You are on page 1of 16

Journal of Cleaner Production 241 (2019) 118327

Contents lists available at ScienceDirect

Journal of Cleaner Production


journal homepage: www.elsevier.com/locate/jclepro

A system dynamics approach to product design and business model


strategies for the circular economy
Maria A. Franco
Department of Engineering and Information Technology, Bern University of Applied Sciences, Quellgasse 21, CH-2501, Biel, Switzerland

a r t i c l e i n f o a b s t r a c t

Article history: By means of a system dynamics computer simulation model, this paper analyzes the systemic effects of
Received 8 June 2018 combining multiple product design and business model strategies for slowing and closing resource loops
Received in revised form in a circular economy. In the model, green (i.e., recycled) and brown (i.e., traditional linear) products, as
3 September 2019
well as products under a PSS model, flow in a closed-loop supply chain. The model structure covers
Accepted 6 September 2019
issues related to product design, product use and replacement, second-hand use, product discard, and
Available online 6 September 2019
finally, collection and processing for product recycling. Results from this research provide insights into
Handling Editor: Prof. Jiri Jaromir Klemes the relationship between design considerations at the beginning of a product's life and their implications
for the product's take-back stage when a new manufacturing cycle begins. The main contributions of this
Keywords: paper are the considerations of time delays and reinforcing feedback loops in the design of an effective
Circular economy closed-loop supply chain.
Circular product design © 2019 Elsevier Ltd. All rights reserved.
Circular business models
System dynamics
Simulation

1. Introduction policymakers, and market players alike. In fact, most of the research
related to product design and business model innovation for the
It is nowadays widely acknowledged that for the circular purpose of the CE is, so far, mainly qualitative and often supported
economy (CE) to advance at the firm level, corporate strategies by case studies only (Bocken et al., 2016; Manninen et al., 2018).
should focus on both product design (PD) and business model (BM) This is not a surprise given that qualitative methodologies are well
innovation (Bocken et al., 2014; den Hollander et al., 2017; Lieder suited for studying novel phenomena such as the CE, where the
and Rashid, 2016; Rashid et al., 2013; Urbinati et al., 2017). Imple- knowledge of relevant actors, strategic processes, and new forms of
menting both types of strategies in conjunction is expected to allow network organizing is still limited (van Dijk et al., 2014).
stakeholders across the value chain to reap all the economic and The research question that this study proposes to investigate is:
environmental outcomes that the CE has promised to deliver. While What interesting dynamics can be observed when testing isolated
eco-design encompasses the principles, strategies, and methods and combined applications of various product design and business
used to design products with environmental considerations in model strategies for slowing down and closing resource loops in a
mind (e.g., design for longevity, design for maintenance, and design CE? To conceptualize and test various combinations of product
for recycling) (Bovea and Pe rez-Belis, 2012), sustainable business design and business model strategies, a system dynamics computer
models, such as leasing or sharing, concentrate on replacing capital simulation model is developed. Simulation models mirror the
ownership with the use of services (Boons and Lüdeke-Freund, operation of real world processes, systems, or events (Law and
2013). Despite the apparent agreement on the need for firms to Kelton, 2000) and are especially useful when the problem under
incorporate circular design and business model principles into their study is longitudinal, and nonlinear, or when empirical data are
strategies, quantitative studies aimed at testing the wider effects difficult to obtain (Davis et al., 2007). Rather than looking at the
(i.e., benefits, trade-offs, and impacts) of isolated and/or hybrid specific financial, environmental, and social value that is created
strategies for circularity are still not available for academics, through product design and business models for circularity, this
paper looks at the broader picture to understand the general dy-
namics present in a closed-loop value chain where a hypothetical
E-mail addresses: maria.francomosquera@bfh.ch, maria.francomosquera@bfh. manufacturing firm commercializes a generic product.
ch.

https://doi.org/10.1016/j.jclepro.2019.118327
0959-6526/© 2019 Elsevier Ltd. All rights reserved.
2 M.A. Franco / Journal of Cleaner Production 241 (2019) 118327

flows used in industrial ecology (IE) (Stahel, 2010). According to the


List of abbreviations IE tradition, the basic idea in a CE is that the economic and envi-
ronmental utility and value of materials in products are preserved
BM Business model at their highest level for as long as possible, either by lengthening
CBM Circular business model products' lifetimes or by looping products’ components back into
CE Circular Economy the system to be reused and/or recycled (den Hollander et al., 2017;
DfD Design for disassembly Saavedra et al., 2018). In other words, in a closed-loop or circular
DI Disassembly index production and consumption system, waste becomes input for new
EOL End-of-life processes, thus reducing the need for virgin raw materials in new
GIF Green image factor production cycles. In contrast to an open-loop model where
IE Industrial ecology products become waste at end-of-life, the CE provides a framework
M&P Mature and pervasive where resources are kept in loops in order to create more systemic
MCI Material compatibility index value and for longer periods (McDonough and Braungart, 2002; Su
PSS Product service systems et al., 2013).
RI Recyclability index For circular products to mimic a closed-loop system, material
SD System dynamics flows need to remain accounted for, before, during, and after the
T&E Trendy and emotional products' end of life (den Hollander et al., 2017). It is within this
context that certain guidelines on product design and business
models must be considered by product manufacturers in the
development of truly circular industrial systems. While design
The model's feasibility is tested by using data derived from the practices focus on maintaining materials at their highest value at
literature and by conducting structure and behavioral tests, as well any point in the products’ lifecycles, business models are mainly
as by testing model assumptions for selected product categories. concerned with the way products are commercialized and
The main purpose of this article is to provide a holistic and reliable “consumed.” A summary of the relationship between these two
decision-support tool for industrial stakeholders at the intersection concepts and the CE is presented next.
of product design (and consequently of post-use) and business
model strategies in a CE. All the possible configurations associated 2.2. Eco- and circular product design
with different strategy bundles bring manufacturing companies to
an uncertain and even risky position because their effects are un- The idea of eco-design is not new and has been present since the
known. This work is therefore aimed at promoting the systematic 1970s (De los Rios and Charnley, 2016; Lieder and Rashid, 2016).
exploration and quantification of different product design, business Design strategies addressing ecological challenges typically fall
models, and post-use strategies, within the context of the under the umbrella of various concepts, including product stew-
manufacturing industry. ardship (Hart, 1995), green design (Fullerton and Wu, 1998), design
This paper is structured as follows: Section 2 provides a brief for the environment (DfE) (Chen, 2001), sustainability-driven
review of the circular economy, circular product design and busi- product design (Byggeth et al., 2007; McLennan, 2004), design for
ness models for circularity. Sections 3 and 4 introduce the system sustainability (DfS) (Arnette et al., 2014), ecological product design
dynamics methodology, describe the steps in the model-building (EPD) (Hartmann and Germain, 2015), eco-design (Deutz et al.,
process, and, most importantly, review in detail the body of liter- 2013), cradle-to-cradle design (Braungart et al., 2007), and regen-
ature used to conceptualize the model structure. Section 5 presents erative design (Lyle, 1996). Most of these concepts emphasize the
the analysis of the simulation results, while Section 6 discusses the need to promote resource efficiency, the restriction of hazardous or
conclusions of this study, its practical and theoretical implications, non-renewable materials in manufacturing, efficiency during the
and finally its main limitations and future research avenues. product's use phase, and component reuse or recycling at the end of
the product's lifetime (Calcott and Walls, 2005; Dyllick and Rost,
2. Literature review 2017; Snir, 2001; Wong et al., 2012). In general, the bundle of
product materials and production technologies chosen during the
2.1. The circular economy (CE) design stage of a product will determine the pollutants and wastes
it will release during its lifetime, the energy it will consume, and
The CE is perceived as representing the business operationali- the ease with which its components will be reutilized in subse-
zation of the much-discussed and sometimes blurry concept of quent uses and manufacturing cycles (Rashid et al., 2013; Tsoulfas
sustainable development, because it offers firms the opportunity to and Pappis, 2006).
achieve product differentiation, competitive advantage, and ulti- Circular design rests on the notion that products must be
mately growth (Ellen MacArthur Foundation, 2013; Kirchherr et al., conceived with multiple life cycles in mind, where material quality
2017; Murray et al., 2017; Saidani et al., 2017). As industries and is to be maintained and waste is to be avoided. That is, a product is
economies face the devastating effects of pollution, global warm- considered circular when it is engineered to be remain in a state
ing, and resource scarcity, legislation in favor of the CE has gained closer to the original product (i.e., product integrity), ideally
increased popularity across the globe. In 1996, Germany integrated “eliminating environmental costs when performing interventions
a semi-circular directive called the “Closed Substance Cycle and to preserve or restore the product's added economic value” (den
Waste Management Act” into its national laws, an action that was Hollander et al., 2017; Lieder and Rashid, 2016). Previous publica-
later followed by Japan's 2002 “Basic Law for Promoting the Crea- tions in connection with circular product design have been mostly
tion of a Recycling Oriented Society” and China's 2009 “Circular conceptual in nature, covering topics such as product design
Economy Promotion Law” (Geissdoerfer et al., 2017; Korhonen frameworks for the CE (Bocken et al., 2016), the difference between
et al., 2018; Lieder and Rashid, 2016). circular and eco-design (den Hollander et al., 2017), product life
Although different definitions have been proposed for the CE extension through design (Bakker et al., 2014), the new set of skills
(Ghisellini et al., 2016; Lieder and Rashid, 2016; Su et al., 2013), the and capabilities required by circular designers (De los Rios and
description used in this paper is based on the concept of material Charnley, 2016), product design and consumer behavior in a CE
M.A. Franco / Journal of Cleaner Production 241 (2019) 118327 3

