Hi Christian
I believe that Code Aster comes with Salome Meca install, or you install it on top of Salome. From Linux you may want to install from source, but that's more of a PITA. From Windows I will definitely use the binaries of course!
You can install Code Aster separately, but Salome will I think give us the 'royal road' to simply getting to a working model, meshed and with a valid config file. After that we can look at what has been created, run it from the command line with Code Aster directly and then think further how to best integrate it into our workflow. Salome is worth having anyway for it's CAD features.
This looks like the place to get it: https://www.salome-platform.org/?p=2446
I think I would be aiming to use Code Aster from the command line, and still build my meshes using gmsh from Python. But Salome has advanced meshing capabilities too. It's really quite advanced but the menus seem like a Labyrinth. I will try to install tomorrow night.
Have not had much time for any computing work in the last few days but I looked at a few YouTube vids that show how to use Code Aster separately from Salome. It seems Python is quite deeply embedded in Code Aster, and you can include Python in the command (.com) file.
What's your timeframe? Are you facing a decision of whether to go full-steam on this, or can we just take it slowly? You were thinking of buying a Lisa licence, is that correct? My preference would be to go slowly and do a few experiments as time permits, mainly because I have have been following the path of Rayleigh integral/BEM, and have several layers of work stacked up, with little progress yet due to other commitments. I fear I will become lost if I go too far down another path
. Baby steps
Paul
I believe that Code Aster comes with Salome Meca install, or you install it on top of Salome. From Linux you may want to install from source, but that's more of a PITA. From Windows I will definitely use the binaries of course!
You can install Code Aster separately, but Salome will I think give us the 'royal road' to simply getting to a working model, meshed and with a valid config file. After that we can look at what has been created, run it from the command line with Code Aster directly and then think further how to best integrate it into our workflow. Salome is worth having anyway for it's CAD features.
This looks like the place to get it: https://www.salome-platform.org/?p=2446
I think I would be aiming to use Code Aster from the command line, and still build my meshes using gmsh from Python. But Salome has advanced meshing capabilities too. It's really quite advanced but the menus seem like a Labyrinth. I will try to install tomorrow night.
Have not had much time for any computing work in the last few days but I looked at a few YouTube vids that show how to use Code Aster separately from Salome. It seems Python is quite deeply embedded in Code Aster, and you can include Python in the command (.com) file.
What's your timeframe? Are you facing a decision of whether to go full-steam on this, or can we just take it slowly? You were thinking of buying a Lisa licence, is that correct? My preference would be to go slowly and do a few experiments as time permits, mainly because I have have been following the path of Rayleigh integral/BEM, and have several layers of work stacked up, with little progress yet due to other commitments. I fear I will become lost if I go too far down another path
. Baby steps Paul
Elmer creates a complex velocity field for a given frequency in the harmonic analysis, normalised (I believe) by modal mass. It does not use mode superposition, but I think gives an equivalent result. I beleive that you should end up with a complex field for your speed shapes. The equivalent equation from Putra and Thompson paper does not take the magnitude of the bottom expression. I think the complex value needs to be preserved? There is a phase shift at each resonance.
I asked to my son (informatic is his job now) about the best solution : Code Aster by compiling sources or already integrated to Salome and we came to the same conclusion : Salome even if it is bigger in memory space. Double advantage : no need to compile, further possibilities for the future. I didn't know the source you link. That's the right one : from EDF-CEA and multi OS.
Yes. That's my approach too
No real time frame. Several topics also at the same time. About Lisa, the idea was not too buy a license but too install the free version on a Windows partition. I keep that aside. For now I am taking time to clean my current laptop and make room to install Salome.. So baby steps is ok.What's your timeframe? Are you facing a decision of whether to go full-steam on this, or can we just take it slowly? You were thinking of buying a Lisa licence, is that correct? My preference would be to go slowly and do a few experiments as time permits, mainly because I have have been following the path of Rayleigh integral/BEM, and have several layers of work stacked up, with little progress yet due to other commitments. I fear I will become lost if I go too far down another path
Christian
Christian
You probably know about it already, but it ocurred to me that the way you are mixing math expressions and python code would be a good case for a Jupyter notebook. You can have markdown (with math), python, graphs etc all integrated together in an active notbook. You can export to PDF too, either using the browser for rough draft or using Latex. Just a thought.
