Method for generating and / or adapting a geometric meta-model for a molded part to be produced in a primary forming process, in particular for an injection molded part to be produced in an injection molding process

Abstract

The invention relates to a method for generating and / or adapting a geometric meta-model for a molded part to be produced in a primary forming process for use as simulation geometry, based on process data, wherein in a learning process step a good part is produced, the good part process is characterized and to determine a metamodel in a geometry determination step at least one of the learned characteristics is iteratively approximated in simulations with intermediate geometries until a sufficiently accurate metamodel is achieved, wherein a virtual structure is used which is at least a geometric rule structure or is composed of one.

Description

[0001] The invention relates to a method for generating and / or adapting a geometric meta-model for a molded part to be produced in a primary molding process, in particular for an injection-molded part to be produced in an injection-molding process, wherein first a learning process step is carried out and after the learning process step a determination of the geometric meta-model is carried out.

[0002] In the area of ​​networking injection molding simulation tools, or the results calculated from them, with injection molding machines, it is generally known to use simulation data to perform machine setup, for example, by setting target values ​​and / or parameters. It is also known to compare simulation data with real-life tests on the injection molding machine.

[0003] Furthermore, software products are known that are used to configure the injection molding process, i.e., the actual injection molding process, based on manually performed simulations. The basis for the state-of-the-art simulations is previously known spatial data, for example, CAD data, of the component to be manufactured and / or spatial data, for example, CAD data, of an injection mold cavity.

[0004] For example, EP 3 291 959 B1 describes the determination and display of process parameter values ​​in an injection molding process. This involves characterizing pressure curves from the simulation and a real process.

[0005] From EP 3 892 440 A1 a method for adjusting an injection molding machine is known, which is also based on the provision and / or selection of at least one 3D model of at least one geometry of at least one molded part to be molded and / or at least one geometry of at least one mold cavity.

[0006] DE 10 2021 109 851 A1 discloses a method and a computer program product for comparing a simulation with the actual process. For this purpose, at least one simulation curve of a characteristic variable for the process, in particular a simulated pressure curve, is calculated within the simulation. During the actual process, at least one measurement curve of the characteristic variable, in particular a measured pressure curve, is measured, and a correlation is established between the simulation process and the actual process based on characteristic points of the simulation curve and the measurement curve.

[0007] From AT 513.481 B1 a method for simulating an injection molding process is known, wherein in a machine simulation for simulating the injection molding machine, optionally without an injection molding tool, at least one first parameter is calculated in an injection unit of an injection molding machine to be simulated on the basis of a physical behavior model, and in a process simulation for simulating the injection molding material to be processed and / or an injection molding tool, at least one second parameter, preferably a pressure, is calculated in the injection molding material to be simulated on the basis of a physical model, wherein the at least one second parameter is communicated to the machine control and / or the at least one first parameter is communicated to the process simulation.

[0008] The above-mentioned prior art has one significant common disadvantage. Performing the simulation based on the CAD data of the component to be manufactured or on the basis of the CAD data of the mold cavity requires, firstly, that such data is known. Furthermore, such a simulation is associated with a high computational effort, since, particularly in the case of complex components or complex, for example, branched mold cavities, a considerable amount of computational effort is required to perform the simulation, e.g., the flow simulation. This computational effort is particularly time-consuming. In general, therefore, the simulation of the injection molding process based on the previously known CAD data of the component to be manufactured or on the basis of the previously known CAD data of the mold cavity is only used for the initial setup of a process.However, this also requires that the person who is supposed to carry out this process simulation must also have knowledge of the mentioned CAD data, since it is impossible to set up a meaningful simulation process without knowledge of this data.

[0009] This in turn results in the disadvantage that, due to the need for at least a realistic 3D model as a starting point, complicated and time-consuming, especially manual, data handling results.

[0010] Particularly in the case of more complex geometries or more complex simulation models, a high level of manual processing effort with regard to initial model optimization and the like is unavoidable.

[0011] The object of the invention is therefore to automate the creation of physical models from real process data in order to enable simulation without knowledge of the real geometry of the component and / or the mold cavity. A further object is to significantly simplify the complexity of simulation-related calculations of primary forming processes, in particular injection molding processes, both qualitatively and / or quantitatively. Furthermore, the object of the invention is to enable the determination of machine setting data / process parameters without knowledge of the CAD data or the geometric three-dimensional shape of the molded part to be produced and / or the CAD data or the three-dimensional shape of the cavity in the mold. Furthermore, it is desirable to significantly reduce the manual effort required for simulation and data management.In particular, the manual meshing effort of a simulation geometry, as currently required in state-of-the-art injection molding simulations, is to be significantly reduced.

[0012] The above-mentioned objects are achieved according to the invention by a method having the features of claim 1. Advantageous embodiments are specified in the claims dependent on claim 1.

[0013] The invention is based on the core idea of ​​using findings from a learning process step to determine a geometric meta-model on the basis of these findings, resulting in a substitute geometry in the form of a geometric meta-model that is greatly simplified compared to the real component or the real tool cavity. This geometric meta-model is created without knowledge of the real spatial shape or the CAD data of the real component and / or the tool cavity. A substitute geometry is determined from greatly simplified geometries, namely from geometric rule structures, solely on the basis of the at least one learning geometry variable acquired in the learning process step, at least one learning process variable acquired in the learning process step, and at least one learning design target variable acquired in the learning process step. This substitute geometry makes it possible to determine a behavior orto simulate a process behavior, particularly with regard to suitable characteristics, sufficiently closely as the real process or the real geometry would do in reality.

[0014] This makes it possible to create a geometric meta-model without knowledge of the actual geometry of the molded part to be produced or without knowledge of the actual geometry of the mold cavity. This geometric meta-model is significantly simplified compared to the real geometries or the simulation geometries in the learning process step, yet still enables a reliable, optimized process setup of the primary forming process, especially the injection molding process. The geometric meta-model is particularly suitable as a simulation geometry for conducting a process simulation of the primary forming process, especially the injection molding process.

[0015] For the understanding of the present invention, the terms listed below are defined as follows, where appropriate, in addition to or alternatively to a relevant technical term definition. If the definition provided here overlaps with or even contradicts a possibly existing technical term definition, the definition as provided in this application shall prevail in case of doubt. Geometric rule structure:

[0016] According to the invention, a geometric rule structure is to be understood as a 2-dimensional or a 2.5-dimensional or a 3-dimensional non-free-form structure.

[0017] A 2-dimensional geometric rule structure is a 2-dimensional polytope or a plane projection of a 2.5-dimensional or a 3-dimensional geometric rule structure.

[0018] A 2.5-dimensional geometric control structure is understood to be a 2-dimensional control structure with at least partially constant thickness information in a third dimension.

[0019] A 3-dimensional geometric regular structure is understood to be a 3-dimensional non-free-form structure, whereby in particular, but not exclusively, the following basic geometric bodies are to be regarded as 3-dimensional regular structures: - 3-dimensional polytopes, e.g. - Cuboid - Dice - Prism - Pyramid - 3- or multi-corner plate - truncated pyramid and / or - 3-dimensional bodies with at least one curved boundary surface, e.g. - cylinder, - Bullet - Cone - Cylindrical disc with a height h and a diameter d, where h <d gilt - partial sphere, especially hemisphere - truncated cone. Virtual entity:

[0020] According to the invention, a virtual entity is understood to be the result of a Boolean addition or a Boolean subtraction of at least two geometric rule entities. In extreme cases, a virtual entity can also be formed from only a single geometric rule entity. Injection phase:

[0021] The period from the start of the melt filling process to complete volumetric mold filling, at which the mold filling process is complete except for the necessary material transport to compensate for shrinkage. The start of the melt filling process approximately coincides with the closing of the non-return valve. Reprint phase:

[0022] The period from the switchover point until the end of pressurization to compensate for material shrinkage. While material is transported into the mold for this purpose, this shrinkage compensation is no longer counted as part of the injection phase. Molding process:

[0023] The period during which melt enters the cavity. Typically, the molding process covers a period from the beginning of the injection phase to the end of the holding pressure phase.

