ISO 2010; Leipziger 2001; Kausek 2007; Vezzoli et al. 2014). These indicators have been selected according to a two-level process starting with selection and ending with validation. The initial selection was performed by the authors according to several criteria and adapted in order to ease their evaluation. The two main criteria are as follows: (1) relevance to the decision-making process during the design stage, (2) comprehensiveness in terms of ability to provide an estimate of the indicator value. The final validation is based on experts’ judgement allowing to filter the initially selected list. Experts have backgrounds in different domains, in particular, sustainability, design, and industrial engineering. The 80 indicators are classified into 18 classes corresponding to the sustainability dimensions. The set of indicators is presented in the APPENDIX and will be integrated into the approach developed in section 3.

Table 1. Sustainability dimensions (Vezzoli et al. 2014).

Sustainable pillars	Sustainability dimensions
Social	Improve employment/working conditions
	Improve equity and justice in relation with stakeholders
	Enable responsible/sustainable consumption
	Favour/integrate the weak and marginalised
	Improve social cohesion
	Empower/valorise local resources
Economic	Market position and competitiveness
	Profitability/added value for businesses
	Added value for clients
	Long-term business development
	Partnership/cooperation
	Macro-economic effect
Environment	System life optimisation
	Transportation/distribution reduction
	Resource reduction
	Waste minimisation/valorisation
	Conservation/biocompatibility
	Toxicity reduction

2.2.5 Needs for multi-criteria analysis

A sustainability holistic assessment relies on heterogeneous and numerous indicators, which are likely to confuse decision-makers and impede the usability of the method. Moreover, the actors involved in the PSS engineering and delivery have usually different expectations concerning the PSS offer. This leads to several contradictory points of view, which generates a distinct evaluation of PSS design scenarios. In order to get to feasible scenario(s) is then needed to identify trade-off situations, meeting as many of the expectations as possible. To overcome these issues, multi-criteria analysis is proposed. In order to identify the most suitable method, a selection procedure of multi-criteria analysis method, developed by Benoit and Rousseaux (2003), was applied with the following characteristics of the decision-making problem:

• The objective is to select scenarios and not ranking neither classifying them.

• The decision criteria are real variables.

• Criteria are weighted.

• There is no hierarchical structure of the criteria.

According to this procedure, the ELECTRE I method has been identified as one of the most suitable ones according to the above characteristics (Vincke 1989). The basic principle of ELECTRE relies on checking over-ranking hypothesis (allowing to rank the scenarios according to the given criteria) through concordance and discordance tests.

3 Method for the selection of the most sustainable PSS scenarios

This section presents our proposition for a PSS assessment method applicable to the early PSS design phases, which intends to integrate the various requirements underlined in section 2.2. This method is evaluating PSS scenarios (combination of PSS functions and actors) through a multi-perspective sustainability indicator system and multi-criteria analysis. Ultimately, the aim is to eliminate non-acceptable scenarios in terms of sustainability issues. The sustainability assessment is enabled through a four-step process (Figure 1), namely initial inputs, scenario definition, scenario assessment, multi-criteria analysis. These steps will be detailed in the next sections.

Figure 1. the selection method of sustainable PSS scenarios.

3.1 Step 0: initial inputs

The pre-requisites of the selection method are customer needs, sustainability dimensions, and the matrix of the potential impact of the former on the latter.

The early-design phase is based on the identification of the customers’ needs, which define the specific customer expectations from the PSS solution. These customers’ needs are one of the main inputs of the method. In the following, customer needs are denoted by B (Equation 1).

(1) $B=\left\{B_l\right\};l\in \left\{1,2,\dots ,b\right\}$(1)

Where

• B is the set of the customers’ needs.

• B_l is a grammatical expression composed of a verb or a verbal group characterizing the expectation of the customers, and of complements representing the element of the external environment concerned.

• b is the number of customers’ needs.

Based on the sustainability dimensions identified in the state-of-the-art, the set of the 18 dimensions might need to be prioritised so as to make the design process converge smoothly into feasible systems. From a decision-making point of view, having so many and equally relevant sustainability dimensions is likely to overwhelm the designers given the dimensions’ complexity and heterogeneity. Consequently, the idea of prioritising those dimensions helps focusing on the most impacted dimensions by the working PSS design project. The sub-set of sustainability dimension is selected through an analysis by the users, guided by their own expertise and the following two main criteria: first, the sub-set should be fairly distributed among the three sustainability pillars, second, the remained dimensions are identified as having the most potential impact $u_{l,k}$. It is then necessary to provide a sub-set of those 18 sustainability dimensions, denoted by E. E represents the sustainability dimensions, which are selected as pertinent for the PSS under design (Equation 2).

(2) $E=\left\{E_k\right\};k\in \left\{1,2,\dots ,e\right\}$(2)

Where

• E is a set of relevant sustainable dimensions.

• E_k is the k sustainability dimension.

• e is the number of sustainability dimensions of the set E.

The last required initial input of the method is the matrix linking the customers’ needs to the sustainability dimensions (Equation 3). Each of the matrix coefficients represents the relative potential impact of a given customer need on a given sustainability dimension. Due to the uncertainty underlying early-design stages, only estimates of the impacts can be provided. In practice, the matrix is filled in through answering the following question: To which extent is the need l likely to impact on the sustainability dimension k? The answer is a number belonging to ${\mathbb{R}}_{+}$. Depending on the PSS design context, the users of the method are free to define their own scale of the coefficients and even to use other methods for estimating them.

(3) $U=(u_{l,k}{)}_{l=\mathrm{1..}b;k=\mathrm{1..}e};u_{l,k}\in {\mathbb{R}}_{+}$(3)

Where

• U is the matrix of dimension (b * e) with variables taking positive real numbers.

• $u_{l,k}$ denotes the potential impact of the customer needs B_l on the sustainable dimension E_k.

