Digital twins<\/a> play a key role in this context because they digitally represent real-world processes, integrate internal and external data, and enable simulations and optimizations [8, 9]. In manufacturing<\/a>, digital twins are increasingly being used for dynamic production planning tasks, particularly to improve real-time capabilities, although this field of research is still in its early stages [10].\u00a0<\/p>\n\n\n\n The research project greenProd is currently developing green digital energy twins (gDEZ) to support the manufacturing industry\u2019s transition to renewable energy sources. The gDEZ is a digital representation of production steps, products, and energy sources and serves as a concrete key component for emissions-oriented production planning. It makes process-related energy consumption transparent, links it to emissions data, and thereby supports emissions-oriented decisions in operational production planning [11]. In addition to internal process information, external data is crucial. Digital twins become practically useful and decision-relevant only when these data sources are combined [12]. <\/p>\n\n\n\n Unlike a purely static digital model, the digital twin enables the continuous integration of external data (e.g., time-varying emission intensities) as well as the dynamic evaluation of alternative planning scenarios [8, 12]. The digital twin does not serve as a direct control system, but rather as a decision support system for operational production planning. <\/p>\n\n\n\n For industrial practice, it is crucial to assess the benefits of the gDEZ prior to implementation. The literature emphasizes that the introduction and operation of digital twins are often associated with high costs, which can represent a significant barrier to their implementation [8]. Meta-reviews also show that the concrete benefits of digital twins are often not reliably quantified under realistic operational conditions [13]. <\/p>\n\n\n\n Against this backdrop, it remains unclear whether digital twins can significantly reduce emissions under practical constraints and thus whether the benefits justify the implementation effort. This leads to the following research question: How large is the organizationally achievable emission savings potential through the use of a conceptual digital twin under realistic practical constraints? <\/p>\n\n\n\n To determine the organizationally achievable emission reduction potential, a case study on a hydraulic pump test bench calculates GHG emissions based on real production and energy data and compares them to an optimized case derived through rule-based, emission-oriented rescheduling under practical constraints. The approach is based on a conceptual digital twin as a target system that links internal planning data with process-related energy information and external emissions data, thereby enabling time-resolved emissions assessment and the computational comparison of alternative production plans. <\/p>\n\n\n\n The paper thus presents a transparent methodology for quantifying production-related GHG emissions and emission reduction potential through an ex-ante benefit assessment.<\/p>\n\n\n\n The case study is conducted on a test bench in an industrial production environment. The underlying production system is classified as series production and is used for the recurring manufacture of both standardized and customer-specific hydraulic pumps. Operational test planning is carried out using a two-stage system consisting of high-level and detailed planning. High-level planning covers several weeks, while detailed planning is created at the beginning of each calendar week. Scheduling and execution are carried out via test orders, with different sizes corresponding to varying test durations.<\/p>\n\n\n\n Setup operations are required between individual test orders, and these take varying amounts of time depending on the sequence. In current planning practice, scheduling is optimized to minimize setup time without explicitly accounting for GHG emissions. The load profile of the test bench is determined by the test sequence and generates characteristic consumption patterns during multi-shift operation. Operational test planning is particularly relevant for analyzing emission reduction potential, as electrical energy consumption falls into intervals of varying emission intensity depending on the time of scheduling.<\/p>\n\n\n\n The study uses an exploratory case study methodology to systematically estimate potential GHG savings. This methodology combines established approaches to emissions-based accounting with application-specific rule-based rescheduling. The tasks of the conceptual digital twin include linking historical planning and energy consumption data from the year 2025 with external emissions information as a basis for deriving the load profile and the total emissions.<\/p>\n\n\n\n Building on this, a rule-based resequencing is performed using defined optimization rules, and a load profile for the optimization case is created. The choice of a rule-based approach is deliberate, as it enables a transparent and practical estimation of the emission reduction potential under operational constraints while also being implementable even with limited data availability.