[0024] In detail, the method according to the invention for generating and / or adapting a geometric meta-model for a molded part to be produced in a primary molding process for use as simulation geometry comprises the following steps: a) in a learning process step: a.1) Producing a good part of the molded part in a real good part manufacturing process and / or carrying out a simulation of the good part manufacturing process, a.2) Recording at least one learning process variable and at least one learning geometry variable and at least one learning design target variable of the good part manufacturing process and / or the simulation of the good part manufacturing process b) in one step to determine the geometric meta-model: b.1) Definition of at least one process MUST criterion corresponding to the learning process size and definition of at least one geometry MUST criterion corresponding to the learning geometry size b.2) Configuration of a simulation environment comprising at least: b.2.1) a geometry-based simulation component containing at least one basic geometry, b.2.2) a process-based simulation component, b.2.3) a design-based simulation component of a test plan design, which defines at least one simulation optimization objective to be achieved in the simulation, wherein the simulation optimization objective is formed from the learning design objective or is the learning design objective; b.3) Carrying out the simulation using the simulation environment configured in step b.2 to obtain a simulation result; b.4) Evaluation of the simulation result by comparing the simulation result with the defined simulation optimization objective; b.5) ​​if the simulation optimization goal is not achieved in step b.4), b.5.1) Iterative adaptation of the base geometry to obtain an intermediate geometry; b.5.2) Modification of the geometry-based simulation component from step b.2.1) by replacing the base geometry with the intermediate geometry; b.6) Repeating steps b.3) to b.5) using the intermediate geometry until the specified simulation optimization objective is achieved with the intermediate geometry; b.7) When the specified simulation optimization goal is achieved, definition of the intermediate geometry as the geometric meta-model; b.8) wherein a virtual structure is used for the basic geometry and the intermediate geometry, which is at least one geometric rule structure or which is formed from two or more geometric rule structures.

[0025] In a particular embodiment, the virtual structure is formed or composed of 2-dimensional, 2.5-dimensional and / or 3-dimensional rule structures.

[0026] These represent desirable basic alternatives regarding rule formation types.

[0027] Advantageously, at least one of the following 3-dimensional rule structures is selected from the group: - 3-dimensional polytopes, e.g. - Cuboid - Dice - Prism - Pyramid - 3- or multi-corner plate - truncated pyramid and / or - 3-dimensional bodies with at least one curved boundary surface, e.g. - cylinder, - Bullet - Cone - Cylindrical disc with a height h and a diameter d, where h <d gilt - partial sphere, especially hemisphere - Truncated cone used.

[0028] These relatively simple 3-dimensional geometric structures, which are mathematically representable, enable particularly low computational effort in simulation calculations based on geometric metamodels formed from these 3-dimensional geometric structures. In contrast to, for example, freeform surfaces or freeform bodies, which can only be determined numerically, such as bodies containing spherical surfaces or oscillating surfaces, the geometric metamodel formed from such structures significantly simplifies simulation calculations.

[0029] Advantageously, a 2-dimensional rule structure is understood to be a 2-dimensional polytope or a planar projection of a 3-dimensional rule structure.

[0030] The use of only 2-dimensional control structures also makes it possible to reduce the simulation computation effort and thus the time required to calculate a simulation based on the geometric meta-model created from it.

[0031] Furthermore, a 2.5-dimensional control structure is to be understood as a 2-dimensional control structure together with at least partially constant thickness information relating to a third dimension.

[0032] Such a 2.5-dimensional control structure can advantageously help to reduce the computational effort and thus the time required for a simulation, especially compared to the use of 3-dimensional control structures.

[0033] It is particularly expedient that the virtual structure and / or at least one of the geometric control structures is designed to be parameterizable in the direction of at least one spatial axis (X-axis and / or Y-axis and / or Z-axis), and in step b.5.1) at least one of the size values ​​of the spatial axes (X-axis and / or Y-axis and / or Z-axis) is changed for the iterative adaptation.

[0034] With the above measure, in order to optimize the basic geometry or the intermediate geometry towards the optimized geometric meta-model, it is possible to only vary the spatial shape and / or the size of at least one control structure and thus to iteratively adapt the basic geometry or the intermediate geometry to the desired target geometry only in this way or in conjunction with further measures, ie to obtain the geometric meta-model.

[0035] According to a further embodiment of the method according to the invention, outer edges and / or inner edges of the virtual structure or transition edges between two standard structures of the virtual structure are rounded or have a chamfer.

[0036] By using fillets or chamfers, it is possible in a particularly simple way, for example, to adapt the flow behavior of a material melt within the geometric meta-model to the findings from the learning process step.

[0037] Advantageously, the primary mold manufacturing process is a casting process, e.g., an injection molding process or a die casting process.

[0038] The method according to the invention is particularly suitable for the above primary forming processes.

[0039] According to a further embodiment of the invention, the geometric meta-model with respect to the geometry-based simulation component satisfies at least one geometry MUST criterion within predefined geometry MUST criterion limit values ​​and / or the geometric meta-model behaves with respect to the geometry MUST criterion within the predefined geometry MUST criterion limit values ​​in the same way as the real geometry of the good part or the simulation geometry of the molded part to be produced, ie like the learning geometry variable.

[0040] The properties of the geometric meta-model outlined above enable a particularly accurate and realistic determination of the geometric meta-model in comparison to corresponding properties of the real geometry or the simulation geometry.

[0041] In a further embodiment of the method according to the invention, the geometric meta-model with respect to the geometry-based simulation component fulfills at least one geometry CAN criterion within predefined geometry CAN criterion limits and / or the geometric meta-model behaves with respect to the geometry CAN criterion within the predefined geometry CAN criterion limits in the same way as the real geometry of the good part or the simulation geometry of the molded part to be produced, ie like the learning geometry variable.

[0042] The properties of the geometric meta-model outlined above enable a further, particularly precise determination of the behavior of the geometric meta-model in comparison to that of the real geometry or the simulation geometry.

[0043] It has proven particularly useful that the geometry MUST criterion is a total volume to be filled (V fill,tot )is.

[0044] According to a particular embodiment, the geometry CAN criterion is a relation of dimensions of the rule structures of the geometric meta-model, e.g. L / D (length to diameter) for a cylinder.

[0045] By specifying such a geometry-CAN criterion, melt flow velocities in a geometric meta-model underlying a simulation can, for example, be approximated at least in some areas to actual flow velocities during the good part process or to flow velocities resulting from the simulation of the good part process. Such approximations, at least in some areas, can, in particular, accurately simulate or adapt average flow front velocities of the melt in the meta-model.

[0046] A further embodiment of the invention is characterized in that the geometric meta-model, with respect to the process-based simulation component, fulfills at least one process MUST criterion within predetermined process MUST criterion limits and / or that a process simulated with the meta-model behaves with respect to the process MUST criterion within the predetermined process MUST criterion limits like the real good part manufacturing process or like the simulation of the good part manufacturing process, ie like the learning process variable.

[0047] With the properties of the geometric meta-model outlined above, a particularly accurate and realistic determination of the geometric meta-model with regard to its process variables in comparison to corresponding process variables of the good part manufacturing process or the simulation of the good part manufacturing process is possible.

