Consumers might be willing to repair their broken devices as long as the associated repair costs do not exceed an undesirable threshold. However, in many cases, the technological obsolescence actuates consumers to retire old devices and replace them with new ones rather than extending the product lifecycle through repair. In this paper, we aim to investigate the impact of components' deterioration profiles and consumers' repair decisions on the lifespan of devices, and then assesse the anticipated life cycle environmental impacts. A Monte Carlo simulation is developed to estimate the life cycle characteristics such as the average lifespan, the number of failed components' replacement, and the total repair cost per cycle for a laptop computer. The lifecycle characteristics estimated from simulation model further have been used in a life cycle assessment (LCA) study to quantify the environmental impact associated with different design scenarios. The results reveal the impact of product design as well as consumers' repair decisions on the product lifespan and the corresponding environmental impacts.
Consumers stop using electronic products for many reasons such as technical and functional obsolescence. In the case of functional obsolescence, if the cost to repair a product is greater than the replacement cost, consumers may decide to replace the product , despite their willingness to repair it . A significant number of properly working devices are also discarded and replaced annually because owners perceive products to be technically obsolete . Consequently, either scenario creates massive streams of end-of-use/life (EoU/L) products  that results in significant environmental burden and pollution problems.
These consumer behaviors speed up the rate of EoU/L generation; however, consumer behavior is inextricably linked to manufacturer's planned obsolescence along with the release time of new technologies . Regardless of the reasons, the environmental burden is a point of issue in “throwaway societies” where durable products are short-lived due to overconsumption . Therefore, it would be beneficial if all factors related to consumer behavior and product design features be considered at the same time to study the associated environmental impacts. In other words, it is important to know how consumers' propensities for repair may affect the life cycle of a product. The impact of consumers' decision on extending products lifecycles through repair and reuse has not been sufficiently explored in the literature. To overcome this gap, we will consider the consumers' repair behavior in the life cycle analyses, in addition to the design characteristics. Again, it is worth noting that manufacturers' technology management strategies (e.g., planned obsolescence) may correlate with consumers' repair behavior. Hence, it is important to consider the existing correlation despite the difficulty of capturing it.
The lifespan of most consumer electronics is becoming shorter and shorter  (i.e., the average lifespan of mobile phones was about 17.3 months in 2015 compared to 18.4 months in 2002). It is estimated that 72 million tons of electronic waste will be produced in 2017, 33% more than what was generated in 2012 . Although there are no accurate data on the functionality status of unwanted products, it is estimated that many of them are still functional or need some minor repairs .
We aim to provide an analytical framework to study the environmental impacts associated with the life cycle of a device by combining consumers' repair behavior with the preknown deterioration process. To further understand the linkage between product design and consumer behavior, we ask the following questions: Considering the point that consumers may allocate limited budget for repair activities, how long would be the average lifespan of a product with specific design characteristics? What would be the impact of technological obsolescence perceived by consumers on the lifespan? And finally, what are the anticipated environmental impacts across the life cycle of a product? In the proposed model, consumers may keep using a device until they perceive that there is no benefit to repair it or it is sufficiently obsolete from a technical standpoint. Three factors are considered in this study to obtain more accurate results: (1) consumers' expectations of the total repair cost relative to the initial price of a product over the use cycle (cumulative repair cost ratio (CRCR)) before making a replacement decision, (2) perception of technical obsolescence of the device, and (3) the time-to-failure for every single component. These are not the only factors contributing to consumers' repair decisions. Factors such as values, beliefs, norms, rate of technology progress, availability of repair infrastructure, and consumer-product attachment may also influence consumers' repair decisions. Most of these factors are particular to given demographics, products and brands in which it is difficult to be taken into account. Therefore, we have only considered the above-mentioned three factors. Finally, a Monte Carlo simulation framework is implemented to obtain the required inputs for the environmental impact analyses part.
The proposed analytical framework assists designers to select the best set of technical features that minimize the anticipated environmental impacts as much as possible over the entire product life cycle by allowing them to consider consumers' postpurchase behavior in the design process. It should be noted that design decision making in companies are not only based on technical features. In fact, many factors should be considered in this highly strategic decision. Our proposed approach can be incorporated in a multilevel design problem, in which other factors such as price, market, and regulations are considered in an integrated analytical framework. Especially as for now companies have not yet integrated environmental impact estimation due to the lack of data and difficulty of estimating them. The proposed model helps companies simulate consumers' behavior and evaluate different design and economic policies. In case of the lack of data, consumers' repair behavior can also be evaluated by survey-based methods.
