A multi-scale calculation and prediction method for post-weld deformation of cold plate friction stir welding
By employing a multi-scale calculation method, the problem of rapid and reliable prediction of post-weld deformation of long-path welds in the integrated connection of cold plate and base plate was solved, enabling optimization of welding parameters and processes, and improving assembly accuracy and sealing consistency.
Patent Information
- Application Number
- CN202611600948.X
- 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
Existing technologies struggle to quickly and reliably predict the post-weld free state morphology of long-path welds during the integrated connection of cold plates and base plates. This is especially true under clamping constraints and springback effects, making it difficult to achieve precise process design and solution iteration.
By employing small-scale local weld zone calibration and inherent strain extraction, a mapping between process parameters and inherent strain is established. Combined with medium-scale welding path segmentation correction and large-scale macrostructure prediction, post-weld deformation is predicted through multi-scale calculation methods. This includes establishing a small-scale calibration calculation model for the local weld zone, a medium-scale welding path segmentation correction model, and a large-scale macrostructure prediction model, which progressively maps the effects of heat accumulation and clamping/unclamping during the welding process.
It enables rapid and reliable prediction of post-weld deformation of long weld seam structures, provides a basis for the coordinated optimization of welding parameters, path sequence, clamping and thermal boundary, and improves the assembly accuracy and sealing consistency of integrated cold plate structures.
Smart Images

Figure CN123116728A_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of welding forming and deformation control technology, and relates to computer-aided engineering analysis and process optimization design in the integrated friction stir welding manufacturing process of cold plate and automobile floor or battery pack floor. Specifically, it is a multi-scale calculation and prediction method for post-weld deformation of cold plate friction stir welding using small-scale, medium-scale and large-scale models, inherent strain mapping and finite element simulation. Background Technology
[0002] As new energy vehicle power battery systems develop towards structural integration and lightweighting, cold-plates have gradually evolved from single thermal management components into functional components that are integrated with the battery pack base plate or vehicle body base plate. Cold-plates typically consist of thin cover plates, flow channels, and frames. The internal flow channels alter the heat conduction path and weaken local stiffness, making them highly sensitive to precision during manufacturing and assembly. Friction stir welding, as a solid-state joining process, is suitable for manufacturing continuous long welds in thin-walled aluminum alloy structures and has become one of the commonly used processes for the integrated connection of cold-plates and base plates.
[0003] In actual production, cold-rolled steel plate weld paths are often long and contain corners, start and stop points, or closed loops. Uneven distribution of heat input, heat dissipation, and structural stiffness during welding easily leads to non-uniform shrinkage and residual stress. Differences in thermal conductivity and constraint within the internal flow channel region amplify the temperature gradient near the weld, resulting in varying degrees of heat accumulation across different path segments and consequently, regional differences in post-weld deformation. In production, strong clamping conditions are typically used to suppress immediate displacement during the welding stage, and backplate heat conduction and cooling measures are employed to reduce heat accumulation. However, the redistribution of residual stress after unclamping triggers springback, significantly amplifying warping in the final free state, local waviness, and height fluctuations in the sealed area. Existing prediction methods, if employing full-process fine-grid temperature field and structural elastoplasticity calculations, are computationally expensive and difficult to use for rapid multi-parameter evaluation in different areas. If only a single scale is used or path heat accumulation and boundary changes are ignored, it is difficult to reflect the dominant role of long-path welding and unclamping springback on the final morphology, making it difficult to support process design and scheme iteration.
[0004] Therefore, an engineering method is needed that can both quantify the welding drive of local weld zones and characterize the heat accumulation of long welds, enabling rapid output of free-state deformation after unclamping at the macroscopic structural level. This invention employs small-scale local weld zone calibration and inherent strain extraction to establish a mapping between process parameters and inherent strain, allowing local plastic deformation to be transferred to subsequent models in the form of equivalent parameters. It uses a medium-scale approach to segment the welding path and continue the previous temperature state, correcting for heat accumulation and start / stop point effects on inherent strain, forming segmented inherent strain factors that can be directly used for macroscopic loading. Finally, it uses large-scale macroscopic structural inherent strain prediction and simulation of the entire clamping and unclamping process, outputting key indicators such as overall warpage and local waviness, thereby achieving rapid prediction of post-weld deformation and providing feasible computational basis for the coordinated optimization of welding parameters, path sequence, clamping, and thermal boundaries. Summary of the Invention
[0005] The purpose of this invention is to provide a multi-scale calculation and prediction method for post-weld deformation of cold-plate friction stir welding, addressing the problem of rapid and reliable prediction of the free-state morphology of long-path weld structures under clamping constraints and springback after unclamping. This invention parameterizes the irreversible plastic deformation generated by local welding as inherent strain and maps it step-by-step to the macroscopic structure, achieving rapid prediction and process optimization under multi-process parameter conditions in different zones. To achieve this objective, the invention provides the following technical solution: A multi-scale calculation and prediction method for post-weld deformation of cold-rolled steel plate by friction stir welding includes the following steps: Step 1: Establish a small-scale calibration calculation model for the local weld zone, and use the coupled Euler-Lagrange method to numerically simulate the friction stir welding process to obtain the welding thermal cycle history; Step 2: Based on the temperature field and stress-strain field calculated by the small-scale model, extract the irreversible plastic strain of the weld center and heat-affected zone, obtain the equivalent inherent strain through regional averaging, change the process parameters and repeat the calculation to establish the mapping relationship between process parameters and equivalent inherent strain. Step 3: Establish a segmented correction model for the mesoscale welding path. Divide the weld path into multiple segments according to its characteristics and use the termination temperature of the previous segment as the initial temperature of the next segment. Conduct mesoscale welding experiments to obtain the temperature time history curves of representative positions in each segment to quantify the heat accumulation effect. Combine the mapping relationship to correct the inherent strain of each segment and form the inherent strain factor of the segment. Step 4: Establish a large-scale macroscopic structural inherent strain prediction model, apply the segmented inherent strain factor in the weld path area, apply clamping constraints and thermal boundary conditions consistent with the actual tooling to perform mechanical analysis of the welding stage, and release the constraints step by step in a preset order to simulate the unclamping process and obtain the deformation results of the free state after welding. Step 5: Based on the results of Step 4, establish an approximate relationship between the segmented inherent strain factor and key deformation index to achieve rapid prediction of macroscopic deformation under multi-parameter conditions in the partitioned area, and perform collaborative optimization of welding parameters, path sequence, start and stop point positions, and clamping and heat conduction boundaries.
