Composite Wing Free Size Optimization
Answers
-
So, if I chose a area ratio of 0.5, each ply I get after smoothening will be at least half the size of the original ply?
Thank you very much!
Gonçalo Pimenta
0 -
If you see the PDF there is a tiny area which was taken care of by OSSmooth as the area falls within the tolerance.
If you have a higher ratio, small regions will be ignored,
0 -
Hello @Prakash Pagadala,
Thanks for the reply!
I have two more questions:
1 - When I export the .stp from ply smoothening I sometimes get the plies perfeclty split up as shown now
<?xml version="1.0" encoding="UTF-8"?>
And other times I get just three components where I can't distinguish each ply? Has this ever happened to you before?
<?xml version="1.0" encoding="UTF-8"?>
2 - When I try to run a size optimization after the free size optimization it gives me the following error:
*** See next message about line 256 from file: size.fem
'STACK 4 SMEAR'
*** ERROR # 1901 *** in the input data:
STACK must have either a list of PLYs or substacking data.
Could this be because when I run ply smoothening of ossmooth it creates/deletes existing plies so I don't have enough or have too much design variables?
Thank you very much for your help!
I'll send the .fem and .out files so you can take a look!
Once again, thank you so much!
Gonçalo Pimenta0 -
Hi,
Can you check if there are any missing plies in STACK 4?
0 -
When you are exporting to .stp are you changing any options?
0 -
When you say stack 4, do you mean laminate 4?
I always use the same export options for .stp I think. I have them recorded on a file
0 -
Just checked, after smoothening, stacks 4 and 5 don't have any plies on them. Do you have any idea why they are gone?
Do you think they are gone due to the area ratio value? Wasn't area ratio supposed to be related to the original ply size?
Only stacks 4 and 5 are gone, that are the design variables constrained to a maximum size.
They are the ones highlighted below:
<?xml version="1.0" encoding="UTF-8"?>
Thanks!!
Gonçalo Pimenta
0 -
Hi,
I am checking a similar case with experts. I will get back to you soon,
0 -
Hello @Prakash Pagadala,
Thank you for taking the time to look at it!
In the meantime I've been running different iterations of OSSMOOTH with different area ratios and the result seems to always be the same which is odd.
So far, I've tried with .1, .2, .3, .4, .5 and .6 on the area ratio (and 40 iterations, with no geometry creation nor splitting) and I always get 28 final plies spread across 3 laminates.
The other two laminates that area the areas that I've limited to a maximum thickness now have no plies and their type is now 'Interface Ply'.They have an interface definition but they are empty.
I've also tried instead of OSSMOOTH just running Ply Geometry Smoothening but it doesn't seem to change much (with the same area ratios from OSSMOOTH) - While they both start with 50 plies to clean - OSSMOOTH always go to 28 plies while Ply Geometry Smoothening goes to 49 plies.
If you need any files, let me now!
Thank you very much!
Gonçalo Pimenta
0 -
An update:
I just tried ply geometry smoothening with a .6 area ratio and the final number of plies was 29. There was also another difference - the laminates that with OSSMOOTH were changed to 'Interface Ply', with Geometry Smoothening stayed 'Ply Laminate'
Thanks!
Gonçalo Pimenta
0 -
0
-
Please confirm you received via file transfer link the 3 files I sent you yesterday.
Thank you very much!
Anything else, please ask!
Gonçalo Pimenta
0 -
Hi,
I got the files in my box (yet to download)
I will let you know if I need any other information from you,
Thank you,
0 -
Hello @Prakash Pagadala, how are you?
Do you have any news regarding the ply smoothening?
Thank you very much!
Gonçalo Pimenta
0 -
Hi,
Not yet, I will reply to you as soon as I get a feedback from experts.
0 -
0
-
Hello @Prakash Pagadala,
I just did my exams and I'm going back to work on the wing. Do you have any news?
Thank you very much!