[0048] A further embodiment of the invention is additionally characterized in that the geometric meta-model, with respect to the process-based simulation component, fulfills at least one process CAN criterion within predetermined process CAN criterion limits and / or that a process simulated with the geometric meta-model behaves with respect to the process CAN criterion within the predetermined process CAN criterion limits like the real good part manufacturing process or like the simulation of the good part manufacturing process, ie like the learning process variable.

[0049] If, in addition to the process MUST criterion or the process MUST criteria, at least one further process CAN criterion is met or the geometric meta-model in a simulation process behaves with regard to the process CAN criterion like the real good part manufacturing process or like the simulation of the good part manufacturing process, the quality and meaningfulness of the geometric meta-model can be further improved as a simplified substitute geometry.

[0050] In a particular embodiment, it has proven useful that the process MUST criterion includes at least one or more selected from the group: - Plasticizing screw stroke curve of the injection phase (s inj ) and / or values ​​calculated therefrom; - Injection pressure curve (ie a course of injection pressure values ​​p inj over a time, e.g. injection time, or a distance, e.g. the screw stroke) or singular injection pressure values ​​(p inj) and / or values ​​calculated therefrom; - Physical material data, e.g. in the form of material models such as pvT and cross-WLF models; - Temperature value (T inj ), which represents a temperature of the material melt, especially near the nozzle, or is / are a hot runner temperature.

[0051] The above-mentioned process MUST criteria, of which at least one must be fulfilled, create a basis for setting up and / or optimizing a simulated process on the basis of the geometric meta-model in terms of essential process data in accordance with the real good part process or with the simulation of a real good part process, in particular within specified process MUST criteria limit values.

[0052] In further embodiments of the invention, the process CAN criterion is one or more selected from the group: - Nozzle pressure curve (p noz) and / or values ​​calculated therefrom; - Volume flow curve of the injection phase (V̇ inj ) and / or values ​​calculated therefrom; - cavity pressure curve (p Cav ) and / or values ​​calculated therefrom; - Closing force curve (F CIp ) and / or values ​​calculated therefrom; - Current curve (I Drv ) and / or values ​​calculated from them.

[0053] Such process CAN criteria make it possible to meaningfully select a geometric meta-model within a simulation with regard to realistic process properties of the good part manufacturing process or the process properties of the simulation of the good part manufacturing process. The latter current curve, in particular, can provide easily measurable information about pressure conditions within the screw antechamber and / or within the mold cavity. The current curve, for example, is the temporal progression of the electrical current consumption of a drive motor for a hydraulic unit that advances the plasticizing screw in a plastics injection molding machine. Similar statements are also possible based on knowledge of the temporal progression of the clamping force required to keep a mold closed.

[0054] Alternatively or additionally, it may be expedient for the geometric meta-model to satisfy at least one design MUST criterion within predefined design MUST criterion limits with respect to the design-based simulation component and / or for a process simulated with the meta-model to behave like the real good part manufacturing process or like the simulation of the good part manufacturing process, ie like the learning design target variable, with respect to the design MUST criterion within the predefined design MUST criterion limits.

[0055] A design must-have criterion is a process criterion toward which the simulation process is optimized using the baseline geometry / intermediate geometry. It is therefore a criterion of a test plan design that should be within the narrowest possible design must-have criterion limits.

[0056] An example of the design MUST criterion is the mold filling time (t fill ), which corresponds, for example, to the duration of the injection phase, under a boundary condition of the total volume to be filled (V fill,tot ).

[0057] The above design MUST criterion gives, under the boundary condition of the total volume to be filled (V fill,tot ), which is component-specific and / or cavity-specific, represents an easy-to-measure and easily achievable optimization target.

[0058] In addition to a design MUST criterion, the geometric meta-model can expediently fulfill at least one design CAN criterion within predefined design CAN criterion limits with regard to the design-based simulation component and / or the geometric meta-model can be characterized in that a process simulated with the meta-model behaves like the real good part manufacturing process or like the simulation of the good part manufacturing process, ie like the learning design target variable, with regard to the design CAN criterion within the predefined design CAN limit values.

[0059] As a design CAN criterion, at least one or more has been selected from the group: - Solidification time (t solid ); - Volume flow curve and derived values, e.g. - maximum volume flow (V̇ max ), average volume flow (V̇ avg ) has proven itself.

[0060] With regard to the iterative optimization to achieve the geometric meta-model, it has proven useful that the iterative adaptation of the base geometry or the intermediate geometry is carried out by changing a geometric extension of one or more of the geometric rule structures in one or more of the spatial axes (x-axis; y-axis; z-axis).

[0061] With such a parameterization of the spatial extents of at least one, preferably several of the rule structures from which the basic geometry, the intermediate geometry or the geometric meta-model is formed, preferably with the aid of Boolean addition and / or Boolean subtraction, adaptations of the basic geometry and / or the intermediate geometry to achieve the geometric meta-model are possible in a particularly simple manner.

[0062] Alternatively or cumulatively to the above-mentioned parameterization of individual rule structures, it may be particularly useful to achieve the optimization goal for the iterative adaptation of the base geometry or the intermediate geometry to be carried out by Boolean addition of another rule structure to the existing base geometry or to the existing intermediate geometry.

[0063] This measure makes it possible to achieve an improved approximation of the geometric meta-model to the corresponding properties of the good part in a particularly simple manner and, above all, without unduly increasing the complexity of the geometric meta-model. Furthermore, characteristic process variables of the good part manufacturing process or the simulation of the good part manufacturing process can be easily adapted by adding rule structures to an existing base geometry or an existing intermediate geometry.

[0064] Alternatively or cumulatively to the above-mentioned parameterization of individual rule structures or the above-mentioned Boolean addition of rule structures, it may be particularly useful to achieve the optimization goal for the iterative adaptation of the base geometry or the intermediate geometry to be carried out by Boolean subtraction of another rule structure from the existing base geometry or from the existing intermediate geometry.

[0065] This measure makes it possible to achieve an improved approximation of the geometric meta-model to the corresponding properties of the good part in a particularly simple manner and, above all, without unduly increasing the complexity of the geometric meta-model. Characteristic process variables of the good part manufacturing process or the simulation of the good part manufacturing process can also be easily adapted by removing standard structures from an existing base geometry or an existing intermediate geometry.

[0066] It may also be appropriate for the iterative adaptation of the base geometry or the intermediate geometry to be carried out by exchanging a rule structure of the existing base geometry or the existing intermediate geometry with another rule structure.

[0067] This represents another cumulative or alternative possibility for meaningful adaptation or generation of the geometric meta-model.

[0068] Advantageously, it is also an aim of the method according to the invention that a characteristic tool system response of the geometric meta-model generates the same tool system response as the real good part manufacturing process or as the simulation of the good part manufacturing process within predetermined tool system response limits.

[0069] This approach assumes that in a simulation using the geometric meta-model, a characteristic tool system response in response to predetermined process sequences, such as an injection pressure curve, is a particularly suitable characteristic, which leads to a calculated process simulation based on the geometric meta-model corresponding to a high degree to the good part manufacturing process or the simulation of the good part manufacturing process despite significantly lower and therefore less time-consuming calculation effort, and thus a suitable replacement geometry for further simulation calculations is found in the form of the resulting geometric meta-model.

[0070] Furthermore, it may be advantageous that a total volume of the geometric meta-model corresponds to a total volume of the good part, in particular including ancillary structures or the volume used in the simulation of the good part process.

[0071] Such an approach simplifies the determination of the total filling volume of a cavity without having to separately know the total volume of the actual molded component or the ancillary structures.

[0072] It is particularly expedient that, from a plurality of intermediate geometries which satisfy both the geometry MUST criterion and the process MUST criterion as well as the design MUST criterion, that intermediate geometry is selected as the meta-model which, with respect to at least one process CAN criterion, comes closest to the learning process variable and / or with respect to at least one geometry CAN criterion to the learning geometry variable and / or with respect to at least one design CAN criterion.

