Keywords

1 Motivation and Problem Scope

The pathway to reach a sustainable supply of energy in the future is challenging. Nevertheless, a reduction of greenhouse gas (GHG) emissions is a pressing issue that needs to be addressed as soon as possible in order to stay below the limit of 1.5 °C of global warming [1]. The current primary energy supply is mainly based on fossil fuels. In 2016 fossil fuels accounted for 81.1% of the world’s primary energy supply [2]. This applies globally as well as in Germany where in 2017 80.3% of the primary energy consumption was based on fossil fuels (hard coal, lignite, gas, oil) [3]. In order to fulfill the political agenda for decarbonisation a lot of research is conducted to illuminate the different routes towards sustainable energy supply. One main approach to model future energy supply are energy scenarios that are integrated in energy system models [4, 5]. A methodical framework is needed to assess the sustainability of these pathways. Life Cycle Assessment (LCA) provides such a framework and evolved to an accepted tool with international standards and norms [6], to evaluate the life cycle impact of different energy technologies [7, 8] and systems [9,10,11]. This article contains a detailed discussion of the requirements and limitations of the current state-of-the art of LCA of energy scenarios. It also makes a proposal to extend the current methodology to a dynamic and modulised LCA using the example of Power-to-X (PtX).

1.1 Deep Decarbonisation Energy Scenarios and the Relevance of Power-to-X Technologies

Most energy scenarios that are based on the political will to cut GHG emissions yield results that point into two slightly different directions—an electrification of energy applications or an electrification in combination with the production of synthetic fuels respectively PtX fuels [12, 13]. The main difference in whether the first or second alternative dominates, depends upon the level of decarbonisation. Especially ambitious levels of decarbonisation of up to 95% GHG emission reduction in 2050 compared to 1990, as for example aspired by the German government [14], favor a development of the energy sector towards an integration of PtX technologies [13].

The term PtX is not yet used uniformly in the literature and often causes confusion due to the variety of synonyms. Synthetic fuels, green fuels, renewable fuels, alternative fuels or electrofuels are just some of the many terms referring to fuels provided by similar technologies. In the present case PtX describes any technology that produces liquid or gaseous fuels which are produced from electricity. The key technology is the electrolysis for the separation of water into hydrogen and oxygen [15], which can be subdivided into alkaline electrolysers (AEL), proton exchange membrane (PEM) electrolysers and solid oxide electrolysers (SOEC). They differ mainly with regard to their efficiency, charge carriers and operating temperatures [16].

The resulting hydrogen can then either be used directly as a final energy carrier [17] or, in a further step, be synthesised into a carbon-containing energy carrier [18] (see Fig. 1). Currently, in literature there are two main synthesis routes for liquid energy carriers, namely methanol and Fischer-Tropsch synthesis [19]. Methanol synthesis uses CO2 in combination with hydrogen as process inputs [20]. In the case of Fischer-Tropsch synthesis, CO is required. Therefore, CO2 has to be converted to CO in a preceding reverse water gas shift reaction [21]. For the production of gaseous carbon-containing fuels methanation is one possible synthesis process to convert hydrogen to methane [22].

Fig. 1
figure 1

Schematic representation of the production of PtX fuels

Full size image

In any case, a CO2 source is necessary. For an ultimately CO2-neutral energy carrier, the CO2 must not be obtained from the combustion of fossil energy sources. Direct air capture, the combustion of biomass or process emissions in the industry are suitable options [23].

For ambitious levels of decarbonisation these fuels are important as they possess two distinct advantages in contrast to the direct use of electrical energy:

  1. 1.

    PtX fuels have a high energy density. A high energy density is crucial in applications that are critical in respect to space or weight limitations such as e.g. the aviation sector.

  2. 2.

    PtX fuels offer the opportunity to store electricity surpluses from renewables over long periods of time.

Even though the energy efficiency is significantly lower than in case of a direct use of electrical energy [20,21,22], the advantages are taken to be crucial for an ambitious decarbonisation scenario.

Due to the fact that these technologies have not yet been assigned such a major role in the energy system transformation process, sustainability analyses are very rare in the literature. In addition, PtX is integrated into a highly dynamic energy system. A static assessment assuming a constant power supply is only valid to a certain degree. In the case of larger PtX plants, the repercussions on the energy system would also have to be considered. Furthermore, even with the integration of small plants and the associated marginal change in load without an effect on the energy system, it is not possible to depict the effects using a classical LCA.

