Constrain extra nod...
 
Notifications
Clear all

Constrain extra nodes  

 

viky113
(@viky113)
Rookie
Joined: 2 months ago
Posts: 6
September 5, 2020 2:37 pm  

I am constraining the nodes of a body to a rigid body to which I am prescribing acceleration. However, I need some inertial effect in the body which is constrained. Is it possible to constrain the nodes at a specific time point or is there another way of adding the inertial effect?


Quote
Negative Volume
(@negativevolume)
Senior Member Admin
Joined: 2 years ago
Posts: 326
September 8, 2020 9:13 pm  

Hi @viky113

You can add inertia to rigid parts using *Part_inertia, so you may be able to duplicate the elements on the deformable body and assign them to a rigid part and then define that part with *Part_inertia? Just an idea. I'm confused though what you mean by "constrain nodes at a specific time point"


ReplyQuote
viky113
(@viky113)
Rookie
Joined: 2 months ago
Posts: 6
September 8, 2020 9:22 pm  

@negativevolume Thanks for your response. By "constrain nodes at a specific time point", I mean that, for instance the rigid body starts moving at time 0, and I want the nodes of the deformable body to be constrained to the rigid body at lets say time 10 ms. I want that delay in the nodes to be constrained. Do you think *PART_INERTIA would be the right way of going about it? And I am assuming I will have to prescribe motion separately to the the rigid body already defined and the rigid body duplicated from the deformable body to see that delay between the two bodies, right?


ReplyQuote
Negative Volume
(@negativevolume)
Senior Member Admin
Joined: 2 years ago
Posts: 326
September 8, 2020 9:44 pm  

I see. Normally that would be called a "birth time" but I'm not sure if that is possible for a constraint. You can use birth times for tied contacts which would be similar to what you want but you cannot use them with rigid parts. Maybe there is another way to model what you are trying to do. What exactly is it that you are trying to simulate with that delay?

Honestly, I'm not sure if that is the best way to go about it because I don't know what the simulation is doing or what the goal is. Generally, you want to limit the amount of "artificial-ness" that you add to one of these simulations. If your materials of that deformable part that is constrained to the rigid part are reasonable then I wouldn't try to introduce a defined inertia. I would change the way that you constrain that part to allow rotation of the constrained part. If this is something that you are interested in then I can help find a solution that allows this. 


ReplyQuote
viky113
(@viky113)
Rookie
Joined: 2 months ago
Posts: 6
September 8, 2020 9:58 pm  

@negativevolume I am trying to simulate a rear impact sled test. There are load plates behind my seat that restrict seatback rotation. In the physical test, the acceleration of the load plates start increasing approximately 10 ms after the initiation of the pulse. Since my load plates are constrained to the rigid floor of the sled to which the pulse is prescribed, my load plates start accelerating at time 0. Since my plates are rigidized because of constrained nodes, I am unable to match the physical test data. I believe its the absence of inertial effect as well as the slight rotation of the plates that we see in the physical test.


ReplyQuote
Negative Volume
(@negativevolume)
Senior Member Admin
Joined: 2 years ago
Posts: 326
September 8, 2020 10:13 pm  

@viky113 Ah okay now we are getting somewhere. Cool, so in the physical test are the load plates physically connected to the sled floor like you have in the simulation? If so, then I can only imagine that 10 ms delay is due to the plates flexing or something?

If they aren't connected to the sled floor, how are they connected? And do you have the acceleration for both the sled and the plates as well as rotation? You could always prescribe the motion of the sled and the plates separately to match the physical test exactly. 


ReplyQuote
viky113
(@viky113)
Rookie
Joined: 2 months ago
Posts: 6
September 8, 2020 10:36 pm  

@negativevolume The plates have load cells behind. The load cells have a big plate behind. That big plate has supporting structures that attach to the floor. The supporting structures allow for some flexing like you mentioned. However, I dont have the CAD file for the supporting structures. Hence, I have limited my simulation to the plates, load cells and the big plate. Its actually the nodes of the big plate that are constrained to the sled floor. As the plates, load cells and the big plates have been made using biased meshing and share nodes, everything seems rigidized once the pulse initiates. I would like to allow for some rotation in the plates. I have the acceleration data from the accelerometers attached to the plates, and force and moments data from the load cells, but I do not have the rotation data from angular rate sensors. Of course, I could prescribe the sled and plate accelerations separately, but I want to see if the load cells in the simulations can measure the same force as in the tests with just one pulse prescribed.