[0006] This technical solution extracts inherent strain and establishes a process mapping through small-scale calibration, equating localized complex plastic behavior to parameterized inputs; mesoscale experiments correct for the effects of heat accumulation, yielding segmented inherent strain factors; large-scale loading factors are applied and the entire process is simulated, rapidly outputting free-state deformation; ultimately achieving integrated prediction and optimization. The entire process avoids directly solving local plasticity using macroscopic models, balancing computational efficiency and accuracy, and can be applied to engineering scenarios involving long welds, multiple segments, and variable boundaries.
[0007] In a preferred embodiment, the coupled Eulerian-Lagrange method described in step one defines the stirring tool as a Lagrangian body and the welding material region as an Eulerian body. Mesh refinement is applied to the weld center, heat-affected zone, and shoulder action area. Backplate contact heat conduction, surface convection, and radiative heat boundary conditions are set. The temperature field, stress-strain field, and material flow characteristics are obtained through transient thermo-mechanical coupling calculations. This accurately describes the strong plastic flow and friction-plastic heat input coupling during friction stir welding, ensuring the accuracy of local temperature cycling and deformation history calculations.
[0008] In a preferred embodiment, the extraction of the equivalent inherent strain in step two is as follows: the longitudinal and transverse irreversible plastic strains are averaged along the transverse width and thickness of the weld within a specified extraction area to obtain the equivalent inherent strain containing both longitudinal and transverse components. The mapping relationship is established using methods such as table lookup, interpolation, or regression approximation. This method simplifies the unevenly distributed plastic strain field into a uniform equivalent characteristic quantity, facilitating its transfer to the macroscopic model. Simultaneously, database mapping supports rapid retrieval of any combination of process parameters.
[0009] In a preferred embodiment, the segmentation of the weld path in step three is based on the tooling partition location, the location of abrupt changes in structural stiffness, the flow channel turning point, or the location sensitive to heat accumulation. The segmentation types include straight segments, corner segments, start-stop segments, and closed segments. This can accurately capture the plastic strain changes caused by differences in heat dissipation, stiffness, and heat accumulation in different segments, making the inherent strain factor more physically consistent.
[0010] In a preferred embodiment, the measuring points for the mesoscale welding experiment in step three are arranged near the weld centerline and typical locations in the heat-affected zone, covering the middle of the straight section, the corner transition section, the neighborhood of the start and stop points, and the end of the closed section. Peak temperature, heating rate, cooling rate, and initial temperature rise between segments are extracted from the measured temperature time history curve to quantify the heat accumulation effect. This measuring point arrangement and index extraction can quantitatively characterize the thermal pre-setting effect of preceding segments on subsequent segments during continuous welding, providing a direct basis for correction.
[0011] In a preferred embodiment, step three further includes: conducting comparative experiments and computational analysis by changing the path sequence or start and stop points to verify the applicability of the mapping relationship under different welding sequences, and determining the range of segmented inherent strain correction amplitudes. This step enables the mesoscale correction model to have good generalization ability, ensuring that reliable segmentation factors can be obtained under different welding path planning.
[0012] In a preferred embodiment, the large-scale macroscopic structural model in step four retains the internal flow channels and frame structure features that significantly affect the overall stiffness and heat conduction path, and refines the mesh in the weld path and its neighborhood; the thermal boundary conditions include the contact thermal boundary between the workpiece and the back plate. This modeling method retains the main structural features and heat dissipation conditions that affect deformation while reducing the computational scale, making the macroscopic prediction results closer to the actual working conditions.
[0013] In a preferred embodiment, the release sequence of the step-by-step constraint release simulation unclamping process in step four is from the edge to the center, from the symmetrical point to the center, or from the corner to the straight line segment, solving for the stress redistribution and structural springback after each stage of constraint release. Step-by-step release can realistically reproduce the actual unclamping process, accurately capture the springback deformation process caused by the gradual release of residual stress, and avoid the dynamic effect distortion caused by single-step release.
[0014] In a preferred embodiment, the key deformation indicators in step five include the overall maximum out-of-plane displacement, the height fluctuation deviation of the sealing area, and the waviness of the area adjacent to the weld. The collaborative optimization, under the constraint of meeting weld quality requirements, optimizes the rotational speed, welding speed, axial force, path sequence, start and stop point positions, clamping point distribution, and backplate thermal conductivity. By targeting these indicators directly related to assembly accuracy and sealing performance, the optimal process and tooling scheme can be systematically provided, significantly reducing post-weld deformation.
[0015] In a preferred embodiment, the cold plate includes a cover plate, a flow channel plate, and a frame, and is integrally connected to the base plate by friction stir welding. The weld path is a long weld path, including corners, start and stop points, or closed loops. This method is particularly suitable for thin-walled integrated structures with internal flow channels and multiple long weld segments, and can solve the complex deformation control problems caused by their unique structure and heat accumulation.
[0016] In the above technical solutions, each dependent technical feature can be combined independently or in combination with the technical solution of the independent claim to form a better implementation method for different application scenarios and precision requirements.
[0017] Compared with existing technologies, this invention has the following advantages: at the small scale, it simplifies the local welding process through heat input calibration and inherent strain extraction, avoiding the direct solution of complex local plastic behavior by macroscopic models; at the medium scale, it improves the characterization ability of thermal cycle differences in different regions of long welds through path segmentation and heat accumulation correction, making the inherent strain input more consistent with the actual welding sequence; at the large scale, it obtains the post-weld free state morphology through inherent strain mapping combined with clamping and unclamping process simulation, thereby improving the predictive specificity while ensuring engineering computability, and providing an implementable parameter optimization basis for the assembly accuracy and sealing consistency control of integrated cold plate structures. Attached Figure Description
[0018] Figure 1 This is an overall flowchart of the method for predicting and controlling residual deformation in integrated friction stir welding of cold plate and base plate according to the present invention.
[0019] Figure 2 This is a schematic diagram of a friction stir welding head model.
[0020] Figure 3 The equivalent stress (Von Mises stress) contour map and displacement contour map are obtained from the calculation of the small-scale model.
[0021] Figure 4 The equivalent stress (Von Mises stress) contour plot and displacement contour plot are obtained from the mesoscale model calculation.
[0022] Figure 5 This is a schematic diagram of the finite element model of the integrated macroscopic structure of the cold plate and the base plate.
[0023] Figure 6 This is a cloud map showing the distribution of residual stress during the welding stage.
[0024] Figure 7 The image shows the overall warping deformation cloud diagram of the structure in its free state after welding ((a) without crossbeams and rivets; (b) with crossbeams and rivets). Detailed Implementation
[0025] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are for illustrative purposes only and are not intended to limit the scope of protection of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without inventive effort are within the scope of protection of this invention.
[0026] This implementation method establishes a small-scale local weld zone calculation model, a medium-scale welding path segmentation correction model, and a large-scale macroscopic structure prediction model. It maps the material flow and thermo-mechanical behavior of the local weld zone during friction stir welding from small scale to large scale step by step. Based on considering the welding process parameters, welding path sequence, heat accumulation effect, and changes in clamping and unclamping boundary conditions, it predicts the post-weld residual deformation of the integrated cold plate and base plate structure.