(Poppelaars et al., 2018), tools for the design of circular products consumption systems in a circular economy. System dynamics is a
(Sinclair et al., 2018; Wastling et al., 2018), and the challenges cir- computer simulation methodology that has its roots in the theory
cular product manufacturers face during the design stage (Franco, of nonlinear dynamics and feedback control developed in mathe-
2017). Later sections will provide a more detailed account of matics, physics, and engineering. It is aimed at enhancing the un-
which circular design strategies were included and how they were derstanding of complex feedback systems while simultaneously
quantified in the simulation model featured in this article. supporting the policy formulation process for decision makers
(Sterman, 2000). SD asserts that the dynamic complexity in a sys-
2.3. Circular business models (CBMs) tem is derived from the presence of various factors, namely: (i)
accumulation (i.e., the accumulation and depletion of resources),
Transitioning to a CE will require not only changes to the way (ii) non-linearities (i.e., an effect is rarely proportioned to its cause),
products are designed, but also radical changes to how they are (iii) feedback structures (i.e., both balancing and reinforcing), and
commercialized and consumed. Circular business models lie at the (iv) time delays (i.e., adding a time delay to a negative feedback
core of the CE (Lewandowski, 2016) and constitute a prerequisite loop creates oscillations, thus reducing a decision maker's ability to
for CE diffusion, because they “enable economically viable ways to control for confounding variables and design-appropriate policies).
continually reuse products and materials” (Bocken et al., 2016). A system dynamics model consists of an interlocking set of
Furthermore, CBM innovation allows firms to go beyond prevalent differential and algebraic equations developed from a broad spec-
sustainable business model configurations that focus only on effi- trum of relevant data and depicted by means of a simulation soft-
ciency, productivity, and greening the supply chain (Bakker et al., ware (Homer and Hirsch, 2006). SD makes use of stocks, flows, and
2014; Geissdoerfer et al., 2018). parameters to model the complex behaviors of different systems.
Due to the early character of the CE and the many in- Stocks or “state variables” represent the accumulation of the in-
terpretations that are currently associated with it, a consolidation flows minus the outflows and are expressed in whole units (e.g.,
and understanding of the concept of circular business models is just widgets, units, people). Changes in stocks can only result from
emerging (Lewandowski, 2016; Nubholz, 2017; Pieroni et al., 2019). “flows” or “rates” (i.e., inflows and outflows), which are always
All in all, CBM is the label used to describe business models in a CE expressed as a function of a unit of time (e.g., widgets/year, units/
that are aimed at achieving resource efficiency by incorporating month, people/day). Hence, stock levels increase when the inflows
strategies that slow and close resource loops (Bocken et al., 2016; exceed the outflows, fall when outflows exceed inflows, and remain
Geissdoerfer et al., 2018; Nubholz, 2017). Because the CE demands in equilibrium when inflows equal outflows. Overall, the stocks will
the lifetime maximization of products and materials, CBMs must be integrate the difference between the inflows and the outflows,
oriented towards designing systems around products that can while the flows will be functions of the stock and other parameters
generate revenue by creating, delivering, capturing, and main- (Sterman, 2000). Parameters are numerical values that remain
taining value over time (Bakker et al., 2014; Boons and Lüdeke- constant throughout the simulation and that affect other variables
Freund, 2013; Osterwalder et al., 2005). Extant publications in and flows in the model.
relation to CBM have conceptually discussed topics such as the A systemic approach to the CE has been proposed by various
barriers and enablers to circular business model implementation authors, including Ghisellini et al. (2016), who conceptualize the CE
(Linder and Williander, 2017; Rizos et al., 2016; Vermunt et al., as “systemic” by highlighting the need to cause simultaneous
2019), the role of CMBs in sustainability and resource efficiency changes at a macro (i.e., city, province, region, nation), meso (i.e.,
(Hofmann, 2019; Manninen et al., 2018; Whalen, 2019) and clas- eco industrial parks), and micro (i.e., single firms) systems-level.
sification frameworks for CBMs (Bocken et al., 2014; Rosa et al., Furthermore, it is widely acknowledged that the circular economy
2019; Urbinati et al., 2017). has the ambition of “optimizing systems rather than components”
As pointed out before, although various typologies for classi- (Ellen MacArthur Foundation, 2013) and as such, requires a tool
fying CBMs have been proposed in the literature (see van that is capable of quantitatively modeling the diverse number of
Renswoude et al. (2015); Moreno et al. (2016); Bakker et al. elements, links, and non-linearities that circular value chains
(2014)), most authors agree that the value proposition of a firm exhibit (Velte and Steinhilper, 2016).
operating in a CE should entail schemes to slow and close resource Given its suitability for the analysis of complex industrial and
loops (Bocken et al., 2016; Nubholz, 2017). On one hand, BMs to environmental systems, system dynamics has already been widely
slow loops encourage long product life and reuse and are classified used in applications related to the CE, including sustainable
according to three main models: (i) access and performance (e.g., development (Meadows et al., 1972), closed-loop supply chains
product-service system (PSS) models including sharing and leas- (Georgiadis and Besiou, 2008; Georgiadis et al., 2006; Golroudbary
ing), (ii) extending product value (e.g., exploitation of the product's and Zahraee, 2015; Spengler and Schro € ter, 2003; Vlachos et al.,
residual value through reuse, refurbishing, and remanufacturing), 2007), recycling and remanufacturing (Wang et al., 2014), sus-
and (iii) classic long life (linked to design for long life and design for tainable business models (Abdelkafi and Ta €uscher, 2016), product-
maintenance). On the other hand, BM for closing loops involve service systems (Lee et al., 2012), and the CE itself (Asif et al., 2016).
capturing value from “waste” or “by-products” through schemes As a methodology, SD is not free of criticism though. The utility of
such as industrial symbiosis. The PSS model (coupled with repair SD is often questioned because of the models’ complexity and the
and maintenance services) as well as the reuse model will be the great amount of quality data they demand. In despite of this, their
two main CBMs conceptualized in this article. Section 4.1.2 elabo- power to provide useful insights and policy recommendations,
rates more on each of these strategies and on how they were even when confronted with limited data, is undisputable. The
incorporated to the model structure. subsections below describe in greater detail the model formulation
process.
3. Methodology
3.2. Model building stages
3.1. System dynamics (SD)
Modeling is an iterative process of scope selection, hypothesis
This work draws on the system dynamics methodology to cap- generation, causal diagramming, quantification, reliability testing,
ture the complex and systemic nature exhibited by production and and scenario analysis (Homer and Hirsch, 2006; Sterman, 2000).
4 M.A. Franco / Journal of Cleaner Production 241 (2019) 118327

The following sections describe these steps and how they were
adapted for this research.

3.2.1. 1st stage: Problem identification


The first stage in the modeling process, problem identification
and definition, is concerned with building a model for the purpose
of understanding or analyzing a specific problem rather than
modeling a whole system (Sterman, 2000). The initial character-
ization of the problem in this paper is given by the research
question introduced in Section 1 and the model assumptions out-
lined in Table 1, which analyzes the dynamic implications stem-
ming from different combinations of circular product design and
business model strategies.

3.2.2. 2nd and 3rd stages: System conceptualization and model


formulation
The second modeling step is concerned with the system
conceptualization or the development of what is called a “dynamic
hypothesis.” The dynamic hypothesis tries to explain the
researched problem's dynamics, and it is embedded in a model's
underlying feedback and stock-and-flow structure. The formal
modeling process started by building and testing small model
structures and then building progressively toward complex ones
until the model was developed completely (Martinez-Moyano and
Richardson, 2013). To formulate and test the model, I used the
proprietary system dynamics modeling software Dynaplan® Smia,
developed by Dynaplan AG. Smia allows modelers to make use of
advanced modeling functions along with easy-to-build model in-
Fig. 1. Overview of the model formulation stage. Adapted from Asif et al. (2016).
terfaces and professional output graphs.
For the different iterations of the model structure, information
for the relevant variables was collected by reviewing prominent
literature on the topics of sustainable and circular product design 4. Description of the conceptual model
and business models, as well as on closed-loop or circular supply
chains. The use of qualitative data from published literature is, in The following subsections describe in the detail the theoretical
fact, one of the most popular techniques for building SD models foundations used for the development of the model structure (see
(Luna-Reyes and Andersen, 2003). The coding software ATLAS. ti Table 2).
was used to systematically record, review and code (by means of
open coding) the different concepts that emerged throughout the
literature review. Codes were then grouped according to their place
in the circular value chain and the most prominent ones were used 4.1. Circular product design
to conceptualize the different model subsectors. The literature re-
view by means of concept coding not only facilitated the identifi- To model product design, I adhered to the taxonomy proposed
cation of key themes and concepts (i.e., stocks, flows, and by Bocken et al. (2016), who suggest categorizing product design
parameters), but also of the relationships among them (see Fig. 1). for a CE according to strategies for slowing loops (S) and strategies
From the beginning of the modeling process, dimensional or for closing loops (C). While the first group includes designing for
unit consistency for all model equations was also ensured longevity and designing for ease of maintenance and repair, the
(Martinez-Moyano and Richardson, 2013). Fig. 2 combines causal- second group incorporates designing for disassembly and reas-
loop and stock-and-flow diagrams to present the structure of the sembly as well as designing for a technological cycle (i.e., recycling)
simulation model. (see Table 3).

Table 1
Overview of model assumptions.

 The time horizon of the simulation is 40 years in order to witness the delayed and indirect effects of potential policies (2017e2057).
 The model is built from the perspective of a single manufacturing firm that sells generic product units.
 The product sales rate determines the number of products that are purchased and are in-use, regardless of the firm's manufacturing capacity. This means that the SD
model structure does not account for flows of product ordering or work-in-progress (i.e., raw material conversion to the final product).
 The model does not account for any dynamics related to product storage and transportation.
 Although certain products exhibit a “baseline” demand in society (i.e., household electronics, bicycles, etc.), the model assumes all product types must be adopted first
by curious consumers by means of the Bass model.
 Customer demand for brown and green product purchases, as well as for customers opting for PSS, is exogenous and deterministic.
 Return and reusability rates are also exogenous.
 The initial number of adopters or customers is 1000 people, with each person buying 1 product unit.
M.A. Franco / Journal of Cleaner Production 241 (2019) 118327 5

Fig. 2. Overview of the SD model structure.

