Computer-implemented method, computer program and computer for generating digital twins of chemical process or plant or system of chemical plant

Abstract

The invention relates to a computer-implemented method for generating a digital twin (20) for a chemical process or plant or system of a chemical plant, in particular for substance synthesis and / or substance separation, comprising the following steps: a) inputting a plurality of boundary conditions (1) into a computational model (2), the computational model (2) comprising equations, wherein each boundary condition (1) contains a plurality of parameter values (4), b) inputting an initial condition for solving the system of equations, c) in a simulation step, solving the system of equations using each of the boundary conditions (1) in order to provide a corresponding simulation result, an individual one of the parameter values (4) of the individual boundary conditions (1) changes from one simulation step to the next while the other parameter values (4) belonging to the individual boundary conditions remain unchanged, and d) providing a digital twin (20) as a set of data points obtained from the simulation steps. The invention also relates to a computer program and a computer (55) for generating a digital twinning for a chemical process or plant or system for a chemical plant, in particular for substance synthesis and / or substance separation.

Description

Technical Field

[0001] The present invention relates to a computer-implemented method, a computer program and a computer for generating a digital twin of a chemical process or an apparatus or a system for a chemical plant. Background Art

[0002] It is generally known that gas mixtures can be separated by gas separation membranes due to the different permeabilities of the individual gases. To produce such gas separation membranes, polymers are processed into hollow fibers or flat sheet membranes. The membranes are characterized by a very thin separation layer, so that the permeability of the membrane is as high as possible.

[0003] In addition to new membrane materials, various ways of connecting membranes have been investigated in the prior art.Many single-stage or multistage membrane compounds for gas separation are known in the literature.

[0004] Exemplary literature sources include: Baker, Ind Eng Chem Res, Natural Gas Processing with Membranes, 47 (2008); Bhide Mem Sci, Hybrid processes for the removal of acid gases from natural gas, 1998. The disadvantage of this particular process is that they partly include multiple recompression steps, or only high-purity permeate gas or only high-purity retentate gas is available.

[0005] WO 2012 / 00727; WO 2013 / 098024; WO 2014 / 075850; KR 10-1327337; KR 10-1327338; US 6,565,626 B1; US ​​6,168,649 B1; JP 2009-242773A; WO 2014 / 183977; EP 0799 634 each disclose a membrane separation process with three membrane separation stages, wherein the retentate stream from stage 3 and the permeate stream from stage 2 are recycled into the crude gas stream. WO 2012 / 00727; WO 2013 / 098024 and WO 2014 / 075850 represent the most optimized of all these processes. In the patents, an apparatus and a process are described which are optimized with regard to product purity combined with minimum energy consumption. In other words, these processes provide two high-purity product streams in an energy-optimized manner.

[0006] However, a new problem has recently emerged that is not fully solved by the devices and processes of the prior art. The problem is that when processing different natural or artificial gas sources (such as fermenters), optimization is usually required. For example, changes or fluctuations in various raw gas compositions or impurities, etc., lead to optimization requirements. In addition, specific requirements for the location of the plant lead to the need for optimization. For example, in the case of an offshore drilling rig, it is necessary to minimize the space and weight of the equipment without negatively affecting the performance and / or capacity of the plant. Therefore, there is a multifaceted demand for solutions to more effectively simulate the devices and processes for gas separation, such as for considering the different requirements of existing plants and facilities, different optimization targets, different materials and / or different architectures, properties or capacities. No appropriate solution to this problem has been found in the prior art. In addition, in order to make these processes commercially viable, these processes are usually carried out on a large scale; therefore, it is usually impractical to run a test process to determine the optimal conditions. In addition, it would be impractical to physically test a wide range of situations, and there would be a risk of damaging the equipment.

[0007] The problem is solved by a computer-implemented method for generating a digital twin of a chemical process or a device or a system for a chemical plant, a computer program for generating a digital twin of a chemical process or a device or a system for a chemical plant and a computer according to the independent claims. Advantageous embodiments of the computer-implemented method for generating a digital twin of a chemical process or a device or a system for a chemical plant are given in the dependent claims. If the embodiments of the invention are not mutually exclusive, they can be freely combined with each other. Summary of the invention

[0008] A first aspect of the invention relates to a computer-implemented method for generating a digital twin of a chemical process or a device or a system for a chemical plant, in particular for substance synthesis and / or substance separation.

[0009] A digital twin may be a digital representation of a system or device. In some cases, a digital twin may include one or more of the following: at least partially identical and / or proportional geometric properties of an actual system or device, material properties, initial conditions, and boundary conditions that are at least partially identical to those of an actual system or device. The geometric properties may include geometric dimensions, which may also be scaled or modified at least in part using methods known to those skilled in the art. The material properties may include the density of the material, the conductivity of the material, the porosity of the material, and other material properties known to those skilled in the art. A digital twin may be generated by simulation software, in which a physical transport equation or a numerical model is solved for a predefined set of predefined boundary and operating condition values ​​of a corresponding actual system or device to provide the obtained simulation parameter values. For a predefined set of similar predefined boundaries and operating conditions, the obtained parameter values ​​may be verified with the corresponding actual parameter values ​​obtained from experiments on the corresponding actual system or device. A digital twin may include all data values ​​obtained as a result of multiple simulation runs, or a subset thereof, such as a subset of confirmed parameter values ​​or a subset of parameter values ​​that meet other criteria.

[0010] Boundary conditions define the inputs to a simulation model. Some boundary conditions, such as velocity and volume flow rate, determine how fluids enter or leave the model. Other constraints, such as heat flow, determine the energy exchange between the model and its environment. Boundary conditions connect a simulation model to its environment. Most boundary conditions can be defined as either steady state or transient. Steady state boundary conditions persist throughout the simulation. Transient boundary conditions change over time and are often used to simulate events or cyclic phenomena.

[0011] When the geometry of the digital twin is at least partially changed or scaled, the boundary conditions can be adjusted so that the effect of the adjusted boundary conditions and the at least partially changed or scaled geometry of the digital twin is the same as that of the actual system or device. Such adaptation of boundary conditions is known to those skilled in the art.

[0012] Unlike boundary conditions, initial conditions are only implemented at the beginning of the analysis. Initial conditions define the initial values ​​of each solution field. Therefore, they can play an important role in the stability and computational time of steady-state simulations. To ensure a good convergence rate for steady-state simulations, it can be a good practice to initialize the domains to be close to the expected solution. For example, if we are studying the cooling effect of a heat exchanger, the time required to reach convergence can make a big difference whether we initialize the surfaces of the heat exchanger to 350K or 800K. In the case of transient analysis, initial conditions can be critical to the setup. They define the state of the system at time zero and play an important role in the simulation.

[0013] One embodiment of the computer-implemented method comprises a first step of inputting a plurality of boundary conditions into a computational model, the computational model comprising a system of equations, wherein each boundary condition respectively comprises a plurality of parameter values.

[0014] The system of equations can be a system of partial differential equations, such as a system of transport equations, and / or algebraic equations, such as thermodynamic equations of state, which are related to the chemical process or equipment or system of a chemical plant, particularly for material synthesis and / or material separation. The boundary conditions comprising multiple parameter values ​​can be assigned by a set of discrete time / attribute pairs (such as tables). In addition, boundary conditions can be assigned as a continuous set of time / attribute pairs, such as curves or mathematical functions. Parameter values ​​can include pressure, temperature, mass flow, molar flow, volume flow, feed composition, purity of material in a specific stream, the ratio of membrane capacity to another stage in a stage, the ratio of retentate pressure to permeate pressure in a specific stage, the quotient of the pressure ratio of a stage to another stage, the selectivity of the membrane, permeability, permeability and / or area, and / or any other characterizing properties or parameters. This can make it possible to take into account process parameters which are particularly likely to change during the lifetime of a chemical process facility, or to adjust process parameters in the design of chemical or other industrial plants (e.g. gas separation facilities) in such a way that they provide good results with regard to the optimization target, or provide good results despite restrictive, difficult framework conditions, etc.

[0015] The input of multiple boundary conditions can be performed by selecting a set of parameter values ​​corresponding to each boundary condition from a database or another external source or multiple external sources such as a cloud service / system. In addition, multiple boundary conditions can be manually input through a data input device. Multiple boundary conditions can also be input through a user-defined function, which can be in the form of a computer code or a computer program. Multiple boundary conditions can be selected and then provided to the computational model. The boundary conditions including the system of equations can be solved in software, such as Aspen Custom Modeler (ACM), however, other software such as Aspen Plus, Aspen Hysis, ProMax, MATLAB, MathCad can also be used.

[0016] According to some embodiments, the system of equations is solved by software, which may include a numerical solver, i.e., a solver for numerical equations. A numerical solver may use one or more numerical approximations to find an approximate solution to a problem. The incremental optimization and corresponding simulation results obtained in a given simulation step may be maintained, such as stored in a main memory and / or a database, and may be reused in the next simulation run. This may have the advantage of improving performance because the simulation converges to the (local or global) optimization optimum more quickly, and may reduce the number of simulations (also referred to herein as "simulation steps"), as well as the associated time and computing resources required to load input data and / or update the solver.

[0017] The software may contain a numerical solver or solvers that can solve the system of equations in a coupled manner, ie, solving the equations together, or in a decoupled manner, ie, solving the equations one after the other.

[0018] The second step of the embodiment of the computer-implemented method includes inputting initial conditions for solving the system of equations.

[0019] The third step of the embodiment of the method implemented by the computer is included in the simulation step and each boundary condition is used to solve the system of equations to provide corresponding simulation results. According to the present invention, for a given chemical process, equipment or system, a system of equations for steady-state process simulation is set up, which is known per se. In addition, a simulation space that can contain a large number of predetermined boundary conditions is defined. For each of these given boundary conditions, the simulation results will be determined. This will be completed with the shortest possible computing time.

[0020] A solution to the potential problem is provided by a computer-implemented method of the present invention, according to which a single one of the parameter values ​​of each boundary condition is changed from one simulation step to the next, while the other parameter values ​​belonging to each boundary condition remain unchanged. This means that only one of the parameter values ​​is changed from one simulation step to the next, while the other parameter values ​​belonging to the boundary condition remain unchanged. For example, the only one parameter may not be changed arbitrarily in the data space, but the closest data point or value of the parameter is selected, particularly in the boundary condition set. This approach to the boundary condition will cause the simulation to converge faster. Therefore, the speed of the entire simulation will be improved.

[0021] In other words, if simulation is used to solve an equation containing three parameters, namely, a first parameter X, a second parameter Y, and a third parameter Z. The first parameter value set contains a first plurality of data points, such as X 1 To X n data points, the second parameter value set contains the second plurality of data points, such as Y 1 To Y mdata points, and the third parameter value set includes a third plurality of data points, such as Z 1 to Z o Then, according to the computer-implemented method of the present invention, the first parameter X may be varied from one simulation step to the next through its value set 1 to n, while the values ​​of the other parameters belonging to the respective boundary conditions are varied in Y 1 and Z 1 That is, for the first 1 to n simulation steps, the values ​​of the first parameter set are varied, covering all n data points. However, the values ​​of the second and third parameters remain unchanged, for example, the second parameter can have the same 1 To Y m The data point corresponds to the invariant value of the first data point of the corresponding number m of data points, and the third parameter may have a corresponding number o of data points Z 1 to Z o The unchanged value of the first data point.

