An investigation of the most important factors for sustainable product development using evidential reasoning

Those working in product development need to consider sustainability, being careful not to compromise the future generation's ability to satisfy its needs. Several strategies guide companies towards sustainability. This paper studies six of these strategies: eco-design, green design, cradle-to-cradle, design for environment, zero waste, and life cycle approaches. Based on a literature review and semi-structured interviews, it identifies 22 factors of sustainability from the perspective of manufacturers. The purpose is to determine which are the most important and to use them as a foundation for a new design strategy. A survey based on the 22 factors was given to people working with product development; they graded each factor by importance. The resulting qualitative data were analyzed using evidential reasoning. The analysis found the factors "minimize use of toxic substances, " "increase competitiveness, " "economic benefits, " "reduce material usage, " "material selection, " "reduce emissions, " and "increase product functionality" are more important and should serve as the foundation for a new approach to sustainable product development.


1.
Introduction. It is increasingly important for industries to reduce their environmental impact. Initially, they were simply worried about the levels of emissions they produced but have now broadened their scope to include the concept of sustainability [24]. The first international conference on sustainability, held in 1972, led to the establishment of the United Nations Environmental Program (UNEP). The term sustainable development was first articulated in the 1987 Brundtland report as development which meets the needs of the present without compromising the ability of future generations to meet their own needs [47]. Sustainable development is applied to economic development, social equity and environmental protection but is most frequently associated with the latter. The term is now used worldwide, with new interpretations being generated [15]. One of the 17 goals created by the United Nations Sustainable Development Summit was to ensure sustainable consumption and production patterns [42]. To do so, it suggested reducing resources and pollution along the whole lifecycle, something requiring cooperation in the whole supply chain, from producer to final consumer [43].
Sustainability is an increasingly important factor in product development. Even though the effort may increase the cost of developing products, more companies are integrating sustainability into their product development processes, largely in response to consumers preferences for eco-friendly products [23]. Myriad methods and strategies for developing sustainable products have been introduced, all considering the distinct relationship between environment and product [44]. The question is how to weight and choose an approach and then apply it in an organization [11].
This paper looks at the possibility of developing a new approach to sustainable design. The goal is to offer a more holistic view of sustainable design, especially for companies working with product development. The proposed approach needs to be efficient, reliable and practical to implement. Through a literature review, it identifies and integrates the advantages of six design strategies: eco-design, green design, cradle-to-cradle, design for environment, zero waste, and life cycle approaches. To validate the findings in the literature and identify any new concerns and advantages, managers in Swedish companies working in product development were interviewed. Once all factors were identified, a survey was given to people involved in product development; they were asked to rate the importance of each.
Because of the qualitative nature of the data, the Evidential Reasoning (ER) approach was used to rank the identified advantages of each approach. ER advocates a general, multi-level evaluation process for dealing with Multiple Criteria Decision Making (MCDM) problems. Because ER can handle qualitative and quantitative uncertain information of any kind, e.g., lacking data, missing data, incomplete information, it is preferable to fuzzy logic or any other approach to ranking. ER is based on Dempster-Shafer evidence theory [40], the evaluation analysis model [56], and decision theory. When the concepts of belief structure [56], [51], [53], and the belief decision matrix [50] are introduced to ER, it is possible to model various types of uncertainties in a unified format [49]. In the present case, for example, the most important factors in sustainability can be identified and a new sustainable design approach suggested. The paper is organized as follows. Section 2 comprises an introduction to sustainable product development. Section 3 outlines the ER approach, while Section 4 provides the results and a discussion. Section 5 offers a conclusion and explores further research directions.
2. Sustainable product development. Companies have many different tools, methods, and strategies to use if they wish to establish a more sustainable method of product development. The difficulty is selecting one strategy: they all have different objectives, and each has its own advantages and disadvantages. Section 2.1 lists some sustainable design strategies; Section 2.2 gives their advantages and disadvantages; Section 2.3 describes the interview process used in this study and explains how the survey was developed.
2.1.1. Eco-design. Eco-design strives to minimize the negative environmental impact of all product parameters [7], [46], [36]. This can be done by avoiding toxic substances and reducing resource and energy use throughout the product life cycle. It can also be done by prolonging product life, for example, by repairing and recycling, or providing the possibility of upgrading the product. Alternatively, the strategy may focus on minimizing the weight and the number of joints in a product [32]. Eco-design can increase a products functionality and improve its economic and environmental performance [36]. Several advantages result from eco-design, including increased competitiveness, lower costs, and better company image [24]. When looking more closely at the economic factors of eco-design, Plouffe et al. [36] find sales volume and revenue increase, while variable costs decrease. However, they also find fixed costs for eco-design products are higher, and the economic benefits of eco-design are mainly short term.
2.1.2. Green design. Green design, or sustainable design as it is also called, aims to reduce the negative impact on the environment by minimizing waste, reducing the use of non-renewable resources [6], [19], [34], and optimizing operational practices [5], [17], [19]. Another way of reducing the negative environmental impact is to minimize the hazardous inputs and outputs from energy and materials and use renewable materials and energy. Green design is a process, with products designed to have an afterlife [3], [34]. Green design should incorporate ecological and social business strategies for sustainability [10]. To work with green design, companies need to invest in specific analytical and organizational tools, systems and software [5], [17]. The tools for green design include checklists and guidelines. One identified disadvantage to green design is that it contains too many suggestions that have not been practically tested [6].