Table 2
Design and business model strategies included in the SD model.

Design strategies For slowing resource flows (S) 1. Design for long-life
2. Design for ease of maintenance and repair
For closing resource flows (C) 1. Design for dis- and reassembly
2. Design for a technological cycle/recycling
Business model strategies For slowing resource flows (S) 1. Access and performance (PSS and maintenance)
2. Extending product value (Reuse)

Table 3
Product design strategies for slowing (S) and closing loops (C) and their corresponding relevant model variables.

Design strategy Quantitative operationalization in the SD model Source Relevant model variables

1. Design for long-life (S) Shape of the distribution using the Bass model (short- and Georgiadis et al.  Adoption fraction
long-life products) (2006)  Contact rate
Sterman (2000)  Product lifetime
 Residence time
 Fraction of products-in-use being replaced
 Fraction of discarded units being replaced
2. Design for ease of maintenance A higher breakdown rate diverts man-hours to solve Sterman (2000)  Breakdown rate
and repair (S) unplanned repair tasks first  Disassembly index
 Man-hours available for repair tasks
 Man-hours available for scheduled maintenance
3. Design for dis- and reassembly (C) Disassembly Index (DI) ranging from 1 to 4 de Aguiar et al. 1: easiest to disassemble
(2017) 4: most difficult to disassemble
4. Design for the technological cycle Material Compatibility Index (MCI) ranging from 1 to 4 de Aguiar et al. 1: parts are made only of one material
(C) (2017) 4: parts and modules are composed of fully
incompatible materials

4.1.1. Design strategies for slowing loops durable products that do not to break down easily (Bocken et al.,
2016). Central to the idea of product longevity is the well-known
4.1.1.1. Design for long-life and the demand distribution. Design for four-stage process of a product's life cycle, consisting of an
durability, longevity, or long life is concerned with developing
6 M.A. Franco / Journal of Cleaner Production 241 (2019) 118327

introduction, growth, maturity, and decline stage. Similar to the SD rate.


formulation proposed by Georgiadis et al. (2006), the concept of After repair, units become available again and the stock of units
short and long product lifecycles is used to emulate products under reactive maintenance is reduced. The remaining number of
designed for short and long lifetimes, respectively. While products man-hours, discounting those being used for repair, are assigned to
purposely built for obsolescence (i.e., short lifecycle) display a scheduled maintenance tasks. Finally, it is also assumed that the
narrow-spread distribution with a high peak, products designed for disassembly index (explained below) affects the number of man-
longer lifetimes display a wider-spread distribution with a lower hours required by planned and reactive maintenance jobs. Thus,
peak (Briano et al., 2010; Georgiadis et al., 2006). In the model, when products are difficult to disassemble, both reactive and
lifetime distributions determine the product demand distribution, scheduled maintenance tasks take longer.
and consequently the shape of the new product adoption flow.
To mirror the distributions for short and long product lifecycles, 4.1.2. Design strategies for closing loops
the model uses the Bass diffusion model in which word of mouth 4.1.2.1. Design for disassembly (DfD) and the disassembly index (DI).
and advertising move customers from a stock of potential adopters Design for disassembly is a key element for transitioning to the CE
to a stock of adopters. The Bass model, however, is often described (Andrews, 2015). A product designed to be easily disassembled
as a first-purchase model, since it does not capture situations brings about various benefits throughout the product's lifecycle,
where the product is consumed and discarded, a situation that including efficiencies in manufacturing and assembly, maintenance
would naturally lead to repeat purchases (Sterman, 2000). As a or servicing, and recovery at end-of-life both through remanu-
result, first-time and repeat purchases are modeled separately. The facturing and recycling (Gu and Sosale, 1999; Vanegas et al., 2017).
purchase rate then consists of the sum of two flows: the repeat Although many analytical models have been developed that eval-
purchase flow and the new product adoption flow. The model uate the degree of a product's disassembly capacity (see, for
formulation for product replacement by means of repeat purchases instance, Desai and Mital (2003)), the model uses the disassembly
is described next. measures proposed by de Aguiar et al. (2017). The authors compile
4.1.1.1.1. Product replacement rate. The product replacement published research to estimate 4 disassembly indexes that range
rate will depend on the product's lifetime and on the residence from 1 to 4, where 1 is an ideal state (i.e., easiest to disassemble),
time. A product's lifetime, or physical time, is the average time and 4, an undesired state (i.e., almost impossible to disassemble).
during which a product is operational before it experiences a The disassembly index score influences three variables in the
functional failure that is beyond recovery (den Hollander et al., model, namely: (i) the recycling delay, (ii) the fraction of recyclable
2017). On the other hand, the residence time, or use time, refers materials in products, and (iii) the man-hours required for sched-
to the amount of time the product is kept by the user, before it uled and reactive maintenance tasks. First, the less a product has
becomes unwanted and is discarded or replaced (Murakami et al., been designed for disassembly, the more the time a recycling party
2010). will need to sort, clean, and process its parts. I assume that when
Past research suggests product replacement can be determined the disassembly index equals 1, the recycling time can be cut by a
by: (i) mechanical, physical, or performance obsolescence (i.e., end maximum of 50% (e.g., see Vanegas et al. (2017), where disassembly
of the product's lifetime), (ii) fashion or technological motives that time is cut by approximately 40% when DfD guidelines are intro-
induce the consumer towards the replacement of well-functioning duced in the design stage). The 50% ceiling assumes that the
products (i.e., products have become outmoded) (Bakker et al., recycling delay is not only affected by disassembly, but also by other
2014; Guiltinan, 2009; van Nes and Cramer, 2005), or (iii) a com- activities.
bination of both wear-and-tear and new desires (van Nes and The disassembly index also affects the fraction of recyclable
Cramer, 2005). The model assumes that more than half of the materials in a product. Hence, when a product has not been
short life cycle products which are in use and are still in good designed for disassembly, component parts cannot be efficiently
condition are replaced due to fashion trends or impulse purchases. separated and reused, thus reducing the product's fraction of
These products stop being used or are discarded, not because they recyclable materials. The same is assumed with the number of
have broken down or are worn out, but because customers perceive man-hours assigned to scheduled and reactive maintenance tasks.
them as being outdated. In contrast, I assume that long life cycle When the disassembly index equals 1 (i.e., the most desirable
products are mostly discarded, and replaced, because they have state), the number of man-hours allotted for unplanned and
broken down (Cox et al., 2013). In contrast to short life cycle scheduled maintenance will remain unchanged, and when it equals
products, the fraction of buyers who replace long-life products 4 (i.e., undesired state), the time for both types of maintenance
prematurely because of fads is relatively low. Note also that some tasks will double.
products may be discarded after a certain period of use, even
though they are still fully functional or in good working order. This 4.1.2.2. Design for the technological cycle and the recyclability index
group of products is represented in the model by the outflow (RI). A product designed for the technological cycle is a product
“discard rate of functional products.” whose technical component parts (e.g., metals, and plastic) can be
recycled into high-quality raw materials to be looped back into the
4.1.1.2. Design for ease of maintenance and repair. Design for ease of industrial system. The degree of material recycling is primarily
maintenance and repair is the second major strategy for slowing affected by the number of materials used in a product and their
resource loops (S). Accordingly, the model displays structures for mutual compatibility. Overall, the fewer the materials in a product,
both reactive (repair) and scheduled (proactive) maintenance. A the easier the recycling, since less separation work is involved
static fraction determines the proportion of product units that (Franco, 2017; van Schaik and Reuter, 2007). This principle has
move from being available to being in the stock of unplanned or proved to be true in products ranging from textiles to common
reactive maintenance. When the ratio of “damaged units to total household electronics.
units” (i.e., the breakdown rate) surpasses a threshold, priority is Although various possible levels of detail can be defined when
given to resolve reactive maintenance tasks. That is, the total conceptualizing the degree of material recycling (see, for example,
number of man-hours available, given the number of available van Schaik and Reuter (2010)), the model uses the Material
mechanics, is split between scheduled and reactive maintenance Compatibility Index (MCI) suggested by de Aguiar et al. (2017). The
jobs, with the latter receiving priority depending on the breakdown authors propose a MCI that ranges from 1 to 4, where 1 refers to a
M.A. Franco / Journal of Cleaner Production 241 (2019) 118327 7

