Final Report Summary - DEEPWELD (Detailed multi-physics modelling of friction stir welding)
The ultimate aim of the DEEPWELD project was the development of a new multi-physics and multi-scale numerical software tool for the accurate modelling of the Friction stir welding (FSW) process. This tool would help shorten the design cycle and decrease cost by reducing the number of experimental prototypes by replacing them by virtual prototypes. The new tool would be a large step forward compared to current solutions, as it would be equipped with a thermo-fluid module in order to simulate the important material flow around it and an advanced metallurgy model in order to predict the evolution of microstructures. Specifically instrumented experiments would be conducted in order to define accurate thermally varying friction laws, material constitutive laws and data in order to validate the new numerical tool.
The software simulation introduced was based on a multi-scale approach in which a new advanced finite element sover, based on a velocity-pressure formulation, solves the material flow and thermal effects around the tool taking into account complex thermally varying friction laws. The material flow solver at the lowest scale was coupled with a state-of-the-art industrial software to compute the complete process from the starting transitory phase to a steady phase and eventually to the final phase of the process. New metallurgy models were implemented in the industrial code in order to take into account the changes in micro-structure due to the stirring and cooling of the metal. Attention was paid to the applicability of the new technique to an industrial basis.
The project introduced the following eight innovations:
Multi-scale multi-physics solution based on industrial software
DEEPWELD did not aim to re-invent existing, proven, industrial finite element solvers, but to complement them with sufficiently advanced features in order to simulate specifically a process such as friction stir welding. The objective was to perform analysis at different scales. First, a thermo-fluid simulation was performed at the scale of the flow material around the tool, advanced friction models and metallurgy were implemented in this new solver. Information such as thermal fields, strain, strain rates of the material flow region were then passed onto the industrial thermo-mechanical solver in order to compute a global scale analysis for the whole process. The thermo-mechanical solver was equipped with metallurgy modules capable of accurately predicting the evolution of microstructures during the process. Coupling between local scale thermo-fluid analysis and global process thermo-mechanical modelling was achieved.
Material flow visulatisation
Experiments were conducted to visualise material flow. Those data were used to validate the thermo-fluid finite element tool and to calibrate the analytical flow model, as well. Techniques used to visualise the material flow laid on the introduction of elements such as very small balls, very thin wires or sheets that are used as tracers of the material flow. Also, an innovative method was proposed which consisted of using a number of tracers with known low-melting temperatures, e.g. solder alloys to be selected on the basis of their melting temperature. By using such tracers, it was unfortunately not possible to obtain information neither on the tracer particles displacements during FSW nor on the maximum temperatures attained during their displacement, in particular in the nugget where thermocouples and thermographic measurements are impossible. Therefore, this method was demonstrated not efficient during the project.
No experimentally calibrated heat flux
DEEPWELD focused on eliminating equivalent heat flux used in current state-of-the-art thermo-mechanical simulation of FSW process. Indeed, in the latter, an equivalent heat flux should be determined on a case by case basis from experiments measuring heat input and tool loads and tuned to obtain good correlations between simulation and measurements. Hence, this methodology was a major burden for any optimisation of welding parameters or application to different alloys and tools. The objective was to replace it by a thermo-fluid calculation to predict the amount of heat generated through plastic work and friction as described before. Therefore, the thermo-fluid module was enhanced by the implementation of a complex friction law.
Understanding friction effects
New friction laws depending on temperature and the mechanical status of the material were developed. Those served as input to the new thermo-fluid simulation tool. Instrumented experiments were conducted to determine the parameters of the laws by inverse analysis and they mainly focused on direct measurement of the local friction (shear) interface stress as a function of the temperature and pressure, as well as on the measurement of forces and torques on FSW equipment in order to obtain global information as a function of operating welding parameters (rotational speed, tilt angles, advancement speed).
Further develop micro-structure thermal evolution
The accurate characterisation of the evolution of the microstructures of the different zones of the weld during the process was very important to predict the mechanical properties of the resulting welded parts. In particular, in the nugget zone when a high strain rate was directly related to the flow. The thermo-fluid module provided the temperature/strain rate data to the metallurgical module to predict accurately the material behaviour. A global simulation of the microstructural evolutions during the process was performed; grain/sub-grain size, textures, hardening precipitation. Validations were carried out by experimental data and microstructures were investigated by means of the Field emission gun scanning electron microscope (FEG SEM).
Validation based on experiments on force and displacement controlled machines
The consortium conducted experiments on two types of machines, namely force-controlled machine and on displacement-controlled machine. These two controlling techniques were widely used for FSW and it was expected that using them both in the same research programme would be a good opportunity to provide a better understanding of the phenomena occurring at the tool-workpiece interface, as well as of the respective advantages and disadvantages of both control techniques.
Optimisation of welding parameter
The new multi-physics multi-scale solver was developed to minimise the number of experiments required to calibrate the different parameters of the numerical solution. The objective to be able to apply the solver as is to different geometries, plate thickness, etc, provided the material metallurgy module remain within the validity limits. Hence, it was possible to optimise process parameters in order to obtain required microstructures of the welds.