[0027] There is a defined data transfer relationship among the three models: the small-scale model extracts the irreversible plastic strain from the local welding thermo-mechanical coupling process and parameterizes it into an equivalent intrinsic strain, establishing a mapping relationship between process parameters and this equivalent intrinsic strain, providing a reference plastic shrinkage amount for subsequent models; the medium-scale model takes the welding path segments as units, obtains the actual temperature history of each segment through experiments, quantifies the influence of heat accumulation effect on local thermal cycling during continuous welding, and corrects the intrinsic strain of each segment accordingly, generating segment intrinsic strain factors related to the path position; the large-scale model uses the segment intrinsic strain factors as initial strain loads applied to the macroscopic structural model, respectively simulating the clamping constraint state during the welding stage and the step-by-step release process of the constraint during the unclamping stage, and calculates the structural deformation field in the completely free state after welding.
[0028] The technical principle behind this step-by-step mapping approach is that the macroscopic structural deformation caused by friction stir welding is essentially determined by the spatial distribution of irreversible plastic strain within the weld and its heat-affected zone. The amplitude of this plastic strain depends on the local thermal cycle characteristics and mechanical constraints. The small-scale CEL thermo-mechanical coupling model can obtain the temperature and plastic strain fields corresponding to the actual process parameters by solving for the frictional heat generation and plastic dissipation at the tool-material interaction interface, without relying on a preset heat source form. Extracting and parameterizing this plastic strain as an equivalent intrinsic strain allows for model order reduction while preserving deformation-driving information. This eliminates the need for the macroscopic model to directly solve for the transient temperature field of the moving heat source and the local large deformation plastic constitutive model, significantly reducing the computational scale. Mesoscale experimental corrections address the problem that the small-scale local model cannot consider the influence of the preceding weld pass on the initial temperature field of the subsequent weld pass during continuous welding, enabling the intrinsic strain input to have spatial differences along the weld path, thereby improving the accuracy of predicting welding deformation in long welds. Based on this, the large-scale model further introduces clamping constraints and backplate thermal boundary conditions consistent with the actual tooling, and adopts a step-by-step release constraint method to simulate the redistribution of residual stress and structural springback during the unclamping process, so that the final deformation calculation results are consistent with the actual production.
[0029] The relationship and calculation steps of the three models are as follows: Figure 1 As shown, it includes: Step S1: Establish a small-scale local weld zone calculation model, use the Coupled Euler-Lagrange (CEL) method to describe the relative motion between the tool and the material and the plastic flow behavior of the material during friction stir welding, output the temperature field, stress-strain field and irreversible plastic strain distribution of the weld zone, extract the equivalent inherent strain, and establish the mapping relationship between process parameters and equivalent inherent strain.
[0030] The technical benefits of this step are as follows: the CEL method solves the mesh distortion problem caused by large material deformation and flow in friction stir welding, making the thermo-mechanical coupling solution of the local welding process numerically stable; by extracting the equivalent intrinsic strain, the complex local plastic behavior is parameterized into a transferable scalar or vector characteristic quantity, providing deformation-driven input for subsequent macroscopic analysis that does not depend on transient calculations of moving heat sources.
[0031] Step S2: Establish a segmented correction model for the medium-scale welding path. Divide the long weld into straight segments, corner segments, start-stop segments, and closing segments. Combine the temperature time history curves of each typical segment position with medium-scale welding experiments to quantify the heat accumulation effect. Based on the small-scale mapping relationship, correct the inherent strain of each segment to obtain the inherent strain factor of the segment.
[0032] The technical benefits of this step are as follows: by using path segmentation and experimental temperature data, the influence of residual heat from the initial welding stage on the initial temperature and cooling rate of the subsequent welding stage during continuous welding of long welds is quantitatively characterized; by converting the heat accumulation effect into path position-related corrections of inherent strain, the shrinkage driving force input of the macroscopic model has a reasonable spatial distribution, thus solving the problem of insufficient prediction accuracy when a single inherent strain value is used for the entire long weld.
[0033] Step S3: Establish a large-scale macroscopic structural prediction model. Apply inherent strain factors in segments along the weld path in the overall cold plate and base plate structure. Perform mechanical analysis of the welding stage under clamping constraints and thermal boundary conditions. Simulate the unclamping process by releasing constraints step by step to obtain deformation results such as overall warping, local waviness, and height fluctuation of the sealing area in the free state after welding.
[0034] The technical benefits of this step are as follows: replacing transient thermo-mechanical coupling analysis with inherent strain loading allows the calculation of post-weld deformation of macroscopic structures to be completed within the framework of conventional static analysis, with computational efficiency meeting the needs of engineering iteration; at the same time, the simulation of clamping constraints and step-by-step unclamping enables the realistic reproduction of residual stress redistribution and springback process, and the output free-state deformation can be directly used for product assembly accuracy and sealing performance evaluation.
[0035] The specific implementation methods for each step are described in detail below.
[0036] The modeling and calculation method of the small-scale local weld zone calculation model described in step S1 is as follows: Figure 1 As shown, the specific steps include: Step S1-1: Establish a geometric model of the local weld zone. A three-dimensional geometric model of the local weld area is established based on the actual thickness combination of the cold plate and the base plate. A typical length is taken along the welding direction, preferably more than twice the diameter of the stirring tool shoulder, to ensure that the welding area contains complete temperature and plastic strain field distribution characteristics. The lateral width of the model covers the weld center area and the heat-affected zone, and the thickness direction is consistent with the actual structure. Since friction stir welding exhibits approximately quasi-steady-state characteristics along the weld direction during the stable welding stage, taking a representative length segment can effectively reduce the model size while ensuring calculation accuracy.
[0037] Step S1-2: Mesh generation and material property setting. Locally refine the mesh in the weld center region, the area affected by the stirring tool shoulder, and the heat-affected zone. The feature size of the elements in the refined region should be controlled within, for example, approximately 0.2–0.5 mm, to meet the requirements for temperature and strain gradient resolution. Gradually increase the mesh size away from the weld region to control the total number of elements. Input the thermophysical and mechanical parameters of the workpiece material, including density, thermal conductivity, specific heat capacity, elastic modulus, Poisson's ratio, yield strength, and coefficient of thermal expansion. These parameters are preferably set as functions of temperature to reflect the evolution of the physical and mechanical properties of the aluminum alloy material from room temperature to the solidus temperature range during the welding thermal cycle.
