This chapter outlines the goal and scope of the study, system boundaries and allocation (see Sect. 2.2), the life cycle inventory (see Sect. 2.3), and impact assessment method (see Sect. 2.4).
Goal and scope
This study aims to quantify the environmental impacts over the life of the human being “Dirk” by applying the Life-LCA approach (Goermer et al. 2019). As the key applications of Life-LCA are individual performance tracking and life optimization, the term reporting unit is used for this individual Life-LCA, which was previously introduced for the O-LCA method (UNEP/SETAC 2015). Thus, the reporting unit is set to the life of Dirk (current life span: 0–49 years) and the range of his consumed products (reporting flow). A further goal was the analysis of Dirk’s consumption impacts over a current 1-year period (baseline scenario before his consumption patterns changed) (see Sect. 3.2) compared to an optimized scenario after changing his consumption behaviour (see Sect. 3.3). For the optimized scenario, improvement measures for reducing environmental impacts of certain consumption behaviors, as recommended by two German environmental NGOs, were assessed. Further, the results are evaluated for their plausibility by comparing them with results calculated by the German Environmental Agency’s (UBA) carbon calculator (see Sect. 4.3).
System boundaries and allocation
Life-LCAs have two-dimensional assessments, where system boundaries are defined for both dimensions - the higher-level human life cycle (dimension 1) and the lower-level product life cycle (dimension 2) (Goermer et al. 2019). For determining the environmental impacts of Dirk’s life, the temporal system boundaries for dimension 1 were set to his current life span, i.e., from his birth until his 49th birthday. In the absence of detailed data, the “childhood and youth stage” (0–17 years) was considered by assuming that in those first life years, the consumption was just half (50%) of the consumption in the baseline year for all product categories excluding “transport” (see Sect. 2.3.2 for a list of product categories). Transport was excluded as a large share of the baseline year’s impacts was due to work-related travels. As Dirk obtained his driver’s license (18 years old), all means of transport were considered. This simplified approach for the “childhood and youth stage” was applied due to a missing methodology and specific consumption data. However, for some product categories, more specific past acquis data and assumptions on past consumption behaviour were considered (for details, see Sect. 2.3.1).
As a second scope of the analysis, the system boundaries for dimension 1 were set to a 1-year period to compare different consumption patterns (baseline and optimized scenario). Data collection was carried out during a 2-month period in the years 2017 (baseline scenario) and 2018 (optimized scenario), respectively. Consumed products (e.g., food, cosmetic products, energy use at home) were documented and extrapolated to the full calendar year with a relative correction of any seasonal effects (e.g., increased heat consumption in winter) (see Sect. 3.1). The system boundaries of dimension 2 (the product life cycle) were set to include all consumed products and services. In both cases, the environmental impacts of the consumed products were covered from cradle-to-grave; i.e., impacts of all life cycle stages (production of raw materials, product production, use, end-of-life) were considered.
The inclusion or allocation of products consumed by his children that are potentially influenced by Dirk’s decisions (e.g., in early life years, parents choose food and clothes for their children) were excluded for consistency reasons. Otherwise, Dirk’s childhood impacts would have to be allocated to his parents as well. Another allocation option would be to allocate the burden of each child partly to Dirk. Therefore, the exact moment of free will and ability to decide would have to be evaluated for each of Dirk’s children. Due to a missing methodology and the complexity of this question, this allocation option was not applied. It is suggested to carry out different childhood allocation options in future studies to close this methodological gap (see Sect. 4.2: Allocation of the childhood phase and other socio-psychological interactions between human beings). Further exclusions were financial investments beyond physical goods (e.g., investments in stocks) and goods received as an inheritance (e.g., properties and houses), which might have a significant impact and should be considered in future studies, if applicable and meaningful. For allocation, the co-product allocation rules apply. As for end-of-life-allocation, the avoided burden approach (0:100) was applied; i.e., credits were given for recycling, while secondary materials carried the burden of primary material production. Recycling credits were modelled based on current German recycling quotas for the different materials (Statistisches Bundesamt 2017).
For Life-LCA, some additional allocation rules need to be specified. For products shared with other persons, Dirk documented the product use ratio (see Sect. 2.3.1). For instance, amounts of food (e.g., shared dishes) and drinks (e.g., a shared bottle of wine or water) served at restaurants were estimated this way. The shared use of the same products (e.g., housing or electronics products) was allocated according to the product sharing ratio. Two additional family members are living in Dirk’s household. Thus, the environmental impacts of a device (e.g., television) allocated to Dirk were assumed to be one-third. The effective usage of a shared product from each household member was not considered. It would have been associated with high efforts for data collection and the other family members’ willingness to play a substantial role in monitoring their behaviour, which was not a feasible option for this study.
This chapter explains the method for data collection and calculation (see Sect. 2.3.1) and the subsequent establishment of product clusters (see Sect. 2.3.2).
Data collection and calculation
To calculate Dirk’s Life-LCA, a relatively comprehensive and time-demanding data collection had to be performed to cover all consumed products. Therefore, data collection was performed with tailor-made sets of data collection sheets (available at: https://www.see.tu-berlin.de/menue/research/data_tools/life_lca_life_cycle_impacts_of_a_human_being/parameter/en/). The different sheets cover continuously (e.g., food, toilet paper), discontinuously consumed products (e.g., clothing), and acquis data (e.g., desk, television). Acquis data refers to all consumed products with an average life span of a year or more and cannot be covered within a short (weeks, months) monitoring period. Digital drop-down tables were created in Excel®, where consumed products were documented according to pre-defined product categories (e.g.,”food” or “energy and water”) and clusters (e.g., non-alcoholic beverages in the product category “food”), sub-clusters (e.g., juice), and types (e.g., orange juice) to simplify the subsequent data analysis and establishment of clusters (see Sect. 2.3.2 for a detailed explanation of the product cluster). The product categories were chosen in accordance and close cooperation with Dirk to cover all necessary products and facilitate his data collection through a clear structure. Where relevant, Dirk documented more specific information to characterize the product, e.g., if the product was bought second-hand (e.g., for clothing) or of which material the product is made of (e.g., fleece or cotton).
Following aspects were applied for data collection:
Continuously consumed products (e.g., meat): data collection on the consumed amount, number of persons consuming it, additional specific information (e.g., in the case of dear “own shot game”), the geographical origin (limited to “Germany,” “Europe,” and “worldwide”), and primary packaging of the consumed product (e.g., plastic bottle for orange juice) as well as classification into disposable and reusable.
Discontinuously consumed products (e.g., clothes): additional information, such as average life span or used material of products.
Acquis data (e.g., desk, television): additional information such as the average life span or user ratio of the products.
To collect data for Dirk’s consumption in the past, data of previous years for the categories transport, energy, water, food, and pets were collected as detailed as he could remember. Past acquis data were derived based on his memories or records (e.g., invoices, logbooks). Dirk’s mobility behaviour (e.g., vehicle type, km driven per vehicle, fuel type) could be documented rather precisely from the age of 18 on, covering nine cars and several motorcycles. Until he started hunting in 2015, all meat consumed was assumed to be from industrial production. His pets (a hunting dog and a cat), including accessories (e.g., bucket, leash) and feed, were taking into account from 2004 onwards. The pets are fully allocated (conservative approach) to Dirk, who constantly cared for the pets in all years. For future individual Life-LCAs, the allocation of the pet’s environmental impacts have to be chosen according to the respective circumstances of the study objects’ living situation.
His past living conditions in several apartments and with different energy supplies were also included. For example, he has only been using electricity and gas from renewable sources since 2014. In contrast, before 2014, he used an average electricity consumption mix and oil as a thermal energy source.
