Numerical Implementation Method and Apparatus for Adaptive Constitutive Model of Additive Manufacturing Stainless Steel
By establishing an adaptive constitutive model and optimizing the parameters of yield strength, hardening model, and damage model using stress-strain data, the problem of large simulation errors in existing technologies was solved, and accurate mechanical response and structural performance evaluation of stainless steel materials under different loading conditions were achieved.
Patent Information
- Application Number
- CN202611600953.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-10-16
- Publication Date
- 2026-11-17
AI Technical Summary
In the existing technology, during the calibration of constitutive model parameters for stainless steel materials manufactured by arc additive manufacturing, the coupled influence of yield parameters, hardening parameters, and damage parameters makes it difficult to simultaneously ensure the simulation accuracy of each deformation stage. This results in a large error between the numerical simulation results and the experimental results, affecting the reliability of the load-bearing performance assessment of structural components.
By acquiring stress-strain data of stainless steel in multiple sampling directions, the yield strength and shear yield strength are determined, an anisotropic yield criterion is established, the parameters of the hardening model and the damage model are fitted, and the anisotropy coefficient, the parameters of the hardening model and the damage model are optimized to minimize the error between the constitutive model prediction results and the stress-strain data, thus constructing an adaptive constitutive model.
This improves the accuracy of the mechanical response of stainless steel materials under different loading conditions, and significantly enhances the accuracy and reliability of structural load-bearing performance assessment.
Smart Images

Figure CN123116729A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of metal material processing, and specifically to a method and apparatus for numerically realizing an adaptive constitutive model of additive manufacturing stainless steel materials. Background Technology
[0002] Currently, when obtaining constitutive model parameters for stainless steel materials manufactured by arc additive manufacturing, the basic mechanical parameters such as yield strength and tensile strength are typically obtained by fitting experimental data first. These parameters are then input into the isotropic constitutive model built into the finite element software for numerical simulation. When the simulation results deviate from the experimental data, researchers manually adjust the hardening and damage parameters based on experience until the simulation results and experimental curves are basically consistent in macroscopic trends.
[0003] However, in the process of implementing the relevant technologies, it was found that the influence of yield parameter, hardening parameter and damage parameter on the macroscopic stress-strain response of materials is coupled. It is difficult to ensure the simulation accuracy of each deformation stage at the same time by calibrating each parameter independently or correcting them separately based on experience. This leads to a large error between the numerical simulation results and the experimental results, which in turn affects the reliability of the load-bearing performance assessment of structural components. Summary of the Invention
[0004] In view of the above problems, the present invention provides a method and apparatus for numerically realizing an adaptive constitutive model of additive manufacturing stainless steel materials.
[0005] According to a first aspect of the present invention, a numerical implementation method for an adaptive constitutive model of additively manufactured stainless steel is provided, comprising: acquiring stress-strain data of stainless steel specimens in multiple sampling directions, and determining the yield strength and shear yield strength of the stainless steel in multiple preset principal axis directions based on the stress-strain data; determining the anisotropy coefficient in the anisotropic yield criterion based on the yield strength and shear yield strength to reflect the difference in yield strength of the stainless steel in different preset principal axis directions; and fitting the parameters of a hardening model based on the data of the plastic deformation stage in the stress-strain data, wherein the hardening model is used to describe the change of flow stress with cumulative plastic strain during the plastic deformation process of the stainless steel. Based on the unloading stiffness degradation data in the stress-strain data, the parameters of the damage model are determined, and the damage variables determined by the parameters of the damage model are introduced into the anisotropic yield criterion. The damage model is used to describe the variation law of internal damage in stainless steel during plastic deformation. With the goal of minimizing the error between the stress-strain results predicted by the constitutive model and the stress-strain data, the anisotropy coefficient, the parameters of the hardening model, and the parameters of the damage model are optimized to obtain the target material parameters. The constitutive model is composed of the anisotropic yield criterion, the hardening model, and the damage model. The target material parameters are substituted into the constitutive model for numerical solution to obtain the mechanical response results used for structural performance evaluation.
[0006] A second aspect of the present invention provides a numerical realization apparatus for an adaptive constitutive model of additively manufactured stainless steel material, comprising: a data acquisition module for acquiring stress-strain data of stainless steel specimens in multiple sampling directions, and yield strength and shear yield strength of stainless steel material in multiple preset principal axis directions determined based on the stress-strain data; an anisotropy determination module for determining anisotropy coefficients in the anisotropy yield criterion based on the yield strength and shear yield strength, to reflect the difference in yield strength of stainless steel material in different preset principal axis directions; and a hardening determination module for fitting parameters of a hardening model based on data from the plastic deformation stage in the stress-strain data, wherein the hardening model describes the variation law of flow stress with cumulative plastic strain during the plastic deformation process of stainless steel material. The damage determination module determines the parameters of the damage model based on the unloading stiffness degradation data in the stress-strain data. The damage variables determined by the damage model parameters are then incorporated into the anisotropic yield criterion. The damage model describes the variation of internal damage in stainless steel during plastic deformation. The parameter optimization module optimizes the anisotropy coefficient, the parameters of the hardening model, and the parameters of the damage model to minimize the error between the stress-strain results predicted by the constitutive model and the actual stress-strain data. The constitutive model is composed of the anisotropic yield criterion, the hardening model, and the damage model. The numerical solution module substitutes the target material parameters into the constitutive model for numerical solution, obtaining the mechanical response results used for structural performance evaluation.
[0007] A third aspect of the present invention provides an electronic device comprising: one or more processors; and a memory for storing one or more computer programs, wherein the one or more processors execute the one or more computer programs to implement the steps of the method described above.
[0008] A fourth aspect of the present invention also provides a computer-readable storage medium having a computer program or instructions stored thereon, wherein the computer program or instructions, when executed by a processor, implement the steps of the above-described method.
[0009] A fifth aspect of the present invention also provides a computer program product, including a computer program or instructions that, when executed by a processor, implement the steps of the above-described method.
[0010] According to embodiments of the present invention, an anisotropic yield criterion reflecting the anisotropic yield behavior of materials is established based on stress-strain data of specimens from multiple sampling directions and yield strength and shear yield strength in the preset principal axis directions. Simultaneously, hardening model parameters are fitted using data from the plastic deformation stage, damage model parameters are determined using unloading stiffness degradation data, and damage variables are introduced into the anisotropic yield criterion, so that yield judgment is simultaneously controlled by anisotropy, hardening, and damage. Furthermore, with the goal of minimizing the error between the constitutive model prediction results and experimental data, the anisotropy coefficient, hardening model parameters, and damage model parameters are uniformly optimized to obtain target material parameters. Finally, the target material parameters are substituted into the constitutive model composed of the above three sub-models for numerical solution. This method at least partially overcomes the shortcomings of existing technologies that treat anisotropy, hardening, and damage separately or use fixed parameters, leading to large simulation errors. It can more realistically reflect the mechanical response of stainless steel materials under different loading conditions, improving the accuracy and reliability of structural load-bearing performance evaluation. Attached Figure Description
[0011] The above-mentioned contents, as well as other objects, features and advantages of the present invention, will become clearer from the following description of embodiments of the present invention with reference to the accompanying drawings.
[0012] Figure 1 The illustration shows an application scenario of the numerical implementation method, apparatus, equipment, medium, and program product for an adaptive constitutive model of additive manufacturing stainless steel material according to an embodiment of the present invention.
[0013] Figure 2 A flowchart illustrating a numerical implementation method for an adaptive constitutive model of additive manufacturing stainless steel material according to an embodiment of the present invention is shown.
[0014] Figure 3 A schematic diagram of the specimen sampling direction is shown in the numerical implementation method of the adaptive constitutive model of additive manufacturing stainless steel material according to an embodiment of the present invention.
[0015] Figure 4A The uniaxial tensile true stress-strain curves of the numerical implementation method of the adaptive constitutive model of additive manufacturing stainless steel material according to an embodiment of the present invention are shown.
[0016] Figure 4B The principal plane pure shear stress-shear strain curves of the numerical implementation method of the adaptive constitutive model of additive manufacturing stainless steel material according to an embodiment of the present invention are shown.
[0017] Figure 4C A stable hysteresis loop is shown in the numerical implementation method of the adaptive constitutive model for additive manufacturing stainless steel materials according to an embodiment of the present invention.
[0018] Figure 4DThe graded unloading stiffness degradation curve of the numerical implementation method of the adaptive constitutive model of additive manufacturing stainless steel material according to an embodiment of the present invention is shown.
[0019] Figure 5 A schematic diagram of the material coordinate axes and stress components is shown in the numerical implementation method of the adaptive constitutive model of additive manufacturing stainless steel material according to an embodiment of the present invention.
[0020] Figure 6 A schematic diagram of the stable hysteretic back stress point selection method for the numerical implementation method of the adaptive constitutive model of additive manufacturing stainless steel material according to an embodiment of the present invention is shown.
[0021] Figure 7 A flowchart illustrating the static constitutive model construction process of the numerical implementation method for the adaptive constitutive model of additive manufacturing stainless steel material according to an embodiment of the present invention is shown.
[0022] Figure 8A A schematic diagram illustrating the parameter acquisition of the uniaxial tensile elastic segment in the numerical implementation method of the adaptive constitutive model for additive manufacturing of stainless steel materials according to an embodiment of the present invention is shown.
[0023] Figure 8B A schematic diagram illustrating the parameter acquisition of the planar pure shear elastic segment in the numerical implementation method of the adaptive constitutive model of additive manufacturing stainless steel material according to an embodiment of the present invention is shown.
[0024] Figure 8C A flowchart illustrating the multi-parameter adaptive optimization process of the numerical implementation method of the adaptive constitutive model for additive manufacturing stainless steel materials according to an embodiment of the present invention is shown.
[0025] Figure 8D A flowchart illustrating the cyclic constitutive model construction process of the numerical implementation method for the adaptive constitutive model of additive manufacturing stainless steel material according to an embodiment of the present invention is shown.
[0026] Figure 9 A structural block diagram of a numerical realization apparatus for an adaptive constitutive model of additive manufacturing stainless steel material according to an embodiment of the present invention is shown.
[0027] Figure 10 A block diagram of an electronic device suitable for implementing a numerical method for an adaptive constitutive model of additively manufactured stainless steel materials, according to an embodiment of the present invention, is shown. Detailed Implementation
[0028] Hereinafter, embodiments of the present invention will be described with reference to the accompanying drawings. However, it should be understood that these descriptions are exemplary only and are not intended to limit the scope of the invention. In the following detailed description, numerous specific details are set forth to provide a thorough understanding of the embodiments of the invention for ease of explanation. However, it will be apparent that one or more embodiments may be practiced without these specific details. Furthermore, descriptions of well-known structures and techniques are omitted in the following description to avoid unnecessarily obscuring the concept of the invention.
[0029] The terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the invention. The terms “comprising,” “including,” etc., as used herein indicate the presence of the stated features, steps, operations, and / or components, but do not exclude the presence or addition of one or more other features, steps, operations, or components.
[0030] All terms used herein (including technical and scientific terms) have the meanings commonly understood by those skilled in the art, unless otherwise defined. It should be noted that the terms used herein are to be interpreted in a manner consistent with the context of this specification, and not in an idealized or overly rigid way.
[0031] When using expressions such as "at least one of A, B and C", they should generally be interpreted in accordance with the meaning that is commonly understood by those skilled in the art (e.g., "a system having at least one of A, B and C" should include, but is not limited to, a system having A alone, a system having B alone, a system having C alone, a system having A and B, a system having A and C, a system having B and C, and / or a system having A, B and C, etc.).
[0032] In the technical solution of this invention, the user information (including but not limited to user personal information, user image information, user device information, such as location information) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved are all information and data authorized by the user or fully authorized by all parties. Furthermore, the collection, storage, use, processing, transmission, provision, disclosure, and application of related data all comply with relevant laws, regulations, and standards, take necessary confidentiality measures, do not violate public order and good morals, and provide corresponding operation entry points for users to choose to authorize or refuse.
[0033] This invention provides a numerical implementation method for an adaptive constitutive model of stainless steel material in additive manufacturing. The method acquires stress-strain data of stainless steel specimens in multiple sampling directions, and determines the yield strength and shear yield strength of the stainless steel material in multiple preset principal axis directions based on the stress-strain data. Based on the yield strength and shear yield strength, an anisotropy coefficient in the anisotropic yield criterion is determined to reflect the difference in yield strength of the stainless steel material in different preset principal axis directions. Based on the data from the plastic deformation stage in the stress-strain data, parameters of a hardening model are fitted, wherein the hardening model describes the variation of flow stress with cumulative plastic strain during the plastic deformation process of the stainless steel material. Based on the unloading stiffness degradation data in the stress-strain data, the parameters of the damage model are determined so that the damage variables determined by the parameters of the damage model can be introduced into the anisotropic yield criterion. The damage model is used to describe the change law of internal damage in stainless steel during plastic deformation. With the goal of minimizing the error between the stress-strain results predicted by the constitutive model and the stress-strain data, the anisotropy coefficient, the parameters of the hardening model, and the parameters of the damage model are optimized to obtain the target material parameters. The constitutive model is composed of the anisotropic yield criterion, the hardening model, and the damage model. The target material parameters are substituted into the constitutive model for numerical solution to obtain the mechanical response results used for structural performance evaluation.
[0034] Figure 1 The illustration shows an application scenario of the numerical implementation method, apparatus, equipment, medium, and program product for an adaptive constitutive model of additive manufacturing stainless steel material according to an embodiment of the present invention.
[0035] like Figure 1 As shown, application scenario 100 according to this embodiment may include a first terminal device 101, a second terminal device 102, a third terminal device 103, a network 104, and a server 105. The network 104 serves as a medium for providing a communication link between the first terminal device 101, the second terminal device 102, the third terminal device 103, and the server 105. The network 104 may include various connection types, such as wired or wireless communication links, or fiber optic cables, etc.
[0036] Users can use the first terminal device 101, the second terminal device 102, and the third terminal device 103 to interact with the server 105 via the network 104 to receive or send messages, etc. Various communication client applications can be installed on the first terminal device 101, the second terminal device 102, and the third terminal device 103, such as shopping applications, web browser applications, search applications, instant messaging tools, email clients, social media platform software, etc. (for example only).
[0037] The first terminal device 101, the second terminal device 102, and the third terminal device 103 can be various electronic devices with displays and support web browsing, including but not limited to smartphones, tablets, laptops, and desktop computers.
[0038] Server 105 can be a server that provides various services, such as a backend management server that supports websites browsed by users using the first terminal device 101, the second terminal device 102, and the third terminal device 103 (this is just an example). The backend management server can analyze and process data such as received user requests, and feed back the processing results (such as web pages, information, or data obtained or generated according to user requests) to the terminal devices.
[0039] It should be noted that the numerical implementation method of the adaptive constitutive model of additive manufacturing stainless steel material provided in the embodiments of the present invention can generally be executed by server 105. Correspondingly, the numerical implementation device of the adaptive constitutive model of additive manufacturing stainless steel material provided in the embodiments of the present invention can generally be located in server 105. The numerical implementation method of the adaptive constitutive model of additive manufacturing stainless steel material provided in the embodiments of the present invention can also be executed by a server or server cluster that is different from server 105 and can communicate with the first terminal device 101, the second terminal device 102, the third terminal device 103 and / or server 105. Correspondingly, the numerical implementation device of the adaptive constitutive model of additive manufacturing stainless steel material provided in the embodiments of the present invention can also be located in a server or server cluster that is different from server 105 and can communicate with the first terminal device 101, the second terminal device 102, the third terminal device 103 and / or server 105.