ReplyQuote
Negative Volume
(@negativevolume)
Senior Member Admin
Joined: 2 years ago
Posts: 326
September 10, 2020 1:10 am  

@viky113 I see. So first off, I would start with the question of what is the ultimate goal of this setup? Is it to use with a future project like simulation a pulse with a ATD dummy model? Or is it purely to create a validated test sled setup? 

If the goal is to use this for a later simulation then it could be justified to apply separate accelerations to the load cells and sled, because you would be simply replicating what was seen physically in an effort to use for an ultimate goal. But like you mentioned, you don’t have the rotation data so the best you could do is linear acceleration.

The other option would be to just attach the load cells to the sled base with something like a 1D steel beam (if it was steel in the physical test). If you don’t have CAD, do you have rough measurements of these connections and their shape? That way you could mesh them and assign you material property.

i would choose either of these options over prescribing inertia since it is a unique situation.


ReplyQuote
viky113
(@viky113)
Rookie
Joined: 2 months ago
Posts: 6
September 10, 2020 1:30 am  

@negativevolume the ultimate goal for now is to validate the seat model using the seat back forces measured from the load cells. It would then be used with a human body model. But for now, it’s a seat-only simulation. I wanted to check if the plate attain the same accelerations as in the test when this pulse is applied. However, I don’t see that happening because of the simplicity in my model right now. I don’t have any material properties as of now. I’m simply assuming it’s steel. I have masses of the plates and the load cells.
I have two questions related to this -

1. If 1D steel beams are attached to the sled, how will inertia come into play? They would still travel with the sled right? 
2. I am trying to introduce some inertia by letting the big back plate strike the load cells from a distance. But when I use tied contacts, my forces and moments are not getting transferred. I just see the attachment and displacement. How do I deal with this?


ReplyQuote
Negative Volume
(@negativevolume)
Senior Member Admin
Joined: 2 years ago
Posts: 326
September 15, 2020 1:39 pm  

@viky113 1. If you connect the plates to the sled with beams or some other type of mesh then you would only apply the motion to the sled. The plates would then move according to the connection with the sled. So yes, they would still move together but they may have the slight delay that you are seeing physically. 

2. Only a select few tied contacts allow for force to transfer through the contact. Usually, these are tied contacts with the _beam_offset option. Below is an overview of this topic:

https://www.dynasupport.com/howtos/contact/tied-tied-offset-and-tiebreak-contacts


ReplyQuote
viky113
(@viky113)
Rookie
Joined: 2 months ago
Posts: 6
September 16, 2020 2:56 pm  

@negativevolume Thanks for your input. As of now, I am providing the same pulse to the plate behind the load cells by making it rigid and giving it local constraints about a coordinate system defined on the sled floor. That way the fixture behind the seatback has some rotation about the sled floor, while having the same pulse (see attached picture).

For the delay in acceleration between the floor and the plates, I am thinking about having something like a pretension cable behind the big plate, which will hold the fixture in place for 10 ms or until the accelerating force reaches a certain value, and then get cut off. Is there an easy way to model this?

1600264594-SledSetup.png

ReplyQuote
Negative Volume
(@negativevolume)
Senior Member Admin
Joined: 2 years ago
Posts: 326
September 23, 2020 3:45 am  

@viky113 If you are trying to validate the response of a physical setup, then I really think that the best way to approach this would be to attempt to model the physical mechanisms of the seat. Meaning, I wouldn't try to add anything to the simulation that wouldn't be in the actual physical test (like the pretension cable that you mentioned). I really think that your best course of action will be modeling the connecting structures between the sled base and the load plates. 


ReplyQuote
Share: