A blade structure creation method based on bearing force flow topology optimization

Abstract

The application discloses a kind of based on bearing force flow topology optimization blade structure creation method, it is related to wind power generation equipment technical field.The method comprises the following steps: establishing a three-dimensional finite element benchmark model of wind turbine blade, and identifying and intercepting a local design domain containing web-beam cap connecting region from the model;Finite element modal analysis is carried out on the blade or its segment, and one or more vibration node lines are extracted from its vibration mode;In the local design domain, a local topology optimization mathematical model is established and solved, with the optimization objective of minimizing the peak stress of material interface, and the dynamic geometric constraint that the material distribution generated by optimization follows vibration node line in space as core constraint condition;According to the result of optimization solution, the geometric model of blade connecting structure is generated.The application can prospectively and integrally improve the static reliability and dynamic fatigue resistance of key connecting region of blade.

Description

Technical Field

[0001] This invention relates to the field of wind power generation equipment technology, and in particular to a method for blade structure creation based on load-bearing flow topology optimization. Background Technology

[0002] As components that convert wind energy into mechanical energy, the structural performance of wind turbine blades determines the overall operational reliability and service life of the turbine. To reduce blade weight while meeting strength and stiffness requirements, topology optimization-based structural design methods have been adopted in this field. A key technical idea behind these methods is to analyze the principal stress trajectories of the structure under load, i.e., the so-called load-bearing flow, to guide the optimal distribution of materials on a macroscopic scale, aiming to achieve a high degree of unity between the mechanical transmission path and material layout. This global optimization approach based on load-bearing flow has been applied to determining the macroscopic orientation of the main load-bearing components such as the main beam and web within the blade.

[0003] However, existing optimization design methods based on macroscopic load-bearing flow have an unresolved technical problem: they fail to consider the reliability and dynamic response characteristics of local connection interfaces when analyzing and applying load-bearing flow. Blades are discontinuous structures assembled from multiple composite material components using adhesive bonding. At the connection interfaces of different components, the macroscopic load-bearing flow can experience severe "flow around" and "congestion" due to abrupt changes in geometry and material properties, resulting in static stress concentration. Simultaneously, the periodic excitation experienced by the blade during operation triggers a dynamic response in the structure; this dynamic effect can be considered a "dynamic force flow disturbance," which is also amplified at the weak connection interface. Summary of the Invention

[0004] This invention provides a blade structure creation method based on load-bearing flow topology optimization, aiming to solve the technical problem that existing large composite blades, in structural optimization design, fail to consider the reliability and dynamic response characteristics of local connection interfaces in the structure, thus failing to effectively suppress the failure risk caused by static and dynamic coupling effects in the connection area during the forward-looking design stage.

[0005] In view of the above problems, the present invention provides a blade structure creation method based on load-bearing flow topology optimization, the method comprising the following steps:

[0006] A three-dimensional finite element reference model of a wind turbine blade is established, and a predetermined connection area formed by at least two blade components by adhesive bonding is identified and extracted from the reference model as a local design domain.

[0007] Finite element modal analysis is performed on the blade or its segment to identify a target vibration suppression mode, and one or more vibration nodal lines are extracted from the mode shape of the target vibration suppression mode.

[0008] In the local design domain, a stress monitoring zone representing the adhesive layer or material interface is defined; a topology optimization mathematical model for the local design domain is established and solved, wherein the optimization objective of the model is to minimize the peak stress in the stress monitoring zone, and the constraints of the model include at least one volume constraint and one dynamic geometric constraint, wherein the dynamic geometric constraint requires that the optimized load-bearing path be spatially adjacent to the vibration node line;

[0009] Based on the results of the topology optimization solution, a geometric model of the connected region is generated.

[0010] Preferably, the predetermined connection area is a T-shaped connection area between the web and the beam cap.

[0011] Preferably, the step of extracting the vibration node line specifically includes:

[0012] Using an isosurface extraction algorithm, a set of spatial points with zero or less than a preset threshold vibration displacement is identified in the modal displacement field of the target suppressed vibration mode.

[0013] The set of spatial points is parameterized into one or more continuous spatial curves using a curve fitting algorithm.