[0022] After the first n simulation steps by varying the first parameter X, then for the second n simulation steps, the value of the first parameter X is varied for the n data points covering X. However, the value of the second parameter Y changes from Y 1 Change to Y 2 , and the third parameter Z is in Z 1 For example, if n = 4, that is, the first parameter is X 1 , X 2 , X 3 and X 4 , then m = 3, that is, the second parameter is Y 1 , Y 2 and Y 3 , and o = 2, that is, the third parameter is Z 1 and Z 2 .

[0023] Then, for the first set of n=4 simulation steps, in this case, the following points are simulated: (X 1 ,Y 1 ,Z 1 )、(X 2 ,Y 1 ,Z 1 )、(X 3 ,Y 1 ,Z 1 ) and (X 4 ,Y 1 ,Z 1 ).

[0024] Then, for a second set of n=4 simulations, the following points are simulated: (X 4 ,Y 2 ,Z1 )、(X 3 ,Y 2 ,Z 1 )、(X 2 ,Y 2 ,Z 1 ) and (X 1 ,Y 2 ,Z 1 ).

[0025] Then, for a third set of n=4 simulations, the following points are simulated: (X 1 ,Y 3 ,Z 1 )、(X 2 ,Y 3 ,Z 1 )、(X 3 ,Y 3 ,Z 1 ) and (X 4 ,Y 3 ,Z 1 ).

[0026] Then, for the fourth set of n=4 simulations, the following points are simulated: (X 4 ,Y 3 ,Z 2 )、(X 3 ,Y 3 ,Z 2 )、(X 2 ,Y 3 ,Z 2 ) and (X 1 ,Y 3 ,Z 2 ).

[0027] Then, for the fifth set of n=4 simulations, the following points are simulated: (X, Y 2 ,Z 2 )、(X 2 ,Y 2 ,Z 2 )、(X 3 ,Y 2 ,Z 2 ) and (X 4 ,Y 2 ,Z 2 ).

[0028] Then, for the sixth set of n=4 simulations, the following points are simulated: (X 4 ,Y 1 ,Z 2 )、(X 3 ,Y 1 ,Z 2 )、(X2 ,Y 1 ,Z 2 ) and (X 1 ,Y 1 ,Z 2 ).

[0029] Therefore, the process is repeated until all possible combinations of the three parameters have been simulated.

[0030] Thus, according to an embodiment, the simulation steps are preferably performed so that no parameter value "jumps" (increases or decreases) from one step to another by more than a predetermined increment, such as the integer "1" or another value, depending on the parameter. This means that there is preferably no "jump" from the lowest to the highest possible value in a predefined set or range of parameter values. In the above example, there is no "jump" of parameter X from 1 to 4 or from 4 to 1. This increment may also be referred to as a "step size."

[0031] Parameter values ​​used for the simulation increase or decrease in predetermined minimum increments, as if they were selected by a movable slider.

[0032] Furthermore, the choice of the parameter whose value is to be changed while keeping other parameters constant may depend on the time required to complete the entire simulation process, ie, may be selected to achieve completion of the simulation process in the shortest possible time interval.

[0033] Therefore, due to this measure, the solver can calculate new simulation results accordingly quickly because it converges quickly.

[0034] The simulations of the computer-implemented method of the present invention can be calculated in parallel and / or sequentially. In the case of parallel calculation, the simulation space for each set of boundary conditions is divided into several subspaces to be simulated in parallel. The respective simulation results for each subspace of a corresponding boundary condition in the plurality of boundary conditions are determined, all simultaneously or with a time offset, depending on the embodiment. However, in the case of serial calculation, one simulation result for one of the boundary conditions is determined at a time, i.e., each respective simulation result for each respective boundary condition in the plurality of boundary conditions is determined one after another in a sequential manner.

[0035] The fourth step of the embodiment of the computer-implemented method includes providing a digital twin as a set of data points obtained from the simulation step. The digital twin can also be a parametric model or a numerical expression. The parametric model can be in the form of a polynomial function, which can be obtained by interpolation or curve fitting, which interpolates or curve fits the simulation results to a function of corresponding boundary conditions from a plurality of boundary conditions. The parametric model can also be in the form of a Gaussian function. This is advantageous because such a parametric model is computationally cheap and fast compared to the calculation of the simulation results of the set of boundary conditions using software.

[0036] The method of the present invention enables efficient and rapid calculation of simulation results for chemical processes or equipment or systems of chemical plants, in particular for large sets of boundary conditions for material synthesis and / or material separation. In addition, the use of a single one of the parameter values ​​of the various boundary conditions being changed from one simulation step to the next, preferably to the closest data point or value of the parameter, while other parameter values ​​belonging to the various boundary conditions remain unchanged, allows a setup that can be run in a short time on a standard computer. Similarly, the computer-implemented method steps can be performed on at least one processor of a computer. In addition, due to the complexity and sheer amount of data that must be processed and calculated by the processor, it may be advantageous to use an array of parallel processors to perform the calculations necessary for performing the computer-implemented method steps.

[0037] It should be understood that additional intermediate steps may be performed between any consecutive steps of the computer-implemented method of the present invention, which steps may be known to those skilled in the art.

[0038] According to some embodiments of the present invention, a single one of the parameter values ​​of each boundary condition can change from the starting data point to the ending data point in the simulation within its entire data point value set, wherein in a subsequent simulation series, a single one of the parameter values ​​of each boundary condition changes backward, i.e., changes from the ending data point to the starting data point. This "back and forth" mechanism, wherein in a simulation series, i.e., in the fourth step, a single one of the parameter values ​​of each boundary condition can change from the starting data point to the ending data point in the simulation within its entire data point value set, and in a subsequent series, i.e., in the latter step, a single one of the parameter values ​​of each boundary condition can change backward, i.e., from the ending data point to the starting data point. This "back and forth" mechanism can continue to be used for subsequent subsequent simulation series.

[0039] Preferably, the parameter to be changed is changed from the starting parameter forward to the ending parameter in the simulation throughout its data point value set, and from the ending parameter backward to the starting parameter in a subsequent series of calculations, etc. For example, referring to the above example, if the simulation is used to solve an equation containing three parameters (i.e., a first parameter X, a second parameter Y, and a third parameter Z). The first parameter value set contains a first plurality of data points, such as X 1 To X n data points, the second parameter value set contains the second plurality of data points, such as Y 1 To Y m data points, and the third parameter value set includes a third plurality of data points, such as Z 1 to Z o Data point. 1 To X n In the first simulation, n To X 1 In the second simulation, 1 To X n In the third simulation of , the parameter X can be varied. It has been found that the "back and forth" mechanism can save significant computation time compared to methods that always start at the starting parameter or randomly select parameters for each series of calculations. n Y 1 Z 1 The simulation results start from X n Y 2 Z 1 The solution can be compared to X 1 Y 2 Z 1 The solution is closer and can be faster. Figure 2 Further details of the method explained above will be provided as the time goes by.

[0040] In a preferred embodiment, the "back and forth" mechanism can be combined with a specific mode for determining a preferred, usually parameter-specific increment size ("step size") and / or another mode for determining the order according to which multiple parameters will be selected for changing the individual parameter values ​​"back and forth" during the simulation, thereby further reducing the total simulation time. Compared with conventional methods, this combination of mechanism and calculation rules can lead to a reduction in calculation time of up to 99%, and can therefore provide a basis for simulating a larger data space with more boundary conditions in an economically acceptable time. For example, if the pressure value changes from an upper limit to a lower limit, the simulation may not converge at all, for example in a complex process, such as a three-stage biogas process. The next point can then be "approached" by homotopy. This may then take 30 to 60 seconds, while this method may only take 0.5 to 1.5 seconds. This may lead to a reduction in calculation or simulation time of up to 99%.

[0041] According to some embodiments of the present invention, the simulation results from one simulation step can be used as the initial conditions for the next simulation step. In other words, the solution found by the solver is used by the solver in a subsequent simulation step for the changed boundary conditions, such as as an initial condition or initial value, so that a solution for the changed boundary conditions can be found more quickly. This is also referred to as a gradient-based method.

[0042] This is advantageous because for the current simulation step, the initial conditions obtained from the simulation results of the previous simulation step will then be close to the final simulation results of the current simulation step, which will ensure that the solution converges in a much shorter time. In cases where the solver cannot find a solution to the boundary conditions, homotopy can be used. Two continuous functions from one topological space to another are called homotopies, and if one can be "continuously deformed" into the other, then this deformation is called a homotopy between the two functions.

[0043] Homotopy can enable one to move from one steady-state solution to another in small increments. In some cases, a simulation may not converge from the current steady-state solution to the target steady-state solution. Homotopy can allow one to approach the target steady-state solution in stages, thereby improving the chances of reaching the target steady-state solution.

[0044] Homotopy can provide a way to move from one converged solution to another solution with different values ​​for one or more homotopy variables. It is a useful technique where convergence is difficult to obtain for a particular specification, but a converged solution is already available for a different specification. Homotopy can work by moving along a path to a new solution, and solving for multiple intermediate points along that path. For example, let HOM1 be a vector of values ​​of the homotopy variables at points that have already been solved, and HOM2 be a vector of values ​​of the points that you want to move to. In a homotopy simulation, the simulation software attempts to solve for multiple points at the following values ​​of the homotopy variables:

[0045] Homotopy = HOM1 + θ × (HOM2 – HOM1)

[0046] Where θ is the homotopy parameter. This is a number that moves from 0 to 1 on successive solutions. When θ is 1, this corresponds to the canonical HOM2. The way θ varies between successive solutions can be controlled.

[0047] In the case of using homology, the saved snapshot (i.e., the solution of the point) is loaded, and then the non-converged point is approached again by homology. Here, preferably, only one snapshot is loaded before starting the simulation, and it does not matter which point has not converged. In addition, for each individual simulation, the saving of snapshots can be turned off so that no extra time is wasted saving snapshots or the simulation software crashes, because 10,000 or more snapshots can now be saved. The value of the changed parameter value of the current boundary condition is subdivided into multiple values ​​of the corresponding parameter value between the parameter value of the previous simulation step and the final parameter value of the current simulation step. In this case, several simulation steps are incorporated between them, and the simulation results of each simulation step are used as the initial conditions of the subsequent simulation steps. Repeat this step until the final simulation result of the final changed parameter value using the boundary condition is obtained.

[0048] According to some embodiments of the present invention, multiple boundary conditions can be obtained from the simulation space, wherein the boundary conditions for the first simulation step are selected from the central region of the simulation space. In other words, the starting boundary conditions are selected not at the edge of the simulation space, but from the central region of the simulation space, because in the boundary regions of the simulation space there is a risk that the solver will not find a solution there, because usually the boundary regions of the simulation space contain the extreme values ​​of the parameter value range of the boundary condition for which the simulation result needs to be obtained.

[0049] The boundary region of the simulation space can contain a maximum value, while another boundary region of the simulation space can contain a minimum value of the corresponding parameter value. For example, if the simulation result is to obtain the reaction rate of the reaction mechanism, and a parameter value of the boundary condition of the change corresponds to the temperature of the reactant mixture. In a boundary region, the parameter value will contain the maximum possible value of the temperature of the reaction mixture, which will result in a very high reaction rate, because the reaction rate is an exponential function of the temperature. Similarly, at another boundary region, the parameter value will contain the minimum possible value of the temperature of the reaction mixture, for which the reaction rate will be a very low value or even an unrealistic negative value. The simulation under such boundary conditions will take a long time to converge or not converge or diverge, thereby resulting in unrealistic values. This results in a disproportionate time loss, which occurs when trying to find a solution numerically in a physical or chemical boundary region. Therefore, bypassing such a boundary region can lead to avoiding the use of homotopy or reducing the use of homotopy.

[0050] Alternatively, the boundary conditions of the first simulation step may be selected from arbitrary random points from the simulation space or also partially from points close to the edge of the simulation space.

[0051] According to some embodiments of the present invention, the path of the boundary condition selected for the next simulation step can be dynamically adjusted through the simulation space based on the time required for the simulation result to be obtained in the previous simulation step. In other words, for example, if the time required by the solver for the simulation result becomes longer and longer, it is conceivable to dynamically adjust the path through the simulation space during the simulation execution. In this way, it is possible to "bypass" the physical boundary or infeasible area. For example, multiple boundary conditions can also be obtained from the simulation space, wherein the boundary conditions for the first simulation step are selected from the area between the boundary area and the central area of ​​the simulation space. Here, the selection path of the boundary condition for the next simulation step can be selected based on whether the time required by the solver for the simulation result becomes longer than the previous simulation step, if the required time is longer than a specific boundary condition, the specific boundary condition can be bypassed, and the selection path of the boundary condition in the simulation space can be adjusted accordingly, for example, the path can be in a tortuous form. Here, the boundary conditions that require a lot of time for simulation are avoided, so that the use of homotopy can be avoided or reduced.

[0052] According to some embodiments of the present invention, the system of equations can be a system of steady-state equations for a quasi-steady-state process. This is advantageous because it leads to higher speed and easier to analyze simulations. However, the solver can also perform unstable or transient simulations.

[0053] According to some embodiments of the present invention, multiple boundary conditions and the input of initial conditions of simulation step can be realized by the interface to external source.In order to access external source or resource or multiple external sources or resource, external source can be selected so that it is compatible with the software for solving the simulation equations.In order to be compatible with external source, the computer-implemented method can include receiving data from external source or resource or system and / or sending data to external source or resource or system.For versatility, the computer-implemented method can further include accessing external software and providing data to external software and / or receiving data from external software.For compatibility, it is also possible to adopt the interface that is suitable for receiving action prescription from another simulator.

[0054] According to some embodiments of the present invention, a digital twin is a set of data points containing multiple simulation results corresponding to various boundary conditions. This is advantageous because it will enable a user to select or filter out specific groups of points from the digital twin as required in order to obtain various simulation results corresponding to various boundary conditions in a fast and efficient manner. It is also conceivable that a parametric model or numerical expression can be obtained from the set of data points of the digital twin. In this way, the parametric model or numerical expression can be easily combined to obtain real-time similar to an application or real-time in other computer programs or computer software.

[0055] According to some embodiments of the present invention, a chemical process may correspond to a process of a gas separation membrane or a gas separation device having a plurality of gas separation membranes, or an equipment or system of a chemical plant may be a gas separation membrane or a gas separation device having a plurality of gas separation membranes. This is advantageous because it will enable testing and development of a gas separation membrane or a gas separation device having a plurality of gas separation membranes, or an equipment or system of a chemical plant may be a gas separation membrane or a gas separation device having a plurality of gas separation membranes that can operate under a wide range of operating conditions in an efficient and economical manner.

[0056] According to some embodiments of the present invention, multiple equations can be solved in parallel in the simulation step, and each equation uses corresponding boundary conditions to provide corresponding simulation results. For this purpose, the simulation space can be divided into several parallel "slices", i.e., mutually parallel subspaces of the simulation space. The path for parallel simulation can run through each subspace. It is also possible that the simulation space is subdivided into subspace blocks that are not in the form of parallel slices, or it can be a combination of subspace blocks and parallel slices, wherein in the simulation step, each subspace of the simulation space is solved in parallel to each other. Such parallel simulation can be run on a single computer or a cluster or a cloud computer system.

[0057] According to some embodiments of the present invention, a time offset can be used to start solving each of the set of equations in parallel relative to solving one of the set of equations, wherein the time offset is a predetermined fraction of the average time of the simulation steps. Starting the parallelized simulation with a time offset is advantageous because it will smooth the total required processor power over time.

[0058] The predetermined fraction of the average time of a simulation step may be between 5% and 10% of the average time of a simulation step. For example, a simulation step may take 0.5 to 1.5 seconds and then an offset may be selected or set to 45 seconds. Typically, starting a simulation takes longer. A converged solution from a previous simulation may be saved as a "snapshot" file and this "snapshot" may then be loaded into appropriate software, such as ACM software. This snapshot is then brought to converge for a new set of boundary conditions. If convergence is not achieved, homology may be used and a simulation step may take approximately 30 to 60 seconds.

[0059] According to some embodiments of the present invention, simulation results can be stored in a data file or in a plurality of data files. This makes it possible to effectively store and conveniently use the big data of simulation results as required. For clarity and convenience, the computer-implemented method can further comprise providing a series of simulation results, and allows to review each result in the sequence in detail. For convenience, simulation results can be shown in the form of a chart, a line graph, a list, a table and / or a chart. The progress of simulation can be shown in the form of the cumulative value of the ongoing simulation.

[0060] According to some embodiments, the number of simulation steps performed is at least 100,000, in particular at least 250,000, in particular at least 500,000 and in particular at least 1,000,000. The number of data points each comprising a respective one of the simulation steps may also be at least 100,000, in particular at least 250,000, in particular at least 500,000, in particular at least 1,000,000.

[0061] According to some embodiments, the number of boundary condition parameters whose values ​​are varied during the simulation step is at least 3, in particular at least 5, in particular at least 7, in particular at least 9.

[0062] According to some embodiments, the average number of predetermined values ​​assigned to one of the boundary condition parameters changed during the simulation step is at least 2, in particular at least 4, in particular at least 8, in particular at least 64, such as at least in particular at least 256.

[0063] The specific way of performing the simulation by increasing only one parameter at a time, and preferably only a single increment, and preferably according to a specific method of prioritizing specific parameters and selecting appropriate parameter specific step sizes, thus makes it possible to calculate a large number of simulation results, and thus to cover a huge combinatorial space that could not previously be covered by empirical, observational methods or stray force simulation methods. According to some embodiments of the present invention, the simulation results (e.g., Fig.10 ). The AR-glasses may be controlled by AR-software configured to create and display one or more digital twins (digital representations) of a chemical process or equipment or system of a chemical plant. This may enable a user wearing the AR-glasses to immediately recognize any problems that may arise under certain process conditions and to be able to take immediate action to prevent or remedy critical situations. For example, according to one embodiment, the controller is configured to calculate a digital twin of a user wearing the AR glasses and enable the user to move his or her virtual twin relative to a digital twin representing, for example, a portion of a chemical reaction or a portion of a chemical plant by moving his head and / or body in the real world.

[0064] According to some embodiments, the computer-implemented method further includes displaying the digital twin. For example, the display may include displaying the digital twin via an augmented reality display system (e.g., via AR glasses).

[0065] Because the embodiments provide for particularly fast calculation of data points / digital twins, they also provide for particularly fast methods of displaying digital twins: graphical representations of the digital twin, such as multi-dimensional line plots of the data points or more complex graphical representations of the individual hardware components of a plant, can be quickly read into a frame buffer without having to solve numerical equations for determining the predicted state of a chemical process or plant.

[0066] According to some embodiments, display of the digital twin includes loading a graphical representation of the digital twin or a portion thereof into a frame buffer of a display system, and displaying the data contents of the frame buffer on the display system. A frame buffer is a portion of random access memory that contains a bitmap that drives a video display. It is a storage buffer containing data representing all pixels in a complete video frame.

[0067] In some embodiments, the method includes subsequently displaying at least two different states of a chemical process, plant, or machine on a display system, wherein the display includes: selecting a first subset of data points obtained in the entire simulation, generating a first graphical representation of the process, plant, or machine using the selected first subset, loading the first graphical representation into a frame buffer, displaying the contents of the frame buffer containing the first graphical representation on the display system, selecting a second subset of data points, generating a second graphical representation of the process, plant, or machine using the selected second subset, loading the second graphical representation into the frame buffer so that it replaces the first graphical representation, and displaying the contents of the frame buffer containing the second graphical representation on the display system.

[0068] For example, a rule engine can be used to identify a subset of data points corresponding to two or more different states of a chemical process, machine or plant of interest. In some cases, a subset can contain a single data value. For example, if a user is interested in the state (defined by multiple parameter values) of a gas separation facility that requires a certain purity when a certain pressure is given, he or she can use a rule engine to obtain a first data point containing the required purity and annotated with temperature T1, and obtain a second data point containing the required purity and annotated with a different temperature T2. Different temperatures represent different states of a process, machine or plant. A user can navigate in real time between graphical representations of different states of a chemical process, machine or plant, which is particularly useful in the environment of an industrial control process, particularly if control is performed via an augmented reality display system. Typically, the state selection rules for traversing a set of data points obtained by simulation will be more complex and may contain filtering criteria for multiple different parameters, multiple different parameters such as gas (e.g., O 2 or N 2 ), gas purity, membrane surface area, temperature, pressure, feed gas flux, permeate gas flux, retentate gas flux, etc.

[0069] According to some embodiments of the present invention, digital twins can be used to train artificial intelligence models or models based on machine learning. Artificial intelligence models or AI twins or models based on machine learning can be advantageous for the visualization of big data of simulation results together with corresponding boundary conditions, and rapid optimization can be achieved, and real-time optimization and advanced process control can also be achieved. In addition, artificial intelligence models can be used for gray box models and proxy models, such as neural network models, radial basis function models, support vector machine models, or Gaussian process regression models. The model based on machine learning can be a model based on a k-nearest neighbor algorithm, which is a nonparametric supervised learning method, wherein supervised learning is a machine learning task of learning to map inputs to functions of outputs based on exemplary input-output pairs. The model can be used for performance prediction, and can also be used for other unit operations, such as distillation, absorption, and other operations in the field of chemical processes, which are known to those skilled in the art.

[0070] According to some embodiments of the present invention, the digital twin may be a set of data points containing multiple simulation results corresponding to various boundary conditions, and is graphically represented so that a specific set of data points in the set of data points of the digital twin can be selected or filtered, for example, by a user according to requirements. This is advantageous because it can directly display good and bad solutions and make them comparable and understandable, even allowing the user to move around in or through the data space, for example, with the help of a digital slider of a 2D-line graph or a 3D-line graph.

[0071] A second aspect of the invention is a computer program comprising instructions that, when executed by a computer, cause the computer to perform the steps of a computer-implemented method of the invention for generating a digital twin of a chemical process or device or system for a chemical plant, in particular for substance synthesis and / or substance separation. For example, substance separation may be performed by permeation, adsorption, absorption, distillation, filtration or other technical means. Separation may be performed to fractionate or purify a mixture of substances, such as a gas mixture or other type of mixture. The computer program may also comprise a computer program product comprising instructions that, when executed by a computer, cause the computer to perform the steps of a computer-implemented method of the invention.

[0072] In particular, as set out in the context of the first aspect of the invention, the set of parameters may be provided to the computer program from a non-transitory storage medium of the computer.

[0073] A third aspect of the invention is a computer for generating a digital twin of a chemical process or device or system for a chemical plant, in particular for substance synthesis and / or substance separation, the computer being configured to use the steps of the computer-implemented method of the invention. In addition, the computer executing the computer program can be connected to a display configured to display any output of the computer program.

[0074] Another aspect of the present invention is a display system comprising an electronic display and a computer configured to generate a digital twin. For example, the computer can be a distributed or single-chip computer system. The electronic display can be an LCD screen, an OLED screen, or any other type of electronic display. The computer is further configured to generate 2D or 3D computer graphics (e.g., line graphs, charts, or approximate true representations of chemical processes, or plants or machines in which chemical processes are performed) that visualize the data in the digital twin. The computer is configured to display 2D or 3D computer graphics on the electronic display.

[0075] Another aspect of the present invention is a data structure configured as a digital twin of a chemical process or device or system for visualizing a chemical plant, the data structure comprising a plurality of data points obtained by a method according to any one of the embodiments and examples described herein. The data structure is configured to cause the display system to generate a 2D or 3D computer graphic that visualizes the data in the digital twin when processed by the display system; and to display the 2D or 3D computer graphic on an electronic display of the display system.

[0076] This can have the advantage of providing a particularly fast and lightweight display system that is able to very quickly visualize a chemical process or a plant or machine performing a chemical process. The graphical representation can be derived directly from the set of data points without having to solve a complex simulation, and the calculations on the set of data points are also performed very quickly. In fact, the display system can be a particularly fast and high-resolution display system because the density of data points in the multi-dimensional parameter space reflects the resolution of the 2D or 3D model of the entity represented by the digital twin. Given a certain amount of computing time and computing resources, the way the data points are calculated by changing only one parameter value from one simulation to the next can provide better resolution because more simulations can be performed and converged per the available amount of time / CPU power.

[0077] The term "computerized apparatus," "computerized system," or similar terms refers to a device including one or more processors operable or operating according to one or more programs.

[0078] The term "computer" or system thereof may be used herein in the ordinary context of the art, such as a general purpose processor or microprocessor, a RISC processor or a DSP, which may include additional elements such as memory or communication ports. Alternatively or additionally, the term "computer" or its derivatives refer to a device capable of executing a provided or incorporated program and / or capable of controlling and / or accessing data storage devices and / or other devices such as input and output ports. The term "computer" also refers to multiple processors or computers connected, and / or linked and / or otherwise communicating, possibly sharing one or more other resources, such as memory information of non-transitory storage media.

[0079] As used herein, the term "server" or "client" or "backend" refers to a computer or computerized device that provides data and / or operational service(s) to one or more other computerized devices or computers.

[0080] The terms "software", "computer program", "software program" or "steps" or "software code" or "code" or "application" or "app" may be used interchangeably depending on the context and refer to a product or method that contains one or more instructions or instructions or circuits for performing a series of operations that generally represent an algorithm and / or other process or methodology. The program may be stored in or on a medium such as RAM, ROM or disk, or embedded in circuitry accessible and executable by a device such as a processor or other circuit.

[0081] The processor and the program may constitute at least partially the same device, for example an electronic gate array, such as an FPGA or an ASIC, designed to carry out a programmed sequence of operations, optionally containing or linked to a processor or other circuits.

[0082] As used herein, without limitation, a process refers to a collection of operations for achieving some objective or result.

[0083] Similarly, a model can represent a collection of operations used to achieve some goal or outcome.

[0084] The terms "configure" and / or "adapt" as used for an object or variations thereof, mean that an object is achieved using at least software and / or electronic circuitry and / or auxiliary equipment that is designed and / or implemented and / or operable or operational.

[0085] A device storing and / or containing a computer program and / or data, such as a non-transitory storage medium, in particular constitutes an article of manufacture. Unless otherwise stated, the program and / or data is stored in or on a non-transitory medium.

[0086] In the context of embodiments of the present disclosure, by way of example and not limitation, terms such as "operation" or "execution" also imply capabilities such as "operable" or "executable," respectively. BRIEF DESCRIPTION OF THE DRAWINGS

[0087] In the following, embodiments of the invention are explained in more detail, by way of example only, with reference to the accompanying drawings, in which:

[0088] Figure 1 is a block diagram of one embodiment of a system of the present invention for generating a digital twin of a chemical process or equipment or system for a chemical plant;

[0089] Figure 2 is a parameter table containing multiple boundary conditions, and the multiple boundary conditions include parameter value sets;

[0090] Figure 3 is the digital twin as the set of points obtained from the simulation step;

[0091] Figure 4 It is the first path to select boundary conditions from the simulation space;

[0092] Figure 5 is the second path for selecting boundary conditions from the simulation space;

[0093] Figure 6 is a flow chart of one embodiment of a computer-implemented method, wherein a digital twin is provided as a set of data points obtained from a simulation step;

[0094] Figure 7 is a comparison of simulation results of the ACM solver and the artificial neural network model; and

[0095] Figure 8 is a flowchart embodiment of a computer-implemented method, wherein in the case of parallel simulation, a time offset is used to start solving each of the system of equations in parallel relative to solving one of the system of equations;

[0096] Fig. 9 is a schematic diagram of a computer for generating a digital twin of a chemical process or a device or system for a chemical plant, in particular for substance synthesis and / or substance separation, the computer being configured to use a computer-implemented method;

[0097] Fig.10 is the filtering of data obtained from the digital twin;

[0098] Fig.11A ,B is an illustration of the effect of step selection on simulation performance;

[0099] Fig.12 is a set of tables used to illustrate the identification of appropriate step widths;

[0100] Fig.13 is a set of tables used to illustrate the identification of appropriate parameter sequences;

[0101] Fig.14 is a line graph illustrating the simulation times observed for three different parameters assuming a different number of allowed values ​​for each parameter; and

[0102] Fig.15 It is a flowchart of an embodiment of a computer-implemented method of the present invention for generating a digital twin of a chemical process or equipment or system for a chemical plant. DETAILED DESCRIPTION

[0103] Fig.15 FIG. 1 is a flow chart of an embodiment of a computer-implemented method of the present invention for generating a digital twin of a chemical process or device or system for a chemical plant. Figure 1 A corresponding system configured to perform the method is shown. Figure 1 and Fig.15 . The computer-implemented method includes a first step 150 of inputting a plurality of boundary conditions 1 into a computational model 2. The computational model 2 comprising a set of equations is solved in a process simulation software 8 (such as Aspen CustomModeler, ACM). Each boundary condition 1 comprises a plurality of parameter values ​​4. The boundary condition 1 comprising a plurality of parameter values ​​4 may be assigned by a set of discrete attribute pairs in a steady state. The parameter values ​​4 of the respective boundary conditions 1 may be stored in a tabular database 6, such as Microsoft Excel. The tabular database 6 may be input into the computational model 2. The parameter values ​​4 of the respective boundary conditions 1 read from the tabular database 6 may be stored in the form of parameter table data 5, such as Figure 2 . Parameter values ​​4 may include pressure, temperature, mass flow, molar flow, volume flow, feed composition and / or any other characteristic property or parameter, each of which is shown in a respective column 7 within the parameter table data 5. Input of multiple boundary conditions 1 may be performed by selecting sets of parameter values ​​4 corresponding to respective boundary conditions 1 from a tabular database 6, such as Microsoft Excel in this case, or another external source, such as a cloud service / system, or multiple external sources. It should be noted that in Microsoft Excel, variables, such as temperature or pressure, may be defined by entering their maximum and minimum values ​​as outer limits for the data points to be simulated and the step size for calculating other data points in between. Alternatively, a sequence of data points or a grid of data points may be entered, or a tabular database may be entered, which may be created in software such as, for example, Konstanz Information Miner, KNIME.

[0104] The second step 152 of the computer-implemented method of the present invention comprises inputting the initial conditions for solving the system of equations into the computational model 2 of the process simulation software 8. The input of the boundary conditions 1 from the tabular database 6 to the computational model 2 is automatically performed by data analysis software, such as KNIME. The double arrows indicate that the simulation results of one simulation step from the computational model 2 are used as the initial conditions 3 for the next simulation step.

[0105] Figure 1 Shown is a graphical user interface 9 of a process simulation software 8. A situation including boundary conditions 1 and a system of equations can be solved in the process simulation software 8. In addition, for speed, the GUI may not be opened by the ACM, but may be allowed to open automatically.

[0106] The third step 154 ​​of the computer-implemented method includes solving the system of equations using each of the boundary conditions 1 in the simulation step respectively to provide corresponding simulation results.

[0107] A fourth step 156 of the computer-implemented method comprises providing a digital twin 20 as a set of data points 21 obtained from the simulation step.

[0108] Figure 2 A representative portion of parameter table data 5 is depicted to explain the "back and forth" described further above. The first column 10 represents the simulation step number, starting with 1, which represents the first simulation run using boundary condition 1, which contains the corresponding parameter value 4 corresponding to the first row 11 within parameter table data 5.

[0109] The third step of the computer-implemented method includes solving the system of equations using each of the boundary conditions 1 in the simulation step to provide a corresponding simulation result. In this case, for the first simulation step shown in the first column 10 and the first row 11, a first parameter value, a first value 12 from the second column 13, a second parameter value, a first value 14 from the third column 15, and a third parameter value, a first value 16 from the fourth column 17 are used. Here, the parameter values ​​12, 14, and 16 of the first row 11 represent the first boundary condition 18.

[0110] According to the computer-implemented method, a single one of the parameter values ​​4 of each boundary condition 1 is changed from one simulation step to the next, while the other parameter values ​​4 belonging to each boundary condition 1 remain unchanged. Figure 2As shown, considering the first five simulation steps, as shown by the numbers 1 to 5 in the first column 10, it can be seen that the value of the first parameter value 12 changes from 0.400 to 0.600, while the value of the second parameter value 14 remains unchanged at 7.000, and the value of the third parameter value 16 remains unchanged at 10.000. Due to this measure, the process simulation software 8 can calculate new simulation results quickly accordingly because it converges quickly. It can also be noted that the step size of the first parameter value 12 is taken as a small value of 0.5, because a smaller step size leads to better and faster convergence of the simulation results. In addition, the simulation results from one simulation step are used as the initial conditions of the next simulation step. For example, the simulation results of the first simulation step using the first boundary condition 18 will be used as the initial conditions of the second simulation step using the second boundary condition 19.

[0111] After the first five simulation steps, as indicated by numbers 1 to 5 in the first column 10, for the sixth simulation step, the value of the second parameter value 14 from the third column 15 is changed from 7.000 to 7.500, while the other two parameter values ​​are maintained. In the sixth simulation step, the first parameter value 12 in the second column 13 is maintained at the same value of 0.600 as in the fifth simulation step. Similarly, the third parameter value 16 from the fourth column 17 is maintained constant at the same value of 10.000 as from the fifth simulation step. Once the second parameter value is changed to a value of 7.500 in the sixth simulation step, the value of the first parameter value is changed from 0.600 to 0.400, while the second parameter value 14 and the third parameter value 16 remain unchanged at the values ​​of 7.500 and 10.000 for the simulation steps 6 to 10. Then, a similar simulation process is further repeated until all parameter values ​​4 are considered to obtain the respective simulation results.