[0073] This measure can ensure that a simulation based on the determined geometric meta-model achieves a particularly high level of accuracy.

[0074] Furthermore, it may be advantageous to determine the total volume to be filled (V fill,tot) taking into account a melt compression and / or taking into account an elasticity of a plasticizing screw and / or taking into account a closing behavior and / or a leakage behavior of a non-return valve.

[0075] Such a measure also increases the accuracy of the simulation calculation based on the geometric meta-model thus determined as a substitute geometry, since material-specific idealized assumptions as well as machine-side idealized assumptions are replaced by corresponding real material behavior and / or real machine behavior. Regarding the implementation of these measures, reference is made to the technical teachings of European patents EP3359365B1, EP3870422B1, and EP3055116B1, merely as examples.

[0076] It is particularly easy to carry out if the tool system response is a pressure curve p(s) which results from an identification speed signal v(s) of the plasticising screw with which the tool is filled.

[0077] This allows the measurement effort, both in terms of equipment and computational effort, i.e. the computational processing of measurement data, to be significantly reduced.

[0078] As an alternative to the tool system response defined above, e.g. a form of a pressure curve, the tool system response can also be a speed curve of the plasticizing screw v(s), which results from an identification pressure signal p(s) with which the tool is filled.

[0079] This also allows the cost of both the instrumental and computational measurement to be significantly reduced.

[0080] It is particularly useful that the test plan design includes, in addition to the simulation optimization objective, at least information on an iteration step size for the iterative adaptation in step b.5.1) and / or upper and lower limit values ​​for the iterative adaptation in step b.5.1).

[0081] By appropriately adjusting the iteration step size within the experimental design, the computational and / or iterative effort required to obtain a geometric meta-model within the required limits can be reduced and / or even optimized. This is achieved, in particular, by reducing the number of required iteration steps.

[0082] It is advisable that a mesh is assigned to the base geometry, the intermediate geometry and the geometric meta-model within the framework of a meshing and / or that an inlet is assigned to the base geometry, the intermediate geometry and the geometric meta-model.

[0083] To make both the base geometry and intermediate geometries, as well as the geometric meta-model, available for simulation calculations, it is advisable to perform meshing and define an inlet that will simulate an injection point. These are common preparatory measures for making a geometric structure accessible to simulation calculations, i.e., creating a simulation-capable geometry.

[0084] To minimize the computational effort required to obtain a simulation-capable base geometry, a simulation-capable intermediate geometry, or a simulation-capable geometric meta-model, the inlet remains unchanged, at least with regard to its geometry, during the iterative adaptation from the base geometry through the intermediate geometries to the geometric meta-model. The shape and / or dimensions of the inlet can be appropriately adapted to the control structure downstream in the melt flow direction.

[0085] The invention is explained in more detail below using the drawings as examples. They show: Fig. 1: schematically on the left side of the figure a view of a real component, namely a holder for wrenches, which can be replaced by a highly simplified replacement geometry in the form of a meta-model (right side of the figure) which was obtained using the method according to the invention; Fig. 2: schematically a flow diagram of the method according to the invention; Fig. 3: a detailed representation of the process step “Configuration of the simulation environment” from Fig. 2; Fig. 4 schematically shows a flow chart for parameter definition, for example of radius and length of a three-dimensional control structure, for example a cylinder; Fig. 5: schematically a flow diagram of a sub-process for determining the cylinder radius from Fig. 4; Fig. 6: schematically a flow diagram of a sub-process for determining the cylinder length from Fig. 4; Fig. 7: schematically a flow chart for obtaining a design of experiments (=DoE: Design of Experiments).

[0086] Fig.1 shows a comparison of a good part 1 with the real geometry of a component produced in a primary forming process. Knowing the real geometry of the good part 1, the prior art involves very time-consuming and complex data processing, such as meshing, to create a model that is accessible for further process simulation. Creating a model based on the real geometry in this way is complex and computationally intensive, and the invention aims to avoid this, or at least greatly simplify it. For this purpose, the invention specifies how a substitute geometry, such as a geometric meta-model 2, can be determined. This substitute geometry, such as a geometric meta-model 2, can be determined. This substitute geometry is geometrically and mathematically much simpler and less complex than the good part 1, yet nevertheless offers a good simulation geometry for an almost realistic tool response in process simulations based on the geometric meta-model 2 determined according to the invention.Due to the significantly simplified complexity of the geometric meta-model 2, simulations to be carried out or calculated with it are significantly less time-consuming and require considerably less computational effort.

[0087] The present invention therefore enables the performance of flow simulations for a primary forming process, in particular for an injection molding process, without knowing the real three-dimensional component geometry or cavity geometry typically required for such a simulation. On the basis of process and setting data of a machine that, for example, carries out a good part manufacturing process, a geometric substitute model, i.e. a geometric meta-model, is derived, which can be used in a flow simulation instead of the more complex original geometry. If such a geometric substitute model fulfills certain minimum process requirements described in more detail below, this geometric substitute model is referred to as geometric meta-model 2 in the context of this invention.

[0088] Calculation results based on such a geometric meta-model 2 are comparable, in particular with regard to mold filling times, injection pressures and volume flows, to those that would have been calculated using the original geometry. The core idea of ​​the invention is therefore to design a geometric meta-model 2 based on, among other things, process data which, despite considerably reduced geometric complexity compared to the good part 1 or the real geometry, i.e. despite greatly reduced computing and data handling effort, can offer reliable statements within the framework of a simulation with which a manufacturing process of the good part or the real geometry can be simulated with sufficient approximation. These basic ideas are fundamentally applicable to all primary forming processes, but in particular to an injection molding process and / or a die casting process, such as an aluminum die casting process.

[0089] The method according to the invention for obtaining such a replacement geometry in the form of a geometric meta-model 2 is described in terms of the process flow in Fig. 2. The method according to the invention is essentially divided into two process steps, namely a learning process step 3 and a step 4 for determining the geometric meta-model.

[0090] Learning process step 3 can fundamentally be performed alternatively or cumulatively in two different ways. Firstly, it is possible to perform a good part manufacturing process on a machine 5 to produce a good part and to determine process parameters or process signals 6 from this good part manufacturing process.

[0091] Alternatively or additionally, a simulation 7 of the good part manufacturing process can also be performed based on the real geometry. Such a simulated good part manufacturing process results in so-called simulated process signals 8 or simulated process parameters, which are determined by simulating the good part manufacturing process. At the end of the learning process step 3, real good part manufacturing process data or simulated good part manufacturing process data are thus available, on the basis of which, as will be described below, the geometrically highly simplified meta-model 2 according to the invention is determined.

[0092] In learning process step 3, therefore, at least one learning process variable and at least one learning geometry variable as well as at least one learning design target variable of the good part manufacturing process and / or the simulation thereof are recorded or determined.

[0093] These, particularly high-resolution, process and setting data are recorded and stored via a data acquisition system and made available for further processing.

[0094] The data supplied, in particular the scope of the data supplied, includes, for example, measured variables that are recorded directly on the machine during the good part manufacturing process or that are calculated in the good part simulation. Alternatively or cumulatively, modeling can run on the machine, in the cloud, or on a local computer alongside the data acquisition. In addition to this data, data acquired using so-called soft sensors can be used, for example, if the soft sensors are of sufficient quality and process-specific requirements regarding resolution, accuracy, etc. are met.