<\/p>\n\n\n\n In contrast to exact optimization methods or complex heuristic approaches, rule-based rescheduling allows for a transparent representation of operational constraints with minimal implementation effort. The goal here is not to determine a globally optimal solution, but rather to provide a realistically implementable estimate of the achievable potential. The individual steps of the methodology are outlined below.<\/p>\n\n\n\n The test orders from the original test plan are first assigned to a uniform 15-minute interval grid. Using average energy consumption data, the order-specific energy consumption is then determined, distributed across the respective intervals, and aggregated into a consumption profile. By linking this to the time-dependent 15-minute emission intensity values of grid electricity, a continuous emission profile for the reference case is created, from which total emissions are calculated.<\/p>\n\n\n\n Taking into account operational and organizational constraints, optimization rules are defined that relate exclusively to the scheduling of test orders. Changes to processes or technical parameters are explicitly excluded. Rescheduling follows application-specific optimization rules and does not claim mathematical optimality in the sense of a globally optimal solution.<\/p>\n\n\n\n Rule-based rescheduling is limited to shifts within regular shift hours (6:00 a.m.-10:30 p.m.) and is performed on a weekly basis. To enable rescheduling to time slots with lower emissions, larger test orders can be divided into subsections if necessary; in doing so, it must be ensured that all scheduled orders remain scheduled within the respective calendar week. Since test orders in the initial plan may extend over the weekend into the next calendar week, a consistent week-to-week transition must be ensured in these cases.<\/p>\n\n\n\n The consumption profile of the reference case forms the basis for rule-based rescheduling, in which energy-intensive tests are shifted to time intervals with lower emission intensity. The rescheduling takes place on the basis of an emission intensity profile that has been averaged out over hours and weeks, derived from the 15-minute emission intensities of grid power consumption (Fig. 1<\/strong>).<\/p>\n\n\n\n This reflects the realistic forecasting horizon in operational practice, since while sufficiently reliable forecasts for approximately 24-48 hours are available during detailed planning at the beginning of the week, increasing uncertainty must be expected beyond that. Rescheduling based on historical 15-minute values for the entire week, by contrast, would imply an unrealistic level of information availability. The comparison of the reference and optimization cases continues to be based on the time-resolved 15-minute emission intensities. The emission profile of the optimization case is derived from the rescheduled test plan, from which the total emissions are in turn determined.<\/p>\n\n\n\n The splitting of individual test orders results in additional setup operations that affect both energy consumption and test sequence. These are not considered separately in the present analysis, as setup times are pump-specific and the necessary data is not available. Furthermore, energy consumption during setup is low when compared to the actual test process, so its impact on potential savings is expected to be limited. <\/p>\n\n\n\n The rescheduling is also based on weekly averaged emission factors, meaning that short-term emission minima within a calendar week are not fully exploited. The emissions assessment considers only grid electricity consumption; site-specific self-generation (e.g., photovoltaics) is not taken into account. Overall, the reported savings potentials should therefore be understood as a conservative estimate.<\/p>\n\n\n\n The data basis for the case study consists of internal planning and energy consumption data as well as external emissions data. The internal test plan data is available as an Excel list and documents the actual completion of the tests based on the reported test times. In addition, an average energy consumption per pump determined internally by the company is available, which was derived from real measurements on the test bench. <\/p>\n\n\n\n The internal data was provided by the company and used as actual data in this study. Data gaps are filled by documented assumptions, and obvious errors are corrected. The external emissions data for grid electricity from the public grid are based on historical emission intensities from the data source electricityMaps. A carbon intensity signal in gCO2<\/sub>eq\/kWh with a 15-minute temporal resolution from the electricityMaps zone for Germany is used.<\/p>\n\n\n\n <\/strong>The following section first presents the results of the GHG accounting for the reference case and the optimization case. Subsequently, the determined emission savings potential is quantified, and the change in the temporal distribution of emissions between the two cases is illustrated using an example comparison profile.