Vous êtes sur la page 1sur 17

International Journal of Production Economics 208 (2019) 483–499

Contents lists available at ScienceDirect

International Journal of Production Economics


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

Warranty and maintenance analysis of sensor embedded products using T


internet of things in industry 4.0
Ammar Y. Alqahtania,∗, Surendra M. Guptab, Kenichi Nakashimac
a
King Abdulaziz University, Faculty of Engineering, Department of Industrial Engineering, Jeddah, Saudi Arabia
b
Northeastern University, Mechanical and Industrial Engineering, Boston, USA
c
Waseda University, School of Social Sciences ,Japan

ARTICLE INFO ABSTRACT:

Keywords: The Internet of Things plays a critical role as a facilitator of concrete planning for both warranty and main-
Internet of Things tenance procedures for remanufactured products by significantly reducing or nearly eliminating ambiguity about
Preventive maintenance the existence, conditions, and types as well as the remaining lives of individual constituents of an End-of-Use
Warranty policies product. Since remaining useful life serves as a consistent measure of reliability, an immediate determination of
Remanufacturing
a product's remaining useful life facilitates an optimal decision on recovery by meeting the minimum quality
Extending product life-cycle
requirement defined in the underlying system criteria. This paper studies and scrutinizes the potential effect by
offering one-dimensional renewing/non-renewing warranties on remanufactured products. The study deliber-
ates on one type of product recovery system: The Advanced Remanufacturing-To-Order (ARTO) system. A dis-
crete-event simulation model was developed from the view of remanufacturer for remanufactured items sold
with one-dimensional warranty, in which, an End-Of-Life product is subjected to preventive maintenance action
when the product's remaining life reaches a pre-determined value in order to maximize expected profit for the
remanufacturer. Experiments were designed using Taguchi's Orthogonal Arrays to embody the full recovery
system and observe its performance under various experimental parameters. In order to assess the effect of both
warranty and preventive maintenance on remanufacturer total cost, pairwise tests were conducted as well as
one-way analyses of variance (ANOVA) and Tukey pairwise comparisons test for each performance measure of
the ARTO system.

1. Introduction acquired components of EOUP and remanufactured products.


As a network of uniquely identifiable components – or things, the
The recent intensification of the exploitation of natural resources, as Internet of Things (IoT) augments both decision-making and planning
well as the corresponding decline in the overall number of landfill sites for remanufacturing operations nearly eliminating or significantly re-
has raised several environmental concerns. Consequently, a manifold of ducing inconsistencies and ambiguities. The accessibility of quality
government regulations have been formulated to ensure that manu- product information imposes a positive effect on decisions about pro-
facturers are accountable for their End-of-Use products (EOUPs). The duct recovery – product identification technologies that are built upon a
management of EOUPs involves an array of operations such as dis- robust Radio Frequency Identification (RFID) framework provide the
assembly, cleaning, sorting, inspection, recovery, and disposal. Notably, capability and potential to acquire such information. The role of RFID
the outcome of EOUP recovery operations is normally indeterminate in improving efficiency along the product lifecycle becomes more
because of the inability to predict both the quantity and the quality of visible in the extent to which such technologies reduce ambiguities by
retrieved EOUPs. This level of ambiguity makes the task of reliability providing accurate data on product attributes such as location and
management highly problematic within the context of reverse logistics characteristics. A networked RFID infrastructure provides automated
(RL). Being a major component of RL, the process of remanufacturing and highly efficient strategy for collecting and delivering comprehen-
presents the most daunting challenge for operations management due sive product information at the component level in real-time, thereby
to the unpredictability and inconsistency in the quality of returned facilitating both process improvements and high-level critical decisions
products, as well as the considerable variations in the overall quality of along the entire product recovery lifecycle (Paksoy et al., 2016).


Corresponding author.
E-mail addresses: aaylqahtani@kau.edu.sa (A.Y. Alqahtani), s.gupta@northeastern.edu (S.M. Gupta), nakasima@kanagawa-u.ac.jp (K. Nakashima).

https://doi.org/10.1016/j.ijpe.2018.12.022
Received 21 December 2017; Accepted 22 December 2018
Available online 26 December 2018
0925-5273/ © 2018 Elsevier B.V. All rights reserved.
A.Y. Alqahtani et al. International Journal of Production Economics 208 (2019) 483–499

It is noteworthy that IoT facilitates a seamless monitoring and deliberated on in this study is a sort of product recovery system. A
tracking of objects in the network infrastructure through RFID. Apart discrete-event simulation model was developed from the view of re-
from providing insight into the possibilities of component or product manufacturer for remanufactured items sold with one-dimensional Free
failure, the data obtained from the embedded sensors provide estimates Replacement Warranty (FRW), Pro-Rata Warranty (PRW) or
for the remaining useful life (RUL) of the components of products as Combination FRW/PRW warranty, in which, an End-Of-Life product
they approach their End-of-Use (EOU). The RUL is a subjective estimate (EOLP) is equipped with sensors to facilitate the process of collecting
of the number of remaining years that an item, component, or system is and transferring the necessary information via IoT system. EOLP is
estimated to be able to function in accordance with its intended purpose subjected to preventive maintenance action when the remaining life of
before warranting replacement. The remaining useful life is estimated the product reaches a pre-specified value so that the remanufacturer's
based upon observations, or average estimates of similar items, com- expected profit can be maximized. Experiments were design using
ponents, or systems, or a combination thereof. It is noteworthy, more- Taguchi's Orthogonal Arrays to represent the full recovery system and
over, that any value of RUL represents a significant measure of quality observe its behavior under different experimental conditions. In order
and reusability, enabling the design of superior recovery models that to assess the impact of warranty and preventive maintenance on re-
guarantee the minimum quality measures and system criteria (Fang manufacturer total cost, pairwise t tests were carried out along with
et al., 2016a). one-way analyses of variance (ANOVA) and Tukey pairwise compar-
Once embedded to a product, RFID sensor tags monitor all critical isons test for each performance measure of ARTO system.
components of the product during its entire economic lifecycle. The This study was able to determine the optimal costs of warranty and
data acquired from the RFID tags can be integrated with external da- preventive maintenance for one-dimensional renewable warranty of-
tabases that consist of product information as provided by Original fered on sensor embedded remanufactured products using the simula-
Equipment Manufacturers (OEMs). At the beginning-of-life (BOL) tion model and design of experiments analysis. Moreover, the optimum
period, the tags record information such as models, serial numbers, prices for remanufactured product to gain attraction in the eyes of the
location, maintenance instructions, terms for warranty, date of manu- buyer.
facture, and guidelines for processing EOU. Comparatively, the data This is the first study that use IoT to maximize the remanufacuterer
recorded in the tags during the middle-of-life (MOL) phase include the profit, minimize the warranty and preventive maintenance cost, max-
date of sale, customer identification number, conditions for optimal imize the confidence of the consumers toward buying a remanufactured
performance, critical failures, run cycles, and maintenance information. products and maximize the attractiveness of the remanufactured pro-
Courtesy of the information fusion and integration capabilities provided ducts prices.
by the IoT framework, component-level information informs decisions The remainder of the paper is organized as follows: section 2 lists all
about product design and maintenance schedule while guiding the relevant findings in literature. Section 3 and section 4 respectively
design and development of a warning system in both BOL and MOL present system description and one-dimensional warranty. Section 5
stages. Yet, relatively, complete information about the quantity and presents the notations. Section 6 provides maintenance and warranty
quality of EOUPs and their RULs make the costly initial inspection and formulation. Lastly, Section 7 and Section 8 offer results and conclu-
disassembly operations unnecessary while optimizing remanufacturing sions.
planning (Fang et al., 2016b).
Products that are tagged with embedded sensors are popularly 2. Literature review
known as sensor-embedded products (SEPs). Such products have im-
planted sensors that monitor and record critical product attributes This section provides a literature review on the issues considered in
during use. An Advanced Manufacturing-to-Order (ARTO) is an ad- this research. First, a brief review on environmentally conscious man-
vanced system that subjects SEPs to an array of recovery operations ufacturing and product recovery is presented. There are two important
ensuring that resource demands and product qualities are guaranteed review papers that are available in the literature and are directly related
within the scope of the stipulated system goals. The use of the IoT to the subject area of this paper. The first one is the state of the art
framework in an ARTO context not only gives customers the chance to survey paper by Gungor and Gupta (1999) covering papers in En-
define minimum quality demands for their orders; it guarantees that the vironmentally Conscious Manufacturing and Product Recovery
remaining-life-time-based component and product requirements are (ECMPRO) published through 1998. The second one is also the state of
fulfilled. Similarly, perhaps noteworthy is that the determination of the art survey paper in the same area covering papers published be-
RULs enables the definition of warranty levels based on real and reli- tween 1998 and 2010 by Ilgin and Gupta (2010a,b). Together, they
able data instead of estimations. Therefore, the information obtained classified more than 870 papers (330 and 540 respectively) under four
from RFID sensor tags enables the preparation of customer orders that main categories, viz. environmentally conscious product design, reverse
exceed the defined minimum quality requirements by customers to and closed-loop supply chains, remanufacturing and disassembly. Then,
minimize chances of warranty claims. a section on Radio-frequency identification (RFID), Internet of Things
This paper's chief contribution is that it provides a quantitative (IoT), Industry 4.0, warranty and maintenance are presented and re-
appraisal of the impact of offering warranties on remanufactured items. lated literature is highlighted to show their role in the Environmentally
This assessment represents a remanufacturer's perspective by proposing Conscious Manufacturing and Product Recovery.
an appealing price for the buyer as well. Although developmental stu-
dies exist regarding warranty policies for brand new products and 2.1. Environmentally conscious manufacturing and product recovery
secondhand products, no study currently exists to evaluate the possible
benefits of warranties on remanufactured products in a measurable and Over the last few years, the volume of research studies addressing
all-inclusive manner. In the afore-mentioned studies, the profit im- environmentally conscious manufacturing and product recovery
provements achieved through warranty options control the range of (ECMPRO) issues has gained tremendous notice among in the research
financial investment in a warranty while simultaneously maintaining community (Gungor and Gupta, 1999; Gupta and Lambert, 2007; Ilgin
overall profit. To that end, this paper dissects and evaluates the impact and Gupta, 2010c; Ilgin et al., 2015; Gupta and Ilgin, 2018). Some
of making renewing warranties available on remanufactured products. reasons that researchers have become highly invested in ECMPRO is-
Specifically, this paper proposes a methodology that both minimizes sues could be attributed to environmental concerns, legislative regula-
remanufacturers' costs and maximizes consumer confidence in buying tions, and public opinion; however, researchers’ interest can also be
remanufactured products. linked to the sheer increase in profitability acquired by implementation
The Advanced Remanufacturing-To-Order (ARTO) system of reverse logistics and resolution of product recycling (Bouzon et al.,

