I am modelling a pump discharge line that leaves the pump nozzle horizontally with a bend about 2 feet away and then has another horizontal run before going up into a pipe rack.
I have had difficulty reducing pump nozzle moments to acceptable levels but am able to get Caesar to produce acceptable moments by installing an anchored vertical dummy leg off this first bend. Putting an anchor so close to the pump is foreign to me, but if the results are below pump allowable levels do you see any problems?

Hi Mike,

Some thoughts:

The loadings are going somewhere, make sure you agree with the credibility of the resulting load path. If you have modeled the dummy leg with an "anchor" (six degrees of fixity) at its end, I would ask you how that could be done, how would it be built.

Be careful of the boundary conditions that you are describing in the model - especially at the dummy leg. A dummy leg can NEVER be an anchor, simply because it is too flexible. Typically, the supporting structure under the dummy leg will also have some flexibility.

Typical dummy legs rest upon some supporting structure and are free to move up (with expansion) and usually free to slide in 2 mutually perpendicular horizontal directions (with some amount of associated friction). As such the dummy leg would only be relieving the vertical (down, maybe weight) component of the original nozzle load. The rest of the loading would follow its original load path to the discharge nozzle. Also, be sure that you are not "pushing off" at the dummy leg location and lifting your pipe off the first support on the rack.

If you find that the loading that was previously being applied to the pump discharge nozzle is now going through the "dummy leg" and to the "ground anchor", something is very wrong from a practical "nuts and bolts" point of view. A (rigid) support should not be located too close to the nozzle or it will either force a vertical expansion force into the nozzle, create a fulcrum for the pipe to "see-saw" on, or both. You can use guides, line stops and limit stops to force the expansion thrusts back towards the rack where you may have more flexibility and more opportunity to locate proper supports.

Are you using the hot (more flexible) young's modulus for the operating case load? That would be appropriate for the pump loading analysis(albeit, not for stress calculations).

I am sure that you will get more opinions from the group.

Regards, John.

Bolts are probably the most common solution, although I have seen welded base plates from time to time. That's about as fixed as you can get.

Hi Ed and all,

Yep, that is the way we all felt about it for a lot of years. Then the NRC came along and took a look at the way we were modeling piping systems. They were uncomfortable and they voiced their discomfort by issuing 79-07 and 79-14 (they were especially worried that those "anchors" would not give us as much control under seismic loadings as we were claiming they would). Under those NRC directives, the plant owner was required to go into the existing plants and first, confirm the accuracy of the "as-built" drawings (making sure the boundary condition that was actually there was as it was modeled in the piping analyses) and second, checking all the "anchors" to see how much flexibility they really had. So, flexibility surveys were done in which known "unit loads" were applied at the "anchors" and the resulting displacements (translations and rotations) were accurately measured. That gave us a set of "spring coefficients" to be applied to the piping models where we originally had six degrees of fixity modeled. For example, a typical "anchor" would comprise a vertical pipe trunion welded to the bottom of the pipe at the trunnion's top and welded to a "base plate" at its bottom. The "as-built" surveys showed that most commonly the trunnions were of a diameter that was half that (or less) of the process pipe it supported (this to facilitate welding). The flexibility surveys showed that the trunion had significant flexibility, that the "base plate" had significant flexibility (the AISC base plate design methodology addresses this but we did not at the time) , and that each of the anchor-bolts had significant (non-uniform) flexibility (sometimes not all the anchor-bolts had been installed or they were sometimes loose). We have been referring to "base plates" as flex-plates ever since.

What we learned was first, quite often the execution of the design during construction was not faithful to the drawings (e.g., 2 "U-bolts" looped over a pipe and bolted to a C6x12.5 provided almost no rotational restraint), and second, many of the constructions that we called "anchors" were really a lot more flexible than we thought.

I think for many of us "olde-timers" those revelations lit a light bulb. We became more aware of how we were modeling non-nuclear piping designs and what the implications were. We needed to be more careful about how much of the loading (forces and moments) were actually diverted to the "structure" by the "anchor" and how much of it was still passed on down the line to the terminal equipment (pump, vessel, turbine....or branch connection). It is really very difficult to take piping MOMENT loadings into a structure by constructing an "anchor". The olde Kellogg book ("Design of Piping Systems") shows such a device (on page 244 for those who have the book) and it has been modeled by FEA by several "fun seekers" (I know, "get a life") though the years - and it has been found to be "good" but not really perfect in its function. The point is "it ain't easy to design such a contraption". The good news is that it is not often necessary to try to directly resist a moment - often by judicious location of guides and stops we can break the moment up into a "force-couple" and forces are easier to accommodate structurally.