[0040] It should be understood that Figure 1 The number of first terminal devices, second terminal devices, third terminal devices, networks, and servers shown in the diagram is merely illustrative. Depending on implementation needs, any number of first terminal devices, second terminal devices, third terminal devices, networks, and servers can be included.
[0041] The following will be based on Figure 1 The described scene, through Figures 2-6 The numerical implementation method of the adaptive constitutive model for additive manufacturing stainless steel material according to the embodiments of the invention is described in detail.
[0042] Figure 2 A flowchart illustrating a numerical implementation method for an adaptive constitutive model of additive manufacturing stainless steel material according to an embodiment of the present invention is shown.
[0043] Figure 3 A schematic diagram of the specimen sampling direction according to an embodiment of the present invention is shown.
[0044] Figure 4A The uniaxial tensile true stress-strain curves of the numerical implementation method of the adaptive constitutive model of additive manufacturing stainless steel material according to an embodiment of the present invention are shown.
[0045] Figure 4B The principal plane pure shear stress-shear strain curves of the numerical implementation method of the adaptive constitutive model of additive manufacturing stainless steel material according to an embodiment of the present invention are shown.
[0046] Figure 4C A stable hysteresis loop is shown in the numerical implementation method of the adaptive constitutive model for additive manufacturing stainless steel materials according to an embodiment of the present invention.
[0047] Figure 4D The graded unloading stiffness degradation curve of the numerical implementation method of the adaptive constitutive model of additive manufacturing stainless steel material according to an embodiment of the present invention is shown.
[0048] like Figure 2 As shown, this embodiment includes operations S210 to S260.
[0049] In operation S210, stress-strain data of stainless steel specimens in multiple sampling directions are acquired, as well as yield strength and shear yield strength of stainless steel in multiple preset principal axis directions determined based on the stress-strain data.
[0050] In operation S220, the anisotropy coefficient in the anisotropic yield criterion is determined based on the yield strength and shear yield strength to reflect the difference in yield strength of stainless steel in different preset principal axis directions.
[0051] In operation S230, based on the data from the plastic deformation stage in the stress-strain data, the parameters of the hardening model are fitted. The hardening model is used to describe the variation of flow stress with cumulative plastic strain in stainless steel during plastic deformation.
[0052] In operation S240, the parameters of the damage model are determined based on the unloading stiffness degradation data in the stress-strain data, so that the damage variables determined by the parameters of the damage model are introduced into the anisotropic yield criterion. The damage model is used to describe the variation law of internal damage of stainless steel material during plastic deformation.
[0053] In operation S250, with the goal of minimizing the error between the stress-strain results predicted by the constitutive model and the stress-strain data, the anisotropy coefficient, the parameters of the hardening model, and the parameters of the damage model are optimized to obtain the target material parameters. The constitutive model is composed of the anisotropic yield criterion, the hardening model, and the damage model.
[0054] In operation S260, the target material parameters are substituted into the constitutive model for numerical solution to obtain the mechanical response results used for structural performance evaluation.
[0055] According to an embodiment of the present invention, based on the anisotropic characteristics of WAAM-316L stainless steel material, uniaxial tensile tests, torsion tests, cyclic loading-unloading tests, and graded loading-unloading damage tests are carried out on specimens in multiple different sampling directions, and stress-strain data in each sampling direction are obtained, specifically including true stress-true strain curves, shear stress-strain curves, stable hysteresis loops, and unloading stiffness degradation data at different loading stages.
[0056] like Figure 3 As shown, the multiple sampling directions include multiple preset main axis directions and auxiliary sampling directions such as 45°. The multiple preset main axis directions include a first preset main axis direction, a second preset main axis direction, and a third preset main axis direction, which correspond to the X-axis (nozzle movement direction), the Y-axis (material longitudinal stacking direction), and the Z-axis (material transverse stacking direction), respectively.
[0057] like Figure 4A As shown, the uniaxial tensile true stress-true strain curves of the four sampling directions (0°, 45°, 90° and Z direction) basically overlap in the elastic segment, and show obvious differentiation after entering the plastic stage. The yield strength and hardening level are the highest in the 0° direction, followed by the 90° direction, while the 45° and thickness directions are relatively low, indicating that the material has significant anisotropic characteristics and that the flow stress and hardening behavior are different in different directions.
[0058] like Figure 4B The figure shows the pure shear stress-shear strain curves in the principal planes. The horizontal axis represents engineering shear strain, and the vertical axis represents shear stress. The three curves represent the pure shear response in planes 1-2, 1-3, and 2-3, respectively. τ12 represents the shear stress in plane 1-2, τ13 represents the shear stress in plane 1-3, and τ23 represents the shear stress in plane 2-3. The slopes of the three curves are similar in the elastic stage, but the hardening trends differ slightly after entering the plastic stage, indicating that the shear yield strength and shear hardening behavior of this material in different shear planes have a certain direction dependence, but the degree of difference is weaker than that under uniaxial tension. This curve is used to determine the shear yield strength and shear modulus in the planes perpendicular to each principal axis.
[0059] like Figure 4C As shown, the steady hysteresis loop is a closed loop curve obtained after cyclic loading reaches a steady state. Its shape and width reflect the hardening characteristics and energy dissipation capacity of the material under cyclic loading. Its horizontal axis is the total strain and the vertical axis is the stress. The back stress evolution data can be extracted from this curve and used to calibrate the kinematic hardening parameters in the cyclic constitutive model.
[0060] like Figure 4D As shown, the unloading stiffness gradually decreases during the staged loading-unloading process. The horizontal axis represents the actual strain, and the vertical axis represents the actual stress. The three vertical dashed lines (orange, green, and red) represent the nodes at the first, second, and third unloading points, respectively, and the corresponding unloading moduli are denoted as follows: , and As the loading level increases, the slope (unloading modulus) of each unloading segment compared to the initial elastic modulus... The stiffness degradation shows a decreasing trend, and the degree of stiffness degradation intensifies with the increase of accumulated plastic strain, reflecting the gradual accumulation of internal damage in the material. (Formula in the figure) Used to calculate the test damage variables corresponding to each unloading node, where This represents the initial elastic modulus obtained from the first loading of the elastic segment. This curve can be used to calculate the evolution of damage variables and calibrate relevant parameters in the damage model.
[0061] Based on the acquired stress-strain data, the yield strength and shear yield strength of WAAM-316L stainless steel in each preset principal axis direction were further determined. For the uniaxial tensile curves in the first, second, and third preset principal axis directions, linear regression was performed on their initial linear segments to determine the slope of the elastic line. A parallel line parallel to this elastic line and offset by 0.2% along the positive axial strain direction was then drawn. The stress at the intersection of this parallel line and the test curve was defined as the yield strength in the corresponding preset principal axis direction. The yield strengths in the three preset principal axis directions were denoted as follows: , and These correspond to the first preset spindle direction, the second preset spindle direction, and the third preset spindle direction, respectively.
[0062] Linear regression was also performed on the initial linear segments of each pure shear test curve, and a straight line parallel to the shear elastic line and offset by 0.2% along the positive direction of engineering shear strain was plotted. The shear stress at the intersection of this line and the test curve was taken as the shear yield strength in the plane perpendicular to each preset principal axis direction. The first shear yield strength in the plane perpendicular to the first preset principal axis direction was denoted as... The second shear yield strength on a plane perpendicular to the second preset principal axis direction is denoted as... The third shear yield strength on a plane perpendicular to the third preset principal axis direction is denoted as... .
[0063] Based on the obtained yield strength and shear yield strength in each preset principal axis direction, substitute them into the mathematical expression of the anisotropic yield criterion (such as the Hill48 yield criterion), solve the system of equations simultaneously and supplement with least squares fitting, and then deduce the anisotropic coefficients of the normal stress term and the shear stress term in the anisotropic yield criterion.
[0064] These anisotropy coefficients (including normal stress and shear stress terms) can quantitatively reflect the inherent differences in yield strength of WAAM-316L stainless steel in different preset principal axis directions. Meanwhile, the uniaxial tensile test data in the 45° sampling direction are specifically used to calibrate the anisotropic behavior in the plane (i.e., the XY plane) containing the first and second preset principal axis directions to ensure an accurate description of the yield trajectory in this plane.
[0065] Based on the data from the plastic deformation stage in the stress-strain data, a power function hardening model, a saturated hardening model, or a hybrid hardening model combining these are selected. With cumulative plastic strain as the independent variable and flow stress as the dependent variable, a nonlinear regression algorithm is used to iteratively fit the parameters of the hardening model (e.g., the saturated hardening increment and hardening rate parameters in the isotropic hardening model, or the kinematic hardening modulus and dynamic recovery coefficient in the kinematic hardening model). This allows the hardening model to accurately describe the evolution of flow stress in WAAM-316L stainless steel during plastic deformation as cumulative plastic strain increases.
[0066] Meanwhile, based on the unloading stiffness degradation data in the stress-strain data, namely the unloading elastic modulus recorded at different loading stages in the graded loading-unloading damage test, the damage variable is defined as the relative decay of the current elastic modulus relative to the initial undamaged modulus, based on the continuous damage mechanics framework.
[0067] By tracking the decreasing trajectory of unloading stiffness with increasing cumulative plastic strain, and assuming that the damage evolution rate, cumulative plastic strain increment, and current stress state satisfy a power function or exponential function relationship, key parameters in the damage model, such as the damage threshold, damage energy scale parameter, and damage evolution index, are determined using numerical regression methods. The damage variables described by this damage model will be incorporated into subsequent calculations using an anisotropic yield criterion to achieve a description of the yield behavior of damage coupling.
[0068] After completing the initial parameter calibration of the aforementioned component models, a multi-objective optimization function is constructed with the optimization objective of minimizing the error between the stress-strain results predicted by the constitutive model composed of the anisotropic yield criterion, hardening model, and damage model and the measured stress-strain data in all sampling directions. A global sensitivity analysis method (Sobol exponent method) is used to perform global sensitivity analysis on each parameter to be optimized. By calculating the first-order and total-order Sobol exponents, a subset of key parameters with the most significant impact on output uncertainty is selected to reduce the dimensionality of the parameters to be optimized. The selected key parameters are divided into anisotropic yield parameter group, hardening parameter group, and damage parameter group according to their physical source. Optimization is performed in stages according to the order of anisotropic yield parameter group, hardening parameter group, and damage parameter group, while fixing the parameter values of other parameter groups when optimizing the current parameter group.
[0069] During the calculation of the objective function, the weights of each set of experimental data are dynamically adjusted based on the proportion of the normalized error between each set of experimental data and the model prediction results to the total normalized error. Data sets with larger normalized errors are assigned higher weights. Through the above-mentioned adaptive strategy of phased optimization and dynamic weight adjustment, until the objective function value meets the preset convergence condition, a set of consistent and physically meaningful target material parameters is obtained.
[0070] After obtaining the target material parameters, the Markov chain Monte Carlo method is used to sample and analyze the optimal parameters, obtain the probability distribution and confidence interval of each parameter, and quantify the parameter uncertainty caused by process fluctuations.
[0071] The final optimized target material parameters are fully substituted into the constitutive model. A two-stage strategy of incremental step elastic prediction and plastic correction is used to rigorously solve for the stress state and damage variables at each incremental step. The constitutive model is then embedded into the boundary value analysis of the actual structural component using the user material subroutine interface of the finite element software. Specifically, Newton's iteration method is used for iterative solving in the plastic correction stage, and the consistent plastic tangent modulus of the missile is calculated after the solution is completed to ensure the convergence speed and stability of the overall Newton iteration.
[0072] Through the above nonlinear finite element numerical solution, the mechanical response results such as load-displacement curves, stress-strain distributions across the entire field, and damage evolution cloud maps of the structure under complex load paths are finally obtained. These results can be directly used for the ultimate bearing capacity assessment, ductile fracture prediction, and overall safety verification of stainless steel additive manufacturing structural components, thereby fully realizing the entire chain calculation process from material parameter calibration to structural performance evaluation.
[0073] To ensure statistical reliability during the entire parameter identification process, three sets of parallel samples were set up in the same sampling direction. Each set of curves independently completed zero-point correction, elastic segment regression, yield point identification, and constitutive parameter fitting. Finally, the elastic modulus, yield strength in each preset principal axis direction, shear yield strength, and arithmetic mean of each constitutive parameter were used as model inputs, and the standard deviation was used to quantitatively characterize the data dispersion.
[0074] Based on stress-strain data from specimens sampled from multiple directions, and the yield strength and shear yield strength along the preset principal axis, an anisotropic yield criterion is established to reflect the anisotropic yield behavior of the material. Simultaneously, hardening model parameters are fitted using data from the plastic deformation stage, and damage model parameters are determined using unloading stiffness degradation data. Damage variables are then introduced into the anisotropic yield criterion, ensuring that yield judgment is synergistically controlled by anisotropy, hardening, and damage. Furthermore, with the goal of minimizing the error between the constitutive model predictions and experimental data, the anisotropy coefficient, hardening model parameters, and damage model parameters are uniformly optimized to obtain target material parameters. Finally, these target material parameters are substituted into the constitutive model composed of the three sub-models for numerical solution. This method at least partially overcomes the shortcomings of existing technologies that separately handle anisotropy, hardening, and damage or use fixed parameters, leading to large simulation errors. It can more realistically reflect the mechanical response of stainless steel under different loading conditions, significantly improving the accuracy and reliability of structural load-bearing performance assessment.
[0075] Figure 5 A schematic diagram of the material coordinate axes and stress components is shown in the numerical implementation method of the adaptive constitutive model of additive manufacturing stainless steel material according to an embodiment of the present invention.
[0076] According to an embodiment of the present invention, the plurality of preset principal axis directions include a first preset principal axis direction, a second preset principal axis direction, and a third preset principal axis direction; the shear yield strength includes a first shear yield strength on a plane perpendicular to the first preset principal axis direction, a second shear yield strength on a plane perpendicular to the second preset principal axis direction, and a third shear yield strength on a plane perpendicular to the third preset principal axis direction; when the yield strength of the first preset principal axis direction is used as the reference yield strength, the anisotropy coefficient in the anisotropic yield criterion is determined based on the yield strength and the shear yield strength, including: based on the reference yield strength, the second preset principal axis direction, and the third preset principal axis direction; The anisotropy coefficient of the normal stress term in the anisotropic yield criterion is calculated based on the yield strength in the axial direction and the yield strength in the third preset principal axis direction. The anisotropy coefficient of the normal stress term is used to reflect the difference in yield strength between the three preset principal axis directions. Based on the reference yield strength, the first shear yield strength, the second shear yield strength, and the third shear yield strength, the anisotropy coefficient of the shear stress term in the anisotropic yield criterion is calculated. The anisotropy coefficient of the shear stress term is used to reflect the difference in shear yield strength between the three preset principal axis perpendicular planes. The anisotropy coefficients of the normal stress term and the anisotropy coefficients of the shear stress term are used as the anisotropy coefficients.
[0077] When using the yield strength in the first preset principal axis direction as the reference yield strength, the anisotropy coefficient in the anisotropic yield criterion is determined according to the following process based on the yield strength and shear yield strength.
[0078] Based on the reference yield strength, the yield strength along the second predetermined principal axis, and the yield strength along the third predetermined principal axis, the anisotropy coefficient of the normal stress term in the anisotropic yield criterion is calculated. The anisotropy coefficient of the normal stress term reflects the difference in yield strength among the three predetermined principal axis directions. In the Hill anisotropic yield criterion, there are three anisotropy coefficients for the normal stress term, denoted as F, G, and H. Specifically, the yield strength along the first predetermined principal axis is used as the reference yield strength. As a reference yield strength, the reference yield strength and the yield strength in the second preset principal axis direction will be used. Yield strength in the third preset principal axis direction Substitute the values into the following formulas to perform the calculations:
[0079] ;
[0080] ;
[0081] ;
[0082] This yields three anisotropy coefficients for normal stress terms, F, G, and H, which respectively reflect the degree of difference in yield strength between different preset principal axis directions.
[0083] Based on the reference yield strength, first shear yield strength, second shear yield strength, and third shear yield strength, the anisotropy coefficient of the shear stress term in the anisotropic yield criterion is calculated. The anisotropy coefficient of the shear stress term reflects the difference in shear yield strength between the three planes perpendicular to the three predetermined principal axes. In the Hill anisotropic yield criterion, there are three anisotropy coefficients for the shear stress term, denoted as L, M, and N. Specifically, the reference yield strength... The shear yield strength on each of the three principal axes perpendicular to the plane is combined and substituted into the following formula for calculation:
[0084] ;
[0085] ;
[0086] ;
[0087] in, The first shear yield strength on a plane perpendicular to the first preset principal axis direction. The second shear yield strength is defined as the shear yield strength on a plane perpendicular to the second preset principal axis direction. The third shear yield strength is defined as the shear yield strength in a plane perpendicular to the third predetermined principal axis. This yields three shear stress terms with anisotropy coefficients L, M, and N, which reflect the degree of difference in shear yield strength between planes perpendicular to different principal axes.
[0088] The calculated anisotropy coefficients F, G, and H for the normal stress term and L, M, and N for the shear stress term are collectively used as the anisotropy coefficients of the anisotropic yield criterion, namely the six anisotropy coefficients of the Hill anisotropic yield criterion. In other words, the anisotropy coefficients F, G, and H for the normal stress term and L, M, and N for the shear stress term together constitute a complete set of anisotropic coefficients describing the anisotropic yield behavior of stainless steel. When F, G, and H are all equal to 1 and L, M, and N are all equal to 3, the anisotropic yield criterion degenerates into the isotropic Mises yield criterion.
[0089] By calculating the anisotropy coefficients of the normal stress term and the shear stress term respectively, the yield strength differences of the material in the three preset principal axis directions and the three shear planes can be accurately reflected, making the anisotropic yield criterion more realistically describe the anisotropic yield behavior of stainless steel.
[0090] like Figure 5 As shown, the three-dimensional material coordinate axes and stress components of stainless steel are defined. The three preset principal axes are the first preset principal axis direction, the second preset principal axis direction, and the third preset principal axis direction. The first preset principal axis direction corresponds to the printing direction, and the second and third preset principal axes directions correspond to two orthogonal directions of the material in the printing plane that are perpendicular to the printing direction.
[0091] The normal stress components are marked in the figure along the three preset principal axes. , and The subscripts 1, 2, and 3 represent the normal directions of the surface, respectively, and the normal stress components act along their respective principal axes. This represents the normal stress acting on a cross section perpendicular to the first preset principal axis direction. and And so on.
[0092] The shear stress components include the shear stresses acting on the planes perpendicular to each principal axis, as shown in the figure. , , , , and There are a total of six shear stress components, of which This represents the shear stress components acting along the first preset principal axis and along the second preset principal axis. This represents the shear stress component acting in the direction of the second preset principal axis and acting in the direction of the first preset principal axis; the other shear stress components are deduced similarly.
[0093] Shear stresses are paired and equal. equal , equal , equal Therefore, only three independent shear stress components are needed to describe the shear stress state. These stress components together constitute the complete stress tensor of the material under three-dimensional stress, which is used to calculate the equivalent stress in the anisotropic yield criterion.
[0094] According to an embodiment of the present invention, when the hardening model is an isotropic hardening model, the parameters of the hardening model are fitted based on the data of the plastic deformation stage in the stress-strain data, including: determining the plastic strain of each data point in the plastic deformation stage based on the uniaxial tensile stress-strain data in the stress-strain data, and accumulating them to obtain the cumulative plastic strain; using the isotropic hardening model, calculating the model flow stress based on the cumulative plastic strain, the initial flow stress determined based on the yield strength, and the saturated hardening increment and hardening rate parameters in the isotropic hardening model, wherein the saturated hardening increment and hardening rate parameters are the parameters to be fitted, and the model flow stress increases with the increase of the cumulative plastic strain, and the increase gradually decreases and tends to the upper limit value jointly determined by the initial flow stress and the saturated hardening increment; with the goal of minimizing the error between the model flow stress and the experimental flow stress, the parameters to be fitted are fitted to obtain the fitted value of the saturated hardening increment and the fitted value of the hardening rate parameter, wherein the experimental flow stress is the flow stress in the uniaxial tensile stress-strain data that has the same cumulative plastic strain as the model flow stress; and using the fitted value of the initial flow stress, the fitted value of the saturated hardening increment, and the fitted value of the hardening rate parameter as the parameters of the hardening model.
[0095] The isotropic hardening model adopted is the Voce hardening model, which describes the variation of flow stress with cumulative plastic strain during the plastic deformation of stainless steel. The mathematical expression of the Voce hardening model is:
[0096] ;
[0097] in, The model flow stress is calculated for the hardening model under the current accumulated plastic strain. The initial flow stress, σ is the saturation hardening increment, representing the maximum increase in flow stress relative to the initial flow stress; b is the hardening rate parameter, controlling how quickly the flow stress approaches the saturation value; and p is the cumulative plastic strain, a state variable that gradually increases with the loading process.
[0098] During the fitting process, the initial flow stress Based on the yield strength obtained above, it can be specifically taken as the yield strength in the first preset principal axis direction. This is a known quantity. The parameter to be fitted is only the saturation hardening increment. And the hardening rate parameter b.
[0099] Based on the uniaxial tensile stress-strain data in the stress-strain data, the plastic strain of each data point in the plastic deformation stage is determined. Specifically, for the i-th data point of the uniaxial tensile test curve, its true total strain is known. Actual stress And the elastic modulus E in that direction, the plastic strain at this data point is calculated using the following formula:
[0100] ;
[0101] in, The experimental plastic strain is obtained by subtracting the elastic strain from the total experimental strain at the i-th data point.
[0102] Then, the plastic strain at each data point is summed to obtain the cumulative plastic strain. Specifically, the plastic strain increment between adjacent data points is first calculated. :
[0103] ;
[0104] The cumulative plastic strain is then obtained by adding the strains point by point:
[0105] ;
[0106] in, The initial value of the accumulated plastic strain up to the i-th data point is zero.
[0107] Using the isotropic hardening model (Voce hardening model), based on the cumulative plastic strain p and the initial flow stress... , saturated hardening increment The model flow stress is calculated using the hardening rate parameter b. Due to the initial flow stress... Given that the parameters to be fitted are And b. The model flow stress increases with increasing cumulative plastic strain, and the rate of increase gradually decreases and tends to the upper limit determined by the initial flow stress and the saturation hardening increment. .
[0108] The goal is to minimize the error between the model flow stress and the experimental flow stress, and to fit the parameters. Fit the parameter to b. Define the error function for fitting the Voce parameter as follows:
[0109] ;
[0110] in, This represents the number of data points used in fitting the experimental curve. The experimental true stress (i.e., the experimental flow stress with the same cumulative plastic strain as the model flow stress) is the i-th data point in the uniaxial tensile stress-strain data. The model flow stress is calculated for the model under the same cumulative plastic strain. The saturation hardening increment is determined by minimizing this error function. The optimal fitted value of the hardening rate parameter b, The mean square error (i.e., the squared mean of the residuals) between the model flow stress and the actual experimental stress corresponding to the experimental curve is used as the objective function for parameter fitting to quantify the degree of deviation between the model prediction and the experimental value.
[0111] Finally, the initial flow stress , Saturated hardening increment fitting value Together with the fitted value b of the hardening rate parameter, they are used as parameters of the hardening model.
[0112] By fitting saturated hardening increment and hardening rate parameters using uniaxial tensile data, the law that the flow stress increases with the increase of cumulative plastic strain and tends to saturate under isotropic hardening can be accurately described, thereby improving the simulation accuracy of material mechanical response under monotonic loading conditions.
[0113] Figure 6 A schematic diagram of the stable hysteretic back stress point selection method for the numerical implementation method of the adaptive constitutive model of additive manufacturing stainless steel material according to an embodiment of the present invention is shown.
[0114] According to an embodiment of the present invention, when the hardening model is a kinematic hardening model, the parameters of the hardening model are fitted based on the data of the plastic deformation stage in the stress-strain data, including: obtaining the stable hysteresis loop obtained from the cyclic loading and unloading test from the stress-strain data, determining the plastic strain of each data point in the plastic deformation stage based on the stable hysteresis loop, and accumulating them to obtain the cumulative plastic strain, and determining the total back stress data points in the plastic deformation stage; using the kinematic hardening model, based on the plastic strain increment and cumulative plastic strain increment of each increment step in the strain history of the cyclic loading and unloading test, and In the kinematic hardening model, the kinematic hardening modulus and dynamic recovery coefficient of each back stress component are incrementally updated to obtain the total back stress of the model. The kinematic hardening modulus and dynamic recovery coefficient of each back stress component are the parameters to be fitted. With the goal of minimizing the error between the total back stress of the model and the total back stress of the experiment, the parameters to be fitted are fitted to obtain the fitted values of the kinematic hardening modulus and the dynamic recovery coefficient. The total back stress of the experiment is the total back stress data point extracted from the stable hysteresis loop. The fitted values of the kinematic hardening modulus and the dynamic recovery coefficient are used as the parameters of the hardening model.
[0115] The kinematic hardening model employs the Charpy hybrid strengthening model, which describes the translational evolution of the yield surface in stainless steel under cyclic loading. The Charpy hybrid strengthening model characterizes the movement of the yield surface in stress space through multiple back stress components. The total back stress is the superposition of these back stress components, and its mathematical expression is as follows:
[0116] ;
[0117] in, For the total back stress, , and There are three sets of back stress components, which are added together according to the above formula to obtain the total back stress. The evolution of each set of back stress components is controlled by the following formula:
[0118] ;
[0119] in, Let be the kinematic hardening modulus of the k-th back stress component. Let be the dynamic recovery coefficient of the k-th back stress component. For a small increment of the plastic strain tensor, This represents the cumulative plastic strain increment. Kinematic hardening modulus. The hardening rate and dynamic recovery coefficient used to determine the back stress as a function of plastic strain increment. Used to control the saturation level of back stress.
[0120] During the fitting process, the parameters to be fitted are the kinematic hardening modulus of each group of back stress components. and dynamic response coefficient That is, there are six parameters to be fitted in three sets of back stresses.
[0121] First, the stable hysteresis loop obtained from the cyclic loading and unloading test is acquired from the stress-strain data. A stable hysteresis loop refers to the hysteresis loop after the material reaches a stable state under cyclic loading, at which point the cyclic hardening or softening of the material has stabilized, and the shape of the hysteresis loop no longer changes significantly with the number of cycles. Based on this stable hysteresis loop, the plastic strain of each data point in the plastic deformation stage is determined, and the cumulative plastic strain is obtained by summing them. Specifically, for the i-th data point on the stable hysteresis loop, its true total strain is known. and actual stress The plastic strain increment for this data point is calculated as follows.
[0122] First, subtract the elastic strain from the total experimental strain to obtain the experimental plastic strain:
[0123] ;
[0124] Where E is the elastic modulus in this direction. Then calculate the plastic strain increment between adjacent data points:
[0125] ;
[0126] Then, calculate the cumulative plastic strain increment:
[0127] ;
[0128] in, It represents the cumulative plastic strain increment of the i-th data point relative to the previous data point.
[0129] Simultaneously, the total back stress data points during the plastic deformation stage are determined based on the stable hysteresis loop. The test back stress points are taken from a complete hysteresis loop after it has reached stability. Within the plastic segment of the hysteresis loop, the i-th total strain coordinate is first selected, and then points at the same total strain coordinate are taken on both the increasing and decreasing strain supports. and Points near inflection points and near elastic segments are not included in the point selection. and They are respectively and The actual test stress at the coordinate point is averaged with the actual test stress at the coordinate point to obtain the total back stress at that coordinate point.
[0130] ;
[0131] By selecting multiple coordinates at equal intervals along the effective plastic segment, a set of total back stress data points for fitting can be obtained.
[0132] Using a kinematic hardening model, the back stress components are updated incrementally based on the plastic strain increment and cumulative plastic strain increment at each increment step in the strain history of the cyclic loading and unloading test, as well as the kinematic hardening modulus and dynamic recovery coefficient of each back stress component in the kinematic hardening model. The kinematic hardening modulus of each back stress component is then used. and dynamic response coefficient These are the parameters to be fitted. Given a set of... and Then, the back stress of each group is calculated step by step from the previous increment according to the following formula:
[0133] ;
[0134] in, and These represent the k-th back stress components before and after the incremental step calculation, respectively, with initial values... Set to zero. The total back stress of the model is obtained by superimposing all back stress components. .
[0135] The goal is to minimize the error between the total back stress of the model and the total back stress of the experiment. The parameters to be fitted are then fitted. The error function for fitting the Charpy parameters is defined as follows:
[0136] ;
[0137] in, The number of back stress data points selected within the effective plastic segment of the hysteresis loop to stabilize it. Let be the total back stress at the i-th coordinate. The total back stress is calculated according to the corresponding model. During the fitting process, the experimental curve can only directly provide the total back stress. They cannot be measured separately. , and Therefore, the experimental strain history was input into the model, with the goal of minimizing the error between the calculated total back stress and the experimental total back stress, and three sets of... and , The mean square error (i.e., the squared mean of residuals) between the total back stress calculated by the model and the total back stress in the effective plastic segment of the stable hysteresis loop is used as the objective function for fitting the back stress parameters, and is used to quantify the degree of deviation between the total back stress predicted by the model and the total back stress in the experiment.
[0138] During fitting, let Greater than , Greater than , making This indicates components that change rapidly after reverse loading. The component representing a change in speed at a moderate rate, It represents components that change slowly and control long-term translation, thereby ensuring that the physical meaning of each back stress component is clear and non-redundant.
[0139] By minimizing the above error function, the kinematic hardening modulus fitting values of each set of back stress components are obtained. , , Fitted values of dynamic recovery coefficient , , The fitted values of the kinematic hardening modulus and the fitted values of the dynamic recovery coefficient are used together as parameters of the hardening model.
