Previous studies
The dosing of the detergent turned out to be the first or second most important factor in this study depending on the system type (private or shared). This is consistent with more recent LCAs on domestic laundry (see for example Yamaguchi et al. (2011) or Shahmohammadi et al. (2017)) but contrary to findings of previous studies for shared laundry services which concluded that the energy usage was the most important factor (Garcilaso et al. 2007). Unfortunately, since the scope of this study differs from previous LCA models for private versus shared systems for domestic laundry, it is hard to compare results for GHG emissions. For example, both Garcilaso et al. (2007) and Haapala et al. (2008) used Eco-Indicator Points to illustrate environmental impacts, and the study done by Amasawa et al. (2018) has a different functional unit. Re-creating the studies for comparison is theoretically possible but lies outside the scope of this study. However, the results from such a procedure would probably yield much lower impacts on GHG emissions since neither of the studies include the building use nor the drying process.
A study of laundry systems in Melbourne presented results in a comparable form to ours. Koerner et al. (2010) estimated the GHG emissions of domestic laundry to be 0.21 kg CO2 eq./kg when drying the clothes on a drying line or 1.3 kg CO2 eq./kg when using a tumble dryer. These results are much higher than the results in our study (0.190 kg CO2 eq./kg) especially since our study also includes emissions from building construction and usage. This could be a result of regional variations, much like the calculations done for European countries by Shahmohammadi et al. (2017). For example, electricity production in Victoria, Australia, is heavily dependent on black and brown coal. In addition, the washing machine used in the study is somewhat larger (7.03 kg) than the machine used in our study (5 kg), while at the same time using more energy per wash cycle (5.72 MJ versus 3.44 MJ). This is also the case for the tumble dryer, where the energy consumption per dry cycle in our study was 7.74 MJ compared with 17.42 MJ in Koerner et al. (2010). Lastly, Koerner et al. (2010) write that “Australian detergents are specially formulated for local conditions and are significantly different compared to other parts of the world”. However how, and to what extent, this has influenced the result is unclear without access to the model data.
Variability and uncertainty
Data sources
A common challenge for many LCA studies is to account for uncertainties and variability in data as well as in the models used (Ascough et al. 2008). The basic LCA model used in this study is mainly based on Swedish and European average values. Since these values are taken from, and used in similar way as, the European Commission’s own preparatory studies of Eco-design requirements of EuPs, the validity of the data is assumed to be good. On the other hand, the EuPs for washing machines and dryers were finalized during 2005–2011, and since technological advancements are proceeding within the washing sector, emission estimates should be viewed as indicative of previously installed whitegoods rather than the latest equipment. This relates not only to changes in resource consumption efficiency for running newer machines, but also to changes in size for domestic washing machines. For example, a previous study by Schmitz et al. (2016) has shown that the average rated capacity of newer private washing machines sold within EU is growing, from 5 kg in 2003 to 7.5 kg in 2014. How these changes in size might affect resulting emissions would be interesting to further investigate but are outside the scope of this study. In any case, being aware of these changes indicate that the results from this simulation are more suitable for comparative uses between systems, rather than seen as absolute values for current emissions from domestic laundry in Sweden.
Another uncertainty concerns the emissions associated with the background system of the building. Although the values used in our model (500–1400 kgCO2e/m2) are lower than previous studies, the differences are small. For example, a recent international review of 95 residential buildings found that embodied carbon emissions varied between 179 and 1050 kg CO2e/m2 and for the operating phase between 156 and 4050 kg CO2e/m2 (Chastas et al. 2018). In France, the total emissions for multi-family buildings have been estimated to somewhere between 575 and 1910 kg CO2e/m2 (Hoxha et al. 2017). Swedish examples include values for a wooden single-family house with a total emission of 567 kg CO2e/m2 for stage A1-A5, B1-B7, and C1-C4 (Petrovic et al. 2019). Lastly, Andersson et al. (2018) found that embodied emissions for a newly built multi-family house in Sweden are approximately 391 kg CO2e/m2 for module A1-A5.
Even though the data is newer, the emissions associated with square meter floor area used during the whole life cycle of the house only accounts for GHG emissions (Liljenström et al. 2015). This means that all the other indicators are underestimated in the model. Additionally, the emissions calculated in the report are based on a recent, low-energy apartment building in Sweden, with an expected lifetime of 50 years. This introduces uncertainty to the model when comparing with European and Polish data since the building may not be representative of current average European or Polish buildings. However, since we used Swedish data for heating and energy use, the GHG emissions from the use phase are likely to be a conservative estimate and lower than European and Polish averages (due to the dominance of renewable electricity sources in Sweden). Consequently, despite its large contribution to overall impacts, the relative contribution to GHG emissions from building use should be seen as a best-case scenario in the context of current European and Polish housing stock. However, due to increased EU regulations on building efficiency, it is not unreasonable to view the contribution to GHG emissions from the building as a proxy for newly built houses in Europe. Despite this low starting point for building use, the large contribution to overall GHG emissions from capital goods raises the question whether previous authors have underestimated the emissions associated with domestic laundry. This may have occurred because the manufacturing of appliances and building space is typically considered to be relatively small in comparison with the user-phase in the LCA, and thus often excluded from the analysis (Shahmohammadi et al. 2017).
