Not much happening in this forum at the moment, so maybe some food for thought....

I recently dealt with a compressor discharge line 24" dia, 1.75" thk A106B. The line was not insulated. Four fluid temperature probes at the machine nozzle all registered 170 deg C but the pipe surface, accurately measured was 70 deg C. Inner pipe wall temperature is unknown. This significant thru' wall temperature differential clearly exists in heavy pipe, yet the induced stress is not generally recognised. It must be axial tensile at the pipe outer wall, thus additive to longitudinal bending and pressure stresses.

What do you think ?

(Assuming the pipe is not rubber lined or similar) The temperature gradient effect is touched on in ASME B31.3 under 301.5 Dynamic Effects, Loads due to Temperature Gradients and also F301. The alternative way forward, I think, is not without some amount of pain. Have a look at PD5500 Annex G (the last section). This gives a calculation procedure for thermal stress which you may be able to adopt/adapt.

I don't suppose the 'thermal bowing delta temperature' can be used here? Otherwise you'll need to do an FEA or a Captain Kirk.

I would be carefull with the inside metal temperature. Somebody clever with heat transfer program can possibly calculate it. If the temp inside is high you may have micro cracks on the inner surface at the end (If the inside is yielding during operation and not when cooling down)

There is nothing happening here today. Did I miss nice stress engineer's party somewhere?

Nope, thermal bowing won't help here. Thermal bowing models a differential temperature top-to-bottom (rather than inside-to-outside) by applying a bending moment along the pipe element. Useful for partially-filled liquid line or solar heating.

See the new Roark (7th edition) pages 761-762, for a discussion of the two relevant cases (15 and 16). The same discussion is on page 585 of the 5th edition, except that the notation used in the formulas has been changed.

It's not obviously clear whether 1.75 wall on a 24" OD is "thin-walled" or "thick-walled," but I bet a few minutes with a computer would make it very clear. I also doubt that there's much of a temperature differential through the wall of the pipe (unless you're using one of the stainless grades with low thermal conductivity). You can probably use an insulation thickness software package like 3EPlus to estimate the delta T through the pipe wall.

Assuming an ambient temperature of 20C, your measurements imply that the film coefficient on the ID of the pipe is about twice as high as the one on the OD of the pipe. This is believeable for moderate external wind speeds, particularly if the process fluid's properties are such as to enable the formation of a large boundary layer at the inner pipe wall. As you stated that the process fluid was compressed air, this is more than possible, it's very likely. The viscosity of air rises as the temperature increases. It's also very likely that the line velocity of the compressed air is much higher than that of the wind; this will contribute to a higher Reynolds number, and therefore thicker boundary layer.

Section III NC addresses this directly get a copy and spend some reading time I don't recall where its covered but I know it has some specific direction.