Analysis of Uncertainty Factors in Part-Specific Greenhouse Gas Accounting

Abstract

Due to the ongoing climate crisis and the resulting targets of several countries, as well as the EU, manufacturing companies face the need to address and reduce their greenhouse gas (GHG) emissions. To calculate these emissions, the product carbon footprint (PCF) can be a helpful tool, both to come up with reduction measures internally as well as to communicate it externally as competitive advantage. However, the PCF is subject to uncertainty factors that hinder its use and need to be systematically assessed. For this reason, this paper first collects and clusters relevant uncertainty factors resulting from differences between accounting standards, but also from imprecisions within the standards. Subsequently, the PCF is determined in different scenarios of uncertainty factors, applied to an industrial case study. This is used to be able to put the deviation in the final results into perspective. Recommendations for action are finally derived from the analysis of the literature and the results of the use case.

1. Introduction

Due to the rising average global temperature and the drastic consequences for the world’s population [1], the global community agreed in Paris in 2015 on a binding international treaty to limit global warming to well below 2 °C, ideally below 1.5 °C [2]. This goal is to be achieved primarily through the reduction of greenhouse gas (GHG) emissions, which has now been translated into country-specific legislation. Germany, for example, has committed to reduce GHG emissions to net zero by 2045 [3]. As a result, the production industry, as one of the largest emitters of GHG emissions [4], has an obligation to act with appropriate measures and reduce its emissions. The accounting and visualization of GHGs over the life cycle of a product, named product carbon footprint (PCF), is an important means to initiate changes and improvements in products and production [5]. The PCF can help acquire new customers who value companies’ efforts to increase sustainability and address new plans regarding a product pass published by the European Union [6].

Despite the advantages of PCFs as basis for improvements in production industry, companies hesitate to perform the necessary accounting. This is due to the fact that different accounting standards are available, the application of which can lead to sometimes significant quantitative variations in the overall result [7]. These variations are partly due to differences within the standards, regarding topics such as allocation rules or system boundaries [8]. However, the standards themselves also offer room for interpretation due to imprecise formulations, the impact of which has not been analyzed in detail. Therefore, the aim of this paper is to systematically investigate which factors cause uncertainties in part-specific GHG accounting and why. If uncertainty factors are analyzed in existing studies available in the literature, they are mentioned in the respective chapters. This is done in Section 3, after giving some theoretical background regarding PCF in general as well as uncertainty factors in Section 2. In addition, the literature and a concrete use case are used to illustrate how high this uncertainty can be and what impact it has on the useability of the result of the PCF in Section 4. That is followed by Section 5, giving a conclusion and outlook at the end.

2. Theoretical Background

2.1. Product Carbon Footprint

The principle of carbon footprints is used on different hierarchy levels such as organizations, states, projects, and persons [9]. However, this paper examines the PCF. The PCF represents a quantifiable metric for the comprehensive assessment of GHG emissions throughout the entire life cycle of a product or service. It encompasses a thorough consideration of all life cycle stages, spanning from the initial raw material extraction, through manufacturing and utilization, to the disposal phase [10]. Various GHG exhibit distinct greenhouse effects due to their varying heat-trapping properties, necessitating the attribution of a global warming potential (GWP) to each GHG. The introduction of GWP facilitates the aggregation of diverse GHG impacts within the framework of the PCF. These GWPs are expressed in relation to the reference GHG, carbon dioxide, using the standardized unit of carbon dioxide equivalents (kgCO₂e) [11,12].

Several standards exist for quantifying the PCF. The most globally established standards are DIN EN ISO 14067 [11], the Product Life Cycle Accounting and Reporting Standard (GHG Protocol) [13], and PAS 2050 [14]. As several existing studies on the comparison of different accounting standards emphasize their international relevance [15,16] and international studies on product-specific CO₂-accounting in particular make use of them [7,8], this paper focuses on the examination of these three standards. In addition, multiple national initiatives were introduced which, in contrast to the aforementioned international guidelines, possess limited suitability for facilitating the comparative evaluation of the GHG impact associated with products [12].

DIN EN ISO 14067 was first published in 2013 by the International Organization for Standardization (ISO) and is based directly on the life cycle assessment method (LCA) already established and standardized in DIN EN ISO 14040/14044 [11,17,18]. PAS 2050:2011 was developed by the British Standards Institution (BSI) as a national standard and published in 2008, but due to its worldwide recognition, it has assumed a higher status than other national standards [16]. In contrast, the Product Life Cycle Accounting and Reporting Standard, which is a further development of the Greenhouse Gas Protocol, was not developed by a standards body, but by the non-governmental organizations (NGO) World Resource Institute (WRI) and the World Business Council for Sustainable Development (WBCSD) in cooperation with various NGOs and companies [13].

