METHOD AND DEVICE FOR MODELLING AND FATIGUE STRENGTH ASSESSMENT OF WELD SEAMS BETWEEN MECHANICAL PARTS

Abstract

A method for modelling and fatigue strength assessment of weld seams between mechanical parts includes providing a finite element model for an assembly, in which a first finite element mesh for a mechanical part, a second mesh for a second mechanical part, and a third mesh for a weld seam joining the mechanical parts comprising a number of notches are generated. The third mesh has fewer than 20 finite elements in cross-section, the notches are modelled sharp-edged, and the distribution of the finite elements follows a defined mesh pattern. The method includes calculating the finite element model. Result values of the finite elements and nodes are provided from the defined mesh pattern of the third mesh. The method includes applying an effective notch stress prediction algorithm matched to the defined mesh pattern to predict occurring notch stresses in the notches using the provided result values as input parameters.

Claims

1. Computer-implemented method for model generation and fatigue strength assessment of weld seams between mechanical parts of an assembly with the aid of a finite element method, characterized by: a) providing a finite element model for the assembly, in which a first finite element mesh for a mechanical part, a separate second finite element mesh for a second mechanical part and a third finite element mesh for a weld seam joining the first mechanical part and the second mechanical part comprising a number of notches are generated, wherein the third finite element mesh has a number of less than 20 finite elements in cross-section, the notches of the weld seam are modelled sharp-edged and the distribution of the finite elements follows a defined mesh pattern, b) calculating the finite element model, wherein result values of the finite elements and nodes are provided from the defined mesh pattern of the third finite element mesh for the weld seam, and c) applying an effective notch stress prediction algorithm matched to the defined mesh pattern of the third finite element mesh to predict occurring notch stresses in the notches using the provided result values as input parameters.

2. Method according to claim 1, characterized in, that the first finite element mesh and the third finite element mesh are coupled by means of a first number of coupling elements and the second finite element mesh and the third finite element mesh are coupled by means of a second number of coupling elements.

3. Method according to claim 1, characterized in, that step b) is formed by: b1) calculating the finite element model, and b2) evaluating the result values of the finite elements and the nodes of the defined mesh pattern for the weld seam on the basis of the calculated finite element model.

4. Method according to claim 3, characterized in, that in step b2) result values are exclusively evaluated within the finite elements and the nodes of the third finite element mesh for the weld seam.

5. Method according to claim 3, characterized in, that the result values, which are evaluated in step b2), include stress results, reaction force results, geometry parameters, and/or material parameters.

6. Method according to claim 3, characterized in, that the result values, which are evaluated in step b2), consist of stress results, reaction force results, geometry parameters, and/or material parameters.

7. Method according to claim 1, characterized in, in that the effective notch stress prediction algorithm is trained with a plurality of weld seam parameter variants using the defined mesh pattern before the application of step c)

8. Method according to claim 1, characterized in, in that in step c) a plurality of parameters of the notches are predicted by means of the effective notch stress prediction algorithm.

9. Method according to claim 8, characterized in, that the parameters include: normal stresses, shear stresses, Von Mises equivalent stresses, radial-, tangential stress components, and/or axial stress components in the notch radius of the respective notch.

10. Method according to claim 1, characterized in, that the predicted notch stresses are then used to perform fatigue strength assessments of the assembly.

11. Computer program product which, on a program-controlled device, initiates the execution of the method according to claim 1.

12. Computer-implemented device for modelling and fatigue strength assessment of weld seams between mechanical parts of an assembly with the aid of a finite element method, characterized by: a first unit for providing a finite element model for the assembly, in which a first finite element mesh for a first mechanical part, a separate second finite element mesh for a second mechanical part and a third finite element mesh for a weld connecting the first mechanical part and the second mechanical part comprising a number of notches, wherein the third finite element mesh comprises a number of less than 20 finite elements in cross-section, wherein the notches of the weld seam are modelled with sharp edges and the distribution of the finite elements follows a defined mesh pattern, a second unit for calculating the finite element model, wherein result values of the finite elements and nodes are provided from the defined mesh pattern of the third finite element mesh for the weld seam, and a third unit for applying an effective notch stress prediction algorithm matched to the defined mesh pattern of the third finite element mesh to predict occurring notch stresses in the notches using the provided result values as input parameters.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

[0058] FIG. 1 shows a cross-section of a welded assembly;

[0059] FIG. 2 shows as the first example the Structural Hot Spot Stress method as the state of the art;

[0060] FIG. 3 shows as the first example the Effective Notch Stress method as the state of the art;

[0061] FIG. 4 shows the cross-section of a multibody FE model with the weld seam modeled according to the invention;

[0062] FIG. 5 shows the cross-section of a multibody FE model with a variant of a weld seam modelled according to the invention;

[0063] FIG. 6 shows the cross-section of a multibody FE model with a further variant of a weld seam modelled according to the invention;

[0064] FIG. 7 shows cross-sections of FE-models of assemblies with different variants of welds modelled according to invention;

[0065] FIG. 8 shows various applications of welds modelled according to the invention;

[0066] FIG. 9 shows a schematic flow chart of an execution example of a computer-implemented method for modelling and fatigue strength assessment of welds between mechanical parts of an assembly using a finite element method; and

[0067] FIG. 10 shows a schematic block diagram of an embodiment of a computer-implemented device for modelling and fatigue strength assessment of welds between mechanical parts of an assembly using a finite element method.

DETAILED DESCRIPTION

[0068] In the figures, identical or functionally identical elements have been provided with the same reference signs, unless otherwise indicated.

[0069] Embodiments for model generation and fatigue strength assessment of welds 4a, 4b between mechanical parts 2, 3 of an assembly are explained with common reference to FIGS. 4 to 9. FIGS. 4 to 7 show examples of FE models of an assembly with a weld seam according to the invention. Furthermore, FIG. 8 shows applications of weld seams modelled according to the invention. Furthermore, FIG. 9 shows an implementation example of a computer-implemented method for model generation and strength evaluation of welds 4a, 4b between mechanical parts 2, 3 of an assembly with the aid of a finite element method.

[0070] Starting with FIG. 4, it shows an FE model 40 of a welded assembly. A first mechanical part 2 and a second mechanical part 3 of the assembly are welded by means of two welds 4a, 4b. The notches of weld 4a are marked with the reference signs 5a, 5b, 5c and the finite elements representing weld 4a are marked with the reference signs 6a, 6b, 6c. For reasons of clarity, only the notches and finite elements of weld 4a are marked with reference signs, but not the notches and finite elements of weld 4b.

[0071] The finite elements 6a, 6b, 6c are especially designed as 3D volume elements for 3D models and as 2D elements for 2D models and supplement the FE model 40 of the welded component. The components 2, 3 and the welds 4a, 4b can be meshed either with common nodes or independently of each other with separate nodes. A separated, independent meshing has the advantage that the variation of the weld seam geometry is easier to achieve and no changes to the basic model of the mechanical part 2, 3 are necessary.

[0072] With independent meshing, the weld seam elements 6a, 6b, 6c are connected to mechanical parts 2, 3 with the aid of FE coupling elements 7a, 7b (see also FIG. 5). Examples for coupling elements 7a, 7b include FE contact elements, FE coupling bars or coupling equations.

[0073] The independent meshing and connection with FE coupling elements 7a, 7b is made possible, since preferably result values are only evaluated from within the weld seam elements or nodes.

[0074] The welds 4a, 4b are meshed with a defined mesh pattern, whereby these mesh patterns are matched to a subsequently used effective notch stress prediction algorithm. The weld seam elements 6a, 6b, 6c preferably have a predefined number, a predefined distribution and a predefined position within the weld seam 4a, 4b. The notches 5a, 5b, 5c of the weld 4a are not rounded but modelled sharp-edged. This allows a relatively coarse meshing and thus saves considerable calculation effort and calculation time. The geometry, the dimensions and the position of the respective weld 4a, 4b are preferably modelled realistically, which is why the stiffness is represented with good accuracy (betters than with FE shell modelling) and the defined mesh pattern is proportionally adapted to the given weld geometry. FIG. 7 shows some examples of structured mesh patterns 8a, 8b, 8c for welds 4a, 4b. It should be noted that the effective notch stress prognosis algorithm used in the following is adapted to the mesh pattern 8a, 8b, 8c used. The creation of a weld seam mesh with defined mesh pattern 8a, 8b, 8c can be carried out automatically using software routines (see method step S1 of FIG. 9).

[0075] The FE model of the assembly prepared in this way, including the weld seams 4a, 4b, is then solved using an established FE calculation method and the results are evaluated (see process step S2 of FIG. 9).

[0076] A number of parameters of the weld seams 4a, 4b are evaluated and made available to the effective notch stress prognosis algorithm as input data. The parameters can be stresses, strains and/or reaction forces of the weld seam elements 6a, 6b, 6c and nodes. In addition, material and/or geometry parameters such as the dimensions of the weld cross-section, relative position coordinates of individual nodes within the weld cross-section or connection angles of the connected geometry in the individual weld cross-sections and notches can be used.