2.1.3.
Cradle-to-cradle. The cradle-to-cradle concept divides materials into two categories: biological and technical. The biological category contains biodegradable materials; the technical category contains materials that can be reused or recycled without losing their properties. The life-cycle of the material flow in cradle-to-cradle is circular, with material reused to reduce negative effects on the environment [9]. As the method mimics the metabolism of a body, the materials are considered food and nutrients for new products [8]. Products and materials meeting cradle-tocradle standards comply with certain principles and criteria, including elimination of the concept of waste, use of renewable energy, carbon management, water stewardship, and social fairness [33]. Because of the material types, the availability of raw materials is ensured. In addition, the reprocessing of materials may lead to increased economic activity and more job opportunities [9]. Even so, there can be over-confidence in the approach; people may believe the method has the answers to everything, and this can harm its implementation [4].
2.1.4. Design for environment. Design for environment (DfE) can be used to evaluate progress towards cradle-to-cradle products [39]. The DfE tool is intended to prevent built-up waste caused by the product and its production [16]. Products can be evaluated by scoring them in three categories: material chemistry, disassembly, and recycling [39]. Many different tools and techniques can be used when working with DfE. The plethora of choices causes designers to struggle when choosing tools to work with [45].
2.1.5. Zero waste. Zero waste is a natural cycle defined by Zero Waste International Alliance (ZWIA) in which waste is eliminated by designing it to become a new resource [55] and increasing recyclability by leaving nothing for disposal [18]. Zero waste can be achieved by redesigning products and their packaging, thus putting the responsibility for waste on the producer [31]. By implementing zero waste, environmental factors such as air and water are greatly improved, and there is much less pollution [55], [30]. The zero waste strategy can result in economic benefits by decreasing costs of waste disposal [18], [32], [12] and by increasing earnings by selling material that can be recycled [14]. To adopt a zero waste product development strategy, most companies must transform their current systems to a refined zero waste system [54]. The implementation of a zero waste strategy may result in increased short-term costs; some businesses may consider this incurs too much risk [14].
2.1.6. Life-cycle approaches. Life-cycle approaches are divided into life-cycle analysis and life-cycle design. Life-cycle analysis (LCA) evaluates the impact of a product on the environment throughout its entire life-cycle [38], [57]. The tool analyzes such factors as raw materials, usage, disposal, recycling, reuse, and end of life [29]. The evaluation is often used for benchmarking studies, trade-off decisions, and hot-spot analysis and evaluation. Life-cycle analysis from older projects can also be used as a source of knowledge [4]. Unfortunately, the potential of LCA is not fully developed; companies more commonly use a retrospective view and/or take a marketing perspective [28], [38]. Another weakness is that it cannot evaluate the balance between material and energy consumption, especially when the life cycle has loops in terms of remanufacturing, reuse, and recycling [41].
Life-cycle design is a holistic concept where products are developed for the public social and environmental needs without compromising the clients requirements [4], [38], [57]. The use of products is increasing worldwide; when existing products are compared to new and environmentally improved ones, most can be seen as waste [25]. The concept also decreases the cost of maintenance services [41].

2.2.
Advantages and disadvantages of sustainable design strategies. The literature review shows there are several advantages and disadvantages of these various strategies, from both an environmental and a business perspective. Most strategies have more advantages than disadvantages, however. Table 1 lists the various design strategies, along with their advantages and disadvantages.
As Table 1 shows, eco-design has the most advantages as well as disadvantages together with green design, zero waste and life-cycle approaches. When we look at the advantages, we see patterns emerging, more clearly shown in Table 2, where the advantages are grouped under factors. By finding the most common patterns, we may be able to determine which factors are most important to sustainable design. Note that there may be more advantages than disadvantages because most articles focus on the benefits of the strategies. However, some strategies do not have a clear line to follow, involve too many tools for measurements, and it may be difficult to implement them.

