https://www.sciencedirect.com/search?qs=Stress%20Relaxation%20Testing%20%28SRT%29&show=25&sortBy=relevance

I agree this needs consideration but we need to be clear about WHY!

I apologise that my discussion will be about terms and terminology and not about anything technical. Nevertheless:

There are serious problems with the experimental error and interpretation of uniaxial constant load creep tests. This is reason enough for looking at alternatives and I applaud David Woodford for thinking about alternatives. However, we must be clear about the final application before we perform any testing. This is the WHY!

The discussion mentions “Creep Strength” and one of the papers attached by Md. Washim Akram mentions “Design Data”. So is this the WHY? If so then the next question is can a stress relaxation test (SRT) be used to Design an Elevated Temperature Pressure Vessel?

So some technical discussion finally, well SRT has some features going for it: The total strain is small therefore, the engineering stress and strain are close to the true stress and strain. Creep means that the stress relaxation rates are initially high and might be termed primary creep. Nevertheless, the stress relaxation rate falls and approaches the minimum creep rate (or steady state creep rate if you prefer). Now I could go off on a tangent about minimum creep rates etc etc. But better I keep on message. It seems to me that the next question is: Can the stress to give a certain minimum creep rate be used for Design? Well actually it can in ASME BPVC they use Stress to produce a creep rate of 0.01%/1000 hour as one of the criterion to set the Allowable Stress. So in principal yes SRT could be used to derive one of the inputs to estimating the Allowable Stress. However, the other inputs to the Allowable Stress could well dominate and the creep rate criterion might be superfluous most for most cases. The other criterions are based on Rupture Strength and Stress to Produce X% Inelastic Strain.

But WHY WHY WHY?

Do the design code committees know what they are doing????

NOT

Now I am going to take this into discussion into Design Code land. The Design codes claim to protect against a number of degradation modes. However, let’s be clear about it First and foremost AMSE BPVC is trying to limit the Primary Stress in a pressure retaining part. Hence, the use of the Constant Load Creep Test and the smörgåsbord of criteria that ASME have come up with.

Personally, I have a lot of sympathy for David Woodford 's suggestion of the SRT. But I worry about it as well. Stress relaxation tests require very stable and high resolution extensometers and a method to reduce the load during the test. It is much easier and cheaper to perform constant load creep tests and a low resolution extensometer.

Furthermore, if all the design code committees actually need is some way to limit the primary stress then why not rationalise the smörgåsbord of criterion and choose one! I would suggest that they simply use “The Stress to Produce an inelastic strain of X% in Y00,000hours at temperature Z” Note when I use inelastic I mean plastic plus creep strain, sometimes this is called Total Plastic Strain (after all creep strain is just a time dependent plastic strain “like as if all plastic strains were not rate dependent!”).

Of course X,Y,Z need to be chosen but this is easy. X will be 0.2% or 1% and this shall be consistent with the proof strength (for the European’s) (Offset Yield Strength if you are American). Z is just temperature. Y is whatever you want it to be 100,000hours might be fine for ASME .

Personally, I think it would be a really great move to use X% Inelastic strain to limit primary stresses as it means that a consistent Allowable Stress can be derived all the way from ambient temperature up to elevated temperature. We can even stop using Tensile Strength data at the same time!

The other reasons for limiting primary stresses in this way is (i) we don’t need to mess around with Rupture Strength in a Design Code.

Hurray, Mike Spindler is on a mission to wean Design Codes off Rupture Strength. Personally, I think creep rupture lifetimes should be the preserve of “Fitness for Service Codes” and “Life Assessment” procedures and are best avoided by Design Codes.

I love trying to be controversial. David I hope you like my contribution. Feel free to shoot it down.

Michael, I'm not sure how to respond since there are so many thoughts and ideas here. I'll comment on a few:

Provided the SRT test is started at a strain just beyond yield (~0.4%), the derived stress vs. creep rate is really a locus of initial creep rates that would be measured in constant load tests (not mcr). This is made clear in the attached paper.

This response could be used directly in the ASME code to give a lower bound on stress (i.e. creep rates higher than the mcr), and would require a reduced safety factor. Since it is a nearly constant state response it should provide a more rigorous measure of creep strength.

In addition to the benefits of testing time and cost, the SRT framework provides a definable end of life involving creep strength and ductility values that are no longer acceptable for the application.

David

I agree SRT is a very good way to measure the initial creep rate (call it primary or transient I don't care which). Indeed a well controlled SRT test compared to a dead weight loaded constant stress creep test there is no contest. I personally doubt the acuracy of primary creep measurements from constant load tests. I must upload a plot, two creep tests one on a dead weight loaded machine and one on a servo-electric machine. The initial creep rates and primary strain are massively different! However, once the tests settled down the min creep rates were similar.

I also agree that an Allowable Stress could be defined for design using the stress for a certain Initial Creep Rate, with a suitable safety factor. I suspect we only use rupture strength for the simple reason that there was lots of it available even many decades ago! The question is Why are we still using Rupture Strength in Design?

As an aside there are some nice transient/primary creep equations that model the decay of an initial rate down to the minimum rate, very nicely. In all likelihood the initial rate and min rate have the same stress dependency, anyway! Of course there are many complicating factors such as using engineering rather than true stress/strain, softening (thermal and strain), plastic loading strains, loading rates, etc.........., which can confuse the interpretation of initial and min. strain rates. Nevertheless, these are often worse in the constant load test than the SRT. In principle I can see no reason why SRT could not be used for Design. We just need a lot more SRT data for a large range of materials.

I am keeping my power dry on the Intrinsic Ductility issue for the simple reason that ductility is not used in Design Codes. So until we define WHY we need ductility and what we are going to use it for I am going to stick with the R5 Ductility Exhaustion methods.

Article Advanced Ductility Exhaustion Methods for the Calculation of...

Well I had to plug my own work didn't I?