[0038] Steps S1-3: Establish the CEL model. The stirring tool is defined as a Lagrangian body, described using a rigid body model, and given the actual tool geometry (including the stirring shoulder and stirring pin) and preset rotational and translational motion parameters. The workpiece material region is defined as an Eulerian body, using a deformable material model, allowing material flow within a fixed Eulerian mesh. The Lagrangian-Eulerian hybrid description of the CEL method allows for precise application of contact and friction boundary conditions at the tool-material interface to the tool's geometric surface. Simultaneously, material flow within the Eulerian domain is not limited by mesh deformation, avoiding the computational divergence problem caused by mesh distortion in traditional Lagrangian finite element methods when handling large deformation flow of friction stir welding materials.
[0039] Steps S1-4: Set contact relationships and thermal boundary conditions. Define the contact relationship between the tool's outer surface and the workpiece material's Eulerian domain. Use a Coulomb friction model or a shear friction model to describe the interfacial tangential stress. The friction coefficient can be determined through calibration experiments. The heat input during welding comes from two parts: frictional heat generated at the tool-material interface and heat dissipation from the material's plastic deformation. Both are naturally generated by the material constitutive model and contact model during the CEL thermo-mechanical coupling solution process, without the need to preset the spatial distribution function of the volume heat source or surface heat source. Set convective heat transfer boundary conditions and radiative heat transfer boundary conditions on the upper and side surfaces of the workpiece. The convective heat transfer coefficient and radiative emissivity are set according to environmental conditions. Set contact thermal conduction boundary conditions in the contact area between the lower surface of the workpiece and the back plate. The contact thermal conduction is determined based on the contact pressure and surface roughness. The above thermal boundary conditions together determine the heat dissipation rate of the workpiece to the surrounding environment, directly affecting the cooling rate of the weld zone and the final residual plastic strain.
[0040] Steps S1-5: Perform thermo-mechanical coupling calculations. Use an explicit or implicit time integration scheme to perform transient thermo-mechanical coupling solutions on the CEL model. The calculation time should cover the complete thermal cycle from the start of welding to the weld zone cooling to room temperature. The solution output includes: temperature time history curves for each node in the weld zone, temperature field distribution of a typical cross-section (perpendicular to the welding direction), material flow velocity field, stress field, and plastic strain field. By comparing the calculated peak temperature and temperature time history curves with measured temperature data under the same process conditions, adjust model parameters such as the friction coefficient and contact thermal conductivity to ensure the calculated temperature matches the measured temperature within a preset error range, thus completing model calibration. The preset error range can be, for example, ±5%.
[0041] Steps S1-6: Post-processing the calculation results. Based on the plastic strain field calculated in step S1-5, spatial averaging is performed within a preset extraction area of the weld center region and the heat-affected zone. The extraction area covers the weld core region corresponding to the diameter of the stirring pin and the heat-affected zone extending to both sides along the transverse width of the weld, and covers the entire thickness range of the cover plate and the bottom plate along the thickness direction. Within the extraction area, the longitudinal plastic strain component along the weld direction and the transverse plastic strain component perpendicular to the weld direction are respectively weighted by volume or area to obtain the equivalent inherent strain including longitudinal and transverse components under this set of process parameters. The physical basis for spatial averaging is that the weld path element size of the macroscopic model is usually much larger than the local plastic strain gradient scale. The response of the macroscopic structure to the weld plastic strain mainly depends on the total amount of plastic strain in this region rather than its precise distribution. Therefore, regional averaging can achieve reasonable model reduction while ensuring the accuracy of macroscopic deformation calculation.
[0042] Figure 3The equivalent stress (Von Mises stress) contour plot and displacement contour plot calculated from the small-scale model in this embodiment are shown, which are used to illustrate the equivalent stress distribution and displacement distribution in the small-scale model, respectively. Von Mises stress is the equivalent stress calculated based on the stress tensor.
[0043] Steps S1-7: Repeat the calculation by changing welding process parameters. Within the preset process parameter window, change parameters such as welding rotation speed, welding speed, and axial force, and repeat steps S1-1 to S1-6 to obtain the equivalent inherent strain corresponding to multiple combinations of process parameters, establishing a mapping relationship database between process parameters and equivalent inherent strain. The mapping relationship can be stored and retrieved using multidimensional lookup tables, linear or nonlinear interpolation, or regression approximation methods based on response surfaces or artificial neural networks. This mapping relationship allows for the direct acquisition of the corresponding benchmark equivalent inherent strain from any set of process parameters in subsequent mesoscale correction and macroscopic optimization analyses, eliminating the need for repeated small-scale CEL calculations.
[0044] The modeling and calculation method of the mesoscale welding path segmentation correction model described in step S2 is as follows: Figure 1 As shown, the specific steps include: Step S2-1: Segment the weld path according to its overall characteristics. Divide the long cold-rolled plate weld into multiple segments based on its geometric features and the changing characteristics of its thermo-mechanical boundary conditions. Segment types include straight segments, corner segments, starting segments, ending segments, and closed segments. The segment boundaries are determined based on: tooling partition locations (corresponding to changes in clamping constraints and heat dissipation conditions), locations of abrupt changes in structural stiffness (such as the junction of flow channel stiffeners and frame connections), geometric turning points in the weld path (corner start and end points), and locations sensitive to heat accumulation (the middle and rear parts of long straight segments, and the intersection of closed and starting segments). A reasonable segmentation strategy ensures that the thermal cycle characteristics and constraint conditions within each segment are approximately consistent, while showing significant differences between segments, providing a physical basis for subsequent segmentation corrections.
[0045] Step S2-2: Conduct medium-scale welding experiments. Select medium-scale specimens that are consistent with or representative of the actual cold-rolled plate structure material specifications, thickness combination, and joint type. The specimen size should be able to accommodate all typical segments of the complete weld path. Select experimental welding parameters according to the typical process parameter range calibrated in Step S1, and conduct friction stir welding experiments. Arrange thermocouple temperature measuring points at representative positions of each path segment. The measuring points are located in the vicinity of the weld centerline and typical positions of the heat-affected zone. Thermocouples can be fixed by pre-drilling or surface mounting. The measuring point arrangement covers the middle of the straight section (reflecting the steady-state welding thermal cycle characteristics), the corner transition section (reflecting the influence of welding direction change on heat input and heat dissipation), the vicinity of the start and stop points (reflecting the transient temperature rise at the start of welding and the continuous heating effect after termination), and the end area of the closed section (reflecting the repeated heating effect at the closure junction).