<\/p>\n\n\n\n Total emissions for the reference scenario in 2025 amount to 4,674.49 kg CO2<\/sub>eq. The total electricity consumption of the reference case is approximately 17 MWh. The original test plan includes a total of 344 setup changes. Figure 2 <\/strong>shows the load profile and the temporal distribution of GHG emissions for the reference case over the course of calendar week 36.<\/p>\n\n\n\n Total emissions in the optimization case amount to 4,554.68 kg CO2<\/sub>eq. The total electricity consumption of the optimization case is identical to that of the reference case and amounts to 17 MWh. In the optimized test plan, the number of setup changes increases to 513. Figure 3 <\/strong>shows the load profile and the temporal distribution of GHG emissions for the optimization case over the course of calendar week 36.<\/p>\n\n\n\n The total emission savings potential at the pump test bench is calculated as the difference between the total emissions of the reference and optimization cases and amounts to 119.81 kg CO2<\/sub>eq. This corresponds to a saving of 2.56%. The rescheduling resulted in 169 additional setup changes, corresponding to an increase of 49.13% compared to the reference case.<\/p>\n\n\n\n No deviations were found during the plausibility check of the rescheduled test plan; all test orders were fully scheduled, and the defined restrictions were adhered to. Figure 4 <\/strong>shows a direct comparison of the load profiles for the reference and optimization cases, as well as the emission intensity trend, using calendar week 36 as an example. <\/p>\n\n\n\n This case study demonstrates that emissions-oriented detailed planning based on a digital twin is fundamentally suitable for reducing GHG emissions in operational production planning. The 2.56% reduction in emissions should be regarded as a moderate result, yet one that is plausible given real-world practical constraints. In a simulation-based case study on aviation manufacturing, a CO2<\/sub>eq reduction of up to 7% was achieved [14]. The lower potential in this particular case is primarily due to the existing setup-time-optimized planning, which bundles similar test orders and thus already achieves a certain degree of energy homogenization.<\/p>\n\n\n\n Furthermore, there is a pronounced trade-off of objectives between emissions-oriented and setup-time-optimized planning: The achieved emissions reduction is accompanied by a significant increase in setup operations (+49.13%). This highlights the need for an integrated consideration of both objectives in future planning approaches. At the same time, it becomes clear that greater emission reductions can be expected if additional degrees of freedom are given, for example by extending the planning period beyond individual calendar weeks or through greater integration into upstream planning levels.<\/p>\n\n\n\n In summary, the following key takeaways can be derived for practical application:<\/p>\n\n\n\n The digital twin must be able to link planning information with energy consumption values, integrate external emissions data, and generate and compare planning variants based on rules. This requires robust data integration and scenario comparison so that reference and alternative plans can be compared and evaluated. Equally crucial is the formalized mapping of use-case-specific constraints, as these define the feasible scope of action.<\/p>\n\n\n\n The presented methodology is particularly transferable to testing and manufacturing processes as well as to batch processes in the chemical industry that exhibit varying electrical energy consumption and where organizational flexibility exists. Examples of application areas include end-of-line test benches, machine tools, hardening furnaces, and industrial washing and drying systems, provided there is temporal flexibility in the order sequence. For practical application, a consistent basis of data is essential, as data gaps have a significant impact on the results.<\/p>\n\n\n\n This case study is based on a one-time input dataset of historical data. For long-term operational use, however, automated data processing is in planning. This would mean that planning information, energy consumption values, and emissions data are continuously consolidated.<\/p>\n\n\n\n Furthermore, the current state of research indicates that short-term forecasts of time-varying CO\u2082 emission intensities are feasible and can thus be used as an information basis for emission-oriented detailed planning [15]. The investigations also yield concrete starting points for future research and development tasks. In particular, the methodology should be further developed to explicitly integrate changeover effort as constraints into the rescheduling process. <\/p>\n\n\n\n Additionally, an empirical validation of the ex-ante benefit assessment is required, in which the projected GHG reductions are compared with the emission effects observed in real-world operation. Lastly, for a comprehensive scientific evaluation, an accounting of the GHG emissions caused by the additional digital infrastructure is necessary to systematically compare these with the expected emission reductions.<\/p>\n\n\n\n The research project is funded by the Federal Ministry for Economic Affairs and Energy as part of the \u201cDevelopment of Digital Technologies\u201d funding program under <\/em>grant number<\/em> 01MN23003B.<\/em><\/p>\nThe pump test bench case study<\/h2>\n\n\n\n