The remaining sections are organized as follows: Section 2 briefly summarizes the relevant literature that supports our approach. In Sec. 3, the consumer repair behavior is characterized and combined with the deterioration process of a consumer electronic device. The life cycle of a product is simulated applying the Monte Carlo method, and the results are provided as a numerical example in Sec. 4. The total life cycle environmental impact is calculated for different product design scenarios using simapro software in Sec. 5. Finally, the limitations of the current study and future work are discussed in Sec. 6.
Relevant literature includes studies on the environmental impact of product life cycles covering the following two areas: (1) technical aspects of the product life cycle and (2) life cycle analyses of consumer behavior.
In the first category of literature, studies attempt to assess, control, and mitigate environmental impacts by developing innovative technical solutions for different stages of a product's life cycle. Developing ecodesign tools , capturing environmental impact of manufacturing processes , advancing green energy technologies , and proper recovery of EoU/L items  are just a few of the many contributions in this area. We survey the literature related to these design concepts as other aspects are beyond the scope of this paper.
The main questions answered in the literature so far include “How can the environmental burden associated with a product be evaluated at the early phase of design?,” “How can a greener product be designed while delivering the expected functionality?,” and “How to recover EoU/L products that exerts less impact on the environment?” To resolve the conflict between the product functionality and environmental impacts, Fitzgerald et al.  proposed an analogical approach to design more ecofriendly products without losing major functionality. In another study, Kishita et al.  explained how to reduce the environmental impact of a product by applying a weighted checklist assessment method in the design phase. There are some recent studies that develop methodological tools to assess and improve the sustainability of design alternatives (see paper by Bovea and Pérez-Belis ). In all the above-mentioned developed methods, the environmental concerns are considered in the early stage of product design. In other words, the product is designed for environment.
Since a main part of this paper is devoted to understanding the role of product deterioration and repairability in the replacement decision process, the relevant literature on these matters is henceforth reviewed. Although a general prescription for optimizing product lifespan and minimizing environment impact is not available, analytical tools for a particular product type can be used for these ends. For instance, Kim et al.  proposed a dynamic model to obtain the optimal lifespans by comparing the environmental scores in two scenarios: continuation of use and replacement. However, making this kind of decision is largely dependent on product design.
In terms of product design, modular design is discussed in the literature as a promising strategy to increase the sustainability of products [18,19]. Although serviceability and upgradability of modular design are more achievable, Agrawal and Ülkü  showed that modular design may decrease the environmental impact if the introduction rate of new modules is controlled somewhat. Reliability is also a determinant factor in quantifying the lifespan of a product. Mazhar et al.  developed a two-stage approach to estimate the remaining useful life of products to facilitate replacement decisions.
In addition to the above-mentioned design-related research, the sustainability level of postsale services (i.e., proper treatment of failed parts, use of nontoxic glues, lubricants, fluids, and solvents in the repair process ) should be considered in assessing the environmental impact associated with the product life cycle.
The second category of literature is concerned with how consumer behavior may affect the life cycle of a product. For example, consumers' willingness to purchase ecofriendly products , their lifestyle and consumption behavior , and their decisions in handling EoU/L items  are linked to the life cycle assessment studies.
Generally, greater repairability, serviceability, and reusability of products encourage people to delay a replacement decision . Consumers may be inclined to repair their failed products as long as the associated cost is reasonable. On average, the amount that consumers spend on repairing a product is 20% of the replacement cost . On the other hand, a still-functional device may be discarded when it becomes or is perceived to be technically obsolete. As such, the entire life cycle of a product and the chance that the product is being reused for another use cycle are closely tied to consumer behavior. In one of our previous studies , we observed that the reusability of the Li-ion laptop batteries is highly affected by consumer usage behavior within the initial first life cycles.
In this context, another question may rise: what should be done with retired products, assuming the mandated responsibility of producers through the entire life cycle? King et al.  concluded that to reduce environmental impacts, the remanufacturing (the recovery process of a used product to restore it as-good-as new one .) of waste items is superior to extending product lifespan through repair. Remanufacturing is ideal given the variability in the quality of product repaired. Remanufactured products can essentially be remarketed as new.
In concluding the literature review, there is a research gap in understanding of how consumer postpurchase behavior (e.g., their repair propensity) and expected lifetime of critical components may affect the lifespan of a product and, consequently, the associated environmental impacts at the same time. Although the design for environment concept is not new and has been already applied in the literature [30,31], the role of consumer repair behavior is not yet well-studied.