[0112] Figure 3 The digital twin 20 is shown as a data point set 21 containing a plurality of simulation results corresponding to respective boundary conditions 1. Here, each of the three axes represents a first parameter value 12, a second parameter value 14, and a third parameter value 16 of the boundary condition 1. The color of each data point 21 represents the corresponding simulation result. The value of the simulation result of each data point of the data point set 21 can be obtained by a color scale 22.

[0113] For ease of understanding, the digital twin 20 describes the first parameter value 12 as being dependent on the other two parameter values ​​14 and 16. However, in a simulation, the digital twin 20 may contain multiple parameter values, for example, more than 200 parameter values. The values ​​and dependencies of each parameter value may be displayed in the form of a multi-dimensional graph or line graph or other representative forms known to those skilled in the art.

[0114] Figure 4Shown is a first path 24 for selecting boundary condition 1 from simulation space 23, with the x-axis representing rotation speed 27 and the y-axis representing torque 28. Here, multiple boundary conditions 1 are obtained from simulation space 23, wherein the first boundary condition 18, i.e., boundary condition 1 for the first simulation step, is selected from the central region 42 of simulation space 23. Because usually the boundary region 25 of simulation space 23 contains the extreme values ​​of the range of parameter values ​​4 of boundary condition 1, simulation results need to be obtained for this boundary condition. Simulations at such boundary regions 25 will take a long time to converge or not converge or diverge, resulting in unrealistic values. This results in a disproportionate time loss, which occurs when trying to find a solution numerically in a physical or chemical boundary region. Such a boundary region 23 is not selected, and therefore, the first path 24 is adjusted, as shown in adjustment path 26. Therefore, the use of homotopy can be avoided or reduced.

[0115] Figure 5 Shown is a second path 29 for selecting boundary conditions 1 from the simulation space 23, with the x-axis representing the rotation speed 27 and the y-axis representing the torque 28. Here, a plurality of boundary conditions 1 are obtained from the simulation space 23, wherein the first boundary condition 18, i.e., the boundary condition 1 for the first simulation step, is selected from the extremely low rotation speed region 27 in the simulation space 23. The second path 29 for selecting boundary conditions 1 for the next simulation step is dynamically adjusted through the simulation space 23 based on the time required to obtain the simulation result in the previous simulation step. Here, the selected second path 29 for the boundary condition 1 for the next simulation step is selected based on whether the time required by the solver for the simulation result becomes longer than the previous simulation step, and if the required time is longer than a specific boundary condition 1, for example, in the boundary region 25, the specific boundary condition 1 can be bypassed, as shown in the adjustment path 26, and the selected second path 29 for the boundary condition 1 within the simulation space 23 is adjusted accordingly, for example, the second path 29 is in the form of a broken line 30. Here, boundary conditions 1 that require a lot of time for simulation are avoided, especially in the boundary region 25, so that the use of homotopy can be avoided or reduced.

[0116] Figure 6 A flow chart of one embodiment of the computer-implemented method is shown, wherein a digital twin 20 is provided as a data point set 21 obtained from a simulation step. The data point set 21 is stored in the form of a data file 32, such as comma separated value data "CSV data". The digital twin 20 is a data point set 21 containing multiple simulation results corresponding to various boundary conditions 1, such as Figure 1 This will enable the user to select or filter out specific point groups from the digital twin 20 as required, so as to obtain various simulation results corresponding to various boundary conditions 1 in a fast and efficient manner. The digital twin 20 can be represented by a visual line graph, such as Fig.10, where the user can select or filter out a specific set of points from the digital twin 20 as needed, for example, by means of a digital slider 35, as shown in FIG. Fig.10 As shown. The digital twin 20 is used to train an artificial intelligence model 34 or a model based on machine learning (not shown in the figure). The artificial intelligence model 34 or the model based on machine learning can be advantageous for the visualization of big data of simulation results together with the corresponding boundary conditions 1, can achieve rapid optimization, and can also achieve real-time optimization and advanced process control. In addition, the artificial intelligence model 34 can be used in gray box models 36 and proxy models such as neural network models 37.

[0117] Figure 7 Shown is a comparison of simulation results obtained by the ANN model 34 and the ACM solver 8 for a 2-step biogas process. It can be seen that the predictions made by the ANN model 34 and the ACM solver 8 match very well, since all the obtained points lie around a straight line 43 inclined at 45° to each axis.

[0118] Figure 8 1 is a flow chart embodiment of the computer-implemented method, wherein in the case of parallel simulations, a time offset 50 may be used to start solving each of the sets of equations in parallel relative to solving one of the sets of equations. Such parallel simulations may be run on a single computer 55, such as Fig. 9 As shown, or run on a cluster. For example, a simulation may usually take 0.4 to 2 seconds. For parallelization, there can be an offset of 45 seconds. In order to load, open and start a simulation file or step with ACM, read in the start value, and the first point can be approached by homotopy, which may take about 1 to 2 minutes. But this depends largely on the complexity of the mathematical model and the duration of the homotopy. For example, about 6 to 8 parallel simulations can also be started simultaneously on a server computing system.

[0119] like Figure 8 As shown, three simulation files or steps are shown, and for ease of understanding only, the first duration 44 of the first simulation file 45 is shown to be greater than the second duration 46 of the second simulation file 47, and the second duration 46 is further greater than the third duration 48 of the third simulation file 49. Here, a time offset 50 is used to start solving each set of equations in one simulation file 45, 47, 49 in parallel with solving a set of equations in another simulation file 45, 47, 49. The time offset may be a predefined value that may vary for different simulation steps.

[0120] For example, if 30,000 simulation points need to be simulated, and it is assumed that each simulation point takes an average of 1 second to simulate. People will not wait for 30,000 seconds, i.e., about 8.4 hours. Therefore, in this case, the ACM simulation can be parallelized in three simulation files or steps (i.e., the first simulation file 44, the second simulation file 47, and the third simulation file 49). In each simulation file, 10,000 simulation points need to be calculated, which means about 3 hours. However, if all three simulation files are started at the same time, the computer / server may stop running. Therefore, the first simulation file 45 can be opened first, and a simulation with 10,000 points can be started. Loading, opening, and starting may require a large amount of computer resources. Therefore, only one simulation file is opened, which is the first simulation file 45 in this case. In addition, the first point of the first simulation file 45 is always started by homology, because there is no starting value when running for the first time, which takes about 1 to 2 minutes. The second simulation point only takes 0.5 to 2 seconds, because the old result of the simulation from the first point can be used as the starting value of the second point. Furthermore, each of the following simulations at the following points may take only about 0.5 to 2 seconds respectively.

[0121] With a time offset of 50, which may be 45 seconds, the second simulation file 47 is started. Then, 45 seconds later, the third simulation file 49 is started again. The choice of the value of the time offset 50 will depend on how long it takes to start the simulation file and how long the computer spends "busy".

[0122] It should be noted that any of the durations 44, 46, 48 may be greater or less than any of the other durations 44, 46, 48. In addition, the division of the simulation points into three simulation files is merely an example. One skilled in the art may parallelize the simulation into a selected number of simulation files. The selection of the number of simulation files may depend on the total number of simulation points that need to be simulated, the complexity of the simulation itself, the available computing resources, the number of points that do not converge and where various homotopies need to be considered, and the time required to complete the simulation. It is advantageous to start the parallelized simulation with a time offset 50 because this will smooth the total required processor power over time.

[0123] Fig. 9 The computer 55 is a schematic diagram of a computer for generating a digital twin of a chemical process or a device or a system for a chemical plant, in particular for substance synthesis and / or substance separation, wherein the computer is configured to use a computer-implemented method; the computer 55 comprises an input unit 51, which is configured to receive a digital twin of a chemical process or a device or a system for a chemical plant, in particular for substance synthesis and / or substance separation ... Figure 1The boundary condition 1 shown. The transmission of the boundary condition 1 between the external source 54 and the input unit 51 can be performed in a wired or wireless manner. The computer 55 includes a storage unit 52, which is configured to store the received boundary condition 1. The computer 55 includes a processor 53, which includes a calculation model 2, such as Figure 1 As shown, and is configured to solve a system of equations with boundary conditions 1. The simulation results obtained from the processor 53 can be stored in the storage unit 52 as a digital twin 20, such as Figure 1 The computer 55 includes an output unit 54 configured to output simulation results from the digital twin 20 from the storage unit 52 .

[0124] Fig.10 Filtering of data obtained from a digital twin 20 is shown. The digital twin 20 is a data point set 21 containing multiple simulation results corresponding to various boundary conditions 1. A specific set of data points in the data point set of the digital twin 20 can be selected or filtered as needed to obtain various simulation results corresponding to various boundary conditions 1 in a fast and efficient manner. The digital twin 20 can be represented by a 3D-line graph 33, where a user can select or filter out a specific set of points from the digital twin 20 as needed, for example, by Figure 8 The digital slider 35 shown is used to obtain a filtered 3D-line graph 38. Here, the ranges of the three parameter values ​​4, namely pressure 39, CO 2 Concentration 40 and CH 4 Purity 41 varies as required, for example, maximum pressure 39 is 14 bar, maximum CO 2 Concentration 40 is 0.35 and maximum CH 4 Purity 41 is a pressure range of 0.9825. The color of each data point 21 represents the corresponding simulation result. The value of the simulation result for each data point of the data point set 21 can be obtained by a color scale 22. The visualization and filtering process can be incorporated as a web application. Therefore, the computer-implemented method thus enables the generation of a digital twin of a chemical process or a device or system for a chemical plant, in particular in the field of gas mixtures that are separated by gas separation membranes due to the different permeabilities of the individual gases.

[0125] Figure 11 is a diagram of the effect of step selection (step width and path topology) on simulation performance. Typically, each simulation uses the values ​​of multiple boundary condition parameters as input, whereby the number of boundary condition parameters considered in some cases is greater than 3, or greater than 5, or greater than 7 or greater than 9. The number of values ​​evaluated for each parameter can be averaged, for example, at least 2, at least 4, at least 8, at least 64 or at least 256. Thousands or even tens of thousands or millions of simulations can be performed. Here, for simplicity, simulations have been performed on a simplified grid of 3 different boundary condition parameters and 6 different values ​​for each parameter. Therefore, in order to simulate all possible value combinations, 6×6×6=216 simulations must be performed.

[0126] To further simplify the description of the simulation, Fig.11A The grids depicted in the figure show only two of the three different boundary condition parameters: The x-axis of each of the grids 102-108 can represent the boundary condition parameter B, such as the A1 / A3 membrane surface ratio of the three-stage gas separation device, thereby depicting six different parameter values ​​B1, B2, B3, B4, B5 and B6. The y-axis of each grid 102-108 can represent the boundary condition parameter A, such as the recycle of the three-stage gas separation device, thereby depicting six different parameter values ​​A1, A2, A3, A4, A5 and A6.

[0127] For simplicity, a grid with 6 possible values ​​for 3 different parameters was used as the basis for the simulation, whereby the increments between different values ​​of the same parameter were assumed to be equidistant, and the grid depicted in Figure 11 has equidistant grid sizes in both dimensions. Typically, however, these parameters differ in the number of possible different values ​​and in the step size and units (if any).

