Above Ground Pipeline Stress Analysis

[Headache2013]

Dear Specialists;

I'm asked to evaluate stresses and anchor force(at anchor block) for a pipeline laid on the ground (not buried), but I don't know how to model it using C2 or Autopipe.
For buried pipeline, there are a lot of literature explaining how to model it and how to simulate soil effect on the pipeline. However for AG lines, I couldn't find any literature.
I made some simulation using both Autopipe and Caesar II softwares and the results are abnormal.
Pipe: 6", Mat'l: API 5L X60, Design T°= 105 °C, Inst T°:15°C, Design Press: 45 BarG. Applicable code is ASME B31.8.
Pipeline shape: "L" shape, with large cold bend (40D) at the corner. An anchor block at the end of small leg of the "L". The long leg of the "L" is terminated at virtual anchor point.
I simulated two cases:
a) I assumed minimum soil cover (50mm). Stresses are Ok but the anchor forces ranges from 13KN (Autopipe) to 888KN (Caesar II).
b) I manually calculated soil springs Ki and Pi. For longitudinal and transversal, I assumed Pi as a friction force (Pipe + content weight) X friction factor (0.5). For downward, Pi and Ki calculated using ALA method.
The results are strange. Anchor forces range from 4KN (Autopipe) and 960 KN (Caesar II).
For your information, with Caesar II I changed the default post yield stiffness (K2=1) to K2=0 , otherwise the system doesn’t converge.
Something is wrong with my modeling. Could you advise the best modeling approach.

Best regards

 

[BigInch]

Do it by hand. A basic L configuration isn't that difficult. See B31.1 Appendix.

Independent events are seldomly independent.

 

[LittleInch]

You might be better for analysis and in practice to actually install sleepers on your A/G pipeline , but then you have the issue of steel / steel or concrete friction.

given that it is above gorund, I can't see your virtual anchor being anything other than hundreds of meters away, but a 6" pipe will snake all over the place in reality and hence the forces on your end point should really be quite low.

If it's on the surface, why have you assumed 50mm soil cover? The real problem will be modelling how the pipe digs into the soil (sand?) as it moves about.

True surface laid is not that common anymore so isn't really supported as a stress analysis issue and due to the small lines (usually flowlines) and inherent flexibility of them the anchor forces tend to be quite low. You might want to end this post and try in the CAESAR II forum.

My motto: Learn something new every day

Also: There's usually a good reason why everyone does it that way

 

[BigInch]

and what are the lengths of each leg?



Independent events are seldomly independent.

 

[BigInch]

If above ground, with no guides, the axial force cannot be greater than the buckling load of the longest leg.

Independent events are seldomly independent.

 

[Headache2013]

Thank you BigInch;
I think your guideline to limit the axial force based on buckling force is good idea. Would you please to give more details such as:
a) Do you mean lateral or upheaval buckling?
b) Incase of lateral buckling, do you know the formula to apply to get that force?

Best Regards

 

[LittleInch]

Your op is asking about modelling. Just model it as normal piping with a lot of infinite length or say 10m wide simple supports with sliding friction which simulates the soil pipe interaction. You could put an anchor point at some distance from the bend and then play with the friction numbers to see what is the most sensitive or realistic number and the program will tell you if you have buckled, got excessive lateral deflection or have lifted off the supports, i.e upheavel buckling.

At 105 C will certainly move around a fair bit....

So long as your short leg is a reasonable distance, say 20m or so, your anchor loads shouldn't be that high.

I'm actually quite surprised that a gas line isn't buried. Nearly all countries I know require gas lines to be buried, even if they let liquid lines be surface run and it's normally a very good idea for damage protection if nothing else.

Post a schematic or the issue which is causing you problems and you may get some better help.

My motto: Learn something new every day

Also: There's usually a good reason why everyone does it that way

 

[Headache2013]

Dear LittleInch;
a) The pipeline is actually carying oil hydrocarbon, but in the future the GOR will increase that is why the Owner decided B31.8 but not buried. This is Owner decision.
b) the short leg is 15m long and the long leg is 5000m ( the virtual anchor length as computed by Caesar is 3.9 KM).
c) you said: "Just model it as normal piping with a lot of infinite length or say 10m wide simple supports ...etc". Using which software can we do this? Autopipe and Caesar provide supporting "points" only, not wide supports.
d) You said "...and then play with the friction numbers to see what is the most sensitive or realistic number": this is the question: what is this relaistic number?

