Solidworks finite element analysis is applied to the design and development of machinery, automobiles, household appliances, electronic products, furniture, architecture, medical orthopedics and other products. Its function is to ensure the safety and rationality of product design, and at the same time find out the best scheme of product design through optimization design, so as to reduce the consumption or cost of materials;

Identify potential problems in advance before product manufacturing or engineering construction; Simulate various test schemes to reduce test time and cost;

It is the core technology of product design and development. Kanban is based on more than ten years of project experience and training experience, reminding friends that finite element analysis is different from drawing. The following is the application method of solidworks finite element analysis summarized by Kanban. I hope it is useful to everyone.

First, the software form:

(a) solidworks built-in form:

Simulation Xpress-only static analysis of some parts with simple load and support type.

(B) SolidWorks plug-in form:

Simulation Works Designer- static analysis of parts or assemblies.

Simulate working spring-static, heat conduction, deformation, frequency, drop test, optimization and fatigue analysis of parts or assemblies.

Simulation work specialty-nonlinear and advanced dynamic analysis has been added to all functions of simulation work.

(3) Form of separate issuance:

Simulation design star- the function is the same as that of Simulation Works Advanced Professional.

Second, the general steps of using finite element analysis:

FEA = Finite Element Analysis-It is an engineering numerical analysis tool, but it is not the only numerical analysis tool! Other numerical analysis tools include: finite difference method, boundary element method and finite volume method.

Methods/steps

1

(A) the establishment of a mathematical model

Sometimes, it is necessary to modify the CAD geometric model to meet the needs of grid division (that is, from CAD geometry to →FEA geometry), * * * has the following three methods:

1. Feature blanking: refers to merging and eliminating unimportant geometric features in analysis, such as rounded corners, edges and marks.

2. Idealization: Idealization is a more active work, such as representing a thin-walled model with a plane (note: if you choose "Use shell grid in the middle" as the grid type, SimulationWorks will automatically create surface geometry).

3. Clear: Because the geometric model used to divide the grid must meet higher requirements than the solid model. For example, the quality problems such as slender surfaces, multiple entities and moving entities in the model will make grid division difficult or even impossible-at this time, we can use CAD quality inspection tools (namely SW menu:

Tools → Check …) to check the problem, and small features with very short sides or faces must also be deleted (small features mean that their feature size is very small relative to the whole model size! However, if the purpose of the analysis is to find out the stress distribution near the fillet, then a very small internal fillet should be kept at this time.

2

(2) Establishing a finite element model, which is the pre-processing part of finite element analysis, includes five steps:

1. Select the grid type and define the analysis type (* * * including static, heat conduction, frequency, etc. )-A FEA instance will be generated at this time, and the configuration name is in brackets after the instance name in the left browser;

2. Add material attributes: Material attributes are usually selected from the material library, and it does not consider factors such as defects and surface conditions. Compared with the geometric model, it has more uncertainties.

(1) Right-click the Solids folder and choose Apply Material to All-all components will have the same material properties.

(2) Right-click a specific part folder under "Entity Folder" and select "Apply Material to All Entities"-all entities (multi-entities) of the part will be given the specified material properties.

(3) Right-click the Solid of a specific part under the Solid Folder and select Apply Material to Solid-only the Solid is given the specified material properties.

3. Impose constraints: Defining constraints is the most error-prone. The usual error comes from the over-constrained model, and the consequence is that the structure is too rigid and underestimates the actual deformation and stress. For assembly, a special "constraint" of "contact/clearance" needs to be defined. The purpose of the constraint is to prohibit the rigid displacement of the model.

There are ten kinds of constraints (excluding "contact/gap") in SimulationWorks. This also means that all these nodes on the specified "point, line and surface" are constrained.

The arrow in the constraint symbol represents the "translation" constraint, and the disk represents the "rotation" constraint (each node of the solid element has only 3 degrees of freedom to move, while the shell element has 6 degrees of freedom).

Right "solid"

Mesh ",because the node has no rotational freedom, the effect of choosing" fixed "and" immovable "is exactly the same. After defining constraints, the spatial position of the model is fixed. At this point, the model can't have any displacement except elastic deformation (in the static analysis of FEA, there may only be elastic displacement), which is called "the model has no rigid displacement".

4. Define the load: In reality, only the size, distribution and time dependence of the load can be roughly known. Therefore, in finite element analysis, approximate estimation must be made through simplified assumptions. Therefore, defining the load will produce a large modeling error (idealization error).

Note: The first four items are collectively referred to as "pretreatment" of finite element analysis, and the uncertainty from high to low is: constraint, load, material and geometric model.

