A Comparative Life Cycle Assessment of Electronic Retail of Household Products

: Electronic shopping is getting more and more popular, and it is not only clothes and electronics that people buy online, but groceries and household products too. Based on real-life data from a major cosmetics and household products retailer in the Czech Republic, this study set to assess the life cycle environmental impact of parcel delivery. Two archetype parcels containing common household and hygiene products were designed and packed in two distinct ways, and the environmental impact was quantiﬁed using the Life Cycle Assessment method. It showed that it is environmentally beneﬁcial to use plastic cushions to insulate the goods instead of paper. However, the most important process contributing to the environmental burden was found to be electricity consumption in the logistics center. Hence, the importance of energy e ﬃ ciency and e ﬃ cient space utilization was demonstrated on alternative scenarios. Since the cardboard box the goods are packed in turned out to be another important contributor, an alternative scenario was designed where a reusable plastic crate was used instead. Even though the scenario was based on several simplistic assumptions, it showed a clear potential to be environmentally beneﬁcial. In the study, contribution of other processes was scrutinized, as well as sensitivity to variation of parameters, e.g. transportation distances. The main scientiﬁc contribution of this work is to show the importance of logistics and distribution of products to end customers in the rapidly developing ﬁeld of electronic retail of household products.


Introduction
The electronic retail is getting an ever bigger market share. More and more people adopt this way of shopping. In the EU, the number of individuals who have purchased goods online grew from 40% in 2010 to 63% in 2019 [1]. While the most prevalent online purchases in the EU are of clothes and sports goods (purchased by 65% of the internet shoppers [1]), the share of people buying food and groceries online more than doubled over the last ten years (from 13% in 2010 to 27% in 2019) [1]. This all means that there were over 87 billion parcels delivered globally and 9.3 billion parcels delivered in Europe in 2018 [2]. With such a massive movement of goods come potentially great environmental consequences. On this scale, even small improvements might bring substantial benefits and vice versa.
In the past years, several authors have set to investigate the environmental impact of e-commerce, usually in comparison to conventional brick-and-mortar shops. For example, Weber et al. [3] compare electronic and traditional retail of flash drives and conclude that the electronic retail results in lower carbon emission, similarly to Edwards et al. [4] or Hischier [5]. Authors often focus on the transportation impact coupled with shopping. Wiese et al. [6] compare the CO 2 emission of traditional and online freight transport are excluded up to the moment when the goods are packed into the box. The parcels are eventually delivered to one of the many brick-and-mortar shops of the retailer, where they are collected by a customer. As handling of the online shopping parcels, so far, holds only a marginal role in the activities of the shops, no energy consumption from the shops is assumed to be attributed to the parcel. The last-mile phase is extremely variable, as the mode of transport and distance depends on the customer's preferences entirely. Although-depending on the often unrealistic assumptions [9]-the customer transport might lead to a dominant fraction of the environmental burden of shopping, including the last-mile transport, would shape the results disproportionally, as the scope of this study is mainly on activities of the retailer. The end-of-life phase of the packaging materials is modelled based on common waste management in the Czech Republic.
The functional unit was designed as "delivery of one parcel". There are two kinds of model parcels containing similar products that are directly comparable. Parcels no. 1 and no. 2 contain similar products but are packed in a distinct way, and the same stands for parcels no. 3 and no. 4. For each of the four scenarios, an alternative was designed where a reusable plastic crate is used instead of cardboard box.