In a theoretical study , Lemke and Luzio surveyed a sample of consumers to capture their opinions about green products and further to drive market-driven design strategies. Interestingly, extending warranty time was demanded by consumers as a purchase decision criterion. In an LCA study, Iraldo et al.  showed that improving the durability of products is economically preferred by consumers. However, whether the associated environmental impact is also reduced or not depends on the production, usage, and waste recovery stages. The role of usage and disposal behaviors of consumers in limiting the lifespan of products has also been studied by Krystofik et al. . However, there is a need to build an integrated analytical model that enables us to estimate the lifecycle data and particularly capturing consumer behavior, beyond just implementing LCA for specific case studies. In this model, all consumers–manufacturers interactions can be tested simultaneously in order to design greener products. Particularly, the focus of this paper is on understanding the repair behavior and studying the impact of influencing factors such as perceived technical obsolescence and repair costs on consumers' decision to extend product lifecycle.
In summary, measuring the degree of sustainability requires an understanding of how willing the consumers are to extend the product lifespan through repair services and how fast the product deteriorates. Thus, the objective of the present study is to assess the impact of consumers' repair decisions on a product's life cycle and the subsequent environmental effects. simapro software is used to assess the associated impact. The results provide new insights on product design for environment and consumer-oriented design concepts.
LCA Model Description
Life cycle assessment (LCA) is an analytical technique to capture all environmental impacts caused by the lifespan of a product from cradle to grave. According to the most developed standards related to environment management (e.g., ISO 14000), an LCA study consists of four main stages: (1) definition of goal and scope, (2) life cycle inventory, (3) life cycle impact assessment, and (4) interpretation of results . After clearly defining the scope and goal of the LCA study, all inputs and outputs that flow through the product life cycle stages are recognized and inventoried—including raw materials, energy, water, and chemical pollutants. The environmental impacts of the flows are quantified based on the selected characterization indices and model. Finally, the obtained results are interpreted to find potential life cycle improvements like reducing waste and environmental burden.
where β and γ are shape parameters of the distribution, and the mean threshold is μA. It should be noted that there is no data on the consumers' willingness to pay for repair in order to fit an exact distribution. Therefore, we assume that beta distribution could be a good option as it is flexible enough to characterize the consumers' behavior and has been used by other researchers [36,37].
where γCi is the shape parameter, and λCi is the scale parameter of the distribution. Weibull distribution is widely used to characterize the reliability of a system and is highly flexible to capture a variety of failure behavior .
Figure 1 schematically demonstrates the proposed life cycle model. One may ask question that “how does the random nature of consumer behavior affect the lifespan of products?” To clarify this point, consider two consumers where one has more propensity to repair a product (a higher value of A) compared with the other consumer. In the case that both consumers have almost the same products in terms of the deterioration process, the consumer who has less propensity for repair will retire the product sooner than the other consumer. There is another example that may show the randomness in consumers' behavior. Assume a consumer who would not spend much money to repair a product. Then, there might be a chance that the product would be retired even after occurrence of the first failure if the cost of repair is more than the consumer's expectation.
Simulation Implementation of Life Cycle Model
In this section, a numerical example is provided to illustrate the proposed model. Table 1 summarizes the data for the simulation analysis. The replacement costs for a spare component—including the labor cost—are arranged in descending order. To calculate the repair cost ratio (α), the replacement costs of components are divided by the replacement cost of the device. The laptop is priced at $1650, and it is assumed that the salvage price for a relatively old laptop is negligible. Therefore, a consumer has to pay almost $1650 to replace the old laptop with a new one.
Let's assume that the psychological and technical obsolete time TO follows a log-normal distribution with mean 3.77 yr and standard deviation 0.56 yr (and hence μ = 1, σ = 0.2, and η = 1). This is not an unrealistic assumption as the previous literature estimated the lifespan of computers to be between 3 and 5 yr . However, it is necessary to collect real data to better estimate the parameters of the distribution for a specific type of products. Therefore, the obsolete times are randomly generated in that range. As discussed in Sec. 3, the costs for repair that consumers are willing to pay for a device along the lifespan is bounded. A beta distribution with mean 20%  and standard deviation 12% represents the threshold for CRCR.
The parameters of the time-to-failure distributions are randomly generated because we do not have access to the actual failure data. Such after-sale data are usually available to manufactures and are not necessarily publicized. To derive the results of the Monte Carlo simulation in Matlab® 2013, 1000 samples have been used as a reasonable choice to estimate the 95% confidence interval for characteristics . To reduce the length of confidence interval, the simulation should be run for more samples. This may increase the simulation time.