3.2 Step 1: scenario definition

When the customers’ needs are identified and their potential impacts are calculated, the PSS design team needs to work on the answers to those customers through the definition of the PSS scenarios which consists of a set of PSS functions and a set of organisational actors involved in these functions. The definition of the scenarios provides valuable input for the allocation of the potential sustainability impacts among the actors. The scenario definition steps are the following:

• Function identification

• Computation of the potential impact of the functions

• Actors’ involvement identification

• Computation of the potential impact of the actors

• Scenario establishment

3.2.1 Function identification

Customer needs are translated into functions which are seen as the means to fulfil these needs (Equation 4). This task is performed by the PSS designers who identify for each customer need a set of functions contributing partially or totally to their fulfilment.

A set of functions is written as:

(4) $F=\{F_j{\}}_{j=1\dots f}$(4)

• $F$ is a set of functions proposed to fulfil all the customer needs in a relevant way.

• $F_j$ represents a function which meets one or many customers’ needs, partially or totally. ‘A function is a grammatical expression composed of a verb or a verbal group characterizing the action, and of complements representing the element of the external environment concerned’ (AFNOR).

• f is the overall number of functions necessary for the entire PSS.

It is then possible to create a functions-needs contribution matrix D (Equation 5): a binary matrix, linking the needs with the functions, depending on the contribution or not of each function to fulfil a given need.

(5) $D=(d_{l,j}{)}_{l=\mathrm{1..}b;j=1\dots f};\{d_{l,j}\in \mathbb{N}|d_{l,j}=\left\{0,1\right\}\}$(5)

Where

• D is a matrix of dimension (b * f).

• $d_{l,j}$ represents the contribution of the function j to fulfil the customers’ need l. If the function fulfils the needs, at least partially, then $d_{l,j}=1$ else $d_{l,j}=0$.

Afterwards, D is normalised, leading to the matrix D’. The values $d{ }_{l,j}^{\prime }$ of D’ are normalisation values of $d_{l,j}$ such that the sum of the contributions of all the functions for a specific need is equal to 1 (Equation 6).

(6) $d{\prime }_{l,j}=\frac{d_{l,j}}{{\sum }_{j=1}^f{d}_{l,j}}$(6)

This normalisation is necessary for the next steps, notably to allocate the potential impact$u_{l,k}$ of a given need $B_l$ on a given dimension $E_k$ among the set of PSS functions $F_j$ contributing to $B_l$.

3.2.2 Computation of the potential impact of the functions

After establishing the link between customer needs and the set of PSS functions, the potential impact of these functions on sustainability dimensions could be defined as follows (Equation 7):

(7) $C=D{\prime }^T\ast U={\left(c_{j,k}\right)}_{j=\mathrm{1..}f,k=\mathrm{1..}e};\left\{c_{j,k}\in {\mathbb{R}}_{+}\right\}$(7)

Where

• $C$ is a matrix of dimension (f * e).

• The coefficient $c_{j,k}$ denotes the potential impact of the function j on the sustainability dimension k. The value of $c_{j,k}arebetween0and{u}_{j,k}$ where uj,k is the potential impact of customer need j on sustainability dimension k.

3.2.3 Actors’ involvement identification

One of the aims of the selection method is to provide estimates of the actors’ respective contributions to the PSS impact on sustainability. The underlying assumption of the method is that such a contribution is correlated with the involvement of the actors in the value network. Actors are considered through their roles with regards to the PSS functions, as formalised below (Equation 8). This information is useful to establish two elements required for the method:

1. To quantify the involvement of each actor according to the sustainability dimensions identified.

2. To assess the specific point of view of each actor, according to an “impact coefficient” defined later.

First, the set of actors is defined as:

(8) $A={\left\{A_i\right\}}_{i=\mathrm{1..}a}$(8)

Where:

• A is the set of actors involved in the value network of the PSS.

• $A_i$ represents the actor i.

• a is the number of actors involved.

Each actor can be involved in the value network through one or several roles regarding the functions of the PSS to be designed:

1. When the actor is involved in the production or implementation process of a function, it plays the role of a function provider.

The identification of the roles results in the matrix G, which expresses the existence of roles linking the actors and the functions (Equation 9).

G can be expressed as:

(9) $G={\left({\beta }_{i,j}\right)}_{i=\mathrm{1..}a,j=\mathrm{1..}f};\{{\beta }_{i,j}\in \mathbb{N}|{\beta }_{i,j}=\left\{0;1\right\}\}$(9)

Where:

• G is a matrix of dimension (a * f).

• ${\beta }_{i,j}$ is a binary variable. If there is at least one role, linking the actor i with the function j then ${\beta }_{i,j}=1$, else ${\beta }_{i,j}=0$.

3.2.4 Computation of the potential impact of the actors

The actors, through their various roles, impact the sustainability in several ways. It is then necessary to identify the potential impact of each actor in order to identify the most impacted sustainability dimensions for each of them (Equation 10).

This potential impact is expressed by a matrix H:

(10) $H=G\ast C={\left(h_k^i\right)}_{i=\mathrm{1..}a;k=\mathrm{1..}e};\left\{h_k^i\in {\mathbb{R}}_{+}\right\}$(10)

Where

• H is a matrix of dimension (a * e).

• The entry $h_k^i$ denotes the potential impact of the actor i on the sustainability dimension k. The range of the value of $h_k^i\mathrm{a}\mathrm{r}\mathrm{e}\mathrm{b}\mathrm{e}\mathrm{t}\mathrm{w}\mathrm{e}\mathrm{e}\mathrm{n}0\mathrm{a}\mathrm{n}\mathrm{d}{\mathrm{C}}_{\mathrm{j},\mathrm{k}}$.

3.2.5 Scenario establishment

A design scenario is an organisational and functional answer to meet the customers’ needs. During the design process, several scenarios are explored.

