Method for manufacturing part made of composite material comprising fatigue design
By defining individual damage events in the finite element model using the discrete damage mechanics method, the problem of inaccurate fatigue life prediction of composite laminates in existing technologies is solved, and high-precision damage prediction and design optimization in complex structures are achieved.
Patent Information
- Application Number
- CN202480066573.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Priority Date
- 2023-10-17
- Filing Date
- 2024-10-16
- Publication Date
- 2026-05-15
AI Technical Summary
Existing technologies cannot accurately account for individual damage events such as matrix cracking, delamination, and reinforcement failure when predicting the fatigue life of composite laminates, especially when the geometry is irregular or discontinuous, leading to inaccurate predictions and requiring extensive experimental verification.
By employing the discrete damage mechanics method, the location and size of individual damage events are defined in the finite element model. Combined with fatigue strength parameters and load history, damage prediction is performed, achieving accurate correlation of laminate stiffness. This method is applicable to complex structures.
It enables accurate damage prediction in high stress gradient regions, reduces experimental verification time and cost, and improves the accuracy of fatigue life design.
Smart Images
Figure CN122055596A_ABST
Abstract
Description
[0001] This invention relates to a method for manufacturing components made of composite materials, including a fatigue design step thereof.
[0002] In this document, "parts made of composite materials" refers to articles comprising one or more layers, which are reinforced, for example, with continuous, discontinuous or particulate fibers, and are associated with each other according to a specific overlapping sequence such that the respective reinforcements are oriented in a specific manner relative to a reference direction; the overlapping sequence of layers is referred to as a "laminate".
[0003] In this document, “lay-up” refers to a set of information intended to define the construction of a laminate; such information may include, for example, the material of the matrix of each layer, the material of the reinforcement of each layer, the elastic properties of each layer, the orientation of the reinforcement of each layer relative to a reference direction, the thickness of each layer, and the layer overlap sequence.
[0004] Various types of structural components, such as wind turbine blades, bicycle frames, pressure vessels, vehicle or aircraft parts, can be manufactured using the method according to the present invention.
[0005] When subjected to time-varying loads, composite laminates progressively deteriorate; the gradual evolution of damage causes the stiffness of the laminate to decrease from the first load application cycle, and the eventual failure of the component (separation into two or more parts) may occur after a very high number of load cycles (even millions of cycles).
[0006] It should be noted that major damage events in composite laminates can include matrix cracking, delamination (separation between adjacent layers), and reinforcement failure.
[0007] Currently, there are known methods for manufacturing composite laminates through an initial fatigue design step, which is based on the so-called "damage tolerance" method and uses a so-called "damage evolution" model.
[0008] It should be noted that fatigue design based on the damage tolerance method is performed by allowing damage events to occur, provided that these events do not cause, for example, a reduction in component stiffness below a certain threshold, or affect the component's ability to withstand predetermined operating loads.
[0009] It should also be noted that the damage evolution model is a model that allows prediction of the stiffness changes and fatigue life of composite laminates over time.
[0010] Siemens patent EP2803968B1 describes a method for manufacturing structural elements, for example, from composite materials, including its fatigue design to predict the fatigue life of the structure. The fatigue design comprises several steps. In the first step, the structure is modeled using a series of calculation points. Subsequently, for each point, the stresses and deformations caused by load cycles are determined, and their hysteresis branches are defined. Then, the cumulative damage caused by the load cycles is predicted and recorded. For the prediction, firstly, using a hysteresis operator, the stress variation along a portion of the hysteresis branch is calculated based on the time-varying load, and secondly, the damage variation is calculated using the stress variation and the recorded cumulative damage. Then, changes in properties including structural stiffness are also calculated. Further changes in stress and deformation are then calculated based on these changes in properties to determine newly adjusted hysteresis branches. Then, further changes in stress along other portions of the adjusted hysteresis branches are calculated based on additional time-varying loads.
