Seismic/Transient loading on anchor

[CRG]

I wanted to bring this thread378-107915 back to see if anyone had any further comments. Basically at issue is a 2500 meter straight run of pipe (assume expansion joints for thermal expansion) in a tunnel (no room for expansion loops) with a valve closed at each end. There are multiple anchors on the straight run of pipe; however, during a seismic event that is in the axial direction, not all of the anchors may see a proportional load to accelerate the fluid in the pipe. The last anchor in the straight run adjacent to the closed valve would carry most of the load to accelerate the fluid in the pipe. Rebis AutoPipe incorrectly divides up the fluid acceleration evenly based on the distance between the anchors. This is not correct. Does Caesar handle this transient case correctly? What is the best approach to determine the potential loading on the end anchors?

One could assume that the end anchor carries the load to accelerate the entire 2500 meter fluid load however; I do not believe this is the case. There should be some attenuation from the pipe expansion, friction, compressibility of the fluid, etc. Obviously, if the line was an infinite length, there would not be an infinite load on the anchor. Any new ideas?

 

[chicopee]

My first question is ,are the anchors design to let the pipe slip thru during expansions and contractions? Do you have expansion joints on the pipe? Also what is the net reaction forces on the bends from the movement slurry during non-seismic events and during seimic events in which case the inertia forces from the pipe and slurry should be considered.

Finally have you had any problems since 2004 during you initial inquiry with anchors and leaking joints

 

[BigInch]

What makes you think that acceleration forces should not be distributed along the pipe? Isn't that where the fluid received the forces that were necessary to accelerate it in the first place? I would think that Autopipe is correct because the fluid acceleration along the axial length would have been started by shear forces coming first from the walls of the pipe to the fluid film adjacent to the wall, then from the film inwards towards the next fluid particle towards the center of the pipe and so on. Once accelerated, flow resistance would tend to distribute fluid shear forces immediately back to the pipe wall where they could be resisted by the first rigid anchor they 100% rigid anchor it encountered. The only fluid force you would have to resist at a concentrated point would be those from a change of momentum of the moving fluid as it went around a pipe bend, if there was one. The resistance of that load would be distributed to all anchors in proportion to the axial flexibility of the pipe and the spring constant of each anchor in the axial direction.

An axial load anywhere along the pipe must be distributed to each anchor according to the axial flexibility of the pipe, L/A/E, between the load and each anchor minus the movement of each anchor x the anchor's spring constant.


Going the Big Inch!

http://virtualpipeline.spaces.msn.com

 

[CRG]

BigInch, thank you for your reply.

I may have over complicated the problem by not accepting/agreeing/understanding the industry standard for dividing up the acceleration of the fluid in proportion to the lengths between the anchors. This usually works well because what are the odds of having pipeline of significant length that is straight and seismic acceleration that is aligned with this perfectly straight pipe. Slim to none. For me this is more of an understanding of the physics of what takes place. What is the difference between (1) the pipe being quickly accelerated by seismic motion when the fluid is stationary and (2) fluid that is flowing coming to a sudden stop when a valve is quickly closed. In both cases the fluid is undergoing acceleration, one case would increase the speed the other would decrease the speed. In both cases the fluid is being quickly accelerated. If one was to look at the peak velocity of the pipe and the rate that it accelerated, there should be a correlation to moving fluid and the rate that a valve shuts on a pipeline causing a water hammer.

 

[BigInch]

I think you're getting it now.

I specialize in transient analysis of petroleum pipelines and do a lot of start-up and shut-downs and when you look at the simulations of fluid flow in a pipeline and how they are started and stopped, it makes little difference if the motive forces are generated by a pump, a closing valve a spinning can with water flowing out from a hole in the side. Energy->shear forces->pressure->acceleration->steady state->deceleration->static condition. In one case, energy provided by fluid contact with an impeller, the other by a moving pipe wall.

As I see it, the frictional forces at the pipe wall are all transferred to the pipe and become axial stress in the cross section, perhaps tension on one side of an infitesimal length of pipe and compression on the other side, depending on the local "fixity" of the pipe, so that even without anchors, fluid loads can usually be transferred by the wall of the pipe through the coating and into the soil by cohesive forces, or via tension or compression (or both) in the pipe to nearby bend location where the soil finally acts as a virtual anchor sufficient to resist all and no relative movement passes that point.

Consider only the vertical plane of a pipe running along undulating terrain. Even in extreme cases, at severe overbends for example, where there are two underbends (acting as anchors in the vertical plane nearby, because you can't push them down any more into the dirt) nearby, may force the pipe to rise out of the soil at the overbend (especially if the product is HOT), but the system still manages to come to rest in a static condition. It must be that all vertical forces, pipe, fluid and soil, are obviously carried by the pipe wall back to the underbends and transferred to the dirt, and thereby are totally absorbed by the dirt, more or less locally, because it did it again a couple of miles away. So, I take what my eyes have seen as proof that it can happen, but I ran the numbers just to make sure.

Now for example, if I have D-12 tugging (along the axial direction) on a cable attached to a branch-off from a section of mainline pipe between two anchors, and the branch is 66 feet from one anchor and 33 feet from the other, I but 66% of the load on the anchor 33 feet away from the branch (in tension) and 33% (in compression, or visa versa depending on direction) of the load on the anchor 66 feet away (assuming both anchors have the same spring constant) and I'm usually correct as to which anchor loses first. When I'm not correct, is usually because the pipe segment on the opposite side buckled in compression. So, best advice is, keep thinking about it, but don't get so theoretical that you forget practicality. Always start with a simple approach and add complications only when necessary to get a more exact solution, and then, only if the final solution is worth the effort it takes to arrive at it.



Going the Big Inch!

http://virtualpipeline.spaces.msn.com

 

[BigInch]

Next... Poisson axial shortening.

Going the Big Inch!

http://virtualpipeline.spaces.msn.com