Gonçalo Pimenta
0 -
Hi,
I have escalated the issue to experts. Looks like the issue is with memory management.
They are working on it. I will let you know as soon as I get the updated version,
0 -
Do you have any idea if the memory management problems are due to the model?
If so, is there anything I can do in the meantime to try and get different results? Like applying some good practice method when setting up the method that may have escaped me (from what you've seen so far)?
Or the problems are related to something else?
Thank you very much!
Gonçalo Pimenta
0 -
Hello @Prakash Pagadala, how are you?
As I told you, I've been running some smoothening simulations to try to find out what was going on. After some attempts I found that the value ratios we were using were to high. Check the table below:
<?xml version="1.0" encoding="UTF-8"?>
With an area ratio of around 0.0875 the result of the smoothening process stagnates and it doesn't matter if we use a higher value.
With this, I have three annotations:
1) Before the Free Size optimizations, I set up a symmetric constraint. The plies resulting from the free size optimization are not symmetric
2) On Hyperworks, the shape of the ply before and after smoothening seems to be the same. I get a new component created after smoothening named after the ply but it doesn't have any elements, just geometry with the contour of the new smoothened ply
3) I get a weird point collection shaped in U that I can't hide
Take a look at the print screens below:
<?xml version="1.0" encoding="UTF-8"?>
fig. 1) This is ply #102200 before the smoothening operation. As you can see, the plies are not symmetric although they are constrained to be as mentioned on 1)
<?xml version="1.0" encoding="UTF-8"?>
fig. 2) This is the same ply #102200 after smoothening. See the U shape points mentioned on 3 and also how the ply it is the same shape as before as mentioned on 2)
<?xml version="1.0" encoding="UTF-8"?>
fig. 3) This is ply #102200 as exported on .stp. This is the shape I have on the geometry of the component with the same name of the ply created after the smoothening process
Is there a step I'm missing?
Thank you for your help! As always, if any file is needed, just ask!
Gonçalo Pimenta
0 -
Altair Forum User said:
) Before the Free Size optimizations, I set up a symmetric constraint. The plies resulting from the free size optimization are not symmetric
2) On Hyperworks, the shape of the ply before and after smoothening seems to be the same. I get a new component created after smoothening named after the ply but it doesn't have any elements, just geometry with the contour of the new smoothened ply
Hi @goncalop
Is it not at all symmetric or not even close to symmetric shape?
In Ply geometry smoothing, did you select Update ply element sets? This will create a new ply set with updated elements. However, we noticed that the card image for the new sets are not updated and I have created a ticket to solve this problem and our developers are working on it.
0 -
Hello @Prakash Pagadala
Altair Forum User said:Is it not at all symmetric or not even close to symmetric shape?
As you can see on the pictures, the plies can look symmetric but if you look closely they are not.
<?xml version="1.0" encoding="UTF-8"?>
Take a look here at the areas highlighted in red and yellow. As you can see, the gurney of the wing on the right side is gone (although it is still on the left side) and the right side is shorter and with more cuts (compared to the left side that goes all the way to the end of the wing).
Altair Forum User said:In Ply geometry smoothing, did you select Update ply element sets?
Yes, I did. On the options, the only one I did not select was Split Disconnected Ply Regions...
Thanks!
0 -
Altair Forum User said:
This will create a new ply set with updated elements. However, we noticed that the card image for the new sets are not updated and I have created a ticket to solve this problem and our developers are working on it
What you mean by this is that the smoothening operation creates new ply sets but the elements that are in each set are not updated so the shape of the new 'smoothened ply' is the same as the ply resulting from the free size optimization?
0 -
Hi,
If you see the Smoothing GUI, you will notice a checkbox for sets update. This is supposed to update the existing sets but somehow this is not working as expected.
This is reported and developers are working on it.
0 -
Hi @Prakash Pagadala, thanks for the answer!