[0140] like Figure 6 As shown in the schematic diagram of the back stress points of the stable hysteresis loop, the horizontal axis represents the total strain, and the vertical axis represents the actual stress. The diagram selects corresponding points on the hysteresis loop located on the increasing strain branch and the decreasing strain branch. This represents the i-th point on the strain support. Indicates the strain reduction support and Corresponding points with the same total strain coordinates.
[0141] Within the plastic deformation segment of the hysteresis loop, first select the i-th total strain coordinate, and then take points at the same total strain coordinate on both the increasing strain support and the decreasing strain support. and The points near the inflection point and the near-elastic segment are not included in the point selection. Multiple coordinates are selected at equal intervals along the effective plastic segment to obtain a set of total back stress data for fitting. Point and The actual stress at point is denoted as . and The average of the two values yields the total back stress at that coordinate point. .
[0142] The figure schematically marks the locations of different total strains. Point and The figure illustrates the method for extracting the total back stress data points from the experiment. The extracted back stress data is used to calibrate the back stress evolution parameters of the kinematic hardening model, enabling the model to accurately describe the translational behavior of the yield surface under cyclic loading. As the total strain changes from negative to positive, the corresponding actual stress varies along the hysteresis loop path. The stress differences at each point reflect the evolution of the back stress during cyclic loading.
[0143] By fitting the kinematic hardening modulus and dynamic recovery coefficient of multiple back stress components using a cyclic loading stable hysteresis loop, the movement of the yield surface center and the Bauschinger effect under kinematic hardening can be accurately described, thereby improving the simulation accuracy of the material mechanical response under cyclic loading conditions.
[0144] According to an embodiment of the present invention, the parameters of the damage model are determined based on the unloading stiffness degradation data in the stress-strain data, including: obtaining the unloading stiffness degradation data of the graded loading and unloading test from the stress-strain data, determining the plastic strain of each data point in the plastic deformation stage based on the graded loading and unloading test data, and accumulating the cumulative plastic strain, as well as determining the initial elastic modulus of the first loading and the unloading modulus corresponding to each unloading; calculating the test damage variable based on the unloading modulus corresponding to each unloading and the initial elastic modulus; using the damage model, when the cumulative plastic strain is less than or equal to the damage threshold, the damage variable remains zero; when the cumulative plastic strain is greater than the damage threshold, the damage variable remains zero. Under these conditions, the damage energy release rate is determined based on stress-strain data. The damage variables are then updated incrementally based on the damage energy release rate, the damage energy scale parameter and damage evolution index in the damage model, and the increment of cumulative plastic strain between adjacent data points. The damage threshold, damage energy scale parameter, and damage evolution index are the parameters to be fitted to the damage model. With the goal of minimizing the error between the damage variables and the experimental damage variables, the parameters to be fitted are fitted to obtain the fitted values for the damage threshold, damage energy scale parameter, and damage evolution index. These fitted values are then used as the parameters of the damage model.
[0145] Damage models are used to describe the evolution of internal damage in stainless steel during plastic deformation. In the Lemaître damage model, the damage variable D characterizes the degree of stiffness degradation caused by the evolution of microcracks and micropores within the material, ranging from 0 (no damage) to 1 (complete fracture). The parameters to be fitted in the damage model include the damage threshold. Damage energy scale parameters And the damage evolution index s.
[0146] The unloading stiffness degradation data of the graded loading-unloading test are obtained from the stress-strain data. The graded loading-unloading test is a test method that sets up multiple unloading-reloading cycles during monotonic tensile process, and measures the stiffness degradation of the material by measuring the slope change of each unloading segment. Based on the graded loading-unloading test data, the plastic strain of each data point in the plastic deformation stage is determined, and the cumulative plastic strain is obtained by summing them. Specifically, for the data point corresponding to the i-th loading-unloading cycle, it is processed as follows.
[0147] Plastic strain is obtained by subtracting elastic strain from total strain:
[0148] ;
[0149] Accumulated plastic strain is obtained by progressively adding the increments of plastic strain point by point:
[0150] ;
[0151] Simultaneously, the initial elastic modulus of the first loading and the unloading modulus corresponding to each unloading are determined. The initial elastic modulus of the elastic segment during the first loading... The stress-strain curve of the first loading stage is obtained by linear regression of the elastic segment. For each unloading stage in the graded loading and unloading test, linear regression is performed on the stress-strain data of the unloading stage to obtain the unloading modulus corresponding to that unloading. .
[0152] Then, based on the unloading modulus and initial elastic modulus corresponding to each unloading, the experimental damage variables are calculated. In damage mechanics, the degree of stiffness degradation of a material can be characterized by the decrease in the unloading modulus. When the material is not damaged, the unloading modulus is equal to the initial elastic modulus; as damage develops, the unloading modulus gradually decreases. The experimental damage variables are calculated using the following formula:
[0153] ;
[0154] in, Let i be the experimental damage variable corresponding to the i-th unloading. Let be the unloading modulus obtained from the linear regression of the i-th unloading segment. This is the initial elastic modulus obtained by first loading the elastic segment. The physical meaning of this formula is that the damage variable is equal to the ratio of stiffness degradation to the initial stiffness.
[0155] When updating damage variables using a damage model, the evolution of damage variables with cumulative plastic strain follows these rules: When the cumulative plastic strain is less than or equal to the damage threshold... In this case, the material has not yet been damaged, and the damage variable remains at zero:
[0156] ;
[0157] When the accumulated plastic strain exceeds the damage threshold Under these conditions, damage begins to initiate and gradually evolves, with the damage variable increasing with the increase of cumulative plastic strain. At this point, the damage energy release rate Y is determined based on stress-strain data. The damage energy release rate is the driving force of damage and characterizes the degree of influence of the stress state on damage evolution. In the Lemaître damage model, the damage energy release rate is calculated using the following formula:
[0158] ;
[0159] in, The equivalent stress calculated using the anisotropic yield criterion (Hill 48 yield criterion), where E is the elastic modulus. Poisson's ratio, This represents the hydrostatic stress (i.e., the average of the normal stresses along the three principal axes). .
[0160] Then, based on the damage energy release rate Y and the damage energy scale parameter And the damage evolution index s, and the increment of cumulative plastic strain between adjacent data points. The damage variables are updated incrementally.
[0161] The evolution of damage variables with cumulative plastic strain is described by the following rate-form governing equation:
[0162] ;
[0163] in, The rate of change of the damage variable, To accumulate plastic strain rate, The damage threshold, is the damage energy scale parameter, and s is the damage evolution index.
[0164] In numerical implementation, the above rate-form equation is discretized and solved using the following incremental approach:
[0165] ;
[0166] in, and These are the damage variables at the beginning and end of the increment, respectively. The damage energy release rate calculated at the end of the increment. This represents the increase in cumulative plastic strain within the current increment. Damage threshold. Critical cumulative plastic strain controlling damage initiation, damage energy scale parameter The magnitude of the damage evolution rate is controlled by the damage evolution index s, which controls the nonlinearity of the damage evolution. All three are parameters to be fitted in the damage model.
[0167] The goal is to minimize the error between the model damage variable and the experimental damage variable, and to fit the parameters. , We fit the damage parameters to s. The error function for fitting the damage parameters is defined as follows:
[0168] ;
[0169] in, This represents the number of effective unloading segments that meet the linear regression requirements in a graded unloading test. Let i be the experimental damage variable corresponding to the i-th unloading. The damage variables are calculated by the damage model using the incremental update method described above after inputting the same experimental strain history.
[0170] During the fitting process, the experimentally known quantity is the initial elastic modulus. and each unloading modulus Therefore, the experimental damage variables corresponding to each unloading are calculated. The parameter to be fitted is the damage threshold. Damage energy scale parameters Damage evolution index s; model damage variables It is calculated incrementally from the damage model.
[0171] The damage threshold is determined by minimizing the above error function. Damage energy scale parameters The optimal fitting values for the damage evolution index s were determined. Finally, the fitting values for the damage threshold, damage energy scale parameter, and damage evolution index were used together as parameters for the damage model.
[0172] By utilizing stiffness degradation data from graded loading and unloading to determine the damage threshold, damage energy scale parameter, and damage evolution index, the damage evolution of materials during plastic deformation can be quantitatively characterized, providing reliable parameters for introducing anisotropic yield criteria for damage variables.
[0173] According to an embodiment of the present invention, with the goal of minimizing the error between the stress-strain results predicted by the constitutive model and the stress-strain data, the anisotropy coefficient, the parameters of the hardening model, and the parameters of the damage model are optimized to obtain target material parameters. This includes: performing a global sensitivity analysis on each parameter among the anisotropy coefficient, the parameters of the hardening model, and the parameters of the damage model to determine multiple key parameters from the anisotropy coefficient, the parameters of the hardening model, and the parameters of the damage model based on the degree of influence of each parameter on the stress-strain results predicted by the constitutive model; classifying the multiple key parameters into anisotropic yield parameter group, hardening parameter group, and damage parameter group according to their physical origin, wherein each of the anisotropic yield parameter group, hardening parameter group, and damage parameter group includes at least one key parameter, and the stress-strain data includes multiple sets of experimental data, each set of experimental data being... Do not associate the anisotropic yield parameter group, hardening parameter group, or damage parameter group with the actual parameters. Repeat the following steps until the error between the stress-strain result predicted by the constitutive model and the stress-strain data meets the preset convergence condition. Obtain the optimized values of the anisotropic yield parameter group, hardening parameter group, and damage parameter group, and use the optimized values as the target material parameters. Optimize the anisotropic yield parameter group, hardening parameter group, and damage parameter group in sequence. When optimizing the current parameter group, fix the parameter values of other parameter groups and adjust the parameters in the current parameter group to minimize the error between the stress-strain result predicted by the constitutive model and the experimental data corresponding to the current parameter group. During the error calculation process, dynamically adjust the weight of each set of experimental data according to the normalized error between each set of experimental data and the stress-strain result predicted by the constitutive model.
[0174] A global sensitivity analysis was performed on each parameter in the anisotropy coefficient, the hardening model parameters, and the damage model parameters. Specifically, the exponential method was used to quantitatively assess the contribution of each parameter to the uncertainty of the stress-strain results predicted by the constitutive model by calculating the first-order and total-order exponents of each parameter. The first-order exponent reflects the impact of a single parameter changing independently on the model output, while the total-order exponent reflects the total impact of the parameter and its interaction with other parameters on the model output. Based on the calculation results, several parameters with the greatest impact on the model output were selected as key parameters, thereby reducing the dimensionality of subsequent parameters to be optimized and improving optimization efficiency.
[0175] The selected key parameters were categorized into three groups based on their physical origin: anisotropic yield parameters, hardening parameters, and damage parameters. The anisotropic yield parameters group includes the anisotropy coefficient from the anisotropic yield criterion, expressed as... Where F, G, and H are the anisotropy coefficients of the normal stress term, and L, M, and N are the anisotropy coefficients of the shear stress term. The hardening parameter set includes the parameters of the hardening model. When the hardening model is an isotropic hardening model (Voce model), it is represented as follows: ,in The initial flow stress, Let be the saturation hardening increment, and b be the hardening rate parameter. When the hardening model is a kinematic hardening model (Chaboche model), it is expressed as: ,in Let be the kinematic hardening modulus of each back stress component. These are the dynamic recovery coefficients for each back stress component. The damage parameter set includes the parameters of the damage model, expressed as... ,in The damage threshold, is the damage energy scale parameter, and s is the damage evolution index.
[0176] In addition, elastic parameters The stress-strain data is determined directly by experiments and is not involved in subsequent optimization. The stress-strain data includes multiple sets of experimental data, such as uniaxial tensile test data, cyclic loading and unloading test data, and graded loading and unloading damage test data. Each set of experimental data corresponds to a specific set of parameters. The anisotropic yield parameter set mainly corresponds to uniaxial tensile and torsion test data, the hardening parameter set mainly corresponds to monotonic tensile or cyclic loading test data, and the damage parameter set mainly corresponds to graded loading and unloading damage test data.
[0177] The anisotropic yield parameter group, hardening parameter group, and damage parameter group are optimized sequentially. When optimizing the current parameter group, the parameter values of other parameter groups are kept constant; only the parameters in the current parameter group are adjusted to minimize the error between the stress-strain results predicted by the constitutive model and the corresponding experimental data. This process is repeated until the total error between the stress-strain results predicted by the constitutive model and the stress-strain data satisfies a preset convergence condition. The optimized values of the anisotropic yield parameter group, hardening parameter group, and damage parameter group are then obtained and used as the target material parameters. The preset convergence condition can be that the objective function value is less than a preset threshold, or that the change in the objective function value between two consecutive iterations is less than a preset threshold.
[0178] A dynamic weight adjustment strategy is employed in the error calculation process. Specifically, the following objective function is established to quantify the total error between the constitutive model's prediction results and the experimental data. :
[0179] ;
[0180] Where θ is the parameter vector to be optimized in the current stage. The number of experimental data groups, Let k be the number of data points in the k-th experimental group. Let i be the experimental value of the i-th data point in the k-th experimental data set. This corresponds to the model's predicted value. The normalization factor is the maximum absolute value of the response in the k-th test group. Let be the dynamic weights for the k-th group of experimental data. In the first round of optimization, the weights for each group... Pick .
[0181] In subsequent optimization rounds, the weights of each group are dynamically updated based on the proportion of the normalized error between the experimental data and the model predictions from the previous round to the total normalized error. Data groups with larger normalized errors are assigned higher weights. This dynamic weight adjustment strategy ensures that data groups with larger fitting errors receive more attention during the optimization process, thereby balancing the contribution of different types of experimental data to parameter optimization and reducing the dominance of any one type of data due to its large numerical magnitude.
[0182] By using the above-mentioned global sensitivity analysis to screen key parameters, to optimize each parameter group in stages, and an adaptive strategy of dynamic weight adjustment, the final target material parameters can make the constitutive model fit the best with all input experimental data in a global sense.
[0183] Figure 7A flowchart illustrating the constitutive model construction process of the numerical implementation method for the adaptive constitutive model of additive manufacturing stainless steel material according to an embodiment of the present invention is shown.
[0184] like Figure 7 As shown, this embodiment includes operations S701 to S711.
[0185] In operation S701, input the predetermined material parameters, including the anisotropy coefficient, the parameters of the isotropic hardening model, and the parameters of the damage model, and read in the macroscopic mechanical test data.
[0186] In operation S702, stress, cumulative plastic strain, and damage variables are initialized to zero as the initial state at the start of the current load increment step.
[0187] In operation S703, the test elastic strain is calculated based on the total strain increment of the current load increment step and the elastic strain at the end of the previous increment step. The test elastic strain is then multiplied by the elastic stiffness matrix to obtain the test stress.
[0188] In operation S704, the test stress is substituted into the yield function determined by the anisotropic yield criterion, hardening model and damage model to calculate the yield function value under the test state.
[0189] In operation S705, determine whether the trial state has yielded. If not, proceed to operation S706; if yes, proceed to operation S707.
[0190] In operation S706, the trial stress is used as the stress of the current load increment step, while the cumulative plastic strain, damage variable, and hardening variable remain unchanged. The elastic update of the current increment step is completed, and the process returns to operation S703.
[0191] When operating S707, the Newton-Raphson iteration method is used to solve for the plastic multiplier increment, and the yield function is brought back to zero through iterative calculation.
