Real-Time Data Assimilation in Welding Operations Using Thermal Imaging and Accelerated High-Fidelity Digital Twinning
Round 1
Reviewer 1 Report
Dear Authors,
thank you very much on the work you have done on combining FE-Models and in-situ measurments for qualty controlled welding. In the introduction you only mention that hot-cracking is one of the occuring defects, but in fact the quality of the weld is not only influenced by hot cracks and resudial stresses, but most likely due to the occurance of pores, lack of fusion, non metallic inclusion and others. You have chosen a FEM model without simulation the solidification, to predict the reaction of the material to the induced heat, but nearly all mentiond defects are solidification/remelting induced, including hotcracks. I would suggest that you add a lot more information on the nature of welding defects and put your work in perspective to that.
In the experimental setup, please add information(measured) of the chemical composition of the basematerial, due to the fact that the occurance of hotcracks and that the materials mechanical properties are directly related to the chemical compostion, inculding the content of nitrogen. Please add also the measured mechanical properties you have measured (on the experimentally used materials). Furthermore give a propper microsturctural analysis for your material. This has also a big influence on the occcuring defaults, and mechanical propertes and on the reaction of the material to the heat (f.e. releasing internal residual stresses).
Secondly, you refer to the nuclear energy as one area where welding is used, but in a big part of the welded joint in that area of industire undergo a post weld heat treatment, which can be madatory for those materials.
In the results-section you show that you can predict a internal stress field due to thermal imaging, which is very nice work, on the mathematical side, but due to missing microsturcure, weld seam geometry and missing mechanical properties of the weld metal, it is not possible to relate the work shown to actual welds. Furthermore, please show, that you have done real welding with a molten phase, and not only a heating of the material. In my oppinion the used welding energy is quiet low for really perform a weld.
Lately there, is no experimantal data given wich relates your result to the real measurement like in-situt measurement of strains, stresses or so on.
Please add the requested information from above to make your wore more valuable. Please furthermore take note that, the materials properties, e.g. yield strength and ultimate tensile strength are changing with the termerature, wich can have a significant influence on the stress state due to deformations.
Besides of that, I have some minor comment to improve your work:
- The introduction needs to be re-written, because in my oppinion there are sentences that does't make any sense and are incomplete.
- Introduction Line 26: There is an incomplete sentence.
Overall, i think, that this paper includes great work on the modelling side of the problem, but due to the missing information on the material and the weld seam generated in the expermient, i cannot estimate how good this model is in terms of "real world" welding, Hopefully you add tha data.
Best regards
a Reviewer
Author Response
Answer to Reviewer 1:
In the introduction you only mention that hot-cracking is one of the occuring defects, but in fact the quality of the weld is not only influenced by hot cracks and resudial stresses, but most likely due to the occurance of pores, lack of fusion, non metallic inclusion and others. You have chosen a FEM model without simulation the solidification, to predict the reaction of the material to the induced heat, but nearly all mentiond defects are solidification/remelting induced, including hotcracks. I would suggest that you add a lot more information on the nature of welding defects and put your work in perspective to that.
Indeed, hot cracking is not the only defect that can occur in welding. The focus on hot cracking is due to the type of experiment (PVR tests) that was performed at EDF. To make this clear, we have included mentions of other types of defects in the introduction.
In the experimental setup, please add information(measured) of the chemical composition of the basematerial, due to the fact that the occurance of hotcracks and that the materials mechanical properties are directly related to the chemical compostion, inculding the content of nitrogen. Please add also the measured mechanical properties you have measured (on the experimentally used materials). Furthermore give a propper microsturctural analysis for your material. This has also a big influence on the occcuring defaults, and mechanical propertes and on the reaction of the material to the heat (f.e. releasing internal residual stresses).
