Dear Caesar Forum Members,

I am doing Time History analysis of an EXISTING system with following features:

- 32" UG 98 pipe
- 19 mm wt
- Carbon Steel
- Above ground
- Design T 80 °C
- Ambient T 21 °C
- Design P 46 barg
- Operating P 5 barg
- Code B31.3.

Because of ESD valve closure, surge pressure is generated. ESD valve takes about 20 seconds to be totally close. As per project scope, I have to analyze only the most critical portion of the pipeline. You can find a snapshot of the selected portion (along with length of the pipe legs, node numbers, coordinate system and wave direction) in the link here below:

http://img196.imageshack.us/img196/2818/ig9u.png

This portion has been selected as there are many horizontal direction change.

ESD valve is around 200 m after the last node, i.e 870.

First and last node, i.e. 10 and 870, are restrained with guide, limit stop and rest support. Another limit stop is located between node 430 and 460. Straight portions of the model have guides and rest supports every 20 m. Concerning the restraints on the loop, you can refer to below snapshot:

http://img31.imageshack.us/img31/1429/3dz9.png

As you can see, only rest support are present in the 2 loops.

I am applying the surge load on elbows 160, 200, 250, 280 and 430. Load directions are respectively +X, +Z, +X, -Z and +X. Such directions are as per SNFAIL paper (http://www.coade.com/Uploads/mechanical-engineering-news/jun94.pdf Page 8 to 13). Load is applied only in such nodes because they have the highest displacements as per static analysis results.

In the link below you can see time vs. pressure spectrum I got from Process engineer. This spectrum occurs exactly at the location that we are considering for the analysis. Max surge pressure is 46.9 barg @ 197 s.

http://img266.imageshack.us/img266/5378/te8t.png

In this other link you will find my input in Caesar.

http://img833.imageshack.us/img833/3585/f7n5.png

As you can see, I manipulated a bit the original spectrum. Here below what I modified and why:

- The load is normalized. Obviusly, In Caesar I will consider the actual load in N (which is calculated from pressure values, i.e. F = P*A).
- I took around 30 points (time/pressure) from the original spectrum. There is no need to input in Caesar more than 1000 points.
- The initial steady state load is now zero, so the peak of pressure now represent only the Delta P between operating pressure and surge pressure. This is because in the combined load case, system already consider the static load which come from the operating pressure. Hence the actual load I am considering for the dynamic analysis is the differential pressure load (46.9-5 = 41.9 barg).
- We made the initial and the final steady state phases shorter because for the sake of dynamic analysis there is no reason to keep them long.
- Time is now in ms because this is the unit which Caesar requests.

I am using this spectrum as Time History profile for all the 5 Force sets.

As the wave speed is very high and the surge profile takes around 20 s to reach the peak, 100% of differential pressure load is considered only for pipe legs which are long enough. As a consequence, 50% of differential pressure load is considered at elbow 200, 250 and 280. I think 50% is a very conservative value, considering the high speed of the wave and the short length of the the loop legs. Since elbows 160 and 430 have a long pipe leg ahead, 100% of the load is applied only at these points (and even here 100% is very conservative, these two legs are not long enough to accommodate the complete pressure raise).

Stiffness factor for friction I put 100.

Now, here comes the riddle. Static analysis is passing without any problem, max code stress ratio is around 60% (Sustained Case). On the other hand, dynamic analysis is miserably failing, 157% of code stress ratio at node 160.

How come such huge percentage????

As I mentioned at the beginning, pipeline system is already constructed, I am just doing a “Verification” job. Hence, huge modifications of the system are not feasible at this stage.

Thank you in advance for your valuable inputs.

Your approach appears sensible.
Are you relying on friction to provide horizontal restraint for these dynamic loads? Is that wise?
Your sustained stress ratio is (if I may...) "orthogonal" to your occasional stress ratio. Those weight loads work in the vertical while your dynamic loads are horizontal. There is little relation between them (other than they are added to evaluate the occasionals).

CAESAR II time history output identifies which modes cause the high stress. Work back from there.

By the way, I will be presenting a webinar on time history analysis TOMORROW at 10am Houston time. Maybe you can attend.

Reserve your Webinar seat now at:
https://www1.gotomeeting.com/register/217984225

Hello Dave,

Thank you for your suggestions and for your invitation. The presentation was very good.

Coming back to my case, yes i'm relying only on friction for the horizontal restraint on the horizontal loops. This is the actual condition at site.

Btw, I managed to have reasonable Code Stress Ratio by doing below two modifications:

- Stiffness factor for friction from 100 to 1000
- I am considering operating temperature instead of design temperature