Interview and survey.
To support the literature review and find factors other than those mentioned in the literature, we conducted semi-structured interviews with managers from three Swedish companies working with product development; see appendix A.
The first company has 44000 employees worldwide. The managers said the most important concerns in their product development are selecting sustainable materials and reducing energy use both in the manufacturing process and later when the product is in use. The company works with sustainability and follows ISO14001 standards. ISO14001 is an international standard for environmental management systems [21].
The third company is based in Sweden and has about 50 employees. The company focuses its environmental work on minimizing the use of toxic substances and complying with existing rules and regulations. It works with sustainability by following ISO14001, as well as EU-directives for the environment [21]. Managers mentioned the effects of pressure from their customers to develop sustainable products and use sustainable materials.
According to our interviewees, important concerns in developing a sustainable strategy are: • Sustainable material selection • Reduced energy usage • Reduced emissions • Minimized use of toxic substances • Increased competitiveness • Economic benefits Most correspond directly to the factors identified in the literature, with the addition of two new ones: sustainable material selection and reduce emissions.
Next, a survey was designed based on the 20 factors collected from the literature review and the two new factors; see appendix B. The survey was distributed, along with instructions, to people working in product development. Respondents were asked to rate the importance of each factor in sustainable product development based on five grades: H = {H1 = unimportant, H2 = not very important, H3 = quite important, H4 = important, H5 = very important}. They were asked to answer the questions by assigning a degree of belief, from 0 to 100 percent in different grades and for different answers. If they were unsure of the importance of a factor, they could respond dont know. Because of time constraints, the surveys were answered by only 10 respondents with an average of eight years of experience in product development, but it is advisable to have more respondents.
3. Methodology. MCDM methods aid decision-making in situations where there are multiple, often conflicting, criteria like multiple quality and quantity attributes. Many MCDM methods have been developed, such as Multiple Attribute Utility Theory (MAUT) [22], and Analytical Hierarchy Process (AHP) [27], [35]. Most are suitable for solving small scale MCDM problems without uncertainty. In uncertain situations, the Fuzzy Multi-Criteria Decision Making (FMCDM) [20], [26] approach provides an ideal option; it has been tested by a number of researchers to rank alternatives in a variety of situations. However, the fuzzy approach is used only when uncertainty is predominant. In other words, when a particular parameter is quantifiable with fair degree of accuracy, or there are a missing or incomplete data, this approach need not be used.
Most real-life decisions use a mixture of qualitative and quantitative attributes with varying degrees of uncertainty, increasing the need for the development of scientific methods and tools that are rational, reliable, repeatable, and transparent. ER is one of the latest developments in the MCDM literature and appears to be the best fit to handle uncertain information [48]. It can model multiple attribute decision problems, using both quantitative and qualitative attributes. The approach is flexible and can process objective or subjective information. It can also handle uncertain, incomplete, or missing information [1], [2]. Given this, we selected ER to interpret the data and to rank the factors.
3.1. Basic evaluation framework. The ER algorithm can be used on MCDM problems; with this algorithm, a complex general property which is usually difficult to assess directly can be broken down and operationalized by using well-defined, measurable concepts that together constitute the general property. The result of such a breakdown is a multiple attribute framework taking the shape of a tree (hierarchy) structure, with assessable basic attributes at the lowest level. The assessment of these basic attributes can be aggregated to an assessment of the upper level of the tree (Figure 1).
Upon assessment of the basic attributes, however, there is always a certain level of uncertainty. Dempster-Shafer mathematics are designed to aggregate the uncertainties in the basic attributes to a total uncertainty of the total assessment [37].

ER approach based on Dempster-Shafer theory.
Steps for the overall assessment of the complex general property are suggested by Yang [52] based on

AHMADZADEH, JEDERSTRÖM, PLAHN, OLSSON AND FOYER
Dempster-Shafer theory [40] and summarized by Pontus et al. [37]. The steps are given below and shown in Figure 1. As the figure shows, the application of Dempster-Shafer theory to one sub tree (dashed rectangular) and reasoning can be generalized to an entire tree consisting of several sub trees.