Life Cycle Inventory
The primary data describing the system were acquired directly from the retailer. The retailer designed two exemplary parcels containing common household and hygiene products like paper towels, diapers, or paper tissues in parcels no. 1 and no. 2; and shower gel, laundry detergent or nail polish in parcels no. 3 and no. 4. The parcels were packaged using two distinct standard methods: parcels no. 1 and no. 4 were insulated using paper, while in parcels no. 2 and no. 3, plastic cushions were used. Due to the nature and weight of the goods, additional zip bags and bubble wrap were used to insure the goods in the parcels no. 3 and no. 4. Finally, the box was secured by paper adhesive tape, and plastic adhesive tape in case of heavy parcels (no. 3 and no. 4). The materials used for packaging, parcel weights, and transportation distances are presented in Table 1. In the alternative scenario, instead of the cardboard box, a 1.5 kg reusable plastic crate made from recycled plastic is used. It is only a hypothetical scenario and an exact design of the crate is yet undetermined. Therefore, the way the parcel is packed (amount of packaging material) was assumed to be similar as in the baseline scenarios. The plastic crate was assumed to be recirculated 200 times before disposal as waste. This number corresponds to the reuse rate presented by Levi et al. [13] and Tua et al. [14], while Koskela et al. [12] estimated up to 700 circulations. The transportation distance of a new plastic crate to packaging was assumed to be similar to that of a cardboard box (180 km) and the transport distance for recirculation was assumed to be the same as the distance to the logistics center (110 km). The distance the parcel is transported from the logistics center to a customer was set to 100 km, which is the average distance. A simplified diagram of the LCA model is presented in Figure 1. For expedition and intermediate storage of the parcels, the retailer rents an area of 2184 m 2 out of the total 7170 m 2 of a logistics center. In the logistics center, electricity is consumed mainly for lighting and partly in the administrative area, natural gas is used for heating, and some drinking water is consumed. A fraction of the consumption of energy and resources was allocated to the retailer according to the area proportion. On average, there are 8672 parcels delivered per month. The total energy consumption and the amount attributed to one parcel is presented in Table 2. In the model, the average grid mix in the Czech Republic was assumed to be used. The packaging necessary for electronic shopping results in the generation of additional waste that needs to be disposed of. In the model, all of the material used for packaging of the parcels is assumed to be disposed of immediately. The waste management was modelled following the current waste management system in the Czech Republic-combination of recycling (32%), incineration (18%), and landfilling (50%) [19]. For expedition and intermediate storage of the parcels, the retailer rents an area of 2184 m 2 out of the total 7170 m 2 of a logistics center. In the logistics center, electricity is consumed mainly for lighting and partly in the administrative area, natural gas is used for heating, and some drinking water is consumed. A fraction of the consumption of energy and resources was allocated to the retailer according to the area proportion. On average, there are 8672 parcels delivered per month. The total energy consumption and the amount attributed to one parcel is presented in Table 2. In the model, the average grid mix in the Czech Republic was assumed to be used. The packaging necessary for electronic shopping results in the generation of additional waste that needs to be disposed of. In the model, all of the material used for packaging of the parcels is assumed to be disposed of immediately. The waste management was modelled following the current waste management system in the Czech Republic-combination of recycling (32%), incineration (18%), and landfilling (50%) [19].

Results and Discussion
For the environmental characterization of the emission flows, the ReCiPe 2016 (H) Life Cycle Impact Assessment (LCIA) methodology was employed [20]. This methodology converts emissions to environmental impact categories both on midpoint and endpoint levels. The chart in Figure 2 shows a relative importance of each impact category in the context of Europe, based on normalized and weighed endpoint results. As normalization factors for the ReCiPe 2016 LCIA methodology are not yet available, the older ReCiPe 1.08 (H) [21] was used. However, since this methodology is already outdated, these results have only indicative value. According to the normalization results, the most significant impact categories are Climate Change and Fossil Depletion, followed by Particulate Matter Formation and Agricultural Land Occupation; nevertheless, the latter is no longer present in the new methodology. The importance of the other impact categories appears low in the European context.

Results and Discussion
For the environmental characterization of the emission flows, the ReCiPe 2016 (H) Life Cycle Impact Assessment (LCIA) methodology was employed [20]. This methodology converts emissions to environmental impact categories both on midpoint and endpoint levels. The chart in Figure 2 shows a relative importance of each impact category in the context of Europe, based on normalized and weighed endpoint results. As normalization factors for the ReCiPe 2016 LCIA methodology are not yet available, the older ReCiPe 1.08 (H) [21] was used. However, since this methodology is already outdated, these results have only indicative value. According to the normalization results, the most significant impact categories are Climate Change and Fossil Depletion, followed by Particulate Matter Formation and Agricultural Land Occupation; nevertheless, the latter is no longer present in the new methodology. The importance of the other impact categories appears low in the European context.