[0192] In operation S708, determine whether the residuals of the Newton iteration meet the preset convergence conditions. If yes, proceed to operation S709; otherwise, proceed to operation S707 and continue the iteration.
[0193] In operation S709, the stress, cumulative plastic strain, damage variable, and flow stress are updated, and the values after iterative convergence are used as the final result of the current load increment step.
[0194] In operation S710, the elastoplastic uniform tangent modulus is calculated based on the updated state variables and used for iterative calculation of the next load increment step.
[0195] In operation S711, the stress, cumulative plastic strain, damage variable, and elastoplastic consistent tangent modulus of the current load increment step are output as numerical solution results.
[0196] According to an embodiment of the present invention, the target material parameters are substituted into the constitutive model for numerical solution to obtain the mechanical response results for structural performance evaluation. This includes: substituting the anisotropy coefficient in the target material parameters into the anisotropic yield criterion; substituting the parameters of the hardening model in the target material parameters into the hardening model; substituting the parameters of the damage model in the target material parameters into the damage model; obtaining the load acting on the stainless steel structural member and dividing the load into multiple load increment steps; initializing the stress, strain, cumulative plastic strain, and damage variable of the structural member to zero, and initializing the hardening variable according to the type of hardening model. Specifically, when the hardening model is a kinematic hardening model, multiple back stress components in the kinematic hardening model are initialized to zero, and each back stress component is used as the hardening variable; when the hardening model is an isotropic hardening model, the hardening variable is the cumulative plastic strain; and determining the elastic stiffness matrix of the stainless steel material based on the elastic stage data in the stress-strain data.
[0197] For each load increment step, perform the following operations: Calculate the test stress using the elastic stiffness matrix based on the stress, strain, cumulative plastic strain, damage variable, and hardening variable at the end of the previous load increment step, and the total strain increment corresponding to the current load increment step; substitute the test stress into the yield function jointly determined by the anisotropic yield criterion, the hardening model, and the damage model to determine whether the test stress satisfies the yield condition. The yield function is obtained by calculating the equivalent stress using the anisotropic yield criterion, calculating the flow stress using the hardening model, and performing damage correction on the equivalent stress using the damage model.
[0198] If the test stress does not meet the yield condition, the test stress is used as the stress of the current load increment step, and the cumulative plastic strain, damage variable, and hardening variable remain unchanged. If the test stress meets the yield condition, the current load increment step is subjected to plastic correction iteration. During the iteration process, the hardening variable is updated using the parameters of the hardening model, the damage variable is updated using the parameters of the damage model, and the stress and cumulative plastic strain are adjusted until the yield function meets the preset convergence condition.
[0199] If the yield function satisfies the preset convergence condition, the elastic-plastic uniform tangent modulus is calculated for the iterative calculation of the next load increment step. After completing the iterative calculation of all load increment steps, the stress, strain, damage variables and displacement results of the structural members are output as mechanical response results for structural performance evaluation.
[0200] According to an embodiment of the present invention, after obtaining the target material parameters, the target material parameters are substituted into the constitutive model for numerical solution to obtain the mechanical response results for structural performance evaluation. This numerical solution process involves discretizing the continuous external load into multiple load increment steps, and updating the stress and state variables in each increment step using a two-stage strategy of elastic prediction and plastic correction.
[0201] Substitute the anisotropy coefficients from the target material parameters into the anisotropic yield criterion, substitute the parameters of the hardening model from the target material parameters into the hardening model, and substitute the parameters of the damage model from the target material parameters into the damage model. At this point, all parameters of the constitutive model have been determined. This constitutive model is composed of the anisotropic yield criterion, the hardening model, and the damage model.
[0202] The load acting on the stainless steel structural member is obtained and divided into multiple load increment steps. The load increment steps are divided as follows: the total load is divided into several sub-steps based on time or amplitude, with each sub-step corresponding to a total strain increment. The number of load increment steps directly affects the accuracy and efficiency of numerical solutions. More increment steps result in higher computational accuracy but also a greater computational burden; conversely, fewer increment steps lead to higher computational efficiency but lower accuracy. In engineering practice, the number of load increment steps is determined based on the required accuracy.
[0203] Before starting the incremental step iteration, the stress of the structural components is... ,strain The cumulative plastic strain p and damage variable D are initialized to zero. Simultaneously, the hardening variables are initialized according to the type of hardening model: in the case of a kinematic hardening model, multiple back stress components in the kinematic hardening model are initialized. , and Initialized to zero, and each back stress component is used as the hardening variable. When the hardening model is an isotropic hardening model (i.e., the Voce hardening model), the hardening variable is the cumulative plastic strain p. The cumulative plastic strain p and the back stress components... All of these are historically relevant state variables, and their initial values are all zero.
[0204] The elastic stiffness matrix of the stainless steel material is determined based on the elastic stage data in the stress-strain data. Specifically, the elastic modulus of the stainless steel material in multiple preset principal axis directions is first determined based on the elastic stage data in the stress-strain data. , , and shear modulus in multiple principal axis perpendicular planes , , The modulus of elasticity is obtained from the slope of the elastic segment of the uniaxial tensile curve in the corresponding direction:
[0205] ;
[0206] The shear modulus is obtained from the slope of the elastic segment of the corresponding pure shear curve:
[0207] ;
[0208] Where Poisson's ratio is taken as 0.3, and according to the orthogonal anisotropic elastic symmetry condition, the symmetry requirement of the compliance matrix is... Therefore, the conversion relationship for the inverse Poisson's ratio is: .in, , , Take 0.3, , , The value is determined by the ratio of elastic moduli according to the above conversion relationship.
[0209] Then construct the orthogonal anisotropic compliance matrix. And then Inverse the matrix to obtain the elastic stiffness matrix The orthogonal anisotropic compliance matrix is in the form of:
[0210] ;
[0211] The elastic stiffness matrix is: .
[0212] For each load increment step, the test stress is calculated using the elastic stiffness matrix based on the stress, strain, cumulative plastic strain, damage variable, and hardening variable at the end of the previous load increment step, as well as the total strain increment corresponding to the current load increment step. Specifically, the test elastic strain is first calculated. :
[0213] ;
[0214] Then calculate the trial stress:
[0215] ;
[0216] in, The elastic strain after the nth increment converges. The total strain increment given by this increment, Here is the elastic stiffness matrix. The calculated test stress.
[0217] The test stress is substituted into the yield function, which is jointly determined by the anisotropic yield criterion, the hardening model, and the damage model, to determine whether the test stress satisfies the yield condition. The yield function is obtained by calculating the equivalent stress using the anisotropic yield criterion, the flow stress using the hardening model, and then applying damage correction to the equivalent stress using the damage model. Specifically, the form of the yield function is:
[0218] ;
[0219] in, This represents the yield function considering damage coupling effects. The equivalent stress calculated based on the test stress for the anisotropic yield criterion (Hill 48 yield criterion) is expressed as follows:
[0220] ;
[0221] in, , , Represents three normal stress components. , , This represents the three shear stress components.
[0222] D represents the current damage variable. The flow stress is calculated by the hardening model based on the current accumulated plastic strain p. When When ≤0, the material is in an elastic state; when When the value is greater than 0, the material yields.
[0223] When the test stress does not meet the yield condition (i.e.) When the stress is ≤0, the trial stress is used as the stress in the current load increment step, while keeping the cumulative plastic strain, damage variable, and hardening variable constant; that is, the current increment step is a purely elastic step. In this case, the tangent stiffness of the current increment step is the elastic stiffness matrix. .
[0224] When the test stress satisfies the yield condition (i.e.) When the load increment is greater than 0, a plastic correction iteration is performed on the current load increment step. The plastic correction iteration uses Newton's iteration method, which solves for the plastic multiplier increment. This brings the yield function back to zero. In each iteration, the consistency residual is calculated. and its derivative with respect to the increment of the plastic multiplier , and according to Update the plastic multiplier increments until the convergence condition is met.
[0225] In each iteration, the plastic strain increment is first calculated based on the plastic multiplier increment of the current iteration:
[0226] ;
[0227] in, This represents the initial value of the plastic multiplier increment before the start of the plastic correction iteration (iteration 0), where m is the direction of plastic flow. Then, the cumulative plastic strain increment is calculated:
[0228] ;
[0229] Determine the update stress: .
[0230] During the iteration process, the hardening variables are updated using the parameters of the hardening model. For isotropic hardening models, the cumulative plastic strain p is updated; for kinematic hardening models, each back stress component is updated using the following formula:
[0231] ;
[0232] Update the damage variable using the parameters of the damage model. When hour, ;when First, calculate the damage energy release rate Y according to the formula, and then update the damage variable according to the following formula:
[0233] ;
[0234] Simultaneously adjust the stress and cumulative plastic strain until the yield function satisfies the preset convergence condition. The convergence condition is:
[0235] ;
[0236] in, For consistent residuals, To preset the convergence tolerance (usually taken as...) to The iteration ends when the convergence condition is met, and the plastic multiplier increment of the current increment is... This is the value after convergence.
[0237] Under the condition that the yield function satisfies the preset convergence condition, the elastoplastic uniform tangent modulus is calculated for iterative calculation in the next load increment step. In the case of static constitutive model (isotropic hardening), the elastoplastic uniform tangent modulus is calculated using the following formula:
[0238] ;
[0239] in, For isotropic hardening tangent modulus, This represents the elastoplastic uniform tangential modulus under isotropic hardening conditions. This represents the elastic stiffness tensor. In the case of cyclic constitutive modeling (kinematic hardening), the elastoplastic uniform tangent modulus must also consider the contribution of kinematic hardening.
[0240] ;
[0241] ;
[0242] in, The contribution of kinematic hardening tangent reflects the tangent response of back stress to the increment of the plastic multiplier. This represents the elastoplastic uniform tangent modulus under cyclic constitutive (kinematic hardening) conditions. This represents the contribution of the kinematic hardening tangent.
[0243] After completing iterative calculations for all load increment steps, the stress, strain, cumulative plastic strain, damage variable, back stress components (if a kinematic hardening model) and displacement results of the structural members are output as mechanical response results for structural performance evaluation. These results can comprehensively reflect the stress state, deformation mode, plastic development region, and damage evolution distribution of the structural members under external loads, and can thus be used to evaluate the load-bearing capacity, deformation capacity, and damage state of the structural members.
[0244] Figure 8A A schematic diagram illustrating the parameter acquisition of the uniaxial tensile elastic segment in the numerical implementation method of the adaptive constitutive model for additive manufacturing of stainless steel materials according to an embodiment of the present invention is shown.
[0245] Figure 8B A schematic diagram illustrating the parameter acquisition of the planar pure shear elastic segment in the numerical implementation method of the adaptive constitutive model of additive manufacturing stainless steel material according to an embodiment of the present invention is shown.
[0246] like Figure 8A The figure shows the stress-strain curves of the uniaxial tensile elastic segment in three preset principal axis directions (0°, 90°, and thickness). The horizontal axis represents the elastic axial strain, and the vertical axis represents the actual stress. The three curves represent the uniaxial tensile elastic response of stainless steel in the three preset principal axis directions. As can be seen from the figure, during the elastic deformation stage, the stress in all three directions increases linearly with increasing strain, and the slopes of the three curves are basically the same, indicating that the elastic modulus of the material is similar in different principal axis directions.
[0247] In the 0° uniaxial tensile elastic segment 801, the stress increases linearly with the increase of strain, and the slope of this elastic segment is the elastic modulus in the 0° direction. This can be obtained by calculating the ratio of the stress increment to the strain increment in that segment.
[0248] In the 90° uniaxial tensile elastic segment 802, the stress also increases linearly with the increase of strain, and the slope of this elastic segment is the elastic modulus in the 90° direction. This can be obtained by calculating the ratio of the stress increment to the strain increment in that segment.
[0249] In the uniaxial tensile elastic segment 803 along the thickness direction, the stress increases linearly with the increase of strain, and the slope of this elastic segment is the elastic modulus in the thickness direction. This can be obtained by calculating the ratio of the stress increment to the strain increment in that segment.
[0250] The slopes of the elastic segments in the three directions are basically the same, indicating that the elastic modulus of the material is similar in different principal axis directions. The obtained elastic modulus is used to construct the orthogonal anisotropic elastic stiffness matrix.
[0251] like Figure 8B The figure shows the shear stress-shear strain curves for the pure shear elastic segment in three principal axis vertical planes (planes 1-2, 1-3, and 2-3). The horizontal axis represents engineering shear strain, and the vertical axis represents shear stress. The three curves represent the pure shear elastic response of stainless steel in the three principal axis vertical planes. As can be seen from the figure, during the elastic deformation stage, the shear stress in all three planes increases linearly with increasing shear strain, and the slopes of the three curves are basically the same, indicating that the shear modulus of this material is similar in different principal axis vertical planes.
[0252] In the pure shear elastic segment 804 of the 1-2 plane, the shear stress increases linearly with the increase of shear strain. The slope of this elastic segment is the shear modulus in the 1-2 plane. The ratio of the increase in shear stress to the increase in shear strain in that segment can be calculated.
[0253] In the purely shear elastic segment 805 of the 1-3 plane, the shear stress also increases linearly with the increase of shear strain. The slope of this elastic segment is the shear modulus in the 1-3 plane. The ratio of the increase in shear stress to the increase in shear strain in that segment can be calculated.
[0254] In the pure shear elastic segment 806 of the 2-3 plane, the shear stress increases linearly with the increase of shear strain. The slope of this elastic segment is the shear modulus in the 2-3 plane. The ratio of the increase in shear stress to the increase in shear strain in that segment can be calculated.
[0255] The slopes of the elastic segments in the three planes are basically the same, indicating that the shear modulus of the material is similar in different principal axis perpendicular planes. The obtained shear modulus is used to construct the orthogonal anisotropic elastic stiffness matrix.
[0256] By substituting the target material parameters into the constitutive model and using load increment steps, elastic trial stress, and plastic correction iterations for numerical solution, the stress, strain, damage, and hardening evolution process of structural components under load can be accurately simulated, thereby improving the accuracy of structural load-bearing performance assessment.
[0257] Figure 8C A flowchart illustrating the multi-parameter adaptive optimization process of the numerical implementation method of the adaptive constitutive model for additive manufacturing stainless steel materials according to an embodiment of the present invention is shown.
[0258] like Figure 8C As shown, this embodiment includes operations S801 to S809.
[0259] When operating S801, input multiple sets of experimental data and the initial values of each parameter to be optimized.
[0260] In operating S802, a global sensitivity analysis was performed on each parameter to be optimized. The first-order Sobol exponent and the total-order Sobol exponent of each parameter were calculated using the Sobol exponent method, and several parameters with the greatest impact on the model output were selected as key parameters.
[0261] When operating S803, the selected key parameters are divided into anisotropic yield parameter group, hardening parameter group and damage parameter group according to their physical source. The parameters are optimized in the following order: first anisotropic yield parameter group, then hardening parameter group and then damage parameter group. When optimizing the current parameter group, the parameter values of other parameter groups are kept unchanged.
[0262] In operation S804, during the current stage of optimization, the parameters in the current parameter set are adjusted to minimize the error between the stress and strain results predicted by the constitutive model and the experimental data corresponding to the current parameter set, and the objective function value for the current stage is calculated.