Both PAS 2050 and the GHG Protocol adhere to the LCA methodology, similar to DIN EN ISO 14067, but deviate slightly from the conventional LCA standard in their approach to formulate certain procedural steps [12]. According to DIN EN ISO 14067, the procedure comprises four steps:

• Goal and scope definition: In this stage, a specification is made regarding the intended application of the PCF, the product system under consideration, and the selection of temporal, geographic, and process-related contextual parameters.
• Life cycle inventory analysis: All relevant resource inputs and outputs are quantified. This includes data acquisition, data validation, and referencing to the product system under consideration.
• Impact assessment: In this step, the impact assessment of the climate change effects of all recorded resource consumptions is carried out by converting them into CO₂ equivalents.
• Interpretation: Finally, the results of the PCF study are summarized in a report regarding the chosen methodological approach.

A thorough evaluation of each step as well as differences between the standards is conducted in Section 3.

2.2. Uncertainty Factors

Uncertainty factors are understood as parameters “associated with the result of a measurement, that characterizes the dispersion of the values that could reasonably be attributed to the measurand” [19] (p. 2). This definition follows the understanding of uncertainty in natural science disciplines. Other research disciplines use a different understanding of the term (e.g., social sciences), and these will not be discussed in the context of this paper [20].

The characterization of uncertainties can be either qualitative or quantitative, although a quantitative approach is preferable if scientifically applicable [21]. Uncertainty can be the result of either random deviations or systematic deviations. Random deviations are those that occur unpredictably and irregularly. Systematic deviations are effects that occur regularly and reproducibly. They can usually be compensated by correction [19]. In practice, uncertainties can originate from a multitude of sources and, in addition to external influences, may be attributed to subjectivity as well as assumptions and approximations inherent in the methodological approach [19]. Possible causes may be limitations in the database (e.g., gaps in the database or measurement uncertainty), as well as an inadequate study design [21].

Following this definition, this paper defines the PCF as a measurand, for which possible parameters that have an impact on the PCF result are thoroughly identified and assessed.

3. Analysis of Uncertainty Factors

As explained in the previous chapter, the ISO 14067 standard, PAS 2050, and the GHG Protocol are particularly relevant in the context of product-specific carbon accounting. To systematically investigate the uncertainty factors contained in these standards, first the topics addressed in the standards were identified. Then, by comparing the standards with each other and analyzing the relevant literature, uncertainty factors are identified in these areas. Subsequently, it is determined in more detail what causes the uncertainty in the final result in each case. The result of this systematic analysis is presented in the following for each topic. Finally, based on the literature, a first assessment is given about which factors can cause uncertainty, and whether it will be investigated in the case study in Section 4. Figure 1 provides an overview of the determined uncertainty factors and those further examined in the case-study.

3.1. General Specifications

This section contains topics that relate to overarching aspects of carbon accounting. In this sense, the three aforementioned standards overlap to a large extent. All standards pursue the goal of life cycle assessment, i.e., they consider the entire life cycle of a product. DIN EN ISO 14067 introduces the term partial carbon footprint, which corresponds to the PCF of one or more selected processes within this life cycle. The commonality of the standards also lies in the consideration of the impact category of global warming as the only impact category. In addition, all three standards specify key calculation principles that must be considered when conducting a PCF study. These include, for example, completeness, transparency and accuracy, so that the accounting result is comprehensible, and uncertainties are reduced. However, these principles are formulated in a general way and do not represent precise rules whose influence on the PCF can be assessed qualitatively or quantitatively. As already mentioned in Section 2, there are no major differences between the standards in terms of the accounting methodology. In all three standards, the definition of a functional unit is mandatory, meaning the definition of a quantifiable benefit as a unit of comparison. In practice, this is a very relevant aspect when it comes to comparing similar products. For example, if the influence of an alternative raw material is evaluated, possible changes in durability or efficiency must be considered. For the comparability of products, it must therefore be ensured via the functional unit that the resulting benefit is quantified and remains the same.

The only uncertainty factor in the area of general specifications is the use of product category rules. This refers to more precise specifications for products of the same category. All three standards recommend their use if there is a corresponding specification, but leave room for interpretation so that this does not necessarily have to be performed. Since the use of product category rules can influence and improve the quality of the results [22], this can create uncertainty depending on whether existing rules are used or not. The standards thus create an uncertainty factor here by imprecise formulation. In the case study in Section 4, the factor is not examined due to a lack of product category rules for the specific product.