We cannot easily perform a detailed microstructural analysis of the material being, as this would involve extending the project to another branch of the company and would therefore delay the review process by an unknown amount of time (months). However, we did include the chemical composition of the steel in the appendix of the paper. The boron content is indeed critically important according to:
38.Tran Van, G. Determination of a liquation hot cracking criterion as a function of boron content and its location for 316L austenitic stainless steel. Theses, Université de Bretagne Sud, 2018
The temperature dependent material properties have also been added in a table in the appendix. They were obtained in previous research works by EDF R&D in collaboration with CEA Saclay and Université de Bretagne Sud.
Secondly, you refer to the nuclear energy as one area where welding is used, but in a big part of the welded joint in that area of industire undergo a post weld heat treatment, which can be madatory for those materials.
It is true that heat treatments are often used in the nuclear industry. In this lab experiment, which is only meant as a technological demonstrator for our digital twinning methodology, we did not perform any pre-heat or post weld heat treatment. However, such treatments could easily be taken into account by representing them in the numerical model.
In the results-section you show that you can predict a internal stress field due to thermal imaging, which is very nice work, on the mathematical side, but due to missing microsturcure, weld seam geometry and missing mechanical properties of the weld metal, it is not possible to relate the work shown to actual welds. Furthermore, please show, that you have done real welding with a molten phase, and not only a heating of the material. In my oppinion the used welding energy is quiet low for really perform a weld.
The energy was chosen to melt the material yet not pierce the 3,5 mm thin specimen. We added a picture of the specimen after the experiment to prove that the material has indeed melted.
Lately there, is no experimental data given wich relates your result to the real measurement like in-situt measurement of strains, stresses or so on.
Currently, we cannot obtain such measurements as our experimental setup is not designed in this way. Therefore, only the thermal predictions are validated, using the thermocouples as independent measurements after assimilation of data from the IR camera. For the mechanical fields, we only compare the predictions to the calibrated FE simulations with no indication of hierarchy in terms of precision, which is a limitation of our work. Future work may include stereo-correlation data to measure the strain field.
Please add the requested information from above to make your wore more valuable. Please furthermore take note that, the materials properties, e.g. yield strength and ultimate tensile strength are changing with the temperature, which can have a significant influence on the stress state due to deformations.
The temperature dependent material properties have been added to a table in the appendices.
Besides of that, I have some minor comment to improve your work:
- The introduction needs to be re-written, because in my oppinion there are sentences that does't make any sense and are incomplete.
- Introduction Line 26: There is an incomplete sentence.
Thank you for pointing out these mistakes, we have revised the incomplete sentences and hope that the text is now clearer.
Reviewer 2 Report
The work presented is complete and well discussed. Further theoretical improuvement and application could be explored starting from this paper. The paper could be accepted in the present form.
Author Response
Answer to Reviewer 2:
Thank you for your comments on our research. Future experiments are already programmed and should extend the current results.
Reviewer 3 Report
This is fine manuscript that should be published. almost as is The authors have produced a practical, useful solution to the problem of online monitoring and predicting weld failures. I have only minor edits that must be attended to.
Minor edits
line 70 replace 'allows' with 'allows us'
line 134 replace ' allows to control ' with 'allows the control of'
line 161 γ not used yet. Does not need definition here.
line 164 replace 'is close' with 'is closed'
line 166 α does not need definition here. It has not been used yet. Define it in or after line 169.
After equations 8 and 9 λ is not defined
In figure 2 , what is Qm ?
line 219 What is K?
Equations 26 and 27 What is Hk ?
In Appendix B Figure 1 and 2 What is C.I., as in C. I. theta ?
Author Response
Answer to Reviewer 3:
Thank you for the detailed list of mistakes in the text and the equations. All of them have been checked and the text should be more clear now. We also added a footnote explaining what the variable K is (a ratio between two parameters). Answering your final question, C. I. is the same confidence interval that is shown in all other figures. For lack of space, we decided to only write C. I. in the legend. A comment was added in the caption so it is more easily understandable.
Round 2
Reviewer 1 Report
Dear authors,
all comments are adressed and i can accept your replies.
I have no further comments.
Best regards
a reviewer