A scenario can be defined as:

(11) $S=⟨F,D,A,G,H,C⟩$(11)

Where

• F is a set of functions designed to fulfil the customers’ needs B in a relevant way.

• D is the matrix of the links between the needs B and the functions F.

• A is the set of actors involved in the value network of the PSS.

• G is the matrix of the involvement of the actors A in the functions F.

• H is the matrix of the potential impacts of the actors A on sustainability dimensions E’.

• C is the matrix of the potential impacts of the functions F on sustainability dimensions E’

3.3 Step 2: scenario assessment

This step is concerned with the assessment of the previously defined scenarios from a sustainability perspective using qualitative indicators. The aim of the sustainability assessment of the scenario is to provide to each actor its impact on sustainability. The sustainability assessment relies on the 80 pre-selected indicators from the literature (cf. section 2.2).

3.3.1 Establishment of sustainability indicator sub-sets

For a given actor, only the indicators belonging to the dimensions potentially impacted by this actor are used for the assessment (Equation 12). These dimensions are identified through the matrix H (representing the potential impact of the PSS actors on sustainability dimensions). Consequently, the indicators sub-set for a given actor are as follows:

(12) $I_i={\left\{I_{kd}^i\right\}}_{k=\mathrm{1..}e;d=\mathrm{1..}{c}_k}|h_k^i\ne 0$(12)

Where

• I_i is the set of indicators evaluated by the actor i.

• $I_{kd}^i$ is the d indicator of the k dimension.

• e is the number of sustainability dimensions.

• $c_k$ is the number of indicators of the sustainability dimension k.

3.3.2 Indicators estimated by the actors

The indicator values are estimated based on the auto-evaluation of the actors’ potential impact on sustainability. Due to the heterogeneity of sustainability aspects, two steps-scale of indicator evaluation are used: a 4-step scale {low, average, high, very high}, or a 2-step scale {Yes, No}. In order to homogenise the whole indicators sub-set including both categories, the estimates are quantified using the following scales:

• A 4-step scale {0,1,2,3} where “0” designates a high impact on sustainability and “3” denotes a low impact, and “1” and “2” are intermediary impact levels.

• A 2-step scale {0, 3} where “0” designates high impact on sustainability and “3” denotes a low impact.

The evaluation results in the $J^i$ vector comprised the quantified indicator values (scores) $v{I}_{kd}^i$(Equation 13).

(13) $J^i={\left(v{I}_{kd}^i\right)}_{k=\mathrm{1..}e;d=\mathrm{1..}{c}_k}$(13)

Where

• $v{I}_{kd}^i$ is the sustainability score of the actor i according to the indicator d of the dimension k, $v{I}_{kd}^i\in \left\{0,1,2,3\right\}$.

• e is the number of sustainability dimensions.

• $c_k$ is the number of indicators associated with the sustainability dimension k.

For each actor i, indicator values $v{I}_{kd}^i$ are aggregated using simple average into scores ($L_k^i$) measuring the sustainability according to each of the dimensions k (Equation 14).

(14) $L_k^i=\frac{{\sum }_{d=1}^{c_k}v{I}_{kd}^i}{c_k},k\in \left\{\mathrm{1..}e\right\}$(14)

Where

• e is the number of sustainability dimensions.

• $v{I}_{kd}^i$ is the sustainability score of the actor i according to the indicator d of the dimension k, $v{I}_{kd}^i\in \left\{0,1,2,3\right\}$.

• $c_k$ is the number of indicators of the sustainability dimension k.

At this level, potential impacts $h_k^i$ and sustainability score $L_k^i$ are calculated for each of the PSS actors. This provides valuable insights into the sustainability of a given PSS scenario. The above procedure is replicated for each identified PSS scenario.

3.4 Step 3: multi-criteria analysis

The identification of PSS functions and the actors involved usually leads to many alternative design scenarios. In order for the PSS designers to converge to feasible scenarios, they need to remove the so-called non-acceptable scenarios. The decision-making process on scenarios elimination is quite complex because of the number of criteria to be considered.

Multi-criteria analysis and, more specifically, an ELECTRE I method are used by the selection method to support such decision-making process towards identifying acceptable and non-acceptable scenarios. The ELECTRE I method is based on the over-ranking scenario by pair comparison in order to identify two classes: the class of the non-acceptable scenarios, which are over-ranked by at least one scenario and the class of the acceptable scenarios, which are not over-ranked by any other scenario (Vincke). The most sustainable scenario is included within the acceptable scenarios class. The non-acceptable scenarios will not be kept, which corresponds to the selection objective. Consistently with this principle, in the proposed method, a given scenario A over-ranks scenario B if A is at least as sustainable (or not worst) as scenario B.

The sustainability of a given scenario, from the point of view of a given actor, is assessed through the normalised potential impacts of the actor i (global weight called${\rho }_{i,k}$) $h_k^i,k\in \left\{\mathrm{1..}e\right\}$ and sustainability scores of the actor i $\left(\mathrm{E}\mathrm{q}.15\right)$. These two parameters, namely, potential impact and sustainability scores are used as inputs for the concordance and discordance indexes of the ELECTRE I method.

(15) ${\rho }_{i,k}=\frac{{\sum }_{z=1}^n{h}_{i,k,z}}{{\sum }_{k=1}^e{\rho }_{i,k}}$(15)

Where

• ${\rho }_{i,k}$ is the global weight of the sustainable dimension k for the actor i.

• $h_{i,k,n}$ is the scenario z potential impact of the actors i on the sustainable dimension k.

• $n$ is the number of scenarios developed.

• $e$ is the number of sustainability dimensions.For better understanding, some definitions are needed:

• $E^{+}\left(S_b,S_c\right)=\left\{k\in \left\{1,2,\dots ,e\right\}|L_k^b>L_k^c\right\}$: set of sustainable dimensions where the scenario b is more sustainable than the scenario c.