3.2. System Boundaries

The system boundary defines which processes and activities are included in the PCF calculation. For example, it specifies the examined GHGs. This is where the three standards differ from each other. The GHG Protocol prescribes the use of six GHGs: CO₂, CH₄, N₂O, SF₆, HFCs, and PFCs [13] (p. 27). The PAS 2050 includes additional GHGs recognized by the IPCC [23], [14] (p. 26), whereas the DIN EN ISO also includes additional GHGs, but explicitly excludes water vapor and ozone [11] (p. 19). As a matter of fact, the GHGs that are not prescribed in the GHG Protocol only have a very small share in the global greenhouse effect [24], which is why the influence on the result of the PCF is estimated to be negligible.

The standards agree in the specification of the necessary product life cycle phases. All three prioritize the assessment of the entire life cycle of a product (cradle-to-grave), but also allow consideration from raw material extraction to the factory gate after production (cradle-to-gate) for products. Therefore, this is no uncertainty factor. That is different when the duration of the utilization phase is considered. Here, all three standards leave room for interpretation if there is no precise knowledge of how long a product will be in use. The fact that the time span can then be determined by the companies themselves can lead to different PCF results. Uncertainty therefore exists because of imprecision in the formulation of the standards. The same affects the application of cut-off criteria, which are intended to determine when process steps and modules can be neglected. For example, PAS 2050 states that all processes that cause more than 1% of the total emissions of the product must be included [14] (p. 13). However, it is not possible to make a concrete statement about the share of processes before the analysis, except if it is based on an assumption. It is therefore possible that different assumptions are made, which can lead to different overall results. The exclusion of processes and their components by applying cut-off criteria is possible for all three standards; consequently, it is an uncertainty factor due to imprecision. In the literature, however, there are studies that classify the influence of this uncertainty factor as low [7,8].

The considered GWP-timelines also represents an uncertainty factor, but in this case is not based on the standards. The timeline specifies the period to which the conversion of GHGs refers (see Section 2.1). All three standards agree on the specification of 100 years—even if DIN EN ISO 14067 itself states that this specification has no scientific basis [11] (p. 66). Criticism of this definition can be found in [25] for example, as it does not sufficiently differentiate between the effects of short-lived and long-lived climate emissions. No study could be found in the research for the presented paper that takes a systematic look at the effects of the GWP time horizon on the overall PCF. Since it can be relatively easily assessed by using GWP databases, this uncertainty factor is also taken up in the case study (see Section 4).

3.3. Data

Data form the basis of PCF calculations and come either from measurements, calculations, estimation, or other external sources. All three standards distinguish between primary and secondary data, depending on whether, for example, direct measurements were performed. The standards use slightly different classifications for primary and secondary data [11] (pp. 28–29), [13] (pp. 50–54), [14] (pp. 5–6), which in itself does not lead to uncertainties in the PCF result. However, all three standards do not specify to which extent the use of secondary data are approved. It is only emphasized that primary data should always be preferred. The use of secondary data generally contribute significantly to uncertainties in the PCF result [26], which is why the (missing) specification on the use of secondary data is classified as an uncertainty factor due to imprecision. The consequent differences that arise regarding the PCF result are also examined in the case study in Section 4.

The standards set data quality criteria. For example, the extent to which the data corresponds to the product system in terms of time and geographical region must be documented. The more accurately the real product system can be described by the data, the better. However, only the GHG Protocol names criteria, but still leaves room for interpretation to the user [13] (p. 55). So, it can be concluded that no objective criteria for assessing quality are available in any of the three standards, which is why uncertainties can arise in the PCF result. The case study examines in more detail how the result changes when emission factors from different geographical regions are used.

A further question to be investigated is which source for (secondary) data is obtained. Here, all three standards are vague and only give examples of where secondary data could come from. These include life cycle databases, the published literature, and national life cycle inventories, e.g., for standard emission factors [11] (pp. 28–29), [13] (p. 53), [14] (p. 19). This is often a source of uncertainty, especially for emission factors used to convert specific resources into CO₂ equivalent emissions, as there are sometimes considerable differences between various sources [27]. For this reason, the influence is also investigated quantitatively in the case study.

The emission data for electricity-related emissions are listed separately at this point. The emission factor used for electricity can obviously vary depending of the source, but this can lead to large differences in the result of the PCF [28] (p. 201), especially if renewable sources are used. All three standards prescribe for electricity from renewable sources that a lower emission factor may only be used if the generated ‘renewable’ electricity is then excluded from other calculations regarding the grid. Otherwise, the use of the average factor of the grid has to be used. However, all three standards leave room for interpretation, e.g., with regard to the use of emission factors from suppliers, which is why a quantitative investigation is carried out in Section 4.