[0095] At the beginning of step 4 for determining the geometric meta-model 2, i.e., in the second essential phase of the method according to the invention, at least one process MUST criterion corresponding to the learning process variable and at least one geometry MUST criterion corresponding to the learning geometry variable are defined. The learning process variable and the learning geometry variable are known from learning process step 3. These two MUST criteria must be met by the newly created replacement geometry, namely ultimately by the geometric meta-model 2, within reasonable MUST criterion limits, so that a simulation based on the geometric meta-model can be meaningful for a real or simulated process of the good part or the real geometry. Geometry must-have criteria / geometry can-have criteria and boundary conditions (examples) type criterion Formula symbols Description Geometry Basic geometry [-] necessary initial definition of a geometric form of the basic geometry Mesh constraint Network configuration [-] Boundary condition that is configured differently depending on the simulation environment used. The geometric meta-model contains this information in the form of "metadata," i.e., in the form of accompanying data so that the geometric meta-model can be used for further simulations. Inlet constraint Inlet configuration [-] Boundary condition that describes all relevant factors of the inlet, e.g. position, diameter, etc. The geometric meta-model contains this information in the form of "Meta-data", i.e. in the form of accompanying data, so that the geometric meta-model can be used for further simulations. Geometry Volume to be filled V fill,tot MUST criterion that is initially used for generation Constraint Volumetric dependence [-] CAN criterion that describes a geometric dependency (e.g. L / D for cylinders)

[0096] At the beginning of step 4, data preprocessing (step S105) also takes place to determine the geometric meta-model 2. Data from the data preprocessing (step S105) is either fed directly to a configuration step (S106) or virtual sensor signals can be determined and / or calculated from the data preprocessing in a calculation step S107, whereby the virtual sensor signals thus determined are assigned to a corresponding virtual sensor, which does not have to be physically present but results from measured values. This step is Fig. 2 is referred to as step S108.

[0097] In this simulation configuration step, settings for the geometry to be simulated, the process to be simulated, and the scope of a simulated test design are specified. Details will be provided in the discussion of Fig.3 will be explained below. The definition of the above-mentioned settings is also referred to below as a geometry-based work step, a process-based work step, and a design-based work step.

[0098] In the geometry-based work step, the geometry to be used in the simulation is defined and, if necessary, parameterized, meshed, and provided with an inlet I that matches its dimensions, shape, and position on the geometry. The geometry obtained in the geometry-based work step is a replacement geometry for the good part 1 and represents the beginning of an iteration loop, so this replacement geometry is referred to as the base geometry.

[0099] The process-based settings depend on the process being simulated. For example, the process-based settings required for an injection molding simulation differ from those required for a die casting simulation. Process MUST criteria, process CAN criteria and boundary conditions (examples) type criterion Formula symbols Description Feed / data input Injection pressure p inj MUST criterion: Injection pressure of the injection phase; injection pressure is e.g. input signal at the inlet Feed / data input Physical material data [-] MUST-HAVE criterion: Physical material data, e.g. in the form of material models such as pvT and Cross-WLF models; Configures the parametric models associated with the material used (e.g. Cross-WLF) as well as other properties that the material contains, e.g. pvT properties Feed / data input Temperature value T inj MUST-.Criterion: the temperature value configures e.g. a melt temperature and, if applicable, a hot runner temperature alternatively: a temperature of the temperature zone closest to the molded part near the nozzle, if e.g. no hot runner is available Constraint / boundary condition Plasticizing screw stroke s inj MUST-HAVE criterion: Plasticizing screw stroke of the injection phase; the screw stroke is a calculated value dependent on the screw diameter in order to be able to evaluate the introduced volume or the filling process. Constraint / boundary condition Volume flow V̇ inj CAN criterion: Volume flow of the injection phase Constraint / boundary condition Nozzle pressure p noz CAN criterion: Pressure of the injection phase in the screw antechamber / near the nozzle Constraint / boundary condition cavity pressure P cav CAN criterion: Pressure of the injection phase inside the tool Constraint / boundary condition Closing force F Clp CAN criterion: Closing force of the injection phase Constraint / boundary condition Electricity I Drv CAN criterion: Current consumption of the engine during the injection phase, e.g. as a measure of an injection pressure

[0100] This data is used, for example, to configure the simulation environment.

[0101] The final configuration step of the simulation environment involves design-based configuration. This configuration of a test plan contains variables that specify the optimization goal, such as the learning design target variable determined in learning process step 3, as well as the variables to be varied to achieve the optimization goal. These variables to be varied are test plan-related variables, whose changes support an approximation of the optimization goal. Furthermore, boundary conditions to be considered can be defined to filter the simulation results or define the limits of an optimization problem.

[0102] Following the configuration of the simulation environment, as performed in step S106, the simulation of the experimental design is performed (step S109). Results from the simulation of the experimental design, i.e., results from step S109, are evaluated in an evaluation step S110 with respect to the achievement of the desired optimization goal, i.e., with respect to the achievement of the desired learning design target. Whether the desired learning design target has been achieved, or at least within technically reasonable tolerances, is determined in a decision step S111.

[0103] If the first test plan does not produce a satisfactory result, the geometry-based configuration step is repeated (step S112) with iterative adjustment of the base geometry, resulting in an intermediate geometry. The intermediate geometry is used to simulate the previous or a new test plan, i.e., step S109 is repeated. Design MUST criteria, design CAN criteria and boundary conditions (examples) type criterion Formula symbols Description Design variable Dimensions of the geometry *geometry dependent Parameters that can be varied depending on the chosen geometric shape. Optimization Objective Component filling time t fill MUST criterion that causes an end to the geometric adaptation. Constraint / boundary condition Volume to be filled V fill,tot MUST criterion that must remain constant despite geometric adaptation. Constraint / boundary condition Solidification time T solid CAN criterion that describes solidification at selected positions. Constraint / boundary condition Max. achieved volume flow V̇ max CAN criterion that describes a maximum volume flow / flow velocity. Constraint / boundary condition Average volume flow V̇ avg CAN criterion that describes an average volume flow / flow velocity.

[0104] The scope of the geometric adaptation or geometric iteration is defined by a separate algorithm and can add new geometric rule bodies to the base geometry or the intermediate geometry as well as subtract them from an existing base geometry or an existing intermediate geometry. Furthermore, it is possible to adapt the degrees of freedom present in the geometry while the shape of the base geometry or the intermediate geometry remains fundamentally unchanged. As an example, the shape of a cuboid is generally retained, but the cuboid can be varied in its length, width, or height. The term “parameterization” is used below for such an approach. The geometry used in the simulation, which depends on the simulation environment used, can be parameterized in all three spatial dimensions if necessary in order to achieve the desired optimization goal, i.e.the learning design target. After achieving the optimization goal, the final parameters are determined in addition to the simulation calculation results, with which the geometry used in the simulation can be uniquely described. Thus, if an iteratively obtained geometry achieves certain tolerances, it can be considered a suitable geometry for the simulation and is thus referred to within the scope of the invention as a geometric meta-model, which is the goal of the inventive method (end: S113).

[0105] The following will be based on Fig. 3 the configuration step S106 for configuring the simulation geometry is explained in more detail.

[0106] Within the framework of the geometry-based settings (S320), a base geometry is first generated or selected (S321). According to the invention, this base geometry is at least one geometrically controlled entity, i.e., a basic body that is geometrically and mathematically easily graspable and easily describable, and which, in particular, does not contain any freeform surfaces, such as oscillating surfaces or oscillating surfaces.

[0107] Freeform surfaces are, for example, surfaces that can only be described numerically and / or consist of point clouds, since they cannot be described by a simple, formulaic relationship, such as a sphere or a cuboid or a cube or the like.

[0108] Such freeform surfaces are, for example, scanned strain surfaces or osculation surfaces on CAD models and / or real geometries.