situation in which parts are made only of one material, and 4 refers of a PSS is Rolls Royce “Power by the Hour” scheme, which allows
to a situation in which parts and modules are composed of fully airlines to purchase the use of aircraft engines instead of having to
incompatible materials. To operationalize the MCI in the model, it is buy them (Baines et al., 2007; Ellen MacArthur Foundation, 2016).
assumed that when the MCI equals 1, 100% of the product's mate- Although not evident in the model structure, it can be assumed that
rials can be recycled. Similarly, at an MCI value of 2, 70% of the the manufacturing firm, the PSS provider, and the reverse logistics
materials are recyclable, at a value of 3, 30%, and at a value of 4, only party are all different stakeholders interacting in the same simu-
5% of the materials are recyclable. The MCI or recyclability index lated industrial ecosystem.
partially affects a product's net recyclability fraction and de- Research literature on PSS has investigated some of the factors
termines the number of units flowing from the “collected for that contribute to the low levels of PSS adoption among customers,
recycling” stock to the “recycled product inventory” stock. including: the low level of PSS maturity and lack of enthusiasm, PSS
acquisition costs and risks (Gesing et al., 2014), hygiene and safety
4.1.2.3. Closing the technological cycle through recycling. At end-of- concerns (Catulli, 2012; Catulli and Reed, 2017), the physical as-
lifetime (EOL), only a fraction of all sold products will be collected pects of ownerless consumption (Manzini et al., 2001; Mont, 2002),
for post-processing, with the remaining fraction flowing through and the insecurity of consumers with regard to what they are
the “uncontrollable disposal rate,” whereby products end up in a getting (Rexfelt and Hiort af Orn€
as, 2009). Instead of accounting for
landfill or being incinerated. Upon collection, I assume that some of each adoption factor discretely, the model sets an exogenous frac-
the materials in products will be suitable for recycling according to tion of customers who opt for services instead of acquiring the
the recyclability index described in the previous section, and that physical product. The remaining fraction (1-PSS fraction) is then
the remaining non-recyclable materials will constitute post- split between customers who buy green (recycled) products and
consumer waste. those who buy brown (conventional) products. Note that
The net fraction of recyclable materials in a product is then throughout this paper, green products will refer to sustainable
defined as the percentage of recyclable materials (defined by both goods that contain a fraction of recycled materials, whereas brown
the disassembly and the recyclability indexes) multiplied by a products will refer to conventional products that have been man-
measure, ranging from 0 to 1, that reflects the efficiency of the ufactured for a linear economy. In fields such as marketing and
recycling process. Finally, the flow of recycled products is regulated energy management, these labels have been extensively used to
by a recycling delay that mirrors the time that processing parties distinguish between products that are clean, benign or resourceful
take to sort, clean, disassemble, and process products’ component (i.e., “green”) and polluting or toxic (“brown”) (Champagne and
parts (see Table 3). Matharu, 2016; Garrett-Peltier, 2017; Yenipazarli and Vakharia,
Table 4 displays some of the variables that were used to test 2015).
extreme case scenarios. Note that scenarios 1.1 and 1.2 show the
parameter values of key model variables for short-lifetime prod- 4.2.1.1. Product lifetime and discards under a PSS model. The model
ucts, while scenarios 2.1 and 2.2. Show the parameter values for assumes a product under the PSS model can reach end-of-life either
long-lifetime products. Results from this exercise are presented in because the product's lifetime has ended or because excessive use
Section 5. has shortened it. The first scenario is straightforward and accounts
for the product's lifetime to determine the flow of product discards.
4.2. Circular business models The second scenario accounts for the intensity of use, which is
linked to the decisions taken by the final consumer while using the
4.2.1. Product service systems (PSS) product (Ellen MacArthur Foundation, 2015; Iraldo et al., 2017).
The concept of PSS has now been discussed for more than a Previous research has highlighted the link between user behavior
decade in connection with its role in sustainability (Annarelli et al., and product durability by suggesting that improper use and over-
2016; Beuren et al., 2013; Reim et al., 2015), and more recently in use lead to shorter product lifetimes (Bobba et al., 2016).
relation to its contribution to the CE (Lewandowski, 2016; Lieder The intensity of use is expressed through a measure of “func-
and Rashid, 2016; Linder and Williander, 2017; Tukker, 2015). The tional units,” which reflects the extent to which a product is used to
PSS model shifts the focus from selling products (complemented by its full capacity (e.g., a kilometer for a car or a wash cycle for a
services), to selling services that substitute products, thereby washing machine). I have included this measure, because, accord-
contributing to dematerialization and offering the opportunity to ing to past research, the level of a product's intensity of use depends
decouple economic gain from material consumption. An example on whether the product is purchased or used as a service. Hence,

Table 4
Parameter values for selected extreme-case scenarios.

Key model variable Short-lifetime products Long-lifetime products

Scenario 1.1 Scenario 1.2 Scenario 2.1 Scenario 2.2

Potential adopters 1000 people 1000 people 1000 people 1000 people
Adoption fraction 0.05 0.05 0.02 0.02
Advertising effectiveness 0.01/yr 0.01/yr 0.01/yr 0.01/yr
Contact rate 100/yr 100/yr 60/yr 60/yr
Initial sales per adopter 1 unit/people 1 unit/people 1 unit/people 1 unit/people
Perceived functional risk 0.5 0.5 0.5 0.5
Average residence time 2 yrs 2 yrs 5 yrs 5 yrs
Fraction of products in-use being replaced 0.7 0.7 0.1 0.1
Average product lifespan 3 yrs 3 yrs 6 yrs 6 yrs
Fraction of discarded units being replaced 0.1 0.1 0.7 0.7
Fraction of products collected for recycling 0 0.5 0 0.5
Disassembly index 4 1 4 1
Recyclability index 4 1 4 1
8 M.A. Franco / Journal of Cleaner Production 241 (2019) 118327

customers under an ownership scenario are more careful with the when functional risk reaches its highest value (1), only true green
products, thus leading to less wear and tear. In contrast, when consumers will buy the recycled product. Prior research has esti-
customers utilize but do not own the asset, overuse often occurs mated this last measure to be approximately 10% (Rex and
(Velamuri et al., 2013). Baumann, 2007). In addition to the functional risk, the perceived
Product use is thus modeled as the number of functional units green image factor (GIF) also affects the fraction of customers
accumulated during the product's use phase, divided by the opting for green products. The GIF is a delayed or smoothed vari-
average number of functional units which the product is assumed able modeled as the ratio of the “recycling rate” to the “product
to have according to industry standards. The outflow of products at collection rate for recycling.” This ratio reflects the delayed market
EOL will, again, be the maximum between the flow of products that awareness that a producer cares about remanufacturing or recy-
have been disposed of because their lifetimes have ended (as cling (Vlachos et al., 2007).
dictated by design choices), and the flow of disposal as dictated by
product overuse (as reflected by the functional unit ratio). 4.4. Assessment of selected product groups

4.2.1.2. Product lifetime and scheduled maintenance. In a PSS model, One important question related to this modeling exercise is
service and maintenance tasks are taken over by the manufacturer whether simulation results are different when the model is popu-
or the retailer in order to lengthen the asset's lifetime (Bocken et al., lated with parameter values corresponding to a specific product
2016). Hence, the model assumes that, besides design characteris- group (e.g., household appliances, clothing, cars, etc.) rather than a
tics and patterns of use, the lifetime of a product is also affected by “generic product.” To answer this question, I searched the literature
its level of scheduled or preventive maintenance (e.g., substitution for key parameter values, and in the absence of suitable informa-
of components and cleaning of the product parts) (Bobba et al., tion, assigned certain variables values based on common sense. In
2016). The effect of preventive maintenance is incorporated in the Tables 5 and I have used the label “T&E” (i.e., trendy and emotional)
model as the delayed ratio of the number of product units under to denote products which are fashion-affected (i.e., models are
preventive maintenance divided by the total number of units under more important than functionality), and “M&P” (i.e., mature and
both preventive and reactive maintenance. When this ratio equals pervasive) to refer to mature products, with established design
or is close to 1, a product's average lifetime remains the same. cycles, that are not affected by fashion fads. While the latter co-
However, when the ratio starts falling below 1, because little or no incides with the description of products designed for a long-life, the
preventive maintenance has been performed, so does the value for former matches the definition of short-lived products. Results from
product lifetime. Finally, it is also assumed that regardless of the this exercise are discussed in Section 5.2.
level of scheduled maintenance, product lifetime can never fall
below 50%. 5. Results and discussion

4.2.2. Product reuse 5.1. Extreme-case scenarios


Product reuse is the second CBM portrayed in the model. It is
incorporated as a ratio that diverts a portion of the product Results from the extreme-case scenario test (Table 4) demon-
intended to be discarded to a stock of products that are intended for strate how product longevity affects disposal and replacement rates
second-hand use. This decision rule delays the number of products in such a way that products designed for a short-lifespan exhibit
flowing to recycling or the landfill. I have included a structure for higher and faster disposal and replacement rates than long-lasting
reuse in the model because it is widely acknowledged that many products (Cooper, 2005; Franklin-Johnson et al., 2016). Addition-
products are discarded even though they are still in perfect working ally, these results show the extent to which different lifecycle
order (Bayus, 1991) and because looping products through reuse is lengths and residence times partially determine the shape and
a major CE strategy. number of discarded products as well as the number of products
intended to be remanufactured or recycled. The following sections
4.3. Other relevant model structures present the results of selected model scenarios.

4.3.1. Functional risk and the green image factor 5.2. Short-life products with a poor disassembly index
The fraction of customers opting for green products versus the
fraction of customers opting for brown products is determined by A disassembly index (DI) of 4 in scenario B means that a prod-
the product's degree of functional risk and the green image factor. uct's component parts are extremely difficult to separate, thus
Functional risk refers to the perceived risk that customers attach to complicating service, remanufacturing, and recycling tasks for the
a product, which can be psychological, financial, performance- relevant parties. In scenario B, not only are there fewer products
related, physical, and social. A repaired tire, for instance, may be flowing into recycling (i.e., recycling rate), but there is also evidence
perceived as embodying a high functional risk, in that it poses a of higher amounts of recycling waste as a result of the poor disas-
high risk of accidents, while clothing produced with recycled fibers sembly performance of the products. Additionally, because a DI of 4
is perceived as presenting a low functional risk. Hamzaoui-Essoussi translates into a smaller fraction of recyclable materials contained
and Linton (2010) found that perceived functional risk appears to in products, a much smaller amount of the “products collected for
have a statistically significant impact on consumer purchase de- recycling” becomes converted into the “recycled products in-
cisions, so that when functional risk in recycled products is ventory.” In contrast, top scores for disassembly and recyclability in
perceived as high, customers’ willingness to purchase (WTP) is low, scenario A ensure that the collected products flow smoothly into
and vice versa. the stock of “recycled products inventory.” Although the level of
The measure of functional risk is determined by a lookup green product demand is the same in both scenarios, the “recycled
function of the form Y ¼ f(X) ranging from 0 (low) to 1 (high), such products inventory” stock in scenario A grows significantly more
that at zero functional risk, an estimated maximum of 80% of cus- than it does scenario B. This is because there are many more units
tomers will choose recycled products (this is because even at zero that are in a suitable condition to be recycled in scenario A, and thus
perceived risk there will still be some customers who will prefer to the inventory accumulates more and does not deplete quickly.
buy new products) (Michaud and Llerena, 2011). On the other hand, Overall, in the case of short-life products, a combination of
M.A. Franco / Journal of Cleaner Production 241 (2019) 118327 9

Table 5
Decision rules and parameter values for selected product groups.