[0011] At the end of the process, if no fatigue failure occurs, the structure is manufactured using the materials, properties, and dimensions specified by the model; otherwise, the model is adjusted, and the life performance prediction is repeated.
[0012] In EP2803968B1, finite element analysis is used to determine the stress state in a structure (e.g., a composite structure). A series of loads are applied to the structure, and at each iteration, successive damage parameters for each element are calculated based on local stresses. Based on these damage parameters, the material stiffness is updated, and the occurrence of eventual failure is assessed.
[0013] However, the method described in the Siemens patent has some limitations.
[0014] In practice, it predicts the stiffness of laminates without considering the presence of individual damage events (such as matrix cracking, delamination, and reinforcement failure), instead relating them to a single, continuous damage parameter through empirical relationships (continuous damage mechanics). This type of approach first requires costly and extensive experimental activity to obtain the relationship between the damage parameter and the elastic properties of the laminate, which may also vary depending on the laminate stack and the type of load it experiences. Furthermore, the method uses a continuous relationship between the number of cycles and the laminate stiffness, assuming that the damage is uniformly distributed within a given physical domain (e.g., an element or component in finite element analysis). This makes the method unable to accurately predict fatigue damage in structural elements, especially if their geometry is relatively irregular or if the element includes discontinuities (e.g., holes or surface irregularities) that are critical in real-world applications, where even very steep high stress gradients exist, leading to damage localization.
[0015] The primary objective of this invention is to overcome the aforementioned drawbacks, and in particular, to manufacture components made of composite materials based on fatigue life design. This fatigue life design achieves more accurate damage prediction by predicting individual cell damage events as spatially and temporally defined entities (“discrete damage mechanics” approach) and correlating the laminate stiffness with their presence. This method provides accurate predictions even in the presence of even very steep stress gradients. An additional advantage is that, once some basic material properties are known, the behavior of the component can be predicted even when the laminate stack and load conditions vary.
[0016] Within the scope of the stated task, the object of the present invention is to accelerate the design of composite material components optimized to withstand time-varying loads by providing accurate predictions of physical damage evolution, and thereby accelerate their manufacture, further enabling a reduction in the time and cost associated with experimental material characterization and design verification testing.
[0017] One aspect of the present invention relates to a method for manufacturing a composite component having the features of claim 1.
[0018] Other advantageous features of the invention are reported in the dependent claims.
[0019] The features and advantages of the present invention will become more apparent from the following exemplary and non-limiting description with reference to the accompanying schematic diagrams of one embodiment of the invention, wherein:
[0020] - Figure 1 A component with a regular shape and two layers is illustrated schematically as an example;
[0021] - Figure 2 This schematically illustrates, for example, the use of the finite element method from... Figure 1 Discretized models generated from components;
[0022] - Figure 3 schematically shown Figure 2 The discretized model, which highlights the damage initiation site;
[0023] - Figure 4 A block diagram of the method according to the present invention is shown schematically;
[0024] - Figure 5 schematically shown Figure 4 The step (A) indicated in the block diagram;
[0025] - Figure 6 schematically shown Figure 4 The steps indicated in the block diagram (B);
[0026] - Figure 7 schematically shown Figure 4 The steps indicated in the block diagram (C);
[0027] - Figure 8 schematically shown Figure 4 The steps indicated in the block diagram (D);
[0028] - Figure 9 It schematically shows that in Figure 8 The sub-step (D1) indicated in the block diagram;
[0029] - Figure 10 It schematically shows that in Figure 8 The sub-step (D2) indicated in the block diagram;
[0030] - Figure 11 It schematically shows that in Figure 8 The sub-step indicated in the block diagram (D3);
[0031] - Figure 12 It schematically shows that in Figure 8 The sub-step (D4) is indicated in the block diagram.
[0032] The present invention relates to a method for manufacturing a component 1 made of a composite material, the composite material component comprising at least a first layer 2 and, if appropriate, at least a second layer 3, the two being associated with each other.
[0033] Advantageously, the first layer 2 and / or the second layer 3 are reinforced, for example, with continuous, discontinuous, or granular fibers appropriately oriented relative to the reference direction 11. As mentioned above, it should be noted that the method according to the invention can also be used to manufacture structural elements comprising a single layer or alternatively multiple layers.
[0034] In this document, the configuration of the laminate constituting component 1 (also referred to as “laminate”) is defined, for example, by the following: the sequence of one or more layers 2, 3 constituting therein, the orientation of the reinforcement of each layer 2, 3 relative to the reference direction 11, the thickness of each layer 2, 3, the matrix material, the reinforcement material of each layer 2, 3 (e.g., continuous fiber), and the elastic properties of each layer.
[0035] The method according to the invention includes a series of steps 4, through which fatigue design of component 1 is performed.
[0036] The series of steps 4 includes a first step A, in which a discretized model 5 is generated, for example, from the finite element model of component 1, wherein the first (or single) layer 2 and / or the second layer 3 are respectively divided into two or more first parts 6 and two or more second parts 7.
[0037] In an advantageous embodiment of the method according to the present discovery, the virtual three-dimensional model of component 1, which is not shown in the accompanying drawings, can be generated using any commercially available three-dimensional modeling software (CAD).
[0038] In the advantageous embodiment, first information, which aims to describe the overall geometry of component 1, is provided to finite element software (SW-FEM) through a virtual three-dimensional model; then, the virtual three-dimensional model can be discretized according to finite element discretization using the SW-FEM software to obtain a discretized model 5.
[0039] In other advantageous embodiments of the discovered method, finite element modeling can be performed entirely within the finite element software SW-FEM.
[0040] In an advantageous embodiment, in the finite element software SW-FEM, second information related to the stacking of the laminate 1 and third information intended to describe the elastic properties of the first layer 2 and the second layer 3 can be assigned to the discretized model 5. The third information may include, for example, the longitudinal elastic modulus, transverse elastic modulus, Poisson's coefficient, and shear modulus of each of layers 2 and 3.
[0041] Advantageously, first information intended to describe the overall geometry of component 1, fourth information intended to describe the finite element discretization of discretized model 5 (i.e., dividing it into two or more first parts 6 and two or more second parts 7), second information related to the stacking of laminate 1, and third information intended to describe the elastic properties of first layer 2 and second layer 3 are stored in a first database DB1.
[0042] The series of steps 4 also includes a second step B, in which the positions of one or more first damage initiation sites 8 and one or more second damage initiation sites 9 are defined in each of two or more first parts 6 and in each of two or more second parts 7, respectively.
[0043] In an advantageous implementation, using a suitable computational tool SW-CAL, based on fourth information stored in the first data collection DB1 intended to describe the finite element discretization of the discretized model 5, the positions of one or more first damage initiation sites 8 and one or more second damage initiation sites 9 in each of two or more first parts 6 and each of two or more second parts 7 are defined; by way of example, the positions of one or more first damage initiation sites 8 and one or more second damage initiation sites 9 may be located at the same position as the centroid of one or more first sub-partitions 61 and one or more second sub-partitions 71, respectively; their geometry may be determined by the user.
[0044] The locations of one or more first damage initiation sites 8 and one or more second damage initiation sites 9 are advantageously stored in the first database DB1.
[0045] The location can be chosen arbitrarily and can refer to, for example, the initiation of matrix cracks, delamination, or reinforcement failure. If appropriate, pre-existing damage events can also be inserted into the structure.
[0046] The series of steps 4 further includes a third step C, in which fatigue strength parameters are assigned to each of the first damage initiation site 8 and the second damage initiation site 9, for example, based on fifth information designed to describe the fatigue behavior of the material. The fifth information is stored in a second database DB2 interconnected with the first database DB1.