[0109] Subsequently, dependencies on geometry parameters are configured (S322). Such dependencies can be, for example, but not exclusively, a length-to-diameter (L / D) ratio of a cylinder, a length-to-width ratio of a cuboid, a width-to-height ratio of a cuboid, or a length-to-height ratio of a cuboid. Furthermore, the geometric control bodies can be parameterized, for example, based on individual or multiple degrees of freedom of the geometric control bodies, for example, lengths, widths, heights, angles, diameters, radii, or the like.

[0110] In a final step S323, the geometry thus obtained is meshed and provided with the inlet I to be usable for a simulation calculation. Thus, a base geometry, which is the first substitute geometry, or an intermediate geometry, which is a substitute geometry with at least one iteration stage, for example, steps S109, S110, S111, and S112, is meshed and provided with the inlet I.

[0111] Within the scope of the process-based settings (S330), essential process settings of the primary forming process, in particular the injection molding process, are configured in a step S331.

[0112] As part of the design-based settings (S340), variables of the experimental design are first configured in step S341. Subsequently, boundary conditions that must be met in all cases with regard to the experimental design are configured (step S342). Furthermore, the optimization goal toward which geometric meta-model 2 is to be optimized is configured in step S343.

[0113] After that, step S106 of configuring the simulation environment is completed.

[0114] The following is based on the Fig. Figure 4 explains the basic process that takes place in step S112, namely the geometry-based configuration or geometry-based adaptation of the substitute geometry (base geometry or intermediate geometry). This is done using the example of a base geometry / intermediate geometry in the form of a geometric rule structure of a cylinder with length I and radius r.

[0115] First, in a step S400, it is decided whether a real radius r real is known. If this is the case, the known real radius r real as cylinder radius for the simulation r cyl,sim set (S401). If the radius r real is not known, the cylinder radius for the simulation is r cyl,sim to a cylinder radius r cyl set (step S402).

[0116] Regarding the length l of the cylinder, the procedure is analogous to the radius r of the cylinder. First, in step S403, it is checked whether a real length l real is known. If this is the case, the real length l real as cylinder length for the simulation l cyl,sim set (S404). If the real length l real not known, the cylinder length for the simulation is l cyl,sim to a cylinder length l cyl set (S405).

[0117] Starting values, step sizes of an iteration as well as limit values ​​for the adaptations explained above, namely the radius adaptation and the length adaptation using the example of a cylinder, can basically be freely selected.

[0118] With the help of the Fig. 5, step S402, namely setting / calculating the cylinder radius r cyl,sim described for the simulation if the real radius r real is not known.

[0119] After starting this process (S500), a query is first made as to whether the existing run is a first run (step S501).

[0120] If it is the first pass, for example, in a step S502 the cylinder radius r cyl set to 2 mm. Furthermore, in step S503, a bit lim high , which indicates that the upper limit has been reached, and a corresponding bit lim low, which indicates that the lower limit has been reached, is set to 0. The first run is now complete. You can return to the start of this process S500.

[0121] In a further run, in step S506, it is first checked whether the upper limit of the cylinder radius r cyl is reached, ie whether lim high = true. If this is the case, a query is made in step S507 as to whether the lower limit of the cylinder radius r cyl is reached, ie whether lim low = true. If this is also the case, the radius r cyl retained and the subroutine is terminated. If lim low is not set to true, a step S508 first queries whether a lower limit of the adaptation r min is reached. If this is not the case, the radius r cyl in a step S508' by the step size h cyl,r the cylinder radius adjustment is reduced.

[0122] If the lower limit of the adaptation r min reached, the bit lim low , which indicates that the lower limit has been reached, is set to 1 (=true) in step S511. With this new value of the bit lim low =1, which indicates the lower limit for r cyl , the query of step S507 is returned to.

[0123] If the upper limit of the cylinder radius r cyl is not reached, ie if lim high is not set to true, a step S509 first queries whether an upper limit of the adaptation r max is reached. If this is not the case, the radius r cyl in a step S510 by the step size h cyl,r the cylinder radius adjustment increased.

[0124] If the upper limit of the adaptation r max reached, the bit lim high, which indicates that the upper limit has been reached, is set to 1 (=true) in step S511. With this new value of the bit lim high =1, which indicates the upper limit for r cyl , the query of step S506 is returned to.

[0125] An analogous procedure is used to determine the length of the cylinder l cyl,sim applied if this length l cyl,sim is not known. This is determined below using the Fig. 6 will be explained.

[0126] After starting this process (S600), a query is first made as to whether the existing run is a first run (step S601).

[0127] If it is the first pass, for example, in a step S602 the cylinder length l cyl set to 50 mm. Furthermore, in step S603, a bit lim high, which indicates that the upper limit has been reached, and a corresponding bit lim low , which indicates that the lower limit has been reached, is set to 0. The first run is now complete. You can return to the start of this process (S600).

[0128] In a further run, in step S606, it is first queried whether the upper limit of the cylinder length l cyl is reached, ie whether lim high = true. If this is the case, a query is made in step S607 as to whether the lower limit of the cylinder length l cyl is reached, ie whether lim low = true. If this is also the case, the cylinder length l cyl retained and the subroutine is terminated. If lim low is not set to true, a step S608 first queries whether a lower limit of the adaptation I min is reached. If this is not the case, the cylinder length lcyl in a step S608' by the step size h cyl,l the cylinder length adjustment is reduced.

[0129] If the lower limit of adaptation I min reached, the bit lim low , which indicates that the lower limit has been reached, is set to 1 (=true) in step S611. With this new value of the bit lim low =1, which indicates the lower limit for l cyl , the query of step S607 is returned to.

[0130] If the upper limit of the cylinder length l cyl is not reached, ie if lim high is not set to true, a step S609 first queries whether an upper limit of the adaptation I max is reached. If this is not the case, the cylinder length l cyl in a step S610 by the step size h cyl,I the cylinder length adjustment increased.

[0131] If the upper limit of adaptation I max reached, the bit lim high , which indicates that the upper limit has been reached, is set to 1 (=true) in step S611. With this new value of the bit lim high =1, which indicates that the upper limit for l cyl , the query of step S606 is returned to.

[0132] In the following, the Fig. 7 explains how a design of experiments (DoE) is carried out within the scope of the invention.

[0133] Unless explicitly stated otherwise, the order in which the following process steps are described is not intended to limit the timing of the process steps. Unless explicitly stated otherwise, the individual steps may take place or be performed in a different order or simultaneously.

[0134] In a material data supply step S701, for example, physical data on the material used, in particular the material melt, a process temperature T Prozess and / or a tool temperature T Werkzeug provided.

[0135] In a supply step S702 for high-resolution data, for example, high-resolution individual values ​​and / or high-resolution curves of an injection pressure p inj , a filling time t fill , a total filling volume V fill, tot and a filling volume flow V̇ inj provided.

[0136] A process configuration is determined from the data of step S701 and step S702 in a step S703.

[0137] Furthermore, in a step S704, a geometry data feed is carried out, wherein geometric data relating to the basic geometry or the intermediate geometry, geometric limit values ​​and geometric boundary conditions are provided.

[0138] The geometry data input is supplemented by a mesh configuration step S705 to prepare for the data processing required for performing a simulation. Parallel to this, before or after this, an injection point configuration (inlet definition) is performed in a step S706.

[0139] A test plan is created in step S707 from the process configuration obtained in step S703, as well as from the geometry data (S704), which includes a mesh configuration (S705) and an injection point configuration (S706). This then results in the desired test plan design (DoE) (step S708). Information on a possible upper limit on the number and / or size of the control structures used:

[0140] Such an upper limit on the number of control entities used is essentially component-dependent and cannot be specified or determined across the board. The construction of the geometric meta-model 2 via the initial or intermediate stages of the base geometry and the intermediate geometries takes place by parameterizing existing control entities or by Boolean addition to the virtual entity or Boolean subtraction of control entities from the virtual entity until the desired optimization goal, i.e., the learning design target, is achieved within technically reasonable tolerances.