Do you have a workaround for this that I can use now?
Thank you very much!
0 -
I'm trying OSSMOOTH with the Ply Geometry Smoothening settings I used before but while with Ply Geometry Smoothening it took 3hours, with OSSMOOTH it has been running for the past 24h and is still 'projecting lines: 54' and has been here for a while..
I'll leave it running and see if it returns something
0 -
Altair Forum User said:
Do you have a workaround for this that I can use now?
We, copy the elements from new set to old set is the workaround which I use. But this is very tedious if you have huge ply data.
0 -
Hello @Prakash Pagadala, how are you?
As mentioned on the email I sent you yesterday, I think the problem with the ply smoothening may be solved with this workaround. I will post some pictures to show the results:
<?xml version="1.0" encoding="UTF-8"?>
So, this it the original ply 102200 after the free size optimization.
To begin with, I ran OSSMOOTH with 100 iterations and an Area Ratio of 0.075. I left Split Disconnected and Create Geometry Unchecked. The second one I left unchecked because it seemed that it was causing Hypermesh to fail to respond and start using all the ram available indefinitely.
So, after OSSMOOTH, the result was:
<?xml version="1.0" encoding="UTF-8"?>
On the image above we can see the result from OSSMOOTH. There is probably not much difference between the ply before and after OSSMOOTH but that may be due to the Area Ratio, a difference on the number of plies is definitely visible.
After OSSMOOTH, to check the .stp file I ran Ply Smoothening - 0 iterations just to create the contours and export to .stp to check the result on Catia:
<?xml version="1.0" encoding="UTF-8"?>
<?xml version="1.0" encoding="UTF-8"?>
On Catia we can see that the ply shape is not rounded and is probably defined by the outline of the elements that make the ply.
After this first test, I ran Ply Smoothening again but this time with 100 iterations and the same area ratio as OSSMOOTH. The result was:
<?xml version="1.0" encoding="UTF-8"?>
Althought the outline of the ply stayed the same, we can now see the rounded contour on the geometry that was created by ply smoothening.
Opening the .stp on Catia:
<?xml version="1.0" encoding="UTF-8"?>
We can now definitely check that we have some nice, rounder ply shapes.
0 -
While doing the process on post above, I checked some odd things. They are mentioned below:
1) When I try to run OSSMOOTH with the result from the optimization loaded, I get this message:
<?xml version="1.0" encoding="UTF-8"?>
'Ply Shape Interpretation Failed, please check log file'
However, when I use OSSMOOTH loading the .fem file from the OSSMOOTH menu, everything works fine...
2) After Ply Smoothening (100 iterations), I get these U shaped nodes that I can't hide. They don't seem to be part of any component...<?xml version="1.0" encoding="UTF-8"?>
These nodes have appeared before when using Ply Smoothening.
_______________________________________________________________
Now the next step is to start working on the Free Size to Size optimization. From what I've seen so far, Optistruct is giving me a lot of errors when I use the file with the cleaned plies.For example:
*** ERROR # 605 ***
Element 174360 is referenced by STACK 3 (on PLY 309100) and by
STACK 5 (on PLY 517100) simultaneously.
And*** ERROR # 1901 *** in the input data:
STACK must have either a list of PLYs or substacking data.I haven't checked on this errors yet but I will eventually.
I the meantime I tried a free size to size optimization with the .fem file before cleaning and it worked. It first stopped after the first iteration due to 0 thickness elements but after a search on this forum I found a solution - imposing a composite failure constraint, that seemed to solve the problem.
After checking the result I found that some constraints were violated mainly - maximum laminate thickness (on the areas I mentioned before that are constrained in thickness due to bonding height. I separated the laminates as suggested by you and while it works for the first step of the optimization, on the second it doesn't work no more).
Thank you so much for your help!
Gonçalo Pimenta
0 -
Hi,
Can you share the complete .out file, please?
0
Altair Employee
