**1. Introduction**

Diverse criteria need to be considered when planning future energy systems, such as costs, greenhouse gas emissions, land use, and further environmental impacts. As these criteria are measured in different units, they cannot be directly compared and decision-makers need support to consider conflicting targets adequately. This article, therefore, deals with a systematic analysis of environmental trade-offs to better support the design of decentralized renewable energy systems. It is based on a case study conducted as part of the center for applied research "Urban Energy Systems and Resource Efficiency" (ENsource), an inter-university research network that aims to provide scientific support for the design and operation of sustainable energy systems. The case study took place at Mainau GmbH, a tourist company located on an island in Lake Constance in southern Germany. To become climate neutral, its management decided to further increase the share of renewables in the company's energy supply. In order to design an energy system that reduces greenhouse gas (GHG) emissions without increasing other environmental impacts, life cycle assessment (LCA) studies for four future energy supply options were conducted, focusing on the choice of an appropriate metric for evaluating environmental trade-offs.

With the Federal Climate Protection Act [1], the German government has committed itself to become greenhouse gas neutral by 2050. As electricity and heat generation account for more than one-third of German GHG emissions [2], the use of renewable technologies in these sectors is an important lever for the German energy transition. Hence, Mainau GmbH's objectives and strategy are representative of the current efforts of many companies and municipalities in Germany.

Although renewable energies reduce GHG emissions, their construction, disposal, and in some cases, also operation still cause environmental pressures [3,4]. The design of a decarbonized urban energy supply almost inevitably leads to environmental trade-offs. For mineral resources Vidal et al. even point out the danger of a vicious cycle, where "the shift to renewable energy will replace one non-renewable resource (fossil fuel) with another (minerals and metals)" [5]. Hertwich et al. substantiate this concern in an LCA study of a long-term, wide-scale implementation of renewable electricity generation up to 2050 that indicates an increased global consumption of mineral resources like cement, iron, aluminum, and especially copper [6]. As humankind is already at the limits of or even transgressing multiple planetary boundaries [7], a multi-dimensional view of environmental impacts becomes imperative: An energy system design that only takes GHG emissions into account can lead to adverse environmental effects [8].

Hence, methods are required that enable planners and decision-makers to identify energy supply options with a minimal overall environmental impact. LCA provides valuable decision support in this context, as it considers the whole life cycle of power plants and energy carriers [9]. It moreover incorporates a comprehensive environmental impact assessment. Unfortunately, the results of a multi-dimensional life cycle impact assessment (LCIA) are often ambiguous. The main challenge for practical decision support is, therefore, to make impacts in different categories comparable in a meaningful way. Several existing weighting methods can help to solve this dilemma. Even though this approach is generally criticized for necessarily relying on normative value choices [10], it provides valuable decision support in practice [11].

This paper uses the ecological scarcity method (ESM) [12,13] to normalize and weight LCIA results. ESM is a distance-to-target method that weights different environmental pressures based on the ratio of the current situation to the desired policy target. Due to its mathematical simplicity and because the weights depend in a transparent way on publicly available data from laws or environmental authorities [11], it is particularly suitable for communication with practical decision-makers who are usually not LCA experts. For the same reason, it is easier to adapt the ESM to specific decision contexts than other weighting approaches such as, e.g., monetary, panel, or mid-to-endpoint weighting [11,14]. In the present study, we adapt and apply an ESM to renewable energy systems in Germany.

The paper is outlined as follows. Section 2 introduces the main features of ESM and describes essential assumptions and necessary adaptations for the assessment of renewable energy systems in Germany. Sections 3 and 4 deal with the application of the method to the case study of Mainau GmbH. Different energy supply scenarios are ranked regarding their environmental impact and trade-offs between different impact categories are analyzed in more detail. A contribution analysis shows which energy technologies and life cycle processes cause the highest environmental impacts. Based on these results, Section 5 provides recommendations for company management. Section 6 puts the findings from the case study into the broader context of the energy transition, discusses strengths and weaknesses of the adapted ESM, and points out future research needs and fields of application.

#### **2. Materials and Methods**

#### *2.1. The Ecological Scarcity Method—Basic Structure and Important Properties*

All ESM share the same basic structure [13]: An impact score (IS) is calculated by multiplying elementary flows *ej* from a product system's life cycle inventory by specific eco-factors *EFj* (Equation (3)) and adding them up (Equation (1)). Elementary flows are material, and energy flows between the system under investigation and the natural environment.

$$IS = \sum\_{j} \varepsilon\_{j} \cdot EF\_{j} \tag{1}$$

Depending on the set of elementary flows covered by the summation index *j*, the impact score corresponds either to a specific impact category *i* (e.g., climate change, mineral resources, water pollution, etc.) or to the total environmental impact. The total impact score *IStotal* thus corresponds to the sum of all category impact scores *ISi* (Equation (2))

$$IS\_{\text{total}} = \sum\_{i} IS\_i \tag{2}$$

The eco-factor (*EFj*) combines an external normalization with respect to an appropriate reference value (*Nj*) and a weighting factor: *wj* = *Aj*/*Tj* 2 (Equation (3)). The weight depends on the ratio of the current environmental pressure (*Aj*) and the desired target value (*Tj*). ESM thus belongs to the class of distance-to-target weighting methods (cf. [11,15]).

$$EF\_j = \cdot \frac{1}{N\_j} \cdot \left(\frac{A\_j}{T\_j}\right)^2\tag{3}$$

For practical implementation, some specifications and extensions to this basic concept are necessary (Equation (4)): First, most quantities involved in calculating eco-factors refer to a certain region *x* (e.g., Switzerland, Germany, the World) and time horizon *t* (e.g., 2010, 2020, or 2050). Second, the index *j* in Equation (3) refers to single elementary flows, whereas target values, in some cases, only exist on an aggregated level: For instance, greenhouse gas reduction targets apply to different substances and are therefore expressed as global warming potential (CO2-equivalents). In this case, Equation (4) integrates characterization, i.e., calculating the contribution of elementary flows *j* to specific impact categories *i* via the characterization factor *CFij* and carries out both normalization and weighting on the impact level. Finally, a region-specific scaling factor *s*(*x*) usually assures "reasonable" numerical values for the impact *IS*.

$$\text{CF}\_{\text{j. x.},t} = \text{CF}\_{ij} \cdot \frac{1}{N\_{i, \text{x.},t}} \cdot \left(\frac{A\_{i, \text{x.},t}}{T\_{i, \text{x.},t}}\right)^2 \cdot s\_{\text{X}} \tag{4}$$

Equation (4) reveals an important feature of ESM: its adaptability. So far, it has mainly been used to adapt the original Swiss ESM [13] to other countries including Germany [16,17], Thailand [18], China [19], and the EU [20,21].

#### *2.2. Compilation of the ENsource ESM*

The following section describes the compilation of an ENsource ESM to evaluate environmental trade-offs in the specific context of decentralized renewable energy systems in Germany at the example of Mainau GmbH. To this end, normalization references (*N*), current pressures (*A*), and target values (*T*) of suitable existing ESM had to be adapted to current German conditions. We chose the ESM developed by Ahbe et al. for Germany as a starting point, whose eco-factors basically required a time update (Equation (4)) [17]. In order to further increase the ESM's coverage of relevant environmental issues, the impact indicators land use, carcinogenic substances into air, heavy metals into air, and ozone layer depletion have been adopted from an ESM developed by Muhl et al. for the European Union [21]. Here, some quantities had to be scaled to German conditions. Eventually, the impact category mineral resources, which is missing in the ESMs of both Ahbe et al. [17] and Muhl et al. [21], was integrated following Frischknecht and Büsser-Knöpfel [13] but applying the latest abiotic resource depletion potentials based on the ultimate reserve according to van Oers et al. [22].

All resulting eco-factors refer to *x* = Germany. For consistency reasons, the base year is *t* = 2017 for all normalization and current environmental pressures. The time horizon for the target values is *t* = 2050 unless this was not possible due to the lack of data. The attribution of elementary flows to impact categories (classification) follows the impact assessment method "ecological scarcity 2013" as implemented in the ecoinvent v3.5 database [13,23].

Table 1 provides an overview of all considered impact indicators in the ESM. The asterisks in column "Adapted" indicate that most target values (*T*) have been changed with respect to their respective origins, e.g., if more recent legislative references were available. The column "legitimation" provides a classification of the target value's degree of democratic legitimation according to the following guidelines: A high legitimation (+) is assigned to quantitative targets adopted directly from a national law or a regulation as, e.g., the Federal Climate Protection Act in the case of Global Warming (GW) [1]. Targets based on binding international treaties or guidelines are assigned a medium legitimation (o), e.g., the UN Protocol on Heavy Metals that sets binding targets for Germany for heavy metal emissions into air (HMIA). Eventually, a low legitimation (−) indicates a target derived from a qualitative objective or strategy [24]. This applies, for example, for the targets for mineral resources (MR) and land use (LU), which are deduced respectively from the German resource efficiency program II [25] and sustainability strategy [26]. For a detailed documentation of the eco-factors in the ENsource ESM please refer to the supplementary information (S1\_ENsource ESM, S2\_ENsouce ESM elementary flows).

**Table 1.** ENsource ESM impact indicators. Origin: indicates ESM from which indicator was adopted: A = Ahbe et al. [17], M = Muhl et al. [21], F = Frischknecht and Büsser-Knöpfel [13]. Adapted: Asterisks (\*) indicate adaptations of target value T with respect to original ESM. Legitimation: Different categories: (+) high (e.g., law, regulation); (o) medium (e.g., binding international treaty, EU directives); (−) low (e.g., derived from qualitative goals, strategies).


## *2.3. Comparative Analysis of the ESM Weighting Schemes*

Figure 1 compares the normalized weights of the ENsource ESM with its predecessor methods. The selection of impact categories corresponds to the ENsource ESM. Particularly striking are the differences with Ahbe et al., as this was the starting point for the development of the ENsource ESM [17].

First, the significantly higher share of GW in the ENsource ESM weighting scheme is conspicuous. It corresponds to the more ambitious target value for GHG emission reductions (−90% of 1990 GHG emissions by 2050 as compared to −80% in the ESM of Ahbe et al.) based on the following consideration: The Federal Climate Protection Act [1] strives for climate neutrality. To avoid the mathematical singularity and because we assume that a complete implementation is not to be expected, we use the mean value between the target value used by Ahbe et al. and a reduction to zero. Even though this does not correspond exactly to the legal limit, the resulting weight for global warming certainly better reflects the current political priority setting than the original weight in Ahbe et al. [1,17].

**Figure 1.** Weighting schemes of the different ecological scarcity methods. GW = global warming, LU = land use, APP = main air pollutants and PM, ER = energy resources, HMIW = heavy metals into water, CSIA = carcinogenic substances into air, HMIA = heavy metals into air, WP = water pollutants, MR = mineral resources, WR = water resources, ODP = ozone layer depletion, WTD = non-radioactive waste to deposit.

Furthermore, the large contribution of water pollutants (WP) and heavy metal emissions into water (HMIW) in the weighting scheme of Ahbe et al. is striking. The corresponding target values stem from a personal communication with the Federal Environment Agency [17]. As they could not be updated with publicly available sources, the ENsource ESM uses the original methodology from Frischknecht and Büsser-Knöpfel instead [13]. The target values for HMIW and polycyclic aromatic hydrocarbons (concerning WP) thus correspond to critical concentrations [27]. The remaining WP target values for phosphorus, nitrogen, and chemical oxygen demand correspond to those in Ahbe et al. [17].
