OpenSees analysis

Forum for OpenSees users to post questions, comments, etc. on the use of the OpenSees interpreter, OpenSees.exe

Moderators: silvia, selimgunay, Moderators

Post Reply
Ashok
Posts: 50
Joined: Tue Feb 19, 2019 8:59 pm
Location: IITM

OpenSees analysis

Post by Ashok » Tue Aug 25, 2020 4:01 am

Dear sir,
This may be a weird question but kindly answer this.
Is it possible to include user defined member stiffness matrix into OpenSees? For example, taking a simple 2D frame, in a FEM software we generally discretize the
frame into elements (beam-column elements) which have their own shape functions...., we assemble them...
But let us say, I have my stiffness matrix already ready defined analytically i.e. like our Matrix Structural Analysis. [4EI/L 6EI/L^2 12EI/L^3]...
Can I include such a member in OpenSees. i.e. can I prevent such a member from going through the regular discretization -> shape function -> strain - displacement -> Element stiffness matrix -> global stiffness matrix assembly.. Can I directly get to Element stiffness matrix (determined by some other means). Is it possible?

Thanks in advance.

mhscott
Posts: 872
Joined: Tue Jul 06, 2004 3:38 pm
Location: Corvallis, Oregon USA
Contact:

Re: OpenSees analysis

Post by mhscott » Tue Aug 25, 2020 6:04 am

There's no direct way I can think of to do this. You could look into OpenFresco where your element is an experimental element, but that might get too complicated for what you want to do.

selimgunay
Posts: 913
Joined: Mon Sep 09, 2013 8:50 pm
Location: University of California, Berkeley

Re: OpenSees analysis

Post by selimgunay » Tue Aug 25, 2020 11:54 am

The elasticBeamColumn element works in the way that you mentioned, you don't need the shape functions or displacements as the stiffness matrix is based on the stiffness coefficients. It is a local stiffness matrix and is transformed to global though. You can use that as an example.

mhscott
Posts: 872
Joined: Tue Jul 06, 2004 3:38 pm
Location: Corvallis, Oregon USA
Contact:

Re: OpenSees analysis

Post by mhscott » Tue Aug 25, 2020 12:25 pm

The elasticBeamColumn still does transformations from basic to local to global as Selim mentioned. If you want an element that directly formulates the global 6x6 or 12x12 (for 3D) stiffness matrix and load vectors, you'll have to write a new element.

Ashok
Posts: 50
Joined: Tue Feb 19, 2019 8:59 pm
Location: IITM

Re: OpenSees analysis

Post by Ashok » Tue Aug 25, 2020 11:32 pm

Thanks for your response.
I will try to describe my formulation.

I have a Cold Formed Steel wall-panel frame (an assembly of 4 "pin jointed" beam-column elements in a rectangle forming a mechanism). Now, the wall-panel has a sheathing board which is connected to the frame through fasteners. When a lateral load is applied, these fasteners give the resistance. Now I have a displacement based formulation in which for every lateral displacement of the top of the frame, I can calculate the force developed by the fasteners "analytically", which in turn gives me the resistance developed in the wall panel. There is no requirement of finite elements at all. Now, when this frame is placed in the building model, and analysis is carried using a displacement controlled analysis, it can give the resistance offered by the wall panel at a particular location in the building. Like this we can place as many wall panels as we want, which give the resistance without using finite element at all for the resistance calculation step, so the computation will be very simplified.
The stiffness of the wall panel changes non-linearly. This can also be captured in the analytical formulation.
In summary-> I have an '"element'" which takes lateral displacement as input and gives lateral resistance as output. Inside its an analytical framework.
No, discretisation, no shape function, no quardrature,.... Directly gives resistance vector.
Can this be coded as an element.

mhscott
Posts: 872
Joined: Tue Jul 06, 2004 3:38 pm
Location: Corvallis, Oregon USA
Contact:

Re: OpenSees analysis

Post by mhscott » Wed Aug 26, 2020 5:41 am

If I understand correctly, perhaps you can use zeroLength or twoNodeLink elements for the panel force-deformation.

Ashok
Posts: 50
Joined: Tue Feb 19, 2019 8:59 pm
Location: IITM

Re: OpenSees analysis

Post by Ashok » Thu Aug 27, 2020 3:28 am

Yeah, you are right. I could directly give 2 node link element with multilinear material property. But how do i differentiate if the multilinear material takes stress - strain or force - displacement? In documentation it's given it can be anything !

Moreover the reason I want to include the whole element rather than a simple 2 node link is: to be able get a more comprehensive view of states of elements of the wall panel at different global structure displacements.
Presently the way is, I have to do post processing, by recording the states of the link and then assessing the state of the elements corresponding to states of the link (2 node link).
It is analogous to equivalent lumped mass system analysis Vs 3-D building analysis.
Hope you understand.

Thank you.

mhscott
Posts: 872
Joined: Tue Jul 06, 2004 3:38 pm
Location: Corvallis, Oregon USA
Contact:

Re: OpenSees analysis

Post by mhscott » Thu Aug 27, 2020 5:59 am

It doesn't matter, the element interprets the uniaxialMaterials it uses as force-deformations. All uniaxialMaterials are just functions, they do not have inherent stress-strain or force-deformation meaning.

I recommend you start out with twoNodeLink and see if you can capture the behavior. After that you can move on to more refined models.

Ashok
Posts: 50
Joined: Tue Feb 19, 2019 8:59 pm
Location: IITM

Re: OpenSees analysis

Post by Ashok » Thu Aug 27, 2020 7:50 am

Ok. Thank you. I will do that.

Post Reply