Stress Analysis Exception - Code Change vs. B31.3 319.4

[StressGuy]

I've got a question for the Code experts that I'm looking into.

B31.3 section 319.4 discusses flexibility analysis. 319.4.1 (a) indicates that one of the exceptions to allow the designer to avoid a flexibility analysis is that the system under consideration duplicates an existing system.

I'm looking at essentially that situation. We did "job 1" a few years ago and are now doing "job 2" which is mostly identical to job 1. I'm already accounting for the review needed for the true differences.

However, "job 1" was governed by an earlier edition of B31.3 and the "job 2" copy will be per the 2010 edition. The biggest issue would be changes to allowable stress values in materials compared to what the original job was engineered to.

Now, this is really more of a thought experiment than a real problem as most of the materials used have changed little if any in their allowables.

But, conceptually, if a material's allowable stress had been cut down by 10% across the different versions of the code, how would I reconcile that with 319.4.1?

Edward L. Klein
Pipe Stress Engineer
Houston, Texas

"All the world is a Spring"

All opinions expressed here are my own and not my company's.

 

[BigInch]

The new allowable stresses are 90% of the old allowable stresses? Correct?

OK evaluate quickly; Was the previous analysis stressed to over 90% of the previous allowable stresses? If it was, I'd say that you need a new complete analysis using new material allowable stresses, as I would tend to expect that, if the old analysis predicted stresses within 90-100% of old allowable stresses, the new allowable stresses would most likely be exceeded.

If your old analysis was stressed to 85% or less of previous allowables, I'd tend to let it go as, "OK by inspection".

Independent events are seldomly independent.

 

[SNORGY]

I think the Code implies that the new design must replicate a similar system with a proven acceptable service history, or words to that effect. It is an ambiguity that a Regulator might overturn the decision on. It would be prudent to have the "Job #1" calculation files on record for submission in support of "Job #2".

 

[StressGuy]

Yes, I did simplify the expression, but the code does make note of having a proven service history. Which, in my mind, makes the question even more poignant. How long does it take to develop such a history? 10 years? 5 years?

This is even more likely to result in "job 1" and "job 2" being separated by different editions of the code.

To go with BigInch's followup - We say a particular system on "job 1" has been in service for 10 years and was designed to code edition 1. Now, the client wants to copy that with "job 2". The code is up to edition 5 and the allowables for the material have been cut to 90% of what they were in edition 1.

We manage to find the calcs from "job 1". We find that we were at 95% of the edition 1 allowables. Which means that same design exceeds the allowables compared to edition 5.

On the one hand, 10 years seems like a pretty good bit of service history to justify. Though, if it's a typical refinery/chemical process, it's probably seen no where near the 7000 cycles that the allowables were based on.

On the other hand, one would expect that the committee folks had good reason and new knowledge come up to justify the allowable change. Thus, it would seem that the designer would need to try to sell the purchaser on the need to revise the layout of the new job.

Edward L. Klein
Pipe Stress Engineer
Houston, Texas

"All the world is a Spring"

All opinions expressed here are my own and not my company's.

 

[XL83NL]

@BI: 90% of the old? For all materials? Or is that just a hypothesis?
There certainly must be some materials that have gotten higher allowables over the years ...

 

[XL83NL]

PS, anyone knows the definition of i (in i >= 5), footnote 9 on page 38 of B31.3 2010 edt.?
Acc to app. J (pag. 262) it should be the SIF (acc. para. 319.3.6), but I cant figure how it's defined then.

 

[BigInch]

X8, I'm just trying to confirm StressGuy problem.

Does the job #1 performance history at present show any evidence of relatively insignificant developments, such as stress lines, excessive yielding, excessive hardness, tiny surface cracks, etc. that in your current opinion would progress over time to reach failure levels well before 7000 cycles that job #1 was designed for, or in other words, could you present the argument that the performance history and examinations, or inspections done since, show no evidence to indicate there is any reason to suspect that a failure would occur earlier than 7000 cycles.

Independent events are seldomly independent.

 

[BigInch]

Kind of like evaluating a corrosion problem. given some wall loss over the previous 5 years, and assuming a linear corrosion rate, would that rate cause you to be alarmed, if you made a similar linear extension of the same observed corrosion rate to 30 years and it still resulted in a wall thickness > that required for pressure and stress. You can do more or less the same for any type of observed current defect for which you can reasonably presume a future growth rate. If you see no evidence of fatigue already, assuming there has been sufficient time to see some yet insignificant evidence of it somewhere, why get excited?



Independent events are seldomly independent.

 

[StressGuy]

That is indeed a good point - certainly evaluating the corrosion of the existing system and evaluating it for cracks would be a vital part of demonstrating a good service history.


@XL83NL - material properties for different materails have undergone a number of revisions over changes in the code. Some have been increased, but most of the ones I've seen, where there has been a change, have gone down. I don't remember the exact specs, but I remember there were a group of stainless steels that had been listed at 20,000 psi from 70 up to 400F and were revised down in an edition to 16,400 psi.

Edward L. Klein
Pipe Stress Engineer
Houston, Texas

"All the world is a Spring"

All opinions expressed here are my own and not my company's.

 

[BigInch]

Statistical repair, maintenance and service life prediction techniques even extend to the medical field. The latest theory concerning slow growth prostate cancers, which apparently are quite common, is that if the predicted growth rate after numerous regular examinations, isn't expected to kill you until you reach 162 years old, why take the risks of operating now. Just keep on doing the regular checkups within the "repair and replace" time interval and known +/- confidence levels of the presently observed activity.

Independent events are seldomly independent.

 

[SNORGY]

Is there a strong possibility that Job #1 stress levels were at 90%? I would think that such cases would be exceptional, so that most of Job #1 could be replicated. Moreover, maybe it's just me, but if there was a lot of 90% stress level situations found in Job #1, I would look for ways to better optimize things in Job #2. Sometimes you are stuck with it, but it's better to try to improve the design than simply photocopy it into new title blocks.

 

[BigInch]

Nothing wrong with "improve" as long as it considers, If it ain't broke, don't fix it.

Independent events are seldomly independent.

 

[XL83NL]

PS, anyone knows the definition of i (in i >= 5), footnote 9 on page 38 of B31.3 2010 edt.?
Acc to app. J (pag. 262) it should be the SIF (acc. para. 319.3.6), but I cant figure how it's defined then.