Baseline Scenario
One of the main goals of this study was to compare which of the two distinct packaging systems for electronic retail results in lower environmental impact. When examining the overall results (midpoint results in Table 3, endpoint results in Appendix B), it is apparent that the systems where plastic cushions are used instead of paper are beneficial in all impact categories: Parcel no. 2 shows lower environmental impact than no. 1, and parcel no. 3 lower than no. 4. The difference is driven mainly by the filling paper which, being substantially heavier than the plastic cushions, causes a larger environmental impact throughout the life cycle.
The impact on the Climate Change category has been quantified to range from 3.12 kg CO2 eq. for parcel no. 2 to 3.68 kg CO2 eq. for parcel no. 4 (Table 3). These numbers roughly correspond to those presented by Levi et al. [13] and are in the same range as those of Koskela et al. [12]. As presented in Figure 3, the main contributors to this impact category are the energy consumption in logistics center (2.18 kg CO2 eq.) and the cardboard box (0.83 kg CO2 eq.). In the logistics center, it is mainly the electricity consumption that causes the environmental impact. As mentioned above, when paper is used to insulate the box, it becomes another significant contributor. Transportation of the parcels contributes to the Climate Change impact category by several percent, similarly as bubble wrap and filling cushions, where used. The impact of the adhesive tapes is lower.
When biogenic carbon is included, the impact of paper products is partly reduced and the difference between the parcels is lower. The logistics center then becomes a dominating driver. Considering Fossil Depletion, the logistics center is a dominating driver as well, contributing to between 70% and 80%, depending on the scenario. On the other hand, since the main source of electricity in the Czech Republic is still lignite combustion [22], in another country where cleaner sources are used, the environmental impact of electricity use would be lower and vice versa. Between 13% to 15% of the Fossil Depletion is attributed to the cardboard box, the second most important contributor. Significance of the filling paper is larger in some other impact categories, especially the human toxicity, ecotoxicity, and eutrophication categories, where it contributes with over 20%. Even though it appears that use of plastic cushions instead of paper is a significantly favorable option, it should be noted that the most debated environmental issue caused by the use of plastics-the plastic littering and the omnipresence of microplastic particles [23,24]-is not (yet) included in the LCIA methodologies. Although some indicators have been proposed, e.g., Littering Potential for carrier bags [25], a comprehensive impact category characterizing this issue is yet to be developed. Plastic littering is, however,-in a country where a relatively effective waste management operates-more of a social issue that should be dealt with on another level [26]. Furthermore, this kind of packaging

Baseline Scenario
One of the main goals of this study was to compare which of the two distinct packaging systems for electronic retail results in lower environmental impact. When examining the overall results (midpoint results in Table 3, endpoint results in Appendix B), it is apparent that the systems where plastic cushions are used instead of paper are beneficial in all impact categories: Parcel no. 2 shows lower environmental impact than no. 1, and parcel no. 3 lower than no. 4. The difference is driven mainly by the filling paper which, being substantially heavier than the plastic cushions, causes a larger environmental impact throughout the life cycle.
The impact on the Climate Change category has been quantified to range from 3.12 kg CO 2 eq. For parcel no. 2 to 3.68 kg CO 2 eq. For parcel no. 4 (Table 3). These numbers roughly correspond to those presented by Levi et al. [13] and are in the same range as those of Koskela et al. [12]. As presented in Figure 3, the main contributors to this impact category are the energy consumption in logistics center (2.18 kg CO 2 eq.) and the cardboard box (0.83 kg CO 2 eq.). In the logistics center, it is mainly the electricity consumption that causes the environmental impact. As mentioned above, when paper is used to insulate the box, it becomes another significant contributor. Transportation of the parcels contributes to the Climate Change impact category by several percent, similarly as bubble wrap and filling cushions, where used. The impact of the adhesive tapes is lower.
When biogenic carbon is included, the impact of paper products is partly reduced and the difference between the parcels is lower. The logistics center then becomes a dominating driver. Considering Fossil Depletion, the logistics center is a dominating driver as well, contributing to between 70% and 80%, depending on the scenario. On the other hand, since the main source of electricity in the Czech Republic is still lignite combustion [22], in another country where cleaner sources are used, the environmental impact of electricity use would be lower and vice versa. Between 13% to 15% of the Fossil Depletion is attributed to the cardboard box, the second most important contributor. Significance of the filling paper is larger in some other impact categories, especially the human toxicity, ecotoxicity, and eutrophication categories, where it contributes with over 20%. Even though it appears that use of plastic cushions instead of paper is a significantly favorable option, it should be noted that the most debated environmental issue caused by the use of plastics-the plastic littering and the omnipresence of microplastic particles [23,24]-is not (yet) included in the LCIA methodologies. Although some indicators have been proposed, e.g., Littering Potential for carrier bags [25], a comprehensive impact category characterizing this issue is yet to be developed. Plastic littering is, however,-in a country where a relatively effective waste management operates-more of a social issue that should be dealt with on another level [26]. Furthermore, this kind of packaging is not among the products that are most often littered, like beverage containers or food packaging [27,28]. is not among the products that are most often littered, like beverage containers or food packaging [27,28].

Alternative Scenario-Plastic Crates
As presented in the previous section, the use of cardboard boxes is an important factor in the life cycle environmental impact of electronic retail. Therefore, the hypothetical alternative scenario was designed. The overall results are presented in Table 3. The difference is most apparent in the default Climate Change impact category. Although, as the plastic crate is heavier, the transportation emissions slightly rise, the overall impact of the alternative scenario is significantly lower. As the crate is recirculated, the impact of production and end-of-life is much lower. However, with large transportation distances, the increase of transportation impact caused by greater weight might overcome the benefits, as it happened for Tua et al. [14] or Koskela et al. [12]. Were the transportation distances-to logistics center, to customer, and for the return of the plastic crate-to grow to 500 km, the climate change impact of the parcel no. 1 would grow to 2.93 kg CO2 eq. from the 2.64 kg CO2 eq.; it would grow to 3.30 kg CO2 eq. were they to grow to 1000 km; and at 1500 km it would grow to 3.67 kg CO2 eq., which is a larger impact than the baseline scenario. However, these distances are not likely in a country the size of the Czech Republic. The situation changes when biogenic carbon is included. Then, the climate change impact of the alternative scenarios is almost identical to the baseline, since the plastic crate is made from fossil resources. In most other categories, the impact of the alternative scenario is lower, by a smaller or larger margin. It is therefore apparent that substitution of cardboard boxes for a plastic crate shows a potential to be environmentally beneficial. On the other hand, this alternative scenario is based on several simplistic assumptions. If a retailer decided to employ reusable plastic crates, the system should be scrutinized in more detail (Figure 4).

Alternative Scenario-Plastic Crates
As presented in the previous section, the use of cardboard boxes is an important factor in the life cycle environmental impact of electronic retail. Therefore, the hypothetical alternative scenario was designed. The overall results are presented in Table 3. The difference is most apparent in the default Climate Change impact category. Although, as the plastic crate is heavier, the transportation emissions slightly rise, the overall impact of the alternative scenario is significantly lower. As the crate is recirculated, the impact of production and end-of-life is much lower. However, with large transportation distances, the increase of transportation impact caused by greater weight might overcome the benefits, as it happened for Tua et al. [14] or Koskela et al. [12]. Were the transportation distances-to logistics center, to customer, and for the return of the plastic crate-to grow to 500 km, the climate change impact of the parcel no. 1 would grow to 2.93 kg CO 2 eq. from the 2.64 kg CO 2 eq.; it would grow to 3.30 kg CO 2 eq. were they to grow to 1000 km; and at 1500 km it would grow to 3.67 kg CO 2 eq., which is a larger impact than the baseline scenario. However, these distances are not likely in a country the size of the Czech Republic. The situation changes when biogenic carbon is included. Then, the climate change impact of the alternative scenarios is almost identical to the baseline, since the plastic crate is made from fossil resources. In most other categories, the impact of the alternative scenario is lower, by a smaller or larger margin. It is therefore apparent that substitution of cardboard boxes for a plastic crate shows a potential to be environmentally beneficial. On the other hand, this alternative scenario is based on several simplistic assumptions. If a retailer decided to employ reusable plastic crates, the system should be scrutinized in more detail (Figure 4).

