Modeling a Structure with Spring Nodes

[SubseaDeep]

Hi All,

I had some questions. In our company we provide subsea services and offshore oil & gas engineering. In this we use A-frames and Winches which are placed on vessels/ships/oil platform decks. But we do not place our A-frame and Winches directly on the deck or doubler pads. We use sub-frames (made from beams) between our equipment and the vessel deck. Please see attached file which will make it clear. The A-frame and Winch weigh around 25 and 40 tonnes respectively.


1. Suppose we are designing a Sub-frame (flat horizontal structure with beams only, no columns). Some experienced engineers in our company define/model the "tie-down points of the sub-frame" to the deck as spring connections. As pinned connections will give unrealistic large reactions in the analysis. So we define/model the nodes as springs. Is this correct?

2. The engineers start with modeling the springs with a stiffness of 50 K/in, and then increase it gradually step by step. Sometimes they stop at 100 K/in, or sometimes increase it more. My questions is when do we stop increasing the stiffness of the springs and confirm that a particular stiffness is correct, and the resultant reactions are correct?

3. We specify the same stiffness in all 3 directions. But in practical, the stiffness of the structure underneath is not same in all directions. How do we understand/work with this?

Any advice will be a great help. Please see attached file.

Thanks in advance.

SubseaDeep

 

[GregLocock]

Well, one thing i often do is model the attachment area as well as the part in question.

Static stiffness and stresses is quite tricky to get right precisiely because the BCs are so badly defined, whereas in dynamics we ususally test free-free which most people will get right eventually.

One thing you can do is to measure the static stiffness of the underlying structure with a hammer and an accelerometer, that is very useful, and you will pick up a much better feel for directions and so on.



Cheers

Greg Locock


New here? Try reading these, they might help FAQ731-376

http://eng-tips.com/market.cfm?

 

[weab]

Subsea,

First, I'm not certain if you are trying to analyze the entire deck of just your ties. I'll assume this is a large model with your subframe included.

Since deflection = PL/AE, then spring values for ties = P/defl = AE/L. Estimating from your photo, I come up with spring values in the 5000-10000 kips per inch, not 50-100. However, bolting through tubes as you show is likely not slip critical bolting. This probably allows for slippage as beams bend. I guess that your coworkers are trying to take the stiffness out of your subframes so that the main deck beams are not incorrectly affected.

Laterally speaking, I usually don't model the deck plate. The deck plate ain't goin' nowhere. I'd be satisfied if you released the reactions laterally except to keep the model stable. This won't work for lateral loading though. I need to sleep on this one.

 

[rb1957]

1) yes, i prefer to model supports as springs rather than as rigid constraints.

2) IMHO, this is probably overkill, over-refining the support stiffness, but trying different values lets you understand the sensivity (which i suspect isn't particularly strong, but then you guys have a lot of data to review and reflect on). Applying the same stiffness in each direction is a simplification alittle at odds with the previous over-definition; unless you have similar bolts connection in each direction. I'd've thought that you'd've have some reactions as shear on some bolts (or maybe friction) and others as tension ... shear stiffness is much easier to quantify than tension stiffness, friction is pretty much a guess.

are our answers similar to the ones your guys gave you ?

are you talking about the bolted connection to the sub-frame or the (what looks like) welded connection to the deck ? or both ?? I'd've thought welds were good in axial shear (along the length of the weld), ok in transverse shear (across the weld) and poor in tension ... so applying the same stiffness in all 3 directions at a weld would be "suspect".

 

[SubseaDeep]

Hi All,

I appreciate your answers. rb1957, the main thing which we are looking into is to find the deck reactions, due to the loads generated by the subframe.

1. Normally the subframe will be supported by some beams and columns which are underneath the deck. We design the subframe, and give loads of the subframe (at support nodes) to the vessel owner. We tell the vessel owner that these are the loads/reactions at particular points. So the vessel owner strengthens that beam/column underneath the deck.

2. We already have loads of the winch and A-frame which we input on the sub-frame. So this question is not about the bolting of the winch and A-frame to the sub-frame. Its about finding what loads the sub-frame will put on the deck or its underneath supports.

The sub-frame is modeled in RISA 3D. The nodes, at which the sub-frame is supported are modeled as spring. So the questions is how to accurately define springs so that we get the right loads generated by the sub-frame on the deck.

Thanks,

SubseaDeep

 

[rb1957]

ok, the subframe looks as though it's welded to the deck, though i'd've thought this odd on a ship and particularly as i'd've thought there'd be a bunch of tension on the joint to the deck.

Your support is redundant (if it were statically determinate you'd be able to calc the reactions by hand). As such stiffness plays a role in how the reactions are determined. In your case I'd suspect that each attachment is pretty much the same, I'd suspect a bunch of bolts in tension with the inplane reaction created by friction (I'd suspect the bolts would be in open holes). The bolts are going to be in tension, there's some VDI??? document that exhaustively analyzes bolts and would give you a good idea on the stiffness of the joint.

However it's probably enough, if all the supports are similar, to do a sensivity study and see how the reactions change. I'd expect to see a change when the stiffness of the support approaches the stiffness of the structure, but I'd expect the stiffness of the support to be reasonably higher than the structure so i wouldn't expect to see much change in the reactions.

I'm curious, what are your engineers telling you (or not) that makes you ask the question of strangers ?

 

[GregLocock]

Another thing you can quickly do is a sensitivity analysis. For instance if your reaction forces are unchanged when you change the restraint springs from say 25000 lb/in to 4 times that (you may need to fiddle with the values dpending on the order of magnitude of stiffness of your structures), then you can be pretty confident that the restraint spring stiffness is not of crucal importance.



Cheers

Greg Locock


New here? Try reading these, they might help FAQ731-376

http://eng-tips.com/market.cfm?

 

[kellnerp]

Looking at your subframe I would say that half the contribution to the stiffness of the connection comes from the location on the subframe which is attached to the deck. To illustrate this, the connection on the left is on the edge of a flange. The connection on the right is also on a flange edge but there are two stiffeners involved which make that connection point a whole lot stiffer than the left connection. Loads go towards the stiffer connection in my experience.

As far as spring stiffness goes the stiffness in the plane of the deck needs to be a lot stiffer than that perpendicular to the deck.

You could model the beam and see what kind of stiffness each edge connection gives.

Just my 2 cents.

TOP
CSWP, BSSE

http://www.engtran.com

http://www.niswug.org

"Node news is good news."