Product types Household appliances Industrial products (e.g., Cars IT personal products (e.g., Clothing Furniture
(e.g., refrigerators, forklift trucks, soil small IT accessories, (e.g.,
washing machines) compactors, excavators) mobile phones) pushchairs)

Lifecycle as dictated by current design Long [a] Long [b] Long [c] Short [a] Short [d] Short [e]
Product type M&P M&P M&P T&E [f] T&E [d] T&E [g]
Suitable for remanufacturing (RM) or recycling RM RM RM RM/RC RC RM/RC
(RC)
Suitable for PSS Yes Yes [b][h] Yes No No [i] No [g]
Average residence time in years 9 [author] 5 [j] 8 [author] 1.5 [j] 0.5 [author] 2 [author]
Median lifespan in years 11 [k] 10 [j] 10 [c] 4 [f][k] 2 [f] 5 [e]
Willingness to purchase (WTP) recycled/ Low (high functional High (high functional risk High (low Low (high functional risk) Low (hygiene High
refurbished products as determined by risk) [l] but familiarity) [l] functional [m] and safety (functional
functional (safety) or hygiene risk risk) [m] risks) [n] risk) [o]

Sources: [a] Huisman et al. (2012), [b] Sundin et al. (2009), [c] Kagawa et al. (2006), [d] Birtwistle and Moore (2007), [e] Costa et al. (2015), [f] Cox et al. (2013), [g] Besch (2005),
[h] Dongmin et al. (2012), [i] Armstrong et al. (2015), [j] Georgiadis et al. (2006), [k] Wang et al. (2013) found in Bakker et al. (2014), [l] Hamzaoui-Essoussi and Linton (2014),
[m] Hamzaoui-Essoussi and Linton (2010), [n] Catulli (2012), [o] Catulli and Reed (2017).

conscious design for recycling and disassembly produces a higher scores for disassembly and recyclability, a poor DI in scenario B
amount of recycled product units at an earlier time. It is important reduces and delays the number of product units flowing into the
to highlight that the conclusions outlined in this section do not recycled products inventory. Similar to the test in the previous
significantly change when the scenario that is opposite to scenario section, there is no significant difference between long-life prod-
B is tested (i.e., short-life products designed with a good disas- ucts that are designed with a poor disassembly index and with a
sembly index and with a poor recyclability index). good recyclability index and the scenario where the opposite holds
(see Fig. 5).
5.3. Long-life products with a poor disassembly index
5.4. Short-life products with average indexes for disassembly and
Compared to short-life products, the stock of “products recyclability
collected for recycling” for long-life products in scenario B (Fig. 4)
exhibits a more constant and steady shape than scenario B in Fig. 3. In this test, I compare two scenarios with different disassembly
The number of products waiting to be recycled is both lower and and recyclability scores under the assumption that neither is
more constant in scenario B of this test, not only because products optimal. Although for both scenarios the product collection rate is
are designed for a longer life, but also because replacement pur- the same, scenario A accumulates fewer units in the “recycled
chases caused by technical failure occur primarily at the products’ products inventory” stock when compared to scenario B (see Fig. 5).
end-of-life (see second hump in the “products collected for recy- In contrast, an average index of 2 for both disassembly and recy-
cling” stock). Also, compared to scenario A, where products flow clability indexes in scenario B allows for a greater amount of
directly to the recycled products inventory because of excellent product materials to be recovered and therefore, enables more

Fig. 3. Short-life products with an optimal and a poor disassembly index.


10 M.A. Franco / Journal of Cleaner Production 241 (2019) 118327

Fig. 4. Long-life products with an optimal and a poor disassembly index.

Fig. 5. Short-life products with average indexes for disassembly and recyclability.

products to be recycled and pile up in the recycled products in- 5.5. Variation in the functional risk
ventory (the recycled product inventory stock grows considerably
because demand cannot keep up with the raising inflow of prod- After a product is manufactured, it must flow within the con-
ucts). This test showed that better results are achieved, in terms of sumer system during a certain amount of time before users dispose
the recycled product inventory, when indexes for both disassembly of it and recycling parties can recover and recycle it. Overall, tests
and recyclability are only “good enough.” Model runs involving the involving changes in the degree of functional risk indicated that
same comparison for long-life products yielded the similar results producers must carefully plan the alignment of: (i) production
(see Fig. 6). Finally, in both scenarios almost the same amount of volumes with the desired inventories of recycled products, and (ii)
waste (resulting from non-recyclable materials) was generated. the forecasted inventories of recycled products with the forecasted
demand for such products.
M.A. Franco / Journal of Cleaner Production 241 (2019) 118327 11

Fig. 6. Long-life products with average indexes for disassembly and recyclability.

In scenarios B and D (see Fig. 7), when the degree of functional parties who rely on substantial volumes of product returns to
risk is low, and consequently the demand for green products is sustain their continuity in a CE.
higher than the demand for brown products, the scant number of
units in the recycled product inventory stock at the beginning of the 5.7. The PSS model versus the purchase-only model: a quick test
simulation generates low sales levels for both green and brown
products (the sales rate for green products is set to be the minimum To elucidate some of the differences between a PSS and a
between the demand and supply rates). With low sales levels for purchase-only model, two simple model structures were built.
both product categories, fewer than the desired number of product Results from this test showed that more units need to be purchased
units are sold, used, collected, recycled, and subsequently stored in at time zero under the PSS model, than is necessary under the
the inventory of recycled units. That is, an initially low inventory of product purchase scenario, to satisfy the same level of hypothetical
recycled goods combined with high levels of green demand leads to customer demand. This is so because under a PSS model each unit is
a reinforcing feedback loop, whereby initial low inventories of kept by the user for a certain amount of time, and, only after this
green products cause sales of both green and brown goods to be time has elapsed will the product unit be returned to the stock of
low, in turn leading to low future levels of collected, recycled, and, “PSS units available.” Also, although the initial investment required
again, stored recycled goods. by a PSS model is much higher for the sponsoring firm, at EOL the
In contrast, when the demand for brown products is higher than number of discarded units (regardless of whether they are land-
the demand for green products, as is the case in scenarios A and D filled, remanufactured or recycled) proved to be the same under
(because of a high functional risk), the stocks for sold, disposed of, both the purchase-only model and the PSS model. Finally, because
and recycled products successfully build up. Paradoxically in this the difference between these two models resides in how many
case, the recycled products inventory can cope with the demand for units are in operation throughout the product's useful life (with
green products. Furthermore, it is only at this point where a vari- their corresponding environmental impact), it can be concluded
ation in the functional risk, like the one introduced in scenarios B that PS is not a panacea unless product units under this model are
and D, can generate the desired effects. also designed with circular design guidelines in mind and their use
is regulated somehow (see Fig. 7).
5.6. Variation in the PSS fraction for short- and long-life products
5.8. Testing the model for different product types
With the introduction of PSS demand, fewer products are pur-
chased and go directly into use and disposal flows. That is, products Fig. 8 shows that the simulated behavior for the different model
which might have been used by only one customer can now be used variables is strikingly similar among products in the long-life
by many. Results of this test show that, compared to short-life category (see red, blue and black curves), as well as those in the
products, long-life products exhibit a lower stock of recycled short-life category (see green, pink, and orange curves). This is so,
products. Contrary to the rather logical belief that a combination of because residence times and lifetimes for all items in each of the
long-life design (i.e., the first of the two design strategies for product groups coincide to a great extent. Products designed for a
slowing loops) and a PSS model (i.e., the most common demateri- short-life are landfilled, collected, and recycled in bigger amounts,
alization strategy) would be ideal in a CE, results show that for compared to the products in the long-life category. Also, given that
long-life products, product returns are fewer in quantity because all long-life products in Table 5 are suitable for PSS, a fixed number
they are distributed over a longer time period. These findings are of these products are distributed among the purchases of brown
relevant to manufacturers, as well as to collection and processing and green products, as well as among products under a PSS model.
12 M.A. Franco / Journal of Cleaner Production 241 (2019) 118327

Fig. 7. Average scores for disassembly and recyclability with variations in the functional risk for short- and long-life products.

