Real-time methods for monitoring and simulating a supply chain for a green fuel

Abstract

The invention relates to a real-time method for monitoring and simulating a supply chain for a green fuel, comprising the following components: - in a data storage system, maintaining historical data on an infrastructure for producing a green fuel, where the infrastructure is complex due to a large number of participants; - via a variety of interfaces, receiving real-time data regarding the infrastructure; and - in a computer in a simulation environment, based on the historical data held and the real-time data obtained, predicting at least an optimal utilization of the infrastructure to form an efficient supply chain and / or to efficiently utilize a supply chain, whereby specific specifications are issued in real time for at least several of the participants of the complex infrastructure. The real-time method proposed here for simulating a supply chain makes it possible to produce green fuel cost-effectively and according to demand.

Description

[0001] The invention relates to a real-time method for monitoring and simulating a supply chain for a green fuel.

[0002] One of humanity's most pressing current problems is energy production and conversion with reduced emissions of greenhouse gases, especially carbon dioxide. A promising approach is to produce or synthesize so-called green fuels (also known as eFuel or other hydrogen derivatives) using renewable energy sources and preferably from climate-neutral raw materials. For example, the following approaches are currently being pursued in projects: Power-to-Liquid (PtL) refers to a process in which electrical energy from renewable sources (such as wind turbines or wind farms) is used to produce a liquid fuel. These liquid fuels can be used in existing combustion engines. In one project, exhaust gases from a steel plant are converted into liquid fuels by utilizing the waste heat generated. In Iceland, methanol is produced using geothermal energy.

[0003] The Fischer-Tropsch process synthesizes a high-quality diesel fuel, which can be produced from a variety of carbon-containing raw materials, such as biomass, methanol (produced via power-to-liquid), or carbon dioxide from the air (using carbon capture). Several projects are working on the production of diesel using the Fischer-Tropsch process. This involves, among other things, converting solid coal into a liquid fuel, for example, using renewable energy.

[0004] Methanol is a versatile fuel and chemical feedstock for producing high-quality fuels, and it is easier to store and transport than gaseous fuels. Projects exist for converting carbon dioxide from the air into methanol, as well as for producing methanol from coal.

[0005] Dimethyl ether [DME] can be used as a substitute for diesel or as an alternative to liquefied petroleum gas (LPG). This product burns almost soot-free and with low nitrogen oxide [NOx] emissions. Among other projects, there are initiatives to produce DME from forestry waste for use in heavy-duty trucks. Research is also underway on the production of DME from natural gas sources.

[0006] Producing hydrogen [H2] from seawater seems like an obvious solution, but it requires a significant amount of electrical energy. Particularly windy locations worldwide are often sparsely populated and therefore ideal for the construction of wind farms. However, these remote locations make directly supplying the generated electricity uneconomical. Given the large distances between hydrogen production and the end user, it can be advantageous not to liquefy the hydrogen by compression, but rather to convert it into methanol using carbon dioxide or (via the Fischer-Tropsch process) into diesel fuel. The carbon dioxide required for this can, for example, be extracted from the air.

[0007] The production and supply of green fuels requires the collaboration of numerous highly technological and / or highly specialized actors. A smooth process from the source to the consumer of the green fuel is necessary to make it affordable enough to encourage widespread use of green fuels instead of conventional ones.

[0008] Based on this, the present invention aims to overcome, at least partially, the disadvantages known from the prior art. The features of the invention are defined in the independent claims, for which advantageous embodiments are shown in the dependent claims. The features of the claims can be combined in any technically meaningful way, whereby the explanations in the following description and features from the figures, which comprise supplementary embodiments of the invention, can also be used.

[0009] The invention relates to a real-time method for monitoring and simulating a supply chain for a green fuel, comprising the following components: - in a data storage system, maintaining historical data on an infrastructure for producing a green fuel, where the infrastructure is complex due to a large number of participants; - via a variety of interfaces, receiving real-time data regarding the infrastructure; and - in a computer in a simulation environment based on the historical data held and the real-time data obtained, predicting at least an optimal utilization of the infrastructure to form an efficient supply chain and / or to efficiently utilize a supply chain, whereby specific specifications are issued in real time for at least several of the participants of the complex infrastructure.

[0010] Unless explicitly stated otherwise, ordinal numbers used in the preceding and following descriptions serve solely for unambiguous differentiation and do not indicate any order or ranking of the components referred to. An ordinal number greater than one does not necessarily imply the presence of another such component.

[0011] The proposed monitoring (preferably event-based monitoring) and simulation method is implemented as a digital twin, which is a software-based model representing reality, for example, on a central server. In a preferred embodiment, the measured values ​​used in the (modeled) real-world application or the data used in its control or monitoring software (e.g., already processed data) can be used as input data without requiring any conversion or processing. This input data enables the model to achieve the same results as in the real-world application. This allows us to simulate scenarios more quickly and react promptly to resulting changes based on the system's reasoned recommendations.In many areas, however, a change is not highly time-critical, or an event can be predicted with sufficient accuracy in advance, such as a change in the weather or the arrival of a freighter at a designated port. In many cases, alternatively or additionally, a reaction time is not critical, for example, if there is a delivery delay due to traffic congestion or if a storage facility reaches, or is about to reach, a critical (e.g., upper) fill level.

[0012] At the same time, the overall situation of such a supply chain, for example, from the source (such as seawater and wind energy for the production of green hydrogen) through logistics, potential processing (such as the production of a liquid eFuel), to the points of consumption (such as filling stations), is so complex that it becomes time-critical and requires real-time processing. Such real-time processing takes place on a minute-to-second basis. It should be noted that the data situation is so complex because the participants work directly with real-time data that this is not feasible for humans (at least not with a reasonable amount of effort).

[0013] Historical data is at least advantageous, and usually necessary, to draw reliable conclusions from (current) real-time data. In one embodiment, the historical data is identical to the (current) real-time data, with newly acquired real-time data preferably becoming historical data, for example, supplementing or replacing it. Preferably, historical data is already available before the first application, for example, as training data and / or knowledge data, in order to meaningfully generate corresponding recommendations for action.

[0014] Data storage is a single physical storage device or a composite of multiple storage devices, which are not necessarily configured to communicate with each other. For example, the data storage device, or at least one of the multiple storage devices, may be located on a central server. A computer processor is able to access the data stored in the data storage device (directly and / or indirectly, and in processed and / or unprocessed form).

[0015] In a simple embodiment, the number of interfaces corresponds to the number of participants in the supply chain. Preferably, the number of interfaces is determined by the measuring points of each participant, thus allowing their measurement data to be directly received as real-time data. Alternatively or additionally, the real-time data is processed before being transmitted via the corresponding interface and / or output data is explicitly generated for the proposed method (for example, to protect know-how and / or personal rights). Preferably, a participant is free to choose the format in which the respective real-time data is provided and / or via which and how many (their own) interfaces it is made available for the real-time process to be executed.For example, real-time data (direct and unprocessed) at an interface could include a temperature measurement at a processing unit, such as in an electrolyzer during water splitting to produce molecular hydrogen from, for example, seawater. For example, real-time data (indirect and processed) at an interface could include a status output from an energy storage system (such as a salt storage system for stored thermal energy), such as the currently available or usable electrical energy and / or free storage capacity.

[0016] The simulation environment, as described above, is a model of reality, maintained as a digital twin. In one implementation, insufficient real-time data is supplemented by fictitious scenarios, or appropriate responses are made to scenarios that have already occurred. For example, real-time data can be supplemented by observations such as monitoring cargo ships at sea or in port if information from the shipping company is insufficient. Similarly, weather data based on historical observations can be used to derive the expected maximum wind yield (i.e., the electricity generation capacity) for a wind farm, provided that the wind farm operator's information is limited to current generation (i.e., available energy quantity) and no forecast is provided.

[0017] The model generates clear instructions for action, specifically detailed guidelines that allow for adjustments to the supply chain. For example, if wind power is expected to cease soon at a wind farm, the system will switch to geothermal sources for molecular hydrogen production to continue meeting current demand (cost-effectively and / or in a climate-neutral manner). Conversely, if demand is low, available storage capacities (including ship cargoes, for example) will not be fully utilized, meaning that available energy may not be used or may be used differently. This is implemented, for example, based on market prices, via interfaces to the relevant commodity exchanges (which are monitored) and / or other participants, such as traders.