Thank you

 

[BigInch]

You're going to need more flexibility than what that 15 meter short leg will give you.
Your axial stress near that leg will be [σ] = 1/2 [α] * [Δ]T *E and Force = [σ]*A
Probably far too much moment on the 15 meter short leg.

Independent events are seldomly independent.

 

[LittleInch]

If you look at the scope of 31.4 and 31.8 neither cover two / three phase fluid exactly for that reason, i.e. when is it a gassy oil line and when is it a wet gas line. However you have what you have . I am sure you can tell Caesar and autopilot what type of point it is, sliding support, guided support, friction factor etc

It rather sounds to me like you don't know a great deal about these programs from the questions you are asking so rather than going too and fro on this post I suggest you look for some training.

I agree with BI, 15m looks a bit short, on reflection my 20m in the post before should be more like 40 to 50 m. At that point the short leg will accept much more expansion without creating a big force and moment on your end flange.

Stress analysis is like hydraulic analysis. It's easy to get a program and do some simple analysis but much more difficult to know how to change things or what the inputs should be and very easy to get results which are meaningless. The big difference you have between different models must be telling you that.

My motto: Learn something new every day

Also: There's usually a good reason why everyone does it that way

 

[BigInch]

I'm still not sure if it's buried or not, but if buried, one end takes 3.9 km to reach a virtual anchor, so does the other and that's nearly 8 km, but the length is only 5. That might be some of the problem.

Independent events are seldomly independent.

 

[Headache2013]

Little Inch;

a) I don't need training on Caesar II or Autopipe. I just misunderstood your adivse to provide "10m wide simple supports ".
b) For your information, this is the first approach modeling that I did one month ago. I provided V Stop support (without guide) each 1m with friction set to 0.5 (as soil-steel). The results were rejected by Owner beacause he said "7600 Newtons" is very low force.
c) Additional information: Actually the configuration of our pipeline is not "L" shape. The actual configuation is attached herein: 03 snapshots in the pdf file(overall confi, departure short leg before large cold bend 15m and arrival short leg after large cold bend 20m). In my post I said "L" shape just to simplfy, because my headache is the modeling approach and the big difference between Autopipe and Caesar II results.

Best Regards

 

[LittleInch]

Looking at your layout, the forces are probably not very big as the Z shape at the A48 end is quite a good way to relieve stress and movement as is the box design allowing the pipe to freely flex. You don't say what wall thickness your pipe is, but for a 6", X60 with a DP of 45 barg, it's probably pretty thin and hence can't actually generate a lot of force.

Why do you get a discrepancy - very difficult to tell from a distance, but sounds like the Caeasar model is locking up somewhere close to the bends / anchors points. Check the suppots aren't accidentally fixed and check all the outputs between the two including movement and forces at other nodes to try and see where the two are different. Both use similar techniques and formulae so differences of this magnitude are usually input errors.

My motto: Learn something new every day

Also: There's usually a good reason why everyone does it that way

 

[Headache2013]

LittleInch;
Thank you very much for your swift answer.
a) wall thickness is 9.53mm
b) No input error in my model, I checked it handred times. Moreover, for 1000mm burial depth, Caesar II and Autopipe give almost same forces (2% difference only at both extremities of the model: nodes A00 and A48).
The two softwares give different results only incase of 50mm burial depth simulation or user defined P1 & K1 simulation(based fricton force and downward bearing capacity).
As I said in my first post, for both cases above, Caesar doesnt converge with defaut post yield stiffness (K2=1) but Autopie converge with its default K2. I think the solution of this headache turns around this K2 parameter, but I can't justify or solve it.
c) for 1000mm burial depth, Caesar converges with default K2 value and give same results with Autopipe (as I said in above).

