Hello, thanks for your interest in the channel. You make a very good point that the extraction of stress-strain plot can be done using a Python Script and reading the resulting ODB files. My goals for this channel is 'getting as many people started with modelling' and as a result I always try to do things in a simple way so that many people can understand and follow. The use of a Python script will definitely require the user to know how to write Python codes and that in my opinion is for advanced ABAQUS users. Having said that, the second vision for the channel is helping more advanced users 'curate effective solutions for their modelling challenges,' and in that case the suggestion of Python Scripting for reading ODB files is a more efficient and effective way of doing that. I will consider that suggestion and make a video for this category of viewers of the channel.
Hi Chunxu, I will make a video on how to plot stress-strain curve based on ODB data. I do not know of any open-source code for doing this but it is relatively straightforward. Just keep coming back to the channel and see my video in future about this.
I would image the same way as the implicit solver. You may also track the E-strain as a history variable specified at the start of the simulation. Is this helpful?
@@MichaelOkereke Thank you very much sir , I followed the same way so I checked on strain components E in field output and in history output, but after the end of computing I had logarithmic strain LE instead of strain components E. Does that mean anything. Thank you very much.
You can manually select those nodes/elements and put them into a set. You can also use other methods for extracting stress strain other than the homogenization approach here for example the experimental-equivalence approach (see my video on how to get stress-strain plots in abaqus by 3 approaches).
@MichaelOkereke, in the history output field, for you, the results are resulted only at Integration point 1, when I followed the same, I got 8 integration points. for each element, the S11 at eight points are stored. 3000 elements with 8 points resulted in 24000 temp data in my case, How you get the data from only one IntPoint. Also, in the video, when you click the ok in the avg(XX,XY) menu, for you only one XY data is shown in the left side menu. For me, those 24000 data was written as temp, what I missed from the video?
Hello, I think what is happening here is that your model must have an element that is prone to multiple integration points hence your stress history variables have 8 integration points. If this is the case, then what I ahve shown here will not precisely work for you. You will have to first integrate the values of stress, S11 across the 8 integration points before then volume averaging that integrated value. It is a two step process but not very trivial. It is probably best to do so using a Python script that first loops over the 8 elements to extract a value for S11 for that element before then looping over the total number of elements using the method I showed here. On avg(XX, XY), I think this might be linked to the problem with multiple integration points. If not, then check that you select the averaging function first before giving the data a name and then plotting the result. I hope these help.
Hello @Tesfahun, thanks for your question. I currently do not have any video on cyclic loading but it relatively straight forward. Here is a vidoe on TH-cam by someone else that might be helpful: th-cam.com/video/aGrYJMelrYw/w-d-xo.html Good luck with the research!
Very interesting video but I have tried your way on fracture modeling using XFEM but the strain stress curve I have got from Abaqus is different from your old way to get that curve by turning the force and displacement into stress and strain. Actually the strain stress curve I have got it from Abaqus almost the same identical with the curve I got it by old way from force and displacement but till the peak point (fracture) but the curve from Abaqus after the peak ppoint return directly to zero without softening part which I got it from the old way. So, I wondering is that due to fracture and is your way is valid for fracture modelling too or only for elastic modelling only and if you could help me to improve that I mean to get the strain stress curve for fracture modelling directly from Abaqus. Thanks in advance
Hello @John Nabil, thanks for your continued interest in the channel. I am not sure why you are having varying stress-strain plots, maybe I do not completely understand what you are trying to do. The essense of these different stress-strain plot videos is not to make identical plots. Of course, the approaches I showed would work even in fracture - as long as elements are in the model - these approaches volume-average (i.e. homogenize) the response in the region(s) you have specified so they will work always. It is really aimed at equipping my viewers with different ways they can probe their models to extract stress-strain plot. I will suggest that you discuss your approach with someone more experienced in these things to better understand the mechanics of your implementation. FEM is a wonderful tool but if you put in the wrong values/process, you are bound to get the wrong/misleading outcomes. The overriding principle is that these methods I have shown in the video are right and you should use them to explore certain features of your model. For like-for-like fitting between all of them, that is a different subject and you have to set up the models right to make such comparisons.