• $E^{=}\left(S_b,S_c\right)=\left\{k\in \left\{1,2,\dots ,e\right\}|L_k^b=L_k^c\right\}$: set of sustainable dimensions where the scenario b is as sustainable as the scenario c.

• $E^{-}\left(S_b,S_c\right)=\left\{k\in \left\{1,2,\dots ,e\right\}|L_k^b<L_k^c\right\}$: set of sustainable dimensions where the scenario c is more sustainable than the scenario b.

• ${\rho }^{+}\left(S_b,S_c\right)=\sum {\rho }_k,k\in E^{+}\left(s_i,s_k\right)$: sum of the weights of the sustainable dimensions included in $E^{+}$.

• ${\rho }^{=}\left(S_b,S_c\right)=\sum {\rho }_k,k\in E^{=}\left(s_i,s_k\right)$: sum of the weights of the sustainable dimensions included in $E^{=}$.

• ${\rho }^{-}\left(S_b,S_c\right)=\sum {\rho }_k,k\in E^{-}\left(s_i,s_k\right)$: sum of the weights of the sustainable dimensions included in $E^{-}$.

• $\rho ={\rho }^{+}\left(S_b,S_c\right)+{\rho }^{=}\left(S_b,S_c\right)+{\rho }^{-}\left(S_b,S_c\right)$The over-ranking relation $\phi$ is defined as:

• $S_b\phi S_c\iff f\left(x\right)=\left\{\begin{array}{c}{C}_{bc}\ge c\left(concordancetest\right)\\ D_{bc}\le d\left(non-discordancetest\right)\end{array}\right.$

• $C_{bc}=\frac{{\rho }^{+}\left(S_b,S_c\right)+{\rho }^{=}\left(S_b,S_c\right)}{\rho }$

• $D_{bc}=\left\{\begin{array}{c}0,si{E}^{-}\left(S_b,S_c\right)=\mathrm{\varnothing }\\ \frac{1}{3}\ast Max\left\{L_k^b-L_k^c\right\};k\in E^{-}\left(S_b,S_c\right)\end{array}\right.$

Where

• 3 is the amplitude of the scale associated with the sustainable dimension.

• c is the concordance threshold. The value of c is fixed by the designers, but it is recommended to fix it between 0.5 and 1. The higher c is, the more reliable the over-ranking hypothesis is.

• d is the discordance threshold. The value of d is fixed by the designers but it is recommended to fix it between 0 and 0.5. The lower d the more reliable the over-ranking hypothesis.

It is important to mention that in our investigation of ELECTRE I method, no specific method was identified for determining c and d values.

4 Case study

The following case study was developed to provide a proof of concept of the selection method. It was chosen as it involves a PSS design project putting sustainability as one of the major concerns. The case study context is quite typical when working with SMEs adopting service in the manufacturing sector. The case study involves a small French SME, ECOBEL providing water-saving sanitary equipment for healthcare facilities. The typical customer considered by ECOBEL in this paper is a local hospital. ECOBEL’s ambition is to define an attractive offer for the hospital based on a showerhead (for patients’ rooms) and a set of potential maintenance and cleaning services. The value proposition of ECOBEL’s offering is multiple: minimization of the bacteriological health risk, ease of the installation, tracking and maintenance, reduction of the raw material used, reduction of water consumption, and cost reduction (total cost ownership & hidden cost).

The basic idea of the case study was to support the identification and comparison of some design scenarios for such an offering.

The application of the method on this case study was supported by an Excel-based tool implementing the method and allowing the automation of the assessment calculations. For the sake of readability intermediary, calculations will not be presented and the results will be depicted using tables instead of matrices.

4.1 Initial inputs for the selection method

The definition of the value proposition behind the PSS offering is based on the analysis of customer needs (B). Then, the set of sustainability dimensions (E) were identified resulting in six dimensions distributed among the three sustainability pillars (Table 2). The following table represents these initial inputs of the method, namely the potential impacts of customer needs on each sustainability dimensions (corresponding to the matrix U in section 3.1, linking the sets B (rows) and E (columns)).

The links between customer needs and sustainability dimensions are correlated with the coefficients of U. Since U is context-dependent, the proposed method leaves the possibility to the users to select their own approach for identifying, prioritising and selecting potential sustainability dimensions that are likely to be impacted by the identified customer needs. This contributes to quantifying the links between sustainability dimensions and customer needs leading to building U. Within the limit of the current paper, the pursued process consists of a literature investigation in order to prioritise the links between customer needs and sustainability dimensions. Then, final validation was based on a group of experts’ judgement. These experts have backgrounds in sustainable development, environmental assessment, and in engineering. As an example, ‘reduce health risk’ is linked to ‘transportation & distribution reduction’ because it is identified by the experts’ judgement that a reduction of the health risk has a potential to impact the transportation of people and/or goods (i.e. by reinforcing the human control, higher frequency of goods turnover).

4.2 Scenario definition

The global aim of the system is to provide a solution to the hospital to offer to their patient the possibility to take a shower with less sanitary contamination risk. Three scenarios are then designed to fulfil this aim.

Three types of actors are involved in the scenarios: ECOBEL, which is the focal company and the ‘service provider’, the ‘product manufacturer’ and the ‘customer’.

Scenario 3 is a business as usual scenario, i.e. a traditional selling contract of the showerhead with a set of additional services. The basic idea of scenario 1 consists in renting the showerhead while ensure its lifecycle management (maintenance, take back). It includes the design, production, distribution use and end-of-life. Scenario 1 and scenario 2 only differ by the actor ‘product manufacturer’, which is a social company (scenario 1) or a classical company (scenario 2). The set of PSS functions and consequently the actors’ contributions to these functions can be different from one scenario to the others.

In the following section, the first scenario will be used to illustrate the key information manipulated (functions, potential impacts, actors, actors’ contribution, sustainability impact). For the two other scenarios, only calculation results are provided.