[0047] In an advantageous embodiment, the fifth information may include the relationship between fatigue strength and the number of cycles initiating the damage event (SN curve), the statistical distribution of fatigue strength, the initial size of the damage event, and the damage propagation pattern (e.g., the relationship between the propagation rate and the elastic energy release rate).
[0048] The series of steps 4 also includes a fourth step D, in which the load history to be simulated on the component is established, and the total number of cycles to be simulated NC is divided into discrete cycle blocks, with load conditions assigned to each cycle block. The fewer cycles contained in each block, the more accurate the damage prediction; however, this means a longer analysis time.
[0049] For each loop block, perform the following calculations:
[0050] (D1) Calculate one or more load parameters for each element of the discretized model.
[0051] Then, as needed, the following steps are performed in parallel:
[0052] (D2) For each damage initiation site, the damage initiation parameter PID is calculated using the fatigue damage initiation model M1. If one or more of the parameters at the initiation site reach a predetermined initiation value, a damage event is placed at the initiation site. Furthermore, if the size of the placed damage event exceeds the perimeter of the element where the damage initiation site is located, the excess portion of the damage event is placed in an adjacent element.
[0053] as well as
[0054] (D3) For each damage event spawned in any previous cycle block, the damage propagation control parameter PCPD is calculated using the fatigue damage propagation model M3, and the size of each damage event is subsequently updated. If the size of a damage event after its propagation exceeds the perimeter of the element where the damage event ended before its propagation, the excess portion of the damage event is placed in an adjacent element.
[0055] (D4) Using the appropriate model M4, calculate the updated elastic properties of each part (6, 7) in each layer based on the damage events located in each part (6, 7) of each layer.
[0056] The detailed execution of step D is as follows. The calculation of one or more load parameters for each element in sub-step D1 is performed by a finite element analysis operating on the discretized model 5. This finite element analysis applies load conditions associated with the analyzed cyclic block to the discretized model 5 based on the elastic properties calculated at the end of the previous cyclic block in each of the first layer 2 and the second layer 3 (or the initial properties when calculated for the first cyclic block). The load parameters are stored in a third database DB3.
[0057] Preferably, sub-step D2, based on the load parameters acting on the element where the damage initiation site is located, calculates the damage initiation parameter PID for each of the one or more first damage initiation sites 8 and for each of the one or more second damage initiation sites 9, using an appropriate damage initiation model M1 for fatigue damage and a stress redistribution model M2 in the presence of damage. The damage events to be considered for stress redistribution and PID calculation are those located within the element where the initiation site is located, and, if necessary, also in adjacent elements, up to a distance that can be set by the user, for example, based on laminate characteristics. If the initiation parameter reaches a determined initiation value for one or more of the first damage initiation sites 8, or for one or more of the second damage initiation sites 9, the damage event is placed at the initiation site using information suitable for defining the location of one or more portions of the damage event in one or more portions of the relevant layer.
[0058] In summary, in sub-step D2, all damage events that arise within the analyzed loop block are placed.
[0059] The information was collected in the fourth database, DB4.
[0060] Preferably, in sub-step D3, for each damage event that ignites in the previous cycle block, the damage propagation control parameter PCPD is calculated using the fatigue damage propagation model M3 and the stress redistribution model M2 in the presence of damage. The damage events considered for stress redistribution and PCPD calculations are those within the element whose damage terminates before propagation, and, if necessary, also located in adjacent elements, up to a distance that can be set by the user, for example, based on laminate characteristics. Based on the propagation, the size of each damage event that ignites in the previous cycle block is updated (e.g., the position of its apex in the case of a crack). This information is collected in a fourth database DB4.
[0061] If the size of a damage event after its propagation exceeds the perimeter of the element to which the damage event terminated before its propagation, the excess portion is placed in an adjacent element. For example, the propagated crack may have a vertex whose coordinates are outside the perimeter of the element to which the crack terminated before propagation. In this case, the excess distance beyond the perimeter is reported to the adjacent element.