484
A.Y. Alqahtani et al. International Journal of Production Economics 208 (2019) 483–499

2015). In response to an increased consumer awareness regarding en- developers carefully assess and embrace both opportunities and chal-
vironmental concerns and more stringent environmental regulations, lenges of IoT revolution in the business domain.
manufacturers have specified facilities as a means to minimize waste The main rationale behind the IoT revolution is to guarantee the
accumulation by recovering materials from EOL products (Gungor and real-time interconnectivity and interaction between devices connected
Gupta, 2002). Additionally, researchers have attempted to better un- to a network. As objects become more intelligent and autonomous, they
derstand the panoptic environmental quandaries that arise during have a greater capability to produce intelligence by processing raw data
product manufacturing. Consequently, researchers have published and in real-time, and through machine learning, and use this intelligence to
disseminated appraisals of these panoptic dilemmas associated with make critical decisions. Because of this capability, therefore, the IoT
environmental manufacturing and product recovery (see for example, revolution promises a further development in the interconnectivity
Moyer and Gupta, 1997; Gupta, 2013; Ilgin and Gupta, 2012). The most between devices as well as the evolution of major concepts such as big
crucial step in the study of remanufacturing is disassembly, specifically data, machine learning, semantic technologies, robotics, and cloud
because of the significant role that disassembly plays in the complete computing among others. These concepts have significantly augmented
recovery system. For more detail on various disassembly components, the overall capability of IoT devices to act autonomously while seam-
refer to the book by Lambert and Gupta (2005). lessly sharing data with other devices on the network. Therefore, es-
sentially, the IoT offers technological solutions that are based off on the
2.2. Radio-frequency identification (RFID) integration of several technologies and combinations of software and
hardware elements to retrieve, process, store, and retrieve information
RFID is a key player in the design and operation of an efficient (Gnimpieba et al., 2015).
supply network in that it promises to eradicate redundancies and in- Nonetheless, product production or manufacturing operations must
consistencies resulting from the manual counting of inventory and adapt to the rapid changes in the IoT industry. Research shows that the
order-numbering mistakes. The integration and exploitation of in- future workplace will be characterized by a complex interaction be-
formation technology enables the realization of the actual potential of tween humans and a sophisticated system of processes, machines, ro-
supply chain management. In fact, the development of a robust and botics, sensors, and devices. Therefore, this trend suggests a widespread
adapted supply chain information system justifies the rationale for in- use of sophisticated, efficiently, and self-learning manufacturing or
tegrating latest technology into supply chain operations to ensure that production processes as the ideal measure of competitive advantage
such operations stay up-to-date (Edwards et al., 2001). The available and success. Smart manufacturing processes as well as full automation
literature suggests that organizations that integrate technology into of operations with a strong integration of information systems and
their supply chain operations alongside a flexible, comprehensive mechanical production equipment will characterize the future work-
business redesign are comparatively more successful (Leung et al., place (Nativi and Lee, 2012).
2014). Therefore, research evidence supposes that technologies that This concept of the future factory is inclined toward enabling and
streamline business processes are critical for the sustainability of an guaranteeing the availability of adequate, relevant, and quality product
organization in the global supply chain (Shin and Eksioglu, 2015). For processing information in the correct time and in a correct manner.
instance, Finkenzeller (2003), argues that RFID is a complex technology However, notably, such efficiency will only be realistic with a full in-
that can be integrated into mainstream business processes to automate terconnectivity among all components of the value chain – yet, the
and streamline operations. components might easily adapt to the industrial dynamics by in-
tegrating real-time processes. Therefore, it will be imperative for both
2.3. Internet of things (IoT) manufacturers and customers to conceptualize the future supply chain
as a highly sophisticated ‘system’ of cohesive components, procedures,
Also referred to as the Industrial Internet or the Internet of and machinery. The explosion of the IoT computing paradigm and the
Everything, IoT is an emergent technology framework that is conceived realization of its potential to streamline business and manufacturing
as a universal network infrastructure that is made up of autonomous processes will necessitate the formulation of different concepts to sy-
devices and machines that are capable of interacting and collaborating nergize and optimize the intricate human-machine interaction with a
with each other. Truly, organizations cannot realize the full potential major goal of increasing efficiency, reducing cost overheads, and
and true value of IoT if there is no seamless communication of the maximizing resource utilization (Sun, 2012).
connected devices. Similarly, devices forming the IoT infrastructure
must be able to integrate with all enterprise systems as well as com- 2.4. Industry 4.0
plementary technologies such as cloud computing, decision-support
systems, and business analytics. The recent increase in the proliferation The first industrial revolution saw a widespread use of mechaniza-
of IoT technologies is a result of a complex interplay of several dy- tion and steam power for large-scale manufacturing. However, elec-
namics including societal, competitive, and technological pressures that tricity would evolve into a dominant source of energy for manu-
make innovation and organizational transformation inevitable (For facturing and industrial processes in the second industrial revolution.
further detail on the emerging field of IOT using RFID and its appli- Notably, the third revolution witnessed large-scale computerization or
cations in supply chain management, see Naskar et al., 2017 & Ben- automation of industrial processes by integrating new technologies into
Daya et al., 2017). core manufacturing processes. The fourth revolution, however, is
Factors such as sensor technologies, wireless network capabilities, characterized by the introduction of sophisticated cyber-physical sys-
digital convergence, RFID, and sophisticated mobile computing tech- tems. In this regard, therefore, the inception and integration of com-
nologies have contributed significantly to the rapid rate of transfor- prehensive cyber-physical production systems (CPPSs) tends to obscure
mation in the IoT industry experienced over the past decade. the traditional borders between digital and physical domains (Wang
Nonetheless, even though the benefits of integrating IoT technologies et al., 2016).
into mainstream business processes are diverse in nature, the IoT re- Therefore, the phrase “Industry 4.0” is widely used to refer to the
volution presents critical challenges that are equally diverse and dy- fourth industrial revolution. Comparatively, it is conceived in some
namic in nature. For instance, the constant up scaling of IoT capabilities schools of thought as an emergent culture of control and organization of
will result into data explosion, causing challenges to data security, data supply chain industries as well as their product lifecycle with a primary
privacy, networking of data centers, and storage management. goal of satisfying the unique customer needs. Therefore, the cycle is
Therefore, apart from understanding the critical role of IoT in providing designed such that it begins with the generation of the idea through the
a competitive edge by streamlining business processes and operations, placement of the order and product development to product delivery to

485
A.Y. Alqahtani et al. International Journal of Production Economics 208 (2019) 483–499

the consumer (Wollschlaeger et al., 2017). This paradigm guarantees Table 1


both recovery and post-delivery services. It is conclusive, consequently, Notation used in the warranty formulation.
to assert that the main drive behind the fourth industrial revolution is to Notations
guarantee the effective availability of reliable, complete, and real-time
information by linking together all parties or elements of the value W: Warranty period
W1: Sub-interval of warranty period
chain. It is noteworthy, moreover, that the interlink between the var-
Co: Operating cost of item
ious components of the value chain system – including people, devices, CS: Sale price of item
and things – fosters highly organized, dynamic, and streamlined value- Cp: Cost to remanufacture an item
adding streams for all companies involved in the value chain (Theorin n: Number of components in an item
et al., 2017). RL: Remaining life of item at sale
RLi: Remaining life of component i (1 ≤ i ≤ n)
j: Number of preventive maintenance measures
v: Virtual remaining life
2.5. Warranty analysis vj: Virtual remaining life after performing the jth PM activity
m: Level of PM effort
δ(m): Remaining life increment factor of PM with effort m
A warranty is a manufacturer's (vendor's/seller's) contractual ob-
Ui, Li: Upper and Lower range of replacement component's remaining life
ligation associated with an individual product sale. The main intention Λi: Intensity function of non-stationary Poisson process
of a warranty is to ascertain liability in case of a purchased item Miu: Renewal function associated with Fiu (x)
functionally failing at a premature time. Warranties specify the guar- E [.]: Expected value of expression within [.]
anteed performance of a product. If this anticipated degree of perfor- Fi (x): Failure distribution of a remanufactured component i
Fi1 (x): Distribution function for times to first failure of component i
mance is not met, the buyer can expect to receive compensation
Fi2 (x): Distribution function for times to subsequent failures of component
(Blischke, 1993). A product warranty serves a number of key purposes. i
First, the warranty offers insurance and protection for the buyer, such Fiu (x): Distribution function for times to failure of remanufactured
that product failure is deemed the responsibility of the sellers (Heal, component used in replacement
1977). Furthermore, product warranties can indicate product reliability H (rl): Distribution function for a remanufactured item
Hi (rl): Distribution function for a remanufactured component
in the minds of customers (Blischke, 1995), (Gal-Or, 1989), (Soberman, N (W; RL): Number of failures over the warranty period with remaining life, RL
2003), (Spence, 1977). Meanwhile, sellers can rely on product war- Ni(W; RLi): Number of failures for component, i, over the warranty period
ranties to obtain a supplemental source of profitability from the product Ʌ(t): Intensity function for system failure
(Lutz and Padmanabhan, 1995). Fw(x): Distribution function for the first failure in the period [W1, W)
given by the excess remaining life of renewal process associated
One new research arena, although with a limited number of pub-
with failures in the period [0, W1)
lications, involves conducting cost analysis of warranties and modelling Cd(W; RL): Total warranty cost to remanufacturer
for used products. Shafiee et al. (2011a) discuss the most preferred up- Cj: Cost of replacement/repair jth failure, j ≥ 1
to-date strategies for used product items that involve conducting
screening tests as a method of reliability development. These re-
searchers have built a stochastic prototype intended to assess ideal ways 2.6. Maintenance analysis
of investing in used products as a means to increase individual relia-
bility in these products by using free repair warranty (FRW) policies. Maintenance has a meaningful function in product dependability
Saidi-Mehrabad et al. (2010) and Shafiee et al. (2011a) deduced that a and quality. According to the research publications, two main types of
greater quantity of investments indicated more significant decrease in maintenance exist: corrective maintenance (CM) and preventive
the virtual age and increased reliability on the improved or newer maintenance (PM). CM is performed when a product component fails
product. Indeed, Shafiee et al. (2011b) presented a stochastic depend- and measures are taken to return a faulty component to its oper-
ability improvement prototype for used products, which covered war- ationally functional state. Meanwhile, PM is performed prior to product
ranties and Cobb-Douglas-Type production function to reach the pro- failure in order to reduce the product's potential deterioration and
duct's ideal upgrade capability. Naini and Shafiee (2011) conducted a chance of failure. When a product's remaining life is determined to be
study to ascertain the ideal upgrade for the product, as well as the short, the offered product's warranty is similarly short; for such pro-
market price and maximum anticipated profit with limiting supposi- ducts, CM policy is appropriate. On the contrary, when a product's re-
tions about the product's age distribution. Naini and Shafiee (2011) maining is determined to be long, the product's warranty could be re-
created a mathematical prototype to conduct a parametric examination latively long; for such products, PM policy is offered, and the service
on the chronological ages of product items as a means to identify the and repair costs outlined in the warranty can be reduced due to the
most appropriate warranty policies. Furthermore, Yazdian et al. (2014) offering of PM. Therefore, a known association exists between war-
utilized a model based upon integrated mathematics that did not de- ranties and CM and PM policies.
pend on an obtained product item's age in order to determine the An extensive array of research literature exists on the topic of
commonly practiced decisions of remanufacturers. Liao et al. (2015) maintenance policies. Multiple papers have been published regarding
studies the impact that a warranty policy creates on customer behavior, maintenance policies (Wang, 2002; Garg and Deshmukh, 2006; Sharma
assessing it according to the consumers' mindset. The major problem is et al., 2011). In this paper, we reference the work of Nakagawa (2006)
that warranty cost analysis for remanufactured products has not been for comprehensive understanding about the fundamental study of
significantly studied. However, several studies that have assessed the maintenance theory. Nakagawa (2008) provides a detailed overview of
warranty for the reverse and closed-loop supply chain management of maintenance policy modelling scenarios.
remanufactured products have been published. Alqahtani and Gupta Shafiee and Chukova (2013) suggest that the study of maintenance
(2017a) have indicated that base and extended one-dimensional war- policies for used products during the warranty timeframe have not pi-
ranty may be presented for products that are to be remanufactured qued the curiosity of researchers. Yeh et al. (2011) suggested two
using Free Replacement Warranty (FRW) and Pro-Rata Warranty (PRW) periodical age-reduction PM models to lower the observed frequency of
policy. In addition, these researchers (2016a, 2016b, 2016c) also point failure of the used goods. Kim et al. (2011) studied the optimal periodic
out that an array of warranty policies can be offered for End-of-life PM policies of used items following the expiration of products' war-
(EOL) derived products. These policies include renewable, nonrenew- ranty. Kim et al. (2011) claims that from the manufacturers’ point-of-
able, one- and two-dimensional warranty policies. view, PM activities should only be utilized when the savings of service
costs outlined in the warranty actually exceed the additional costs

486
Table 2
Warranty Formulation for different one-dimensional policies.
Policy First Failure Succeeding Failures virtual remaining life Expected warranty cost Expected warranty cost to the buyer
after performing to the remanufacturer
the jth PM activity
A.Y. Alqahtani et al.