Table 2 shows the obtained results. The average usage time of laptops is 3.25 yr. The distributions for both obsolete and usage times are shown in Fig. 2. Here, the term “usage time” refers to the time interval that a laptop has been used by its owner before being retired due to physical or technical obsolescence. It should be noted that the average usage time is significantly lower than (t-value = 17.48 and p-value = 0.00) the average obsolete time (3.77 yr). In fact, consumers are more sensitive to the physical obsolescence, rather than the technological according to the setting of our case study.
The average number of component replacements ranges from 0.17 to 0.51 units. For example, consumers replace component C1 for 0.22 times on average (i.e., 220 consumers may replace component C1 over the product lifespan within a population size of 1000). In Sec. 5, we use these numbers in the environmental impact analyses.
Among the retired laptops, some of them are still functional, and the rest are failed. Figure 2 illustrates the failed and still-working laptops after replacement decision. Among 1000 samples, 499 are failed with a mean age of 3.61 yr and a standard deviation of 0.49 yr. The mean age of 501 still-working laptops is 3.93 yr with a standard deviation of 0.58. Thus, about 50% of laptops are retired, while there is a potential to be reused for another use cycle without further recovery actions such as repair. The rest of laptops can be collected for remanufacturing process or reuse of functional components.
Sensitivity of Simulation to Consumer Behavior.
As expected, the results of the simulation are sensitive to the parameters of the model. The way that a product is designed in terms of reliability and repairability, the manufacturer's policy to supply the spare parts, and consumer usage and repair behaviors may change the results.
Consumers' willingness to repair is affected by their emotional attachment to the products  and their personal characteristics . Let's assume that consumers become greener (less green) by showing more (less) propensity for repair. A sensitivity analysis is done for the change in the value of repair threshold A. In other words, it is important to know how long the product lifespan can be extended if repair is more attractive to consumers.
Figures 3 and 4 show the results of sensitivity. The baseline value of the average threshold (μA) for CRCR is 20% . The horizontal axis shows the percentage change in the average threshold. For example, 40% indicates that the average threshold for CRCR increases from 20% to 28%. The linear increase in the average threshold (μA) is represented by the black line in Fig. 3. The lower line shows the mean cumulative repair cost ratio (μCRCR). Moreover, Fig. 4 shows that the average usage time of computers increases for the same reason. In this figure, the bounds on the usage time are calculated over the population. The average usage time, however, is considerably affected by consumers' perception about the technological and psychological obsolescence. That is why the gap between two curves increases as the average threshold value increases, and the lifespan trend (μCRCR) converges to the mean obsolete time. In fact, the technical obsolescence is a limiter for restricting the monetary value spent on repair activities over the lifespan of a device. These figures reveal that how consumers' repair behaviors and technical characteristics can affect the life cycle of a product together.
Sensitivity of Simulation to Spare Part Cost.
The spare parts are usually supplied by providers for a limited amount of time. Therefore, the needed parts may be found from the discarded devices or purchased at higher prices than usual. A time-consuming repair process and limited access to the repair shops can also cause inconvenience to consumers .
Figure 5 illustrates the usage time by changing the cost of spare parts. The average usage time decreases by increasing the cost of spare parts. As seen in this figure, the usage time converges to the warranty time offered by manufacturers. It means that consumers may continue to use a device as long as the maintenance services are covered by the manufacturer for free. Hence, the used devices may not be repaired anymore once consumers observe even minor failures. In other words, it is perceived that the warranty time is a clue for understanding the obsolescence mechanism of products. In other words, a product with a relatively short warranty time may become rapidly obsolete.
Environmental Impacts Assessment
If the quality of retired products is relatively low, then the only economic treatment option is to properly dismantle the product and retrieve the precious materials. Otherwise, the retired product can be reused for another use cycle after remanufacturing or refurbishing operations . In this paper, the manufacturer offers a long-term repair service to the consumers. After retirement, the EoU/L laptops are returned to the manufacturer for material recovery. Let's assume that there are two design options. In Table 3, the technical characteristics of design option D1 and D2 are summarized. The level of deterioration rate (L) for different components is given in the table, in which L1 represents the components that deteriorate faster compared with the components with deterioration rates of L2 and L3.
As stated in the model description, the failed components are replaced with new ones for free over the warranty time. RWi represents the number of replacement during warranty time for component Ci. After the warranty time, the consumers have to pay for the repair services. RLi represents the number of replacements after the warranty time. Finally, the retired laptop is recycled at the end of the life cycle. RTi shows the average number of replacements for the entire life cycle.