[0128] Grid 102 represents a single-step data selection (single-step parameter value change) method according to an embodiment of the present invention: In the first five simulation runs, only the first boundary condition parameter A is changed from A1 to A6, while parameter B remains constant at value B1. Then, A is changed from A6 back to A1, while B (but not another parameter C) has a value that increases in a single step from B1 to B2. The total run time required to execute all 216 parameter value combinations based on the parameter value change pattern shown in line graph 102 is 5 minutes and 9 seconds.

[0129] Grid 104 represents a different approach: In the first six simulation runs, A is varied from A1 to A6, while parameter B (and parameter C, not shown) remain constant. Then, B is assigned a new value that increases by 1, but A is now varied again from A1 to A6 (not from A6 to A1, as shown in 102). Then, in the next six simulation runs, B is again assigned new values ​​that increase by 1, and A is again varied from A1 to A6. The time required to execute all 216 parameter value combinations based on the parameter value change pattern shown in line graph 104 is almost twice as long as grid 102, i.e., 9 minutes and 15 seconds.

[0130] Grid 106 shows a selection of boundary condition parameter values ​​that "jumps" from A3 to A4, then from A4 to A2, then from A2 to A5, then from A5 to A1, and from A1 to A6. During these six steps, the value of B remains constant (as illustrated by the depicted parameter B, which always has the value B1). As can be inferred from grid 106, the change of boundary condition A is not based on a "single step", but on a "jump", that is, the parameter value changes by more than one increment. For further simulation runs (not shown in 106), parameter B is changed on a single-step basis. Although only a single parameter is changed from one simulation to the next, the "jump" performed when changing a single parameter over more than one increment results in a significantly extended run time of 28 minutes for the 216 simulations.

[0131] Grid 108 shows a random boundary condition parameter value selection strategy. In the depicted example, the "step width" and the number and identity of boundary conditions that are varied in each simulation are all randomly varied. In the depicted example, a run time of 40 minutes was observed for performing 216 simulations, which is almost ten times longer than the simulation strategy according to the embodiment of the present invention illustrated in grid 102.

[0132] Fig. 11B The line graph 110 shows the cumulative run time for 216 parameter value combinations for executing the above four parameter selection strategies. As can be inferred from the line graph 110, the parameter value selection strategy according to an embodiment of the present invention (HTS Run 0) quickly and significantly outperforms the three alternative methods (HTS Runs 1-3) corresponding to grids 104-106.

[0133] Fig.12is a set of tables illustrating the identification of suitable step widths for increasing simulation speed and minimizing the CPU resources required to perform the simulations. According to a preferred embodiment, multiple simulations are performed such that only one parameter value changes from one simulation to the next, whereby the change is performed so as to use the parameter value that immediately follows the previously used parameter value in a predefined, parameter-specific series of parameter values. This method may also be referred to as a "single-step" parameter value change method (i.e., no jumps to change a parameter value from one simulation to the next by more than one increment). The parameter may be, for example, a parameter of a boundary condition or an initial condition.

[0134] For example, a series of data values ​​dedicated to and assigned to a particular parameter may be a series of discrete and preferably equidistant different values ​​that span the range of values ​​allowed for and assigned to the parameter. The range of values ​​is defined by a minimum and a maximum value. Depending on the parameter, the values ​​may also have units, such as °C or kg or m 2 Different parameters may have different numbers of allowed parameter values ​​assigned to them.

[0135] Applicants have observed that the order in which parameters are varied during multiple simulations and the "step size" (or "increment") that corresponds to and determines the number of allowed parameter values ​​assigned to each parameter have an impact on both the performance and accuracy of the simulation: if the step size is too small, the number of simulations will increase dramatically, resulting in a significant decrease in performance. If the step size is too large, the simulation may not converge or may not provide a digital twin that accurately represents a sufficient number of relevant operating states of a chemical process, device, or system.

[0136] According to the depicted embodiment, there are three boundary condition parameters, namely: A1 / A3 ratio, recirculation and compressor outlet pressure.

[0137] Parameter A1 / A3 ratio represents the ratio of the membrane area used in the first and third stages of gas separation facilities.For example, gas separation facilities can be equipment for separating raw gas flow, which is carried out in equipment including feed flow separation stage (first stage), retentate separation stage (second stage) and permeate separation stage (third stage).The facility can include one or more membranes in each of the three separation stages, and can include at least one compressor.An embodiment of this system is described in EP 2 588217B1.For example, the first stage can be a membrane separation stage for separating feed flow into a first permeate stream and a first retentate stream.The second separation stage can be a membrane separation stage, which can have a structure that is the same or different from the feed flow separation stage, for separating the first retentate stream into a second permeate stream and a second retentate stream.The third separation stage can refer to a membrane separation stage, which can have a structure that is the same or different from the feed flow separation stage and / or the retentate separation stage, and can be used for separating the first permeate stream into a third permeate stream and a third retentate stream. In the depicted embodiment, the A1 / A3 ratio should be allowed to be a value between 0.700 and 1.200, and the minimum and maximum values ​​are set accordingly.

[0138] The parameter recycle represents the fraction of the retentate of the second stage which is compressed and recycled to the previous stage relative to the feed stream of the second separation stage. In the depicted embodiment, recycle should be allowed to values ​​between 0.340 and 0.440, and the minimum and maximum values ​​are set accordingly.

[0139] The parameter "compressor outlet pressure" represents the pressure on the feed side of the feed stream separation stage (first separation stage). The pressure can be generated by a compressor arranged upstream of the feed stream separation stage. In the depicted example, the pressure generated by the compressor should be allowed to be a value between 10.0 bar and 15 bar, and these minimum and maximum values ​​are set accordingly.

[0140] Typically, the minimum and maximum values ​​are set by the user taking into account, for example, literature values ​​and / or parameter values ​​that are known to be tolerable or supportable for the gas separation facility whose operation is to be simulated. In some embodiments, the minimum and maximum ranges simply define the ranges of values ​​that are to be simulated because they are of interest for a particular gas separation facility design or use project.

[0141] Table 1206 not only shows the minimum and maximum values ​​of the three example parameters, but also finds the step size "Δ" which has the advantage of providing particularly fast simulation results. In order to determine the appropriate step size ("Δ" or "increment") for each parameter, the "step size determination method" (SLD-method) is performed by the computer. In order to determine the appropriate order of parameters whose values ​​are to be changed first, the "parameter-order-determination method" (POD-method) is performed by the computer.

[0142] The SLD method includes the following steps:

[0143] a) Assign to a first parameter (e.g.: A1 / A3 ratio) a first series of different, preferably equidistant parameter values ​​that are within the minimum and maximum parameters assigned to that parameter. For example, as shown in Table 1206, four different parameter values ​​ranging from 0.70, 0.825 to 1.200 are assigned to the A1 / A3 ratio. For the second and third parameters, the average values ​​of their respective parameter ranges are assigned (see Table 1208: Average Recirculation (Avg.Rec) and Average Compression (Avg.Comp)).

[0144] b) for each set of parameter values ​​thus obtained, e.g. for each row of table 1208, performing a simulation of the design and / or dynamic behavior of the gas separation facility using said set as input, e.g. as parameter values ​​for boundary conditions and / or initial conditions.

[0145] c) The assignment of the first parameter (here: A1 / A3 ratio) to a second series of different, preferably equidistant parameter values ​​within the minimum and maximum parameters assigned to this parameter, whereby the second series contains more values ​​than the first series. For example, as shown in table 1210, eight different parameter values ​​ranging from 0.70, 0.7625 to 1.200 are assigned to the A1 / A3 ratio. For the second and third parameters, the average values ​​of their respective parameter ranges are assigned.

[0146] d) for each set of parameter values ​​obtained in step c), e.g. for each row of table 1210, performing a simulation of the design and / or dynamic behavior of the gas separation facility using said set as input, e.g. as parameter values ​​for boundary conditions and / or initial conditions.

[0147] e) repeating steps c and d) a plurality of times, whereby in each repetition the number of values ​​in the series of different data values ​​assigned to the first parameter is increased (e.g., may be doubled). The repetitions may continue until a predetermined termination criterion is reached, such as a predetermined maximum number of repetitions, a maximum number of different data values ​​in the series, etc.

[0148] f) Repeating steps a) to e) for a different one of the boundary condition parameters). For example, the number of data values ​​in the series of data values ​​assigned to the parameter "Recirculation" may be increased in each repetition, while a constant mean value is used for the other parameters "A1 / A3 ratio" and "Compressor".

[0149] g) Repeating step f) until each parameter of the boundary condition has been used once, increasing its assigned series of "allowed" different data values ​​in each repetition defined in c) and d).

[0150] As described above, each of a plurality of different series of data values ​​assigned to a given parameter (which corresponds to steps a) and c) and tables 1208, 1210) can be used to perform a set of simulations, whereby each different data value of the series corresponds to a simulation. The total time used to perform all simulations in steps b) (and d)) is measured and stored.

[0151] Applicants have observed that for all simulations, the total simulation time required for each step b) or d) will typically decrease as the number of data values ​​in the series increases. However, after the data value series has exceeded a threshold of different data values ​​contained therein, the total simulation time will increase.

[0152] A suitable sequence of data values ​​to be assigned to a given parameter is typically the sequence of data values ​​having the shortest total simulation time for performing all simulations defined by the combination of parameter values ​​created in step b) for the sequence of data values. The step width of a single step, i.e., "Δ" in Table 1206, is defined by the distance between two subsequent data values ​​in the series of data values ​​that has been observed to provide the shortest total simulation time for all values ​​in the series. Therefore, according to an embodiment of the present invention, the SLD method includes automatically identifying a suitable step size for one or more of the boundary condition parameters (and therefore, a suitable number of different predefined values ​​to be used and changed during the simulation), the identification comprising identifying a series of different data values ​​having the shortest total simulation time for performing all simulations defined by the combination of parameter values ​​created in step b) for the series of data values. In Fig.13 An example of selecting an appropriate step size for a parameter is illustrated in Table 1302 of , which contains the number of discrete values ​​of each parameter evaluated in the first column and the simulation time for a single simulation in the second column. The total simulation time for rows 1-4 is provided in the third column of each table.

[0153] Therefore, for the parameter A1 / A3 ratio, a suitable step size / series of different data values ​​should be chosen so that the predetermined parameter value range contains 4 different parameter values ​​to be changed during the simulation, since 4 different parameter values ​​correspond to the shortest total simulation time of 7.2 seconds.

[0154] Different parameters can be assigned different step sizes / different numbers of discrete data values.

[0155] According to some embodiments, the list of automatically identified best-fitting different parameter values ​​can be manually modified, for example, to obtain better resolution. For example, using 8 instead of 4 different values ​​for the A1 / A3 ratio will double the resolution of the simulation results, thereby only slightly increasing the simulation time.

[0156] In addition, for some parameters, the minimum curve cannot be observed when the total simulation time is evaluated for an increasing number of different data values. For example, the total simulation time for rows 1-4 of table 1304 for parameter recycling will be minimum when only two different parameter values ​​are used. In these cases, the number of different values ​​assigned to the parameter and changed during the simulation can be set manually as a compromise between the simulation run time and the need to evaluate different values ​​to receive a sufficiently fine data point set that reflects the effect of the parameter on the overall gas separation process in sufficient detail. Small step sizes are generally (to some extent) slower than large step sizes (from 4), but may not provide enough detail for each parameter.

[0157] Therefore, the above-described automatically performed SLD method may optionally include a manual step for manually modifying the number of different data values / step sizes identified by the SLD method for each parameter.

[0158] The SLD method is used to identify an appropriate, parameter-specific pattern of value increments, whereby the increments can be of the same or different sizes. If the value increments of a parameter are the same, this means that the parameter value is always increased or decreased by increments of a constant, parameter-specific size ("step size") from one simulation to the next.

[0159] The simulations performed to determine the step size of the parameters correspond to the number of different values ​​that will be assigned to the parameters, using the same set of equations that are later used to perform the simulations that generate the digital twin. However, since the other boundary condition parameters always use a constant average value, the simulations are much faster and are not used to generate the digital twin. Therefore, the simulations performed in the SLD method are also called "preliminary simulations".

[0160] Fig.13 A set of tables is shown for illustrating the identification of a suitable parameter sequence (POD method). The POD method is usually performed after the SLD method has been performed, i.e. after parameter-specific "step sizes" (corresponding to respective series of different parameter values ​​within a range of values ​​defined by given minimum and maximum values) have been identified. This can have the advantage of using e.g. Fig.12The advantages of the calculation times already exist as depicted in the table of . However, the POD method can also be performed after the user has selected the appropriate parameter-specific step sizes completely manually. However, even in this case, a preliminary simulation is performed, since the results may be needed as a basis for determining a sequence of parameters to be varied that is particularly fast.

[0161] The POD method is performed to identify one of the multiple boundary condition parameters as the first boundary condition parameter to be changed during the simulation. This is performed by selecting the parameter to be changed during the simulation as the first boundary condition, which produces the fastest results for each simulation at least for the determined, most suitable number of data values ​​assigned to the parameter in the SLD method. In some embodiments, this is performed by selecting the parameter to be changed during the simulation as the first boundary condition, which produces the fastest results in each simulation for the most series of different data values ​​assigned to the parameter in the SLD method. A preliminary simulation can be used to determine the simulation time, in which the values ​​of other parameters are not changed, but are set to constant values, such as the average of their respective assigned numerical ranges.

[0162] For example, the parameters to be used for the first parameter to be changed can be determined by analyzing the results of the SLD method, in which the simulations are performed based on boundary parameter values, which are selected so that the value of only one parameter at a time is changed and all other parameters have in each case only the average of their value ranges assigned by a minimum and a maximum value. The three tables 1302, 1304 and 1306 illustrate the simulation times for each preliminary simulation (second column) and the simulation times for all simulations for a given number of different parameter values ​​(third column), which are obtained for a specific parameter assuming a given number of different values ​​(e.g., 2, 4, 8, 32 and 64).

[0163] Table 1302 shows the time required to perform various simulations when performing the SLD-method: When the A1 / A3 ratio was assigned as a series with only two different values ​​(while the recirculation and compressor parameters were assigned as constant average values), the average measured time for calculating a single simulation was 4.11 seconds. When the A1 / A3 ratio was assigned as a series with 4 different values ​​(while the recirculation and compressor parameters were assigned as constant average values), the average measured time for calculating a single simulation was 1.81 seconds. When the A1 / A3 ratio was assigned as a series with 8 different values, the average measured time for calculating a single simulation was 1.18 seconds.

[0164] Table 1304 shows the times required to perform various further simulations when performing the SLD method: When the recirculation was assigned as a series with only two different values ​​(while the A1 / A3 ratio and the compressor parameters were assigned as constant average values), the average measured time for calculating the individual simulations was 1.88 seconds. When the recirculation was assigned as a series with 4 different values ​​(while the A1 / A3 ratio and the compressor parameters were assigned as constant average values), the average measured time for calculating the individual simulations was 1.24 seconds. When the recirculation was assigned as a series with 8 different values, the average measured time for calculating the individual simulations was 1.20 seconds.

[0165] Table 1306 shows the time required to perform various further simulations when performing the SLD method: When the pressure generated by the compressor ("Compression") was assigned to a series with only two different values ​​(while the A1 / A3 ratio and the recirculation parameters were assigned to constant average values), the average measurement time for calculating the individual simulations was 6.74 seconds. When Compression was assigned to a series with 4 different values, the average measurement time for calculating the individual simulations was 2.32 seconds. When Compression was assigned to a series with 8 different values, the average measurement time for calculating the individual simulations was 1.41 seconds.

[0166] From the measured times in the three tables it can be inferred that the smaller the step size / the higher the number of different parameter values ​​assigned to the parameters, the faster the individual simulations will be. However, the performance gain of each individual simulation slows down and reaches a plateau.

[0167] A comparison of the simulation times obtained for the different parameters and different numbers of allowed discrete parameter values ​​depicted in Tables 1302-1306 reveals that, for substantially every number of discrete values ​​(e.g., for 2, 4, 6, 8, 16, 32, or 64 discrete values), a single step change in the recirculation ratio has faster performance than a single step change in A1 / A3, and in particular has faster performance than a single step change in the compression parameter.

[0168] For example, assuming that each of the three parameters is assigned a series of 16 discrete values ​​representing the best fit number of different data values, the time for each individual simulation step is 1.06 seconds for the A1 / A3 ratio, which is slower than the 0.86 seconds for recirculation but faster than the 1.22 seconds required for a stepwise change of the compression parameter assuming 16 different values. Therefore, assuming that each of the three parameters will have 8 different values, the order of the parameters that are changed during the simulation will be from left to right: recirculation, A1 / A3 ratio, compression.

[0169] Typically, however, different parameters will have different numbers of data values ​​to be changed. For example, as described above, the most suitable number of different parameter values ​​for the parameter A1 / A3 ratio is 4 because it produces the shortest total time for performing the preliminary simulation: 4×1.81 seconds=7.2 seconds. However, recirculation may have 16 different parameter values ​​assigned to provide sufficient detail during the simulation for that parameter. Compressor may have 4 different values ​​assigned because 4×2.32 seconds=9.28 seconds provides the shortest time for performing all preliminary simulations described in table 1306. Therefore, Fig.13 The situation in the embodiment depicted is:

[0170]

[0171] As shown in the table above, the SLD method is used to identify an appropriate number of different values ​​for each parameter. Then, the order of the parameters to be changed is determined so that the execution time of a single preliminary simulation for the number of SLD-method-determined values ​​is shorter, the earlier the values ​​of the various parameters are changed. Thus, in the depicted embodiment, recirculation will be changed first, then the A1 / A3 ratio, and then the compression.

[0172] It is often the case that the number of different, "allowed" data values ​​for each parameter may not be the same. For example, in the last simulation, recirculation may be assigned 16 different parameter values ​​that must be used as boundary parameter values, A1 / A3 ratio may be assigned 4 (or for example 8) parameter values, and compression may be assigned 4 different parameter values.

[0173] The result of selecting recirculation as the first parameter, the A1 / A2 ratio as the second parameter, and compression as the third parameter can be shown as follows, assuming that each parameter has only two allowed values, referred to herein as V1 and V2:

[0174] Simulation run Recycling R A1 / A2 ratio (AR) Compression 1 R=RV1 AR=ARV1 C=C1 2 R=RV2 AR=ARV1 C=C1 3 R=RV2 AR=ARV2 C=C1 4 R=RV1 AR=ARV2 C=C1 5 R=RV1 AR=ARV2 C=C2 6 R=RV2 AR=ARV2 C=C2 7 R=RV2 AR=ARV1 C=C2 8 R=RV1 AR=ARV1 C=C2

[0175] Thus, the order of the parameters may ensure that the "fastest" parameter, ie, recirculation, is changed most often, while the "slowest" parameter, ie, compression, is changed least often during a simulation run.

[0176] In more general terms, the simulation steps for computing the digital twin may be executed such that the first three boundary condition parameters whose values ​​are to be changed increase or decrease from one step to another according to the following scheme, where "BCP" is "boundary condition parameter" and V1, V2 are respective values, assuming that each parameter is assigned a series of two different values:

[0177] #simulation BCP1 BCP2 BCP3 1 BCP1=BCP1V1 BCP2=BCP2V1 BCP3=BCP3V1 2 BCP1=BCP1V2 BCP2=BCP2V1 BCP3=BCP3V1 3 BCP1=BCP1V2 BCP2=BCP2V2 BCP3=BCP3V1 4 R BCP1=BCP1V1 BCP2=BCP2V2 BCP3=BCP3V1 5 BCP1=BCP1V1 BCP2=BCP2V2 BCP3=BCP3V2 6 BCP1=BCP1V2 BCP2=BCP2V2 BCP3=BCP3V2 7 BCP1=BCP1V2 BCP2=BCP2V1 BCP3=BCP3V2 8 BCP1=BCP1V1 BCP2=BCP2V1 BCP3=BCP3V2

[0178] For simplicity, the value changes during 8 consecutive simulation runs shown in the table above are based on only three parameters that are each assigned a series of only two different values. In practice, the number of parameters and the number of parameter-specific data values ​​can be much larger.

[0179] Fig.14 Shown is a line graph 1400 illustrating the observed average simulation time for a single simulation run when performing the SLD-method assuming 2, 4, 8, 16, 32 and 64 different allowed values ​​for each parameter. The x-axis has a logarithmic scale. From this line graph it can also be inferred that, in addition to the number of 8 parameter values ​​per parameter, the recirculation parameter performs the fastest, while stepwise changes in the compression parameter take the longest time.

[0180] Fig.15 A flow chart of a computer-implemented method for computing a digital twin of a chemical process or a digital twin of a device or system of a chemical plant is shown. Figure 1 The steps of the method are described.

[0181] Embodiments of the present invention may have the advantage of minimizing CPU and memory consumption required to perform a large number of simulations. For example, by changing only a single parameter value from one simulation to the next, while keeping other boundary condition parameters constant, the number of parameter values ​​that must be read from storage into main memory is reduced. Preferably, a single parameter value is changed so that the value is increased or decreased by only a single increment from one simulation to the next, so that no jumps are performed.

[0182] Preferably, the specific manner in which the values ​​of individual boundary parameters are changed (with respect to the order in which the parameters are first changed in value and / or with respect to the increments by which the parameters are changed) is selected so that processing time is reduced. The selection of the order of parameters and the size of the increments ("step sizes") can be based on empirical measurements of the time required to perform preliminary simulations, thereby taking into account the characteristics of the computer system used to perform the simulations (preferably, the preliminary simulations and the final simulations can be performed on the same or similar computer systems). Furthermore, because only a single boundary parameter value is changed at a time in a highly specific manner, the system of equations can be solved more quickly as the equations converge faster.

[0183] Due to the significant performance gain, digital twins can be created that contain a large number of data points, thereby providing a very detailed and therefore highly accurate digital representation of a chemical process or facility. There are several technical devices for digital twins, which consist of hundreds of thousands or even millions of data points, each of which has multiple boundary condition parameter values ​​assigned to it.