This, in turn, influences the number of product units accumulated short-lived products, compared to long-lived ones, generated a
in the stock of “products collected for recycling,” which, as previ- higher amount of disposed of and collected product units earlier in
ously stated, is smaller for long-life than for short-life products. time, not only because their lifetimes were shorter, but also because
Last, the results from this test evidence that even though it is their replacement times were faster. Here, the readiness of
critical to look at different product groups separately to study their manufacturing and processing parties to deal with product returns,
behavior in a CE over time, the decision to use a “generic product” and the preparedness of customers in different markets to buy
for this simulation model is justified. them, will have to be assessed. Also, tests measuring whether it is
better to design a product that excels in one of the design frame-
works (e.g., DI ¼ 1) and not in the other (e.g., RI ¼ 4), or to design a
6. Conclusions
product with good-enough indexes for both disassembly and
recyclability (e.g., DI ¼ 2 and RI ¼ 2), revealed the second scenario
Important conclusions from this exercise are plentiful, both
produced better outcomes since the inventory of recycled goods
theoretically and practically. First, comparisons between scenarios
increased by a higher amount. Naturally, it did so earlier for the
covering design for longevity (i.e., short and long life) showed that
M.A. Franco / Journal of Cleaner Production 241 (2019) 118327 13

Fig. 8. Model output variables for different product groups.

short-lived products than for its long-lived counterparts. This regenerated and resold, despite the high green product demand.
conclusion speaks particularly to firms wanting to redesign their In contrast, when the availability of green products is ensured
product offering according to CE principles or to start-ups wanting beforehand (e.g., through government incentives, for instance),
to achieve the best circular results from the outset. the existing high demand for green products can be satisfied,
Tests examining variations in the functional risk also produced enough products can accumulate for reprocessing, and in-
somewhat counterintuitive results, especially when considering ventories of green products can pile up (see similar results in
the units that accumulate in the inventory of recycled goods. Hence, Franco (2017)).
when a low value is given to the functional risk and therefore green Model results also showed than when the PSS model is intro-
products are preferred over brown products, one would expect duced to the simulation, there is a need for firms to align their
sales and inventories of recycled green products to soar. Counter- future supply and demand recycling flows. This is because, due to
intuitively, CE implementation is sabotaged. One must remember longer and more intermittent use patterns, PSS units accumulating
that the simulation starts with a low inventory of recycled goods (at in the recycled products inventory are distributed over a longer
time 0 the first product purchases are being made and conse- period of time. Although it is believed that a combination of a PSS
quently, there is nothing yet to regenerate), and that the flow of model and products designed for a long life is highly desirable in a
green product sales is set as the minimum between green product CE, it is important that either recycling or remanufacturing firms
supply and green product demand. Although the demand for re- consider the flow of products to be regenerated under a PSS model
generated products in this scenario is high (due to a low functional for their planning activities. Finally, this model building exercise
risk), only a few products can be offered for purchase and, conse- evidenced the extent to which many contextual conditions beyond
quently the number of product units to regenerate and sell back to the manufacturers’ control (e.g., product purchase and replace-
the market is low. ment behaviors, product reuse and return rates, volume, timing,
This situation can be seen as a reinforcing feedback loop of and quantity of returns, as well as unexpected or postponed
scarcity such that in the presence of an initially low production product disposal) need to be satisfied for a circular model to
and sales level of green products, only a few units (designed and properly function and be accurately assessed.
manufactured with sustainable guidelines in mind) can be
14 M.A. Franco / Journal of Cleaner Production 241 (2019) 118327

6.1. Practical and theoretical implications and demand profiles for different business models, or environ-
mental (e.g., CO2 production), organizational (e.g., marketing ini-
Because there are no official indicators, methods, or tools tiatives and policies), and economic variables (e.g., costs of
available for supporting and measuring businesses’ transition from recovered materials and investments for reverse logistics and re-
a linear to a circular industrial system (Ellen Mac Arthur covery infrastructure) could, as examples, increase the utility of the
Foundation, 2015), this system dynamics model might be particu- model. One could also think about endogenizing the model struc-
larly useful to manufacturers, service providers, and collection ture for manufacturing so that, in some cases, primary production is
parties seeking to test or evaluate the performance of CE strategies decreased (with its associated costs) when secondary/recycled
and the viability of self- or third-party imposed collection targets. products are considered suitable replacements for primary/virgin
Each of these players could use the model to assess how different goods. Finally, to ease the use of this tool for managers and other
assumptions in different parts of the value chain affect CE imple- relevant parties, the development of a user interface is
mentation at the macro and micro level (Kalmykova et al., 2018). recommended.
These findings also support policy making and academic
research on EPR (extended producer responsibility), because to Acknowledgements
make EOL management more efficient, legislation and industry
should take into consideration the delayed effects and complex I thank all the reviewers who, throughout the various rounds of
dynamics of combined strategies for design, business models, and revision, contributed to make this manuscript more accurate and
collection schemes. Future model versions could also help insightful.
governmental and non-governmental institutions test the wider
network economic effects of a broad range of CE interventions at Appendix A. Supplementary data
the firm or micro level. For academics, the development of the
proposed model fills the gap in the literature of a systemic, inte- Supplementary data to this article can be found online at
grated, interdisciplinary, and visual tool to support the under- https://doi.org/10.1016/j.jclepro.2019.118327.
standing and performance management of circular production
systems, from product design and manufacturing up business References
models and product return and reprocessing.
Finally, this modeling exercise reveals the paradox of the circular Abdelkafi, N., Ta €uscher, K., 2016. Business models for sustainability from a system
dynamics perspective. Organ. Environ. 29 (1), 74e96.
economy, which seems to require ever increasing volumes of pro-
Ameli, M., Mansour, S., Ahmadi-Javid, A., 2019. A simulation-optimization model for
duction and consumption to make the closing of material loops sustainable product design and efficient end-of-life management based on in-
economically profitable for recycling parties. In fact, in the presence dividual producer responsibility. Resour. Conserv. Recycl. 140, 246e258.
of no link between end-of-life and primary production flows, a CE Andrews, D., 2015. The circular economy, design thinking and education for sus-
tainability. Local Econ. 30 (3), 305e315.
could encourage even more production, partially offsetting the Annarelli, A., Battistella, C., Nonino, F., 2016. Product service system: a conceptual
resource efficiencies gained in the first place (especially when framework from a systematic review. J. Clean. Prod. 139, 1011e1032.
secondary goods are not equivalent replacements for primary Armstrong, C.M., Niinim€ aki, K., Kujala, S., Karell, E., Lang, C., 2015. Sustainable
product-service systems for clothing: exploring consumer perceptions of con-
goods). sumption alternatives in Finland. J. Clean. Prod. 97, 30e39.
Arnette, A.N., Brewer, B.L., Choal, T., 2014. Design for sustainability (DFS): the
6.2. Limitations and future research intersection of supply chain and environment. J. Clean. Prod. 83, 374e390.
Asif, F.M.A., Lieder, M., Rashid, A., 2016. Multi-method simulation based tool to
evaluate economic and environmental performance of circular product sys-
Populating model parameters with data derived from literary tems. J. Clean. Prod. 139 (Suppl. C), 1261e1281.
sources, rather than a real case study, is arguably the most salient Baines, T., Lightfoot, H.W., Evans, S., Neely, A., Greenough, R., Peppard, J., Roy, R.,
Shehab, E., Braganza, A., Tiwari, A., Alcock, J.R., Angus, J.P., Bastl, M., Cousens, A.,
limitation of this work. There are some reasons why data was not
Irving, P., Johnson, M., Kingston, J., Lockett, H., Martinez, V., Michele, P.,
readily available or difficult to obtain (Ameli et al., 2019; Asif et al., Tranfield, D., Walton, I.M., Wilson, H., 2007. State-of-the-art in product-service
2016; Lieder et al., 2017). First, while some firms have focused systems. Proc. Inst. Mech. Eng. B J. Eng. Manuf. 221 (10), 1543e1552.
Bakker, C.A., Wang, F., Huisman, J., den Hollander, M., 2014. Products that go round:
either on circular design, business models, or on product take back
exploring product life extension through design. J. Clean. Prod. 69, 10e16.
and reprocessing schemes, hardly any firm has implemented all Bayus, B.L., 1991. The consumer durable replacement buyer. J. Mark. 55 (1), 42e51.
three sets of strategies simultaneously (see, for instance, the case Besch, K., 2005. Product-service systems for office furniture: barriers and oppor-
studies presented by Franco (2017)). Second, long time delays in tunities on the European market. J. Clean. Prod. 13 (10), 1083e1094.
Beuren, F.H., Gomes Ferreira, M.G., Cauchick Miguel, P.A., 2013. Product-service
product use and returns have prevented those scant number of systems: a literature review on integrated products and services. J. Clean. Prod.
firms that have incorporated any sort of circular strategy to be 47, 222e231.
endowed with detailed, reliable and longitudinal data on the Birtwistle, G., Moore, C.M., 2007. Fashion clothing e where does it all end up? Int. J.
Retail Distrib. Manag. 35 (3), 210e216.
quantity and quality of material flows in their circular value chains Bobba, S., Ardente, F., Mathieux, F., 2016. Environmental and economic assessment
(Gupta et al., 2018; Kirchherr et al., 2018; Whalen et al., 2018). of durability of energy-using products: method and application to a case-study
Because of all this, retrieving fragmented data from previous pub- vacuum cleaner. J. Clean. Prod. 137 (Suppl. C), 762e776.
Bocken, N., de Pauw, I., Bakker, C., van der Grinten, B., 2016. Product design and
lications became imperative for this study. Furthermore, although business model strategies for a circular economy. J. Ind. Prod. Eng. 33 (5),
an attempt to test the model structure for different product groups 308e320.
was made in Section 5.2, future research could assess the applica- Bocken, N., Short, S.W., Rana, P., Evans, S., 2014. A literature and practice review to
develop sustainable business model archetypes. J. Clean. Prod. 65, 42e56.
bility and reliability of this generic tool for a real firm. By param- Boons, F., Lüdeke-Freund, F., 2013. Business models for sustainable innovation:
eterizing the model to fit a specific industrial case, one could check state-of-the-art and steps towards a research agenda. J. Clean. Prod. 45, 9e19.
whether the conclusions from this modeling exercise vary and Bovea, M.D., Pe rez-Belis, V., 2012. A taxonomy of ecodesign tools for integrating
environmental requirements into the product design process. J. Clean. Prod. 20
whether recommendations for the future would be different.
(1), 61e71.
Additionally, the present SD model is not meant to be final and Braungart, M., McDonough, W., Bollinger, A., 2007. Cradle-to-cradle design: creating
comprehensive. Instead, it must be developed further to grasp all healthy emissions e a strategy for eco-effective product and system design.
the missing details that could be relevant to manufacturers, service J. Clean. Prod. 15 (13e14), 1337e1348.
Briano, E., Caballini, C., Giribone, P., Revetria, R., 2010. Using system dynamics for
providers, refurbishing, remanufacturing or recycling parties, and short life cycle supply chains evaluation. In: Proceedings of the 2010 Winter
policy makers. Modeling varying product adoption mechanisms Simulation Conference, pp. 1820e1832.
M.A. Franco / Journal of Cleaner Production 241 (2019) 118327 15

Byggeth, S., Broman, G., Robe rt, K.-H., 2007. A method for sustainable product Hamzaoui-Essoussi, L., Linton, J.D., 2010. New or recycled products: how much are
development based on a modular system of guiding questions. J. Clean. Prod. 15 consumers willing to pay? J. Consum. Mark. 27 (5), 458e468.
(1), 1e11. Hamzaoui-Essoussi, L., Linton, J.D., 2014. Offering branded remanufactured/recycled
Calcott, P., Walls, M., 2005. Waste, recycling, and “Design for Environment”: Roles products: at what price? Journal of Remanufacturing 4 (1), 9.
for markets and policy instruments. Resour. Energy Econ. 27 (4), 287e305. Hart, S.L., 1995. A natural-resource-based view of the firm. Acad. Manag. Rev. 20 (4),
Catulli, M., 2012. What uncertainty?: further insight into why consumers might be 986e1014.
distrustful of product service systems. J. Manuf. Technol. Manag. 23 (6), Hartmann, J., Germain, R., 2015. Understanding the relationships of integration
780e793. capabilities, ecological product design, and manufacturing performance.
Catulli, M., Reed, N., 2017. A personal construct psychology based investigation into J. Clean. Prod. 92, 196e205.
a product service system for renting pushchairs to consumers. Bus. Strateg. Hofmann, F., 2019. Circular business models: business approach as driver or
Environ. 26 (5), 656e671. obstructer of sustainability transitions? J. Clean. Prod. 224, 361e374.
Champagne, P., Matharu, A., 2016. Brown to green and sustainable chemistry. Homer, J.B., Hirsch, G.B., 2006. System dynamics modeling for public health:
Current Opinion in Green and Sustainable Chemistry 2, iiieiv. background and opportunities. Am. J. Public Health 96 (3), 452e458.
Chen, C., 2001. Design for the environment: a quality-based model for green Huisman, J., Maesen, M.v.d., Eijsbouts, R.J.J., Wang, F., Balde , C.P., Wielenga, C.A.,
product development. Manag. Sci. 47 (2), 250e263. 2012. The Dutch WEEE Flows. United Nations University, Bonn.
Cooper, T., 2005. Slower consumption reflections on product life spans and the Iraldo, F., Facheris, C., Nucci, B., 2017. Is product durability better for environment
“throwaway society”. J. Ind. Ecol. 9 (1e2), 51e67. and for economic efficiency? A comparative assessment applying LCA and LCC
Costa, F., Prendeville, S., Beverley, K., Teso, G., Brooker, C., 2015. Sustainable product- to two energy-intensive products. J. Clean. Prod. 140 (3), 1353e1364.
service systems for an office furniture manufacturer: how insights from a pilot Kagawa, S., Tasaki, T., Moriguchi, Y., 2006. The environmental and economic con-
study can inform PSS design. Procedia CIRP 30 (Suppl. C), 66e71. sequences of product lifetime extension: empirical analysis for automobile use.
Cox, J., Griffith, S., Giorgi, S., King, G., 2013. Consumer understanding of product Ecol. Econ. 58 (1), 108e118.
lifetimes. Resour. Conserv. Recycl. 79 (Suppl. C), 21e29. Kalmykova, Y., Sadagopan, M., Rosado, L., 2018. Circular economy e from review of
Davis, J.P., Eisenhardt, K.M., Bingham, C.B., 2007. Developing theory through theories and practices to development of implementation tools. Resour. Con-
simulation methods. Acad. Manag. Rev. 32 (2), 480e499. serv. Recycl. 135, 190e201.
de Aguiar, J., de Oliveira, L., da Silva, J.O., Bond, D., Scalice, R.K., Becker, D., 2017. Kirchherr, J., Piscicelli, L., Bour, R., Kostense-Smit, E., Muller, J., Huibrechtse-
A design tool to diagnose product recyclability during product design phase. Truijens, A., Hekkert, M., 2018. Barriers to the circular economy: evidence from
J. Clean. Prod. 141, 219e229. the European Union (EU). Ecol. Econ. 150, 264e272.
De los Rios, I.C., Charnley, F.J.S., 2016. Skills and capabilities for a sustainable and Kirchherr, J., Reike, D., Hekkert, M., 2017. Conceptualizing the circular economy: an
circular economy: the changing role of design. J. Clean. Prod. 160, 109e122. analysis of 114 definitions. Resour. Conserv. Recycl. 127 (Suppl. C), 221e232.
den Hollander, M.C., Bakker, C.A., Hultink, E.J., 2017. Product design in a circular Korhonen, J., Honkasalo, A., Seppa €la
€, J., 2018. Circular economy: the concept and its
economy: development of a typology of key concepts and terms. J. Ind. Ecol. 21 limitations. Ecol. Econ. 143, 37e46.
(3), 517e525. Law, A.M., Kelton, W.D., 2000. Simulation Modeling and Analysis. McGraw-Hill,
Desai, A., Mital, A., 2003. Evaluation of disassemblability to enable design for New York.
disassembly in mass production. Int. J. Ind. Ergon. 32 (4), 265e281. Lee, S., Geum, Y., Lee, H., Park, Y., 2012. Dynamic and multidimensional measure-
Deutz, P., McGuire, M., Neighbour, G., 2013. Eco-design practice in the context of a ment of product-service system (PSS) sustainability: a triple bottom line (TBL)-
structured design process: an interdisciplinary empirical study of UK manu- based system dynamics approach. J. Clean. Prod. 32, 173e182.
facturers. J. Clean. Prod. 39, 117e128. Lewandowski, M., 2016. Designing the business models for circular economydto-
Dongmin, Z., Dachao, H., Yuchun, X., Hong, Z., 2012. A framework for design wards the conceptual framework. Sustainability 8 (1).
knowledge management and reuse for Product-Service Systems in construction Lieder, M., Asif, F.M.A., Rashid, A., 2017. Towards Circular Economy implementation:
machinery industry. Comput. Ind. 63 (4), 328e337. an agent-based simulation approach for business model changes. Aut. Agents
Dyllick, T., Rost, Z., 2017. Towards true product sustainability. J. Clean. Prod. 162 Multi-Agent Syst. 31 (6), 1377e1402.
(Suppl. C), 346e360. Lieder, M., Rashid, A., 2016. Towards circular economy implementation: a
Ellen MacArthur Foundation, 2013. Towards the Circular Economy: Economic and comprehensive review in context of manufacturing industry. J. Clean. Prod. 115,
Business Rationale for an Accelerated Transition. Cowes, UK. 36e51.
Ellen MacArthur Foundation, 2015. Circularity IndicatorsdAn Approach to Measure Linder, M., Williander, M., 2017. Circular business model innovation: inherent un-
Circularity. Methodology & Project Overview. Cowes, UK. certainties. Bus. Strateg. Environ. 26 (2), 182e196.
Ellen MacArthur Foundation, 2016. Intelligent Assets: Unlocking the Circular Luna-Reyes, L.F., Andersen, D.L., 2003. Collecting and analyzing qualitative data for
Economy Potential. Ellen MacArthur Foundation. system dynamics: methods and models. Syst. Dyn. Rev. 19, 271e296. https://
Franco, M.A., 2017. Circular economy at the micro level: a dynamic view of in- doi.org/10.1002/sdr.280.
cumbents' struggles and challenges in the textile industry. J. Clean. Prod. 168, Lyle, J.T., 1996. Regenerative Design for Sustainable Development. John Wiley &
833e845. Sons, New York.
Franklin-Johnson, E., Figge, F., Canning, L., 2016. Resource duration as a managerial Manninen, K., Koskela, S., Antikainen, R., Bocken, N., Dahlbo, H., Aminoff, A., 2018.
indicator for Circular Economy performance. J. Clean. Prod. 133, 589e598. Do circular economy business models capture intended environmental value
Fullerton, D., Wu, W., 1998. Policies for green design. J. Environ. Econ. Manag. 36 (2), propositions? J. Clean. Prod. 171 (Suppl. C), 413e422.
131e148. Manzini, E., Vezzoli, C., Clark, G., 2001. Product service systems: using an existing
Garrett-Peltier, H., 2017. Green versus brown: comparing the employment impacts concept as a new approach to sustainability. J. Des. Res. 1 (2), 12e18.
of energy efficiency, renewable energy, and fossil fuels using an input-output Martinez-Moyano, I.J., Richardson, G.P., 2013. Best practices in system dynamics
model. Econ. Modell. 61, 439e447. modeling. Syst. Dyn. Rev. 29 (2), 102e123.
Geissdoerfer, M., Morioka, S.N., de Carvalho, M.M., Evans, S., 2018. Business models McDonough, W., Braungart, M., 2002. Design for the triple top line: new tools for
and supply chains for the circular economy. J. Clean. Prod. 190, 712e721. sustainable commerce. Corp. Environ. Strat. 9 (3), 251e258.
Geissdoerfer, M., Savaget, P., Bocken, N., Hultink, E.J., 2017. The Circular Economy e a McLennan, J.F., 2004. The Philosophy of Sustainable Design: the Future of Archi-
new sustainability paradigm? J. Clean. Prod. 143, 757e768. tecture. Ecotone.
Georgiadis, P., Besiou, M., 2008. Sustainability in electrical and electronic equip- Meadows, D.H., Rome, C.o., Associates, P., 1972. The Limits to Growth: a Report for
ment closed-loop supply chains: a System Dynamics approach. J. Clean. Prod. 16 the Club of Rome's Project on the Predicament of Mankind. Universe Books,
(15), 1665e1678. New York.
Georgiadis, P., Vlachos, D., Tagaras, G., 2006. The impact of product lifecycle on Michaud, C., Llerena, D., 2011. Green consumer behaviour: an experimental analysis
capacity planning of closed-loop supply chains with remanufacturing. Prod. of willingness to pay for remanufactured products. Bus. Strateg. Environ. 20 (6),
Oper. Manag. 15 (4), 514e527. 408e420.
Gesing, J., Maiwald, K., Wieseke, J., Sturm, R., 2014. Are IPS2 always a solution? Mont, O.K., 2002. Clarifying the concept of producteservice system. J. Clean. Prod.
Obstacles towards buying industrial product service systems. Procedia CIRP 16, 10 (3), 237e245.
265e270. Moreno, M., De los Rios, C., Rowe, Z., Charnley, F., 2016. A conceptual framework for
Ghisellini, P., Cialani, C., Ulgiati, S., 2016. A review on circular economy: the ex- circular design. Sustainability 8 (9), 937.
pected transition to a balanced interplay of environmental and economic sys- Murakami, S., Oguchi, M., Tasaki, T., Daigo, I., Hashimoto, S., 2010. Lifespan of
tems. J. Clean. Prod. 114, 11e32. commodities, Part I. J. Ind. Ecol. 14 (4), 598e612.
Golroudbary, S.R., Zahraee, S.M., 2015. System dynamics model for optimizing the Murray, A., Skene, K., Haynes, K., 2017. The circular economy: an interdisciplinary
recycling and collection of waste material in a closed-loop supply chain. Simul. exploration of the concept and application in a global context. J. Bus. Ethics 140
Model. Pract. Theory 53, 88e102. (3), 369e380.
Gu, P., Sosale, S., 1999. Product modularization for life cycle engineering. Robot Cim- Nußholz, J.L.K., 2017. Circular business models: defining a concept and framing an
Int Manuf 15 (5), 387e401. emerging research field. Sustainability 9 (10), 1810.
Guiltinan, J., 2009. Creative destruction and destructive creations: environmental Osterwalder, A., Pigneur, Y., Tucci, C., 2005. Clarifying business models: origins,
ethics and planned obsolescence. J. Bus. Ethics 89 (1), 19e28. Present, and future of the concept. Commun. Assoc. Inf. Syst. 15.
Gupta, S., Chen, H., Hazen, B.T., Kaur, S., Santiban ~ ez Gonzalez, E.D.R., 2019. Circular Pieroni, M.P.P., McAloone, T.C., Pigosso, D.C.A., 2019. Business model innovation for
economy and big data analytics: a stakeholder perspective. Technol. Forecast. circular economy and sustainability: a review of approaches. J. Clean. Prod. 215,
Soc. Chang. 144, 466e474. ISSN 0040-1625. https://doi.org/10.1016/j.techfore. 198e216.
2018.06.030. Poppelaars, F., Bakker, C., Van Engelen, J., 2018. Does access trump ownership?
16 M.A. Franco / Journal of Cleaner Production 241 (2019) 118327

Exploring consumer acceptance of access-based consumption in the case of economy business models. J. Clean. Prod. 168 (Suppl. C), 487e498.
smartphones. Sustainability 10 (7), 2133. van Dijk, S., Tenpierik, M., van den Dobbelsteen, A., 2014. Continuing the building's
Rashid, A., Asif, F.M.A., Krajnik, P., Nicolescu, C.M., 2013. Resource Conservative cycles: a literature review and analysis of current systems theories in com-
Manufacturing: an essential change in business and technology paradigm for parison with the theory of Cradle to Cradle. Resour. Conserv. Recycl. 82, 21e34.
sustainable manufacturing. J. Clean. Prod. 57, 166e177. van Nes, N., Cramer, J., 2005. Influencing product lifetime through product design.

Reim, W., Parida, V., Ortqvist, D., 2015. ProducteService Systems (PSS) business Bus. Strateg. Environ. 14 (5), 286e299.
models and tactics e a systematic literature review. J. Clean. Prod. 97, 61e75. van Renswoude, K., ten Wolde, A., Joustra, D.J., 2015. Circular Business Models e
Rex, E., Baumann, H., 2007. Beyond ecolabels: what green marketing can learn from Part 1: an Introduction to IMSA's Circular Business Model Scan. IMSA
conventional marketing. J. Clean. Prod. 15 (6), 567e576. Amsterdam.
Rexfelt, O., Hiort af Orn€as, V., 2009. Consumer acceptance of product-service sys- van Schaik, A., Reuter, M.A., 2007. The use of fuzzy rule models to link automotive
tems: designing for relative advantages and uncertainty reductions. J. Manuf. design to recycling rate calculation. Miner. Eng. 20 (9), 875e890.
Technol. Manag. 20 (5), 674e699. van Schaik, A., Reuter, M.A., 2010. Dynamic modelling of E-waste recycling system
Rizos, V., Behrens, A., van der Gaast, W., Hofman, E., Ioannou, A., Kafyeke, T., performance based on product design. Miner. Eng. 23 (3), 192e210.
Flamos, A., Rinaldi, R., Papadelis, S., Hirschnitz-Garbers, M., Topi, C., 2016. Vanegas, P., Peeters, J.R., Cattrysse, D., Tecchio, P., Ardente, F., Mathieux, F.,
Implementation of circular economy business models by small and medium- Dewulf, W., Duflou, J.R., 2017. Ease of disassembly of products to support cir-
sized enterprises (SMEs): barriers and enablers. Sustainability 8 (11), 1212. cular economy strategies. Resour. Conserv. Recycl. 135, 323e334.
Rosa, P., Sassanelli, C., Terzi, S., 2019. Towards Circular Business Models: a sys- Velamuri, V., Bansemir, B., Neyer, A.-K., Mo €slein, K., 2013. Product service systems as
tematic literature review on classification frameworks and archetypes. J. Clean. a driver for business model innovation: lessons learned from the manufacturing
Prod. 236, 117696. industry. Int. J. Innov. Manag. 17 (01).
Saavedra, Y.M.B., Iritani, D.R., Pavan, A.L.R., Ometto, A.R., 2018. Theoretical contri- Velte, C., Steinhilper, R., 2016. Complexity in a Circular Economy: A Need for
bution of industrial ecology to circular economy. J. Clean. Prod. 170 (Suppl. C), Rethinking Complexity Management Strategies World Congress on Engineer-
1514e1522. ing. London, UK.
Saidani, M., Yannou, B., Leroy, Y., Cluzel, F., 2017. How to assess product performance Vermunt, D.A., Negro, S.O., Verweij, P.A., Kuppens, D.V., Hekkert, M.P., 2019.
in the circular economy? Proposed requirements for the design of a circularity Exploring barriers to implementing different circular business models. J. Clean.
measurement framework. Recycling 2 (1), 6. Prod. 222, 891e902.
Sinclair, M., Sheldrick, L., Moreno, M., Dewberry, E., 2018. Consumer intervention Vlachos, D., Georgiadis, P., Iakovou, E., 2007. A system dynamics model for dynamic
mappingda tool for designing future product strategies within circular product capacity planning of remanufacturing in closed-loop supply chains. Comput.
service systems. Sustainability 10 (6), 2088. Oper. Res. 34 (2), 367e394.
Snir, E.M., 2001. Liability as a catalyst for product stewardship. Prod. Oper. Manag. Wang, F., Huisman, J., Stevels, A., Balde , C.P., 2013. Enhancing e-waste estimates:
10 (2), 190e206. improving data quality by multivariate InputeOutput Analysis. Waste Manag.
€ter, M., 2003. Strategic management of spare parts in closed-loop
Spengler, T., Schro 33 (11), 2397e2407.
supply chainsda system dynamics approach. Interfaces 33 (6), 7e17. Wang, Y., Chang, X., Chen, Z., Zhong, Y., Fan, T., 2014. Impact of subsidy policies on
Stahel, W.R., 2010. The Performance Economy, 2 ed. Palgrave Macmillan UK, recycling and remanufacturing using system dynamics methodology: a case of
Hampshire, UK. auto parts in China. J. Clean. Prod. 74 (Suppl. C), 161e171.
Sterman, J., 2000. Business Dynamics: Systems Thinking and Modeling for a Com- Wastling, T., Charnley, F., Moreno, M., 2018. Design for circular behaviour: consid-
plex World. McGraw-Hill Education. ering users in a circular economy. Sustainability 10 (6), 1743.
Su, B., Heshmati, A., Geng, Y., Yu, X., 2013. A review of the circular economy in Whalen, K.A., 2019. Three circular business models that extend product value and
China: moving from rhetoric to implementation. J. Clean. Prod. 42, 215e227. their contribution to resource efficiency. J. Clean. Prod. 226, 1128e1137.
Sundin, E., Lindahl, M., Ijomah, W., 2009. Product design for product/service sys- Whalen, K.A., Milios, L., Nussholz, J., 2018. Bridging the gap: barriers and potential
tems: design experiences from Swedish industry. J. Manuf. Technol. Manag. 20 for scaling reuse practices in the Swedish ICT sector. Resour. Conserv. Recycl.
(5), 723e753. 135, 123e131.
Tsoulfas, G.T., Pappis, C.P., 2006. Environmental principles applicable to supply Wong, C.W.Y., Lai, K.-h., Shang, K.-C., Lu, C.-S., Leung, T.K.P., 2012. Green operations
chains design and operation. J. Clean. Prod. 14 (18), 1593e1602. and the moderating role of environmental management capability of suppliers
Tukker, A., 2015. Product services for a resource-efficient and circular economy e a on manufacturing firm performance. Int. J. Prod. Econ. 140 (1), 283e294.
review. J. Clean. Prod. 97, 76e91. Yenipazarli, A., Vakharia, A., 2015. Pricing, market coverage and capacity: can green
Urbinati, A., Chiaroni, D., Chiesa, V., 2017. Towards a new taxonomy of circular and brown products co-exist? Eur. J. Oper. Res. 242 (1), 304e315.

You might also like