[0014] Preferably, the optimization objective of the model is to minimize an equivalent peak stress within the stress monitoring zone after processing by an aggregation function; the aggregation function is configured to smoothly transform the stress values ​​of multiple elements within the stress monitoring zone into a globally differentiable scalar.

[0015] Preferably, in the step of establishing the benchmark model, the mechanical behavior of the adhesive layer material within the local design domain is defined by a cohesive zone model; the mechanical behavior of the composite material within the local design domain is defined by an orthotropic material model that includes a progressive damage failure criterion.

[0016] Preferably, the step of identifying and extracting the local design domain specifically includes:

[0017] One or more static limit load conditions are applied to the three-dimensional finite element reference model, and finite element analysis is performed to obtain the global stress distribution;

[0018] Based on the global stress distribution, high stress concentration areas are identified as the local design domains;

[0019] Using sub-model technology, displacement boundary conditions are applied to the intercepted boundary of the local design domain. These displacement boundary conditions are derived from the displacement analysis results of the benchmark model at the corresponding nodes.

[0020] Preferably, the step of generating the geometric model of the blade connection structure specifically includes:

[0021] The topology optimization solution is binarized and isosurfaces are extracted to obtain an initial boundary surface;

[0022] The initial boundary surface is smoothed.

[0023] Based on the smoothed surface, a computer-aided design model that can be used for manufacturing is reconstructed.

[0024] Preferably, the reconstructed computer-aided design model has geometric features including at least one of the following:

[0025] Reinforcing ribs arranged along the vibration node line;

[0026] A wavy or finger-shaped meshing interface.

[0027] The technical solution provided in this application has at least the following technical effects or advantages:

[0028] The method provided by this invention couples the local dynamic characteristics of a structure (represented by vibration node lines) with the static reliability of the interface (targeted by interface stress) within a unified optimization framework. This enables a proactive and integrated solution to the static-dynamic coupling reliability problem in connection areas during the structure creation stage.

[0029] The connection structure generated by the method of this invention has a geometry that is the result of the combined effect of static stress dissipation and dynamic response suppression requirements. Simulation data shows that, compared with connection structures designed using traditional methods, the structure generated by this invention can reduce the peak stress at the connection interface and suppress the vibration response amplitude of the structure at a specific frequency without significantly increasing the amount of material used, thereby improving the static reliability and dynamic fatigue resistance of the connection area.

[0030] This invention significantly reduces the computational cost of this advanced design method by focusing the complex co-optimization process on the highest-risk local connectivity regions, rather than the entire blade. Simultaneously, this localized design makes the manufacturing and quality control of complex geometries more focused and feasible. This allows the static-dynamic co-optimization concept, which was previously confined to the theoretical level or required enormous computational resources, to be applied in engineering practice in a more economical and efficient manner, shortening the development cycle of high-performance blades. Attached Figure Description

[0031] Figure 1 This is a flowchart of a blade structure creation method based on load-bearing flow topology optimization according to the present invention. Detailed Implementation

[0032] To make the objectives, technical solutions, and advantages of this invention clearer and more complete, the following will provide a detailed and reproducible description of a method for generating blade connection structures based on dynamic constraints and interface optimization, as well as a wind turbine blade manufactured by this method, in conjunction with a specific implementation process.

[0033] Please see Figure 1 A method for creating blade structures based on load-bearing flow topology optimization, the method comprising the following steps:

[0034] A three-dimensional finite element reference model of a wind turbine blade is established, and a predetermined connection area formed by at least two blade components by adhesive bonding is identified and extracted from the reference model as a local design domain.

[0035] Finite element modal analysis is performed on the blade or its segment to identify a target vibration suppression mode, and one or more vibration nodal lines are extracted from the mode shape of the target vibration suppression mode.

[0036] In the local design domain, a stress monitoring zone representing the adhesive layer or material interface is defined; a topology optimization mathematical model for the local design domain is established and solved, wherein the optimization objective of the model is to minimize the peak stress in the stress monitoring zone, and the constraints of the model include at least one volume constraint and one dynamic geometric constraint, wherein the dynamic geometric constraint requires that the optimized load-bearing path be spatially adjacent to the vibration node line;

[0037] Based on the results of the topology optimization solution, a geometric model of the connected region is generated.

[0038] The execution flow of the method described in this invention begins with the establishment of a high-fidelity reference three-dimensional finite element model.