[0184] For example, a set of data points generated and provided as a digital twin can be used in full or at least in part to simulate, control, or design a chemical process or equipment or system of a chemical plant. For example, a simulation can be performed during the design phase of a facility to create or optimize the design of a plant or chemical facility, such as a facility for separating gases or other types of materials. The data points can be used directly, such as by executing an algorithm (such as a rule engine or other type of predefined algorithm) that uses the set of data points or a subset thereof as input. Alternatively, at least some of the data points obtained from the simulation step are used as a training data set for training a predictive model using a machine learning approach. In a subsequent step, the trained predictive model is used to simulate, control, or design a chemical process or equipment or system of a chemical plant.

[0185] Reference numerals

[0186] 1 Boundary Condition 19 Second Boundary Condition

[0187] 2 Computational Model 20 Digital Twin

[0188] 3 Initial conditions 21 Data point set

[0189] 4 Parameter value 22 Color code

[0190] 5 Parameter table data 23 Simulation space

[0191] 6 Tabulation database 24 First path

[0192] 7 columns 25 border area

[0193] 8 Process Simulation Software 26 Adjustment Path

[0194] 9 Graphical User Interface 27 Speed

[0195] 10 First column 28 Torque

[0196] 11 First row 29 Second path

[0197] 12 First parameter value 30 Broken line

[0198] 13 Second column 32 data file

[0199] 14 Second parameter value 33 3D-line graph

[0200] 15 Column 34 Artificial Intelligence Model

[0201] 16 Third parameter value 35 Digital slider

[0202] 17 Fourth column 36 Gray box model

[0203] 18 First Boundary Condition 37 Neural Network Model

[0204] 38 filtered 3D-line graphs 51 input units

[0205] 39 pressure 52 storage unit

[0206] 40CO 2 Concentration 53 processor

[0207] 41CH 4 Purity 54 external source

[0208] 42 Central Area 55 Computer

[0209] 43 straight lines 102-108 parameter value grid

[0210] 44 First duration 110 Line graph

[0211] 45 First simulation file 1202 minimum

[0212] 46 Second duration 1204 Maximum value

[0213] 47 Second simulation file 1206 parameter table

[0214] 48 Third duration 1208-1210 Table with parameter value combinations

[0215] 49 The third simulation file 1302-1306 has a table of observed simulation times

[0216] 50 Time Shift 1400 Line Chart

Claims

1. A computer-implemented method for generating a digital twin (20) of a chemical process or device or system for a chemical plant, in particular for substance synthesis and / or substance separation, the method comprising the following steps: a) inputting a plurality of boundary conditions (1) into a calculation model (2), wherein the calculation model (2) comprises a system of equations, wherein each boundary condition (1) comprises a plurality of parameter values ​​(4), b) inputting initial conditions for solving the system of equations, c) in a simulation step, solving the system of equations using each of the boundary conditions (1) to provide a corresponding simulation result, wherein a single one of the parameter values ​​(4) of each boundary condition (1) changes from one simulation step to the next, while the other parameter values ​​(4) belonging to the respective boundary condition remain unchanged, and d) providing the digital twin (20) as a set of data points (21) obtained from the simulation step.

2. A computer-implemented method according to claim 1, wherein the single one of the parameter values ​​(4) of the respective boundary conditions (1) changes from a starting data point to an ending data point in a simulation within its entire data point value set, wherein in a subsequent series of simulations, the single one of the parameter values ​​(4) of the respective boundary conditions changes backward, i.e., changes from the ending data point to the starting data point.

3. A computer-implemented method according to claim 1 or 2, wherein simulation results from one simulation step are used as initial conditions for the next simulation step (3).

4. The computer-implemented method of any of the preceding claims, further comprising displaying the digital twin.

5. The computer-implemented method of claim 4, wherein displaying the digital twin comprises displaying the digital twin via an augmented reality display system.

6. A computer-implemented method according to claim 4 or 5, wherein displaying the digital twin includes generating 2D or 3D computer graphics that visualize data in the digital twin; loading the 2D or 3D graphics into a frame buffer of a display system; and displaying the contents of the frame buffer on the display system.

7. A computer-implemented method according to one of the preceding claims, wherein the plurality of boundary conditions (1) are obtained from a simulation space (23), wherein the boundary conditions (1) for a first simulation step are selected from a central area of ​​the simulation space (23).

8. A computer-implemented method according to one of the preceding claims, wherein the path (24, 29) for selecting the boundary conditions (1) for the next simulation step is dynamically adjusted through the simulation space (23) based on the time required to obtain simulation results in the previous simulation step.

9. The computer-implemented method of any of the preceding claims, wherein the system of equations is a system of steady-state equations for a quasi-steady-state process.

10. The computer-implemented method according to one of the preceding claims, wherein the digital twin (20) is a database containing a matrix of a plurality of simulation results corresponding to the respective boundary conditions.

11. A computer-implemented method according to one of the preceding claims, wherein the chemical process corresponds to a chemical process of a gas separation membrane or a gas separation device having a plurality of gas separation membranes, or wherein the device or the system of the chemical plant is a gas separation membrane or a gas separation device having a plurality of gas separation membranes.

12. A computer-implemented method according to any of the preceding claims, wherein in the simulation step a plurality of systems of equations are solved in parallel, each system of equations using a respective boundary condition (1) to provide a respective simulation result.

13. A computer-implemented method according to claim 12, wherein a time offset (50) is used to start solving each of the sets of equations in parallel relative to solving one of the sets of equations, wherein the time offset (50) is a predetermined fraction of the average time of the simulation steps.

14. The computer-implemented method of claim 13, wherein the predetermined fraction is between 5% and 10% of the average time of the simulation steps.

15. A computer-implemented method according to one of the preceding claims, wherein the digital twin (20) is used to train an artificial intelligence model (34) or a model based on machine learning.

16. The computer-implemented method of any of the preceding claims, wherein the digital twin (20) is a set of data points (21) comprising a plurality of simulation results corresponding to the respective boundary conditions.

17. A computer-implemented method according to any one of the preceding claims, wherein the simulation steps are performed such that from one step to another no parameter value increases or decreases by more than one predefined parameter-dependent increment.

18. A computer-implemented method according to one of the preceding claims, further comprising determining a series of parameter values ​​for each of the boundary condition parameters, which are assigned to the respective boundary condition parameters and change successively during the simulation steps, whereby when the value of the boundary condition changes from one simulation step to the next, one of the data values ​​in the series that immediately precedes or follows the previously used value of the boundary condition parameter is used, whereby the series of parameter values ​​is determined specifically for each of the boundary condition parameters so that the processing time for performing the simulation on all parameter values ​​contained in the series is minimized.

19. The computer-implemented method of claim 18, wherein determining the series of parameter values ​​for each of the boundary condition parameters comprises: A simulation step size determination method-SLD method is executed, wherein the SLD method includes, for each of the boundary condition parameters: a) assigning to one of the boundary condition parameters a first series of different parameter values ​​lying within given minimum and maximum values ​​assigned to the parameter, and assigning to the other boundary condition parameter an average value of the parameter range defined by the respectively assigned minimum and maximum values; b) for each parameter value in the first series, performing a preliminary simulation of the design and / or dynamic behavior of the chemical process or device or system of the chemical plant using the set of equations, whereby for other boundary condition parameters, the preliminary simulation uses the respectively assigned average values ​​as input, whereby while performing the simulation, measuring the time required to perform the preliminary simulation for the parameter values ​​of the first series; c) assigning to said one of said boundary condition parameters a second series of different parameter values ​​within a given minimum and maximum value assigned to that parameter, said second series of different parameter values ​​having more or fewer parameter values ​​than any series of data values ​​previously assigned to said parameter; d) for each parameter value in the second series, performing a preliminary simulation of the design and / or dynamic behavior of the chemical process or device or system of the chemical plant using the system of equations, whereby for other boundary condition parameters, the preliminary simulation uses the respectively assigned average values ​​as input; e) repeating steps c) and d) until a termination criterion is reached, such as a maximum number of parameter values ​​contained in the series; f) analyzing the total time required to execute all preliminary simulations for each series of parameter values ​​to identify one of the series of data values ​​corresponding to the shortest total execution time; and g) using parameter values ​​of the identified series as parameter values ​​to be successively assigned to the respective boundary condition parameters during simulation steps performed to generate the digital twin, wherein the difference between two consecutive parameter values ​​in the identified series defines the simulation step size when a single one of the parameter values ​​of the boundary condition parameter changes from one simulation step to the next.

20. The computer-implemented method according to any one of the preceding claims, wherein the changing of a single one of said parameter values ​​of the respective boundary conditions from one simulation step to the next is performed according to the sequence of the boundary condition parameters, wherein the simulation is performed such that all values ​​in a series of parameter values ​​assigned to the first parameter in the sequence must be traversed from a minimum value to a maximum value and backwards before any value in the value of a subsequent parameter according to the sequence is changed; The method comprises: An order in which boundary condition parameters are to be changed is identified so that the simulation time is minimized.

21. The computer-implemented method of claim 20, wherein each of the boundary parameters is assigned a sequence of parameter values, wherein identification of the sequence comprises: performing a plurality of preliminary simulations using the system of equations, wherein in each preliminary simulation, only the value of one of the boundary condition parameters is changed while the other parameters are assigned constant values, the constant values ​​preferably representing an average of a specific range of values ​​for the assigned parameters, thereby measuring the time required to perform the preliminary simulations; identifying, for each of the boundary parameters, a total time required to perform all preliminary simulations required to go through all values ​​included in the series of data values ​​assigned to the parameter; The sequence is identified such that the shorter the total time required, the higher the priority of the individual parameters within the sequence.

22. The computer-implemented method of any preceding claim, further comprising: using at least some of the data points (21) obtained from the simulation step to simulate, control or design the chemical process or equipment or system of the chemical plant; or Using at least some of the data points (21) obtained from the simulation step to train a prediction model using a machine learning method, and using the trained prediction model to simulate, control or design the chemical process or equipment or system of the chemical plant.

23. A computer-implemented method according to any one of the preceding claims, wherein the number of simulation steps performed and the number of data points comprising the results of a respective one of said simulation steps is at least 100,000, in particular at least 250,000, in particular at least 500,000, in particular at least 1,000,000; and / or wherein the number of boundary condition parameters whose values ​​are changed in the simulation step is at least 3, in particular at least 5, in particular at least 7, in particular at least 9; and / or The average number of predetermined values ​​assigned to one of the boundary condition parameters that changes during the simulation step is at least 2, in particular at least 4, in particular at least 8, in particular at least 64, such as at least in particular at least 256.

24. Computer program comprising instructions which, when executed by a computer (55), cause the computer (55) to perform the steps of the computer-implemented method according to one of the preceding claims.

25. A computer (55) for generating a digital twin (20) of a chemical process or device or system for a chemical plant, in particular for substance synthesis and / or substance separation, configured to use the steps of a computer-implemented method according to one of claims 1 to 23.

26. A display system comprising: Electronic displays; A computer (55) according to claim 25, configured to generate the digital twin, the computer further configured to generate 2D or 3D computer graphics that visualize data in the digital twin; and display the 2D or 3D computer graphics on the electronic display.

27. A data structure configured to visualize a digital twin (20) of a chemical process or equipment or system of a chemical plant, the data structure comprising a plurality of data points obtained by a method according to one of the preceding claims 1 to 23, the data structure being configured to, when processed by a display system, cause the display system to generate 2D or 3D computer graphics that visualize the data in the digital twin; and to display the 2D or 3D computer graphics on an electronic display of the display system.

Images