So, I think that when a piping designer finds relief from excessive loadings upon a pump (or, vessel nozzle, or turbine, or....) by inserting an "anchor" close to the equipment, the "cure" is always suspect. Most of what we do to solve piping system loading problems at terminal equipment is a compromise. These designs include both positive and negative effects and we try to balance the good with the bad. Of course the good news is that with CAESAR II we can "cut and try" and iteratively solve the problem by moving guides, line stops and limit stops around (preferably well away from the strain sensitive equipment) to find the best arrangement for solving the problem. CAESAR II make this a lot less man-hour intensive.

I think that it is important to keep in mind that when we apparently cure a problem in a piping design by a seemingly simple "tweak", maybe we should look further to be sure we completely understand exactly how that "tweak" worked - and how it would be reproduced in actual construction. Just a thought.

I agree with Ed that the constructions that he describes are about as fixed as you can get (or expect in the real world). But I wanted to point out that the fixity provided by these constructions likely do not equal a CAESAR II “anchor”.

What do the rest of you "lurkers" think?

Best regards, John.

Hey guys,

Thanks John for the background info.

This subject is one that I believe is "implied" by many engineers. When I model a restraint, say an anchor at a nozzle or a vertical support in a rack, I know for a fact that it is not infinitely rigid. There will be some "flex" from the vessel wall or the beam. I think we can all agree on this. How many times do you actually model in the beam flexibility? Or that of the vessel wall? I'd venture to say less than 10% of the time.

However, when you give piping loads to a structural or vessel engineer for their analysis and/or approval, we typically assume these restraints to be "infinitely rigid", even though we know this is not the case. Typically, I only go to the trouble to model restraint flexibilities when the restraint loads are borderline using the infinitely rigid model, and then usually as a last resort.

The same goes for trunnions and dummy legs. I often model in the dummy leg diameter (if its pipe) and length (without temperature or pressure) to get an idea of the flexibility of the support. I've also been known to use the structural modeller to model beam trunnions, which can get tedious (watch the orientations). Local stress issues and SIFs still need to be considered.

After having said all that, (looks like one of Mr. Breen's postings) I guess the point is: there are many things we do in our analysis that are assumed or implied, but we need to be aware of the implications.

Dear Mr.Mike and all other friends,

Nozzle loads on equipments should be reduced as much as possible so that qualification of equipment by the vendor is possible.Generally all clients insist that there should not be any load passed on to the equipment at all.But,I am not in favor of modelling dummy leg below the first bend as an anchor.

It is very complicated to understand the local stresses that develop in dummy leg.I have not come across any "ASME Code case for calculating the local stresses due to dummy leg on a elbow".This is because the any program will calculate only the pipe stresses but not local stresses due to external attachment like lugs,pipes etc.,

It is difficult to suggest any solution without having the isometric and the other input parameters.

I can suggest that by altering the location of nearby valves or any concentrated loads we can bring down the loads on equipment.I have also faced similar problem in several problems.However I could solve most of them by simple techniques

C srinivas

Well, from this and other nightmare stress stories that I've heard, I can understand why nuclear plants are so expensive to build.

That said, I would be greatly concerned if someone presented a base support that was less than half the OD of the run pipe and wanted to anchor the base plate. I've never seen such a configuration satisfy local stress requirements. Even with the Kellogg method's conservatism, calculated stresses can easily be twice the allowable.

As well, if you anchor a base plate on a support in front of a pump nozzle, all you've done is guarantee that you're going to overload the nozzle when the line heats up. An anchored baseplate in front of a pump nozzle is about the worst thing I can think of (unless you find a client who's willing to allow an expansion joint on the nozzle. Let me know if such a client actually exists, 'cause I've never met him).

Trying to restrain the baseplate in front of a pump is pretty much just asking for trouble. If you "anchor" the base plate, even with the flexibilities of the support pipe and plate, you still have a lot of stiffness relative to the typical pump nozzle allowables. If you try to put a guide in (say, to try and restrain the growth between a pair of pumps), your model may tell you the guide is taking the load, but it will probably never actually move enough to engage the guide, so your load is still going to the nozzle.

As for the no loads on equipment business, I can't help but laugh when someone suggests such a thing. Even with an expansion joint, there's no way to have a piece of equipment immune from external piping loads. I suspect most companies are at the point of having a standard load table based on nozzle sizes for vendors to satisfy when making their designs. This gives vessel/equipment/exchanger engineers and the pipe stress engineers a threshold below which they don't have to deal with each other and is hopefully robust enough to allow for a piping design that is not excessively flexible.