[0263] In operation S805, determine whether the objective function value satisfies the preset convergence condition. If yes, proceed to operation S807; otherwise, proceed to operation S806.
[0264] In operation S806, the weight of each group of experimental data in the objective function is dynamically adjusted according to the proportion of the normalized error between each group of experimental data and the model prediction results to the total normalized error. The data group with the larger normalized error is assigned a higher weight. After adjustment, the process returns to operation S804 to continue optimization.
[0265] In operation S807, the optimal value of the current parameter group is output as the optimization result of that parameter group.
[0266] In operation S808, after obtaining the optimized values of all parameter groups, the optimized values of each parameter group are merged as the target material parameters. The posterior distribution samples of each parameter are obtained through Markov chain Monte Carlo sampling, and the posterior confidence interval of each parameter is determined accordingly to evaluate the uncertainty of the target material parameters.
[0267] In operation S809, the target material parameters are used to form a material parameter library for numerical simulation of structural components.
[0268] In another embodiment, for cyclic loading conditions, a cyclic constitutive solution scheme using a kinematic hardening model is employed. In this case, the yield function considers the influence of back stress and takes the following form:
[0269] ;
[0270] in, = + + The total back stress is obtained by superimposing multiple back stress components; η is the relative stress, representing the effective stress after subtracting the total back stress from the current stress. Using the Voigt order [11, 22, 33, 23, 13, 12], the relative stress vector is written as... The Hill anisotropy metric matrix A corresponds in the same order and has the following form:
[0271] ;
[0272] Where F, G, H, L, M, and N are anisotropy coefficients, determined by the method described above. Under cyclic loading conditions, the back stress evolves with plastic deformation and is updated incrementally according to the back stress update formula of the aforementioned kinematic hardening model. The numerical solution process for this cyclic constitutive model is consistent with that of the static constitutive model, also employing the load increment step, elastic trial stress, and plastic correction iteration strategy. The difference lies in the introduction of a back stress term in the yield function to describe the Bauschinger effect under cyclic loading.
[0273] Figure 8D A flowchart illustrating the cyclic constitutive model construction process of the numerical implementation method for the adaptive constitutive model of additive manufacturing stainless steel material according to an embodiment of the present invention is shown.
[0274] like Figure 8D As shown, this embodiment includes operations S811 to S821.
[0275] In operation S811, input the predetermined material parameters, including the anisotropy coefficient, the parameters of the kinematic hardening model, and the parameters of the damage model, and read in the cyclic loading test data.
[0276] In operation S812, stress, cumulative plastic strain, damage variable and multiple back stress components are initialized to zero.
[0277] In operation S813, the test elastic strain is calculated based on the total strain increment of the current load increment step and the elastic strain at the end of the previous increment step, and multiplied by the elastic stiffness matrix to obtain the test stress. Then, the test relative stress is calculated based on the total back stress at the end of the previous increment step.
[0278] In operation S814, the test relative stress is substituted into the cyclic yield function determined by the anisotropic yield criterion, the kinematic hardening model and the damage model, and the cyclic yield function value under the test state is calculated.
[0279] In operation S815, determine whether the trial state has yielded. If not, proceed to operation S816; if so, proceed to operation S817.
[0280] In operation S816, the test stress is used as the stress of the current load increment step, the cumulative plastic strain, damage variable and back stress components are kept unchanged, and the operation returns to S813.
[0281] In operating S817, the Newton-Raphson iteration method is used to synchronously update stress, cumulative plastic strain, damage variables, and multiple back stress components.
[0282] In operation S818, determine whether the residuals of the Newton iteration meet the preset convergence conditions. If yes, proceed to operation S819; otherwise, return to operation S817 to continue iteration.
[0283] In operation S819, the stress, cumulative plastic strain, damage variables, and multiple back stress components after iteration convergence are used as the final result of the current load increment step to update each state variable.
[0284] In operation S820, the cyclic elastoplastic uniform tangent modulus is calculated based on the updated state variables and used for iterative calculation of the next load increment step.
[0285] In operation S821, the stress, cumulative plastic strain, damage variable, back stress component, and cyclic elastoplastic consistent tangent modulus of the current load increment step are output as the numerical solution results of cyclic loading.
[0286] According to an embodiment of the present invention, the test stress is substituted into a yield function jointly determined by the anisotropic yield criterion, the hardening model, and the damage model to determine whether the test stress satisfies the yield condition. This includes: calculating the equivalent stress based on the test stress using the anisotropic yield criterion; performing damage correction on the equivalent stress based on the damage variables using the damage model to obtain the damage-corrected equivalent stress; calculating the flow stress based on the cumulative plastic strain using the hardening model; and comparing the damage-corrected equivalent stress with the flow stress to determine whether the test stress satisfies the yield condition if the damage-corrected equivalent stress is greater than the flow stress.
[0287] The equivalent stress is calculated based on the test stress using the anisotropic yield criterion. Specifically, the Hill 48 yield criterion is adopted, which considers the six stress components (including three normal stress components) under the current test stress state. , , and three shear stress components , , Substituting these values into the expression for the Hill 48 yield criterion, the equivalent stress is calculated. The formula for calculating the equivalent stress of Hill 48 is:
[0288] ;
[0289] Wherein, F, G, and H are the anisotropy coefficients of the normal stress term, and L, M, and N are the anisotropy coefficients of the shear stress term. These coefficients have been determined by the aforementioned method based on the yield strength and shear yield strength of stainless steel material in different preset principal axis directions.
[0290] Damage correction is applied to the equivalent stress using a damage model based on damage variables. In the Lemaître damage model, the concept of effective stress describes the weakening effect of damage on the material's load-bearing capacity. When damage exists within the material, the effective area actually bearing the load decreases, thus increasing the effective stress under the same external load. Specifically, the Hill 48 equivalent stress is... Divide by (1-D) to obtain the equivalent stress after damage correction. :
[0291] ;
[0292] Where D is the current damage variable, and its value ranges from 0 (no damage) to 1 (complete fracture). When D=0, the material is undamaged, and the equivalent stress after damage correction is equal to the original equivalent stress; when D>0, the equivalent stress after damage correction is greater than the original equivalent stress, reflecting the weakening effect of damage on the material's load-bearing capacity.
[0293] Simultaneously, the hardening model is used to calculate the model flow stress based on the current cumulative plastic strain p. The hardening model can be either an isotropic hardening model (Voce hardening model) or a kinematic hardening model (Chaboche hybrid hardening model). In the case of the isotropic hardening model, the flow stress is calculated by the following equation:
[0294] ;
[0295] in, The initial flow stress, denoted as saturated hardening increment, b as hardening rate parameter, and p as cumulative plastic strain. In the case of the kinematic hardening model, back stress is introduced into the hardening model to describe the translation of the yield surface, but the flow stress used to determine yield is still calculated based on the isotropic hardening part, and the influence of back stress is reflected in the relative stress term in the yield function.
[0296] The equivalent stress after damage correction With flow stress A comparison is made. If the equivalent stress after damage correction is less than or equal to the flow stress, i.e. ≤ If the material is currently in an elastic state, the test stress does not meet the yield criterion. If the equivalent stress after damage correction is greater than the flow stress, i.e. > If the test stress is found to be in the plastic state, then the material has entered a plastic state, confirming that the test stress satisfies the yield condition. This comparison process can be uniformly expressed in the form of a yield function:
[0297] ;
[0298] when When ≤0, the material is in an elastic state; when When the stress value is greater than 0, the material yields. This yield function simultaneously reflects the calculation of equivalent stress by the anisotropic yield criterion, the damage correction of equivalent stress by the damage model, and the calculation of flow stress by the hardening model. The combined effect of these three factors determines whether the material enters the plastic state under the current stress condition.
[0299] By using damage variables to correct the equivalent stress and comparing the damage-corrected equivalent stress with the flow stress calculated by the hardening model, the synergistic effect of anisotropic yielding, hardening, and damage in yield judgment is realized, making the yield judgment more consistent with the actual behavior of material damage weakening.
[0300] According to an embodiment of the present invention, the numerical implementation method of the adaptive constitutive model for additive manufacturing stainless steel further includes: constructing a likelihood function based on the normalized residual between the stress-strain results predicted by the constitutive model and the stress-strain data; setting the prior distribution of each parameter in the target material parameters; performing Markov chain Monte Carlo sampling based on the likelihood function and the prior distribution to obtain the posterior distribution samples of each parameter in the target material parameters; and determining the posterior confidence interval of each parameter in the target material parameters based on the posterior distribution samples to evaluate the uncertainty of the target material parameters.
[0301] After optimizing the target material parameters, in order to further quantify the parameter uncertainties caused by process fluctuations and experimental errors, this invention uses the Markov chain Monte Carlo method to perform uncertainty analysis on the optimized target material parameters.
[0302] A likelihood function is constructed based on the normalized residuals between the stress-strain results predicted by the constitutive model and the stress-strain data. Within the Bayesian inference framework, the likelihood function is used to quantify stress-strain conditions given a set of parameters. Under the given conditions, the degree of agreement between the model's predictions and the experimental data is considered. Specifically, the normalized residuals used in the optimization process are used as inputs to the likelihood function. The normalized residuals are calculated using the following formula:
[0303] ;
[0304] in, The normalized residual of the i-th data point in the k-th experimental data set is... For experimental data points, This corresponds to the model's predicted value. Let $\frac{k}{k}$ be the maximum absolute value of the response in the $k$-th experimental group (i.e., the normalization factor). Based on this, assuming that the residuals of each data group are independent and follow a normal distribution with a mean of zero, we construct the likelihood function. ,in This represents all experimental data involved in the parameter inversion. Let be the parameter vector to be inverted. represents the standard deviation of the residuals for each data set.
[0305] Define the prior distribution of each parameter in the target material parameters. A prior distribution reflects a reasonable estimate of the range of parameter values based on existing knowledge or experience before experimental data are observed. The prior distribution can be set using a uniform distribution (uniformly distributed within the physically permissible range), a normal distribution (with the mean and standard deviation determined based on existing literature or experience), or other appropriate probability distribution forms. For example, for damage thresholds... Based on its physical meaning (cumulative plastic strain threshold), it can be set to a uniform distribution between 0 and a certain upper limit value; for the hardening rate parameter b, its prior distribution can be set to a normal distribution based on existing data of similar materials. The setting of the prior distribution should cover all possible reasonable values of the parameter, while avoiding the introduction of too much subjective bias.
[0306] Based on likelihood function and prior distribution Markov chain Monte Carlo sampling is performed to obtain the posterior distribution samples of each parameter in the target material parameters. According to Bayes' theorem, the posterior distribution... Proportional to the product of the likelihood function and the prior distribution:
[0307] ;
[0308] The basic idea of the Markov chain Monte Carlo method is to construct a Markov chain with a stationary target posterior distribution. Samples from this target posterior distribution are generated by sampling this chain. Specifically, starting from initial parameter values, new candidate parameter values are continuously generated according to specific transition rules, and acceptance criteria are used to determine whether to accept each candidate value. After a sufficiently long "burning period" (i.e., pre-burning period), the Markov chain approaches a stationary distribution, and the samples collected at this point can be considered as samples from the posterior distribution. In actual sampling, multiple parallel Markov chains can be used for simultaneous sampling to improve sampling efficiency and the reliability of convergence diagnosis. During the sampling process, the parameter values of each chain at each sampling step are recorded. After discarding samples from the burning period, the remaining samples are merged into the posterior distribution sample set. To ensure the representativeness of the samples, the number of sampling steps should be sufficiently large (e.g., tens of thousands to hundreds of thousands of steps), and convergence diagnostic indicators (such as the latent scale reduction factor) are needed to confirm that the sampling has converged.
[0309] Based on the posterior distribution sample, the posterior confidence intervals of each parameter in the target material parameters are determined to assess the uncertainty of the target material parameters. The posterior confidence interval is an important indicator in Bayesian statistics used to quantify parameter uncertainty. It means that, given the observed data, the parameter falls within the interval with a certain probability.
[0310] Specifically, for each target material parameter, the posterior distribution samples are sorted by numerical value, and the 2.5% quantile and 97.5% quantile are taken as the lower and upper limits of the 95% posterior confidence interval, respectively. For example, for parameter b, the 2.5% quantile of its posterior distribution sample is 4.8 and the 97.5% quantile is 5.6. Then the 95% posterior confidence interval of b is [4.8, 5.6], which means that under the given experimental data, b has a 95% probability of falling within this interval.
[0311] By using the posterior confidence interval, the uncertainty range of the target material parameters can be quantitatively assessed, providing a reliable basis for parameter values in subsequent structural performance evaluation. When the confidence interval of a parameter is narrow, it indicates that the parameter identification result is relatively reliable; when the confidence interval is wide, it indicates that the parameter has greater uncertainty, and it may be necessary to supplement experimental data to improve the accuracy of parameter identification.
[0312] Based on the above-mentioned numerical implementation method for the adaptive constitutive model of additively manufactured stainless steel, this invention also provides a numerical implementation device for the adaptive constitutive model of additively manufactured stainless steel. The following will be combined with... Figure 9 The device is described in detail.
[0313] Figure 9 A structural block diagram of a numerical realization apparatus for an adaptive constitutive model of additive manufacturing stainless steel material according to an embodiment of the present invention is shown.
[0314] like Figure 9 As shown, the numerical realization device 900 for the adaptive constitutive model of additive manufacturing stainless steel material in this embodiment includes a data acquisition module 910, anisotropy determination module 920, hardening determination module 930, damage determination module 940, parameter optimization module 950, and numerical solution module 960.
[0315] The data acquisition module 910 is used to acquire stress and strain data of stainless steel specimens in multiple sampling directions, as well as the yield strength and shear yield strength of stainless steel in multiple preset principal axis directions determined based on the stress and strain data.
[0316] The anisotropy determination module 920 is used to determine the anisotropy coefficient in the anisotropic yield criterion based on the yield strength and shear yield strength, so as to reflect the difference in yield strength of stainless steel material in different preset principal axis directions.
[0317] The hardening determination module 930 is used to fit the parameters of the hardening model based on the data of the plastic deformation stage in the stress-strain data. The hardening model is used to describe the variation of flow stress with cumulative plastic strain in stainless steel during plastic deformation.
[0318] The damage determination module 940 is used to determine the parameters of the damage model based on the unloading stiffness degradation data in the stress-strain data, so as to introduce the damage variables determined by the parameters of the damage model into the anisotropic yield criterion. The damage model is used to describe the variation law of internal damage of stainless steel material during plastic deformation.
[0319] The parameter optimization module 950 is used to optimize the anisotropy coefficient, the parameters of the hardening model, and the parameters of the damage model with the goal of minimizing the error between the stress-strain results predicted by the constitutive model and the stress-strain data, so as to obtain the target material parameters. The constitutive model is composed of the anisotropic yield criterion, the hardening model, and the damage model.
[0320] The numerical solution module 960 is used to substitute the target material parameters into the constitutive model for numerical solution to obtain the mechanical response results for structural performance evaluation.
[0321] According to an embodiment of the present invention, the anisotropy determination module 920 includes a normal stress determination submodule, a shear stress determination submodule, and anisotropy determination submodule.