Application to welding of real-life coupons representative of aeronautical applications
The final objective was to validate the predicted solutions of the new solver on coupons representative of aeronautical applications. The solver was able to accurately compute macroscopic properties such as residual stresses, distortions, mechanical properties of the joints. These results were validated by experiments conducted on welded panels. In addition, the mechanical properties of the latter were compared to classical riveted panels, which should in the future allow for the definition of guidelines on the usage of FSW for the design of aeronautical parts.
The software simulation introduced was based on a multi-scale approach in which a new advanced finite element sover, based on a velocity-pressure formulation, solves the material flow and thermal effects around the tool taking into account complex thermally varying friction laws. The material flow solver at the lowest scale was coupled with a state-of-the-art industrial software to compute the complete process from the starting transitory phase to a steady phase and eventually to the final phase of the process. New metallurgy models were implemented in the industrial code in order to take into account the changes in micro-structure due to the stirring and cooling of the metal. Attention was paid to the applicability of the new technique to an industrial basis.
The project introduced the following eight innovations:
Multi-scale multi-physics solution based on industrial software
DEEPWELD did not aim to re-invent existing, proven, industrial finite element solvers, but to complement them with sufficiently advanced features in order to simulate specifically a process such as friction stir welding. The objective was to perform analysis at different scales. First, a thermo-fluid simulation was performed at the scale of the flow material around the tool, advanced friction models and metallurgy were implemented in this new solver. Information such as thermal fields, strain, strain rates of the material flow region were then passed onto the industrial thermo-mechanical solver in order to compute a global scale analysis for the whole process. The thermo-mechanical solver was equipped with metallurgy modules capable of accurately predicting the evolution of microstructures during the process. Coupling between local scale thermo-fluid analysis and global process thermo-mechanical modelling was achieved.
Material flow visulatisation
Experiments were conducted to visualise material flow. Those data were used to validate the thermo-fluid finite element tool and to calibrate the analytical flow model, as well. Techniques used to visualise the material flow laid on the introduction of elements such as very small balls, very thin wires or sheets that are used as tracers of the material flow. Also, an innovative method was proposed which consisted of using a number of tracers with known low-melting temperatures, e.g. solder alloys to be selected on the basis of their melting temperature. By using such tracers, it was unfortunately not possible to obtain information neither on the tracer particles displacements during FSW nor on the maximum temperatures attained during their displacement, in particular in the nugget where thermocouples and thermographic measurements are impossible. Therefore, this method was demonstrated not efficient during the project.
No experimentally calibrated heat flux
DEEPWELD focused on eliminating equivalent heat flux used in current state-of-the-art thermo-mechanical simulation of FSW process. Indeed, in the latter, an equivalent heat flux should be determined on a case by case basis from experiments measuring heat input and tool loads and tuned to obtain good correlations between simulation and measurements. Hence, this methodology was a major burden for any optimisation of welding parameters or application to different alloys and tools. The objective was to replace it by a thermo-fluid calculation to predict the amount of heat generated through plastic work and friction as described before. Therefore, the thermo-fluid module was enhanced by the implementation of a complex friction law.
Understanding friction effects
New friction laws depending on temperature and the mechanical status of the material were developed. Those served as input to the new thermo-fluid simulation tool. Instrumented experiments were conducted to determine the parameters of the laws by inverse analysis and they mainly focused on direct measurement of the local friction (shear) interface stress as a function of the temperature and pressure, as well as on the measurement of forces and torques on FSW equipment in order to obtain global information as a function of operating welding parameters (rotational speed, tilt angles, advancement speed).
Further develop micro-structure thermal evolution
The accurate characterisation of the evolution of the microstructures of the different zones of the weld during the process was very important to predict the mechanical properties of the resulting welded parts. In particular, in the nugget zone when a high strain rate was directly related to the flow. The thermo-fluid module provided the temperature/strain rate data to the metallurgical module to predict accurately the material behaviour. A global simulation of the microstructural evolutions during the process was performed; grain/sub-grain size, textures, hardening precipitation. Validations were carried out by experimental data and microstructures were investigated by means of the Field emission gun scanning electron microscope (FEG SEM).
Validation based on experiments on force and displacement controlled machines
The consortium conducted experiments on two types of machines, namely force-controlled machine and on displacement-controlled machine. These two controlling techniques were widely used for FSW and it was expected that using them both in the same research programme would be a good opportunity to provide a better understanding of the phenomena occurring at the tool-workpiece interface, as well as of the respective advantages and disadvantages of both control techniques.
Optimisation of welding parameter
The new multi-physics multi-scale solver was developed to minimise the number of experiments required to calibrate the different parameters of the numerical solution. The objective to be able to apply the solver as is to different geometries, plate thickness, etc, provided the material metallurgy module remain within the validity limits. Hence, it was possible to optimise process parameters in order to obtain required microstructures of the welds.
Application to welding of real-life coupons representative of aeronautical applications
The final objective was to validate the predicted solutions of the new solver on coupons representative of aeronautical applications. The solver was able to accurately compute macroscopic properties such as residual stresses, distortions, mechanical properties of the joints. These results were validated by experiments conducted on welded panels. In addition, the mechanical properties of the latter were compared to classical riveted panels, which should in the future allow for the definition of guidelines on the usage of FSW for the design of aeronautical parts.