[0141] The size of individual rule structures can also vary, both in the base geometry and in the intermediate geometries as well as in the finished geometric meta-model 2. For example, a specific component to be replicated can be created by using a finite number of rule structures, each with a volume of 0.1 cm 3 up to 1 cm 3or even larger volumes. The volumetric sizes of the individual regular structures within a base geometry or an intermediate geometry or the finished geometric meta-model 2 can be the same or different from each other, even for qualitatively similar regular structures such as cuboids, spheres, cones, or the like.

[0142] A scaling and a specific use of qualitatively different control structures of different basic forms is specified by the application case itself and the representable state variables of the geometric meta-model 2.

[0143] A possible upper limit for the number of control entities to be used can be used, for example, by setting an upper limit for the simulation time required to simulate such a geometric meta-model. It is, of course, desirable to accomplish the additive construction of a geometric meta-model with the simplest possible form and the minimum necessary number of control entities, so that the specified optimization goal, i.e., the learning design target variable, for example, the filling time, is achieved before the required simulation time corresponds to that of the original geometry. The aim of the invention is to reduce the simulation time and thus the computational effort compared to that of the original geometry. Further embodiments of the invention:

[0144] Of course, it is possible that the process of geometric adaptation can be shortened and / or optimized through the use of artificial intelligence or expert knowledge. Information on how to conduct the simulations:

[0145] The simulations to be performed within the scope of the invention on the basis of a base geometry, an intermediate geometry, or on the basis of the geometric meta-model 2 can be carried out, for example, within commercially available simulation applications. However, it should be emphasized that the inventive method for generating the geometric meta-model 2 can be applied independently of the simulation application used. One advantage of using a commercial simulation application is that important steps for a filling simulation, such as generating a mesh, defining the inlet I, and the numerical calculation methods with which the necessary differential equation systems can be solved, are already available and have been tested.In addition, commercial simulation applications usually have a material database to include material-specific processes or data, such as viscosity properties, in the simulation. On the topic of “networking” of the basic geometry, the intermediate geometries and the geometric meta-model 2:

[0146] The term "meshing" in the context of the invention means that a corresponding mesh is placed on the respective geometry in preparation for a simulation in the usual way, on the basis of which the simulation calculations are then carried out. Of course, an adapted mesh will also have to be determined for each geometry adaptation, since this represents the respectively valid geometry, i.e. the base geometry, one or more of the intermediate geometries or the finished geometric meta-model 2. However, due to the geometry of the geometric meta-model 2, which has been greatly reduced in complexity according to the invention, the meshing effort as well as the meshing effort of all geometric intermediate stages (base geometry and intermediate geometry) is considerably reduced, which significantly simplifies the effort required to prepare a simulation and saves computing time. On the topic of “setting an inlet” on the base geometry, the intermediate geometries and the finished geometric meta-model 2:

[0147] Within the scope of the invention, an injection point (inlet I) is expediently also defined on the respective existing geometry, i.e. the basic geometry, the intermediate geometries and the finished geometric meta-model 2 with regard to its size, shape and position on the model. Of course, it may be necessary to reduce the inlet I, which was initially assumed to be of a certain size for a basic geometry, if, for example, an end face of a standard structure, e.g. a cylinder, becomes smaller during adaptation than an area of ​​the originally assigned inlet, or to enlarge it if necessary. In general, it is of course absolutely necessary to adapt the inlet I appropriately. List of reference symbols 1 good part, real geometry 2 geometric meta-model 3 learning process step 4 Step to determine the geometric meta-model 5 machine 6 process signals 7 Simulation 8 simulated process signals 10 virtual entities 11 virtual rule structure S105 to S113 steps S320 to S323 steps S330 to S331 steps S340 to S343 steps S400 to S405 steps S500 to S512 steps S600 to S612 steps S701 to S708 steps I Inlet QUOTES CONTAINED IN THE DESCRIPTION

[0000] This list of documents submitted by the applicant was generated automatically and is included solely for the convenience of the reader. This list is not part of the German patent or utility model application. The DPMA assumes no liability for any errors or omissions. Cited patent literature

[0000] EP 3 291 959 B1

[0004] EP 3 892 440 A1

[0005] DE 10 2021 109 851 A1

[0006] AT 513.481 B1

[0007] EP 3359365B1

[0075] EP 3870422B1

[0075] EP 3055116B1

[0075]