[0322] The normal stress determination submodule is used to calculate the anisotropy coefficient of the normal stress term in the anisotropic yield criterion based on the reference yield strength, the yield strength in the second preset principal axis direction, and the yield strength in the third preset principal axis direction. The anisotropy coefficient of the normal stress term is used to reflect the difference in yield strength among the three preset principal axis directions.
[0323] The shear stress determination submodule is used to calculate the anisotropy coefficient of the shear stress term in the anisotropic yield criterion based on the reference yield strength, the first shear yield strength, the second shear yield strength, and the third shear yield strength. The anisotropy coefficient of the shear stress term is used to reflect the difference in shear yield strength between the three preset principal axis vertical planes.
[0324] The anisotropy determination submodule is used to use the anisotropy coefficients of the normal stress term and the shear stress term as anisotropy coefficients.
[0325] According to an embodiment of the present invention, the hardening determination module 930 includes a strain determination submodule, a flow determination submodule, a hardening fitting submodule, and a hardening determination submodule.
[0326] The strain determination submodule is used to determine the plastic strain of each data point in the plastic deformation stage based on the uniaxial tensile stress-strain data in the stress-strain data, and to accumulate the cumulative plastic strain.
[0327] The flow determination submodule is used to calculate the model flow stress using the isotropic hardening model, based on the cumulative plastic strain, the initial flow stress determined based on the yield strength, and the saturated hardening increment and hardening rate parameters in the isotropic hardening model. The saturated hardening increment and hardening rate parameters are parameters to be fitted. The model flow stress increases with the increase of cumulative plastic strain, and the increase gradually decreases and tends to the upper limit value determined by the initial flow stress and the saturated hardening increment.
[0328] The hardening fitting submodule is used to fit the parameters to be fitted with the goal of minimizing the error between the model flow stress and the experimental flow stress, and to obtain the saturated hardening increment fitting value and the hardening rate parameter fitting value. The experimental flow stress is the flow stress in the uniaxial tensile stress-strain data that has the same cumulative plastic strain as the model flow stress.
[0329] The hardening determination submodule is used to take the initial flow stress, the fitted value of the saturated hardening increment, and the fitted value of the hardening rate parameter as parameters of the hardening model.
[0330] According to an embodiment of the present invention, the hardening determination module 930 further includes a stress determination submodule, a stress update submodule, a follow-up fitting submodule, and a follow-up determination submodule.
[0331] The stress determination submodule is used to obtain the stable hysteresis loop obtained from the cyclic loading and unloading test from the stress-strain data, and based on the stable hysteresis loop, determine the plastic strain of each data point in the plastic deformation stage, accumulate them to obtain the cumulative plastic strain, and determine the total back stress data points in the plastic deformation stage.
[0332] The stress update submodule is used to update each back stress component incrementally using the kinematic hardening model, based on the plastic strain increment and cumulative plastic strain increment of each incremental step in the strain history of the cyclic loading and unloading test, as well as the kinematic hardening modulus and dynamic recovery coefficient of each back stress component in the kinematic hardening model, to obtain the total back stress of the model. The kinematic hardening modulus and dynamic recovery coefficient of each back stress component are the parameters to be fitted.
[0333] The follow-up fitting submodule is used to fit the parameters to be fitted with the goal of minimizing the error between the total back stress of the model and the total back stress of the experiment, and to obtain the fitted values of the follow-up hardening modulus and the dynamic recovery coefficient. The total back stress of the experiment is the total back stress data point extracted from the stable hysteresis loop.
[0334] The follow-up determination submodule is used to use the fitted values of the follow-up hardening modulus and the fitted values of the dynamic recovery coefficient as parameters of the hardening model.
[0335] According to an embodiment of the present invention, the damage determination module 940 includes an unloading determination submodule, a damage calculation submodule, a variable update submodule, a damage fitting submodule, and a damage determination submodule.
[0336] The unloading determination submodule is used to obtain the unloading stiffness degradation data of the graded loading and unloading test from the stress and strain data, and to determine the plastic strain of each data point in the plastic deformation stage based on the graded loading and unloading test data, and to accumulate the cumulative plastic strain, as well as to determine the initial elastic modulus of the first loading and the unloading modulus corresponding to each unloading.
[0337] The damage calculation submodule is used to calculate the test damage variables based on the unloading modulus and initial elastic modulus corresponding to each unloading.
[0338] The variable update submodule is used to utilize the damage model to keep the damage variable at zero when the cumulative plastic strain is less than or equal to the damage threshold; when the cumulative plastic strain is greater than the damage threshold, the damage energy release rate is determined based on the stress-strain data, and the damage variable is updated incrementally based on the damage energy release rate, the damage energy scale parameter and the damage evolution index in the damage model, and the increment of the cumulative plastic strain between adjacent data points. Here, the damage threshold, the damage energy scale parameter, and the damage evolution index are the parameters to be fitted to the damage model.
[0339] The damage fitting submodule is used to fit the parameters to be fitted with the goal of minimizing the error between the damage variable and the experimental damage variable, and to obtain the damage threshold fitting value, the damage energy scale parameter fitting value, and the damage evolution index fitting value.
[0340] The damage determination submodule is used to use the fitted values of the damage threshold, the fitted values of the damage energy scale parameter, and the fitted values of the damage evolution index as parameters of the damage model.
[0341] According to an embodiment of the present invention, the parameter optimization module 950 includes a sensitivity analysis submodule, a parameter grouping submodule, and an iterative convergence submodule.
[0342] The sensitivity analysis submodule is used to perform a global sensitivity analysis on each parameter among the anisotropy coefficient, the parameters of the hardening model, and the parameters of the damage model, in order to determine multiple key parameters from the anisotropy coefficient, the parameters of the hardening model, and the parameters of the damage model based on the degree of influence of each parameter on the stress-strain results predicted by the constitutive model.
[0343] The parameter grouping submodule is used to divide multiple key parameters into anisotropic yield parameter group, hardening parameter group and damage parameter group according to their physical source. Each of the anisotropic yield parameter group, hardening parameter group and damage parameter group includes at least one key parameter. The stress-strain data includes multiple sets of test data, and each set of test data corresponds to the anisotropic yield parameter group, hardening parameter group or damage parameter group respectively.
[0344] The iterative convergence submodule is used to repeatedly perform the following operations until the error between the stress-strain result predicted by the constitutive model and the stress-strain data meets the preset convergence condition, obtaining the optimized values of the anisotropic yield parameter group, hardening parameter group, and damage parameter group, and using the optimized values as the target material parameters: the anisotropic yield parameter group, hardening parameter group, and damage parameter group are optimized sequentially. When optimizing the current parameter group, the parameter values of other parameter groups are fixed, and the parameters in the current parameter group are adjusted to minimize the error between the stress-strain result predicted by the constitutive model and the experimental data corresponding to the current parameter group. In the process of calculating the error, the weight of each group of experimental data is dynamically adjusted according to the normalized error between each group of experimental data and the stress-strain result predicted by the constitutive model.
[0345] According to an embodiment of the present invention, the numerical solution module 960 includes a parameter input submodule, a load acquisition submodule, an initialization submodule, a matrix determination submodule, a trial determination submodule, a yield judgment submodule, a variable preservation submodule, a correction iteration submodule, a modulus determination submodule, and a result output submodule.
[0346] The parameter input submodule is used to substitute the anisotropy coefficients in the target material parameters into the anisotropic yield criterion, substitute the parameters of the hardening model in the target material parameters into the hardening model, and substitute the parameters of the damage model in the target material parameters into the damage model.
[0347] The load acquisition submodule is used to acquire the loads acting on the structural members made of stainless steel and divide the loads into multiple load increment steps.
[0348] The initialization submodule is used to initialize the stress, strain, cumulative plastic strain, and damage variables of the structural components to zero, and to initialize the hardening variables according to the type of hardening model. Specifically, when the hardening model is a kinematic hardening model, multiple back stress components in the kinematic hardening model are initialized to zero, and each back stress component is used as the hardening variable. When the hardening model is an isotropic hardening model, the hardening variable is the cumulative plastic strain.
[0349] The matrix determination submodule is used to determine the elastic stiffness matrix of stainless steel materials based on the elastic stage data in the stress-strain data.
[0350] The trial determination submodule is used to perform the following operations for each load increment step: calculate the trial stress using the elastic stiffness matrix based on the stress, strain, cumulative plastic strain, damage variable, and hardening variable at the end of the previous load increment step, as well as the total strain increment corresponding to the current load increment step.
[0351] The yield judgment submodule is used to substitute the test stress into the yield function jointly determined by the anisotropic yield criterion, the hardening model, and the damage model to determine whether the test stress meets the yield condition. The yield function is obtained by calculating the equivalent stress by the anisotropic yield criterion, calculating the flow stress by the hardening model, and then performing damage correction on the equivalent stress by the damage model.
[0352] The variable preservation submodule is used to use the test stress as the stress of the current load increment step when the test stress does not meet the yield condition, while keeping the cumulative plastic strain, damage variable, and hardening variable unchanged.
[0353] The correction iteration submodule is used to perform plastic correction iteration on the current load increment step when the test stress meets the yield condition. During the iteration process, the hardening variables are updated using the parameters of the hardening model, the damage variables are updated using the parameters of the damage model, and the stress and cumulative plastic strain are adjusted until the yield function meets the preset convergence condition.
[0354] The modulus determination submodule is used to calculate the elastoplastic uniform tangent modulus when the yield function meets the preset convergence conditions, and is used for iterative calculation of the next load increment step.
[0355] The results output submodule is used to output the stress, strain, damage variables and displacement results of structural members after completing iterative calculations for all load increment steps, as mechanical response results for structural performance evaluation.
[0356] According to an embodiment of the present invention, the yield judgment submodule includes an equivalent calculation unit, a damage correction unit, a flow calculation unit, and a yield judgment unit.
[0357] The equivalent calculation unit is used to calculate the equivalent stress based on the test stress using the anisotropic yield criterion.
[0358] The damage correction unit is used to correct the equivalent stress based on the damage variables using the damage model, and obtain the equivalent stress after damage correction.
[0359] The flow calculation unit is used to calculate flow stress based on cumulative plastic strain using a hardening model.
[0360] The yield judgment unit is used to compare the damage-corrected equivalent stress with the flow stress to determine whether the test stress satisfies the yield condition if the damage-corrected equivalent stress is greater than the flow stress.
[0361] According to an embodiment of the present invention, the numerical realization device 900 for the adaptive constitutive model of additive manufacturing stainless steel material includes a function construction module, a distribution setting module, a sample sampling module, and an interval determination module.
[0362] The function construction module is used to construct the likelihood function based on the normalized residuals between the stress-strain results predicted by the constitutive model and the stress-strain data.
[0363] The distribution setting module is used to set the prior distribution of each parameter in the target material parameters.
[0364] The sample sampling module is used to perform Markov chain Monte Carlo sampling based on the likelihood function and prior distribution to obtain the posterior distribution samples of each parameter in the target material parameters.
[0365] The interval determination module is used to determine the posterior confidence interval of each parameter in the target material parameters based on the posterior distribution sample, so as to assess the uncertainty of the target material parameters.
[0366] According to embodiments of the present invention, any plurality of modules among the data acquisition module 910, the heterogeneity determination module 920, the hardening determination module 930, the damage determination module 940, the parameter optimization module 950, and the numerical solution module 960 can be combined into one module, or any one of these modules can be split into multiple modules. Alternatively, at least part of the functionality of one or more of these modules can be combined with at least part of the functionality of other modules and implemented in one module. According to embodiments of the present invention, at least one of the data acquisition module 910, the heterogeneity determination module 920, the hardening determination module 930, the damage determination module 940, the parameter optimization module 950, and the numerical solution module 960 can be at least partially implemented as hardware circuitry, such as a field-programmable gate array (FPGA), a programmable logic array (PLA), a system-on-a-chip, a system-on-a-substrate, a system-on-package, an application-specific integrated circuit (ASIC), or any other reasonable means of integrating or packaging circuitry, or implemented in software, hardware, or firmware, or in any suitable combination of any of these three implementation methods. Alternatively, at least one of the data acquisition module 910, the anisotropy determination module 920, the hardening determination module 930, the damage determination module 940, the parameter optimization module 950, and the numerical solution module 960 can be at least partially implemented as a computer program module, which can perform corresponding functions when the computer program module is run.
[0367] Figure 10 A block diagram of an electronic device suitable for implementing a numerical method for an adaptive constitutive model of additively manufactured stainless steel materials, according to an embodiment of the present invention, is shown.
[0368] like Figure 10As shown, an electronic device 1000 according to an embodiment of the present invention includes a processor 1001, which can perform various appropriate actions and processes according to a program stored in a read-only memory ROM 1002 or a program loaded from a storage portion 1008 into a random access memory RAM 1003. The processor 1001 may include, for example, a general-purpose microprocessor (e.g., a CPU), an instruction set processor and / or an associated chipset and / or a special-purpose microprocessor (e.g., an application-specific integrated circuit (ASIC)), etc. The processor 1001 may also include onboard memory for caching purposes. The processor 1001 may include a single processing unit or multiple processing units for performing different actions of the method flow according to an embodiment of the present invention.
[0369] RAM 1003 stores various programs and data required for the operation of electronic device 1000. Processor 1001, ROM 1002, and RAM 1003 are interconnected via bus 1004. Processor 1001 executes various operations of the method flow according to embodiments of the present invention by executing programs in ROM 1002 and / or RAM 1003. It should be noted that programs may also be stored in one or more memories other than ROM 1002 and RAM 1003. Processor 1001 may also execute various operations of the method flow according to embodiments of the present invention by executing programs stored in one or more memories.
[0370] According to an embodiment of the present invention, the electronic device 1000 may further include an input / output (I / O) interface 1005, which is also connected to a bus 1004. The electronic device 1000 may also include one or more of the following components connected to the input / output (I / O) interface 1005: an input section 1006 including a keyboard, mouse, etc.; an output section 1007 including a cathode ray tube (CRT), liquid crystal display (LCD), etc., and a speaker, etc.; a storage section 1008 including a hard disk, etc.; and a communication section 1009 including a network interface card such as a LAN card, modem, etc. The communication section 1009 performs communication processing via a network such as the Internet. A drive 1010 is also connected to the input / output (I / O) interface 1005 as needed. A removable medium 1011, such as a disk, optical disk, magneto-optical disk, semiconductor memory, etc., is installed on the drive 1010 as needed so that computer programs read from it can be installed into the storage section 1008 as needed.
[0371] The present invention also provides a computer-readable storage medium, which may be included in the device / apparatus / system described in the above embodiments; or it may exist independently and not assembled into the device / apparatus / system. The computer-readable storage medium carries one or more programs, which, when executed, implement the method according to the embodiments of the present invention.
[0372] According to embodiments of the present invention, a computer-readable storage medium may be a non-volatile computer-readable storage medium, such as including, but not limited to: portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. In the present invention, a computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. For example, according to embodiments of the present invention, a computer-readable storage medium may include ROM 1002 and / or RAM 1003 and / or one or more memories other than ROM 1002 and RAM 1003 described above.
[0373] Embodiments of the present invention also include a computer program product comprising a computer program containing program code for performing the methods shown in the flowchart. When the computer program product is run on a computer system, the program code enables the computer system to implement the numerical implementation method for the adaptive constitutive model of additive manufacturing stainless steel materials provided in the embodiments of the present invention.