[0018] In an advantageous embodiment, the output specifications are so detailed that they are returned to the respective participant as (direct or indirect) control variables via a corresponding interface (for example, the same one from which the real-time data originates). These input variables resulting from the specifications then correspond, for example, to a (precisely defined) order, such as a hydrogen producer specifying a desired quantity of hydrogen.

[0019] In a further advantageous embodiment of the real-time method, it is proposed that a plurality of sets of dedicated specifications be output.

[0020] A set of specifications is compiled, for example, for each participant or, alternatively, for a project manager. A set always contains all specifications, including those that do not currently require any changes. This serves to ensure and / or allow for human verification that the specifications are consistent, and / or to document the processes. For example, if specifications are communicated electronically via the interfaces, an error message is generated if a set is incomplete. In one implementation, human intervention is (mandatory) only required in the event of such an error message (or one caused by another).

[0021] In a further advantageous embodiment of the real-time method, it is proposed that the complex infrastructure includes participants that act alternatively to one another.

[0022] In one embodiment, as already indicated above, several participants in a supply chain or in parallel supply chains of a (total) infrastructure are in competition with each other, even though they have different tasks and produce different products. In energy production, these could be, for example, wind farms, geothermal sources, and solar power plants. In fuel production, these could be, for example, the hydrogen producer and the methanol producer, even though the methanol producer is often downstream of the hydrogen producer in a supply chain.

[0023] In a further advantageous embodiment of the real-time method, it is proposed that alternative green fuels can be produced from a common source and / or a common raw material using the infrastructure.

[0024] Alternative fuels are preferably produced based on a predefined and / or dynamically adapted set of rules. Decisions are derived from these rules and supported by real-time data acquisition.

[0025] This infrastructure makes it possible to produce green fuels from hydrogen in the form of liquefied hydrogen (LNG) or, alternatively, by adding carbon dioxide, methanol (liquid under normal conditions). Methanol can then be used as a feedstock for the production of diesel or gasoline. A similar approach applies to the energy sources, such as wind, geothermal, or solar power. In many cases, the raw material and the energy source are linked via the respective plant or its location. For example, wind energy is used in Scotland, geothermal sources in Iceland, waste heat from a steel mill in Germany, and solar thermal or photovoltaic energy in Morocco for water splitting and / or methanol production.However, the location then has a strong influence on the downstream logistics for transporting the green fuel to the point of demand of the customers or processors.

[0026] In an advantageous embodiment of the real-time method, it is further proposed that at least one of the following external influences and / or associated forecasts be taken into account in real time: - Weather; - Prices for producible green fuels, raw materials and / or services; - Demand; - Processing, storage and / or logistics capacities; - unplanned cost changes; - geopolitical influences; and - Deviations from upstream and / or downstream processes of participants in a potentially efficient supply chain.

[0027] In one embodiment of the present invention, the real-time method for simulating a supply chain for a green fuel takes into account several external influences and / or associated forecasts, which are captured by means of a multitude of sensors, IoT devices [Internet of Things], databases and / or interfaces to information services and processed in a computer.

[0028] External influences considered include, for example, the weather, which is monitored in real time (immediately) using weather sensors and satellites or an external service provider, and / or for which forecasts are obtained from meteorological data models. In one embodiment, the prices of producible green fuels, raw materials, and / or services are recorded in real time, with this information potentially originating from commodity exchanges or price monitoring services. The demand for green fuels and the capacities for processing, storage, and / or logistics are factors considered in another embodiment and are taken into account in the computer's simulation environment to ensure efficient utilization of the existing infrastructure.

[0029] In one embodiment, unplanned cost changes (such as additional costs or discounts) that may arise from maintenance work, breakdowns, wage increases, or similar factors are included in the simulation. Geopolitical influences, preferably provided by news agencies and political analysis institutes, are also preferably incorporated into the simulation to identify potential impacts on the supply chain early on and to be able to react accordingly. For example, an impending change of government, based on political affiliation and / or specific campaign promises, could lead to a prediction that changes regarding customs duties, taxes, and / or legal certainty in general are to be expected.

[0030] In one embodiment, deviations in upstream and / or downstream processes are taken into account, for example, because these processes run slower or faster than expected. These deviations can occur, for example, due to strikes, outages, or other unexpected events and have a direct impact on the efficiency of the supply chain.

[0031] By taking such factors into account, it is possible to optimize the supply chain for green fuels at all times and thus ensure the availability of these essential energy sources.