4.2.1 Functions

In order to meet the customers’ needs, nine functions and their contributions to the needs are identified and represented in Table 3. For example, the function F₂ ‘Provide a sufficient water jet for washing of the body’ is only contributing to the customers’ need B₁ ‘Wash the patient’.

Table 2. Potential impacts of needs (B) on sustainability dimensions (E).

	E1 – Improve equity and justice in relation with stakeholders	E2 – Empower or valorise local resources	E3 – Added value for customers	E4 – Long-term business development	E5 – System life optimisation	E6 – Transportation & distribution reduction
B1 – Wash the patient	1	2	3	1	2	1
B2 – Maintain the system	2	0	3	2	1	0
B3 – Reduce health risk	0	0	1	1	2	2
B4 – Control the overall cost	1	1	3	2	0	0

Table 3. Contribution of the actors (A) to the function (F) in the first scenario.

B		B1 – Wash the patient		B2 – Maintain the system		B3 – Reduce health risk		B4 – Control the overall cost
F		F1 – Provide a personalised service to the patient	F2 – Provide a sufficient water jet for washing of the body	F3 – Implement a preventive maintenance process	F4 – Implement a curative maintenance process	F5 – Diagnosis of the behaviour	F6 – Control the health risk	F7 – Reduce the consumption of raw material	F8 – Reduce the consumption of water and energy	F9 – Reduce the workforce
A	A1 – ECOBEL	X	X	X		X	X	X	X
	A2 – Hospital	X	X	X	X	X	X		X	X
	A3 – ADAPEI				X			X

4.2.2 Computation of potential impact of the functions

The potential impacts of the nine identified functions on sustainability dimensions are given by the matrix C, which results from the matrix D’ (normalised values of functions’ contribution to the needs) and the matrix U (representing potential impacts of needs on sustainability dimensions). In matrix C (Figure 2), the rows refer to the functions and the columns stand for their respective potential impacts.

Figure 2. C matrix of potential sustainability impacts of the functions.

4.2.3 Actors

At this point, it is needed to identify the list of the actors, which will take in charge the nine identified functions. Three actors are involved in the PSS value network:

• ‘ECOBEL’ the focal company of the PSS network, initiator of the partnership and of the definition of the value proposition, as well as the service provider;

• ‘Hospital’ is the customer of the PSS, which could belong to the public or private sector.

• ‘ADAPEI’ is the supplier of the showerhead. ADAPEI is an association concerned with the working-based integration of disabled persons.

In order to identify the actor’s contributions, the following table links the actors with the functions and customers’ needs. The crosses mean that a given actor is involved in a given function. This corresponds to matrix G (Equation 9) in the method.

4.2.4 Calculation of the sustainability impacts of the actors

The potential impact of the actors on sustainability is calculated through weighting the functions potential sustainability impact by actors’ contribution to those functions. Table 4 represents the actors' potential impact on sustainability in the PSS designed scenario (matrix H).

Table 4. Potential impact of actors.

	E	E1 – Improve equity and justice in relation with stakeholders	E2 – Empower/valorise local resources	E3 – Added value for customers	E4 – Long-term business development	E5 – System life optimisation	E6 – Transportation/distribution reduction
A	A1 – ECOBEL	8/3	8/3	15/2	13/3	9/2	3
	A2 – Hospital	11/3	8/3	9	16/3	5	3
	A3 – ADAPEI	4/3	1/3	5/2	5/3	1/2	0

4.3 Scenario assessment

The assessment of the PSS design scenario is based on a set of sustainability indicators, which need to be filled out by each of the three actors. For the sake of readability, this section reports only on the assessment of the PSS design scenario from the point of view of ECOBEL. The assessment from the other actors’ points of view follows the same logic.

4.3.1 Establishment of sustainability indicator sub-sets

According to Table 4, all three actors are potentially impacting the following sustainability dimensions: ‘Improve equity and justice in relation with stakeholders’, 'Empower/valorise local resources', 'Added value for customers', 'Long-term business development', 'System life optimisation' and only the actors ECOBEL and Hospital impact ‘Transportation & distribution reduction’. Subsequently, only indicators belonging to each actor-impacted dimensions are kept (Equation 12 of the method). The resulting list is shown in the appendix.

4.3.2 Indicators estimated by the actors

The estimates result in the $J^1$ vector composed of the quantified indicator values (scores) $v{I}_{kd}^1$. For ECOBEL, indicator values $v{I}_{kd}^1$ are aggregated using simple average into scores ($L_k^1$) measuring the sustainability according to each of the dimensions k. At the end of this step, the potential impact of each actor is calculated (results presented in Table 5).

Table 5. Assessment of the impact of the actors involved in scenario 1 on the sustainability dimensions.

	E1 – Improve equity and justice in relation with stakeholders	E2 – Empower/valorise local resources	E3 – Added value for customers	E4 – 0Long-term business development	E5 – System life optimisation	E6 – Transportation/distribution reduction
A1 – ECOBEL	2,5	2,0	1,1	1,4	2,9	2,0
A2 – Hospital	2,5	2,3	2,5	2,4	2,9	2,0
A3 – ADAPEI	3,0	3,0	2,5	2,6	2,9	2,0

4.4 Multi-criteria analysis

The two other scenarios were built and assessed following the procedure illustrated in section 4.3. All three scenarios are compared in order to classify them from the different actors’ points of view, using the ELECTRE method. The actors’ potential impacts on sustainability for scenarios 2 and 3 are presented in Table 6.

Table 6. Assessment of the impact of the actors involved in the other scenarios on the sustainability dimensions.