[0062] Preferably, sub-step D4 calculates the first elastic property and the second updated elastic property (or at the end of the analyzed cycle block) for each of the first layer 2 and (if applicable) the second layer 3, based on the mentioned model M4 (which is used to calculate the apparent elastic properties based on the damage events located therein).
[0063] In practice, the elastic properties of each layer are updated based on the current damage. For this purpose, the model M4, used to calculate the apparent elastic properties of the layers in the laminate, is provided as input. Based on the calculated new values, the elastic properties of the damaged layer for each element are updated in the finite element software.
[0064] Step D specifies that sub-steps D1 through D4 are repeated until an unacceptable situation is reached or the user-specified number of loops NC is reached. Furthermore, in step D, external models M1 through M4 are used as inputs to the method and are known in themselves. For example, Model M1, used for matrix crack initiation and for delamination initiation, is described in the following articles: Carraro PA, Quaresimin M, A damage-based model for crack initiation in unidirectional composites under multiaxial cyclic loading, Composites Science and Technology 2014; 99:154-163 and Maragoni L, Carraro PA, Quaresimin M, A novel method to include crack-induced delamination in a fatigue damage predictive procedure for composite laminates, Composites Science and Technology 2023; 238:110011; Models M2 and M4 are described in the following articles: Carraro PA, Quaresimin M, A stiffness degradation model for cracked multidirectional laminates with cracks in multiple layers, International Journal of Solids and Structures 2015; 58:34-51 and Carraro PA, Maragoni L, Quaresimin M, Stiffness degradation of symmetric laminates with Off-axis cracks and delamination: an analytical model, International Journal of Solids and Structures 2021; 213:50-62; The model M3 can be represented, for example, by the relationship between the rate of damage event propagation and the rate of elastic energy release.
[0065] Finally, step E specifies that component 1 is manufactured when a certain number of load cycles are reached without any specific unacceptable situation occurring.
[0066] For the purposes of this invention, unacceptable circumstances may refer to, for example, the failure of reinforcement in any element of the component, or the reduction of the component stiffness below a certain value, or the occurrence of a certain number of damage events (or even just one).
[0067] During the first calculation cycle, the initial elasticity property is the property of the layer in the absence of damage or in the presence of any pre-existing damage. In subsequent cycles following the first calculation cycle, the initial elasticity property is equal to the updated elasticity property calculated during the sub-step (D4) of the previous cycle.
[0068] Therefore, it has been found that the method according to the present invention solves the above-mentioned tasks and objectives.
[0069] In practice, this method tracks the location and size of individual damage events over time, allowing them to interact between one element and another (both in terms of the initiation and expansion of damage events).
[0070] This enables more accurate damage prediction and allows for better handling of areas with high stress gradients, which is crucial in real-world applications (e.g., considering holes).
[0071] The region is characterized by a significant degree of damage localization, which cannot be covered by continuous damage methods.
[0072] Compared to discrete damage methods applicable to complex structures, this method allows numerical stress calculation tools to interact with analytical models (M1 to M4) used to predict discrete damage events. This method manages the stress analysis of the structure (retained within the numerical model) and damage evolution (whose information is contained in an appropriate database) in separate environments. However, these environments remain tightly interconnected because models M1 to M4 use the results of the numerical analysis and are provided with updated local elastic properties based on the current damage. This separate management enables a significant reduction in computation time compared to currently available discrete damage prediction methods for complex structures.