[0032] In a further advantageous embodiment of the real-time method, it is proposed that artificial intelligence be used to identify possible decisions.

[0033] The uncertainties that arise during the transport of green fuels, especially hydrogen, can be mitigated and minimized through artificial intelligence (AI). For example, containers can be monitored to determine if they could not be loaded onto the ship as planned. In this case, the system can automatically suggest solutions, such as alternative shipping routes, so that the delivery obligation to the customer can still be fulfilled.

[0034] AI-supported analysis of historical and real-time data on the status of all parts of the digital twin enables the identification of anomalies. This allows, for example, the derivation and optimization of maintenance requirements to prevent production downtime and increase the efficiency of the entire production process.

[0035] By taking into account data on future environmental conditions, such as weather forecasts, AI can be used to model various scenarios that provide information about the expected energy supply. Through a continuous learning process of the algorithms, the scenarios become increasingly precise over time, thus enabling accurate forecasts.

[0036] Approaches using so-called neural networks have proven particularly suitable and easy to handle. For large datasets, an approach using a so-called generative pre-trained transformer is especially promising.

[0037] A generative pre-trained transformer (GPT) is an advanced algorithm based on the principles of machine learning. A GPT is capable of processing and analyzing large amounts of data (including so-called Big Data) to independently derive predictions and decisions. Ideally, such a GPT is trained to continuously improve its predictive accuracy by processing new input data. It is also preferably designed to consider so-called invisible or hidden constraints, such as correlations or events typically not recognized or considered by humans.

[0038] In one embodiment, a pre-trained generative transformer is used as the artificial intelligence component. This computer component utilizes pre-trained algorithms to identify specific requirements in real time. This transformer is responsible for analyzing and predicting optimal infrastructure utilization by generating forecasts based on stored historical data and real-time data obtained through a simulation environment. The use of this transformer enables the efficient utilization and design of a supply chain, thus ensuring adaptation to constantly changing conditions and requirements.

[0039] Preferably, the transformer is implemented in a self-learning and / or self-reinforcing manner, so that with each analysis it improves its accuracy in finding effective specifications for the participants in the complex infrastructure. It should be noted that (self-)learning represents a permanent modification of the algorithm for future processing. Self-reinforcing can be performed within a task solution (i.e., approximation or optimization) and does not necessarily lead to a permanent change in the algorithm. However, self-reinforcement is preferred as a suitable implementation of self-learning.

[0040] It should be noted that the algorithm uses exclusively confidential data sources and / or treats the results confidentially. It should also be noted that the simulation environment is a human-created and comprehensible model or (possibly additionally) a machine-learned model, whereby the internal connections within the model are not comprehensible to a human (or at least not without further processing). In one embodiment, the transformer used is the model itself or a user of the model. In another embodiment, it is a mixture of both and / or a plurality of independent transformers.

[0041] In a further advantageous embodiment of the real-time method, it is proposed that the computer automatically outputs at least some of the specific requirements and / or reasons for identifying these requirements in the form of reports that are readable by both humans and computers.

[0042] For quality assurance and information exchange within a network of participants, the creation of reports is essential. These reports are generated automatically but designed for human readability, for example, as a computer file or on paper. In one implementation, this results in a (potentially tiered) certification indicating the climate impact of the respective green fuel. For example, a non-climate-neutral energy source may be used temporarily (e.g., as a supplement) to meet demand and / or maintain a stable process (e.g., to maintain a constant temperature and / or pressure). In such cases, a fuel batch might be assigned a lower quality rating, for example, designated as 80% climate-neutral, which could then affect the available purchase price or any applicable climate levies.An important aspect here is also to identify future improvement potential (for example, regarding ship propulsion and energy backup sources) and to promote innovation in these areas. Specifically, achieving a particularly high level of accuracy is a preferred boundary condition of the real-time method.

[0043] The invention described above is explained in detail below against the relevant technical background with reference to the accompanying drawings, which show preferred embodiments. The invention is in no way limited by the purely schematic drawings, although it should be noted that the drawings are not dimensionally accurate and are not suitable for defining size relationships. It is illustrated in Fig. 1: schematically an infrastructure and its integration into a control system using a digital twin; and Fig. 2: a decision tree for determining a supply chain within an infrastructure.