FRW Fi1 (x ) = [Fi (RLi + x ) Fi2 (x ) = p Fi (x ) vj = vj 1 + (m)(Aj Aj 1) v +W


E [Cd (W; v )] = cd (t ) dt
Fi (RLi)]/[1 Fi (Li)] + (1 p) Fiu (x )
v
N/A
PRW W W
1
E [Cd (W ; v )] = S (x ) dFi1 (x ) = cS (v ) Fi1 (W ) x dFi1 (x )
W
( )
0 0
N/A
FRW/PRW W1
[Cd (W1, W ; v )] = CS Fi1 (W1) + Miu (W1 x ) dFi1 (x )
0
(W W1)
W W1 1
+ cS { W }F W (W W1) W
( ) XdFW (X )
0
N/A
RFRW W1
E [Cd (W1, W ; v )] = Cp Fi1 (W1) + Miu (W1 x ) dFi1 (x ) + FW (W W1 )
0

/F¯iu (W W1 )

N/A
RPRW (W W1)
1
E [Cd (W1, W ; v )] = cS (v ) W1) YdFW (Y )
{ W WW1 } FW (W W
( ) 0
N/A

487
RFRW/RPRW W1 N/A
E [Cd (W1, W ; v )] = Cp Fi1(W1) + Miu (W1 x ) dFi1(x )
0
(W W1)
W W1 1
+ cS (v ) { W1) YdFW (Y )
W } FW (W W
( ) 0
CSW v+ W v+W
E [Cd (W ; v )] = Ck (v ) × p × (v ) dRL E [Cb (W ; v )] = Ck (v ) × (1 p) × (v ) dv
v v
CLW CI (v) v+ W
E [Cd (W ; v )] = Ck (vRL) × g (Ck ) dCk + CI × G (CI ) E [Cb (W ; v )] = {Ck (v ) CI (v )} × g (Ck ) dCk × (v ) dv
0 CI (v) v
v+ W
× (v ) dv
v
MBG k
E [Jd (W ; v ) = n=0
n c¯ Pn (v W , v) + n= k + 1
{(k c¯ + c¯dd )
+ n c¯ (1 )} P n (v W , v)
N/A
International Journal of Production Economics 208 (2019) 483–499
A.Y. Alqahtani et al. International Journal of Production Economics 208 (2019) 483–499

Fig. 1. ARTO smart system's recovery processes.

incurred by conducting the PM work. Thus, creating and implementing environment and insisting upon a higher quality of products, then
PM policies for remanufactured products still calls for additional study warranties for remanufactured products have become more meaningful.
(Alqahtani and Gupta, 2017b). Such influential consumer perceptions cause an increase in the demand
for remanufactured products as well as costs to replace or repair failed
products. Thus, warranty management has become highly essential to
3. One-dimensional warranty
remanufacturing companies of remanufactured products. These re-
manufacturers are responsible for estimating repair and replacement
When deciding whether or not to purchase a product, a buyer will
expenses, and then incorporating these costs into the remanufactured
typically compare the features among similar products. Sometimes, two
product pricing structure. If remanufacturers fail to do so, they hurt
competing brands will produce comparable products with similar
themselves by incurring financial losses, as opposed to profits, as it
highlights, such as purchase price, quality, product reliability, special
pertains to remanufactured product sales. Indeed, analyses of warranty
features, and manufacturer guarantee. In such situations, post-sale add-
expenses for remanufactured products have been demonstrated to be
ons gain relevance, including a manufacturer's discount, warranty, ease
multifaceted in comparison to new product cost analysis, primarily due
of repairs, accessibility of parts, and a myriad of additional services.
to the lack of knowledge about the usage and past maintenance of re-
These post-sale factors may significantly impact the buyer. Likewise,
manufactured products versus new products. Furthermore, warranty
the post-purchase warranty provides an additional assurance to the
policies, relatively consistent to those of new and used products, may
buyer on the product's dependability.
not be financially viable for the remanufacturer. Thus, it is necessary to
A warranty constitutes a promise that the manufacturer will conduct
thoroughly examine and analyze each warranty policy for re-
repairs on any product that fails or appropriately compensate the buyer
manufactured products, and then determine the approximate cost
for any problems with the product during the timeframe of the war-
linked to each policy. Additional issues must be considered, such as
ranty. The warranty's key purpose is to endorse and guarantee the
strategies to service remanufactured spare parts associated with the
quality of the product's performance, thus assuring product's pro-
replacement or repair of failed products while they are under warranty
ductivity for both the customer and the manufacturer. For a specific
(Murthy and Blischke, 2006).
product, the statistical cost of warranty is equivalent across all new
One type of warranty, the one-dimensional warranty, establishes a
items of similar category, provided that the manufacturer has a solid
‘warranty period’ (W). This period of time is determined in accordance
quality assurance in place. In contrast, each EOL product often differs as
with a signal variable, such as age or usage. For non-renewing policies,
a result of certain factors, including age of the item, usage, and past
the item requiring repair or replacement is covered under warranty for
maintenance. Such contributing factors create statistically different
a pre-determined period of time, established as the original warranty
warranty costs for each EOL item-derived remanufactured product.
timeframe. For renewing policies, the warranty period starts fresh each
Since consumers are increasingly more aware of the competitive

488
A.Y. Alqahtani et al. International Journal of Production Economics 208 (2019) 483–499

Fig. 2. Disassembly process for SEPs in ARTO system.

489
A.Y. Alqahtani et al. International Journal of Production Economics 208 (2019) 483–499

Fig. 3. Refurbished process for SEPs in ARTO system.

time that the product is repaired or replaced. This actually creates a Under the CLW policy, the remanufacturer's liability is based upon cost
greater level of uncertainty, since the warranty only expires when the limits on individual claims or total claims for the duration of the war-
item does not fail for a pre-determined period of time. A myriad of one- ranty period. Under a MBG policy, the buyer could return the re-
dimensional consumer non-renewable and renewable warranty policies manufactured product and get a full or partial refund of the sales price
exist. Most products are sold under such warranties as: Non-Renewing from the remanufacturer, provided the product is returned during the
Free Replacement Warranty (FRW) and Non-Renewing Pro-Rata warranty timeframe. Under the FRW/PRW combination policy, the
Warranty (PRW), combination of the both FRW/PRW, Renewing Free remanufacturer replaces failed products at the company's expense up to
Replacement Warranty (RFRW), Non-Renewing Pro-Rata Warranty W1 (W1 < W). If the product failure occurs in the interval [W1, W),
(RPRW), or a combination of the both RFRW/RPRW other policies to then the remanufacturer refunds a percentage of the original sale price;
consider are Cost sharing warranty (CSW), Cost Limit Warranty (CLW) the warranty ends at this time as well. Under RFRW policy, the re-
and Money Back Guarantee (MBG). manufacturer agrees to replace/repair all product failures in the in-
Under the FRW policy the remanufacturer agrees to replace or re- terval [0, W) at zero expense to the buyer. Additionally, the product is
pair all failures throughout the warranty coverage timeframe at zero returned to the consumer with a new warranty of duration W. In this
cost to the buyer. The warranty expires after time W from the time of case, the warranty ends in one of two ways: when there is no failure in
sale. Under the PRW policy the remanufacturer commits to refunding a the original warranty period [0, W), or when there is no failure under
percentage of the sale price, in the event of product failure during the the new warranty period. Finally, under RPRW policy, the re-
warranty period. Furthermore, the buyer is under no obligation to manufacturer provides a replacement product at a reduced cost if a
purchase a replacement product. Under the CSW, the buyer and the problem occurs during the product's warranty timeframe W. In RPRW
remanufacturer share the expense to replace/repair a given product. policy, the buyer is required to apply the refund towards a replacement

490
A.Y. Alqahtani et al. International Journal of Production Economics 208 (2019) 483–499

Fig. 4. Remanufacturing process for SEPs in ARTO system.

product from the remanufacturer. The amount of the refund depends on 4.2. Modelling failures
how much RL is left, and the replacement item carries a new warranty
that is synonymous with the original product warranty. Since the majority of products are multifaceted, then most items can
be viewed and assessed as a system consisting of many parts. When an
item fails, it is because one or more of the components failed. A re-
4. Maintenance and warranty formulation manufactured product or component is designated into one of two ca-
tegories viz., either working or failed. Periods of time between con-
4.1. Modelling preventive maintenance effect secutive failures are random variables, and these are modeled by proper
distribution functions. Interchangeably, a suitable counting process can
Typically, PM activities encompass a series of maintenance activ- model the occurrence of these failures over a period of time.
ities. These activities may include cleaning, lubricating, adjusting and Time to first failure of a remanufactured component is contingent
calibrating, inspecting, or replacing individual components (Ben on the remaining life of the component at the time the remanufactured
Mabrouk et al., 2016). When PM activities are performed effectively, product is sold. If the sensor information about EOL component (the
they can actually lower the occurrences of product failure, conse- component when the item was new) signifies that the component has
quently reducing warranty expenses and increasing consumer satisfac- never failed, or has only required minimal repair, then the remaining
tion. For this particular study, we adopted the modelling framework life of the component at sale is the same as that of the item. Most often,
proposed by Kim et al., 2004, to model the effectiveness of PM activ- the remaining life of remanufactured component at sale differs because
ities. the replacement or repair actions were greater than minimal repair.
Remaining life RL1, RL2, …RLj, …, with RL0 = 0 dictates the PM Thus, the time to first failure under warranty must be defined. Let RLi
activities for a remanufactured item. In this case, PM can result in the denote the remaining life of remanufactured component, i.
actual restoration of an item, which means that the item's virtual re- An embedded product sensor can provide the remanufacturer with
maining life is subsequently increased. Kijima et al. (1988), introduce the remaining life of the sale RL. A point process with intensity function
the concept of virtual age, later extending the notion, (Kijima, 1989). In models the item failure Ʌ (t) whereby t represents the remaining life of
this particular study, the jth PM only reimburses the damage that ac- the item. Ʌ (t) is an increasing function of t, thus indicating that the
crued during the time between the (j − 1)th and the jth PM activities. quantity of failures increases with age. Product failures during a war-
Consequently, a mathematical reduction of virtual remaining life can be ranty timeframe [RL, RL + W) occur according to a non-stationary
acquired (Martorell et al., 1999). Poisson process with intensity function Ʌ (t). This implies that N (W; A),
the number of failures over the warranty period W for an item of