[0046] Step S2-3: Collect and process temperature history data. Record the temperature time history curves of each thermocouple measuring point throughout the welding process and the post-weld cooling stage using a data acquisition system at a sampling frequency preferably not less than 10Hz (e.g., approximately 50Hz). Align the raw temperature data with a time reference, ensuring that the time coordinates of each measuring point are unified to a relative time with the welding start time as zero. Identify and remove abnormal data points caused by thermocouple detachment or electromagnetic interference. Segment and stitch the temperature curves of each measuring point according to the welding path. Extract the characteristic temperature parameters of each segment from the processed temperature time history curves: peak temperature (reflecting the highest heating temperature of that segment), heating rate (the average rate at which the temperature rises from room temperature or initial temperature to the peak temperature), cooling rate (the average rate at which the temperature drops from the peak temperature to a specified temperature), and the increase in temperature between the end of the previous segment and the initial temperature of the next segment.
[0047] Figure 4 The equivalent stress (Von Mises stress) contour plot and displacement contour plot calculated by the mesoscale model are shown, which are used to illustrate the equivalent stress distribution and displacement distribution in the mesoscale model, respectively.
[0048] Step S2-4: Quantifying the path heat accumulation effect. Using the termination temperature of the previous segment extracted in Step S2-3 as the initial temperature of the next segment, a quantitative relationship is established between the initial temperature of each segment and the welding sequence. The physical mechanism of the heat accumulation effect is as follows: after the welding of the previous segment, the weld zone and adjacent areas are still at a relatively high temperature, and the heat has not been fully conducted or dissipated. At the start of the welding of the next segment, the initial temperature of the workpiece is higher than room temperature. This results in a higher peak temperature of the thermal cycle experienced by the subsequent weld segment, a longer high-temperature residence time, and a lower cooling rate, leading to a difference in its irreversible plastic strain compared to the baseline value under isolated welding conditions.
[0049] Step S2-5: Correcting the Segmented Inherent Strain. Based on the mapping relationship between process parameters and equivalent inherent strain established in Step S1, the reference inherent strain is corrected according to the measured thermal cycle characteristics of each segment. The correction principle is as follows: for segments with a large initial temperature rise and a reduced cooling rate, the actual plastic strain is usually greater than the reference value, and the inherent strain amplitude is adjusted upwards accordingly; for segments where the cooling rate increases due to forced cooling by tooling or backplate, the inherent strain amplitude is adjusted downwards accordingly. The correction amount is expressed in the form of a segmented inherent strain factor, i.e., the ratio of the inherent strain of each segment to the reference inherent strain. This segmented inherent strain factor is directly used as the input parameter for the subsequent large-scale macroscopic model.
[0050] Step S2-6: Verify the applicability of the modified model. Repeat steps S2-2 to S2-5 for comparative experiments by changing the welding path sequence or start / stop point positions. Compare the temperature history, heat accumulation degree, and corrected inherent strain factor of each segment under different welding sequences. Analyze the impact of welding sequence changes on the stability of the correction factor, and determine the applicable range and correction amplitude limits of the inherent strain factor for each typical segment. This verification step ensures that the mesoscale modified model has the ability to generalize under different welding path planning conditions.
[0051] The modeling and calculation method of the large-scale macroscopic structure prediction model described in step S3 is as follows: Figure 1 As shown, the specific steps include: Step S3-1: Establish a three-dimensional finite element model of the integrated macroscopic structure of the cold plate and base plate. Based on the actual product design geometric model, establish an overall finite element analysis model. The geometric model should retain structural features that significantly affect the overall stiffness and heat conduction path of the structure, including the internal flow channel stiffener structure, frame structure, thickness distribution of the cover plate and base plate, and the actual direction of the weld path. Minor geometric features that do not affect stiffness and heat conduction (such as chamfers in non-load-bearing areas and small holes) can be appropriately simplified to reduce the difficulty of mesh generation and computational scale. The model coordinate system is aligned with the welding direction, and the weld path is identified as an independent geometric path or element set in the model to facilitate the subsequent application of the inherent strain factor in sections.
[0052] Step S3-2: Input material properties and mesh. Input the corresponding thermophysical and room-temperature mechanical parameters, including elastic modulus, Poisson's ratio, yield strength, and coefficient of thermal expansion, based on the material grades used for the cold plate and base plate. Refine the mesh along the weld path and its adjacent areas. The size of the elements in the refined area along the weld direction should be adapted to the length of each segment, and the lateral width should cover the weld and heat-affected zone. A coarser mesh can be used away from the weld, with the mesh size gradually increasing. The mesh type can be a hexahedral structured mesh or a tetrahedral unstructured mesh. Mapped or swept meshes are recommended for the weld path area to ensure the accuracy of the loading direction.
[0053] Step S3-3: Apply clamping constraints and thermal boundary conditions. Based on the actual welding fixture arrangement and clamping method, apply displacement constraints to the corresponding nodes in the macroscopic model. Clamping point constraints typically restrict the translational degrees of freedom of the clamped area along the normal direction of the plate surface; in some cases, in-plane translational degrees of freedom also need to be restricted to prevent workpiece slippage. Simultaneously, set contact thermal conductivity boundary conditions between the lower surface of the workpiece and the back plate. The contact thermal conductivity value should be consistent with the value used in the small-scale model in Step S1-4 or back-calibrated through mesoscale experiments. Thermal boundary conditions also include convective and radiative heat transfer on the upper and side surfaces of the workpiece, which can be applied during the thermal analysis in the welding stage.
[0054] Step S3-4: Apply inherent strain factors segmented according to the weld path. The segmented inherent strain factors obtained in step S2 are applied to the macroscopic model according to the geometric regions of each segment of the weld path. The application method is as follows: anisotropic initial strain is introduced into the initial strain field of the weld path element or weld neighborhood element. The longitudinal component is obtained by multiplying the segmented inherent strain factor by the longitudinal component of the baseline equivalent inherent strain under the corresponding process parameters; the transverse component is handled similarly. For initial strain discontinuities at segment boundaries caused by spatial variations in the segmented factor, a linear gradient transition zone can be set in the boundary transition region to address this.
[0055] Step S3-5: Perform mechanical analysis of the welding stage. Under the conditions of clamping displacement constraints and initial strain loads, perform static analysis on the macroscopic structural model. Solve for the structural response caused by the inherent strain of the weld under the clamping constraint state. The output includes: the displacement of each node (i.e., the confined displacement), the stress tensor of each element (i.e., the residual stress distribution during the welding stage), and the support reaction force at each clamping point. The welding stage analysis uses a linear elastic or elastoplastic constitutive model. If an elastoplastic model is used, the material yield strength should be taken as the room temperature value, because the workpiece has cooled to room temperature after the welding stage.
[0056] Figure 6 The residual stress distribution cloud map of the cold plate structure during the welding stage is shown. As can be seen from the figure, there is significant residual stress concentration in the weld area, with the maximum residual stress located in the middle region of the weld. The local stress increase occurs at the clamping point due to the concentration of constraints.