1.2 Limitations of Classic LCA Approaches and the Development Towards Dynamic LCA

LCA methodology is commonly used to quantify the environmental impact of products or service systems [24]. For liquid and gaseous energy carriers a lot of research has been done on biofuels [25]. PtX fuels, however, only recently gained more attention. Additionally, only few research has been done on dynamic LCA such as e. g. the integration of LCA methodology into time-dependent energy system modelling.

A big hurdle to apply LCA to complete energy systems is missing data on future key technologies and production plants [10]. Furthermore, there are methodical limitations to standard LCA approaches such as double counting and shifting system boundaries in the analysis of the total energy system [11]. Another problem of LCAs of energy systems is more inherent to the energy system itself. In the context of a decarbonisation of the energy system a greater amount of renewable energies will be integrated into the power plant park. This integration will lead to higher dynamics in the provision of electricity as electricity from wind and photovoltaic plants is subject to volatile generation. However, not only the total amount of generated electricity and its provision is important, but also the flexibility of the demand-side technology and its ramp-up time to react to changes in the load. This may especially be important for the production of PtX fuels as their economic profitability strongly depends on the full load hours [26]. Therefore, the often used assumption of just deploying excess electricity from renewables is not strictly valid, but may possess potential for PtX fuels [27]. For it to be specified, a time-resolved consideration of renewable electricity generation and resulting operation strategies for PtX plants are required.

In the following a modular, dynamised LCA approach is proposed aiming at providing a more detailed assessment of future energy scenarios and the problem of volatility. The developed methodology is applied to the described case of PtX technologies.

2 Developed Methodology and Exemplary Application

According to DIN 14040/14044 an LCA can be structured in four phases: the definition of goal and scope, the life cycle inventory (LCI), the life cycle impact assessment (LCIA) and the interpretation. In Sect. 2.2 a method is explained to modularization and dynamics the determination of the LCI in order to assess the production and application of a PtX fuel in transport. The modularization allows different system boundaries and therefore different scopes of the LCA. This topic is discussed in Sect. 2.1. As an exemplary application of the methodology, in Sect. 2.3, the production of a PtX fuel with a dynamic energy system providing the electrical energy input is discussed.

2.1 Discussion of Goal and Scope for the Proposed Methodology

The proposed methodology does not imply one particular goal. Moreover, the goal has to be defined as soon as the level of modularisation is set. If the environmental impact of the provision of 1 passenger-km (pkm) wants to be assessed, the scope and therefore the level of modularisation has to be set accordingly. But also parts of the life cycle, such as the provision of 1 kWh PtX fuel, can be assessed as explained in the following chapters. The scopes in these two cases are different and hence the system boundaries are different. Consequently, the modularisation adds flexibility to the choice of system boundaries.

In the following, first, Sect. 2.2 explains the method for the system boundaries according to a Well-to-Wheel analysis. Consequently, in addition to the operational considerations, the upstream chains of the individual life cycle stages are also considered. However, in Sect. 2.3 the focus is set on the fuel supply and the assessment method is shown for a functional unit of 1 kWh PtX fuel.

2.2 Modularisation and Dynamisation of LCA

Looking at the whole life cycle of a PtX fuel, it can be divided into energy supply, conversion, distribution and end use. Figure  2 illustrates these four steps within the system boundaries. It also shows which parts of the fuel life cycle are regarded as foreground and background processes. The foreground processes are part of detailed considerations, while the background processes are assessed using the database ecoinvent 3.5.

Fig. 2
figure 2

Modularisation for the LCA of PtX fuels for transport applications with integration of the energy system model ISAaR

Full size image

The energy system model ISAaR, developed at the Research Center for Energy Economics (FfE), provides the hourly German energy supply embedded in a European energy system. Among others, the energy sources considered include electrical energy, district heating and gas. By setting the boundary conditions, including the energy consumption by energy carriers, the dispatch of the supply systems for scenarios up to 2050 can be simulated [28] (see also Fig. 4). Additionally, the results contain economic as well as ecological indicators such as electricity prices and GHG emissions. The output of ISAaR in hourly resolution serves as an input for the LCA. Hence, the use of fuels for power supply can be considered a foreground process. This enables the time-dependent assessment of the environmental impact of the further PtX fuel production in the foreground and thus the inclusion of possible peaks in the power supply of volatile renewable energies. In addition, the applications such as cars and other means of transport are also considered in the foreground. This allows the comparison with transport applications using other energy sources with less favorable storage properties, such as electricity. In addition to the modeled foreground processes, the background processes are assessed with the help of the ecoinvent 3.5 database. Examples of background processes are the construction of required plants and means of transport as well as the provision of raw materials for power generation.