5, grid division:

(1) There are only two types of units in Simulation Works: first-order units (draft quality units) and second-order units (high quality units). Or: solid tetrahedral element and triangular shell element. In this way, SimulationWorks*** has four types of elements: first-order solid tetrahedron element (with only four corner nodes and 1 Gaussian points), second-order solid tetrahedron element (with four corner nodes and six intermediate nodes, * * * counts as 10 nodes and four Gaussian points), and first-order triangular shell element (with only three corner nodes, 1 Gaussian point), the second-order triangular shell element (with three corner nodes and three intermediate nodes, six nodes and three Gaussian points)-the tetrahedron here is not necessarily. In addition, the edges and faces of the second-order element can be bent to simulate the actual deformation of the element due to loading.

(2) You can select the quality of the unit through the SW menu: simulation works→ Options →→→→→ Select Grid tab …

(3) Generally, the element with the least number of nodes in FEA is the beam element, which has only two nodes (i.e. both ends of the beam), but each node has six degrees of freedom (i.e. three translational components plus three rotational displacement components).

(4) The second-order solid tetrahedron element and the second-order triangular shell element are suitable for surface geometry.

(5) Some types of shapes can use both solid elements and shell elements, and the specific choice of which type of elements depends on the purpose of analysis. Generally speaking, however, the natural shape of geometric figures determines the type of elements used. For example, some castings can only be divided by solid grids, while a metal plate is best divided by shell elements.

(6) The degree of freedom in the finite element mesh refers to the degree of freedom of the element nodes. Each node of a solid element has three degrees of freedom (three translational components) and each node of a shell element has six degrees of freedom (three translational components plus three rotational displacement components). The displacement of nodes is the geometric composite vector of these components.

(7) When meshing, the element will be deformed and distorted in the process of matching geometry, but excessive distortion will lead to the degradation of the element, which will lead to an increase in the amount of calculation, greatly reducing the calculation accuracy and even making it impossible to calculate. To do this, you need to control the size of the default cell (i.e. SW menu:

SimulationWorks→Mesh→Create…, where: coarse corresponds to large values and fine corresponds to small values) or local grid control (i.e. SW menu:

Simulation work → grid → application control …) Avoid excessive distortion of the unit.

(8) Grid quality assurance: including aspect ratio check and Jacobian check, which are automatically executed by the program.

Length-width ratio check: the length-width ratio of regular tetrahedron is usually used to calculate the length-width ratio of other units. The aspect ratio of the unit is defined as the length value of the longest side of the tetrahedron/the minimum length value of the normal distance from the vertex of the tetrahedron to its opposite side. Here, the opposite faces of vertices need to be regularized by regular tetrahedron, assuming that the four corners of tetrahedron are connected by straight lines. The aspect ratio of a small regular tetrahedron element can be approximately considered as 1.0. As part of the aspect ratio check, SimulationWorks also automatically performs side length check, inscribed circle and circumscribed circle check and normal length check.

Jacobian check: that is, check the value of Jacobian determinant to judge the bending degree of the unit. The Jacobian of extreme torsion element is negative, which will lead to the termination of finite element program. Jacobian checking is based on a series of points (Gaussian points or nodes) located in each cell. Generally speaking, it is acceptable that the Jacobian ratio is less than or equal to 40. SimulationWorks will automatically adjust the position of the middle node of the twisted unit to ensure that all units can pass the Jacobian test. In the quadratic element, the intermediate nodes on the boundary of the element are placed on the real geometry; However, in sharp wedges and curved boundaries, placing intermediate nodes in real geometry will lead to overlapping twisted elements under the edges. For a regular tetrahedron, all intermediate nodes are exactly located at the midpoint of the straight side, and its Jacobian ratio is 1.0. With the increase of edge curvature, its Jacobian ratio also increases. Jacobian check settings can be realized through the SimulationWorks→options…→Mesh tab.

(9) Local grid control: it is controlled by three parameters-the cell size of the selected entity, the cell size ratio between layers, and the number of cells affected by local optimization. Their default values are 2.2, 1.5 and 3, respectively. Mesh control can be used for points (vertices), lines (boundaries), faces (surfaces) and assembly components. Three control parameters can be controlled by commands: SimulationWorks→Mesh→Apply.

Control ... to achieve. In order to find the largest cell that is still working, you can check the automation in the SimulationWorks→options…→Mesh tab.

Cycling option, the function of "Automatic Solid Cycling" requires the gridding program to subdivide the model with smaller global cell size, and the user can control: the maximum number of cyclic experiments, the range of each reduction of global cell size, and the tolerance.