	E1 – Improve equity and justice in relation with stakeholders	E2 – Empower/valorise local resources	E3 – Added value for customers	E4 – Long-term business development	E5 – System life optimisation	E6 – Transportation/distribution reduction
SCENARIO 2
A1 – ECOBEL	2,5	1,5	1,1	1,4	2,9	2,0
A2 – Hospital	1,5	1,5	2,3	2,0	2,9	2,0
A3 – classical manufacturer	0,0	0,5	0,8	1,0	2,9	2,0
SCENARIO 3
A1 – ECOBEL	1.25	2,0	0,38	0	2,9	1,67
A2 – Hospital	2,0	2,0	2,1	0,3	2,9	2,0
A3 – ADAPEI	3,0	3,0	2,0	1,4	2,9	2,0

4.4.1 Computation of the global weights

The global weights of the sustainability dimensions for the different actors are reported on in Table 7. These weights correspond to Equation 15 of the method.

Table 7. global weights of the sustainable dimensions for each actor.

	E1 – Improve equity and justice in relation with stakeholders	E2 – Empower/valorise local resources	E3 – Added value for customers	E4 – 0Long-term business development	E5 – System life optimisation	E6 – Transportation/distribution reduction
A1 – ECOBEL	0,12	0,10	0,31	0,18	0,17	0,11
A2 – Hospital	0,13	0,10	0,32	0,18	0,17	0,10
A3 – ADAPEI	0,23	0,03	0,39	0,26	0,10	0,00

The method of multi-criteria analysis is then applied for each actor to provide them with the concordance and discordance indexes of each pair of scenarios. For all three actors, the concordance threshold is fixed as c = 0,75 and the discordance one as d = 0,25.

4.4.2 Comparison of the three scenarios

According to the point of view of ECOBEL (Table 8), scenario 1 over-ranks scenarios 2 and 3. However, it is not possible to conclude about the over-ranking relation between scenario 2 and 3. Furthermore, as scenarios 2 and 3 are over-ranked by scenario 1, they are already classified as non-acceptable by ECOBEL. For all the actors, scenario 1 seems to be the most sustainable and thus should represent the good trade-off among the three actors in this illustrative case study. Moreover, the ELECTRE method underlines that both scenarios 2 and 3 are non-comparable because the values of the discordance index in all three scenarios exceed the threshold (0,25). A discussion of these results is developed in the next section.

Table 8. Concordance (C Index) and Discordance (D Index) indexes for ECOBEL.

C	Scenario 1	Scenario 2	Scenario 3	D	Scenario 1	Scenario 2	Scenario 3
Scenario 1	-	1*	1*		-	0*	0*
Scenario 2	0,9*	-	0,9*		1	-	1
Scenario 3	0	0,27	-		1	1	-

Actor 1: ECOBEL

According to Table 8 the scenario 1 over-ranks scenarios 2 and 3

Actor 2: Hospital

According to Table 9, scenario 1 is over-ranking both scenarios 2 and 3, which are classified as non-acceptable.

Table 9. Computation of the Concordance (C Index) and Discordance (D Index) indexes for Hospital.

C Index	Scenario 1	Scenario 2	Scenario 3	D Index	Scenario 1	Scenario 2	Scenario 3
Scenario 1	-	1*	1*		-	0*	0*
Scenario 2	0,37	-	0,77*		1	-	0,5
Scenario 3	0	0,5	-		1	0,7	-

Actor 3: ADAPEI

According to Table 10, scenario 1 is over-ranking scenario 3. The scenario 2 is not compared because ADAPEI is not involved in it.

Table 10. Computation of the Concordance (C Index) and Discordance (D Index) indexes for ADAPEI.

C	Scenario 1	Scenario 3	D	Scenario 1	Scenario 3
Scenario 1	-	1*		-	0*
Scenario 3	0,35	-		0,75	-

4.5 Feedback from the case study

The objective of the experimentation was limited to a feasibility study to check the applicability of the selection method, in a way to provide the first proof of concept More specifically, the case study illustrates the required input for the method and the articulation of its steps. The calculations were enabled by a software tool supporting also data collection.

This experimentation illustrates the quite easy access to the required input data despite data scarcity at the early-design stage. The multi-criteria analysis is built at the early stages of the design process and provides the designers with a valuable support to take proper decisions during the design process. This relative easy application could ease a broader deployment of the method into the design community. While helping decision-making, the method also increases the awareness of designers on sustainability drivers and facilitates the selection of the most relevant scenarios from that point of view.

For ECOBEL, the final multi-criteria analysis led to highlight scenario 1: this scenario corresponds to the transformation of the usual ‘selling’ business into a renting system, where the provider (ECOBEL) offers services for all the life-cycle management (maintenance and end-of-life treatment). It is interesting to highlight, according to the mathematical analysis, that the two main actors to decide the scenario (provider and customer) both converge to the conclusion that scenarios 2 and 3 are not acceptable. Here, scenario 3 corresponds to a ‘business as usual scenario’ when scenario 1 corresponds to a PSS solution based on renting economic model: in this (rather limited) example, the decision-aid approach put forth the pertinence of a transition towards sustainable PSS solutions.

Despite this result on a simple case study, the effective implementation of such an offer faces some other issues. Indeed, some actors, although they are convinced of the sustainability benefits, they are still reluctant to implement them in their business model because of the difficulty to perceive the paradigm change towards sustainability. Then, such decision-aid tools become useful, as they could help to change these resistance factors by measuring the global benefits. The objective is to catalyse a transformation in the way of thinking such offers, in the future. The ECOBEL company could be an iconic example of such a transition. The company started its business with launching effectively the scenario 1 (PSS offer, beyond the previous product-selling model) which is still available on the market today. Complementary, the company later developed new innovative PSS offers and started to propose them to some specific clients. The innovation and solution development process is still on the way in the company: an iterative design process is currently running to improve the existing offer, for instance by the extension of the boundaries of the offer and by customisation.

5 Discussion

Sustainability integration within the decision-making process at PSS early-design stages is confronted to a lack of operational tools alongside the fact that most of the existing methods fall short of providing a holistic assessment of sustainability considering its three pillars. The proposed method attempts to fill this gap through a step-wise approach considering the peculiarities of design contexts and integrating seamlessly with the design process. In fact, sustainability indicators can be selected according to the design context and this has no impact on the method steps. Furthermore, the back-and-forth information flows between the design process and the method ensures both reliable assessment of the design scenarios and well-informed decisions on the design alternatives.

The case study presented in the previous section demonstrated the feasibility of the method is being applied to a unique and specific case. In particular, the step by step methodology can support the designers to: define the PSS scenarios designed through a dedicated formalism (functions, actors, potential impact) and to establish relationships between the functions, actors, and potential impacts (Step 1); assess the sustainability of PSS scenarios (Step 2); provide to the decision-makers valuable information about the potential impact of each scenario on sustainability and to identify the one with the highest sustainability potential (Step 3).

In reference to previous contributions in this domain, the method extends the current literature in several ways. First, the method supports a deeper sustainability integration within the design process enabling a filtering of the design scenarios according to their sustainability potential since the early-design stages, this extends research works such as (Geum et al., 2011). The proposed method results of a provided support for the decision-making in the rough context of the early-design step of PSS, taking into account all the specificities of PSS (scenario, actors involved, service/objects), the early-design (blurry information, iterative process) and the integration of sustainability (holistic approach, operationalisation of the concept of sustainability). Second, the method integrates further actors' expectations and provides a sustainability assessment following a multi-actor perspective, thus extending the following researches (Trevisan and Brissaud 2017; Sousa and Miguel 2015; Vezzoli et al. 2014). Third, unlike previous methods such as Trevisan (2016) the proposed method considers all three sustainability pillars, aiming at increasing actors’ awareness of sustainability issues through a comprehensive assessment. Finally, the method allows contextualising the assessment to the design context through building shortlists of sustainability indicators, which ensures further operationalisation, compared to Vezzoli et al. (2014).

Alongside the above multi-fold contribution, some limits of the method open additional perspectives for further research. First, some hypotheses of the assessment (Step 1) could be discussed such as the use of some specific sustainability framework, namely, the one from Vezzoli et al. (2014). While such a framework seemed to be in line with the general objectives of the method (integrating sustainability in PSS design), the peculiarities of design contexts require flexibility in the choice of the sustainability framework and sustainability dimensions. Thus, further improvements could be to introduce some guidance in selecting and customising sustainability dimensions and indicators. Additionally, the method is, at some specific points, based on subjective points of view (Step 2). For example, the auto-assessment by the actors may introduce some biases into the final scores. Also, the translation of the qualitative answers into a quantitative score introduces a progressive bias in the assessment process. The influence of these issues on the result of the method is quite unknown and the estimation of these influences could reinforce the accuracy of the result. Sensitivity analyses would be relevant to estimate the influence of key factors on the final results and to recommend subjectivity-oriented models like fuzzy logic if required.

Furthermore, while in the case study the method provided an acceptable scenario for all actors, in some cases this turns out not to be the case (Step 3). In this case, further steps of the method would be needed to support the designers build new trade-off scenarios, derived from the existing and non-acceptable scenarios. The trade-offs require in turn to go through a negotiation process involving the actors. Proper guidelines and protocols are then needed to support the negotiation and improve its efficiency.

More generally, the information provided by the method is increasing the knowledge of the designers about the system designed. As the design process is iterative, an interesting perspective would be to investigate the influence of the information provided by the proposed method on the new iterations. This will help, firstly, to identify the reused pieces of information which could be focused on for further development of the method; secondly, to better understand the whole mechanism of the PSS design process. This said, the method still requires further validation through additional case studies in order to fine-tune each of its steps. The deployment of additional case studies would also validate the adaptability of the method to different application contexts and is likely to open up new improvement perspectives.

6 Conclusion

Based on observed gaps in PSS engineering literature, some key principles were identified and a 4-step selection method was proposed. The method integrates a sustainability assessment procedure and a multi-criteria analysis method to show to the actors of the PSS network their potential sustainability impacts from the early-design stages. The method helps identifying sustainability hotspots among the PSS scenarios, thus increasing designers' awareness on sustainability issues. In addition, the method supports an iterative design through the back-and-forth information flows informing designers of the potential impacts of their decisions.

The current method applies to early-design stages and only considers function-based scenarios. Further research directions include the extension of the method to detail design stages so as to cover the assessment of solutions implementing the scenarios (PSS structural elements).

Acknowledgments

This research work is funded by the Arc8 network, part of the Rhône-Alpes Region research program, which aims to support the scientific research in the area. The authors warmly thank the company executive of Ecobel, M. Jean-Pierre Bosles.

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Notes on contributors

Benjamin Doualle

Benjamin Doualle is currently a consultant engineer in Industry. He receive his PhD from Ecole des Mines de Saint-Etienne (France) in 2018. He told a MSc of Engineering and Management of Environment and Sustainable development from Université de Technologie de Troyes in 2014.  His scientific interests are : sustainability assessment, circular economy, eco-design.

Khaled Medini

Khaled Medini is currently an Associate Professor at Ecole des Mines de Saint-Etienne, Fayol Institute. He received his PhD from Ecole Centrale de Nantes in 2013. He holds a MSc of Industrial Engineering from Ecole Centrale de Lille, 2010 and an Engineer Diploma from Ecole Nationale d'Ingénieurs de Tunis, 2009. His research interests relate to enterprise modelling (meta-modelling, object oriented modelling, model transformation), decision making support (aggregation method, linear programming), and finally modelling and simulation (analytical, agent based, discrete event), with the following application domains, mass customization and variety management of products and services, product service systems engineering and costing, and sustainability assessment of products and systems at large.

