Hello everyone.
Let's say I have a set of points (Load vs Disp) which gives the ideal "reference" plot. I also have a UMAT which has 4 parameters. Every combination of the parameters generates a load vs disp distribution.
The goal is to write a script that varies the parameters automatically until the reference Load vs Disp profile is obtained. The set of parameters that gives the ideal curve is considered to be the best. The reference curve that I spoke about is basically a set of points in an excel sheet. The goal is to obtain the same (almost same) using the perfect UMAT combination.
I want a suggestion as to how to go about this. Can I write a python script for this?
Yes, possible! You need a coupled procedure to solve this. Maybe an optimization procedure in which the objective function to be minimized will be the sum of the squared differences between the points of the actual UMAT curve and the corresponding points of the reference curve (in Excel) would be perfect.
To set up an optimization procedure you need to begin with an initial combination of the four parameters of the UMAT, then obtain the results from Abaqus analysis (in which the UMAT will be used), and then postprocess these results in order to determine the next combination of the four parameters, which will further define a new UMAT, which will be used in the next Abaqus analysis, etc. This is a cyclic procedure (loop) in which Abaqus and a suitable postprocessing software will be coupled to exchange data from one to the other until the four parameters (formally called as design variables of the optimization procedure) yield a UMAT which will reproduce the reference (Load vs Disp) curve.
Any optimization procedure is comprised of two substantial parts: the optimizaer and the solver. In the above description, Abaqus is used as the solver, and any programming languange can be used as the optimizer. I recommend you to use either Matlab or Python as the optimizer. Personally I prefer the former because it is much more user-friendly. If you are a Matlab user, things will be much more easier for you, given that there exists a special Matlab application which can efficiently postprocess the Abaqus results without much programming effort. This is Abaqus2Matlab (available to download for free at www.abaqus2matlab.com ) which can be installed on your Matlab software. If you decide to use it, I can help you in setting up the optimization process in order to solve your problem.
Yes, possible! You need a coupled procedure to solve this. Maybe an optimization procedure in which the objective function to be minimized will be the sum of the squared differences between the points of the actual UMAT curve and the corresponding points of the reference curve (in Excel) would be perfect.
To set up an optimization procedure you need to begin with an initial combination of the four parameters of the UMAT, then obtain the results from Abaqus analysis (in which the UMAT will be used), and then postprocess these results in order to determine the next combination of the four parameters, which will further define a new UMAT, which will be used in the next Abaqus analysis, etc. This is a cyclic procedure (loop) in which Abaqus and a suitable postprocessing software will be coupled to exchange data from one to the other until the four parameters (formally called as design variables of the optimization procedure) yield a UMAT which will reproduce the reference (Load vs Disp) curve.
Any optimization procedure is comprised of two substantial parts: the optimizaer and the solver. In the above description, Abaqus is used as the solver, and any programming languange can be used as the optimizer. I recommend you to use either Matlab or Python as the optimizer. Personally I prefer the former because it is much more user-friendly. If you are a Matlab user, things will be much more easier for you, given that there exists a special Matlab application which can efficiently postprocess the Abaqus results without much programming effort. This is Abaqus2Matlab (available to download for free at www.abaqus2matlab.com ) which can be installed on your Matlab software. If you decide to use it, I can help you in setting up the optimization process in order to solve your problem.
Hi Jonah,
I agree with George, you would need some sort of optimization procedure to come up with the best fit.
However before you go down the UMAT route, are you sure there is no existing built in Abaqus material that does the same or very similar? Developing a UMAT to be correct and work for every element type, condition, etc can be quite challenging.
If you do want to use UMAT, you should include your 4 parameters in the material definition so that your subroutine picks it up from there, and you can easily control it from within Abaqus.
Do you by any chance have access to Isight? It's also a Dassault/Simulia product, and it is an execution engine that you can use to run parametric studies, optimization, etc, and it integrates seemlessly with Abaqus. If not, you could find a python or matlab based optimizer perhaps as suggested before.
Alternatively, you could also use the "Parametric studies" built in feature in Abaqus to run all sort of combinations of the parameters. This is probably the quickest and simplest (apart from Isight, but you might not have that), you can generate all the result curves, and then post process to find which combination gives the best match.
Hope this is helpful, good luck with your work!
Best regards,
David
http://abaqus.software.polimi.it/v6.12/books/usb/default.htm?startat=pt04ch20s01aus104.html
How can I import a user subroutine into Abaqus component in Simulia Isight?
for optimization my material parameter I use Isight software. I have an input file that simulate a load-unload process via DISP subroutine. How can I import this inp file in the abaqus component of isight for optimization?
- Badges
- Science topic