The modularisation of the LCA along life cycle phases is carried out to facilitate data integration and contribution analysis. It enables the smooth modification of known modules as well as the easy integration of new modules from e.g. PtX implementation projects. Furthermore, this approach allows individual sub-areas of the life cycle to be assessed in a transparent way, which simplifies the identification and communication of possible drivers of environmental impacts. As shown in Fig. 3 each module is represented by a separate technology matrix.

Fig. 3
figure 3

Matrix interpretation of the modules to carry out the LCA. The part of the scaling vector described as si,PM is the one to be handed over to the preceding system [see also Eqs. (4) and (5)]. The inventory vector gi can be derived for each module

Full size image

The matrix representation follows Heijungs et al. [29]. Therein the square technology matrix A is used to carry out the inventory analysis. A contains the input processes associated with a unit output process. For example, 1 m3 hydrogen is assigned a certain amount of power generation. Given a final demand f the scaling vector s is to be carried out with Eq. (1):

$$s = A^{ - 1} f$$
(1)

The scaling vector serves to scale the unit processes of the technology matrix. With the use of s and the intervention matrix B, it is then possible to calculate the environmental flow vector g:

$$g = B s$$
(2)

The emissions related to each unit output process are described in the intervention matrix B. It contains for example the CO2 emissions of the combustion of coal to produce 1 kWh of electricity.

To connect the different modules, the elements in the scaling vector s of the processes to be considered in a preceding module are handed over as elements of a new final demand vector f. This concept is clarified in the following chapter using the example of a PtX fuel.

Eventually, all environmental flow vectors gi are cumulated to derive the LCI. Ultimately, based on the LCI the LCIA can be conducted for each module separately or for all modules collectively.

2.3 Exemplary Application for the Production of PtX Fuel

The case of PtX fuel production is now considered more closely as an exemplary application of the methodological approach explained before. For this purpose, two systems of equations are set up. One represents the production of a PtX fuel and the other one describes the generation of electrical energy. The output of the whole system is a PtX fuel with the functional unit of 1 kWh referred to the lower heating value. According to Eq. (1) the first system of equations consequently results as follows:

$$\left( {\begin{array}{*{20}c} {a_{11,PtX} } & {a_{12,PtX} } & \cdots & {a_{1n,PtX} } \\ {a_{21,PtX} } & {a_{22,PtX} } & \cdots & {a_{2n,PtX} } \\ \vdots & \vdots & \ddots & \vdots \\ {a_{m1,PtX} } & {a_{m2,PtX} } & \cdots & {a_{mn,PtX} } \\ \end{array} } \right)\left( {\begin{array}{*{20}c} {s_{1,PtX} } \\ {s_{2,PtX} } \\ \vdots \\ {s_{n,PtX} } \\ \end{array} } \right){ = }\left( {\begin{array}{*{20}c} {f_{1,PtX} } \\ {f_{2,PtX} } \\ \vdots \\ {f_{n,PtX} } \\ \end{array} } \right) = \left( {\begin{array}{*{20}c} {1 \,{\text{kWh}}} \\ 0 \\ \vdots \\ 0 \\ \end{array} } \right)$$
(3)

Thereby the first row of APtX, sPtX and fPtX represent the PtX fuel production process. The final demand vector consists of the functional unit in the first row and zeros in every other row. Each column in the technology matrix APtX represents a unit output process. The rows contain the respective input processes. For the step of fuel conversion, the technology matrix contains, among others, the processes of operating materials such as water, CO2 and electricity as well as the construction of the synthesis plant. Every entry aij of the matrix where i = j is one. The system of equations is used to calculate the scaling vector as described above. The scaling vector contains entries of processes assessed in the PtX module and processes to be handed over to the previous module. The processes which are not handed over are used to determine the environmental flow of the PtX module gPtX. For the other processes the resulting scaling vector elements are handed over as the final demand vector element of the preceding system of equations (see Fig. 3). Thus, if row α in Eq. (3) represents the provision of electrical energy, the scaling vector can be divided into one to be considered in the preceding module (PM) and one to be considered in the current module (CM):

$$s_{PtX} = s_{PtX,PM} + s_{PtX,CM} = \left( {\begin{array}{*{20}c} 0 \\ \vdots \\ {s_{\alpha ,PtX,PM} = s_{\alpha ,PtX} } \\ \vdots \\ 0 \\ \end{array} } \right) + \left( {\begin{array}{*{20}c} {s_{1,PtX,CM} } \\ \vdots \\ {s_{\alpha ,PtX,CM} = 0} \\ \vdots \\ {s_{n,PtX,CM} } \\ \end{array} } \right)$$
(4)

Consequently, the final demand vector of the energy system (ES) results in:

$$f_{ES} = \left( {\begin{array}{*{20}c} {f_{1,ES} = s_{\alpha ,PtX} } \\ 0 \\ \vdots \\ 0 \\ \end{array} } \right)$$
(5)

Thus, the first row in the system of equations of the energy system corresponds to electrical energy. All other elements of the final demand vector are zero. With the use of the technology matrix of the energy system AES, the underlying environmental flow for the supply of electrical energy can now be determined as explained in Eqs. (1) and (2).

However, the high dynamics of the energy system must be included. Figure  4 illustrates the time dependency of the technology matrix of the energy system. A simulated dispatch for the provision of electrical energy for the year 2030 using the energy system model ISAaR is depicted.

Fig. 4
figure 4

Exemplary electrical dispatch for the hours 2000–3000 of the year 2030 and schematic representation of the technology matrices for two hours (t = 2300, t = 2700)

Full size image

The electrical energy generation of power plants is shown divided by type, such as i.e. gas, lignite and hard coal. In addition, the load and the export of electrical energy are presented. It can be seen that the deployment of the different types of power plants is subject to great variability. As an example, the technology matrices of two hours t = 2300 and t = 2700 are set up.

As a consequence, the resulting environmental flow will also vary over time. Therefore, the described system of equations must be solved for each time step, in this case each hour of the year.

Since in this example the energy system is the last set of processes to be considered, the cumulative environmental flow of the overall system for each time step can finally be calculated as follows:

$$g = g_{PtX} + g_{ES} = B_{PtX} \cdot s_{PtX,CM} + B_{ES} \cdot A_{ES}^{ - 1} \cdot f_{ES}$$
(6)

As the formation of the inverse of a large, sparsely occupied matrix is a complex arithmetic operation, it is important to keep the technology matrix AES and therefore also the intervention matrix BES and the final demand vector fES as small as possible. In contrast to the consideration of the whole system, the modularisation and the associated partitioning of the technology matrix already allows a significant reduction of complexity. Additionally, the second term can be cleared of unused processes. Then, in order to enable the addition with the vector gPtX, the vector gES must again be extended by the corresponding zeros to match the length of the vector gPtX.

In the present case, it is assumed that only the energy system varies over time. For example, in the event of a variable operation of an electrolyser, the system of equations for the conversion can also be time-dependent. In this case, the existing system can also be extended by a time dependency of the load. Therefore, e.g. partial load capable electrolysers can also be investigated.

3 Outlook and Future Projects

The methodology proposed above adds new aspects to classic LCA approaches. By being able to integrate distinct time series for the production of electricity and the related energy generation parks, a new level of dynamisation can be reached. The high temporal resolution adds to the accuracy of the LCA as the highly fluctuating composition of the electricity mix from different generation technologies can be taken into account. Additionally, the environmental impact of different electricity generation technologies can be compared according to different future energy scenarios.

The modularisation of the LCA adds the advantage to easily interchange, modify and add process data without the need to calculate the whole fuel life cycle all over again. Therefore, competing life cycles can easily be compared with regard to their environmental impact and new data of future technologies can be integrated to track environmental improvements. In addition, the computational complexity for the formation of matrix inversions can be reduced by the modularisation.

This newly developed methodology is implemented, in a first step, in a case study of different PtX production chains in the context of the project BEniVer. In a second step, a full integration into the energy system model ISaAR is aspired. By feeding the results of the LCA back into ISaAR it is possible to use them as an optimisation criteria in the plant expansion and deployment planning.