[0057] Step S3-6: Simulate the unclamping process. Based on the analysis completed in the welding stage of Step S3-5, the clamping point displacement constraints are removed in stages according to a preset sequence corresponding to the actual unclamping operation sequence. The constraint release sequence can be selected according to the actual process, proceeding step by step from the structural edge to the center, from the symmetrical position to the center, or from the corner area to the straight segment. After removing part of the constraints in each stage, the static solution is re-performed to update the structural displacement field and stress field; the next stage uses the displacement and stress state of the previous stage as the initial state, and continues to remove the remaining constraints and solve. This iteration continues until all clamping constraints are released. The principle of staged simulation is that the actual unclamping operation is not completed instantaneously; there is a time interval between the release of each constraint. When subsequent constraints are released, the structure has already completed some springback and stress redistribution in the previous release step. Staged solution can more realistically reflect this gradual relaxation process.
[0058] Step S3-7: Output the post-weld free-state deformation results. After all clamping constraints are completely released, the structure reaches a free equilibrium state unconstrained by external forces. The following post-weld deformation evaluation indicators are extracted from the displacement field results of this state: maximum overall out-of-plane displacement (reflecting the overall warping degree of the structure), local waviness in the weld-adjacent area (extracting the fluctuation amplitude of out-of-plane displacement along the weld path), height fluctuation deviation in the sealing area (specifying the peak-to-valley difference of the displacement at the sealing mating surface nodes), and diagonal twist of the structure. These indicators quantitatively characterize the impact of post-weld deformation on product quality from three levels: overall morphology, local flatness, and functional surface fit accuracy.
[0059] Figure 7 The deformation contour plots of two structural schemes in their free state after welding are shown. (a) represents the scheme without crossbeams and rivets, and (b) represents the scheme with crossbeams and rivets. The two plots respectively show the spatial distribution of deformation for each scheme; specific values are read according to the color codes below each plot.
[0060] Example 1 Taking the integrated friction stir welding structure of aluminum alloy cold plate and base plate of a new energy vehicle battery pack as an example, the specific implementation process of the present invention is explained.
[0061] In this embodiment, the cold plate cover is made of 6061-T6 aluminum alloy with a thickness of 2.0 mm; the base plate is also made of 6061-T6 aluminum alloy with a thickness of 3.0 mm. The cold plate contains a flow channel rib structure. The cover and base plates are connected along a rectangular closed path by friction stir welding. The total weld length is approximately 1500 mm, comprising four straight segments and four 90° corner segments. The welding start and end points are located on the same straight segment, forming a closed weld. The welding process parameters are as follows: stirring tool rotation speed 800–1200 r / min, welding speed 80–180 mm / min, and axial force 4–10 kN. The diameter of the stirring tool shoulder is, for example, 12 mm, and the diameter of the stirring pin is, for example, 4 mm, with the pin length slightly less than the total thickness of the cover and base plates.
[0062] This embodiment is implemented in the following three stages: Phase 1: Calculation of CEL model and calibration of inherent strain in small-scale local weld areas.
[0063] Step A1: Establish a 3D geometric model of the local weld zone. The model is 60mm long along the welding direction (approximately 5 times the diameter of the stirring tool shoulder), 40mm wide, and 5mm thick (2mm for the cover plate and 3mm for the base plate). The weld center is located in the middle of the model's transverse direction. This model size ensures that the weld center area and the heat-affected zone are within the model's range, and that it has sufficient length along the welding direction to eliminate the influence of boundary effects on the stable region in the center of the weld.
[0064] Step A2: Create a 3D geometric model of the mixing tool and set up the CEL model. The mixing tool is modeled to actual dimensions, including the cylindrical surface of the shoulder, the cylindrical surface of the mixing pin, and end face features. The tool is defined as a Lagrange rigid body, and the workpiece material region is defined as an Eulerian body. The dimensions of the Eulerian domain are consistent with the geometric model in Step A1, and the material can flow freely within the Eulerian domain.
[0065] Step A3: Input material properties and mesh. Input the material parameters of 6061-T6 aluminum alloy according to the temperature variation curve, covering a temperature range of 25℃ to 580℃ (close to the solidus temperature). The mesh feature size for the weld center region and shoulder action region is, for example, approximately 0.3 mm; for the heat-affected zone, approximately 0.5 mm; gradually transitioning to, for example, approximately 2 mm further away from the weld. Set convective heat transfer boundary conditions on the upper and side surfaces of the workpiece, with a convective heat transfer coefficient of, for example, approximately 15 W / (m²). 2 • K), with an emissivity of approximately 0.3; a contact thermal conductivity boundary condition is set in the contact area between the lower surface of the workpiece and the back plate, with a contact thermal conductivity of approximately 2000 W / (m²). 2 ·K).
[0066] Step A4: Set process parameters and perform CEL thermo-mechanical coupling analysis. Select the following baseline process parameter combination: rotation speed 1000 r / min, welding speed 120 mm / min, axial force 6 kN. The calculation time step is automatically determined based on the element size and material wave velocity, and the total calculation time covers the heating stage of the welding process and the cooling to room temperature stage after welding. Solve for the temperature field distribution, temperature time history curve, material flow velocity field, stress field, and plastic strain field of the weld zone. Compare the calculated peak temperature at the weld center with the measured temperature of the thermocouple under the same process conditions. By fine-tuning the friction coefficient, the deviation between the calculated and measured peak temperatures is controlled within, for example, ±5%, thus completing the model calibration.
[0067] Step A5: Extract equivalent inherent strain. Within the weld center area and heat-affected zone (approximately 8mm wide laterally, covering the diameter of the stirring pin and the shoulder action area), covering the entire plate thickness along the thickness direction, the longitudinal and transverse plastic strain components are respectively volume-weighted averaged. For the baseline process parameter combination, the longitudinal and transverse equivalent inherent strains are obtained.
[0068] Step A6: Establish a process parameter mapping database. Within the range of process parameters, select multiple horizontal combinations (rotation speed of 800, 1000, 1200 r / min; welding speed of 80, 120, 180 mm / min; axial force of 4, 6, 8, 10 kN), and perform the calculations from steps A1 to A5 one by one to obtain the equivalent inherent strain corresponding to each set of parameters. Establish a three-dimensional process parameter space mapping database that supports linear interpolation calls.
[0069] Phase 2: Establishment and experimental calibration of the segmented correction model for the medium-scale welding path.
[0070] Step B1: Weld Path Segmentation. The rectangular closed weld path is divided into four segments: the starting segment (from the welding start point to the first corner), the straight segment (the middle area of the four sides), the corner segment (the four 90° corners and the 20mm transition zones before and after them), and the closing termination segment (from the last corner to the welding end point on the last side). The total plan is divided into 14 segments (4 straight segments + 4 corner segments + 1 starting segment + 1 termination segment + 4 supplementary transition segments). The physical basis for segmentation is as follows: the straight segment corresponds to steady-state welding conditions, the corner segment involves tool dwell and heat input accumulation caused by changes in welding direction, and the starting and termination segments correspond to transient heating and repeated heating during weld closure, respectively.