[0039] In a preferred embodiment, the process is executed on a software package platform integrating computer-aided design and computer-aided engineering capabilities (e.g., Abaqus / CAE, ANSYS Workbench). The process begins by importing or creating a preliminary three-dimensional geometric model of the wind turbine blade, hereinafter referred to as the baseline geometric model. The geometric information of this model can be derived from an existing, aerodynamically validated blade design database, and the data format can be CATIA, Pro / ENGINEER, or the general STEP format. The baseline geometric model should topologically encompass all major load-bearing and non-load-bearing components of the blade, such as the upper and lower beam caps (forming the main beams) made of unidirectional or multidirectional glass fiber / carbon fiber reinforced composite materials, the web for connecting the beam caps and transmitting shear forces, the skin for maintaining the aerodynamic shape of the blade, and the epoxy resin-based structural adhesive layer for bonding the aforementioned components.

[0040] In the finite element mesh generation stage, in order to achieve a balance between computational accuracy and computational efficiency, an adaptive and differentiated mesh generation strategy based on geometric features and expected stress distribution can be adopted.

[0041] Specifically, for regions in the blade's main structure where the stress gradient changes relatively gently, such as most of the skin surface, computationally efficient structured hexahedral elements or two-dimensional shell elements can be used. However, for critical areas with complex geometries and anticipated significant stress concentrations based on engineering experience, such as the T-shaped connection between the web and the beam cap, the connection between the blade root and the hub, and the trailing edge bonding area, a finer unstructured mesh is used. This unstructured mesh preferably uses second-order tetrahedral elements that can well fit complex curved surfaces, or a hybrid mesh using pyramidal elements for smooth transitions between different types of mesh regions.

[0042] Specifically, the mesh generation process can be executed by an automated mesh generator based on a set of preset control parameters. For example, a higher mesh density is required in the T-junction region, and the element feature length can be set to a smaller value, such as five millimeters; while on the skin far from this region, the feature length can be increased to fifty millimeters. All generated meshes, including but not limited to element aspect ratio, maximum / minimum interior angle, skewness, and Jacobian determinant, should be automatically evaluated using the software's built-in mesh quality check tool. This ensures that, for example, over 95% of the elements have quality indicators within the preset, engineering-calculation-allowed good range, thus avoiding numerical singularities or distorted calculation results due to poor mesh quality during subsequent finite element solution processes.

[0043] Precise physical definitions of material properties are fundamental for finite element models to accurately reflect real-world physical behavior. The main structure of a blade is typically composed of composite laminates with significant anisotropic characteristics, and its mechanical behavior is defined using an orthogonal anisotropic material model. The constitutive relations of this model require inputting a series of engineering constants obtained through physical experiments or consulted authoritative material handbooks into the material property definition module of the CAE software. This set of constants should include at least the elastic modulus (E1, E2, E3) in the three principal directions, the shear modulus (G12, G23, G13) in the three planes, three independent Poisson's ratios (V12, V23, V13), and the material density. To accurately predict the damage initiation and evolution process of composite materials in ultimate load analysis, a progressive damage model should be defined for it.

[0044] In a preferred embodiment, a Hashin failure criterion capable of independently determining four different failure modes—fiber tensile failure, fiber compressive failure, matrix tensile failure, and matrix compressive failure—can be employed. The underlying logic is based on a continuous comparison between the stress state tensor at each integration point within each element of the finite element model and pre-input parameters characterizing material strength (such as fiber tensile strength XT, fiber compressive strength XC, etc.). Once the stress state at a certain integration point satisfies the strength envelope defined by any of the failure criteria, the material stiffness at that point is considered to have begun to degrade. This degradation process can be described by a reduced stiffness matrix associated with internal damage variables (e.g., dft, dfc, dmt, dmc), which represent the damage degree of the four damage modes, and their evolution can be defined by a damage evolution rule based on energy release rate or equivalent displacement. For adhesive materials connecting various components, to accurately capture potential interfacial delamination and shear failure under complex stress states, a cohesive zone model (CZM) based on fracture mechanics should be used to define the entire nonlinear physical process of the adhesive interface, from the initial elastic bearing to damage initiation and complete stiffness loss leading to ultimate failure, through a predefined traction-separation rule with a specific mathematical shape (e.g., trapezoidal, bilinear, or exponential). The initial damage of the adhesive layer can be determined using a quadratic nominal stress criterion that comprehensively considers normal stress and stresses in both shear directions. The stiffness degradation process of the adhesive layer element after damage initiation is controlled by the material's critical strain energy release rate (GIC, GIIC, GIIIC). This set of parameters is numerically equal to the area enclosed by the complete traction-separation curve and the horizontal axis, and is a key physical quantity characterizing the material's fracture toughness. The accurate acquisition of the aforementioned complex material parameters usually requires a series of standard material mechanical property tests on the material sample, such as uniaxial tensile, compression, and shear tests, as well as double cantilever beam (DCB) and end notch bending (ENF) tests used to obtain fracture toughness.