[0077] The effective notch stress prediction algorithm includes in particular metamodels or response surface methods, such as [0078] Global polynomials [0079] Moving leased squares [0080] Kriging [0081] Radial basis functions [0082] Neuronal networks

[0083] These models are created (fitted or trained) using: [0084] Regression [0085] Interpolation [0086] Extrapolation

[0087] The effective notch stress prediction algorithms are each fitted (trained) to a given weld modelling method with a given mesh pattern. Input data of the effective notch stress prediction algorithm is a relevant subset of the above mentioned parameters. Output data are effective notch stresses and notch stress components for each weld notch per weld cross-section.

[0088] In order to fit (train) the effective notch stress prediction algorithm, preferably a sufficient number of weld seam constellations is calculated. Each weld constellation has different geometric dimensions (parameters) of the components and the weld as well as different loads (parameters) and represents a design point in the parameter space. From each design point, preferably firstly the present modelling method and secondly a variant with a standard notch rounding radius and very fine meshing as shown in FIG. 3 is calculated. The second model provides the reference results (target values) of the weld seam notch stresses and the first model provides the input data for the effective notch stress prediction algorithm. Thus, the effective notch stress prediction algorithm is fitted (trained) to the existing meshing pattern, the given notch radius and the existing modelling method of the first model. The trained algorithm can then be applied to productive FE models to predict weld notch stresses. Even though the prediction accuracy may be slightly lower than with the classical rounded and finely meshed Effective Notch Stress method, the method according to the invention still results in an enormous advantage, since considerably shorter calculation times can be achieved with considerably fewer nodes, or it is only made possible in the first place that Effective Notch Stress method can be applied economically to complex finite element models with a high number of weld seams. Without the present method, the number of elements and nodes for Effective Notch Stress calculations on complex models would be too large to be calculated economically.

[0089] With the effective notch stresses and notch stress components predicted in this way, a fatigue strength assessment of the weld seam can then be carried out.

[0090] This effective notch stress evaluation method is applied to a cross-section of a weld seam, i.e. new local notch stresses can be predicted at defined intervals in the longitudinal direction of the weld seam.

[0091] The modeling method can be used in the same way for different weld seam applications. FIG. 8 shows the possible applications for T-joints 9a, butt joints 9b and overlap joints 9c. For double-sided welded joints, one weld seam model is preferably used on each side. The effective notch stress prediction algorithm can preferably be fitted in such a way that it can be used unchanged for all these applications. For higher prediction accuracy, however, specialized effective notch stress prediction algorithms can also be fitted for individual applications.

[0092] As shown, for example, in FIGS. 7-8b and 8c, simple fillet welds involve welding components without geometric weld preparation. In order to obtain a better and continuous mechanical connection, the mechanical parts are also often connected as shown in FIG. 8a and thus provided with a geometric weld seam preparation. FIG. 4 shows a weld seam modelled in accordance with the invention in which the weld seam preparation on the component is fully modelled and meshed. However, as shown in FIG. 5, the components can also be modelled without weld seam preparation. This is made possible by independent meshing of the weld and the connection of the welds to the neighboring parts via FE coupling elements or coupling equations 7a. This facilitates the variation of the weld seam geometry without having to change the finite element model of the mechanical parts themselves.

[0093] As shown in FIG. 6, the present modeling method also allows the application to shell models 60 in the same way, where components 2, 3 are meshed with finite shell elements in the center plane of the mechanical parts 2, 3 and welds 4a, 4b with the defined mesh pattern and the real weld geometry. The connection is again made with coupling elements or coupling equations 7a, 7b. The present modeling and notch stress prediction method can therefore be used in many different applications (FIG. 4, 5, 6, 8).

[0094] FIG. 9 shows a schematic flowchart of an execution example of a computer-implemented method for model generation and fatigue strength assessment of welds 4a, 4b between mechanical parts 2, 3 of an assembly using a finite element method. The procedure of FIG. 9 comprises the process steps S1 to S3 and is explained with reference to FIGS. 4 to 8:

[0095] In step S1, a finite element model 40 (see FIG. 4), 50 (see FIG. 5), 60 (see FIG. 6) is provided for the assembly. For finite element model 40, 50, 60 a first finite element mesh for a first mechanical part 2, a separate finite element mesh for a second mechanical part 3 and a third finite element mesh for a weld 4a, 4b connecting the first mechanical part 2 and the second mechanical part 3 having a number of notches 5a, 5b, 5c is created. The third finite element mesh has a number of less than 20 finite elements 6a, 6b, 6c in cross-section. The notches 5a, 5b, 5c of the weld 4a, 4b are modelled sharp-edged. The distribution of the finite elements follows a defined mesh pattern 8a, 8b, 8c (see FIG. 8). For example, the first finite element mesh and the third finite element mesh are coupled by a number of FE coupling elements 7a, 7b, 7c and the second finite element mesh and the third finite element mesh are coupled by a second number of FE coupling elements 7a, 7b, 7c.