Figure 1. Generic framework to assess general property
Step 1. Define a set of L basic attributes including all factors influencing the assessment of the upper level attribute as follows: E = {ε 1 , ε 2 . . . ε L }. Now estimate the relative weights of the attributes, where ω 1 is the relative weight for basic attribute ε i and is normalized so that Σ ω i = 1 and 0 ≤ ω i ≤ 1. Define N distinctive evaluation grades H n , n = 1, . . . , N as a complete set of standards to assess each option on all attributes. For example, H = {H 1 = worst, H 2 = poor,. . . , H N −1 =good, H N =excellent}. For each attribute ε i and evaluation grade H n a degree of belief β n is assigned. The degree of belief denotes the sources level of confidence when assessing the level of fulfillment of a certain property.
Step 2. Let m ni be a basic probability mass, representing the degree to which the i th basic attribute ε i supports a hypothesis that the general attribute is assessed to the n th evaluation grade H n . Then, m ni is calculated as follows: 1 (1) Let m Hi be the remaining probability mass unassigned to each basic attribute, ε i so m Hi is calculated as follows: Decompose m Hi intom Hi andm Hi as follows: m Hi =m Hi +m Hi (4) Step 3. The assessments of the basic attributes constituting the general property are aggregated to form a single assessment of the general property. The probability masses assigned to the various assessment grades, as well as the probability mass left In equation 5, we continue to let i = 1. The term m n,1 , m n,2 measures the degree of attributes ε 1 and ε 2 supporting the general attribute y to be assessed to H n ; the term m n,1 , m H,2 measures the degree of only ε 1 supporting y to be assessed to H n ; the term m H,1 , m n,2 measures the degree of only ε 2 supporting y to be assessed to H n .
In equation 7, the termm H,1 ,m H,2 measures the degree to which y cannot be assessed to any individual grades due to the incomplete assessments for both ε 1 and ε 2 . The termm H,1 ,m H,2 measures the degree to which y cannot be assessed due to incomplete assessments for ε 2 only. The termm H,1 ,m H,2 measures the degree to which y cannot be assessed due to incomplete assessments for ε 1 only. The term m H, ,m H,2 in equation 8 measures the degree to which y has not yet been assessed to individual grades due to the relative importance of ε 1 and ε 2 after ε 1 and ε 2 have been aggregated. K I(2) , as calculated by equation 9, is used to normalize m n,I (2) and m H,I(2) so that: Step 4. Let β n denote the combined degree of belief that the higher level property assessed to the grade H n is generated by combining the assessments for all the associated basic attributes ε i . So β n is calculated by: Steps 1-4 can now be employed for the other sub-trees to obtain the combined degree of belief in the higher level of the hierarchy model.
Step 5. In this step, the utilities of the respective assessment grades H ( 1 . . . n) are estimated via utility functions (u(H n )). A range of methods and techniques can be utilized for this purpose. In this paper, we assume the utilities of the respective assessment grades can be appreciated in a linear fashion. Therefore, the top level score of the hierarchy model can be obtained by Σβ n u(H n ), n = 1, . . . , N (cf. Figure  2). Figure 2 gives a visual representation of the five steps in the form of a flowchart.  Figure 1, the application of ER approach described through an example in which a complex property is broken down into two sub trees, but we will focus on only the left sub tree (dashed rectangular).

