Can we control the order that CAESAR II analyses different loads?

We have a model of a system that includes a rupture disc with a large thrust load. When we run the stressing model without considering the rupture disc thrust load the piping is code compliant. However when we include the rupture disc thrust load the piping is no longer code compliant - but it is failing under the thermal load, not the rupture disc load.

What we believe is happening is that the rupture disc thrust load is being considered when the piping is cold, ie before the thermal load. The rupture disc thrust load is large, and hence the frictional component of this is also large.

We believe that because the rupture disc load is being considered before the thermal load, the thermal component is lower than the rupture disc friction component and hence the pipe is not able to expand thermally.

What order are the loads analysed? Can we contol this?



Regards

Say w/o the rupture disk thrust load the load case is :

1) W+P+T -> This is not a code case ( at least pre-2004 )

2)W+P ->Code case.

With rupture disk thrust load,

1) W+P+T+F -> Again not a code case.

2)W+P+F -> Code case or is it W+P as before ? I don't know what you have selected, so only you can answer.

W/o rupture disk load, EXP case =L1-L2 .

With rupture disk load, EXP =L1-L2 which may be (W+P+T+F) - (W+P) or (W+P+T+F )- (W+P+F) . For either case , EXP in the first case and the second case need not be same because of system non- linearity.

Definitely in your case the two EXP values are different , that is why you have failure in EXPANSION STRESS RANGE.

For analysis of a case ( static analysis) the program solves equation :

[P]= [K][D].

I feel that you are not very conversant with the equation above and what it basically stands for and how it is derived.
Once you understand the above equation, you will get the answer to your question yourself.

Regards

No you can't control the order the loads are applied. As Anindya states above, CAESAR II solves the finite element equation [K]{x} = {f}, where {f} is the load vector. The load vector is comprised of all your load components.

So if you have a load case such as W+P1+T1+F1, then {f} is formed with components from "W", and from "P1", and from "T1", and from "F1". Then the system of equations is solved and you obtain the displacement vector {x}. There is no first or last, all of the load components go into the load vector.

Now, if you want to analyze an operating case, such as W+P1+T1, and then from this (displaced) position, apply your thrust load "F1", then what you want to do is what we refer to as load stepping. CAESAR II doesn't do this, and I'm unaware of any software that does, outside of a full blown FEA package like ANSYS (or similar).

I might suggest this...

Your rupture disk event occurs from the hot position. It is not part of your expansion stress range unless there is some temperature change assoiated with the event. I wouldn't look at those stresses from (W+T+P+F)-(W+P+F).

Your issue is with friction, right? The larger normal loads at supports during the event "restrain" the thermal growth. But the pipe is already moved to its hot position before the event, before the normal loads increase. If you want to get a better look at the system response during the event, you could run a separate analysis where you first find out how your rupture disk loads distribute through the system as normal loads on supports and then fiddle with your coefficients of friction for the complete load set so that only the appropriate friction load is applied here.

There's no button to push. It'll all be custom tune-up for this one analysis.

But, then again, you might be fooling yourself with numbers. How do you know what coefficient of friction to use for these impact loads?

Thank you all for your replies.

Anindya, I don't think you fully understood my question. It is a difficult scenario to try and explain, but as always any comments/suggestions are much appreciated.

Dave, I agree with your statements and what you have suggested is pretty much the way we have gone about solving this particular case.

I have to say that I am surprised that the order that the individual load components are applied can not be controlled. Surely load stepping is required for non-linear analysis?

I did not think the principal of load superposition applied in a non linear analysis, and hence the starting point for each load case and the order the loads are applied is important.

Surely the results of the thermal load case for example will depend on the state of the supports (ie guide gaps) at the end of the gravity analysis? Some gaps may be opened, some closed.

Greetings all.

Can anyone throw any further light onto the issue I raised above re order of loads on a non linear system.

My understanding was a non linear system had a non-linear curve on a force versus displacement graph. I also was under the impression that for a non linear system, the starting point for each load case and hence the order the loads are applied was important.

Surely the results of the thermal load case for example will depend on the state of the supports (ie guide gaps) at the end of the gravity analysis? Some gaps may be opened, some closed.

If this in not the case I would like some further enlightenment on this issue if anyone can point me in the right direction.

Regards

In the "real world", you're right. However, in the world of "linear elastic analysis", this isn't going to happen.

Each load case starts from the neutral position, the position you defined in the input. All of the loads for a particular load case are inserted, together, into the laod vector {F}. Then the solution is performed to obtain the displacement vector {x}.

The only non-linearity involved is the iteration necessary to address gaps, friction, and one-way restraints. We call these "non-linear restraints", but we are not doing "load stepping" (where you apply one load and obtain a displaced position, then apply the next load, and so on).

Hello.

I often have to Look at BD systems.
Frequently, the issue of temperatures and pressures comes up.

For us, we do a fire case, which looks at the issue of ensuring the code stresses stay within range if the BD ruptures due to internal vessel pressure rise due to an external fire.

How does this apply to your problem?

We heat the whole system up (as per a fire)
And then let the BD go. (Just as it would)

But in the absence of fire, Any heat gain after the BD is considered to occour after the rupture event, and the cases are coded to suit.
In all cases i look for the worst possible senario.

For example, my latest Calc shows the highest stresses when the system is at -10c and under full vacuum due to bellows etc.

So for your problem, i would model the enviroment accuratly, ie determine the full range of thermal stresses before and after the BD ruptures, which
may mean:
1. Full temp range before BD/ Ambiant after,
2 Apply Rupture forces in this thermal state.
3. Asses thermal stress from heat transfer due to relief after bd failure.
4.(For analysis) we assume the pressure to be max
for all events (Worst case)

Good luck