For some products, the seasonal variation during the monitoring periods in 2017 and 2018 had to be considered. For example, thermal energy consumption was recorded over the 2 months of February and March; thus, a distorted extrapolation to 1 year would overestimate the yearly consumption. Therefore, correction factors based on the average monthly thermal energy consumptions considering German national statistics (Gasverbrauchsrechner 2019) were applied. Based on these numbers, thermal energy consumption in the months February and March accounts for 25.5% of the yearly consumption. Another possibility to account for yearly energy and electricity consumption could be to refer to past gas/electricity bills for missing months (if available). The methodology with the correction factor can be favoured: (a) if someone just moved into an apartment/house and there are no past records available, (b) the study object changed to more efficient energy systems and past records would not be adequate anymore, and (c) the system boundaries are set to specific months. Furthermore, as it was not possible to measure the specific energy and water consumption of hotel stays, average data were used. The energy consumption per hotel stay was estimated at 25 kWh of total energy consumption (thermal and electric) (Beccali et al. 2009; Filimonau et al. 2011). A conservative approach was chosen to account for energy and water consumption during a working day (time not spent at home—no water and electricity meters available) by assuming 40 h of work per week considering 168 total hours available. Thus, time spent at work was calculated at 23.8%—hours spent at his workplace per week, divided by the total hours available per week—and rounded up to 25%. The water and energy consumption measured at home was assumed to be the basis during working hours, assuming a similar consumption behaviour like at home. Therefore, an additional 25% of measured energy and water consumption at home was modelled as consumption at work. It was also assumed that the German electricity mix for energy was used for work and in hotels.
After completing the comprehensive primary data collection described above, the different used products, services, and resources need to be connected with the respective background data. Due to the breadth and sheer number of products needed, different background data sources had to be used. While it is understood that this mix of data sources introduces inconsistencies in the overall result, it is the only option available to date to cover the broad set of products consumed in Dirk’s life applying the LCA method.
The following data sources were used to model the consumed products (see Fig. 1): (1) common LCA databases, which are GaBi database (Thinkstep 2018), Ecoinvent 3.3 (Wernet et al. 2016), and Agribalyse database 1.3 (Koch and Salou 2014); (2) Environmental Product Declarations (EPD) based on Product Category Rules (PCR); (3) documents published by the Product Environmental Footprint (PEF) initiative (European Commission 2019); and (4) additional LCA reports and case studies (later addressed as proxy (own model) and reference materials). The particular order of the hierarchy highlights the decreasing quality of the underlying data, as shown in Fig. 1. The practice has shown that EPD was ultimately not used (see Fig. 2). Nonetheless, the hierarchy should be applied in this order as EDP is the more established database compared to PEF (European Commission 2018; Minkov et al. 2020). In cases where more than one dataset for the same product was available, data were chosen according to the hierarchy presented in Fig. 1 (e.g., when data for apples are available in GaBi and a published case study, GaBi data were chosen).
Moreover, in the case of different options from common LCA databases, Gabi database was preferred over Ecoinvent and Agribalyse, due to having the most state-of-the-art and up-to-date datasets available. The LCI and LCIA were modelled with GaBi software service pack 39 (Thinkstep 2018). Proxy (own model) means that the product itself was modelled based on its bill of material (e.g., soft furniture such as a sofa: no dataset at product level was available). Thus, it is modelled based on a publication in which typical material compositions of average soft furniture are listed (DHP furniture 2019). The term “reference materials” (e.g., GaBi or Ecoinvent reference material) was used when there was neither information about the average composition nor aggregated product datasets or specific case studies available. If such a product consisted mainly of one material, it was modelled based on an average material dataset available (e.g., Dirk has two wooden tables for which no data was available; thus, the table was only modelled with an average dataset for wood mix in Germany. Such a simplification neglected non-wooden parts like screws.