3.4. Allocations

If data is only available for larger areas of a company, it is necessary to allocate it to individual products when calculating the PCF. Allocation methods for production can be used for this purpose, even if all three standards consider it preferable not to use allocations. However, if this is unavoidable, for example if two products are produced in one process, there are various options for allocating the material and energy quantities used for the two products. First of all, physical relations, e.g., via mass allocation, should be used for this purpose. If this is not possible or reasonable, economic relations, e.g., via the generated turnover, can also be used [11] (pp. 50–51), [13] (p. 62), [14] (p. 22). However, all three standards leave the selection of the allocation method up to the user, which may lead to uncertainties in the final PCF [29].

The allocation in recycling of process waste also represents a challenge in the calculation of the PCF. All standards distinguish between open-loops and closed-loops. In the case of open-loops, the waste is an input for another product, which means that the emissions associated with the processing of the waste do not have to be accounted for, according to the standards. In closed-loops, the waste is an input in the same product system, which additionally reduces the emissions of the raw material [11] (p. 52), [13] (pp. 72–73), [14] (p. 31). Here, it is assumed that uncertainties can arise because of the standards’ imprecision in the choice of the specific allocation method used for recycling of waste in the same or in another product system. The influence of both, and the allocation in the production and in the recycling, are examined in more detail in the case study.

3.5. Approach for Special Emissions

In the following, special emissions regarding the PCFs are considered in more detail, which do not apply to all PCF accounting studies. First, this includes capital goods that are important for the manufacturing of the product. In production, these particularly include the machinery and equipment, but also the factory building itself. Only the GHG Protocol states that these capital goods must also be included in the balance sheet [13] (pp. 35–36), the other two standards leave room for excluding them. This results in an uncertainty factor due to differences between the standards. In the literature, there are conflicting assessments of the extent to which these emissions affect the overall result [8,29].

If biogenic materials (e.g., wood) are used in products, an uncertainty factor also arises as to how the carbon storage in the product with this biogenic fraction is to be handled. DIN EN ISO 14067 explicitly excludes offsetting with the emitted emissions [11] (p. 55), but PAS 2050 and the GHG Protocol require this [13] (p. 27), [14] (p. 10). For products with a high proportion of biogenic materials, it can therefore be assumed that a high degree of uncertainty arises, depending on which standard is applied [7,8].

Differences between the three standards also become apparent when dealing with time-delayed emissions that occur over the duration of the product life cycle. If, for example, energy quantities are only required years after the manufacture of a product, these can be accompanied by completely different underlying emissions since renewable energy sources might have been increasingly expanded in the meantime. Only DIN EN ISO 14067 [11] (p. 54), [13] (p. 89), [14] (p. 10) allows the use of justified factors for this purpose; in PAS 2050 and the GHG Protocol, emissions are to be calculated as if the resources were consumed at the time of manufacture. Depending on the standard, this can also lead to different results of the PCF [7,30].

There is also a difference between the standards in the way they deal with transport emissions caused by customers traveling to purchase a product. While PAS 2050 excludes these emissions altogether [14] (p. 17), the GHG Protocol only wants the journey to the place of purchase to be accounted for, but not the return journey [13] (p. 35). DIN EN ISO 14067 leaves a great deal of flexibility as to whether these emissions should be attributed to the product or not. Depending on the product, these transport emissions can have a considerable influence on the final result, but are often very difficult to quantify and are therefore usually based on estimates [29].

If goods are transported by air traffic, emissions occur at high altitudes, hence the impact on the greenhouse effect is higher than that of emissions generated on the ground [12]. The exact factor for plane emissions by which this increase occurs is scientifically disputed [12] (p. 55); DIN EN ISO 14067 [11] (p. 62) as well as PAS 2050 [14] (p. 10) exclude factors in the accounting, where the GHG Protocol [13] (p. 90) leaves this open, but also does not state a concrete factor to use. Based on the literature [31], it is therefore assumed that it is as well an uncertainty factor due to different specifications of the standards, and that differences in the final result of the PCF can occur.

However, the three standards agree on the accounting of off-setting certificates based on measures or avoided emissions by the use of the product (e.g., in the case of insulating materials) ([11] (p. 18), [13] (p. 89), [14] (p. 12, 17)). In all three standards, such offsetting is prohibited in the PCF, so there is no uncertainty factor here.

The product considered in the case study in Section 4 does not contain any biogenic components in the material, it is an intermediate product that does not generate any emissions in the utilization phase itself, and flight emissions also do not apply. In addition, no data are available in connection with the capital goods, so that no uncertainty factor from this topic area is examined in the case study. For other products, however, this topic should always be screened carefully for relevance due to the large number of uncertainty factors.

3.6. Investigation Results

In the previous chapters, various topics were identified that should be considered in a PCF study. It was examined whether, and to what extent, uncertainties can occur in these topics and sub-topics. The predominant amount of uncertainty factors arises from the fact that the three accounting standards under investigation provide imprecise or unspecific information for certain topics. The resulting possibilities for interpretation can lead to considerable variations in the final PCF result, as can already be seen in various studies in the literature.

