Best practice for modelling refractory lined pipes for stress analsis?

[barrind]

What is the best practice for this, or should I say, most efficient practice? Modelling the steel and refractory as either a composite shell or as a solid seems tricky to me due to getting the properties of the refractory right and how it should interact with the steel shell. I have read that you can modify the elastic modulus to combine the stiffness of the steel and concrete and then just model the steel. Is that a bona fide method? The equation they gave for an empirical calculation of the elastic modulus of the lining refered to some concrete standard and I was unble to find the parameters they used, but assuming you could find the property from a table would this be an ok method when you are only interested in the stress in the pipe? Is this appicable to an FE approach in any case?

 

[GBor]

Are you trying to model internal pressure? Fluid Flow? Or thermal expansion? Since you are talking about modulus, I assume it isn't flow through the pipes, but rather internal pressure...

 

[EdR]

Using an equivalent modulus is a reasonable way to model this composite type of problem....however....the results must be treated very carefully....Typically the results for displacements will be correct (assuming you use a correctly calculated equivalent modulus) but the results for stress must be treated as stresses from a composite material...(see a Mechanics of Materials book on the transformed section method for handling composite materials)...thus requiring additional post procesing by hand calculation.

Depending on probles size, etc. and other things we don't know it might be more efficient to use overlaid elements...but this also may have some pitfalls....

Ed.R.

 

[barrind]

Would it be ok then do use the deflection results from the modified modulus model to drive another model which is reverted to the true steel modulus, i.e. using displacement control for the entire model, in order to read stresses directly?

 

[EdR]

Depends on whether it is done correctly or not.....

The composite material causes a shift in the neutral axis which in turn changes the stress and strain distribution...Just taking the results of the composite run (displacements) and applying them to another model the same size made of steel will not give the correct results.....To show this take a simple composite beam and compare the results of a hand calc. with those for a similar beam amde of one material......

Of course I am assuming that the lining has a significant effect on the actual situation....

Ed.R.

 

[jhardy1]

Be VERY careful in how you model the refractory, and how you interpret the stress results. Assuming the pipe is subject to internal pressure, it is possible (probable?) for the refractory to develop at least partial-depth tension cracks. If it cracks, it will not contribute significantly to the overall tensile strength and stiffness of the system. In any case, the applicable design code may require you to design the steel pipe section on the assumption that it carries all tensile stresses due to internal pressure, regardless of whether the refractory cracks or not.

I have encountered instances of refractory which developed significant cracks (due to a combination of tension and thermal effects), which allowed the hot fluid to reach to the steel pipe wall at the root of the cracks, significantly increasing the local steel temperature near the roots of the cracks, well above the theoretical temperature that is calculated for the uncracked refractory. This also ensured that the steel wall was effectively carrying all of the hoop tension, because a full-depth crack in the refractory means that it has no effective tensile strength. (Probably more of an issue for fluid filled pipes than gas filled ducts, because of the higher heat transfer coefficients for typical liquids.)

 

[corus]

If the refractory is internal then it was common practice to model the pipe wall and refractory in one dimension (radial) and calculate the radial pressure produced by differential thermal expansion of the refractoy to the pipe, which is most signifciant during the early stages of a thermal transient. These pressures are simply added to a model of the pipe itself (together with any internal gas pressures), disregarding any additional stiffness provided by the refractory as this tends to be negigilble. The self weight of the refractory is added as an additional load, however. You can also model the refractory and shell together, but I've never seen a composite material/thickness shell used as its conservative to ignore the refractory for stiffness purposes and you can't be sure that the refractory will be tied completely to the shell sufficient to add to the shell stiffnesss. In fact in some thermal cases the refractory can come away from the shell.

corus

 

[4Pipes]

Just to add to comments already made. Keep it simple and agricultural. Use common sense and seek past experience. Don't get brow beaten into making it thinner.
Don't ignore those with hands on operating experience. Trying to model the affects of expansion gaps and what is effectively civil engineering in precise way stands a good chance of coming unstuck if you pardon the pun.
I've seen a couple of failures where the promises of detailed analyis failed. Then you'll get you get the "told you so". One such calc I remember cost the equivalent of a years salary and still went severely pear shaped (literally) costing several times more to fix.

 

[barrind]

Thanks for everyones comments. As a trial I did a run of my model with a steel/refractory composite and you are correct, the resulting reduction in stress did seem excessive. I would add that the stress is due mainly to local bending of the main, not differential thermal expansion. In this case this is analysing something which has already been built and has been up for about 15 years. My analysis without the refractory modelled but including the weight of refractory shows a significant part of this section of main near a bend is almost at yield. I think in my report I will just make comment about the potential extra stiffening provided by refractory but offer these results as a worst case and let them decide.