491
A.Y. Alqahtani et al. International Journal of Production Economics 208 (2019) 483–499

Table 3
Factors and factor levels (1–45) used in design-of-experiments study.
No Factor Unit Levels

1 2 3

1 Mean arrival rate of EOL Smart refrigerators Products/hour 10 20 30


2 Mean arrival rate of Remanufactured Smart refrigerators for CM Products/hour 10 20 30
3 Mean arrival rate of Remanufactured Smart refrigerators for PM Products/hour 10 20 30
4 Probability of Repair EOLPs % 5 10 15
5 Probability of a nonfunctional Solenoid Valve % 10 20 30
6 Probability of a nonfunctional motor % 10 20 30
7 Probability of a nonfunctional Fan % 10 20 30
8 Probability of a nonfunctional Compressor % 10 20 30
9 Probability of a missing Solenoid Valve % 5 10 15
10 Probability of a missing motor % 5 10 15
11 Probability of a missing Fan % 5 10 15
12 Probability of a missing Compressor % 5 10 15
13 Mean non-destructive disassembly time for station 1 Minutes 0.8 1 1.3
14 Mean non-destructive disassembly time for station 2 Minutes 0.8 1 1.3
15 Mean non-destructive disassembly time for station 3 Minutes 0.8 1 1.3
16 Mean non-destructive disassembly time for station 4 Minutes 0.5 0.8 1
17 Mean non-destructive disassembly time for station 5 Minutes 0.8 1 1.3
18 Mean non-destructive disassembly time for station 6 Minutes 1 1.5 2
19 Mean destructive disassembly time for station 1 Minutes 0.4 0.5 0.7
20 Mean destructive disassembly time for station 2 Minutes 0.4 0.5 0.7
21 Mean destructive disassembly time for station 3 Minutes 0.4 0.5 0.7
22 Mean destructive disassembly time for station 4 Minutes 0.3 0.5 0.6
23 Mean destructive disassembly time for station 5 Minutes 0.4 0.5 0.7
24 Mean destructive disassembly time for station 6 Minutes 0.6 0.8 1
25 Mean Assembly time for station 1 Minutes 1 1.3 1.5
26 Mean Assembly time for station 2 Minutes 1 1.3 1.5
27 Mean Assembly time for station 3 Minutes 1 1.3 1.5
28 Mean Assembly time for station 4 Minutes 0.8 1 1.3
29 Mean Assembly time for station 5 Minutes 1 1.3 1.5
30 Mean Assembly time for station 6 Minutes 1.3 1.5 1.8
31 Mean demand rate Metal Cover Parts/hour 10 15 20
32 Mean demand rate for Solenoid Valve Parts/hour 10 15 20
33 Mean demand rate for Temperature Controls Parts/hour 10 15 20
34 Mean demand rate for Evaporator Parts/hour 10 15 20
35 Mean demand rate for Motor Parts/hour 10 15 20
36 Mean demand rate for Condenser Parts/hour 10 15 20
37 Mean demand rate for Fan Parts/hour 10 15 20
38 Mean demand rate for Aluminum Radiator Parts/hour 10 15 20
39 Mean demand rate for Compressor Parts/hour 10 15 20
40 Mean demand rate for 1 Year smart refrigerators Products/hour 5 10 15
41 Mean demand rate for 2 Years smart refrigerators Products/hour 5 10 15
42 Mean demand rate for 3 Years smart refrigerators Products/hour 5 10 15
43 Mean demand rate for Refurbished smart refrigerators Products/hour 5 10 15
44 Mean demand rate for Material Products/hour 5 10 15
45 Percentage of Good Parts to Recycling % 95 0.9 0.8

remaining life RL at the time of sale, is a random variable with renewing/non-renewing warranty policy for remanufactured products.
RL + W The Advanced Remanufacturing-To-Order (ARTO) system is the pro-
RL + W (t)dt duct recovery system utilized for this study. Taguchi's Orthogonal
P{N(W:RL)=n} = (t ) dt e RL /n! Arrays were the reference point for designing the experiments used in
(1)
this study, with an intention of representing the entirety of the product
RL

The expected number of failures over the warranty period is given recovery system in order to observe the system's conduct under dif-
by ferent experimental conditions. Furthermore, numerous warranty and
RL + W
preventive maintenance situations were examined using pairwise t-
E[N(W; RL)] = (t ) dt tests, as well as one-way analysis of variance (ANOVA) and Tukey
RL (2) pairwise comparisons tests for each situation, for the purpose of de-
ducing the optimal strategy for the remanufacturer.
The Advanced Remanufacturing-To-Order (ARTO) system used for
4.3. Modelling warranty policies this study is a type of product recovery system. The product example
selected is a smart refrigerator build with sensor embedded and net-
All notations used in the warranty formulation and warranty for- work connectivity which enable to collect and exchange data. The EOL
mulation for different policies are shown in Tables 1 and 2 respec- smart refrigerator performs a series of recovery procedures based upon
tively.. its condition, as documented in Fig. 1. In order to refurbish and repair
the product, reusable components may be needed in order to meet the
5. Advanced remanufacturing-to-order system product's demands. This necessity satisfies the demands of both internal
and external components using the disassembly of recovered parts. In
The study detailed in this paper utilized a discrete-event simulation the ARTO system, there are three specific types of arrived items: EOL
model to improve the application and operation of a one-dimensional

492
A.Y. Alqahtani et al. International Journal of Production Economics 208 (2019) 483–499

Table 4 is conducted in order to remove each individual component. Nine


Factors and factor levels (46–90) used in design-of-experiments study. components are involved in a smart refrigerator: the Metal Cover,
No Factor Unit Levels Solenoid Valve, Temperature Controls, Evaporator, Motor, Condenser,
Fan, Aluminum Radiator, and Compressor. Each component has a
1 2 3 sensor embedded into it to record and transfer the necessary data.
Exponential distributions are employed to produce the station dis-
46 Mean Metals Separation Process Hour 1 1.3 1.5
47 Mean Copper Recycle Process Minutes 1 1.3 1.5 assembly times, the interarrival times of each component's demand, and
48 Mean Steel Recycle Process Minutes 1 1.3 1.5 the interarrival times of EOL smart refrigerators. After the retrieval of
49 Mean Fiberglass Recycle Process Minutes 1 1.3 1.5 this information, all End-Of-Life products (EOLPs) are either trans-
50 Mean Dispose Process Minutes 0.5 0.8 1 ported to station 1 for disassembly or, if only a specific component of
51 Maximum inventory level for smart Products/hour 10 15 20
the EOLP needs to be repaired, then this part is delivered to its corre-
refrigerators
52 Maximum inventory level for Refurbished Products/hour 10 15 20 sponding station. Depending upon the specific part's condition, one of
smart refrigerators two disassembly operations will be utilized, either destructive or non-
53 Maximum inventory level for smart Products/hour 10 15 20 destructive. Destructive disassembly is utilized if the component is non-
refrigerator Component
functional (which means, it is broken and has zero percent of remaining
54 PM action effort level number 0.2 0.5 0.6
55 Mean measurements time for Metal Cover Parts/hour 10 15 20 life). This destructive disassembly is performed in a manner that pre-
56 Mean measurements time for Solenoid Parts/hour 10 15 20 serves the functionality of the other components. Thus, the cost of unit
Valve disassembly is higher for a functional component than for a nonfunc-
57 Mean measurements time for Temperature Parts/hour 10 15 20 tional component. Additionally, component testing is not necessary
Controls
after disassembly because information pertaining to the state of each
58 Mean measurements time for Evaporator Parts/hour 10 15 20
59 Mean measurements time for Motor Parts/hour 10 15 20 component is available from their sensors. Also, life cycle information
60 Mean measurements time for Condenser Parts/hour 10 15 20 and demands for EOLPs is expected to be known. Furthermore, the
61 Mean measurements time for Fan Parts/hour 10 15 20 retrieval of information from sensors is presumed to cost less than the
62 Mean measurements time for Aluminum Parts/hour 10 15 20
inspection and testing of the components. Figs. 2–4 show the details of
Radiator
63 Mean measurements time for Compressor Parts/hour 10 15 20
the disassembly, refurbished and remanufacturing processes for SEPs in
64 Mean adjustments time for Metal Cover Parts/hour 10 15 20 ARTO System respectively.
65 Mean adjustments time for Solenoid Valve Parts/hour 10 15 20 It is important to note that the recovery operations vary for each
66 Mean adjustments time for Temperature Parts/hour 10 15 20 SEP according to their general condition and approximated remaining
Controls
lifespan. The components that are recovered can be used to meet the
67 Mean adjustments time for Evaporator Parts/hour 10 15 20
68 Mean adjustments time for Motor Parts/hour 10 15 20 demands of spare parts. Meanwhile, fully recovered or refurbished
69 Mean adjustments time for Condenser Parts/hour 10 15 20 products can be used to meet the consumer product demand. In addi-
70 Mean adjustments time for Fan Parts/hour 10 15 20 tion, the use of recycled products and parts meet material demands. The
71 Mean adjustments time for Aluminum Parts/hour 10 15 20
recovered products and parts are categorized according to remaining
Radiator
72 Mean adjustments time for Compressor Parts/hour 10 15 20
lifespan, and then placed in distinct life-bins, such as one year, two
73 Mean part replacement time for Metal Parts/hour 10 15 20 years, or three years. These products and components now wait for
Cover retrieval by customer demand. Underutilization is a distinct risk for any
74 Mean part replacement time for Solenoid Parts/hour 10 15 20 product or component, particularly if a product or component is qua-
Valve
lified to enter a higher life-bin but is moved to a lower life-bin due to
75 Mean part replacement time for Parts/hour 10 15 20
Temperature Controls the higher life-bin's full occupancy. When any product, component, or
76 Mean part replacement time for Evaporator Parts/hour 10 15 20 material inventory is determined to be greater than the maximum in-
77 Mean part replacement time for Motor Parts/hour 10 15 20 ventory permitted, then it is designated as excess and is either disposed
78 Mean part replacement time for Condenser Parts/hour 10 15 20
of entirely or used to meet basic material demand.
79 Mean part replacement time for Fan Parts/hour 10 15 20
80 Mean part replacement time for Aluminum Parts/hour 10 15 20
In order to successfully meet product demand, repairing and re-
Radiator furbishing options may be selected. EOLP may be missing parts or
81 Mean part replacement time for Parts/hour 10 15 20 possess nonfunctional (broken, with zero remaining life) components
Compressor that must be either replaced or restored during the process of repairing
82 Mean cleaning time for Metal Cover Parts/hour 10 15 20
or refurbishing, so that specific remaining life requirements are met. It
83 Mean cleaning time for Solenoid Valve Parts/hour 10 15 20
84 Mean cleaning time for Temperature Parts/hour 10 15 20 is also possible that EOLP possess components that have lesser-than-
Controls preferred remaining lives hence, these components may also have to be
85 Mean cleaning time for Evaporator Parts/hour 10 15 20 replaced.
86 Mean cleaning time for Motor Parts/hour 10 15 20 When SEPs fail during the warranty timeframe, then these failed
87 Mean cleaning time for Condenser Parts/hour 10 15 20
88 Mean cleaning time for Fan Parts/hour 10 15 20
smart refrigerators arrive at the ARTO system so that information can
89 Mean cleaning time for Aluminum Parts/hour 10 15 20 be retrieved by utilizing a radio frequency data reader. Afterward, the
Radiator failed smart refrigerators go through the recovery operations process,
90 Mean cleaning time for Compressor Parts/hour 10 15 20 similar to one described above for an EOLP.
Finally, to lower the potential failure risk, PM activities are per-
formed throughout the warranty timeframe. During these activities, if
products for recovery process, failed sensor-embedded products (SEPs)
the remanufactured smart refrigerator's remaining life reaches a pre-
for rectification, or SEPs for maintenance activities. This propose
determined value, then the remanufactured SEPs arrive at the ARTO
system can be applicable and generalize for a wide range of products
system. Once at ARTO, these remanufactured SEPs undergo information
from appliances to electronics and so more.
retrieval, by utilizing a radio frequency data reader. Afterward, the
The processing of smart refrigerators is complex. First, EOL smart
SEPs undergo four maintenance activities, according to the information
refrigerators reach the ARTO system to retrieve information utilizing a
that the sensor captured about their condition. These four maintenance
radio frequency data reader located in the facility's database. Next, the
activities include the following: measurements, adjustments, parts re-
smart refrigerators move through a disassembly line, which is com-
placement, and cleaning. When PM activities are performed with de-
prised of six different stations. The comprehensive disassembly process
gree δ, then the remaining life of the remanufactured smart