Methodology for quantifying emission reduction potential<\/h2>\n\n\n\n
Determination of greenhouse gas emissions in the reference case<\/h3>\n\n\n\n
Definition of optimization rules<\/h3>\n\n\n\n
Key restrictions on rescheduling<\/h3>\n\n\n\n
Rule-based rescheduling <\/h3>\n\n\n\n
![\"Figure](\"https:\/\/industry-science.com\/wp-content\/uploads\/2026\/06\/Bischoff_I4S-26-3_Figure-1-1024x551.webp\")
Limitations and restrictions<\/h3>\n\n\n\n
Data basis<\/h3>\n\n\n\n
Effects of rescheduling on the pump test bench<\/h2>\n\n\n\n
Results of the reference case<\/h3>\n\n\n\n
![\"Figure](\"https:\/\/industry-science.com\/wp-content\/uploads\/2026\/06\/Bischoff_I4S-26-3_Figure-2-1024x500.webp\")
Results of the optimization case<\/h3>\n\n\n\n
![\"Figure](\"https:\/\/industry-science.com\/wp-content\/uploads\/2026\/06\/Bischoff_I4S-26-3_Figure-3-1024x519.webp\")
Quantification of the savings potential<\/h3>\n\n\n\n
![\"Figure](\"https:\/\/industry-science.com\/wp-content\/uploads\/2026\/06\/Bischoff_I4S-26-3_Figure-4-1024x546.webp\")
Key findings for practical application<\/h2>\n\n\n\n
\n
Requirements for the twin<\/h3>\n\n\n\n
Transferability<\/h3>\n\n\n\n
Outlook<\/h2>\n\n\n\n
Bibliography <\/h2>[1] Shukla, P. R.; Skea, J.; Reisinger, A. R.; IPCC (eds.): Climate Change 2022: Mitigation of Climate Change. IPCC. Geneva 2022.\r
[2] Giannetti, B. F.; Agostinho, F.; Eras, J. J. C.; Yang, Z.; Almeida, C. M. V. B.: Cleaner production for achieving the sustainable development goals. In: Journal of Cleaner Production 271 (2020) , p. 122127. DOI: https:\/\/doi.org\/10.1016\/j.jclepro.2020.122127.\r
[3] Kaiser, M.; Senkpiel, C.; Henning, H.-M.; Kost, C.: Direct and indirect electrification in the German industry from a sector-coupled energy system modeling perspective. In: Energy Conversion and Management 28 (2025) , p. 101305. DOI: https:\/\/doi.org\/10.1016\/j.ecmx.2025.101305.\r
[4] Kono, J.; Ostermeyer, Y.; Wallbaum, H.: The trends of hourly carbon emission factors in Germany and investigation on relevant consumption patterns for its application. In: Int J Life Cycle Assess 22 (2017) 10, pp. 1493-1501. DOI: https:\/\/doi.org\/10.1007\/s11367-017-1277-z.\r
[5] Blizniukova, D.; Holzapfel, P.; Unnewehr, J. F.; Bach, V.; Finkbeiner, M.: Increasing temporal resolution in greenhouse gas accounting of electricity consumption divided into Scopes 2 and 3: case study of Germany. In: Int J Life Cycle Assess 28 (2023) 12, pp. 1622-1639. DOI: https:\/\/doi.org\/10.1007\/s11367-023-02240-3.\r
[6] Germscheid, S.; Mitsos, A.; Dahmen, M.: Demand response potential of industrial processes considering uncertain short-term electricity prices. In: AIChE Journal 68 (2022). DOI: https:\/\/doi.org\/10.1002\/aic.17828.\r
[7] Mencaroni, A.; Leyman, P.; Raa, B.; De Vuyst, S.; Claeys, D.: Towards net-zero manufacturing: Carbon-aware scheduling for GHG emissions reduction. In: Journal of Cleaner Production 529 (2025), p. 146787. DOI: https:\/\/doi.org\/10.1016\/j.jclepro.2025.146787.\r
[8] Attaran, S.; Attaran, M.; Celik, B. G.: Digital Twins and Industrial Internet of Things: Uncovering operational intelligence in Industry 4.0. In: Decision Analytics Journal 10 (2024), p. 100398. DOI: https:\/\/doi.org\/10.1016\/j.dajour.2024.100398.\r
[9] Soori, M.; Arezoo, B.; Dastres, R.: Digital twin for smart manufacturing, A review. In: Sustainable Manufacturing and Service Economics 2 (2023), p. 100017. DOI: https:\/\/doi.org\/10.1016\/j.smse.2023.100017.\r
[10] Ouahabi, N.; Chebak, A.; Kamach, O.; Laayati, O.; Zegrari, M.: Leveraging digital twin into dynamic production scheduling: A review. In: Robotics and Computer-Integrated Manufacturing 89 (2024), p. 102778. DOI: https:\/\/doi.org\/10.1016\/j.rcim.2024.102778.\r
[11] Balduf, F.; M\u00fcller-Zhang, Z.; Kuhn, T.: Praktikable Ermittlung des produktbezogenen CO2-Fu\u00dfabdrucks mithilfe gr\u00fcner Digitaler Energiezwillinge. 2024. DOI: https:\/\/doi.org\/10.25673\/116047.\r
[12] Hildebrandt, G.; Dittler, D.; Habiger, P.; Drath, R.; Weyrich, M.: Data Integration for Digital Twins in Industrial Automation: A Systematic Literature Review. In: IEEE Access PP (2024) , p. 1. DOI: https:\/\/doi.org\/10.1109\/ACCESS.2024.3465632.\r
[13] Kuehner, K. J.; Scheer, R.; Strassburger, S.: Digital Twin: Finding Common Ground \u2013 A Meta-Review. In: Procedia CIRP 104 (2021), pp. 1227-1232. DOI: https:\/\/doi.org\/10.1016\/j.procir.2021.11.206.\r
[14] Haddad, Y.; De Bonneval, E. G.; Afy-Shararah, M.; Carter, J.; Artingstall, J.; Salonitis, K.: Energy flexibility in aerospace manufacturing: The case of low carbon intensity production. In: Journal of Manufacturing Systems 74 (2024), pp. 812-825 . DOI: https:\/\/doi.org\/10.1016\/j.jmsy.2024.05.004.\r
[15] Ostermann, A.; Bajrami, A.; Bogensperger, A.: Short-term forecasting of German generation-based CO2 emission factors using parametric and non-parametric time series models. In: Energy Informatics 7 (2024). DOI: https:\/\/doi.org\/10.1186\/s42162-024-00303-9.<\/div>