Claims

[1] Method for generating and / or adapting a geometric meta-model (2) for a molded part to be produced in a primary molding process for use as simulation geometry, on the basis of process data, comprising the steps: a) in a learning process step (3): a.1) Producing a good part (1) in a real good part manufacturing process and / or carrying out a simulation of the good part manufacturing process, a.2) Recording at least one learning process variable and at least one learning geometry variable and at least one learning design target variable of the good part manufacturing process and / or the simulation of the good part manufacturing process b) in a step (4) to determine the geometric meta-model (2): b.1) Definition of at least one process MUST criterion corresponding to the learning process size and definition of at least one geometry MUST criterion corresponding to the learning geometry size b.2) Configuration of a simulation environment comprising at least: b.2.1) a geometry-based simulation component containing at least one basic geometry, b.2.2) a process-based simulation component, b.2.3) a design-based simulation component of a test plan design, which defines at least one simulation optimization objective to be achieved in the simulation, wherein the simulation optimization objective is formed from the learning design objective or is the learning design objective; b.3) Carrying out the simulation using the simulation environment configured in step b.2 to obtain a simulation result; b.4) Evaluation of the simulation result by comparing the simulation result with the defined simulation optimization objective; b.5) ​​if the simulation optimization goal is not achieved in step b.4), b.5.1) Iterative adaptation of the base geometry to obtain an intermediate geometry; b.5.2) Modification of the geometry-based simulation component from step b.2.1) by replacing the base geometry with the intermediate geometry; b.6) Repeating steps b.3) to b.5) using the intermediate geometry until the specified simulation optimization objective is achieved with the intermediate geometry; b.7) When the specified simulation optimization goal is achieved, definition of the intermediate geometry as the geometric meta-model (2); b.8) wherein a virtual structure (10) is used for the basic geometry and the intermediate geometry, which is at least one geometric rule structure (11) or which is formed from two or more geometric rule structures (11). [2] Method according to claim 1, characterized bythat the virtual structure (10) is formed or composed of 2-dimensional, 2.5-dimensional and / or 3-dimensional control structures (11). [3] Method according to claim 1 or 2, characterized by that as a 3-dimensional control structure (11) at least one is selected from the group: - 3-dimensional polytopes, e.g. - Cuboid - Dice - Prism - Pyramid - 3- or multi-corner plate - truncated pyramid and / or - 3-dimensional bodies with at least one curved boundary surface, e.g. - cylinder, - Bullet - Cone - Cylindrical disc with a height h and a diameter d, where h <d gilt - partial sphere, especially hemisphere - truncated cone is / are used. [4] Method according to claim 2, characterized bythat a 2-dimensional polytope or a plane projection of a 3-dimensional rule structure is used as the 2-dimensional rule structure(11). [5] Method according to claim 4, characterized by that a 2-dimensional control structure (11) is used as the 2.5-dimensional control structure (11) together with at least sectionally constant thickness information relating to a third dimension. [6] Method according to one of the preceding claims, characterized by that the virtual structure (10) and / or at least one of the control structures (11) is designed to be parameterizable in the direction of at least one spatial axis (X-axis and / or Y-axis and / or Z-axis), and in step b.5.1) at least one of the size values ​​of the spatial axes (X-axis and / or Y-axis and / or Z-axis) is changed for the iterative adaptation. [7] Method according to one of the preceding claims, characterized bythat outer edges and / or inner edges of the virtual structure (10) or transition edges between two standard structures (11) of the virtual structure (10) are rounded or have a chamfer. [8] Method according to one of the preceding claims, characterized by that the primary mold manufacturing process is a casting process, e.g. an injection molding process or a die casting process. [9] Method according to one of the preceding claims, characterized by that the geometric meta-model (2) with respect to the geometry-based simulation component satisfies at least one geometry-MUST criterion within predefined geometry-MUST criteria limits and / or the geometric meta-model (2) behaves with respect to the geometry-MUST criterion within the predefined geometry-MUST criteria limits in the same way as the real geometry of the good part (1) or the simulation geometry of the molded part to be produced, ie as the learning geometry size [10] Method according to one of the preceding claims, characterized by that the geometric meta-model (2) with respect to the geometry-based simulation component fulfills at least one geometry CAN criterion within predefined geometry CAN criteria limit values ​​and / or the geometric meta-model (2) with respect to the geometry CAN criterion within the predefined geometry CAN criteria limit values ​​behaves like the real geometry of the good part (1) or the simulation geometry of the molded part to be produced, ie like the learning geometry variable. [11] Method according to claim 9, characterized by that the geometry MUST criterion is a total volume to be filled (V fill,tot )is. [12] Method according to claim 10, characterized by that the geometry CAN criterion is a relation of dimensions of the rule structures (11) of the geometric meta-model, e.g. L / D (length to diameter) for a cylinder. [13] Method according to one or more of the preceding claims, characterized by that the geometric meta-model (2) with respect to the process-based simulation component fulfills at least one process MUST criterion within predefined process MUST criterion limit values ​​and / or that a process simulated with the meta-model (2) behaves with respect to the process MUST criterion within the predefined process MUST criterion limit values ​​like the real good part manufacturing process or like the simulation of the good part manufacturing process, ie like the learning process variable. [14] Method according to one or more of the preceding claims, characterized bythat the geometric meta-model (2) with respect to the process-based simulation component fulfills at least one process CAN criterion within predefined process CAN criterion limit values ​​and / or that a process simulated with the geometric meta-model (2) behaves with respect to the process CAN criterion within the predefined process CAN criterion limit values ​​like the real good part manufacturing process or like the simulation of the good part manufacturing process, ie like the learning process variable. [15] Method according to claim 13, characterized by that the process MUST criterion includes at least one or more selected from the group: - Plasticizing screw stroke curve of the injection phase (s inj ) and / or values ​​calculated therefrom; - Injection pressure curve (p Inj ) or singular injection pressure values ​​(p Inj ) and / or values ​​calculated therefrom; - Physical material data, e.g. in the form of material models such as pvT and cross-WLF models; - Temperature value (T inj ), which represents a temperature of the material melt, especially near the nozzle, or is a hot runner temperature. [16] Method according to claim 14, characterized by that the process CAN criterion is one or more selected from the group: - Nozzle pressure curve (p noz ,) and / or values ​​calculated therefrom; - Volume flow curve of the injection phase (V̇ inj ) and / or values ​​calculated therefrom; - cavity pressure curve (p Cav ) and / or values ​​calculated therefrom; - Closing force curve (F CIp ) and / or values ​​calculated therefrom; - Current curve (I Drv ) and / or values ​​calculated from it. [17] Method according to one or more of the preceding claims, characterized bythat the geometric meta-model (2) with respect to the design-based simulation component fulfills at least one design MUST criterion within predefined design MUST criterion limits and / or that a process simulated with the geometric meta-model (2) behaves with respect to the design MUST criterion within the predefined design MUST criterion limits like the real good part manufacturing process or like the simulation of the good part manufacturing process, ie like the learning design target variable. [18] Method according to claim 17, characterized by that the design MUST criterion is the - Mold component filling time (t fill ) under a boundary condition of the total volume to be filled (V fill,tot )is. [19] Method according to one or more of the preceding claims, characterized bythat the geometric meta-model, with respect to the design-based simulation component, fulfills at least one design CAN criterion within predefined design CAN criterion limits and / or that a process simulated with the geometric meta-model (2) behaves with respect to the design CAN criterion within the predefined design CAN limits like the real good part manufacturing process or like the simulation of the good part manufacturing process, ie like the learning design target variable. [20] Method according to claim 19, characterized by that the design CAN criterion includes at least one or more selected from the group: - Solidification time (t solid ); - Volume flow curve and derived values, e.g. maximum volume flow (V̇ max ), average volume flow (V̇ avg ) is. [21] Method according to one of the preceding claims, characterized bythat the iterative adaptation of the basic geometry or the intermediate geometry is carried out by changing a geometric extension of one or more of the geometric rule structures (11) in one or more of the spatial axes (x-axis; y-axis; z-axis). [22] Method according to one of the preceding claims, characterized by that the iterative adaptation of the base geometry or the intermediate geometry is carried out by Boolean addition of a further rule structure (11) to the existing base geometry or to the existing intermediate geometry. [23] Method according to one of the preceding claims, characterized by that the iterative adaptation of the base geometry or the intermediate geometry is carried out by Boolean subtraction of a further rule structure (11) from the existing base geometry or from the existing intermediate geometry. [24] Method according to one of the preceding claims, characterized bythat the iterative adaptation of the basic geometry or the intermediate geometry is carried out by exchanging a control structure (11) of the existing basic geometry or the existing intermediate geometry with another control structure (11). [25] Method according to one of the preceding claims, characterized by that a characteristic tool system response of the geometric meta-model (2) produces, within predetermined tool system response limits, the same tool system response as the real good part manufacturing process or as the simulation of the good part manufacturing process. [26] Method according to one of the preceding claims, characterized by that a total volume of the geometric meta-model (2) corresponds to a total volume of the good part, in particular including ancillary constructions or the volume used in the simulation of the good part process. [27] Method according to one of the preceding claims, characterized bythat from a plurality of intermediate geometries which satisfy both the geometry MUST criterion and the process MUST criterion and the design MUST criterion, that intermediate geometry is selected as the geometric meta-model (2) which comes closest to the learning process variable with respect to at least one process CAN criterion and / or to the learning geometry variable with respect to at least one geometry CAN criterion and / or to at least one design CAN criterion. [28] Method according to one of the preceding claims, characterized by that a determination of the total volume to be filled (V fill,tot ) taking into account a melt compression and / or taking into account an elasticity of a plasticizing screw and / or taking into account a closing behavior and / or a leakage behavior of a non-return valve. [29] Method according to one of the preceding claims, characterized bythat the tool system response is a pressure curve p(s) resulting from an identification speed signal v(s) of the plasticizing screw with which the tool is filled. [30] Method according to one of the preceding claims, characterized by that the tool system response is a speed curve of the plasticizing screw v(s), which results from an identification pressure signal p(s) with which the tool is filled. [31] Method according to one of the preceding claims, characterized by that the test plan design includes, in addition to the simulation optimization objective, at least information on an iteration step size for the iterative adaptation in step b.5.1) and / or upper and lower limit values ​​for the iterative adaptation in step b.5.1). [32] Method according to one of the preceding claims, characterized bythat a network is assigned to the basic geometry, the intermediate geometry and the geometric meta-model (2) within the framework of a network and / or an inlet I defined in terms of size and / or shape and / or arrangement on the geometry is assigned to the basic geometry, the intermediate geometry and the geometric meta-model (2). [33] Method according to claim 32 characterized by that the inlet I remains unchanged during the iterative adaptation at least with regard to the arrangement of the inlet I on the base geometry, the intermediate geometry and the geometric metamodel.

Images