[0045] After establishing and verifying the high-fidelity benchmark finite element model, the next step is to identify and extract the local design domains that require refined optimization from the global model.

[0046] The identification of the web-beam cap T-junction region, which is the focus of this method, is primarily based on stress analysis results of the baseline model under one or more static ultimate load conditions. By applying load conditions such as the ultimate gust or ultimate operating gust as defined in the International Electrotechnical Commission standard "Design Requirements for Wind Turbine Generators" (IEC 61400-1) to the entire blade model and performing preliminary nonlinear finite element static analysis, a global stress, strain, and damage distribution cloud map of the blade can be obtained. Through post-processing, high-stress concentration areas with the highest stress values, the most drastic stress gradient changes, or von Mises equivalent stresses exceeding the warning threshold can be automatically and quantitatively identified. After identifying this high-risk T-junction region, a segment of this region along the blade span is extracted from the geometry and mesh of the baseline model as the design domain for this local optimization. During the truncation process, it is essential to ensure that the design domain spatially encompasses the ends of the web, the contact area of ​​the beam cap, the upper and lower adhesive layers, and the surrounding composite matrix area. This is crucial to guarantee a smooth mechanical and geometric transition between the subsequently optimized novel structure and the original, unoptimized blade structure. Furthermore, the truncation of the local design domain must inherit its mechanical environment, including the corresponding boundary conditions and loads, from the global analysis results of the baseline model.

[0047] In a preferred embodiment, this can be achieved using the mature submodeling technique in commercial finite element software. The principle is to apply a displacement boundary condition to all nodes of the intercepted boundary of the local design domain. The displacement value is exactly equal to the displacement result calculated by the global model at these corresponding nodes under the same load conditions.

[0048] After obtaining the finite element model of the local design domain with clearly defined boundary conditions, the next step is to calculate and define the co-constraint conditions.

[0049] The first step is the extraction and parameterization of dynamic constraints. To proactively suppress harmful vibrations at specific frequencies during the structural design phase and improve fatigue life, finite element modal analysis (FEM) of the blade segment containing the local design domain is required. This analysis aims to solve for a series of natural frequencies and corresponding vibration modes (i.e., mode shapes) of the structure under undamped free vibration. From the calculated mode shape results, one or more target suppression vibration modes with the greatest impact on the fatigue life of the T-joint region need to be identified. The identification criteria can comprehensively consider multiple factors, such as whether the natural frequency of the mode is close to the frequency of a known external periodic excitation source (e.g., the characteristic frequency generated by the tower shadow effect or wind shear effect), and whether its vibration mode produces significant relative displacement or high local strain energy density in the T-joint region. After determining the target suppression vibration mode, one or more vibration nodal lines are extracted from the mode displacement field of that mode. Physically, a vibration nodal line is a set of spatial points where the vibration displacement of the structure is always zero when vibrating in that specific mode.

[0050] In a specific algorithm implementation, this extraction process can be achieved by using an isosurface extraction algorithm (e.g., the MarchingCubes algorithm) to search for and construct spatial isosurfaces in the discrete finite element nodal displacement field, where all displacement values ​​are zero or close to a minimal threshold. The extracted vibration nodal lines typically represent a set of discrete three-dimensional points. To make them usable as geometric constraints in subsequent optimization algorithms, they need to be parameterized. For example, B-spline curves or non-uniform rational B-spline (NURBS) curves can be used, and the point set can be fitted using the least squares method, transforming it into a continuous and differentiable mathematical expression defined by a few control points.