@@MichaelOkereke Thanks for replying. First I got the strain stress curve from your old way to get that curve by turning the force and displacement into stress and strain in Excel sheet and it looks fine but when I got the strain stress curve directly from Abaqus as you have explained in this video the curve looks not fine as after the peak point the curve returned directly to zero without softening part. By the way the stress curve alone I got it directly from Abaqus looks fine and also the strain curve looks fine both regarding to time as x axis but when I combine the strain and stress together the resulted curve not looks fine as after the peak point it returned directly to zero. Yes I know it should not be identical plots from the two ways but actually I wonder as since I got the stress strain curve by the old way from displacement force curve and it is fine which means that my model working well so why the direct stress strain curve from Abaqus looks strange. I am not sure but think the strain values after the peak point (fracture) is very low compared to the values before fracture or even compared to the size of the elements so the strain values going to zero and the stress values too but how I can modify this I mean how to overcome this to get the good stress strain curve, so if you could help me to improve that. Thanks in advance
Hi, Thanks for sharing. I wonder how the avg(X,X) works, It seems that it calculates the average stress, not the volume average stress. Can you clarify this?
Hello @Mina, this is a very valid question and well spotted. Technically, the volume averaged term will require it expressed as in my video at 1:00 where you sum up all the stresses in the elements and divide by the sum of all the elements in the model. Whilst this makes sense but what is the practical value of such a volume-normalized stress. On its own, it does not mean anything but when you combine this volume-averaged term with say the volume averaged strain, then you have a sensible value for say Young's Modulus which is the ratio of this stress to the strain. In this instance the volume term will be dropped leaving only the summation of the stress and strain values. What will make more sense will be the average stress in comparison with the average of the strain the model over the region of interest. This is why I use this as a proxy for the volume averaged stress, but you are right that avg(x,x,...) is simply an average.
Great video, very instructive !
Glad it was helpful! Thanks Bruno for your interest in the channel.
Hi if i simulate a bending of sandwich how can i get stress strain curve in with set i selected step variables
It is a pleasure to see that you have mastered these things and explained them clearly. A better suggestion is to use Python for ODB data extraction.
Hello, thanks for your interest in the channel. You make a very good point that the extraction of stress-strain plot can be done using a Python Script and reading the resulting ODB files. My goals for this channel is 'getting as many people started with modelling' and as a result I always try to do things in a simple way so that many people can understand and follow. The use of a Python script will definitely require the user to know how to write Python codes and that in my opinion is for advanced ABAQUS users. Having said that, the second vision for the channel is helping more advanced users 'curate effective solutions for their modelling challenges,' and in that case the suggestion of Python Scripting for reading ODB files is a more efficient and effective way of doing that. I will consider that suggestion and make a video for this category of viewers of the channel.
How to use python for ODB data then? Is there open-source code? Thx
Hi Chunxu, I will make a video on how to plot stress-strain curve based on ODB data. I do not know of any open-source code for doing this but it is relatively straightforward. Just keep coming back to the channel and see my video in future about this.
Thank you very much dear Prof . please how can i use the (E, total strain components) in explicite solver ? thank you again.
I would image the same way as the implicit solver. You may also track the E-strain as a history variable specified at the start of the simulation. Is this helpful?
@@MichaelOkereke Thank you very much sir , I followed the same way so I checked on strain components E in field output and in history output, but after the end of computing I had logarithmic strain LE instead of strain components E. Does that mean anything. Thank you very much.
If we did not create or model in abacus we don't get geometry option in sets. What to do instead in such case?
You can manually select those nodes/elements and put them into a set. You can also use other methods for extracting stress strain other than the homogenization approach here for example the experimental-equivalence approach (see my video on how to get stress-strain plots in abaqus by 3 approaches).
@MichaelOkereke, in the history output field, for you, the results are resulted only at Integration point 1, when I followed the same, I got 8 integration points. for each element, the S11 at eight points are stored. 3000 elements with 8 points resulted in 24000 temp data in my case, How you get the data from only one IntPoint. Also, in the video, when you click the ok in the avg(XX,XY) menu, for you only one XY data is shown in the left side menu. For me, those 24000 data was written as temp, what I missed from the video?
Hello, I think what is happening here is that your model must have an element that is prone to multiple integration points hence your stress history variables have 8 integration points. If this is the case, then what I ahve shown here will not precisely work for you. You will have to first integrate the values of stress, S11 across the 8 integration points before then volume averaging that integrated value. It is a two step process but not very trivial. It is probably best to do so using a Python script that first loops over the 8 elements to extract a value for S11 for that element before then looping over the total number of elements using the method I showed here.
On avg(XX, XY), I think this might be linked to the problem with multiple integration points. If not, then check that you select the averaging function first before giving the data a name and then plotting the result.
I hope these help.
how we apply lateral cyclic load on abaqus
Hello @Tesfahun, thanks for your question. I currently do not have any video on cyclic loading but it relatively straight forward. Here is a vidoe on TH-cam by someone else that might be helpful: th-cam.com/video/aGrYJMelrYw/w-d-xo.html
Good luck with the research!
Very interesting video but I have tried your way on fracture modeling using XFEM but the strain stress curve I have got from Abaqus is different from your old way to get that curve by turning the force and displacement into stress and strain. Actually the strain stress curve I have got it from Abaqus almost the same identical with the curve I got it by old way from force and displacement but till the peak point (fracture) but the curve from Abaqus after the peak ppoint return directly to zero without softening part which I got it from the old way. So, I wondering is that due to fracture and is your way is valid for fracture modelling too or only for elastic modelling only and if you could help me to improve that I mean to get the strain stress curve for fracture modelling directly from Abaqus. Thanks in advance
Hello @John Nabil, thanks for your continued interest in the channel. I am not sure why you are having varying stress-strain plots, maybe I do not completely understand what you are trying to do. The essense of these different stress-strain plot videos is not to make identical plots. Of course, the approaches I showed would work even in fracture - as long as elements are in the model - these approaches volume-average (i.e. homogenize) the response in the region(s) you have specified so they will work always. It is really aimed at equipping my viewers with different ways they can probe their models to extract stress-strain plot. I will suggest that you discuss your approach with someone more experienced in these things to better understand the mechanics of your implementation. FEM is a wonderful tool but if you put in the wrong values/process, you are bound to get the wrong/misleading outcomes. The overriding principle is that these methods I have shown in the video are right and you should use them to explore certain features of your model. For like-for-like fitting between all of them, that is a different subject and you have to set up the models right to make such comparisons.
@@MichaelOkereke Thanks for replying. First I got the strain stress curve from your old way to get that curve by turning the force and displacement into stress and strain in Excel sheet and it looks fine but when I got the strain stress curve directly from Abaqus as you have explained in this video the curve looks not fine as after the peak point the curve returned directly to zero without softening part. By the way the stress curve alone I got it directly from Abaqus looks fine and also the strain curve looks fine both regarding to time as x axis but when I combine the strain and stress together the resulted curve not looks fine as after the peak point it returned directly to zero. Yes I know it should not be identical plots from the two ways but actually I wonder as since I got the stress strain curve by the old way from displacement force curve and it is fine which means that my model working well so why the direct stress strain curve from Abaqus looks strange. I am not sure but think the strain values after the peak point (fracture) is very low compared to the values before fracture or even compared to the size of the elements so the strain values going to zero and the stress values too but how I can modify this I mean how to overcome this to get the good stress strain curve, so if you could help me to improve that. Thanks in advance
Hi, Thanks for sharing. I wonder how the avg(X,X) works, It seems that it calculates the average stress, not the volume average stress. Can you clarify this?
Hello @Mina, this is a very valid question and well spotted. Technically, the volume averaged term will require it expressed as in my video at 1:00 where you sum up all the stresses in the elements and divide by the sum of all the elements in the model. Whilst this makes sense but what is the practical value of such a volume-normalized stress.
On its own, it does not mean anything but when you combine this volume-averaged term with say the volume averaged strain, then you have a sensible value for say Young's Modulus which is the ratio of this stress to the strain. In this instance the volume term will be dropped leaving only the summation of the stress and strain values.
What will make more sense will be the average stress in comparison with the average of the strain the model over the region of interest. This is why I use this as a proxy for the volume averaged stress, but you are right that avg(x,x,...) is simply an average.
@@MichaelOkereke Thanks for your reply.