It should also be noted that the datasets used for the calculating GHG emissions were the most up-to-date available in the Gabi Professional database in 2021. However, in practice, the Polish electricity mix used represents data from 2015 and should therefore be considered a “cornerstone scenario” (Pesonen et al. 2000). On the other hand, according to the IEA, the carbon intensity of Polish electricity changed little between 2015 and 2018, having improved considerably since 2006 (IEA 2021). As countries such as Poland wean their electricity sectors off coal and the transition to renewables continues, the relative significance of electricity supplies shown in our results can be expected to reduce in comparison with processes with considerable mineral carbonate or crude oil origins (e.g., building materials and detergents, respectively).
Detergent recipe
It is important to note that the detergent used in the study does not account for the great variety of detergent recipes on the market, but rather represent a generic recipe for the composition based on the EuP-reports. An assessment of the influence of detergent formulation is not within the scope of this LCA but may present an interesting horizon for future research. Likewise, the LCA model does also not include other types of detergent-free machines, e.g., washing with de-ionized water. Since the detergent is such a dominant contributor to all the impact categories, more research is needed to see how these types of alternative systems could influence the environmental impacts from domestic laundry.
Consumer behavior
Regarding consumer behavior, the data used in the model has its origins in self-reporting. This source of behavioral information is characterized by uncertainty since humans tend to have problems remembering the decisions and frequencies of past actions, especially for habitual behaviors (Verplanken et al. 2005). This means that there are some uncertainties regarding the amount of detergent used, whether the machine is fully loaded, and how often the machines are used. For example, this study assumes a linear relationship between dosage and machine size which might not reflect how people behave in real life. Especially since only a fraction of the European consumers state that they follow the recommendations provided on the detergent packaging (Alborzi et al. 2017).
Additionally, the reference flow in the model is 1 kg of laundry cleaned and dried. The calculations were performed by first calculating the emissions from one cycle of washing and drying, and then weighting the results with the amount of laundry in the machine (i.e., 60% of the rated capacity of each type of washing machine). Since this loading rate is an average value for European consumers, it might differ substantially between different individuals and accounting for a different filling percentage would lead to changed emissions per kg laundry. Should the filling percentage also differ between private and shared laundries (e.g., because of availability, time constraints, or direct costs such as for coin-operated machines), the difference between the systems would change accordingly. Higher loading rates for shared laundries would tend to strengthen our conclusion regarding the significance of the impacts of whitegoods manufacturing and building space used. With that said, it should also be noted that this model assumes that the consumer either washes using a private laundry or a shared facility since this historically has been the case for Sweden. This might not be true in other countries and a recent article by Moon (2020) showed that some consumers instead use both of the alternatives, leading to lower filling rates and higher emissions per kg laundry washed.
Lastly, changing from one type of system to another might influence other types of consumer behavior not necessarily included in the functional unit. This in turn might affect the overall emissions from doing laundry but are not included in the LCA model. Although speculative, changes in behaviors might be higher or lower wash frequency (e.g., washing training clothes directly after use) or changes in choice of drying (e.g., line drying inside your own apartment rather than using the communal tumble dryer).
Implications for policy
Shared versus private
According to estimations from housing developers, approximately 80–100% of all new multi-family buildings built in major cities in Sweden since 2012 were equipped with privately owned machines, rather than a shared laundry room (Borg and Hogberg 2014). Looking at the result in this study makes it clear that it would be beneficial from an environmental point of view to introduce policies to reverse that trend. According to Statistics Sweden, approximately 193,300 apartments were completed in multi-family buildings throughout Sweden during the year 2012–2018 (Statistics Sweden 2019). If we assume that each apartment runs 5 washing programs a week (the average wash frequency in Sweden according to Presutto et al. (2007b)), this translates roughly to an annual emissions of 2500 metric tonnes CO2-eq. per year in 2019 (and each year thereafter) that could have easily been avoided, just from domestic laundry in Sweden.
It is also important to mention that reclaiming the shared laundry room concept for multi-family houses would not only limit the emissions from domestic laundry but could also lead to more efficient use of building space. This would mean that there could be economic incentives for shared laundries (especially in densely populated cities), provided that the laundry rooms are designed in an appealing way for consumers (Amasawa et al. 2018).
The role of detergent
Policies concerning domestic laundry in Sweden and in Europe have historically focused on making laundry machines more energy efficient. This has in turn resulted in huge technological advancements over the last decade (Graulich et al. 2011; Presutto et al. 2007a), marketing campaigns (Morgan et al. 2018), and behavioral shifts (Laitala et al. 2012) towards washing at lower temperatures. However, the findings in this study suggest that Sweden are starting to reach a point of diminishing returns regarding lowering emissions using energy efficiency measures, as well as for measures targeting lower wash temperature. Looking at areas with hard water, to further reduce the emissions from domestic laundry, it might actually be preferable if some consumers increased the average washing temperature, provided that this shift was coupled with a reduction of the amount of detergent used. Or in other terms, since the current environmental labeling systems for washing machines in Europe are mainly focused on energy consumption rather than emissions, an energy efficient washing machine/program run in Sweden could actually lead to higher levels of emissions than a “normal” machine/program run at a higher temperature (but with less detergent).
However, these recommendations are foremost a result from the low GHG emissions associated with the electrical grid mix in Sweden. For other European countries, such a change in behavior would instead lead to higher emissions from domestic laundry. With that said, since many European countries are striving for greater use of renewable energy sources, this situation will change. One possible way to better understand when such a point has been reached would be to expand the system boundary for the environmental labeling systems in Europe so that it also includes possible tradeoffs between variables that are codependent (e.g., temperature and detergent). Another possible initiative would be to inform and educate consumers within countries with mainly renewable energy sources (e.g., Sweden) about this dynamic so that they themselves can make informed decisions.