[0071] Step B2: Prepare a medium-scale specimen and conduct welding experiments. Prepare a medium-scale butt joint specimen of the same material and thickness as the actual cold-rolled plate, with specimen dimensions of 300mm × 200mm. The weld seam runs along the length direction, and the path includes straight segments and corner segments corresponding to the complete product. Temperature measuring holes with a diameter of 1.0mm are made at representative positions on each segment of the specimen, and K-type armored thermocouples are embedded therein. The thermocouple measuring ends are approximately 2mm from the weld centerline. Friction stir welding experiments are conducted according to the baseline process parameters (rotation speed 1000r / min, welding speed 120mm / min, axial force 6kN). The data acquisition system records the temperature data throughout the welding process and after 300s of cooling at a frequency of approximately 50Hz.
[0072] Step B3: Temperature Data Processing. Time-align the temperature time-history data at each measuring point, using the welding start time as the zero point; manually check and remove abnormal data points caused by electromagnetic interference; establish a mapping between the temperature curve and the segment number according to the segment correspondence. Extract the characteristic parameters of each segment: peak temperature, average rate of heating from 100℃ to the peak temperature, average rate of cooling from the peak temperature to 200℃, and the initial temperature of the segment (taken as the measured temperature of the end point of the segment before the welding enters this segment).
[0073] Step B4: Quantify the heat accumulation effect. Analyze the initial temperature change trend of each segment. Data shows that as the welding sequence progresses, the initial temperature of subsequent segments gradually increases. The initial temperature of the starting segment is room temperature (approximately 25°C). After the first corner segment, the initial temperature of the subsequent straight segments has risen to approximately 80°C. After several segments, the initial temperature of the closing and terminating segment can reach over approximately 120°C. The corresponding cooling rate gradually decreases from approximately 45°C / s in the starting segment to approximately 30°C / s in the closing and terminating segment.
[0074] Step B5: Correct the segmented inherent strain factor. Using the small-scale calibration reference inherent strain corresponding to the initial segment (initial temperature 25℃) as a reference value of 1.0, determine the correction factor for each subsequent segment based on the differences in its initial temperature rise and cooling rate. The corrected segmented inherent strain factor is approximately 1.05–1.10 for the linear steady-state segment, approximately 1.12–1.18 for the corner segment, and approximately 1.15–1.22 for the closed termination segment. This correction factor reflects the increased plastic strain effect caused by thermal accumulation.
[0075] Step B6: Validate the modified model. Repeat the above experiment by changing the welding path order (using sequential welding instead of the original order) and compare the differences in the segmented correction factors between the two sets of experiments. The results show that the change in welding order mainly affects the correction factors of the closed termination segment and the adjacent start segment, while the fluctuations of other segmented correction factors are within, for example, ±3%, verifying the applicability of the modified model under different path orders.
[0076] Phase 3: Large-scale macroscopic structural modeling and post-weld deformation prediction and optimization.
[0077] Step C1: Establish a finite element model of the integrated macroscopic structure of the cold plate and base plate. Based on the actual product's 3D geometric model, establish a finite element mesh model, retaining key structural features such as the flow channel stiffeners and frame. The model uses a hybrid shell element and solid element approach; the cold plate cover and base plate use shell elements, while the flow channel stiffeners and frame use solid elements. The weld path region is divided into separate element groups for subsequent partitioned loading. The mesh size is, for example, approximately 120,000 nodes, with a mesh size of approximately 5mm in the weld path region and approximately 15mm away from the weld.
[0078] Step C2: Input material properties. At room temperature, the elastic modulus of 6061-T6 aluminum alloy is taken as 69 GPa, Poisson's ratio as 0.33, yield strength as 276 MPa, and coefficient of thermal expansion as 23.6 × 10⁻⁶. -6 / K.
[0079] Step C3: Apply clamping constraints and thermal boundaries. Based on the actual tooling design, 16 clamping points are set around the perimeter of the cold plate and in the middle of its long side. These clamping points restrict the translational displacement of the node along the normal direction of the plate surface, while allowing free movement in the in-plane direction. A contact thermally conductive boundary is set on the lower surface of the base plate to simulate the thermal conduction effect of the back plate.
[0080] Step C4: Apply inherent strain factors in segments according to the weld path. The 14 segmented inherent strain factors obtained in the second stage are applied to the corresponding element groups of the weld path in the macroscopic model, and each element introduces longitudinal and transverse inherent strain components in the form of initial strain.
[0081] Step C5: Mechanical Analysis of the Welding Stage. Static solutions are performed under the conditions of activated clamping point displacement constraints and the application of a thermally conductive boundary in contact with the base plate to obtain the residual stress distribution and confined displacement field in the clamped state. The calculation results show that significant compressive stress is generated in the weld zone under clamping point constraints, with the maximum compressive stress amplitude in the middle region of the weld. Local stress increases occur at the clamping point due to constraint concentration.
[0082] Step C6: Simulate unclamping by releasing clamping constraints in stages. Release the 16 clamping point constraints in four steps, proceeding from the edge to the center. The first step releases the four corner clamping points; the second step releases the outer clamping points on the long side; the third step releases the middle clamping point on the long side; and the fourth step releases all remaining constraints. After each release, resolve the static equilibrium, outputting the displacement and stress fields. The calculation results show that as the constraints are gradually released, the out-of-plane displacement of the structure gradually increases, and the residual stress gradually redistributes and tends to become uniform.
[0083] Step C7: Output post-weld free-state deformation. After all constraints are released, extract the deformation indices: the overall maximum out-of-plane displacement is, for example, 1.52 mm (located in the center region of the cold plate), the height fluctuation deviation of the sealing area is, for example, 0.38 mm, and the waviness amplitude of the area adjacent to the weld is, for example, 0.24 mm. These indices provide a quantitative evaluation benchmark for subsequent process optimization.
[0084] Step C8: Establishing Approximate Relationships and Optimization. Using the response surface methodology, a surrogate model is established with the piecewise inherent strain factor as input variables and the overall maximum out-of-plane displacement and the height fluctuation deviation of the sealed area as output responses. The fitting accuracy of the surrogate model is, for example, R0. 2 >0.95, meeting engineering application requirements. Based on the surrogate model, with deformation minimization as the objective and weld tightness (guaranteed by the small-scale model criterion) as the constraint, the welding rotation speed, welding speed, axial force, welding path sequence, start and stop point positions, clamping point distribution, and backplate heat conduction scheme are synergistically optimized. The optimized scheme reduces the overall maximum out-of-plane displacement to, for example, approximately 0.84 mm, a reduction of approximately 45%, and the height fluctuation deviation of the sealing area to, for example, approximately 0.19 mm, meeting the product assembly accuracy requirements.