[0044] In Fig. Figure 1 schematically depicts an infrastructure 5 and its integration into a control system using a digital twin. On the left side of the diagram, three levels are shown: at the top, a network of participants 6 in a supply chain 1, where, for example, a symbol for a factory 13 represents the source 9; furthermore, a network for logistics 11 (here, pars-pro-toto, the symbols for a freighter 14, a truck 15, a container 16, and a transshipment port 17 are shown); and a filling station 18 (as the end customer 19 for green fuel 2,3) is shown. The middle level shows the currently active participants 6, and at the bottom, the participant 6 (optionally represented here as a truck 15) is shown receiving a directive.

[0045] In the lower right is a computer 7 with a data storage device 4, which contains the simulation environment 8 corresponding to the infrastructure 5 shown on the left, as well as historical data, and a screen (for example, for reporting 12). The computer 7 (for example, a central server) generates specifications for the operation of the infrastructure 5. In the upper right is a satellite, symbolizing data communication 20 of the real-time data transmitted via the interfaces of the individual participants 6 to the computer 7, potentially over long distances.

[0046] Tasks to be accomplished include, for example: - managing digital networks, including, for example, risk management through scenario analysis and / or network control and monitoring, - monitoring and controlling hydrogen production, - controlling transport or logistics 11, including, for example, delivery management for transport to customers and material transport between process steps, - monitoring and optimizing storage capacity or warehouse planning, and - managing cargo handling, including, for example, monitoring container availability and status, and forecasting the number of containers needed.

[0047] In Fig. Figure 2 shows a decision tree for determining a supply chain 1 within an infrastructure 5. At the top of the diagram, a central server is shown as the processing computer 7, with a symbolic representation of an input and output of a multitude of data via a multitude of interfaces in data communication 20 with a real infrastructure 5 (here, top left, with reference to...). Fig. 1 shown).

[0048] Below is a sequence of stations or participants 6, which compete with each other in the parallel diagram or represent alternative branches. The process begins on the left with source 9 for energy and raw materials 10 for a green fuel 2,3. The energy source is represented at the top by a symbol for a wind turbine 21, where, for example, the currently generated or available electrical energy and the usability of wind energy are monitored. In the middle is a symbol for available carbon dioxide 22, where, for example, quantity and quality are monitored, and to the right is a carbon dioxide storage facility 23, where storage capacity and access to the storage facility are monitored. At the bottom is a symbol for water as a raw material source for green hydrogen 2, where, for example, quantity and quality are monitored.

[0049] Subsequently, a factory plant 13 for producing green hydrogen 2 is shown, in which, for example, the production capacity, the production quantity and production quality, the energy source and, if applicable, external influences are monitored.

[0050] The following is a symbol for a hydrogen storage facility, where, for example, the storage capacity and the required containers 16 are monitored.

[0051] The process then branches out into alternative green fuels 2,3. In the upper row, liquefied hydrogen 2 is produced and transported as a green fuel. The first symbol represents the liquefaction process, where, for example, production capacity, production quantity and quality, storage capacity, the required containers 16, and the energy source used are monitored. The second symbol in this upper section of the process shows a truck 15 for transporting the liquefied (green) hydrogen 2, where, for example, real-time geolocation monitoring is carried out, geopolitical and / or environmental conditions are monitored, and various scenarios of this often complex logistics 11 are simulated. Following this, a symbol for the regasification 24 of the hydrogen 2 is shown, where, for example, the quality and the input quantity are monitored.Finally, in the upper process section, the distribution to the end customers 19 (not shown here) is shown with the symbol of a truck 15, whereby, for example, real-time geolocation monitoring and fill level monitoring are carried out, and various scenarios of this mostly complex distribution logistics are played out.

[0052] In the lower row, methanol 3 is produced and transported as a green fuel. The first symbol represents a tank canister for producing methanol 3 by adding carbon dioxide 22, monitoring, for example, production capacity, production quantity and quality, storage capacity, required containers 16, and the energy source used. The second symbol in this lower section shows a truck 15 for transporting the (green) methanol 3, monitoring, for example, real-time geolocation, geopolitical and / or environmental conditions, and various scenarios of this often complex logistics 11. Following this, a symbol for a transshipment port 17 is shown, monitoring, for example, port availability and storage capacity.Finally, in the upper process section, the distribution to the end customers 19 (not shown here) is shown with the symbol of a truck 15, whereby, for example, real-time geolocation monitoring and fill level monitoring are carried out, and various scenarios of this mostly complex distribution logistics are played out.

