Second of Two Parts
To view all of the articles in the Thermal Stress Affects Pipe Forces on Pump series, click here.
Two readers responded to “Thermal Stress Affects Pipe Forces on Pumps,” from the January 2012 issue of Pumps & Systems. The conversation was interesting and worth reviewing this month.
I read your article, “Thermal Stress Affects Pipe Forces on Pumps,” in Pumps & Systems, January 2012. I am not an engineer, just an “old rubber hose salesman.” Maybe I am simplifying the situation, but wouldn’t the logical answer for the thermal forces encountered with the expanding and contracting pipe be the use of an expansion joint…either rubber or stainless steel? We recommend expansion joints for solutions to such situations on a daily basis.
The last spools shown in Figure 2 could very well be replaced with expansion joints to handle the pipe growth and contraction. Thank you very much, Dr. Nelik. I look forward to reading articles authored by you in the future.
With regard to your calculation of axial load on the 2-inch long pipe, the force will not be 60,000 pounds because the calculation assumes that the pipe is not allowed to expand, or each end is infinitely rigid. Obviously, there will be some expansion at both ends of the pipe.
Lev Nelik responds:
Lyn, the pipe is either welded to the tank or bolted to it. It is also bolted to the pump inlet flange. How can it expand?
Lyn Greenhill responds:
If you conducted a finite element analysis (FEA) of the pipe and included a representative section of the pump and tank and constrained the pump and tank as per the drawing, you would find that the pump would be pushed to the left, and the pipe joint at the tank would deform into the tank. The stress in the pipe would not be all that high. However, if you anchored the ends of the pipe, say by bolting the flanges into thick, rigid plates, then the pipe would see the kind of compressive stress and internal load that you calculated. The other comment I guess I could make about your example is that you assumed that the pipe would be the same temperature as the water. Of course, there is the thermal mass of the system to consider, which probably means that the pipe would see nothing close to the water temp.
The bottom line to any pipe installation, whether to a pump, turbine, compressor, etc., is if you do not have expansion joints, you will get pipe strain and/or cracked pipes. I am working on a steam turbine that moves 0.4 inch from cold to hot due to pressure and temperature. Large diameter, hot pipes can generate enormous loads. That is why you see all the pipes in a power plant hanging from springs.
Lev Nelik responds:
You properly mentioned two mitigating effects:
Non-rigid constraints
Temperature of the pipe lower than water
Both are true. However:
Movement of the pipe at the tank will not be too significant because tanks are usually much more rigid than the connecting pipe effects. Perhaps the FEA would show about a 10 percent reduction, not much more. On the other side (at the pump), your comment is much more applicable. Indeed, most pumps are “softer” to resist the movement of the expanding pipe. That, however, is also a point of the article, although not stated explicitly. The good aspect of load reduction due to pump movement is completely negated by the bigger problem of the pump movement causing misalignment and, with it, leaky seals, worn or broken couplings and overloaded bearings.
Assume the pipe temperature is half that of the water, so that the 60,000 pounds is only 30,000 pounds, still a very significant load.
However, there is still a much more dramatic, but fortunately helpful, mitigating effect that you have not yet touched on. Take a closer look at the picture in the article. It has a clue.
Lyn Greenhill responds:
I have looked at the picture, and I do not see anything obvious that would significantly change the pipe restraints. The pump volute appears to be firmly anchored to the baseplate and is attached to a vertical riser. That would tend to serve as an anchoring point.
Quite frankly, if I were looking at this application for pipe strain and knew that the tank only held water that had some sun exposure, I would not be concerned about thermal expansion and load on the pump inlet. I would be much more concerned about the vertical, offset load on the volute from the discharge piping, particularly the isolation and check valves. Those things are heavy and need to be supported.
Lev Nelik responds:
I will take this opportunity to respond to two readers of the article—Lyn and Rand (the first email)—as their feedback touches on a similar aspect.
Lyn, the third missing aspect is gaskets. If the pipe were entirely continual, the high stresses would be as high—or close to as high, if considering Lyn’s point on the mitigating effects of the (somewhat) non-pinned end supports—as the stresses shown in the article. This is because the nearly-rigid supports would not allow the movement, and it will all be absorbed into stress.
However, the good news is that the pipe is allowed room to move. The in-between flanges/gaskets are not infinitely rigid and, even being stressed by the bolts, are still much more flexible to allow the expansion to be absorbed within a bit more compression of these gaskets. For a typical 0.06-inch gasket, the final compressed thickness is likely 0.04 inch, and that still allows some small additional (thermally induced) compression by the “moving” pipe—without the end restraints really experiencing it.
That is why you intuitively—from your experience—said that you “would not worry about it.” True, such installations are common and are everywhere, without problems. However, if the length of the piping is not a few feet (as in the article) but hundreds of feet (a pipeline) with rigid (or mostly rigid) end supports, then it is a different story. In such cases, much more attention would need to be paid to the expansion loops, pipe supports and, as Rand stated in his email, some sort of expansion joints.
Thank you again for an opportunity for a lively discussion. Good luck to you in your work.
Lyn Greenhill responds:
It would be interesting to conduct a complete analysis on the joints in this system to examine the bolt preload, gasket crush, thermal expansion and load relaxation in the bolts. I remember going through this in Shigley or other similar texts in college. Perhaps I will do that in the next few weeks just for my own interest.
Of course, intuition is sometimes backed up with experience. In the pump installations I have worked on, most of the short pipe runs do not use expansion joints.
Some longer ones, with large diameter pipes (say 12-inch and larger), are also rigidly connected (but with gaskets). I have always thought they can get away with that because of the pipe size. Maybe it is the gaskets after all.
I wish I had a good picture of an expansion joint nightmare I was called in to review in 2001. Three pumps in a pump station next to an apartment complex were experiencing vibration that was upsetting the tenants. The owner had an “engineer” come in and isolate the pumps. After he was finished, the vibration was no better and was perhaps worse.
Although he did put isolators everywhere, we measured the transmissibility and found that the isolators were actually amplifying the vibration.
My favorite part of this job was looking at the setup used on the discharge pipe—an isolator at the pump and another at the connection to the force main about 6 feet higher than the pump.
So the entire pipe, check valve and discharge valve was being supported by a set of isolators in shear and was visibly sagging. When I saw that, I told the owner to get rid of one isolator immediately, before the pipe could fall on the ground! We eventually replaced all the isolators with ones more appropriate and fixed the problem.
Lev Nelik responds:
Good to hear! I would love to see your addition to the topic with the analysis you suggested. Back in my younger days as a pump design engineer at Ingersoll-Rand, I used to do that—first by Shigley and similar formulas and then, later, when computers began being used (shows my age) by FEA. If you can do both (Shigley and FEA), even a simplified model will tell the story. As you said, experience matters.
P&S