[0051] Parallel to dynamic constraints is the quantitative definition of the interfacial stress objective. In the established finite element model of the local design domain, the set of elements representing the adhesive layer's cohesive zone, or the first layer of composite material elements in contact with the adhesive layer, is precisely defined as the stress monitoring zone. The objective function of the topology optimization problem in this method is set to minimize the peak stress within this stress monitoring zone to delay or avoid interfacial damage and cracking. Since peak stress typically only occurs on discrete, single finite element elements, directly using it as the optimization objective would make the objective function extremely sensitive to changes in design variables, leading to numerical instability and difficulty in convergence of the optimization process. Therefore, a stress smoothing or cohesion technique is needed to transform the discrete, non-differentiable peak stress into a continuous, differentiable, globally equivalent function.

[0052] In a preferred embodiment, this purpose can be achieved using a P-norm aggregation function or a (KS) function, both of which have the mathematical property that their function values ​​can smoothly approximate the actual maximum value as their internal parameters tend to a large value.

[0053] After completing all the above preparatory work, the system enters the local topology optimization solution stage with cooperative constraints. The first step is the construction of the topology optimization mathematical model. The optimization design variable of this model is the relative density of each finite element within the local design domain, whose value is constrained to a range between zero and a very small positive number (to avoid stiffness matrix singularities during iteration) and one. The objective function of this mathematical model is set to minimize the smoothed equivalent peak stress defined in the stress monitoring zone in the previous steps. The constraints of this model include at least the following two: First, a volume constraint, which specifies an upper limit on the total volume of material allowed to be used to construct the new connection structure; for example, this upper limit can be set to no more than 30% of the volume of solid material in the original design domain. Second, a dynamic constraint. This mandates that the main load-bearing path or geometric center axis of the region generated by topology optimization, composed of high-density material, must coincide in three-dimensional space with the vibration node line calculated and parameterized in the previous steps, or remain within a pre-defined geometric tolerance range. The dynamic constraint can be implemented mathematically in various ways. For example, the spatial distance function between the geometric center point of the high-density region and the vibration node line can be used as a penalty term, multiplied by a large penalty factor, and added to the objective function; or the distance function can be directly used as an inequality constraint condition, requiring its value to be less than a certain given tolerance.

[0054] After the mathematical model is constructed, a suitable optimization solver is used for iterative calculations, such as a level set method or a variable density method (e.g., SIMP, Solid Isotropic Material with Penalization). The iterative steps of the entire topology optimization algorithm are as follows: First, the design variables of all elements are initialized. Then, the main iteration loop is entered. In each iteration step, the finite element solver is first invoked to perform a static finite element analysis to solve for the nodal displacements and stress fields of the structure under the current material distribution. Then, sensitivity analysis is performed, i.e., calculating the partial derivatives of the objective function and all constraints with respect to the relative density of each design element. Next, based on the calculated sensitivity information, a mathematical optimization algorithm (e.g., Moving Asymptote Method (MMA) or Optimality Criterion Method (OC)) is used to update all design variables. After updating the design variables, a filtering technique (e.g., a density filter or a Helmholtz-type partial differential equation filter) is applied to eliminate the common checkerboard phenomenon in the results and introduce a minimum feature size to ensure the manufacturability of the optimization results. Finally, check if the preset convergence criteria (e.g., the relative change of the objective function over several consecutive steps is less than one ten-thousandth, or the maximum change of the design variable is less than one thousandth) are met. If convergence is not achieved, return to the beginning of the iteration loop; if convergence is achieved, the iteration calculation ends, and the final density contour map, continuously distributed between zero and one, is output.