You probably know about it already, but it ocurred to me that the way you are mixing math expressions and python code would be a good case for a Jupyter notebook. You can have markdown (with math), python, graphs etc all integrated together in an active notbook. You can export to PDF too, either using the browser for rough draft or using Latex. Just a thought.
Thank you Paul. Yes I had an eye on Jupyter notebook. I think even if my code is not organized like that my wish is to switch on or off the pdf export depending if the code is running in a development mode (to show how it works) or in a production mode (exporting results). So I make the choice to keep a full script. Jupyter is an other learning curve I don't want to send to much time in.
By the way, good news, I manage with my son's help to enlarge the linux partition on my computer (still a dual boot even if Windows is no more used).
I had also a look to Code Aster... Not clear at the moment between Salome, Salome meca, Code Aster what is in what... I have download different binaries and source to try to see what is inside...
Christian
I installed Salome from solome-platform.org, but it does not come with code aster. I will need to use the versions from code-aster-windows.com. I think you will use the versions from code-aster.org. The sizes are truly huge.
I may experiment with just code aster, there are many test cases for plates available.
I may experiment with just code aster, there are many test cases for plates available.
So it is confirmed now... No Code aster in Salome... but it seems to be in Salome-meca, binary available or from the code source. To be continued.
Yes the number of test cases including 2D plates is impressive.
Don't worry Paul... we have also a life (and also problems) aside those DIY topics! I am in the opinion that started code aster might lead to have to solve computer issues (see the list of prerequisites). Here I just inspected the install file to try to understand what it does/modifies. So no big progress neither.
Christian
@pway
Hi Paul
I finally downloaded Code Aster in a singularity container (named now apptainer) of Salome-meca as the old universal binaries didn't run on my computer. Well... Salome seems working but while defining the model based on the video tutorial something went wrong... I have more and more doubt it is a good solution...
Christian
Hi Paul
I finally downloaded Code Aster in a singularity container (named now apptainer) of Salome-meca as the old universal binaries didn't run on my computer. Well... Salome seems working but while defining the model based on the video tutorial something went wrong... I have more and more doubt it is a good solution...
Christian
Hi Paul
From my little experience with it, I think salome-meca/code aster are complex software very sensitive to a long list of software package versions. The Linux distribution I use (Manjaro/Arch) is unfortunately not in the short list of the tested distrubutions (Ubuntu/debian, CentOS). I hoped the singularity containers would avoid configuration problems but seems not completely. Debugging is not easy when searching for the cause of the errors. I would prefer to put the energy on the center of our topic like with the FDM script (or FENICS?) than on debugging problems of libraries.
By the way, I am wondering if the modal synthesis is the right approach. Let me explain. We have seen from the posts from Steve, Eric or mine that there is a wave absorption with the distance from the exciter that increases with the frequency. This is a key phenomena. I wonder if it is well model with the modal approach. I guess not as in the eigenfrequencies or mode shape calculation the damping factor is not used. I have the feeling the reintroduction of the damping coefficient afterwards in the frequency domain doesn't model the geometric aspect. Your opinion?
RPN? Ah! Reversed Polish Notation from the HP calculator when I was a student! I see. RPN Python implemention. Happy recreational programming!
Christian
I expect FEM analysis will be fruitful only for LF, but that's OK I think.
I have also noticed that at higher frequencies, the response is confined more and more closely to the exciter. This is a benefit, because it limits the SPL at HF which would otherwise be excessive, and it makes it more like a point source, with favourable dispersion characteristic. I also expect that because of this high damping at HF, the model for the panel would actually be quite simple at HF, with no resonance but a damped travelling wave decaying exponentially with radius.
Re the calculator. Yes I still have my HP15C, which I treasure like some kind of Faberge egg - an artifact that is representative of the peak of that technology, never to be bettered. I know there are many RPN calc programs and apps out there (I have a HP42 simulator on my ipad).
My simple efforts are a sort of meditation on the usability of a stack as a general purpose user interface. To try to think for example how it might be applied to other applications like CAD/drawing, OS shell, PIM, text processing etc. I believe that user interfaces went too quickly to so-called direct manipulation/GUI interfaces, which can be very inefficient, 'noisy', and often surprisingly confusing.
Paul
- LF is where low modal density, and uneven modal distribution produces a lumpy frequency response, and where the most gains can be made from understanding individual modes.
- LF is where acoustic short circuiting happens, and where gains might be made from the use of a baffle or perhaps canvas/membrane.
- At higher frequency, the response is smoother anyway, and more amenable to equalisation.
- I think that most changes one might make to even-out the distribution of modal frequencies at LF should also improve the higher frequencies, or at least not affect them adversely.
- I always have had low expectations for the accuracy of eigenfrequency calculations at high frquencies, and Ive only bothered to plot to 500 Hz
I have also noticed that at higher frequencies, the response is confined more and more closely to the exciter. This is a benefit, because it limits the SPL at HF which would otherwise be excessive, and it makes it more like a point source, with favourable dispersion characteristic. I also expect that because of this high damping at HF, the model for the panel would actually be quite simple at HF, with no resonance but a damped travelling wave decaying exponentially with radius.
Re the calculator. Yes I still have my HP15C, which I treasure like some kind of Faberge egg - an artifact that is representative of the peak of that technology, never to be bettered. I know there are many RPN calc programs and apps out there (I have a HP42 simulator on my ipad).
My simple efforts are a sort of meditation on the usability of a stack as a general purpose user interface. To try to think for example how it might be applied to other applications like CAD/drawing, OS shell, PIM, text processing etc. I believe that user interfaces went too quickly to so-called direct manipulation/GUI interfaces, which can be very inefficient, 'noisy', and often surprisingly confusing.
Paul
Fully agree, for the same reasons. that's been my expectaion and focus from the start.
Eric
I have not seen any of my old calculators in a long time, I suspect they are gone. But I do have one of my Dad's slide rules. He was an electrical engineer at Bell Labs in the 60's. The slide rule I have is a pretty small one. He had a much larger one that I suspect my older brother grabbed before I thought to, haha.
Eric
My dad had a very nice slide rule too, not sure where it went.
Logarithm tables obsolete when I started high school, but had to learn them.
Slide rules obsolete at the same time.
Scientific calculators for all but student or occasional casual use obsolete with the PC and spreadsheet
Video tape obsolete with dvd
DVD obsolete with streaming (everyone has a dvd player still but you can’t get the content)
CD obsolete with streaming except for hifi dinosaurs like us
next, AI makes all human thought a quaint, wistful memory….
Logarithm tables obsolete when I started high school, but had to learn them.
Slide rules obsolete at the same time.
Scientific calculators for all but student or occasional casual use obsolete with the PC and spreadsheet
Video tape obsolete with dvd
DVD obsolete with streaming (everyone has a dvd player still but you can’t get the content)
CD obsolete with streaming except for hifi dinosaurs like us
next, AI makes all human thought a quaint, wistful memory….
+ @Veleric
Seems we have the same experience of time flies. I still have my slide rule. Never really used, it comes from a time the teachers were not 100% the calculator would be allowed or not (early 80s?). When I closed my parents's house some years ago, I found my TI58.
Well, lets do by a innovative thinking that your projection doesn't become true... or not too soon. Maybe an opportunity to read Asimov or other SF writers.
A book that might be interesting : "Theory and analysis of Elastic plates and shells" JN Reddy
In page 333 (extract below) is the natural frequencies for a simply supported orthotropic plate. With the criteria you proposed Eric : less a ratio 1.3 between f(1,1) and f(1,3), the ratio b/a can be determined.
In page 333 (extract below) is the natural frequencies for a simply supported orthotropic plate. With the criteria you proposed Eric : less a ratio 1.3 between f(1,1) and f(1,3), the ratio b/a can be determined.
Last edited: