shear flow?

[Mechaero2006]

Dear all,

in the file attached you will see a simple drawing of a part I want to model. The question is about the recommended process to model the boundary conditions and especially the shear flow seen. As stated in the file, everything is known (forces, moments at points 1 and 2), section dimensions and the shear flow values.

What I thought was to put constrains at points 1,2 of my model, insert the shear flow as the load and then check in the solution if the constraint forces and moments match the values given initially. But: How to define the given shear flow in my fea model? Please note that the FEA model created is made of 3D elements (not shells) and the analysis will be linear static (SOL 101 for the MSC NASTRAN users ).

Other approaches are welcome.

Thank you in advance for your assistance.

 

[Mechaero2006]

The following post is relevant:

http://www.eng-tips.com/viewthread.cfm?qid=355927

Note: A link of the present thread to the more appropriate: /simulation/MSC.Software: Patran Forum will be made.

 

[rb1957]

your attachment is very confusing ... everything is known (P and M) but "values are fictitious"?

are the reactions at 2 meant to show that the section is fully effective? you could show this using an RBE between the end of your model and point 2.

if you know all the applied loads and their internal reactions then you load your model that way and constrain a node, all 6 dof, to take out rigid body motion.

how's this relate to shear flow, except that this is one of your "known" loads ?

another day in paradise, or is paradise one day closer ?

 

[Mechaero2006]

Dear rb1957,

thank you for your reply. I think you're right, I should have paid more attention to what I want to do and then write my question. So let me try to explain:

the term "fictitious" for the loads, describes the values of the shear flow shown in the attachment. I meant that these are not the actual values that I have, but on a second thought since I am not saying anything about the other loads, I shouldn't have mentioned it.

second: forces and moments at points 1, and 2 are calculated loads applied on the part from attaching structures. Not constraint loads as initially said. So to wrap it up: what I really wanted was to investigate how to model the force/unit length load on my model, after putting my constraints (x,y,z) on the bottom surface (base) of the modeled part (see attachment). Do you think applying nodal forces with magnitude = (force/length) x (curvature length) is OK? Any other ideas?

Thank you in advance for your assistance and please accept my apologies for the confusion caused by my first post.

 

[rb1957]

i think it's enough to say you have loads at two points and shear flows applied to the boundary; maybe "I have a complete definition of loads and am just looking for internal stresses".

btw, point 1 looks "odd" ... point 2 looks sensible (as at the NA/centroid of the section).

it should be easy to apply loads at a point to mimic loads on a cross-section (by using an RBE or rigid element, to distribute the point loads onto the section). it should be easy to apply loads along a curve. You will need to restrain the model to take out the 6 dof of rigid body motion; your loads may balance perfectly but FE needs this to stop spinning in space. This also checks how well your loads balance (there should be no reaction forces at this constraint).

another day in paradise, or is paradise one day closer ?

 

[Mechaero2006]

Hello rb1957,

Sometimes the urge to over describe a problem makes you write "essays" instead of being precise. I promise that next time I'll be more spartan in my problem definitions. So:

i think it's enough to say you have loads at two points and shear flows applied to the boundary; maybe "I have a complete definition of loads and am just looking for internal stresses".

Correct. Couldn't say it better myself (as proved!)

it should be easy to apply loads at a point to mimic loads on a cross-section (by using an RBE or rigid element, to distribute the point loads onto the section)

That's what I've already done.

it should be easy to apply loads along a curve

Well, that was my problem from the beginning. I thought of forces acting tangential to the curve (magnitude as described before). However there is an option for "distributed load" which I think will do the job.

You will need to restrain the model to take out the 6 dof of rigid body motion;

Agree. I've done that as well.

Thank you very much for your assistance.

 

[rb1957]

depending on the FE program you should be able to load the line element or the edge of the surface with a load per inch, typical options a X lbs/in along the curve, or Y lbs distributed over the curve.

another day in paradise, or is paradise one day closer ?

 

[Mechaero2006]

Hello rb1957,

apologies for not replying sooner. Thank you for your last remark. As to my progress with this problem, I have created a number of successful runs, but I am having some serious thoughts about: a) My boundary conditions at the bottom of the part (constraints all over the bottom surface or only on the holes representing the attachment bolts) and b) the "notorious" shear flow applied on my line elements (that's 1D elements I suppose?). I guess there is much work to be done...

Kind regards.

 

[rb1957]

if you have section loads (like you show at point 2), create an RBE "spider" element between the modelled section nodes and the loaded node.

only you (and your advisor/mentor/lead) know if point 1 is correct (I suspect not).

How did you determine the shear flows ?

Why do you doubt tehm now ? because there's load being reacted at the "dummy" constraint ?

another day in paradise, or is paradise one day closer ?

 

[Mechaero2006]

Dear rb1957,

I really appreciate your interest. Yes, I have to agree that this load on point 1 is "strange" and my experience doesn't allow me to think other ways to apply it on the structure. What I have shown in terms of loads in the original file attached, comes from a report where some static calculations were made for that particular part. Application of loads on section (point 2) was more straightforward: I applied the loads using an RBE2 "spider" as already mentioned. As to the loads on point 1 I can only think of applying them on the respective edge (the structure is 3d, I am using again RBE2s) and constraint the model to see what happens.

I would be interesting to continue this conversation but I do not wish: a) to exploit your good will b) "Inflate" this thread with unnecessary information. Is there a possibility of a PM?

Thanks

PS: This is not by any means a university project or a project made in the context of a business contract of any kind.

 

[rb1957]

not really sure what the question is now ??

if your data says "the loads at point 1 are" then that's what you're got; dumb though it looks.

if the shear flows are defined, then so be it.

if things don't balance, then there's a problem (if they are meant to balance) !

another day in paradise, or is paradise one day closer ?

 

[Mechaero2006]

Dear rb1957,

first let me apologize for taking so many days to reply. I hadn't checked this thread and to be honest I didn't expect any follow ups. Anyway, let me briefly update: In terms of loads application, I decided to apply these wierd looking loads at point 1 as pressure loads (forces over a logically selected area of the part). Then applied BC: a) To the bottom surface (there are some bolts there holding the part attached to another part and b) to the section at point 2 (RBE2s from an indepenedent node to all nodes on the section, then constraints on the independent node).
With all the above, the solution is OK, internal stresses are fine, displacements look correct. Now for the hard part: The notorious shear flow: For every section of interest the force from shear flow was calculated as: F = q x (curve length) (units are in lbs) and applied on the surface nodes as a tangential load (remember the whole part is modelled with 3D elements). That was OK. BUT: Now when I run the problem (as above plus the forces from shear flow), the internal stresses go sky high. Instead of getting stress values in the order of e04 I get e06. Meaning that the part fails royally when it was not supposed to do so. I am breaking my head to find out why but I cannot.

Attached is a pic from the applied load (derived from q).

http://files.engineering.com/getfil...19e0-4254-b045-6b7593834ca0&file=untitled.pdf

 

[Mechaero2006]

Found it I think... It was a mistake with the area where the load derived from q was applied. More info soon...

Thank you rb1957 for your interest.