Xavier Boucher

Xavier Boucher is currently Professor in Industrial Management at the Ecole Nationale Supérieure des Mines de Saint Etienne (France) where he is teaching production management, re-industrialization strategies and industrial transition towards new business models for engineering students. As researcher he is currently Research Director for FAYOL Research Center (interdisciplinary Research Center, developing activities for the Global Performance of Industrial Companies and Territories). His main research orientation is currently focusing on the design of Industrial-Product-Service-Systems (PSS) and the change of industrial Business Model by a transition towards service-integrated activities for manufacturing companies (servitization process). Active member of several scientific societies in the field of Industrial Engineering (EU Leader for IFIP WG 5.5 during many years, active member of CIRP IPSS2 working group on Product-Service-Systems, Active member of SOCOLNET Society on Virtual Organisations), he has participated to several EU projects and has coordinated several national research project for Mines Saint Etienne (ANR ServINNOV, ANR PSS Euro Network, FUI Clean Robot, FUI Affinid). He has published more than 30 articles in International & National scientific journals like Computers in Industry, International Journal of Computer Integrated Systems, Journal of Decision Systems, as well as more than 90 communications in international conferences.

Daniel Brissaud

Daniel BRISSAUD has been a professor of engineering design and eco-design at Grenoble INP since 1998. He is director of School of Industrial Engineering and Management and CIRP fellow. He was vice-president of Grenoble INP for strategic planning and development, head of the research cluster on engineering for industry and innovation at Rhone-Alpes area and has led the French survey on sustainable production systems for the future. His scientific interests are: Eco-design, environmental assessment, lifecyle engineering, clean technologies, human beings in industry 4.0. He authored more than 70 papers in international journals and books and mentored more than 20 PhD theses.

Valerie Laforest

Valerie Laforest is currently Research Director in Environmental Sciences. In 2014, she took responsibility of the research and teaching department named “Organisation and Environmental Engineering” of Fayol research Center of Mines Saint-Etienne. Her major research activities concern the development of environmental performance assessment methods notably for best available techniques, as well as decision support tools for the implementation of industrial ecology strategies. She has published more than 20 articles in International Scientific Journal like Journal of cleaner Production, Waste Managment or Resources, Conservation and Recycling. She is involved in several Scientific Research Groups : UMR 5600 Environment, Town and Society – EEDEMS - EcoSD French Network - ETV AFNOR expertise group – ACPN (Advanced in Cleaner Production Network). Moreover, Valerie is reviewer for several scientific journals such as Elsevier’s Journal of Cleaner Production, and Resources, Conservation and Recycling. Moreover, she teaches industrial ecology, ecotechnologies, environmental processus for French and European regulation application. Valerie Laforest became a Knight of the Order of the Academic Palms in 2013.

APPENDIX

Table

Dimensions	Indicators
Employment/working conditions	Risk of child labour
	Risk of teenagers labour
	Risk of forced labour
	Risk of all forms of discrimination in the workplace
	Risk of intervention on right of workers
	Risk of discrimination about the age of the worker
	Risk of excessive working hours
	Risk of intervention on right to collective negotiation
Equity and justice in relation to stakeholders	Involvement of stakeholders in design processes
	Consider stakeholders’ expectations and potentialities
	Promote the adoption of social certification systems by suppliers and subcontractors
	Promote the quality and accessibility of common goods
Responsible/sustainable consumption	Adopt standards underlining the social sustainability
	Provide the information to educate the end-user on sustainable behaviour
	Responsible/sustainable participation of the end-user
	Involve the end-user in the implementation of his/her own product-service systems towards sustainable behaviour
	Involve the end-user in the design process, of his/her own product-service systems towards sustainable behaviour
Integrate the weaker and marginalised	Involve the conditions for weaker social strata
	Involve marginalised persons offering them qualifying jobs which enhance their competences
	Involve introducing foreigners into the social context
	Develop offer at a cost accessible even for people of low-income
	Diversify the offer to increase the access capacities
	Develop systems with shared economic property
	Develop systems which promote labour services with equitable access
	Develop systems which involve product sharing and cost reductions
	Develop systems which allow easier access to credit
Social cohesion	Risk of gender marginalisation
	Risk of marginalisation between cultures or generations
	Risk of gender discrimination
	Risk of cultural discrimination
Empower/valorise local resources	Influence of the system on the local economy
	Link between the system and the educational and training institutions
Market position	Visibility of the offer on the market
Added value for businesses	Risk of fast evolution of customer’s needs
	Risk of non-responsiveness to the evolution of customer’s needs
	Value of products or support products at their end-of-life
Added value for clients	Reduce the cost of waste disposal
	Increase the revenue
	Cost reduction
	Debt/taxes reduction
	Increase the satisfaction
	Responsibility transfer
	Customisation of the offer
	Flexibility
Long-term business development	Potential improvement of the flexibility
	Potential improvement of the innovation capacity
	Use of a tracking method
	Potential of knowledge integration
	Expertise in information and knowledge management
	Risk investment
	Risk of the evolution of the law
Partnership	Potential partnership identification
	Amount of partnership
	Potential improvement of the organisation
	Potential of price reduction
	Potential improvement of the visibility of the offer
Macro-economic effect	Potential impact on local economy
	Use of local products or support products
	Potential for local job creation
System life optimisation	Amount of disposal products
	Amount of single use products
	Amount of packaging
	Packaging recyclability
	Functional obsolescence
	planned outage
	Expiration by notification
	Obsolescence by incompatibility
	Aesthetic obsolescence
	Existence of more easily weared out part
Transportation reduction	Transport of goods
	Transport of people
	Optimisation of the transport logistic
Resource reduction	Amount of energy used
	Use of eco-design method
	Amount of natural resources or consumables
Waste minimisation/valorisation	Implementation of end-of-life product recovery system
	Amount of waste destined to landfill
Conservation/biocompatibility	Amount of non-renewable resources or with a very low renewal rate
toxicity reduction	Amount of potential hazardous substance
	Existence of a mechanism to prevent contamination risk