Figures 6–8 show the results of simulation. Although the laptop is used for almost the same time (3.10 versus 3.06 yr, p-value = 0.218), the average number of replacements for some components is significantly different before and after warranty time.
However, the numbers for C1 in D2 are 0.03 and 0.19, respectively, and are significantly different (p-value = 0.006 and 0.000). Table 4 summarizes the simulation results.
Now, we have enough information to assess the environmental impact associated with each design. SimaPro® 8 2014 software (PRé Consultants, Amersfoort, The Netherlands) is used as a framework to calculate environmental impacts. The LCA software packages such as simapro may not have information for all materials, components, modules, manufacturing processes, and recovery options that we are looking for. As a solution to deal with the lacking of data, we have selected the most similar elements and processes to the preset ones. An impact assessment method, named ReCiPe Endpoint, is used in simapro. In this method, the environmental impacts are translated into 17 environmental themes ranging from human toxicity level to fossil depletion . The weights of components are collected from the product description provided by the manufacturer. In the usage stage, it is assumed that the laptop is on active mode for 5.5 h/day, 240 days a year. For the EoU/L treatment, a waste scenario for durable goods is considered, in which the materials such as steel, ferro metals, mixed plastic, polypropylene, and polyethylene terephthalate are separated from the used products. After separation, the remaining will end up in landfills or incineration centers.
Figure 9 shows the environmental impact associated with the production stage of all components needed during the lifespan of design 1, D1. In this figure, there are 17 factors that represent the environmental impact. For example, the agricultural land occupation and human toxicity scores of central processing unit production are 6.47% and 10.72%, respectively. The results vary depending on the number of replacements for each component and design feature. In Fig. 10, the 17 themes are categorized into three general groups: human health, ecosystems, and resources. The single score of components is calculated using the ReCiPe endpoint method. Now, the environmental impact of the entire life cycle of D1 is compared with D2 in Fig. 11. D2 shows lower environmental impacts (less score) according to all themes except terrestrial ecotoxicity, freshwater ecotoxicity, and metal depletion.
This proposed LCA framework can be applied to estimate the frequency of replacement for product components over its life span and assess the associated environmental impacts. In our example, the prone-to-failure components of D1 are less environmentally friendly than the prone-to-failure components of the other design.
Limitations and Future Work
In this paper, an analytical framework is presented—integrating the consumer repair behavior and deterioration process of components—to assess the environmental impact associated with the life cycle of a consumer electronic device such as a laptop. This framework can be employed in the early phase of product development to design more sustainable products. In addition, it may be helpful if manufacturers extend after-sale services and assess the environmental impact of their service operations.
simapro software has been used for the purpose of LCA analysis. simapro is one of the most common LCA software that has been widely used in the literature. However, it should be noted that all LCA software products may still be difficult to implement due to the lack of actual data for specific manufacturing processes, materials, and transportation modes. On the other hand, most of the available LCA databases have been developed for specific regions . In the absence of data on LCA, using conceptual design methodologies and guidelines might be an applicable approach .
This paper is limited by the fact that there is not much data on the deterioration of individual device components. We also have assumed that the technical characteristics, except the deterioration process, of two design options do not change. Also, price and replacement costs are relatively static, which does not necessarily represent a real-world scenario. Obtaining more robust results requires having access to accurate life cycle data and conducting survey studies.
Unfortunately, there is no available framework on analyzing consumer's repair decisions. One way to extend this study is to compare the proposed framework with existing frameworks for other types of consumers' behavior (e.g., purchase, adoption, and collection behavior) conceptually.
This paper can be extended to a multi-objective problem where manufacturers determine the period of time that the spare components are produced, develop a pricing strategy for the repair services to maximize profit, and decrease the environmental impact of the entire product life cycle. Consumers' behaviors may change according to the manufacturers' production and marketing policies. Therefore, this problem becomes a strategic decision-making process rather than a tactical one. Therefore, future studies examining the effects of these policy changes on consumer behavior are needed to determine the full environmental impact of EoU/L products.
In addition, relaxing the assumptions of rational decision making of both consumers and businesses is another line for future research. Business decisions are often based on the market feedback and not necessarily based on the technical features of a product, so this study can be extended by integrating market-driven features with technical and design-driven characteristics.
This material is based upon work supported by the National Science Foundation—USA under Grant No. CMMI-1435908. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of the National Science Foundation.