Relating temperature history to susceptibility to thermal shock

[colemanstoops]

Here's the situation: there is a stainless steel vessel outfitted with half pipe jackets that ordinarily use 150 psig saturated steam to heat the vessel contents. On occasions, the batch recipe calls for a cooling step (using cooling water, which yields heat to the atmosphere by way of a cooling tower) after steam use.

Assuming one has a temperature history of the jacket and its contents after admitting cooling water to a still-hot jacket, how would one relate that temperature history to some sort of measure of susceptibility to failure of the jacket (I'm thinking particularly of the welds attaching the half pipe to the vessel wall) due to thermal shock?

I'm aware of the existence of a so-called thermal shock parameter that uses Young's modulus, the material thermal conductivity, and the coefficient of thermal expansion. But so far all I've been able to uncover is a general qualitatitive mention that the higher the value of this parameter, the greater the susceptibility to thermal shock. Is there some relation between this parameter and a prediction? guess? of the likelihood of failure, particularly weld failure? And how might this relate to changes in temperature with time for any given item of equipment?

 

[blacksmith37]

Thermal fatigue is more likely the possible problem. I would say the design ( restraint, flexibility) is most important factor.

 

[corus]

You need to look at the stresses that arise from differential thermal expansion. This could be a combination of stresses across the thickness due to a transient thermal shock and stresses along the length of the shell, say from restraint or differential thermal expansion again. These thermal stresses are in combination with your mechanical stresses, for which you assess the total agaist fatigue limits for the number of cycles you require.

Tara

 

[davefitz]

The thermal fatiuge damage can be monitored by installing a mid wall themrocouple in the thickest component that is subjected to the thermal transients.

The relationship between the rate of change of mid wall temperature, V, to the thermal stress is defined in the new european norm EN-12952-3, and the various thermal cycles are added together using the "rainflow cycle counting algorithm", also defined in that document. If there also occur pressure changes, then the stress caused by the changing fluid pressure also needs to be added when determing total stress range at the assumed max stressed location.

The methods in that document are based on the older german boiler code, trd 301 annex 1, and those approximations in turn are based on the technical article "Instationare warmespannungen in Hohlzylindern" by W. Albrecht , journal named Konstruction, v 18, 1966, heft 6, PP 224-231.

 

[davefitz]

If it is a true thermal shock, then it might not be sensed by the rate of change of mid wall metal temperture of a thick body- in those cases, you need to monitor the temperature difference of the midwall of the thickest body and the surface metal temperature- that can be tricky to accomplish.

The method of monitoring the rate of change of mid wall metal temp works best with steady thermal ramp rates that occur over a timeem period greater than t > 3*s^2/a, where s= wall thickness, and a= thermal diffusivity

 

[unclesyd]

corus is on track as it will definately crack by differential thermal fatigue (DTE) at any change in section if the temperature is cycled. I've never seen any calulations that predicted the real world cracking. One thing mentioned is that I haven't seen is any case where mechanical stresses play a major roll in the DTE cracking problem..
We have one area of our production unit where all equipment undergoes thermal cycling and is cracking due DTE. As you imagine the heavier wall equipment cracks the most while thin sections will crack, it is extremely slow to initiate with swarm of cracks. In our case thick sections meaning above 3/8" thk. This doesn't hold true where cracking was initiated from the mixing hot and cold liquids in piping.
As stated above the number of cycles usually governs but sometimes the magnitude of the thermal gradient will amplify the speed of the cracking. In thick sections the majority of the DTE will originate on the cold side and is very hard to detect in it's early stage. We have found DTE is readily repairable as a single entity but not as swarm of cracks.

In our vessels most subject to DTE there have been many efforts to design around the problem. So far none we have tried has been completely successful under our operating conditions. These vessels are inspected twice a year and in every case need some repair. This is not saying that things cannot be done to mitigate DTE.

Addenda:
In the our case the worst offender a 304L SS separator 5' dia. x 12' long with 3/4" wall cycling from 50°C to 220°C at 8 times a day. The vessel originally built in 1952 has a 24" manway nozzle in the midsection that was originally reinforced with CS, a no no, cracking at the toe and heal of every weld. A SS repad was added with th same results. A belly band of thicker material was added to act as integral reinforcement with all welds blended very smooth. It didn't take long for cracking to start in the nozzle at the toe of the nozzle to pad weld. The next areas to crack was the toe of the fillet welds in the transition area. Cracking in this area is very hard to detect with PT though it can be with due diligence by the PT technician using unclesyd's procedure It may sound strange but I've found cracking in these vessels where the change in section was less than 0.063" in the smooth transitions. We've had the face of the 24" weld neck flange crack with the latest being on the nozzle side groove weld for the flange to nozzle. Cracking in this area was a long time process as it was 15 years before the crack was detectable by PT. Every shell penetration, and support lugs on these vessels has cracked to a varying degree with the exception of the top head area.
One thing I've wanted to do is move the 24" manway to the top head but have been overruled by the powers above.

 

[corus]

Mechanical stresses may not add to the cracking problem if these are constant throughout the operating period as it's the stress range you should be looking at. I should have pointed that out before.

Increasing the thickness of a section could well have an adverse effect on cracking in case of thermal loads as the peak stress from the thermal shock will be greater, ie. the temperature difference between the surface and the mid-section will be greater. Reinforcement will reduce the mechanical stresees but as said previously, may induce greater thermal stresses in the shell, hance a greater stress range to reduce the fatigue life.

To improve the design you could look at reducing the thermal load, say by reducing the nett heat transfer coefficient somehow, or increasing the time period over which the thermal transinet occurs. Another way would be to make the shell walls thinner, whilst satisfying mechanical load limits of course.

Tara