[0374] When the computer program is executed by the processor 1001, it performs the functions defined in the system / apparatus of this invention. According to embodiments of the invention, the systems, apparatuses, modules, units, etc., described above can be implemented by computer program modules.
[0375] In one embodiment, the computer program may rely on a tangible storage medium such as an optical storage device or a magnetic storage device. In another embodiment, the computer program may also be transmitted and distributed in the form of signals over a network medium, and may be downloaded and installed via the communication section 1009, and / or installed from a removable medium 1011. The program code contained in the computer program can be transmitted using any suitable network medium, including but not limited to: wireless, wired, etc., or any suitable combination thereof.
[0376] In such an embodiment, the computer program can be downloaded and installed from a network via the communication section 1009, and / or installed from the removable medium 1011. When the computer program is executed by the processor 1001, it performs the functions defined in the system of this embodiment of the invention. According to embodiments of the invention, the systems, devices, apparatuses, modules, units, etc., described above can be implemented by computer program modules.
[0377] According to embodiments of the present invention, program code for executing the computer programs provided in the embodiments of the present invention can be written in any combination of one or more programming languages. Specifically, these computational programs can be implemented using high-level procedural and / or object-oriented programming languages, and / or assembly / machine languages. Programming languages include, but are not limited to, languages such as Java, C++, Python, "C", or similar programming languages. The program code can be executed entirely on the user's computing device, partially on the user's device, partially on a remote computing device, or entirely on a remote computing device or server. In cases involving remote computing devices, the remote computing device can be connected to the user's computing device via any type of network, including a local area network (LAN) or a wide area network (WAN), or it can be connected to an external computing device (e.g., via the Internet using an Internet service provider).
[0378] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of the present invention. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in a block diagram or flowchart, and combinations of blocks in a block diagram or flowchart, may be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.
[0379] Those skilled in the art will understand that the features described in the various embodiments of the present invention can be combined and / or combined in various ways, even if such combinations or combinations are not explicitly described in the present invention. In particular, the features described in the various embodiments of the present invention can be combined and / or combined in various ways without departing from the spirit and teachings of the present invention. All such combinations and / or combinations fall within the scope of the present invention.
Claims
1. A numerical implementation method for an adaptive constitutive model of additive manufacturing stainless steel materials, characterized in that, The method includes: The stress-strain data of stainless steel specimens in multiple sampling directions are obtained, and the yield strength and shear yield strength of the stainless steel in multiple preset principal axis directions are determined based on the stress-strain data. Based on the yield strength and the shear yield strength, the anisotropy coefficient in the anisotropic yield criterion is determined to reflect the difference in yield strength of the stainless steel material in different preset principal axis directions; Based on the data of the plastic deformation stage in the stress-strain data, the parameters of the hardening model are fitted, wherein the hardening model is used to describe the variation law of flow stress with cumulative plastic strain in the stainless steel material during plastic deformation. Based on the unloading stiffness degradation data in the stress-strain data, the parameters of the damage model are determined so that the damage variables determined based on the parameters of the damage model are introduced into the anisotropic yield criterion. The damage model is used to describe the variation law of internal damage of the stainless steel material during plastic deformation. With the goal of minimizing the error between the stress-strain results predicted by the constitutive model and the stress-strain data, the anisotropy coefficient, the parameters of the hardening model, and the parameters of the damage model are optimized to obtain the target material parameters. The constitutive model is composed of the anisotropic yield criterion, the hardening model, and the damage model. The target material parameters are substituted into the constitutive model for numerical solution to obtain the mechanical response results used for structural performance evaluation.
2. The method according to claim 1, characterized in that, The plurality of preset principal axis directions include a first preset principal axis direction, a second preset principal axis direction, and a third preset principal axis direction; the shear yield strength includes a first shear yield strength on a plane perpendicular to the first preset principal axis direction, a second shear yield strength on a plane perpendicular to the second preset principal axis direction, and a third shear yield strength on a plane perpendicular to the third preset principal axis direction. When the yield strength in the first preset principal axis direction is used as the reference yield strength, determining the anisotropy coefficient in the anisotropic yield criterion based on the yield strength and the shear yield strength includes: Based on the reference yield strength, the yield strength in the second preset principal axis direction, and the yield strength in the third preset principal axis direction, the anisotropy coefficient of the normal stress term in the anisotropic yield criterion is calculated, wherein the anisotropy coefficient of the normal stress term is used to reflect the difference in yield strength among the three preset principal axis directions. Based on the reference yield strength, the first shear yield strength, the second shear yield strength, and the third shear yield strength, the anisotropy coefficient of the shear stress term in the anisotropic yield criterion is calculated, wherein the anisotropy coefficient of the shear stress term is used to reflect the difference in shear yield strength between the three preset principal axis vertical planes. The anisotropy coefficients of the normal stress term and the shear stress term are used as the anisotropy coefficients.
3. The method according to claim 1, characterized in that, When the hardening model is an isotropic hardening model, fitting the parameters of the hardening model based on the data from the plastic deformation stage in the stress-strain data includes: Based on the uniaxial tensile stress-strain data in the stress-strain data, the plastic strain of each of the multiple data points in the plastic deformation stage is determined, and the cumulative plastic strain is obtained by summing them up. Using the isotropic hardening model, the model flow stress is calculated based on the cumulative plastic strain, the initial flow stress determined based on the yield strength, and the saturated hardening increment and hardening rate parameters in the isotropic hardening model. The saturated hardening increment and the hardening rate parameters are parameters to be fitted. The model flow stress increases with the increase of the cumulative plastic strain, and the increase gradually decreases and tends to the upper limit value determined by the initial flow stress and the saturated hardening increment. With the goal of minimizing the error between the model flow stress and the experimental flow stress, the parameters to be fitted are fitted to obtain the saturated hardening increment fitting value and the hardening rate parameter fitting value. The experimental flow stress is the flow stress in the uniaxial tensile stress-strain data that has the same cumulative plastic strain as the model flow stress. The initial flow stress, the fitted value of the saturated hardening increment, and the fitted value of the hardening rate parameter are used as parameters of the hardening model.
4. The method according to claim 1, characterized in that, When the hardening model is a kinematic hardening model, fitting the parameters of the hardening model based on the data from the plastic deformation stage in the stress-strain data includes: The stable hysteresis loop obtained from the cyclic loading and unloading test is obtained from the stress and strain data. Based on the stable hysteresis loop, the plastic strain of each data point in the plastic deformation stage is determined and accumulated to obtain the cumulative plastic strain. The total back stress data points in the plastic deformation stage are also determined. Using the kinematic hardening model, based on the plastic strain increment and cumulative plastic strain increment of each incremental step in the strain history of the cyclic loading and unloading test, as well as the kinematic hardening modulus and dynamic recovery coefficient of each of the multiple back stress components in the kinematic hardening model, each back stress component is updated incrementally to obtain the total back stress of the model, wherein the kinematic hardening modulus and dynamic recovery coefficient of each back stress component are the parameters to be fitted. With the goal of minimizing the error between the total back stress of the model and the total back stress of the experiment, the parameters to be fitted are fitted to obtain the fitting values of the kinematic hardening modulus and the dynamic recovery coefficient, wherein the total back stress of the experiment is the total back stress data point extracted from the stable hysteresis loop. The fitted values of the kinematic hardening modulus and the fitted values of the dynamic recovery coefficient are used as parameters of the hardening model.
5. The method according to claim 1, characterized in that, The step of determining the parameters of the damage model based on the unloading stiffness degradation data in the stress-strain data includes: The unloading stiffness degradation data of the graded loading and unloading test is obtained from the stress and strain data. Based on the graded loading and unloading test data, the plastic strain of each data point in the plastic deformation stage is determined and accumulated to obtain the cumulative plastic strain. The initial elastic modulus of the first loading and the unloading modulus corresponding to each unloading are also determined. The test damage variable is calculated based on the unloading modulus corresponding to each unloading and the initial elastic modulus. Using the damage model, when the cumulative plastic strain is less than or equal to the damage threshold, the damage variable remains zero; when the cumulative plastic strain is greater than the damage threshold, the damage energy release rate is determined based on the stress-strain data, and the damage variable is updated incrementally based on the damage energy release rate, the damage energy scale parameter and the damage evolution index in the damage model, and the increment of the cumulative plastic strain between adjacent data points, wherein the damage threshold, the damage energy scale parameter and the damage evolution index are the parameters to be fitted to the damage model; With the goal of minimizing the error between the damage variable and the experimental damage variable, the parameters to be fitted are fitted to obtain the damage threshold fitting value, the damage energy scale parameter fitting value, and the damage evolution index fitting value. The fitted values of the damage threshold, the damage energy scale parameter, and the damage evolution index are used as parameters of the damage model.
6. The method according to claim 1, characterized in that, The goal is to minimize the error between the stress-strain results predicted by the constitutive model and the stress-strain data. This involves optimizing the anisotropy coefficient, the parameters of the hardening model, and the parameters of the damage model to obtain target material parameters, including: A global sensitivity analysis is performed on each of the anisotropy coefficients, the parameters of the hardening model, and the parameters of the damage model to determine multiple key parameters from the anisotropy coefficients, the parameters of the hardening model, and the parameters of the damage model based on the degree of influence of each parameter on the stress-strain results predicted by the constitutive model. The multiple key parameters are divided into anisotropic yield parameter group, hardening parameter group and damage parameter group according to their physical source. The anisotropic yield parameter group, the hardening parameter group and the damage parameter group each include at least one key parameter. The stress-strain data includes multiple sets of test data. Each set of test data corresponds to the anisotropic yield parameter group, the hardening parameter group or the damage parameter group. Repeat the following operations until the error between the stress-strain result predicted by the constitutive model and the stress-strain data satisfies the preset convergence condition, to obtain the optimized values of the anisotropic yield parameter set, the hardening parameter set, and the damage parameter set, and use the optimized values as the target material parameters: The anisotropic yield parameter group, the hardening parameter group, and the damage parameter group are optimized sequentially. When optimizing the current parameter group, the parameter values of other parameter groups are fixed, and the parameters in the current parameter group are adjusted to minimize the error between the stress-strain results predicted by the constitutive model and the experimental data corresponding to the current parameter group. In the process of calculating the error, the weights of each group of experimental data are dynamically adjusted according to the normalized error between each group of experimental data and the stress-strain results predicted by the constitutive model.
7. The method according to claim 1, characterized in that, The step of substituting the target material parameters into the constitutive model for numerical solution to obtain the mechanical response results for structural performance evaluation includes: Substitute the anisotropy coefficient in the target material parameters into the anisotropic yield criterion, substitute the parameters of the hardening model in the target material parameters into the hardening model, and substitute the parameters of the damage model in the target material parameters into the damage model. The load acting on the structural member of the stainless steel material is obtained, and the load is divided into multiple load increment steps; The stress, strain, cumulative plastic strain, and damage variables of the structural member are initialized to zero, and the hardening variables are initialized according to the type of the hardening model. In the case of the hardening model being a kinematic hardening model, multiple back stress components in the kinematic hardening model are initialized to zero, and each back stress component is used as the hardening variable. In the case of the hardening model being an isotropic hardening model, the hardening variable is the cumulative plastic strain. The elastic stiffness matrix of the stainless steel material is determined based on the elastic stage data in the stress-strain data. Perform the following operations for each of the aforementioned load increment steps: The test stress is calculated using the elastic stiffness matrix based on the stress, strain, cumulative plastic strain, damage variable, and hardening variable at the end of the previous load increment step, and the total strain increment corresponding to the current load increment step. The test stress is substituted into the yield function jointly determined by the anisotropic yield criterion, the hardening model, and the damage model to determine whether the test stress meets the yield condition. The yield function is obtained by calculating the equivalent stress by the anisotropic yield criterion, calculating the flow stress by the hardening model, and then correcting the equivalent stress for damage by the damage model. If the test stress does not meet the yield condition, the test stress is used as the stress of the current load increment step, while the cumulative plastic strain, the damage variable, and the hardening variable remain unchanged. If the test stress satisfies the yield condition, a plastic correction iteration is performed on the current load increment step. During the iteration process, the hardening variable is updated using the parameters of the hardening model, the damage variable is updated using the parameters of the damage model, and the stress and the cumulative plastic strain are adjusted until the yield function satisfies the preset convergence condition. If the yield function satisfies the preset convergence condition, the elastic-plastic uniform tangent modulus is calculated for iterative calculation in the next load increment step. After completing iterative calculations for all load increment steps, the stress, strain, damage variables, and displacement results of the structural member are output as mechanical response results for structural performance evaluation.
8. The method according to claim 7, characterized in that, The step of substituting the test stress into the yield function jointly determined by the anisotropic yield criterion, the hardening model, and the damage model to determine whether the test stress satisfies the yield condition includes: The equivalent stress is calculated based on the test stress using the anisotropic yield criterion. The equivalent stress is corrected based on the damage variables using the damage model to obtain the damage-corrected equivalent stress. The flow stress is calculated based on the accumulated plastic strain using the hardening model. The damage-corrected equivalent stress is compared with the flow stress, and if the damage-corrected equivalent stress is greater than the flow stress, the test stress is determined to satisfy the yield condition.
9. The method according to claim 1, characterized in that, The method further includes: A likelihood function is constructed based on the normalized residual between the stress-strain results predicted by the constitutive model and the stress-strain data. Define the prior distribution of each parameter in the target material parameters; Markov chain Monte Carlo sampling is performed based on the likelihood function and the prior distribution to obtain the posterior distribution samples of each parameter in the target material parameters. Based on the posterior distribution sample, the posterior confidence interval of each parameter in the target material parameters is determined to assess the uncertainty of the target material parameters.
10. A numerical realization device for an adaptive constitutive model of additive manufacturing stainless steel material, characterized in that, The device includes: The data acquisition module is used to acquire stress and strain data of stainless steel specimens in multiple sampling directions, and to determine the yield strength and shear yield strength of the stainless steel in multiple preset principal axis directions based on the stress and strain data. An anisotropy determination module is used to determine the anisotropy coefficient in the anisotropic yield criterion based on the yield strength and the shear yield strength, so as to reflect the difference in yield strength of the stainless steel material in different preset principal axis directions; The hardening determination module is used to fit the parameters of the hardening model based on the data of the plastic deformation stage in the stress-strain data, wherein the hardening model is used to describe the variation law of flow stress with cumulative plastic strain in the stainless steel material during plastic deformation. The damage determination module is used to determine the parameters of the damage model based on the unloading stiffness degradation data in the stress-strain data, so as to introduce the damage variable determined according to the parameters of the damage model into the anisotropic yield criterion, wherein the damage model is used to describe the change law of internal damage of the stainless steel material during plastic deformation. The parameter optimization module is used to optimize the anisotropy coefficient, the parameters of the hardening model, and the parameters of the damage model with the goal of minimizing the error between the stress-strain result predicted by the constitutive model and the stress-strain data, so as to obtain the target material parameters. The constitutive model is composed of the anisotropic yield criterion, the hardening model, and the damage model. The numerical solution module is used to substitute the target material parameters into the constitutive model for numerical solution to obtain the mechanical response results for structural performance evaluation.