493
A.Y. Alqahtani et al. International Journal of Production Economics 208 (2019) 483–499

Table 5
Parameters used in the ARTO system.
No Parameters Unit Value No Parameters Unit Value

1 Backorder cost rate % 40 40 Weight for Compressor lbs. 6


2 Holding cost rate % 10 41 Unit copper scrap revenue $/lbs 0.6
3 Remanufacturing cost/minute $ 1.5 42 Unit Fiberglass scrap revenue $/lbs 0.9
4 Disassembly cost/minute $ 1 43 Unit steel scrap revenue $/lbs 0.2
5 Price for 1 Year Metal Cover $ 10 44 Unit disposal cost $/lbs 0.3
6 Price for 1 Year Solenoid Valve $ 20 45 Unit copper scrap Cost $/lbs 0.3
7 Price for 1 Year Temp. Controls $ 5 46 Unit Fiberglass scrap Cost $/lbs 0.45
8 Price for 1 Year Evaporator $ 5 47 Unit steel scrap Cost $/lbs 0.1
9 Price for 1 Year Motor $ 45 48 Price of 1 Year smart refrigerators $ 100
10 Price for 1 Year Condenser $ 15 49 Price of 2 Years smart refrigerators $ 150
11 Price for 1 Year Fan $ 15 50 Price of 3 Years smart refrigerators $ 185
12 Price for 1 Year Radiator $ 15 51 Price of Refurbished smart refrigerators $ 200
13 Price for 1 Year Compressor $ 50 52 Assembly cost/minute $ 1
14 Price for 2 Years Metal Cover $ 15 53 Measurements cost/minute $ 1
15 Price for 2 Years Solenoid Valve $ 30 54 Adjustments cost/minute $ 1
16 Price for 2 Years Temp. Controls $ 12 55 Replacements cost/minute $ 1
17 Price for 2 Years Evaporator $ 12 56 Cleaning cost/minute $ 1
18 Price for 2 Years Motor $ 55 57 Testing cost/minute $ 1
19 Price for 2 Years Condenser $ 18 58 Operation costs Metal Cover $/unit 4
20 Price for 2 Years Fan $ 18 59 Operation costs Solenoid Valve $/unit 4
21 Price for 2 Years Radiator $ 20 60 Operation costs Temp. Controls $/unit 2.8
22 Price for 2 Years Compressor $ 60 61 Operation costs Evaporator $/unit 1.2
23 Price for 3 Years Metal Cover $ 20 62 Operation costs Motor $/unit 4.
24 Price for 3 Years Solenoid Valve $ 35 63 Operation costs Condenser $/unit 1.66
25 Price for 3 Years Temp. Controls $ 15 64 Operation costs Fan $/unit 2.34
26 Price for 3 Years Evaporator $ 15 65 Operation costs Radiator $/unit 0.6
27 Price for 3 Years Motor $ 60 66 Operation costs Compressor $/unit 3.4
28 Price for 3 Years Condenser $ 25 67 Operation costs smart refrigerators $/unit 55
29 Price for 3 Years Fan $ 20 68 Repair costs Metal Cover $/unit 8
30 Price for 3 Years Radiator $ 20 69 Repair costs Solenoid Valve $/unit 8
31 Price for 3 Years Compressor $ 65 70 Repair costs Temp. Controls $/unit 5.6
32 Weight for Metal Cover lbs. 8 71 Repair costs Evaporator $/unit 2.4
33 Weight for Solenoid Valve lbs. 4 72 Repair costs Motor $/unit 8
34 Weight for Temp. Controls lbs. 2 73 Repair costs Condenser $/unit 3.32
35 Weight for Evaporator lbs. 2 74 Repair costs Fan $/unit 4.68
36 Weight for Motor lbs. 6 75 Repair costs Radiator $/unit 1.2
37 Weight for Condenser lbs. 12 76 Repair costs Compressor $/unit 6.8
38 Weight for Fan lbs. 3 77 Repair costs smart refrigerators $/unit 85
39 Weight for Aluminum Radiator lbs. 3

refrigerators will be δ units of time greater than before. In the mean- Overall, the quantity of experiments must be greater than or equal
time, any product or component failures occurring between two suc- to a system's degrees-of-freedom. Specifically, we selected L242(390)
cessive PM activities throughout the warranty timeframe are resolved Orthogonal Arrays since the ARTO system's degree of freedom is 242.
without costing anything to the customer. This calculated degree of freedom means it necessitates 242 experi-
Ilgin and Gupta (2011) conducted a study on measurable evaluation ments to account for 90 factors upon three separate levels.
of SEPs on disassembly line performance. The researchers showed that Furthermore, orthogonal array employs the assumption that no two
smart SEPs favorably address customer doubt for remanufactured pro- factors interact.
ducts. To assess this claim on ARTO, we constructed a simulation model Additionally, for the purpose of both validation and verification,
that represents the full recovery system. Then, we watched the system's animations of the simulation models were constructed along with
behavior under various experimental circumstances. We used ARENA multiple dynamic and counters plots. There were 2000 replications
program, Version 14.5, to build the discrete-event simulation models. during the course of six months (eight hours per shift, one shift per day,
We also used a three-level factorial design with 90 factors, with each and 5 days per week) utilized to conduct each experiment. The fol-
considered at 3 levels – low, intermediate, or high. These three-level lowing equation was used to calculate the profit according to arena
designs were intended to model potential curvature in the response models:
function and to address the case of nominal factors taking place at 3
levels. Tables 3–5 present the factors, factor levels, and parameters Profit = SR + CR + SCR HC BC DC DPC TC RMC
respectively. A full-factorial design with 90 factors at 3 levels calls for TPC WC PM (3)
substantial experimentation (viz., 8.727E+42). Thus, to lower the
number of required experiments for practicality purposes, we selected a where SR represents the total revenue generated by the product, in-
small set of all the possible combinations. This selection method of an cluding both component and material sales during the simulated run
experiment's number, known as a partial fraction experiment, yields the time; CR is the total revenue generated by the collection of EOL smart
greatest volume of information possible but requires the least number refrigerators during the simulated run time; SCR is the total revenue
of experiments possible. Taguchi (1986), established precise guidelines generated by selling scrap components during the simulated run time;
for such an experiment – a new method of managing the experimental HC is the total holding cost of products, components, material and EOL
design by using an exclusive set of arrays called orthogonal arrays smart refrigerators during the simulated run time; BC is the total
(OAs). These orthogonal arrays enabled conducting a minimal number backorder cost of products, components and material during the si-
of experiments - which can be determined using the degrees of freedom mulated run time; DC is the total disassembly cost during the simulated
approach. run time; DPC is the total disposal cost of components, material and

494
A.Y. Alqahtani et al. International Journal of Production Economics 208 (2019) 483–499

Table 6 (a)
Expected cost for remanufactured smart refrigerator and components for one-dimensional policies.
Components W FRW PRW FRW/PRW RFRW RPRW