Best Regards

 

[Headache2013]

Dear BigInch;
I'm confused by your statment: "I'm still not sure if it's buried or not, but if buried, one end takes 3.9 km to reach a virtual anchor, so does the other and that's nearly 8 km, but the length is only 5. That might be some of the problem."
I'm confused because it seems we have different understand of virtual anchor length. In my opition, if the virtual anchor length is 3.9km, I can terminate my model (long leg) at 3.9 km or 5km or 10km, the results(stresses and forces) impact on the short leg will not change . Don't you think so?

Thks

 

[LittleInch]

I think what is confusing the inches is this 50mm burial. Can you sketch a cross section to show what you mean. I think the programs will have a lot of difficulty if it thinks a 6" pipeline has 50mm cover?? or do you mean the pipe base is 50mm below the ground level. If I'm confused, then I can understand why the program might be. 50mm of cover is very strange, so I'm not surprised the programs are having difficulty.

If it is 50mm cover,then I think in reality any answer you get is meaningless as the simualtions just fall over as their correlations and calcualtions needs more cover depth and soil pressures to work.

Also I have seen some long virtual anchor lenghts, but 3.9km for a 6" pipe is huge. If the line was perfectly straight then the expansion at the end point would also be huge, but for a surface laid 6" just doesn't seem right. Your original post intimated a 5km long straight pipeline. Hence the virtual anchor would go for 3.9 km from both end points. 3.9 + 3.9 does not equal 5.

My motto: Learn something new every day

Also: There's usually a good reason why everyone does it that way

 

[Headache2013]

LittleInch;

Thank you for your answer.
a) 50mm burial means 50mm soil cover (50mm from top of pipe to ground level). As you may know, Caesar II requires positive and non zero burial depth, that is why I assumed 50mm as soil cover just to simulate the soil effect on the pipe. I also simulated the case of 1mm soil cover. The problem is same, C2 doesn't want to converge.
b) As I explained in the begining of this post, my pipeline is not buried , it is laid on the ground. Since I don't know how to model it, I supposed that simulation based on minimum cover (50mm) is one good option to get soil interaction soil-pipline. I hope there is no more confusion now.

c) Regardinh the virtual anchor length, I'd like to exaplain as follows: 3.9km is the virtual anchor length computed manually based on 50mm cover. Manual calculation, Caesar & Autopipe computed value are same and no doubt on this.
In my model, I set the long leg of "L" shape 5km, because in my understanding , any straight bigger than 3.9 km hase same effect (stress and forces) on the system. in other words, stress and forces on the system having the long leg 3.9km or 5km or 99km is same since the pipeline beyond 3.9km is fully restrained.

Regards

 

[LittleInch]

Ok, I think we're going round in circles a bit, but it is clear to me that the 50mm cover bit is what is causing your programs to go loopy. Try varying the depth from 1000mm cover down in 100mm stages to find out when your program falls over. This will give you the limit of its ability to converge and you might also see a trend in the anchor lenght and anchor force.

3.9km may be what is calcualted, but is not feasible in the real world with a 6" pipe - more like 390m. Your 5km leg with 3.9km virtual anchor, doesn't work because unless the length before an end or bend is more than twice the virtual anchor leg, you will get forces and movements coming into your 3.9 km length from the other direction. This could also be causing your program some issues.

Caesar II will analyse A/G piping very well so just analyse it using the sleeper method, maybe even with sleepers every 500mm. In your previous post you said it was a V stop support? What exactly is that? I looked it up and it doesn't look lie a simple sliding support (bar or rod type), which is what you need.

My motto: Learn something new every day

Also: There's usually a good reason why everyone does it that way

 

[Headache2013]

Thank you LittleInch;

Regarding V-Stop meaning, I'd like to say: Vertical Stop (like sleeper). No provision of guide to let the pipe displace largely.

 

[BigInch]

It will buckle upward. There is no weight holding it down.
Just looking at the "buried" length. You have a moving pipeline in the 0 to 3.9 km length. You also have a moving pipeline in the remaining length. There is no virtual anchor. The pipe is essentially anchored in the mid point (due to symmetry) and growth occurs from there towards each end.

Independent events are seldomly independent.