[0096] In step S2, the finite element model 40, 50, 60 is calculated. From the defined mesh pattern 8a, 8b, 8c of the third finite element mesh for the welds 4a, 4b result values of the defined elements and nodes are provided.

[0097] For example, step S2 comprises the following substeps: [0098] Calculating the finite element model 40, 50, 60, and [0099] Evaluating the result values of the finite elements and the nodes of the defined mesh pattern 8a, 8b, 8c for the weld 4a, 4b on the basis of the calculated finite element model 40, 50, 60.

[0100] Preferably, result values are evaluated exclusively within the finite elements and nodes of the third finite element mesh for the weld 4a, 4b. The result values preferably comprise and consist of stress results, reaction force results, geometry parameters, and/or material parameters.

[0101] In step S3, an effective notch stress prediction algorithm is applied to predict occurring stresses in notches 5a, 5b, 5c using the provided result values as input parameters. The effective notch stress prediction algorithm predicts the occurring notch stresses in the notches 5a, 5b, 5c in their rounded state. The applied effective notch stress prediction algorithm is adapted to the defined mesh pattern 8a, 8b, 8c of the third finite element mesh.

[0102] The effective notch stress prediction algorithm is preferably trained before its application with a plurality of weld parameter variants for weld 4a, 4b using the defined mesh pattern 8a, 8b, 8c. By means of the effective notch stress prediction algorithm a plurality of parameters is predicted. This plurality of parameters comprises: principal stresses, shear stresses, Von Mises equivalent stresses, radial-, tangential- and/or axial-stress components in the notch radius of the respective notch 5a, 5b, 5c.

[0103] The predicted notch stresses can then be used to perform fatigue strength assessments of the assembly.

[0104] FIG. 10 shows a schematic block diagram of a design example of a computer-implemented device 100 for modelling and fatigue strength assessment of welds 4a, 4b between mechanical parts 2, 3 of an assembly using a finite element method.

[0105] The device 100 comprises a first unit 101, a second unit 102 and a third unit 103.

[0106] The first unit 101 is configured to provide a finite element model 40, 50, 60 for the assembly, in which a first finite element mesh for a first mechanical part 2, a separate second finite element mesh for a second mechanical part 3 and a third finite element mesh for a weld 4a, 4b connecting the first mechanical part 2 and the second mechanical part 3 comprising a number of notches 5a, 5b, 5c are generated. The third finite element mesh has a number of less than 20 finite elements 6a, 6b, 6c in the notches 5a, 5b, 5c of the weld 4a, 4b are sharp-edged and the distribution of the finite elements follows a defined mesh pattern 8a, 8b, 8c.

[0107] The second unit 102 is configured to calculate the finite element model 40, 50, 60, whereby 8a, 8b, 8c of the defined mesh pattern of the third finite element mesh are provided for the weld 4a, 4b result values of the finite elements and nodes.

[0108] The third unit 103 is configured to apply an effective notch stress prediction algorithm matched to the defined mesh pattern 8a, 8b, 8c of the third finite element mesh for predicting occurring notch stresses in the notches 5a, 5b, 5c using the provided result values as input parameters.

[0109] Although the present invention was described by means of design examples, it can be modified in many ways.

LIST OF REFERENCE CHARACTERS

[0110] 2 mechanical part [0111] 3 mechanical part [0112] 4 weld seam [0113] 4a weld seam [0114] 4b weld seam [0115] 5a notch [0116] 5b notch [0117] 5c notch [0118] 6a finite element [0119] 6b finite element [0120] 6c finite element [0121] 30 finite-element-model [0122] 7a coupling element [0123] 7b coupling element [0124] 7c coupling element [0125] 8a mesh pattern [0126] 8b mesh pattern [0127] 8c mesh pattern [0128] 9a T-joints [0129] 9b butt welds [0130] 9c lap joint [0131] 10 assembly [0132] 40 finite-element-model [0133] 50 finite-element-model [0134] 60 finite-element-model [0135] 100 device [0136] 101 first unit [0137] 102 second unit [0138] 103 third unit [0139] S1 method step [0140] S2 method step [0141] S3 method step

REFERENCES

[0142] [1] IIW Fatigue Recommendations: Recommendations for Fatigue Design of Welded Joints and Components from International Institute of Welding (IIW) A. F. Hobbacher [0143] [2] FKM Guideline: Analytical Strength Assessment of Components from Forschungskuratorium Maschinenbau (FKM) (VDMA Verlag) [0144] [3] CN103838975A [0145] [4] DE102012023670A1 [0146] [5] DE102014224129A1 [0147] [6] EP1337942B1 [0148] [7] EP3267338A1 [0149] [8] JP2003080393A [0150] [9] US2013325417A1