RL = 1 RL = 2 RL = 3 RL = 1 RL = 2 RL = 3 RL = 1 RL = 2 RL = 3 RL = 1 RL = 2 RL = 3 RL = 1 RL = 2 RL = 3

Metal Cover 0.5 $5.96 $6.83 $5.46 $7.52 $8.63 $6.90 $6.69 $7.68 $6.14 $8.45 $9.69 $7.75 $9.49 $10.89 $8.70
1 $6.61 $7.49 $5.56 $8.34 $9.45 $7.02 $7.43 $8.41 $6.25 $9.37 $10.62 $7.88 $10.53 $11.93 $8.86
2 $9.92 $9.86 $5.72 $12.53 $12.45 $7.22 $11.15 $11.08 $6.43 $14.07 $13.98 $8.11 $15.81 $15.71 $9.11
Solenoid Valve 0.5 $5.85 $6.79 $5.45 $7.38 $8.57 $6.88 $6.57 $7.62 $6.12 $8.29 $9.62 $7.73 $9.31 $10.81 $8.68
1 $6.90 $7.28 $5.53 $8.71 $9.19 $6.98 $7.75 $8.18 $6.21 $9.78 $10.32 $7.84 $10.99 $11.60 $8.81
2 $9.81 $9.69 $5.69 $12.39 $12.22 $7.18 $11.02 $10.88 $6.39 $13.91 $13.73 $8.06 $15.63 $15.43 $9.06
Temperature 0.5 $2.93 $2.74 $2.80 $3.70 $3.46 $3.54 $3.29 $3.08 $3.15 $4.16 $3.89 $3.98 $4.67 $4.36 $4.47
Controls 1 $3.86 $4.86 $2.87 $4.87 $6.13 $3.62 $4.33 $5.46 $3.22 $5.47 $6.89 $4.07 $6.14 $7.74 $4.57
2 $5.32 $5.97 $2.96 $6.72 $7.54 $3.74 $5.98 $6.71 $3.33 $7.54 $8.47 $4.20 $8.48 $9.52 $4.72
Evaporator 0.5 $1.61 $1.58 $1.34 $2.03 $1.99 $1.69 $1.81 $1.77 $1.50 $2.28 $2.24 $1.90 $2.56 $2.51 $2.13
1 $2.33 $2.15 $1.48 $2.94 $2.71 $1.87 $2.61 $2.42 $1.66 $3.30 $3.05 $2.10 $3.71 $3.43 $2.36
2 $3.09 $3.06 $1.58 $3.90 $3.86 $1.99 $3.47 $3.44 $1.77 $4.38 $4.34 $2.24 $4.92 $4.87 $2.51
Motor 0.5 $6.10 $5.85 $5.72 $7.70 $7.38 $7.22 $6.85 $6.57 $6.43 $8.65 $8.29 $8.11 $9.72 $9.31 $9.11
1 $6.71 $6.26 $5.81 $8.46 $7.90 $7.34 $7.53 $7.03 $6.53 $9.51 $8.88 $8.24 $10.68 $9.97 $9.26
2 $9.43 $8.09 $5.88 $11.90 $10.21 $7.42 $10.60 $9.09 $6.60 $13.37 $11.48 $8.34 $15.02 $12.89 $9.36
Condenser 0.5 $1.91 $1.62 $1.58 $2.41 $2.05 $1.99 $2.15 $1.83 $1.77 $2.71 $2.30 $2.24 $3.05 $2.59 $2.51
1 $2.77 $2.33 $1.74 $3.50 $2.94 $2.19 $3.11 $2.61 $1.95 $3.93 $3.30 $2.46 $4.42 $3.71 $2.77
2 $3.20 $2.69 $1.83 $4.04 $3.40 $2.31 $3.60 $3.02 $2.06 $4.54 $3.82 $2.60 $5.10 $4.29 $2.92
Fan 0.5 $3.65 $3.09 $2.96 $4.60 $3.90 $3.74 $4.10 $3.47 $3.33 $5.17 $4.38 $4.20 $5.81 $4.92 $4.72
1 $5.05 $3.60 $3.04 $6.37 $4.54 $3.84 $5.67 $4.04 $3.42 $7.16 $5.11 $4.31 $8.04 $5.74 $4.85
2 $6.28 $5.02 $3.20 $7.92 $6.33 $4.04 $7.05 $5.64 $3.60 $8.90 $7.12 $4.54 $10.00 $7.99 $5.10
Aluminum 0.5 $0.97 $0.76 $0.54 $1.23 $0.97 $0.68 $1.09 $0.86 $0.61 $1.38 $1.08 $0.77 $1.55 $1.22 $0.86
Radiator 1 $1.50 $1.21 $0.65 $1.89 $1.53 $0.82 $1.68 $1.36 $0.73 $2.12 $1.72 $0.93 $2.39 $1.93 $1.04
2 $2.63 $1.75 $0.70 $3.32 $2.21 $0.88 $2.95 $1.97 $0.79 $3.73 $2.48 $0.99 $4.19 $2.79 $1.12
Compressor 0.5 $4.24 $3.97 $3.79 $5.35 $5.01 $4.79 $4.76 $4.46 $4.26 $6.01 $5.62 $5.38 $6.75 $6.32 $6.04
1 $5.46 $5.13 $4.11 $6.90 $6.47 $5.19 $6.14 $5.76 $4.62 $7.75 $7.27 $5.83 $8.70 $8.17 $6.55
2 $7.42 $6.71 $4.25 $9.37 $8.46 $5.37 $8.34 $7.53 $4.78 $10.53 $9.51 $6.03 $11.83 $10.68 $6.78
Smart 0.5 $89.80 $85.85 $85.10 $113.34 $108.35 $107.41 $100.89 $96.45 $95.61 $127.33 $121.73 $120.67 $143.05 $136.76 $135.57
Refrigerator 1 $92.81 $93.75 $88.78 $117.14 $118.33 $112.05 $104.27 $105.32 $99.74 $131.60 $132.94 $125.89 $147.85 $149.35 $141.43
2 $105.52 $84.32 $72.76 $107.94 $106.43 $91.84 $96.08 $94.74 $81.75 $121.27 $119.57 $103.18 $67.76 $66.81 $57.65

Table 6 (b)
Expected cost for remanufactured smart refrigerator and components for one-dimensional policies.
Components W RFRW/RPRW CSW CLW MBG

RL = 1 RL = 2 RL = 3 RL = 1 RL = 2 RL = 3 RL = 1 RL = 2 RL = 3 RL = 1 RL = 2 RL = 3

Metal Cover 0.5 $4.10 $4.70 $3.75 $4.72 $5.41 $4.33 $5.30 $6.08 $4.86 $4.20 $4.82 $3.85
1 $4.56 $5.15 $3.83 $5.24 $5.93 $4.40 $5.88 $6.66 $4.95 $4.66 $5.28 $3.92
2 $6.84 $6.80 $3.95 $7.86 $7.81 $4.53 $8.83 $8.78 $5.09 $7.00 $6.95 $4.03
Solenoid Valve 0.5 $4.03 $4.68 $3.75 $4.63 $5.38 $4.32 $5.20 $6.04 $4.85 $4.12 $4.79 $3.84
1 $4.75 $5.02 $3.80 $5.47 $5.77 $4.38 $6.14 $6.48 $4.92 $4.86 $5.13 $3.90
2 $6.76 $6.67 $3.92 $7.77 $7.67 $4.51 $8.73 $8.62 $5.06 $6.92 $6.83 $4.01
Temperature Controls 0.5 $2.02 $1.89 $1.93 $2.32 $2.17 $2.22 $2.61 $2.44 $2.50 $2.07 $1.93 $1.98
1 $2.66 $3.34 $1.97 $3.05 $3.85 $2.27 $3.43 $4.32 $2.55 $2.72 $3.43 $2.02
2 $3.66 $4.12 $2.05 $4.22 $4.73 $2.35 $4.74 $5.32 $2.64 $3.75 $4.21 $2.09
Evaporator 0.5 $1.11 $1.09 $0.93 $1.27 $1.25 $1.06 $1.43 $1.40 $1.19 $1.13 $1.11 $0.94
1 $1.61 $1.48 $1.02 $1.84 $1.70 $1.17 $2.07 $1.91 $1.32 $1.64 $1.52 $1.04
2 $2.13 $2.11 $1.09 $2.45 $2.42 $1.25 $2.75 $2.72 $1.40 $2.18 $2.16 $1.11
Motor 0.5 $4.20 $4.03 $3.94 $4.83 $4.63 $4.53 $5.43 $5.20 $5.09 $4.30 $4.12 $4.03
1 $4.62 $4.31 $4.01 $5.31 $4.96 $4.61 $5.97 $5.57 $5.18 $4.73 $4.42 $4.10
2 $6.50 $5.58 $4.05 $7.47 $6.41 $4.66 $8.39 $7.20 $5.23 $6.65 $5.71 $4.15
Condenser 0.5 $1.32 $1.12 $1.10 $1.51 $1.29 $1.25 $1.70 $1.45 $1.40 $1.35 $1.15 $1.11
1 $1.91 $1.61 $1.21 $2.20 $1.84 $1.38 $2.47 $2.07 $1.55 $1.95 $1.64 $1.22
2 $2.21 $1.85 $1.27 $2.54 $2.13 $1.45 $2.85 $2.40 $1.63 $2.26 $1.90 $1.29
Fan 0.5 $2.52 $2.13 $2.05 $2.89 $2.45 $2.35 $3.25 $2.75 $2.64 $2.57 $2.18 $2.09
1 $3.47 $2.49 $2.10 $4.00 $2.85 $2.41 $4.49 $3.20 $2.71 $3.56 $2.54 $2.15
2 $4.32 $3.45 $2.21 $4.97 $3.98 $2.54 $5.59 $4.47 $2.85 $4.43 $3.54 $2.26
Aluminum Radiator 0.5 $0.67 $0.52 $0.37 $0.77 $0.61 $0.43 $0.86 $0.68 $0.48 $0.69 $0.54 $0.38
1 $1.03 $0.84 $0.45 $1.19 $0.96 $0.52 $1.33 $1.08 $0.58 $1.06 $0.85 $0.46
2 $1.81 $1.22 $0.48 $2.08 $1.39 $0.56 $2.34 $1.56 $0.62 $1.85 $1.24 $0.49
Compressor 0.5 $2.92 $2.74 $2.62 $3.36 $3.14 $3.00 $3.77 $3.53 $3.37 $2.99 $2.80 $2.67
1 $3.75 $3.53 $2.84 $4.33 $4.06 $3.26 $4.86 $4.57 $3.66 $3.85 $3.62 $2.90
2 $5.12 $4.62 $2.93 $5.88 $5.31 $3.37 $6.61 $5.97 $3.79 $5.24 $4.73 $3.00
Smart Refrigerator 0.5 $55.01 $52.95 $52.49 $71.15 $68.02 $67.42 $79.93 $76.41 $75.75 $63.33 $60.54 $60.02
1 $57.22 $57.78 $54.75 $73.53 $74.28 $70.34 $82.61 $83.45 $79.02 $65.45 $66.12 $62.61
2 $66.46 $65.53 $56.63 $67.76 $66.81 $57.65 $76.13 $75.06 $64.77 $76.13 $75.06 $64.77

495
A.Y. Alqahtani et al. International Journal of Production Economics 208 (2019) 483–499

Table 8

$1,690,650.89
$4,282,157.60
$2,591,506.70
$1,009,952.03
ANOVA table and tukey pairwise comparisons for warranty cost.

$169,388.47

$361,244.66
$34,071.30

$68,400.78

$35,508.52
$10,140.15

$1944.97
ANOVA: Warranty Cost

13,460
MBG

Null All means are equal


hypothesis

Alternative At least one mean is different

$2,014,579.52
$3,906,807.98
$1,892,228.46
$1,193,194.43 hypothesis
$200,121.77

$426,787.72
$40,253.09

$80,811.20

$41,951.08
$27,444.09

$4016.15
Significance α = 0.05
19,547 level
CLW

SUMMARY

Models Count Sum Average StDev 95% CI


$1,793,197.33
$3,564,359.38
$1,771,162.04
$1,062,074.26
$178,130.38

$379,888.01

FRW 2000 28,756,990 14,378.50 506.17 (14356,


$35,829.67

$71,930.85

$37,341.07
$24,428.26

$3574.82

14401)
17,399

PRW 2000 36,295,776 18,147.89 512.42 (18125,


CSW

18170)
FRW/PRW 2000 32,307,233 16,153.62 518.75 (16131,
16176)
RFRW 2000 48,856,525 24,428.26 525.15 (24405,
$3,399,318.99
$3,906,807.98
$2,030,667.08

24451)
RFRW/RPRW

$507,488.98
$340,582.11

$726,339.10
$137,530.51

RPRW 2000 54,888,189 27,444.09 531.64 (27421,


$68,505.69

$71,395.46
$20,388.37

$3910.67

27467)
27,064

RFRW/RPRW 2000 40,776,732 20,388.37 538.20 (20365,


20412)
CSW 2000 22,784,042 11,392.02 544.84 (11368,
11416)
CLW 2000 25,596,884 12,798.44 551.57 (12774,
$3,025,767.76
$4,282,157.60
$1,256,389.84
$1,807,517.03

12823)
$303,155.54

$646,521.68
$122,417.28
$60,977.60

$63,549.81
$18,147.89

MBG 2000 20,280,303 10,140.15 565.27 (10115,


$3480.93

10165)
24,090
RFRW

ANOVA

Source of SS df MS F-Value P-value


Variation
$2,693,266.07
$3,906,807.98
$1,213,541.91
$1,608,888.95
$269,841.77

$575,475.40
$108,964.84

Model 5.60E+11 8 7.01E+10 246628.57 0


$54,276.77

$56,566.32
$16,153.62

$3098.41

Error 5.11E+09 17991 2.84E+05


21,442
RFRW

Total 5.66E+11 17999

Tukey Pairwise Comparisons


Grouping Information Using the Tukey Method and 95% Confidence
$2,133,863.33
$3,564,359.38
$1,430,496.05
$1,274,715.91

Model N Mean Grouping


$213,794.49

$455,946.73
$43,003.25

$86,332.38

$44,817.26
$12,798.44
FRW/PRW

$2454.86

MBG 2000 10,140.15 A


16,989
Results of performance measures for different models with warranty and PM.