Sensitivity Analysis
The consumption of electricity in the logistics center was found to be the most important contributor to the life cycle environmental impact of the electronic retail. The impact of the electricity to climate change was quantified to 1.72 kg CO 2 eq. Because in the e-retail scenario described in this model quite old-school fluorescent lights were used, the sensitivity was analyzed to the case where 18 W or 24 W LED tube lights were used, instead of the old 58 W fluorescent lights. To simplify the calculation, it was assumed that all of the electricity is used for lighting of the storage area, although some fraction is also used in the administrative area. It turned out that use of more efficient lights would lead to tremendous environmental benefits. If 18 W LED lights were used, the impact of the electricity use in the logistics center to Climate Change would decrease to 0.36 kg CO 2 eq. and it would decrease to 0.71 kg CO 2 eq. with 24 W lights. Comparison of the Climate Change impacts of the scenarios is presented in Figure 5. There is a similar decrease in the other impact categories. Although the actual electricity use reduction would not be as large as assumed in this simple scenario, it shows the importance of energy efficiency and efficient space utilization in the logistics centers. It also means that the environmental impact of electronic retail, in general, might easily be lower than what was quantified in this scenario. The environmental impact of electricity consumption also highlights the need to look for cleaner sources of energy. However, this is a broader socio-economic issue that is usually beyond direct influence of a retailer or a consumer.

Sensitivity Analysis
The consumption of electricity in the logistics center was found to be the most important contributor to the life cycle environmental impact of the electronic retail. The impact of the electricity to climate change was quantified to 1.72 kg CO2 eq. Because in the e-retail scenario described in this model quite old-school fluorescent lights were used, the sensitivity was analyzed to the case where 18 W or 24 W LED tube lights were used, instead of the old 58 W fluorescent lights. To simplify the calculation, it was assumed that all of the electricity is used for lighting of the storage area, although some fraction is also used in the administrative area. It turned out that use of more efficient lights would lead to tremendous environmental benefits. If 18 W LED lights were used, the impact of the electricity use in the logistics center to Climate Change would decrease to 0.36 kg CO2 eq. and it would decrease to 0.71 kg CO2 eq. with 24 W lights. Comparison of the Climate Change impacts of the scenarios is presented in Figure 5. There is a similar decrease in the other impact categories. Although the actual electricity use reduction would not be as large as assumed in this simple scenario, it shows the importance of energy efficiency and efficient space utilization in the logistics centers. It also means that the environmental impact of electronic retail, in general, might easily be lower than what was quantified in this scenario. The environmental impact of electricity consumption also highlights the need to look for cleaner sources of energy. However, this is a broader socio-economic issue that is usually beyond direct influence of a retailer or a consumer. What may also vary substantially is the distance a parcel is transported. Sensitivity of the results to changes in transportation distance was evaluated for the parcel no. 1. In the baseline scenario, the distance the parcel travels from the logistics center to a customer was set to 100 km, which corresponds to the average distance. In the case where the distance is 50 km, the overall impact to climate change decreases from 3.439 kg CO2 eq. for the baseline scenario to 3.420 kg CO2 eq., and increases to 3.466 kg CO2 eq. when the distance is 200 km. As larger distances are unlikely in a country the size of the Czech Republic, it may be concluded that the results are not sensitive to the location of customer. Additionally, the distance between packaging and a logistics center might differ in other situations. If the distance changes from the baseline 110 km, to 50 km, the climate change impact is reduced from 3.439 kg CO2 eq. to 3.417 kg CO2 eq., and grows to 3.478 kg CO2 eq. when the distance is 250 km. If the truck utilization efficiency decreased from the average 61%-that is, the default value What may also vary substantially is the distance a parcel is transported. Sensitivity of the results to changes in transportation distance was evaluated for the parcel no. 1. In the baseline scenario, the distance the parcel travels from the logistics center to a customer was set to 100 km, which corresponds to the average distance. In the case where the distance is 50 km, the overall impact to climate change decreases from 3.439 kg CO 2 eq. For the baseline scenario to 3.420 kg CO 2 eq., and increases to 3.466 kg CO 2 eq. when the distance is 200 km. As larger distances are unlikely in a country the size of the Czech Republic, it may be concluded that the results are not sensitive to the location of customer. Additionally, the distance between packaging and a logistics center might differ in other situations. If the distance changes from the baseline 110 km, to 50 km, the climate change impact is reduced from 3.439 kg CO 2 eq. to 3.417 kg CO 2 eq., and grows to 3.478 kg CO 2 eq. when the distance is 250 km. If the truck utilization efficiency decreased from the average 61%-that is, the default value in the GaBi database-to 40%, the overall contribution to the Climate Change category would increase to 3.454 kg CO 2 eq. The trend is similar in the other impact categories, hence, the transportation attributes only by several percent in most likely scenarios.
The hypothetical alternative scenario is based on several assumptions that might affect the results. Sensitivity of the results to variation in the assumptions was evaluated for the parcel no. 1. With the assumed crate weight of 1.5 kg, the impact on Climate Change was 2.640 kg CO 2 eq. Were the crate lighter, at 1 kg, the burden decreases to 2.626 kg CO 2 eq., and use of a 2 kg plastic crate would lead to 2.653 kg CO 2 eq. Furthermore, the number of times the crate is recirculated may vary. If the crate is recirculated 50 or 500 times, the results remain nearly unchanged. However, when the crate is reused only 10 times, the climate change burden increases to 2.754 kg CO 2 eq. This finding corresponds to that of Tua et al. [14], who concluded that, in their scenario, the results do not change much when the reuse rate is over 20. The burden on the other impact categories changes in a similar manner. Hence, the results are not sensitive to the assumed weight of the used plastic crate nor the number of times the crate is reused.

Conclusions
As the electronic retail sector is rapidly growing, this study set out to scrutinize the life cycle environmental impact of online shopping for household products. It was found that the additional packaging of products in electronic retail of household products represents an additional environmental burden. It has also been shown that logistics has a significant environmental impact that should be reduced in the future. Using the Life Cycle Assessment methodology with the functional unit designed as "delivery of one parcel", two distinct packaging systems were compared based on real-life data from a major cosmetics and household products retailer in the Czech Republic. Analyzing the life cycle of e-retail of two kinds of archetype parcels-manufacturing and freight transport of the products were assumed to be outside the system boundaries, as well as the last-mile transport-it showed that using plastic cushions causes a smaller environmental burden than using paper to insulate the goods. However, the main driver of the environmental impact was electricity consumption in the logistics center. Seeing that most of the electricity is needed for lighting of the storage area where old-school fluorescent lamps were used, the importance of more efficient lighting, as well as efficient space utilization in the logistics, was demonstrated on the alternative scenarios. The second most important aspect of the electronic retail appeared to be the use of cardboard boxes. Therefore, the alternative scenario was designed, where reusable plastic crates are used instead. Although the alternative scenario is only hypothetical and is based on several simplistic assumptions, it showed a great potential to be environmentally beneficial, regardless of changes in the assumed parameters. The transportation of the parcels-to the logistics center and to the brick-and-mortar shop where the customer collects it-contributed to the environmental impact only by several percent, even considering the variability of the distances. Similarly, the filling cushions and bubble wrap, where used, contributed to the overall burden by several percent. The share of the adhesive tapes does not appear important.
What does this mean then? The results appear conclusive: It is environmentally beneficial to use the much lighter plastic cushions instead of paper. Similarly, it would be worthwhile to design a reusable plastic crate that would substitute the single-use cardboard box. However, it is questionable whether a commercial enterprise would endorse this significantly better option when public image of plastics is contrary, particularly when the plastic littering problem is not yet included in such environmental analyses. On the other hand, it is intuitive: The less material used, the less energy and resources are needed. On the scale that the electronic retail sector operates, even small improvements in the single parcels would lead to great benefits overall. This is blatantly manifested with the energy use in the logistics center. Nevertheless, in the perspective of today's climate and ecological crisis, it would be best if the customers avoided excessive consumerism whatsoever.
Author Contributions: V.K. outlined the research topic, performed data collection and LCA analysis. J.M. also performed and improved LCA analysis and wrote the whole manuscript. All authors have read and agreed to the published version of the manuscript.