Evalutation Grade Weight
Belief  Table 3. Assigned weights, belief degrees and calculated probability masses Step 3. We use the recursive ER algorithm to aggregate the probability masses of the basic attributes ε 1 and ε 2 to the intermediate property Statement 1. Putting the values from Table 3 into equations 5 In the same way, assessments of the Statement1 property for the grade H 2 = average, H 3 = good are equal to 0.5401 and 0.0751 respectively.
Step 4. By using equations 11, 12 and with the values calculated in step 3, we get the combined degrees of belief for the property Statement1 for the different grade values. These can be expressed as:  Table 4 shows the mean value for each grade (H = {H1 = unimportant, H2 = not very important, H3 = quite important, H4 = important, H5 = very important}.) and factor based on the survey results. The mean value is calculated by adding up the respondents degree of belief in each grade. A Windowsbased Intelligent Decision System (IDS) is applied to implement the ER approach. IDS is a general-purpose multiple criteria decision analysis tool; it provides graphical interfaces to build a decision model and can be assessed on a hierarchy of criteria. The factors of sustainability considered here are not arranged by hierarchy; rather, all are assumed to be top-level criteria. The mean values for each factor were entered into the IDS. The results are shown in Figure 3 and Table 5.
The results from the IDS shows all factors are important, with average score of more than 53%, but the most important ones, i.e., those with a score of over 65% (the mean value of all factors), are the following: minimize use of toxic substances (82%), increase competitiveness (76%), economic benefits (75%), reduce material usage (74%), sustainable material selection (72%), reduce emissions (69%), and increase product functionality (69%). In Table 4, the column marked unassigned indicates the percentage of degree of belief that was not given; a high unassigned percentage may indicate uncertainty in the answers.  Table 4. Factors identified in sustainable design and the corresponding strategies When we look at the unassigned percentage for each factor, a few things stand out; notably, the factor circular material flow has 49% unassigned, holistic view has 37% unassigned, and sustainable social standards has 26% unassigned. This indicates respondents are uncertain of their answers or unsure of the definition of the factor or how it is connected to sustainable product development. Several factors have a high unassigned percentage at around 20%; this may affect the reliability of the answers; when the information is complete, the results are more reliable. Table 2 and Table 6 shows that most of the important factors are covered by the Eco-design strategy, apart from the two mentioned during the interviews. Clearly, Eco-design is the dominant environmental strategy  · · · Reduce emissions (69%) · · · Increase product functionality (69%) Eco-design Table 6. Important design factors and corresponding design strategy 4.2. Sensitivity analysis. IDS software can be used to visually conduct a sensitivity analysis to indicate how much the various factors influence the results. The weight of each factor is set to a value of 1 in the default settings of the software; by decreasing the weight of one or more factors, the effect on the average score of the complete analysis on different factors (advantages) can be determined. When the weight of the seven most important factors is decreased, one by one, to a value of 0, the average score for the analysis decreases from 65 to 64%. When the weight of all of the most important factors is decreased at the same time, the average score of the analysis decreases to 60%.
In Table 4, several factors have a high unassigned percentage; as noted above, this may indicate uncertainty in the answers. The factors with the highest unassigned percentages are circular material flow, holistic view, sustainable social standards, eliminate emissions, minimize waste, increase use of renewable materials and increase use of renewable energy. To evaluate how much the factors with high uncertainty affect the outcome of the analysis, the weight of those factors is reduced, one by one, to a value of 0; in this case, the average for the complete analysis stays the same. When we decrease the weight of those factors together, the average score of the analysis is increased to 66%. The factors with a high unassigned percentage affect the average score very little, suggesting the results are reliable. In sum, the factors scoring the highest and found to be the most important have the highest impact on the average score of all the factors.

5.
Conclusion. The results suggest the need to develop a new sustainable design strategy, as no single existing strategy meets all the most important needs. The study considers a wide range of sustainable design strategies proposed in the literature and selects six common strategies focusing on the environment and business: eco-design, green design, cradle-to-cradle, design for environment, zero waste, and life-cycle approaches. Although the literature mentions both advantages and disadvantages for each strategy, we focus on advantages, mainly because the aim is to lay the groundwork for a new design strategy with advantages from a business perspective. Data evaluation is simplified by using such qualitative terms as unimportant, not very important, quite important, important, and very important. This might have induced some overly simplified answers for very complex questions on sustainability, but it makes the results much more comprehensible.
The data from the IDS rank all factors as important, but seven stand out: minimize use of toxic substances, increase competitiveness, economic benefits, reduce material usage, sustainable material selection, reduce emissions and increase product functionality. Most are covered by the eco-design strategy; however, that strategy also contains certain factors that are not considered important, while two of the more important factors are not found in any strategy, only in the interviews. It is possible that these two factors are more fundamental to sustainable product development and manufacturing in real life and, thus, have not been considered in design strategies in the literature.
When developing a new approach to sustainability, it is important to consider possible disadvantages and ensure they are not incorporated into it. In this sense, eco-design can serve as an inspiration, but more work is needed to develop new guidelines, tools and methods. In addition, as a new method is developed, it would be advisable to test and evaluate it continuously in collaboration with product development companies. Two of the most important factors are economic benefits and increase competitiveness; these are proof that companies need a strategy with advantages for both business and the environment.
The results may be questioned because of the uncertainty of some factors, but the sensitivity analysis shows the factors with a high unassigned percentage of uncertainty affect the average scores very little, indicating that the results are reliable. However, the research was limited by its timeframe and we consulted only respondents working with product development. In the process of developing a tool for sustainable product design, it would be advantageous to collect more surveys and interviews from companies and compare them with the results from this study. Arguably, this study has found the most important factors, but having more respondents complete the survey and possibly clarifying some factors in the survey might help.
The results lay the foundation for a new tool combining advantages that are attractive from a business perspective, but for future research, it would be advisable to consider different approaches. If the evaluation base is in pure mathematics, i.e. a purely quantitative approach, Preference Function Modeling (PFM) is preferable. If it is a mathematically rigorous synthesis of judgements, i.e. a qualitative approach, researchers could use Direct-Interactive Structured-Criteria Utility Scoring (DISCUS MCDM).