Grid control applied to components is controlled by Component Importance.

(Effective number of parts) ". For different slide positions, instruct the gridding program to select different cell sizes for gridding each selected component. But if "using the same"

If Element Size is selected, all components will be divided according to the same cell size specified in the grid control window.

(10) The actual grid division process is divided into three steps:

The first step is to evaluate the geometric model-check whether the CAD geometry is defective;

The second step is to deal with the boundary-that is, put the nodes on the boundary first, which is called surface division;

The third step is to create a mesh-fill the solid volume with tetrahedral elements.

(1 1) If the first step fails, it is most likely that the geometric model is wrong. In order to verify whether the geometric model is wrong, use IGES to output the model, and observe whether the error message "Failed to process trimmed surface entity" appears.

(12) If the second step fails, there are two situations: First, if there is an error before the progress indicator bar reaches the extreme right, it means that at least one face is divided incorrectly. At this time, right-click the grid and select "Fault Diagnosis" to find out the problematic surface, and then there is a dividing line or grid control to help divide the surface; Two. When the progress indicator bar reaches the far right and an error occurs before the third step, it is necessary to increase the tolerance from 5% (default) to 10%, and then re-divide the grid. However, if it still fails when the tolerance is 10%, you can continue to increase the tolerance, but the maximum tolerance shall not exceed 25%. Set the command as: SimulationWorks→Mesh→Create…→…

(13) If the third step fails, the error occurred in the volume filling stage. At this time, the tolerance of unit size can be reduced from 5% to 1%. If it still fails, the cell size can be reduced by 25% and the tolerance is set to 1%.

(14) The "Fault Diagnosis" tool is only valid for solid units, but not for shell units.

(15) Since the 2008 edition, SimulationWorks has realized automatic "local grid control", so "grid division" does not need manual intervention at all.

three

(3) solving the finite element model

In structural analysis, FEA first calculates the displacement (vector) of each node in the grid, and then calculates other physical quantities such as strain and stress on this basis; In thermal analysis, FEA first calculates the temperature (scalar) of each node in the grid, and then calculates other physical quantities such as temperature gradient and heat flow.

In general, if the model can be meshed, it can be solved, but if no material or load is defined, the solution will be terminated. The solver can also check the rigid body motion caused by insufficient constraints. However, rigid body motion can be handled by solver options, such as using soft springs to stabilize the model, or using in-plane motion and inertial unloading. Five factors affecting the selection of suitable solvents:

1, the size of the problem-Generally speaking, FFEPlus is faster when its DOF exceeds 100000. As the problem changes, FFEPlus becomes more efficient.

2. Computer resources-When the computer has enough available memory, the DirectSparse solver is faster.

3. Analysis options;

4. Unit type;

5. Material properties-When the elastic modulus of materials used in the model is quite different (such as steel and nylon), the accuracy of FFEPlus (iterative method) is lower than that of DirectSparse (direct method).

If you are not sure which solver is the best choice for analysis, you can set the type of solver to Automatic.

The command to select the solver is: simulation works→ Options …→ Select the Results tab.

four

(4) result analysis

The correct interpretation of the results requires us to be familiar with and understand: I) various assumptions, such as those in static analysis.

Material linearity hypothesis, small deformation hypothesis and static load hypothesis; Ii)。 Simplify the agreement; Iii) Errors generated in the first three steps, such as modeling errors (also called idealized errors), discrete errors (errors generated during gridding) and numerical errors (errors generated during solving). Of these three kinds of errors, only the discretization error is unique to FEA, so this error can only be controlled when using FEA-the smaller the grid element, the lower the discretization error; Modeling errors that affect mathematical geometric models are introduced before finite element analysis, so they can only be controlled by correct modeling techniques. The numerical error (solution error) is generated in the calculation process, which is difficult to control, but usually very small.

Execute "Simulation Works → Options …→ Results Tab →Automatic Results Plots Button" to determine the results of calculation items to be displayed in the program interface.

The node value in the result refers to the stress on the node of the element, while the element value refers to the stress on the Gaussian point of the element.

Element stress and node stress are generally different, but if the difference between them is too big, it means that the grid division is not fine enough.

The analytical solution (the solution obtained by mathematical formula) is effective only when the thickness of the plate is very thin under the assumption of plane stress-because it does not consider the distribution of stress along the thickness direction (gradient distribution: the largest in the middle and the smallest at the two edges), and thinks that the stress on the thickness direction section is equal everywhere. Therefore, the finite element analysis solution can truly reflect the actual state of stress.