Modelling of product clusters
Dirk recorded a broad variety of products during the data collection and monitoring phase, based on predefined product categories (see Sect. 2.3.1). Not all of these products could be modelled in detail due to limited data availability. Thus, a bottom-up product clustering scheme was applied according to Goermer et al. (2019), which is based on a concept to establish product clusters according to data availability. Products with missing LCA data were assigned to similar products with available LCA datasets (e.g., buns, pastry, and bread are clustered as “bakery products” and modelled with the dataset for bread).
Overall, 10 product categories cover 213 product clusters (e.g., the product category “food” covers the cluster tropical fruits, which refers to oranges, kiwis, and pineapples among others). N refers to the number of considered product clusters (e.g., “Bakery products” equals one product cluster out of 41 in the product category “food” (N = 41)). The following product categories were determined:
“clothes and jewellery” (N = 23),
“cosmetics, hygiene, and cleaning” (N = 17),
“electronics” (N = 15),
“energy and water” (N = 6),
“food” (N = 41),
“health and medical equipment” (N = 5),
“hobbies, leisure, and pet” (N = 29),
“house” (N = 16),
“living, household, and home office” (N = 49),
“transport” (N = 12).
The highest number of product clusters were established for the product categories “living, households, and home office” (covering 49 clusters) and “food” (covering 41 clusters), whereas “health and medical equipment” just covered 5 clusters. Reasons for fewer clusters were either a limited number of products in general (e.g., “energy and water”) or the fact that Dirk consumed only few of the products covered by the product category (e.g., only contact lenses or painkillers in the product category “health and medical equipment”). All product clusters are documented in detail in the supplementary material SM1 (see SM1: 3. Clusters and data sets).
Figure 2 summarizes the different data sources used for modelling and different product clusters (datasets) of the 10 product categories for this specific case study. Some categories (e.g., the product categories “energy and water,” “house,” and “transport”) are mainly modelled using existing databases; i.e., aggregated datasets were available. In contrast, other product categories (e.g., “clothes and jewellery”) are mainly modelled based on reference materials (see Sect. 2.3.1 for an explanation of the term), case studies, and proxies. The differentiation between “GaBi” and “Gabi reference materials” (same applies for Ecoinvent and PEF) was made for transparency reasons and to indicate that current LCA databases lack datasets, especially for these product categories.
Data availability in LCA databases and case studies differs per product category. For “transport, “housing and energy,” and “water” as well as for “food” and “electronic products,” a fair amount of data on product level is available (> 75% of the clusters were modelled by using aggregated product-based datasets from databases, case studies, or PEF).
The life cycle impact assessment (LCIA) phase included the calculation of the Global Warming Potential (GWP), Acidification Potential (AP), Eutrophication Potential (EP), and Photochemical Ozone Creation Potential (POCP). The impact assessment method CML-IA (version 2016) (Guinée et al. 2001; van Oers et al. 2002; CML-Department of Industrial Ecology 2016) was applied for all impact categories. CML-IA was chosen as it is one of the most commonly applied approaches for assessing environmental impacts in LCA (Bach and Finkbeiner 2017). Additionally, most of the reviewed case studies, which served as data sources for modelling the different impact categories, used CML-IA. Therefore, the effort of results conversion was reduced.
However, when different impact assessment methods were applied for some products, case study results had to be converted into CML-IA results. This conversion was done by a re-calculation via the characterization factors of the corresponding reference substances. For example, the results of the impact category acidification applying the characterization model of Seppälä et al. (2006) (accumulated exceedance expressed in [molc H + eq.]) had to be translated to results of the model developed by Hauschild and Wenzel (1998) used in CML-IA (expressed in [kg SO2-eq.]). Therefore, the amount of SO2-eq. was determined by dividing the LCIA result based on the model of Seppälä et al. (2006) through the respective characterization factor for this substance; i.e., 1.31 mol of H + -eq. is translated into 1 kg SO2 emission into air. Again, this procedure implies obvious simplifications and inconsistencies. However, it was inevitable to cover the broad set of products consumed by Dirk as no consistent data is available.