[0085] Example 2 Taking another type of cold-plate structure with a combination of non-closed long straight lines and local corner welds as an example, the thickness of the cold-plate cover plate is 2.0 mm, the thickness of the bottom plate is 2.5 mm, and the weld path consists of a starting straight section (200 mm long), a middle straight section (400 mm long), a corner transition section (90° corner), and a termination section (150 mm long), which is a non-closed path.
[0086] The implementation steps of this embodiment are basically the same as those of Embodiment 1, except that the path segmentation type includes a starting segment, a long straight stable segment, a corner transition segment, and a termination segment, totaling four segments. Thermocouple measuring points for the mesoscale experiment are primarily arranged at the rear of the long straight segment (150mm from the starting point, used to detect the heat accumulation effect during long-distance welding) and the corner transition segment (used to detect thermal cycling changes when the welding direction changes). The loading region of the inherent strain factor in the large-scale macroscopic model is re-divided according to the actual path geometry, and the segmentation factor correction amplitude is calibrated with reference to the data range determined in Embodiment 1 and combined with the measured temperature data of this embodiment.
[0087] Using the method of this invention, the maximum out-of-plane displacement of the entire weld in its free state is, for example, 1.18 mm, and the waviness of the weld-adjacent area is, for example, 0.21 mm. By optimizing the welding path sequence (changing the original unidirectional welding from one end to the other to symmetrical welding from both ends towards the middle) and adjusting the distribution of clamping points, the deformation index is further reduced. This embodiment verifies the adaptability of the method of this invention to non-closed long weld paths, as well as the effectiveness of deformation prediction and process optimization under different weld geometries.
[0088] The above description is merely a preferred embodiment of the present invention and does not limit the patent scope of the present invention. Any equivalent structural or procedural modifications made based on the content of the present invention's specification and drawings within the spirit and principles of the present invention, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of the present invention. The scope of protection of the present invention is determined by the appended claims.
Claims
1. A multi-scale calculation and prediction method for post-weld deformation of cold-rolled steel plate by friction stir welding, characterized in that, Includes the following steps: Step 1: Establish a small-scale calibration calculation model for the local weld zone, and use the coupled Euler-Lagrange method to numerically simulate the friction stir welding process to obtain the welding thermal cycle history. Step 2: Based on the temperature and stress-strain fields calculated by the small-scale model, extract the irreversible plastic strain of the weld center and heat-affected zone, obtain the equivalent inherent strain through regional averaging, and repeat the calculation by changing process parameters to establish the mapping relationship between process parameters and equivalent inherent strain. Step 3: Establish a medium-scale welding path segmentation correction model, dividing the weld path into multiple segments according to characteristics and using the termination temperature state of the previous segment as the initial temperature state of the next segment; conduct medium-scale welding experiments to obtain the temperature time history curves of representative locations in each segment. The process involves: 1) Quantifying the heat accumulation effect; 2) Correcting the inherent strain of each segment based on the mapping relationship to form a segmented inherent strain factor; 3) Establishing a large-scale macroscopic structural inherent strain prediction model; 4) Applying the segmented inherent strain factor in the weld path region segment by segment; 5) Applying clamping constraints and thermal boundary conditions consistent with the actual tooling to perform mechanical analysis of the welding stage; 6) Releasing constraints step by step in a preset order to simulate the unclamping process and obtain the deformation results in the free state after welding; 7) Based on the results of step four, establishing an approximate relationship between the segmented inherent strain factor and key deformation indicators to achieve rapid prediction of macroscopic deformation under multi-parameter conditions in different zones; and 8) Co-optimizing welding parameters, path sequence, start and stop points, and clamping and thermal conductivity boundaries.
2. The method according to claim 1, characterized in that, In step one, the coupled Eulerian-Lagrange method defines the stirring tool as a Lagrange body and the welding material region as an Eulerian body. The mesh is refined in the weld center, heat-affected zone, and shoulder action area. Boundary conditions for backplate contact heat conduction, surface convection, and radiation heat are set. The temperature field, stress-strain field, and material flow characteristics are obtained through transient thermo-mechanical coupling calculation.
3. The method according to claim 1, characterized in that, The method for extracting the equivalent inherent strain in step two is as follows: the longitudinal and transverse irrecoverable plastic strains are averaged along the transverse width and thickness of the weld within a specified extraction area to obtain the equivalent inherent strain containing longitudinal and transverse components; the method for establishing the mapping relationship includes table lookup, interpolation, or regression approximation.
4. The method according to claim 1, characterized in that, The segmentation of the weld path in step three is based on the tooling partition location, the location of sudden change in structural stiffness, the flow channel turning point, or the location of heat accumulation sensitivity. The segmentation types include straight segments, corner segments, start-stop segments, and closed segments.
5. The method according to claim 1, characterized in that, In step three, the measuring points for the mesoscale welding experiment are arranged near the weld centerline and in typical locations of the heat-affected zone, covering the middle of the straight section, the corner transition section, the neighborhood of the start and stop points, and the end of the closed section. The peak temperature, heating rate, cooling rate, and initial temperature rise between segments are extracted from the measured temperature time history curve to quantify the heat accumulation effect.
6. The method according to claim 1, characterized in that, Step three also includes: conducting comparative experiments and calculations by changing the path sequence or the start and stop points to verify the applicability of the mapping relationship under different welding sequences, and determining the range of segmented inherent strain correction amplitude.
7. The method according to claim 1, characterized in that, In step four, the large-scale macroscopic structural model retains the internal flow channels and frame structure features that have a significant impact on the overall stiffness and heat conduction path, and performs mesh refinement on the weld path and its neighborhood; the thermal boundary conditions include the contact thermal conduction boundary between the workpiece and the back plate.
8. The method according to claim 1, characterized in that, In step four, the release sequence of the step-by-step constraint release simulation unclamping process is from the edge to the middle, from the symmetrical point to the center, or from the corner to the straight line segment. The stress redistribution and structural springback after the constraint release at each stage are solved.
9. The method according to claim 1, characterized in that, The key deformation indicators in step five include the overall maximum out-of-plane displacement, the height fluctuation deviation of the sealing area, and the waviness of the area adjacent to the weld. The collaborative optimization optimizes the design of rotational speed, welding speed, axial force, path sequence, start and stop point positions, clamping point distribution, and back plate heat conduction capacity under the constraint that the weld quality meets the requirements.
10. The method according to any one of claims 1 to 9, characterized in that, The cold plate includes a cover plate, a flow channel plate and a frame, and is integrally connected to the base plate by friction stir welding. Its weld path is a long weld path, including corners, start and stop points or closed loops.