[0055] After the optimization solution is completed and the final density cloud map is obtained, the process enters the final stage: the generation and application of novel connection structures. First, the continuous density values ​​output by optimization are binarized by setting a density threshold (e.g., 0.5). Cells with densities above this threshold are defined as solid materials, while those below are defined as voids. Then, an isosurface extraction algorithm (such as MarchingCubes) is used to extract the boundary surfaces between the material and voids from this binarized 3D scalar field. This initially extracted surface, which may exhibit discretized step-like effects, undergoes a series of geometric post-processing operations, such as Laplacian smoothing or more advanced surface fitting algorithms, to obtain a smooth, continuous, and self-intersecting geometric model. Finally, the optimized geometric data is imported into computer-aided design (CAD) software for the final reconstruction of a CAD 3D model that can be used to guide manufacturing. This reconstruction process requires accurately reproducing the organic geometric morphology revealed by the optimization results and conforming to mechanical principles. For example, it ensures that the geometric central axis of the generated, skeletal-like reinforcing ribs precisely matches the vibration node lines serving as optimization constraints, and reproduces the optimized special contact interfaces, which may have wavy or finger-like meshing patterns and enhance interfacial bonding performance. Finally, the CAD model of this reconstructed novel connection structure is replaced back into the initial, global baseline blade model using Boolean operations and other methods, replacing the original T-shaped connection portion, thus forming the final, performance-optimized complete blade design. This design can be used to guide subsequent blade mold manufacturing and composite material layup process design.

[0056] The preferred embodiments of the present invention disclosed above are merely illustrative of the invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the invention to specific implementations. Clearly, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize the invention. The invention is limited only by the claims and their full scope and equivalents.

Claims

1. A method for creating blade structures based on load-bearing flow topology optimization, characterized in that, The method includes the following steps: A three-dimensional finite element reference model of a wind turbine blade is established, and a predetermined connection area formed by at least two blade components by adhesive bonding is identified and extracted from the reference model as a local design domain. Finite element modal analysis is performed on the blade or its segment to identify a target vibration suppression mode, and one or more vibration nodal lines are extracted from the mode shape of the target vibration suppression mode. Specifically, the step of extracting the vibration node line includes: Using an isosurface extraction algorithm, a set of spatial points with zero or less than a preset threshold vibration displacement is identified in the modal displacement field of the target suppressed vibration mode. The set of spatial points is parameterized into one or more continuous spatial curves using a curve fitting algorithm. Within the local design domain, a stress monitoring zone representing the adhesive layer or material interface is defined. A topology optimization mathematical model for the local design domain is established and solved. Specifically, the optimization objective of the model is to minimize an equivalent peak stress within the stress monitoring zone after processing by an aggregation function. The aggregation function is configured to smoothly transform the stress values ​​of multiple elements within the stress monitoring zone into a globally differentiable scalar. The constraints of the model include at least one volume constraint and one dynamic geometric constraint. The dynamic geometric constraint requires that the geometric center axis of the region composed of high-density material generated by topology optimization coincides with the vibration node line in three-dimensional space, or remains within a pre-defined geometric tolerance range. Based on the results of the topology optimization solution, a geometric model of the connected region is generated.

2. The blade structure creation method based on load-bearing flow topology optimization as described in claim 1, characterized in that, The predetermined connection area is a T-shaped connection area between the web and the beam cap.

3. The blade structure creation method based on load-bearing flow topology optimization as described in claim 1, characterized in that, In the step of establishing the benchmark model, the mechanical behavior of the adhesive layer material within the local design domain is defined by a cohesive zone model; the mechanical behavior of the composite material within the local design domain is defined by an orthotropic material model that includes a progressive damage failure criterion.

4. The blade structure creation method based on load-bearing flow topology optimization as described in claim 1, characterized in that, The step of identifying and extracting the local design domain specifically includes: One or more static limit load conditions are applied to the three-dimensional finite element reference model, and finite element analysis is performed to obtain the global stress distribution; Based on the global stress distribution, high stress concentration areas are identified as the local design domains; Using sub-model technology, displacement boundary conditions are applied to the intercepted boundary of the local design domain. These displacement boundary conditions are derived from the displacement analysis results of the benchmark model at the corresponding nodes.

5. The blade structure creation method based on load-bearing flow topology optimization as described in claim 1, characterized in that, The steps for generating the geometric model of the blade connection structure specifically include: The topology optimization solution is binarized and isosurfaces are extracted to obtain an initial boundary surface; The initial boundary surface is smoothed. Based on the smoothed surface, a computer-aided design model that can be used for manufacturing is reconstructed.

6. The blade structure creation method based on load-bearing flow topology optimization as described in claim 5, characterized in that, The reconstructed computer-aided design model has geometric features including at least one of the following: Reinforcing ribs arranged along the vibration node line; A wavy or finger-shaped meshing interface.

Images