Long piping - expansion

[johnp]

Long lengths of piping resist expansion movement. Local movement takes place within 40 - 60 diameters of a bend. Calculations show extent of "possible" movement, but "probable" movement actually witnessed on site over time is far less.<br>
This is particularly true with PE piping. Simple calculations show massive movements which are not seen on site, even with allowance made for theoretical snaking.<br>
I don't know how to quantify this. The simple coefficient of friction is clearly inadequate. Perhaps a "terrain factor" could be added to the equation.

 

[ingmap]

Contact Eng. R Etcharte of PLAPIQUI in Argentina, Bahía Blanca. He have information about your problem, maybe he has something about you're looking for.His mail is<br>
recharte@criba.edu.ar<br>
I hope this could be helpfull.

 

[TedT]

Hello<br>
I am going to give a general answer.<br>
Pipe expansion is a complex field. Calculated expansions must take place, however piping restraints, especially rigid anchors provide resistance to movement. The various bends in the pipe allow for flexibility. That is why bends are always recommended. If the bends are not sufficient, bending moments exceed which surpass that allowable and failure takes place.<br>
See "Piping Design for Process Plants" by H. Rase, Wiley publication. Or, look up "the Kellogg Method"<br>
<br>
Ted Tracewicz<br>
Project Engineer<br>
ATS Spartec<br>
Burlington, ON

 

[johnp]

Thanks for your comment. Agreed that pipe expansion is a complex field but that is where it gets interesting. The “Kellogg Method” is based on bending stress resulting from the sideways movement of a simple section of pipe clamped by anchors at each end. By limiting stress to an allowable value for steel the familiar DY/(L-U2) results. It is simple to apply and useful around plant design, but is hardly applicable for pipe lengths of 300 to 500 m.<br>
To get a better understanding we can use Autopipe (which is available on the Internet) to do some calculations. Assumptions have to be made for temperature extremes and friction factors. These show “possible” movement and stress. Actual movements witnessed on site over time are far less (“probable” movements). This is particularly noticeable with PE pipe. Do some simple calculations. For a 300 m length of pipe calculated movements can easily be several meters. Then walk around any mining site and look for evidence of such movement. The emerging picture is quite different. Even a coefficient of friction of 1.0 may not adequately reflect what is happening. More practical understanding is needed, hence the need to raise questions on this forum.<br>

 

[TedT]

Interesting indeed! However, can there be an "equivalent sliding support anchor" somewhere along midspan. I think it should be termed equivalent since you probably have a UDL reaction. Maybe this is indeed what you are applying with friction factors. Does this also include a spring constant for the earth at this equivalent anchor? You may actually have localized spongy pockets which provide a loy of "give". In order to simulate what's going on I think you have to simulate a multi-anchor system with spring supports. What equations are you using, and what are the sources? Have you got a web address for Autopipe? What is "PE pipe"?<br>
Ted Tracewicz<br>
Project Engineer<br>
ATS Spartec<br>
Burlington, ON

 

[helmley]

This is interesting! Pipe expansion is based on the installed temperature and the operating temperature or design temperature. One would only think that the difference in the operating temperature and the design temperature are wide apart so that the anticipated thermal expansion did not materialized (over designed system). If PE stands for polyethylene, then perhaps manufacturer's data is a miss. Certainly, the data used in the calculation shall be checked against field verification and other sources. Only one can have a definite answer. Friction factor is seldom included in the calculcation of pipe flexibility, for conservative reasons. The only reason I can think of is that, pipe expansion is not cummulative on a plant, unless it is a 100% straight run pipe. The bends and turns take some of the expansion.

 

[johnp]

Rebis AutoPIPE

http://www.ceanet.com.au/engineer/autopipe.htm

offer a “live” pipe stress analysis version limited to 20 node points. That is an advance on using the “Kellogg Method” or some of the graphical or chart solutions given in handbooks.<br>
Let us stay with polyethylene (PE) pipe because the effects are more visible. For a 20 deg C difference PE expands 5 m/km. The coefficient of expansion is given by the manufacturer, based on lab test results. We must make a number of assumptions to do the calculations.<br>
<br>
1 Coefficient of friction. <br>
In a straight length of pipe continuously supported there comes a distance along the pipe at which thermal forces are balanced by frictional forces and no more movement takes place, but there is a buildup of internal temperature stress. That length depends on the coefficient of friction, as well as on the temperature range and weight of pipe and contents. <br>
There is also starting friction and sliding friction. Movement does not take place until starting friction forces are overcome. The appropriate value to use is therefore open to interpretation.<br>
<br>
2 Straight and Level. <br>
How much does the installed pipe deviate from assumptions of straight and level? Slight deviations can have a marked effect on the results. (Long runs without bends or loops are common practice).<br>
<br>
3 Supports. <br>
Software has a problem dealing with continuous supports. Most only allow for discreet supports at node points. And perhaps we need to allow for the compaction of the soil as a “spring” effect?<br>
<br>
4 Buckling or “Snaking”. <br>
The Euler buckling of a straight length of pipe into an S-curve is quite acceptable when working with flexible piping systems. Most software does not take this effect into account.<br>
<br>
5 Temperature Difference. <br>
With water pipe one usually is given the maximum air temperature. From that one has to guess what the empty pipe design temperature might be, and guess at what season and time of day the pipe would be installed. Once the pipe is full the temperature range will be less severe and the pipe will be heavier.<br>
<br>
6 Restraints at Tees and Valve Stations. <br>
Some are bolted down to concrete blocks, some are just guided.<br>
<br>
To get a better picture of what is going on it looks like we need an extensive stress analysis model. And it needs to be “checked against field verification and other sources”. Looks like a trap for young players here. I think we underestimate the effect of friction, overestimate the material temperature on site and ignore completely Euler buckling effects. Any engineering tips based on experience to make things easier?<br>

 

[dprice]

PE is a favorite subject of mine. It is a most interesting material. No most piping programs do not take into account the important effects. However the performance is predictable if you have the time to account for all the factors.<br>
<br>
It is useful in analyzing any piping system to compare to steel which most engineers can relate to. Unlike steel PE has a high stress/strain curve. Also unlike steel PE has a low thermal conductivity. Also as you already deduced the coefficient of expansion for PE is greater than steel. Also the coefficient of friction is quite alot lower for PE than steel. <br>
<br>
I think you are having difficulty accounting for all the factors that can impact your situation. The first problem to assess is what is the temperature of your pipe. If your fluid is at the air temperature and there is no solar radiation and you have a foggy night, your pipe will be at a uniform temperature. If you are in the middle of a 100 degree F day, it is high noon, the ground below is at 60 degrees and your process fluid is at 60 degress you have a different situation. In this case the internal part of the pipe is at a low expansion while the upper part is at a high expansion. As long as the pipe is not bending lots, these internal expansion differences can be taken up in the pipe. <br>
<br>
I think your situation is much like the one I have outlined above, but without reviewing your calculations and the situation I cannot comment on the results you are comparing. I could help you more if you need detailed assistance with piping problems.<br>
<br>
Dan Price