[0053] The IT infrastructure required for communicating data includes, firstly, the diverse interfaces along the supply chain 1, which are consolidated on a central server, and secondly, the fact that all relevant data will be displayed in a user-friendly manner, with the possibility of accessing additional detailed views, such as the integrated digital partial twins.

[0054] To optimize the process, simulations are performed using historical and real-time data to analyze various scenarios and identify the optimal course of action. For example, artificial intelligence algorithms (preferably a neural network) are used to independently derive specific and ideal control strategies for subsequent measures.

[0055] Below are two application examples with specific (simplified and therefore comprehensible) scenarios or framework conditions: Example 1:

[0056] Wind capacity (e.g., strength and duration) is expected to decrease next week, meaning that production targets will not be met. However, current wind capacity this week is so high that more hydrogen 2 could be produced than is needed to meet current demand.

[0057] Relevant parameters that should therefore be given special attention [KPI, English: Key Point of Interest] include, for example: - Wind availability, - Power source, - Production capacity, - Storage capacity, - Customer demand, and - Risk assessment

[0058] Strategies to achieve these parameters include, for example: - All relevant parameters and real-time data are used to centrally develop scenarios, - In order to achieve the set production targets, we will use the available wind capacity this week to replenish the storage facilities, and - More hydrogen 2 will be produced next week to compensate for the reduced production capacity. Example 2:

[0059] A delivery to end customers 19 is delayed due to problems with a participant 6 in the logistics area 11. As a result, end customers 19 receive their ordered quantity of green fuel 2.3 later than requested, and the containers 16 required for storage do not arrive at the production site on time.

[0060] Relevant parameters that should therefore be given special attention [KPI, English: Key Point of Interest] include, for example: - geopolitical and ecological (weather) conditions, - End-user demand 19, - Inventory data, - Real-time container tracking, - Number and / or type of containers required: 16, - Production capacity, - Distribution scenarios, and - Risk assessment

[0061] Strategies to achieve these parameters include, for example: - Centralized development and analysis of supply chain scenarios, - An adjustment of the production quantity is possible because the customers have limited storage capacities, and - Container 16 must be procured by other means for storage in order to continue production.

[0062] The real-time method proposed here for simulating a supply chain makes it possible to produce green fuel cost-effectively and according to demand.

Claims

[1] Real-time method for monitoring and simulating a supply chain (1) for a green fuel (2,3) comprising the following components: - in a data storage (4), holding historical data for an infrastructure (5) for producing a green fuel (2,3), the infrastructure (5) being complex due to a large number of participants (6); - via a variety of interfaces, obtaining real-time data concerning the infrastructure (5); and - in a computer (7) in a simulation environment (8) based on the historical data held and the real-time data obtained, predicting at least one optimal utilization of the infrastructure (5) to form an efficient supply chain (1) and / or to efficiently utilize a supply chain (1), whereby specific requirements are issued in real time for at least several of the participants (6) of the complex infrastructure (5). [2] Real-time method according to claim 1, wherein a plurality of sets of dedicated specifications are output. [3] Real-time method according to claim 1 or claim 2, wherein the complex infrastructure (5) comprises participants (6) that act alternatively to each other. [4] Real-time method according to claim 3, wherein alternative green fuels (2,3) can be produced from a common source (9) and / or a common raw material (10) using the infrastructure (5). [5] Real-time method according to any of the preceding claims, wherein at least one of the following external influences and / or associated forecasts are taken into account in real time: - Weather; - Prices for producible green fuels (2,3), raw materials (10) and / or services; - Demand; - Processing, storage and / or logistics capacities (11); - unplanned cost changes; - geopolitical influences; and - Deviations from upstream and / or downstream processes of participants (6) in a potentially efficient supply chain (1). [6] Real-time method according to one of the preceding claims, wherein artificial intelligence is used to identify possible decisions. [7] Real-time method according to one of the preceding claims, wherein the computer (7) automatically outputs at least part of the specific requirements and / or reasons for identifying these requirements in the form of reports (12) which are readable by both humans and computers (7).

Images