Aside from the examined GHG emissions with presumably low impact, differences between the standards actually only occur in the approach regarding special emissions, such as emissions during the use phase. These special emissions do not apply to all products, so that it can be concluded that the choice of a standard itself is not of decisive relevance. However, this does not apply to products where the special emissions explained in Section 3.5 occur. Here, the standard should be chosen carefully to obtain a PCF result that is as realistic as possible. For that, it is recommended to choose the standard which specifically considers the respective special emissions.

4. Use Case—Quantitative Analysis of Uncertainty Factors

In the following chapter, the previously determined findings are applied to a real industrial use case to determine the influence of the identified uncertainty factors on the result of the PCF for this case. To this end, the product and the associated production system are first presented, followed by an explanation of how the analysis is structured and how it is carried out. The results are then discussed in Section 3.

4.1. Introduction Product and Production System

The examined product is an impeller that conveys the fluid within a chemical pump. The impeller is manufactured in eight process steps. Figure 2 shows the impeller, the eight process steps and the respective input and output resources.

In the first step, a mold is assembled, into which the plastic granulate is pressed in the second step. Afterwards, the mold must be removed by an etching process so that the pressed plastic impeller is left exposed. This is followed by a drying process, after that the impeller is sandblasted. Protruding parts outside of the impeller’s geometry are removed during the sixth step in a turning process. In the seventh step, the impeller is checked to ensure balanced rotation. If this is the case, it is finally tested and packaged. Parallel to the eight process steps, electricity is also needed for the hall lighting and ventilation.

4.2. Structure and Approach

In order quantify the influence of the various uncertainty factors, the CO₂ emissions of the product, described in Section 4.1, are calculated several times depending on the design of the respective uncertainty factors. Based on the differences of the results, the evaluation of the uncertainty level of the respective uncertainty factors is conducted. The results relate to complete PCFs of the impeller as well as the emissions of individual process steps for in depth analyses.

The potentially relevant uncertainty factors have already been identified in Section 3. However, the product, its production system as well as the data available in the use case, limit the practical investigation of all uncertainty factors. Therefore, only specific uncertainty factors were chosen. Regarding these uncertainty factors, different representative settings were used in the scenarios for the calculation. One set of settings was chosen in the so-called base scenario, which is assumed to describe the product and its related emissions most accurately. Using the base scenario, one uncertainty factor is varied at a time to assess the impact of each factor individually. The uncertainty factors as well as the scenario settings are depicted in

Table 1. Scenarios and corresponding characteristics of the uncertainty factors under consideration, base scenario marked in bold in the right column.

Uncertainty Factor	Scenario Settings
1. Data source for emission data	IPCC (ecoinvent)
	ReCiPe (ecoinvent)
	ProBas (UBA)
2. Use of secondary data	Exact consumptions
	Process time & connected power
	Process time & factorized connected power
3. Geographical region of emission data	Global
	European
	German
4. Emission data for electricity	German electricity mix
	European electricity mix
	Electricity from wind
	Electricity from coal
	Electricity from natural gas
	Electricity from nuclear power
	Chinese electricity mix
5. GWP-timeline	20 years
	100 years
	500 years
6. Allocations in production	Without allocations
	Mass related allocations
	Quantity allocations
	Production time allocations
	Value allocations
7. Recycling	Without recycling
	Open-loop circle
	Closed-loop circle

. The base scenario settings are highlighted in bold.

When calculating the GHG emissions, the emissions of all processes and activities in the raw material extraction and production phases (cradle-to-gate) must be recorded and summed for each scenario. In the case of the impeller, there is no direct measured emissions data for either the materials used in the impeller (raw material extraction phase) or the production processes in the manufacturing phase. For this reason, the emissions of the materials and processes are calculated based on emission factors and consumption data.

To calculate the PCF, the consumption data are multiplied with the respective emission factor. The emissions of the individual process streps, so-called activities, can then be summed up in various ways, either to obtain emissions of individual processes, to obtain the emissions of the material or energy separately, or to obtain the PCF. Hottenroth et al. show exemplary formulas for calculating product-specific CO₂e emissions and their intermediate results in their guide [12].

The emission factors are taken from the databases IPCC 2021 (via ecoinvent.org, accessed on 22 November 2023), ReCiPe 2016 (via ecoinvent.org, accessed on 22 November 2023) and ProBas (uncertainty factor 1). For the baseline scenario, the IPCC database is used, as it contains the most emission factors and thus provides the best description of the product system. For the calculation of the different scenarios, among other things, the emission factors are varied to correspond to the changed design form of the uncertainty factor, such as regarding the geographical origin (uncertainty factor 3) or the electricity mix (uncertainty factor 4). For uncertainty factor 2, the exact consumption values measured for each production step with external sensors were varied with values based on the connected power. Since the connected power alone does not indicate the consumption, it was multiplied with the process time in one scenario, and additionally with a usage factor in the second scenario. The usage factor was chosen to consider that only a certain amount of power is actually consumed while the machine is running. In a study examining 32 machine tools, Petruschke et al. [32] found out that only 19% of the connected power was actually consumed on average. Therefore, this factor was applied to the use case as well, to make a statement about the added value of the measurements in relation to the final PCF result in the case study. With uncertainty factor 5, the impact of the GWP timeline is investigated, since the recommended base of 100 years is not scientifically backed [11] (p. 66).

The input and output resources regarding the product are energy consumptions (in kWh for electricity and in m³ for compressed air) and material consumptions (in kg), which were determined at the individual process steps in the manufacturing phase. The consumption of the activities is related to the so-called functional unit, in this case that is one impeller. The consumption data was partly varied while calculating the scenarios regarding uncertainty factor 6 of allocations in production. For the baseline scenario, the values measured with external sensors were used to obtain the value for each machine. In the other allocation scenarios, different allocation methods were used, for example the mass allocation, where the yearly consumption of resources in the production line of impellers is split between different impeller-types according to the mass-relation of these types. For uncertainty factor 7, different possibilities for recycling were considered. In the baseline scenario, no recycling was assumed, whilst in the other scenarios, a closed-loop and open-loop recycling was considered according to the standards (see Section 3.4).

4.3. Discussion of Results

Based on the framework described in the previous chapter, a quantitative assessment of the uncertainty factors is conducted for the impeller in this chapter. To achieve this objective, the initial step involves assessing the baseline scenario. Therefore, a preliminary assessment of the GHG emissions of individual processes and resources is carried out, which serves as a reference for the analysis of all further scenarios. The key results of the baseline are presented in Figure 3. The total GHG emissions calculated for the impeller were distributed between material and energy flows. GHG emissions from material flows account for 72% (134 kgCO₂e, respectively) of the PCF, whilst GHG emissions from energy flows account for 28% (51 kgCO₂e, respectively).

The high proportion of used materials in the PCF is largely due to the assembly process (1), whose GHG emissions of approx. 95.5 kgCO₂e contribute 71% to the total of all GHG emissions from material flows. Those GHG emissions are due to the amount of aluminum used (GHG emissions of approx. 75 kgCO₂e), which acts as an auxiliary material for production and is not part of the impeller. The GHG emissions from this process step also constitute a significant proportion of the total PCF, accounting for half (approx. 51%) of total GHG emissions connected to the product system itself.

Figure 3 shows that process steps (1), (2), and (3) make the largest contribution to the PCF (over 90%). In contrast to process (1), in which the emissions originate almost exclusively from material flows, the GHG emissions of processes (2) and (3) also result to a considerable extent from their energy flows—with 49.5% and 65.2%, respectively. Thus, the product system under consideration and the corresponding manufacturing process are considered suitable for analyzing the selected uncertainty factors, which have an impact on both the energy-induced and the material-induced share of the PCF. In addition to these production-related process steps, energy consumption for lighting and ventilation, calculated together as superordinate processes and allocated per product via processing time regarding the whole area, also accounts for a significant proportion of PCF, namely 7.1 kgCO₂e.

If the results of the baseline scenario are used to derive measures to reduce the use of resources, the process steps mentioned represent the first starting point, with the hotspots of materials for process steps (1) and (2) and the energy used for process steps (2) and (3) standing out.

As previously stated, in accordance with the baseline scenario, the different settings for each uncertainty factor were varied in the calculation one at a time. Therefore, the calculation of different PCF-results was possible, as depicted in Figure 4. Here, the red line marks the PCF of the baseline scenario: 185 kgCO₂e. When using a different data source for the emission factors, this PCF of the impeller is 2% higher in case of the ReCiPe-method and 18% lower in case of the ProBas database. However, it has to be noted that the emission factors of certain resources varied to a much higher degree, so that the GHG emissions related with brass, for example, are 72% lower when using ProBas instead of the IPCC database. Based on the results of the case study, it is therefore assumed that the deviation caused by the uncertainty factor data source can be even higher in other products which use more complex input or output resources.

The uncertainty factor related to the use of secondary data shows a much higher deviation for the PCF: +22%, if only the connected power and the process time is used, and −19% if that is factorized as described before. Firstly, it must be noted that this uncertainty factor influenced only the emissions due to energy flows—the emissions for material flows were calculated in the same way as for the baseline scenario. Therefore, it is assumed that production systems containing more energy intensive processes could result in even higher deviations in the final PCF result. Furthermore, the factor used in the second calculation scenario was determined for machine tools. In the production steps of the impeller, mostly other machines like a press or an oven are used; hence, it is unsurprising that the factor is inaccurate.

For the third uncertainty factor, emission factors regarding different regions were used. All emission factors were available as global values, representing a global average for example for transport emissions between producers and costumers of purchased parts. Wherever possible, emission factors were used that are as close to the production location as possible to minimize deviations based on national differences regarding the environmental impacts. In the case study, this meant using factors related to the European region and Germany, since the production facility is located in Central Germany. Only for the resources leach and sand, no European values were available, whereas for most of the resources, an emission factor related to the German region was not available in the used database. In those cases, the emission factors for the European region or the global region, respectively, were used. Hence, it comes with no surprise that the results for the two scenarios are close together, deviating +26% and +27% from the baseline scenario. The fact that the PCF result is higher in the European and German scenarios is based on the higher emission factors, which are not explainable in detail by the authors. It might be due to longer transportation distances to Europe and Germany, for example in the case of aluminum, which is mostly produced and consumed in China, hence keeping the value with global relation lower.

The emission factors regarding electricity flows were varied with uncertainty factor 4. Six different scenarios were calculated additionally to the baseline scenario, where the value for the German electricity mix was used. In fact, this uncertainty factor showed the highest deviation in the final PCF, ranging from +38% in case electricity from coal is used, and −27%, in case the electricity would come from nuclear power. As mentioned already for the uncertainty factor 2 (use of secondary data), the deviation is expected to be even higher for products with a higher share of emissions coming from electricity flows. However, this uncertainty factor is also expected to be the one most easily eliminable, since the information about which source of electricity a supplier is using should be available in the contract or the energy bill.

A less tangible uncertainty factor is the one in case 5, where the timeline of the GWP is considered differently. Therefore, emission factors for 20, 100 and 500 years were extracted from the database and used to calculate the PCF. As expected, the GWP scenario related to the next 20 years shows the highest PCF with 218.8 kgCO₂e, making it 18% higher than the GWP-100-factors used in the baseline scenario. The values related to a timeline of 500 years, however, resulted in a 7% lower PCF. Since all standards examined in this paper require the use of the GWP 100, the deviation caused by this uncertainty factor might be irrelevant when comparing different products. Nonetheless, it is interesting to see the influence of this decision, which is questionable [25] or at least not scientifically backed [11] (p. 66).

For uncertainty factor 6, overall annual consumptions in the impeller production area are divided with different allocation methods to the specific impeller type considered in the use case. It must be noted that these annual consumption data were only available for the process steps mounting, etching, and sand blasting. Thus, the data was allocated only in these three process steps. For the other process steps, the values of the baseline scenario were used. The other two impeller types produced in the production line are produced less and sold for a lower price (due to lower-value plastic granulate) than the chosen impeller type, but they weigh more. With a minimum deviation by −4%, the mass allocation works best in this case, the allocation based on the economic value of each impeller type results in the highest deviation of +21% for this uncertainty factor. This backs the recommendations made by the standards to use allocation methods which are based on physical relations, rather than economic ones (see Section 3.4). To draw a conclusion about which allocation method is most suitable in production processes like the case study, more data, especially for the neglected process steps, would be necessary.

The last uncertainty factor investigated is related to recycling (7). In the baseline scenario, no recycling is assumed, whereas in the other two scenarios, recycling is assumed for the turning process step, where plastic scraps are created as waste. Assuming recycling of these scraps in an open-loop means the scraps are used as inputs for another product, therefore emissions related with this recycling can be left out of the calculation. These savings even increase when assuming the scraps are used in the same production line again, resulting in less input of raw material for the impeller. Therefore, the closed-loop-scenario results in the highest deviation of −3% regarding this uncertainty factor. In summary, the deviation caused by this uncertainty factor is the lowest compared to the deviation caused by the other uncertainty factors. Nonetheless, since raw material is mostly one of the largest contributors to product emissions, as also backed by the results of this study, using recycled materials is supposed to gain increasing importance, making this factor more relevant in the future.

The comparison of the different scenarios shows that the factors emission data for electricity (4), use of secondary data (2), and allocations in production (6) cause the highest uncertainty in the present case study. However, the options for resolving these uncertainties differ significantly from scenario to scenario.

While the selection of emission factors regarding electricity, the data source or the GWP-timeline can be supported by explicit guidelines to achieve a more homogeneous approach, the resolution of the other two uncertainty factors is much more difficult to achieve as they are highly dependent on the respective product and production system, as well as the existing database. The overall recommendation can be given to use as much primary data as possible to avoid secondary data and allocations, in the best case completely. Therefore, concrete regulations could be made by overarching institutions about which resource consumptions should be gathered with measurements, possibly differentiated by industrial sector.

In addition to the comparative analysis of all scenarios, an exemplary investigation of possible effects on the individual shares of the PCF and the resulting implications is carried out below using uncertainty factor 2: the use of secondary data. Figure 5 shows a process-specific analysis of the shares of the PCF across all scenarios varying the use of secondary data. Each bar in color stands for one of the three scenarios, differentiated between material and energy induced GHG emissions.

Using this visualization as an example for the aspect of secondary data, it can be shown how the varying uncertainty factors can potentially affect the evaluation and decision-making process. First, it is evident that the prioritization of the individual process steps against each other does not change across the different scenarios, meaning the evaluation of which process steps have the largest share of the PCF. However, the proportions of the shares in the PCF change significantly in some cases, and in particular the share of energy induced and material induced GHG emissions for individual process steps also changes. The derivation and prioritization of measures to reduce the PCF can certainly be influenced by these fluctuations caused by uncertainties. In other applications in industrial practice, it is also conceivable that the assessment of the individual process steps changes from scenario to scenario. However, this is not the case in this case study.

To draw a conclusion regarding the question of highest uncertainty for the case study, a minimum and a maximum scenario is constructed based on all possible scenarios of the uncertainty factors. Not all scenarios are reasonable in the case study since, for example, no energy from nuclear power is available in Germany anymore, hence wind energy is considered for the minimum scenario. For the maximum scenario, the ProBas database is chosen, only providing GWP 100 values. So, a combination of ProBas and GWP 500 was not possible but is expected to result in even higher numbers.

The differences between the scenarios, shown in Figure 6, are quite significant. The PCF of the minimum scenario is only about half as high (85 kg CO₂e vs. 185 kgCO₂e), whereas the PCF of the maximum scenario is more than twice as high as that of the baseline scenario (430 kgCO₂e vs. 185 kgCO₂e). In a direct comparison of the PCF of the two new scenarios, the maximum scenario has a PCF more than five times higher than the minimum scenario. In both scenarios, it is also noticeable that the energy-induced GHG emissions vary more than those of the material. In the minimum scenario, this is due to the emission factor of the wind energy, which is almost zero. In addition to the emission factor of pure coal-fired electricity, the maximum scenario also uses the imprecise activity data, which was determined using the product of the connected power and the process time, so that the combination of both scenarios causes an additional increase in emissions from electricity.

5. Conclusions and Outlook

After the basics of PCFs and uncertainty factors were explained, an in-depth analysis of the calculation standards DIN EN ISO 14067, GHG Protocol and PAS 2050 was given. The three standards were compared regarding different aspects and uncertainty factors were identified which could cause varying results in PCFs. Remarkably, these uncertainty factors mostly arise from interpretation possibilities inside the standards themselves, not from differences between the standards. In this context, studies comparing the results with further national standards from other parts of the world like China could be a very interesting focus for future research.

After this analysis, selected uncertainty factors were investigated in a case study. A real industry product, namely an impeller for chemical pumps, was analyzed regarding its PCF and deviations caused in different scenarios. The deviations varied from +38% to −27% when varying one uncertainty factor at a time. If several uncertainty factors were varied at once, a minimum scenario with a deviation of −54% as well as a maximum scenario with a deviation of +134% could be reached for the same product. This deviation in the final result is especially significant if a company is using the PCF in outside communication, making it very difficult to compare similar products from different companies. This could be avoided or at least mitigated by eliminating the uncertainty factors beforehand, for example by concrete customer requirements that each supplier would need to follow for the calculations. How these requirements could be set, and which uncertainty factors actually cause the highest deviations in a specific use case, should be investigated in further research studies, ideally in cooperation with industry partners.

Nevertheless, the case study also showed that, no matter the setting in the different scenarios, the process steps that are identified as most relevant regarding GHG emissions remain the same. In the case study, those are always the first three process steps: assembly, pressing, and etching. Therefore, it is assumed that gathering improvement measures would be the same for the impeller, resulting in the same improvement measures independent of the chosen scenario. That is an encouraging message for companies applying a PCF calculation as hotspot method to eventually identify improvement measures. At the moment, to calculate PCFs, especially in companies with many product variants, is very time-consuming because a high amount of data has to be gathered from different sources or departments. To overcome this obstacle, and possibly reduce uncertainty due to secondary data at the same time, concepts for automated data acquisition combined with the automation of PCF calculations, described for example in [33,34], are expected to offer great potential to especially reduce uncertainty arising from the use of secondary data and allocation. Therefore, more research and case studies should be conducted in this field to persuade more companies to calculate PCFs and subsequently reduce GHG emissions.