Claims
1. A method for manufacturing a component (1) made of composite material, said component (1) comprising one or more layers (2, 3) associated with each other. Its features are, The method involves a series of steps (4) for fatigue design of the component (1), including: (A) First step, in which a discretized model (5) is generated, for example, by means of the finite element model of the component (1), wherein at least the first layer (2) and at least the second layer (3) as needed are divided into two or more first parts (6) and two or more second parts (7), respectively. (B) Second step, in which the positions of one or more first damage initiation sites (8) and one or more second damage initiation sites (9) are defined in each of the two or more first portions (6) and each of the two or more second portions (7), respectively. (C) Third step, in which fatigue strength parameters are assigned to each damage initiation site (8, 9). (D) Fourth step, in which a load history to be simulated on the component is established based on specific load conditions, and the total number of cycles (NC) to be simulated is divided into discrete cycle blocks, and load conditions are assigned to each cycle block. For each loop block, perform the following sub-steps: (D1) Calculate one or more load parameters for each element of the discretized model, as the resulting forces and moments. Then, the following steps are performed in parallel as needed: (D2) For each damage initiation site, the damage initiation parameter PID is calculated using the fatigue damage initiation model (M1). If the parameter reaches a predetermined initiation value for one or more initiation sites, a damage event is placed at the damage initiation site. If the size of the placed damage event exceeds the perimeter of the element where the damage initiation site is located, the excess portion of the damage event is placed in an adjacent element. as well as (D3) For each damage event that originates in any previous cycle block, the damage propagation control parameter (PCPD) is calculated using the fatigue damage propagation model (M3), and the size of the damage event is subsequently updated. If the size of the damage event after its propagation exceeds the perimeter of the element to which the damage event terminated before its propagation, the excess portion is placed in the adjacent element. (D4) Based on the damage events located in each part (6, 7) of each layer, calculate the updated elastic properties of each part (6, 7) using an appropriate model (M4). (E) When a specific number of cycles (NC) is reached without any specific unacceptable situation occurring, the component (1) is manufactured.
2. The method according to claim 1, wherein, By performing analysis, for example using the finite element method, to calculate one or more load parameters for each element in substep (D1), the analysis applies load conditions associated with the analyzed loop block to the discretized model (5) using the elastic properties calculated for each part of the layer at the end of the previous loop block (or the initial properties when the calculation is performed on the first loop block).
3. The method according to claim 1, wherein, The first step (A) and the sub-step (D1) are performed using the first finite element software, and the second step (B), the third step (C), and the sub-steps (D2), (D3), and (D4) are performed using computational tools different from those used in the structural solution steps performed using the first finite element software.
4. The method according to one or more of the preceding claims, wherein, The calculation of the damage initiation parameter (PID) in substep (D2) is performed using an appropriate fatigue damage initiation model (M1) and a stress redistribution model (M2) in the presence of damage events; the damage events to be considered for the stress redistribution and the PID calculation are damage events located within the element where the damage initiation site is located, and, if applicable, in adjacent elements, up to a distance that can be set by the user, for example, based on the laminate characteristics.
5. The method according to one or more of the preceding claims, wherein, The calculation of the damage propagation control parameter (PCPD) in substep (D3) is performed using a fatigue damage propagation model (M3) and a stress redistribution model (M2) in the presence of a damage event; the damage events to be considered for the stress redistribution and the PCPD calculation are damage events located within the element where the damage event terminates before the damage event propagates, and, if applicable, in adjacent elements, up to a distance that can be set by the user, for example, based on the laminate characteristics.
6. The method according to one or more of the preceding claims, wherein, The unacceptable situation refers to the occurrence of the following events before a certain number of cycles: for example, reinforcement failure in any element of the component, component stiffness decreasing below a certain value, or a certain number of damage events occurring.
7. The method according to claim 1, wherein, In the first step (A), information about the discretized model (5) is given in the finite element software SW-FEM, such as first information relating to the geometry of the entire laminate, second information relating to the stacking of the component (1), and third information intended to describe the initial elastic properties of the first layer (2) and the second layer (3) as needed, including, for example, longitudinal elastic modulus, Poisson's coefficient and tangential elastic modulus.
8. The method according to claim 1, wherein, In the second step (B), the positions of one or more first damage initiation sites (8) and one or more second damage initiation sites (9) may be located at the same position as the centroid of one or more first sub-divisions (61), and as needed, may be located at the same position as the centroid of one or more second sub-divisions (71), the geometry of which can be determined by the user.