What are the causes of thermal stress?

[FEAnalyst]

TL;DR

What are all the possible causes of thermal stress ?

Hi,

I would like to know what are all the possible causes of thermal stress. The most common are external constraints that block thermal expansion/contraction (for example a bar fixed at both ends). However I know that there are also at least two additional causes of thermal stress and that they can be classified as internal constraints:
1) bodies made of multiple materials with different properties (i.e. coefficients of thermal expansion)
2) bodies made of single material but subjected to nonuniform temperature (so that some parts are hotter and some are colder which results in nonuniform thermal expansion/contraction)

Are there any other possible causes of thermal stress ? What about thermal shock (sudden heating/cooling) - is that a special case of 2) or different effect ?

When it comes to case 1) is coefficient of thermal expansion the only property that matters or are there also different material constants that cause thermal stresses when they differ for various parts of a single body ? For example can the difference in Young's moduli result in thermal stresses ?

Thanks in advance for your help

 

[berkeman]

Summary:: What are all the possible causes of thermal stress ?

I would like to know what are all the possible causes of thermal stress.

Can you say more about your question? Is this for a report, or some consulting work that you are doing, or for a paper you are working on, etc.? That will help us to try to give you the best-tuned replies.

My experience with "thermal stress" testing comes mainly from my work in electronics design and R&D. I was fortunate to work at Hewlett Packard in the mid-1980s in Silicon Valley, where one of the reliability tests that we did was "STRIFE" testing. It involved 10 degree C temperature ramps from -50C to +95C, with dwell times at low and high temperature. Power cycles were added in at the dwell points, and on the temperature ramps. Test units were operating during this testing, and the test programs were looking for any hints of errors or issues.

STRIFE testing was just one of the required tests for new product introductions at HP, and helped a lot to make our products more reliable in the field and over their lifetimes.

Those of us engineers who moved on to other companies often took this HP company culture of reliability and STRIFE testing to our new companies, and that helped to improve the reliability of many products over the years.

The most common are external constraints that block thermal expansion/contraction (for example a bar fixed at both ends).

So one of the problems that we would routinely find with STRIFE testing was mechanically over-constrained components, like when power transistors had leads soldered to the PCB, but they were also bolted to metal heat sinks that may have a roundabout mechanical relationship with the PCB mounting structure. That would flex the power transistors during thermal STRIFE testing, which would lead to failures with enough cycles.

But the electronic aspect of the thermal STRIFE cycle testing was important as well. Obviously it was a double-check of static high- and low-temperature testing of the circuit operation (plus the 10C temperature margin outside of the Industrial operating temperature specification range). But more than that, adding in the power cycles at the temperature extremes ensured that the power supply circuitry and other circuitry could start up correctly, and the power cycles on the temperature ramps looked for some differential timing issues that can occur when one part of a high-speed circuit is at a slightly different temperature than another part of that circuit.

 

[FEAnalyst]

Thanks for reply and for sharing some interesting real-life examples.

The reason I ask these questions is because I deal with thermal stresses (in numerical simulations) from time to time. This topic is not so well described in literature and I wonder what can cause thermal stress. Besides, I teach some beginners the basics of practical FEA and I would like to have a short slide in my presentation listing all the general reasons of thermal stress. So far I think that those three causes from my first post cover all possibilities but of course I may be wrong. Also I'm not sure if thermal shock belongs to the group of "bodies made of single material but subjected to nonuniform temperature" or maybe it requires separate category.

 

[berkeman]

Okay, that helps.

It seems that part of your simulations should be to include the failure mechanisms. An over-constrained metal bar will fail later than a combined system that includes more brittle materials, or materials that can develop stress cracks and fractures after only modest thermal cycling (like the solder joints for the over-constrained power transistors that I mentioned.

 

[Chestermiller]

In my judgment, your initial post really correctly described mechanistically the causes of thermal stress.

 

[FEAnalyst]

In my judgment, your initial post really correctly described mechanistically the causes of thermal stress.

Thanks, that would be great. So do you agree that thermal shock can be classified as one of the cases when body is subjected to nonuniform temperature ? I think that both high and slow heating/cooling rates can lead to this effect.

Also what about the mismatch in material properties other than CTE for composite bodies ? In my opinion different mechanical properties (Young's modulus or Poisson's ratio) may cause the same effect. Is it also true for specific heat ?

 

[DaveE]

You might also consider phase changes or crystal growth. It's not common, but could be an issue with some materials.

 

[Chestermiller]

Thanks, that would be great. So do you agree that thermal shock can be classified as one of the cases when body is subjected to nonuniform temperature ? I think that both high and slow heating/cooling rates can lead to this effect.

Of course, high heating/cooling rates are going to cause higher thermal stresses than low heating rates (because of the larger spatial temperature variations).

Also what about the mismatch in material properties other than CTE for composite bodies ? In my opinion different mechanical properties (Young's modulus or Poisson's ratio) may cause the same effect. Is it also true for specific heat ?

Anything that causes one part of the material to expand more than another (cte and temperature) can cause thermal stress. Also, when more than one material is involved with mismatch of material properties, this can also do it. Specific heat differences can be responsible for mismatches in transient situations.

 

[FEAnalyst]

Thanks again, that explains everything.