CSW 2000 11,392.02 B


CLW 2000 12,798.44 C
FRW 2000 14,378.50 D
FRW/PRW 2000 16,153.62 E
PRW 2000 18,147.89 F
$2,397,303.00
$3,251,927.88
$1,432,088.11

RFRW/RPRW 2000 20,388.37 G


$854,624.87
$240,188.85

$512,236.39
Mean Value with Warranty and PM

RFRW 2000 24,428.26 H


$48,312.29

$96,990.69

$50,350.25
$14,378.50

$2757.93

RPRW 2000 27,444.09 I


19,086
PRW

Means that do not share a letter are significantly different.


$1,899,373.04
$2,966,882.35
$1,067,509.31
$1,134,637.35

EOL smart refrigerators during the simulated run time; TC is the total
$190,300.61

$405,842.73
$38,277.62

$76,845.32

$39,892.29
$11,392.02

testing cost during the simulated run time; RMC is the total re-
$2185.09

manufacturing cost of products during the simulated run time; TPC is


15,122
FRW

the total transportation cost during the simulated run time; PMC is the
total preventive maintenance cost during the simulated run time; and,
WC is the total warranty cost.
For every EOL smart refrigerator, there are three scrap forms that
Remanufacturing Cost
Performance Measure

Transportation Cost

can be recovered and sold. The Metal Cover and Pump are sold as
Number of Claims
Disassembly Cost

copper scrap, the Chassis and metal covers are sold as steel scrap, and
Backorder Cost

Warranty Cost

Total Revenue
Disposal Cost
Holding Cost

Agitators, Water & Drain Hoses, and Spin Tubs are sold as fiberglass.
Total Cost

The remaining components are categorized as waste. The revenue from


PM Cost
Table 7

Profit

steel, copper, and fiberglass scrap is calculated by multiplying their


weight in pounds by the units of scrap revenue produced by each metal

496
A.Y. Alqahtani et al. International Journal of Production Economics 208 (2019) 483–499

Table 9 Table 10
ANOVA Table and Tukey Pairwise Comparisons for number of claims. ANOVA table and tukey pairwise comparisons for preventive maintenance.
ANOVA: Warranty Claims ANOVA: Preventive Maintenance

Null All means are equal Null hypothesis All means are equal
hypothesis
Alternative At least one mean is different
Alternative At least one mean is different hypothesis
hypothesis
Significance α = 0.05 Significance α = 0.05
level level
SUMMARY SUMMARY

Models Count Sum Average StDev 95% CI Models Count Sum Average StDev 95% CI

FRW 2000 34,797,947 17,399 319.49 (17385, FRW 2000 6,196,821 3098.41 164.19 (3091,
17413) 3106)
PRW 2000 39,093,986 19,547 358.93 (19531, PRW 2000 7,149,633 3574.82 184.46 (3567,
19563) 3583)
FRW/PRW 2000 38,172,305 19,086 403.24 (19068, FRW/PRW 2000 5,515,852 2757.93 207.24 (2749,
19104) 2767)
RFRW 2000 48,179,362 24,090 453.03 (24070, RFRW 2000 8,032,303 4016.15 232.82 (4006,
24110) 4026)
RPRW 2000 54,127,426 27,064 508.96 (27042, RPRW 2000 6,961,860 3480.93 261.57 (3469,
27086) 3492)
RFRW/RPRW 2000 42,884,931 21,442 571.79 (21417, RFRW/RPRW 2000 7,821,348 3910.67 293.86 (3898,
21467) 3924)
CSW 2000 30,243,756 15,122 642.38 (15094, CSW 2000 3,889,946 1944.97 184.46 (1937,
15150) 1953)
CLW 2000 33,977,550 16,989 721.69 (16957, CLW 2000 4,370,186 2185.09 207.24 (2176,
17021) 2194)
MBG 2000 26,920,269 13,460 358.93 (13444, MBG 2000 4,909,715 2454.86 232.82 (2445,
13476) 2465)

ANOVA ANOVA

Source of SS df MS F-Value P-value Source of SS df MS F-Value P-value


Variation Variation

Model 2.97E+11 8 3.71E+10 148601.96 0 Model 9.093E+09 8 1.137E+09 23033.86 0


Error 4.49E+09 17991 2.50E+05 Error 887733999 17991 49343.227
Total 3.01E+11 17999 Total 9.98E+09 17999

Tukey Pairwise Comparisons Tukey Pairwise Comparisons


Grouping Information Using the Tukey Method and 95% Confidence Grouping Information Using the Tukey Method and 95% Confidence

Model N Mean Grouping Model N Mean Grouping


MBG 2000 13,460 A
CSW 2000 15,122 B CSW 2000 1944.97 A
CLW 2000 16,989 C CLW 2000 2185.09 B
FRW 2000 17,399 D MBG 2000 2454.86 C
FRW/PRW 2000 19,086 E FRW/PRW 2000 2757.93 D
PRW 2000 19,547 F FRW 2000 3098.41 E
RFRW/RPRW 2000 21,442 G RFRW 2000 3480.93 F
RFRW 2000 24,090 H PRW 2000 3574.82 F
RPRW 2000 27,064 I RFRW/RPRW 2000 3910.67 G
RPRW 2000 4016.15 G
Means that do not share a letter are significantly different.
Means that do not share a letter are significantly different.

type. In addition, the cost of disposal is calculated by multiplying the


waste weight by the unit disposal cost. The estimated time to retrieve 6.1. Remanufacturing warranty policies evaluation
information from smart sensors is set at 20 s per smart refrigerator. The
estimated transportation cost is set at $50 for each truck round-trip. The ARENA 14.5 program was utilized to obtain results to compute
Lastly, since levels of quality vary, then prices in the secondary market the expected number of failures and expected cost to the re-
of recovery product also vary. manufacturer. We evaluate different warranty periods as well as of-
fering a preventive maintenance policy during each period. For illus-
tration, we will discuss only the results of FRW in details where the rest
6. Results are in similar fashion.
Table 6 presents the expected cost for the remanufactured smart
The results from this study are organized into three sections. Section refrigerator and components for one-dimensional warranty policies. In
7.1 evaluates different warranty options to determine the most appro- Table 6, the remanufacturer's expected cost includes the expense of
priate policy that should be offered. Section 7.2 presents a measurable supplying the original item, Cs. Therefore, the estimated warranty cost
assessment of offering PM as part of a warranty package. Finally, is calculated by subtracting Cs from the expected cost to re-
Section 7.3 offers a quantitative assessment regarding the impact of manufacturer. For example, from Table 6, for W = 0.5 and RL = 1, the
SEPs on costs related to warranties and maintenance. warranty cost for smart refrigerator is |$89.80 – Cs| = |$89.80 -

497
A.Y. Alqahtani et al. International Journal of Production Economics 208 (2019) 483–499

$100.00| = $10.20 which is ([$10.20/$100.00] x 100) = 10.20% increase in total profit when the ARTO system is used.
saving of the cost of supplying the item, Cs, which is significantly less
than that $100.00, Cs. This savings might be acceptable, but the cor- 7. Conclusions
responding values for longer warranty costs are much higher. For ex-
ample, for W = 2 years and RL = 1, the corresponding percentage is Smart products, built with sensors embedded and network con-
([|$105.52 - $100.00|/$100.00] x 100) = 5.52% of the total cost extra nectivity to enable the collection and exchange of data, utilize sensors
for offering FRW on remanufactured smart refrigerator with 1 year that are implanted into products during production. These sensors are
remaining life. used for remanufacturers to predict an optimal warranty policy and
time period that should be offered to customers who purchase re-
6.2. Preventive maintenance evaluation manufactured components and products. Sensor-provided data can help
to evaluate the overall condition of a product, as well as the remaining
Pairwise t tests were carried out for each performance measure to lives of product components, prior to offering a warranty. This data can
examine the impact of PM on warranty cost. Table 7 presents costs for be used to lower the number of warranty claims, identify the most ef-
warranty models with PM respectively. According to the data presented fective preventive maintenance (PM) policy, and to prevent extra ex-
in these tables, PM brings about notable savings in holding, backorder, penses for the remanufacturer. In this paper, the one-dimensional Free
disassembly, disposal, remanufacturing, transportation, warranty, PM Replacement Warranty (FRW), Pro-Rata Warranty (PRW) Renewable
costs and number of warranty claims. Furthermore, SEPs allow for FRW, Renewable PRW, Cos sharing warranty (CSW), cost limit warrant
major increases in total revenue and profit. According to Table 7, of- (CLW), money back guaranty (MBG), and combination FRW/PRW or
fering PM allows the remanufacturer to achieve savings of 18%, 21%, RFRW/RPRW policies’ costs for remanufactured products and compo-
19% and 18% in total cost for Conventional, SEM with FRW, SEM with nents were assessed and evaluated by offering PM for different periods.
FRW, and SEM with FRW respectively. Additionally, this paper examined the effect of offering individual
The lowest warranty average value and the quantity of claims warranty options for disassembled parts as well as sensor-embedded
during the covered period for remanufactured smart refrigerators across remanufactured products. This paper also addressed the impact of
all policies are $10,140.15 and 13,460 claims respectively for the sensor embedded products on the cost of warranties. Lastly, this paper
Sensor Embedded Model with MBG policy, and $1944.97 PM cost for presented and analyzed a case study involving various simulation
the SEPs with CSW policy. In contrast, the RPRW has the worst values conditions to illustrate the applicability of the model.
in a number of categories, including PM costs and the number of
warranty claims during the covered timeframe. References

6.3. Sensor embedded evaluation Alqahtani, A.Y., Gupta, S.M., 2017a. Warranty cost analysis within sustainable supply
chain. In: Akkucuk, U. (Ed.), Ethics and Sustainability in Global Supply Chain
Management. IGI Global, Hershey, PA, pp. 1–25.
Pairwise t tests were carried out for each performance measure to Alqahtani, A.Y., Gupta, S.M., 2017b. Optimizing two-dimensional renewable warranty
examine the effect of SEPs on warranty cost as well. Tables 8–10 pre- policies for sensor embedded remanufacturing products. J. Ind. Eng. Manag. 10 (2),
73–89.
sent a ninety-five percent confidence interval, t value and p value for Ben Mabrouk, A., Chelbi, A., Radhoui, M., 2016. Optimal imperfect preventive main-
each test. Based upon the data displayed on these tables, SEPs achieve a tenance policy for equipment leased during successive periods. Int. J. Prod. Res.
noteworthy savings in holding, backorder, disassembly, disposal, 1–16.
Ben-Daya, Mohamed, Hassini, Elkafi, Bahroun, Zied, 2017. Internet of things and supply
testing, remanufacturing and transportation costs. Furthermore, SEPs chain management: a literature review. Int. J. Prod. Res. https://doi.org/10.1080/
provide measurable increases in both total revenue and profit. 00207543.2017.1402140.
Blischke, W., 1995. Product Warranty Handbook. CRC Press.
The MINITAB-17 program was utilized to conduct one-way analyses Blischke, W. (Ed.), 1993. Warranty Cost Analysis. CRC Press.
of variance (ANOVA) and Tukey pairwise comparisons for the findings Bouzon, M., Spricigo, R., Rodriguez, C.M.T., de Queiroz, A.A., Miguel, P.A.C., 2015.
in this study. Additionally, ANOVA was used to locate measurable Reverse logistics drivers: empirical evidence from a case study in an emerging
economy. Prod. Plann. Contr. 26 (16), 1368–1385.
differences between the warranty costs, number of claims and PM costs Edwards, P., Peters, M., Sharman, G., 2001. The effectiveness of information systems in
for the four different models viz., SEPs with FRW, SEPs with PRW, SEPs supporting the extended supply chain. J. Busin. Logis. 22 (1), 1–27.
with FRW/PRW, SEPs with RFRW, SEPs with RPRW, SEPs with RFRW/ Fang, C., Liu, X., Pardalos, P.M., Pei, J., 2016a. Optimization for a three-stage production
system in the internet of things: procurement, production and product recovery, and
RPRW, SEPs with CSW, SEPs with CLW and SEPs with MBG, while the acquisition. Int. J. Adv. Manuf. Technol. 83 (5–8), 689–710.
Tukey pairwise comparisons identified similarities and differences Fang, C., Liu, X., Pei, J., Fan, W., Pardalos, P.M., 2016b. Optimal production planning in a
hybrid manufacturing and recovering system based on the internet of things with
among models. The data presented in Table 8 shows that a notable closed loop supply chains. Oper. Res.: Int. J. 16 (3), 543–577.
difference in the cost between different warranty policies. Furthermore, Finkenzeller, K., 2003. RFID Handbook: Fundamentals and Applications in Contact-less
Tukey test shows that all the models are different and the SEP model Smart Cards and Identification. John Wiley, England.
Gal-Or, E., 1989. Warranties as a signal of quality. Canadian J. Econ. 50–61.
with MGB policy has the lowest warranty cost. In addition, Table 9 Garg, A., Deshmukh, S.G., 2006. Maintenance management: literature review and di-
provides evidence of drastic differences in the quantity of warranty rections. J. Qual. Maintenance Eng. 12 (3), 205–238.
Gnimpieba, Z.D.R., Nait-Sidi-Moh, A., Durand, D., Fortin, J., 2015. Using Internet of
claims between varying warranty policies. The fewest number of claims
Things technologies for a collaborative supply chain: application to tracking of pallets
is connected with the MBG policy. Lastly, Table 10 demonstrates the and containers. Procedia Comput. Sci. 56, 550–557.
remarkable difference in PM costs between various warranty policies. Gungor, A., Gupta, S.M., 1999. Issues in environmentally conscious manufacturing and
product recovery: a survey. Comput. Ind. Eng. 36 (4), 811–853.
Tukey test shows that all models are different, except for RFRW and Gungor, A., Gupta, S.M., 2002. Disassembly line in product recovery. Int. J. Prod. Res. 40
PRW. There are no significant differences between them in terms of PM (11), 2569–2589.
cost and between RFRW/RPRW and RPRW. The SEP model with CSW Gupta, S.M., Ilgin, M.A., 2018. Multiple Criteria Decision Making Applications in
Environmentally Conscious Manufacturing and Product Recovery. CRC Press, Boca
policy has the lowest cost. Overall, these findings are useful for se- Raton, Florida ISBN: 978-1498700658.
lecting the most economical warranty policy associated with embed- Gupta, S.M., 2013. Reverse Supply Chains: Issues and Analysis. CRC Press.
Gupta, S.M., Lambert, A.F. (Eds.), 2007. Environment Conscious Manufacturing. CRC
ding sensors in smart refrigerators. Press.
Sensor value was calculated for the smart refrigerator by dividing Heal, G., 1977. Guarantees and risk-sharing. Rev. Econ. Stud. 549–560.
the difference in total profit by the total number of sensors used in the Ilgin, M.A., Gupta, S.M., 2010a. Comparison of economic benefits of sensor embedded
products and conventional products in a multi-product disassembly line. Comput.
proposed system. This value is equal to $61.23 which means that the Ind. Eng. 59 (4), 748–763.
price of a sensor should not exceed $61.23 for it to be viable and Ilgin, M.A., Gupta, S.M., 2010b. Evaluating the impact of sensor-embedded products on
profitable if implemented in refrigerator. Most profitable experiment the performance of an air conditioner disassembly line. Int. J. Adv. Manuf. Technol.
53 (9–12), 1199–1216.
had $64.66 as sensor value. This showed that there is a significant

498
A.Y. Alqahtani et al. International Journal of Production Economics 208 (2019) 483–499

Ilgin, M.A., Gupta, S.M., 2010c. Environmentally conscious manufacturing and product decentralized supply chain with reverse logistics operations. Int. J. Prod. Econ. 136
recovery (ECMPRO): a review of the state of the art. J. Environ. Manag. 91 (3), (2), 366–377.
563–591. Paksoy, T., Karaoglan, I., Gokcen, H., Pardalos, P.M., Torgul, B., 2016. An experimental
Ilgin, M.A., Gupta, S.M., 2011. Performance improvement potential of sensor embedded research on closed loop supply chain management with internet of things. J. Econ.
products in environmental supply chains. Resour., Conserv. Recycl. 55 (6), 580–592. Bibliogr. 3 (1S), 1–20.
Ilgin, M.A., Gupta, S.M., 2012. Remanufacturing Modeling and Analysis. CRC Press. Saidi-Mehrabad, M., Noorossana, R., Shafiee, M., 2010. Modeling and analysis of effective
Ilgin, M.A., Gupta, S.M., Battaïa, O., 2015. Use of MCDM techniques in environmentally ways for improving the reliability of remanufactured products sold with warranty.
conscious manufacturing and product recovery: state of the art. J Manuf. Syst. 37, Int. J. Adv. Manuf. Technol. 46 (1–4), 253–265.
746–758. Shafiee, M., Chukova, S., 2013. Maintenance models in warranty: a literature review. Eur.
Kijima, M., 1989. Some results for repairable systems with general repair. J. Appl. J. Oper. Res. 229 (3), 561–572.
Probability 89–102. Shafiee, M., Chukova, S., Yun, W.Y., Akhavan Niaki, S.T., 2011a. On the investment in a
Kijima, M., Morimura, H., Suzuki, Y., 1988. Periodical replacement problem without reliability improvement program for warranted remanufactured items. IIE Trans. 43
assuming minimal repair. Eur. J. Oper. Res. 37 (2), 194–203. (7), 525–534.
Kim, C.S., Djamaludin, I., Murthy, D.N.P., 2004. Warranty and discrete preventive Shafiee, M., Finkelstein, M., Chukova, S., 2011b. On optimal upgrade level for used
maintenance. Reliability Eng. Syst. Saf. 84 (3), 301–309. products under given cost structures. Reliability Eng. Syst. Saf. 96 (2), 286–291.
Kim, D.K., Lim, J.H., Park, D.H., 2011. Optimal maintenance policies during the post- Sharma, A., Yadava, G.S., Deshmukh, S.G., 2011. A literature review and future per-
warranty period for second-hand item. In: The 2011 International Conference on spectives on maintenance optimization. J. Qual. Maintenance Eng. 17 (1), 5–25.
Quality, Reliability, Risk, Maintenance, and Safety Engineering, pp. 446–450. Shin, S., Eksioglu, B., 2015. An empirical study of RFID productivity in the US retail
Lambert, A.J.D., Gupta, S.M., 2005. Disassembly Modeling for Assembly, Maintenance, supply chain. Int. J. Prod. Econ. 163, 89–96.
Reuse, and Recycling. CRC Press, Boca Raton, FL. Soberman, D.A., 2003. Simultaneous signaling and screening with warranties. J. Market.
Leung, J., Cheung, W., Chu, S.C., 2014. Aligning RFID applications with supply chain Res. 40 (2), 176–192.
strategies. Inf. Manag. 51 (2), 260–269. Spence, M., 1977. Consumer misperceptions, product failure and producer liability. Rev.
Liao, B.F., Li, B.Y., Cheng, J.S., 2015. A warranty model for remanufactured products. J. Econ. Stud. 561–572.
Ind. Prod. Eng. 32 (8), 551–558. Sun, C., 2012. Application of RFID technology for logistics on internet of things. AASRI
Lutz, N.A., Padmanabhan, V., 1995. Why do we observe minimal warranties? Marketing Procedia 1, 106–111.
Sci. 14 (4), 417–441. Taguchi, G., 1986. Orthogonal Arrays and Linear Graphs. American Supplier Institute,
Martorell, S., Sanchez, A., Serradell, V., 1999. Age-dependent reliability model con- Inc., Dearborn, Ml.
sidering effects of maintenance and working conditions. Reliability Eng. Syst. Saf. 64 Theorin, A., Bengtsson, K., Provost, J., Lieder, M., Johnsson, C., Lundholm, T.,
(1), 19–31. Lennartson, B., 2017. An event-driven manufacturing information system archi-
Moyer, L.K., Gupta, S.M., 1997. Environmental concerns and recycling/disassembly ef- tecture for Industry 4.0. Int. J. Prod. Res. 55 (5), 1297–1311.
forts in the electronics industry. J. Electron. Manuf. 7 (01), 1–22. Wang, H., 2002. A survey of maintenance policies of deteriorating systems. Eur. J. Oper.
Murthy, D.P., Blischke, W.R., 2006. Warranty Management and Product Manufacture. Res. 139 (3), 469–489.
Springer Sinence & Business Media. Wang, S., Wan, J., Zhang, D., Li, D., Zhang, C., 2016. Towards smart factory for Industry
Naini, S.G.J., Shafiee, M., 2011. Joint determination of price and upgrade level for a 4.0: a self-organized multi-agent system with big data based feedback and co-
warranted remanufactured product. Int. J. Adv. Manuf. Technol. 54, 1187–1198. ordination. Comput. Netw. 101, 158–168.
Nakagawa, T., 2006. Maintenance Theory of Reliability. Springer Science & Business Wollschlaeger, M., Sauter, T., Jasperneite, J., 2017. The future of industrial commu-
Media. nication: automation networks in the era of the internet of things and industry 4.0.
Nakagawa, T., 2008. Advanced Reliability Models and Maintenance Policies. Springer IEEE Ind. Electron. Mag. 11 (1), 17–27.
Science & Business Media. Yazdian, S.A., Shahanaghi, K., Makui, A., 2014. Joint optimisation of price, warranty and
Naskar, S., Basu, P., Sen, A.K., 2017. A literature review of the emerging field of IoT using recovery planning in remanufacturing of used products under linear and non-linear
RFID and its applications in supply chain management. In: The Internet of Things in demand, return and cost functions. Int. J. Syst. Sc. 1–21.
the Modern Business Environment. IGI Global, pp. 1–27. Yeh, R.H., Lo, H.C., Yu, R.Y., 2011. A study of maintenance policies for second-hand
Nativi, J.J., Lee, S., 2012. Impact of RFID information-sharing strategies on a products. Comput. Ind. Eng. 60 (3), 438–444.

499

Vous aimerez peut-être aussi