Charpy U v's V and brittle failure FFS 1

[robsalv]

Just continuing on from thread330-211640

I have access to some oldish Exxon ERE data which correlates Charpy U with Charpy V. For example, some 3.5Ni plate that was Charpy U'd in 1960 to confirm a -101degC MDMT, is said to be the equivalent of a -97.5DegC MDMT. Given that the correlation has never made it into the likes of API579, and that the information is proprietary, I'm reluctant to say more about it, BUT...

The inference is that vessels which we believed to be suitable for certain temperatures based on the old keyhole tests, might not be considered suitable when correlated up to the equivalent and more conservative V notch figures,

What thoughts does this generate in folks? Can we rely on the old Charpy U's?

Were the 1960 codes acceptable enough for low temperature design?

Does this possible correlation give any one in petrochemical industries cause for concern?

I'm wondering just how much a deal to make this since a brittle failure FFS that is being considered at the moment is marginal based on charpy U's and failing on the correlated Charpy V's.

Does the fact that the correlation doesn't appear in API579 suggest that it's not an issue??


Our plant is no longer part of ExxonMobil circuit, so I can't go quizzing the guru's



Now on another but related matter, in the linked thread, the OP had taken some samples of the vessel for analysis. I'm normally a proponent of this method - i.e., add a nozzle to the vessel and recover a coupon... BUT, it must be understood that by cutting the shell and welding in a nozzle or filler piece, you have now modified the vessel and grandfathering is off the table. Not to mention that you MIGHT actually add a defect or change the stress profile sufficiently that you could promote brittle failure... so tread warily if going down this path!



- - - - - - - - - - - - - - -
"Life! No one get's out of it alive."
"The trick is to grow up without growing old..."

 

[metengr]

robslav;

What thoughts does this generate in folks? Can we rely on the old Charpy U's?

From a code viewpoint, you can use this correlation if you have actual test data to allow for a CUN to CVN conversion for the grade of steel. If not, you would need to obtain CVN data per current FFS requirements or attempt to remove actual material samples and run CVN tests.

Were the 1960 codes acceptable enough for low temperature design?

Yes. The concept of impact testing and using code rules to establish minimum design metal temperature based on impact test data and material thickness really has not changed since the late 1950's for pressure vessel steels. What has changed is improved steel making and inspection techniques over the years that allow code users to take advantage of a broader class of steels in PV design.

Does this possible correlation give any one in petrochemical industries cause for concern?

N/A .

I'm wondering just how much a deal to make this since a brittle failure FFS that is being considered at the moment is marginal based on Charpy U's and failing on the correlated Charpy V's.

Does the fact that the correlation doesn't appear in API579 suggest that it's not an issue??

No, not at all. Most codes and standards have provisions to use design by analysis approaches in lieu of design by rules. If you can show by analysis that the CVN to CUN correlation exists for a particular steel used in design, I see no problem with defending this approach. This means that if API 579 requires CVN values for evaluation, and you have a specific correlation developed for this, I believe you should be bound to it. If you did not have the CUV to CVN correlation for your specific grade of steel, than you would have to develop one or obtain test samples to run CVN data. So, if the correlation shows the CVN converted data does not satisfy API 579, you have your answer.

 

[robsalv]

Thanks Metengr.

Good food for thought from the code point of view.

It does however just reinforce the concern in my mind, that from the API579 point of view, there is nothing to tell me that this vessel is not acceptable. I have an inherently consistent design with the standards of the day. It used CUN impacts to confirm MDMT suitability. Without this correlation data from Exxon however, I wouldn't have even doubted the cold temperature design of this vessel - but I may need to derate MDMT... This base circumstance can't be that rare, i.e., Charpy U satisfactory vessels not passing current review.

Surely this should be an industry alert?






Out of curiousity, as I don't have a circa 1960 ASME VIII handy, could someone tell me what the average and minimum CUN impacts were for a 3.5Ni vessel in 1960?



Thanks in advance.

Rob

- - - - - - - - - - - - - - -
"Life! No one get's out of it alive."
"The trick is to grow up without growing old..."

 

[unclesyd]

I have 2 booklets, circa 1965, on low temperature and cryogenic steels and neither have CUN values for 3 1/2% NI steel , both just show CVN and CKN values based on testing to ASTM A300(obsolete). It gives a min use temperature of -150F but shows it has been used to -170F. The info is for ASTM A203 Gr D & E.

Nearly all this data come from the old International Nickel Company or USS.

 

[metengr]

robsalv;
Was this vessel built to ASME Code or to another Code of construction? If this vessel was built to another code of construction, you need to revisit the rules regarding minimum design metal temperature. If the vessel was built to ASME, it would appear that the material of construction was supplied by a source that performed impact quals and decided to use CUN versus CVN. This would explain the correlation. Was this correlation was developed as an after thought?

Here is my view regarding your situation. If the correlation shows that you fall above the MDMT based on actual impact test values of the materials of construction. It would appear to me that you would need to re-rate to a higher MDMT because you are bound by the CVN to CUN correlation that was established.

The only way I see out of this would be to remove some sample and run actual CVN tests to determine if the material exhibits 15 ft-lbs or more at the current MDM

 

[unclesyd]

Is it still permissible to meet the requirements of UG 84 to get a little break on MDMT values?

 

[robsalv]

Just to clarify, the correlation is simply some generic information that generally relates Charpy impacts of one kind to another kind. It wasn't a correlation specifically developed for this vessel or this heat of steel.



The material in question was indeed A203-D and the vessel(s) were built and purchased as Code stamped ASMEVIII's for installation and commissioning into a 1960 built petrochem plant in Australia.

The MDR's include impact results on the vessel plate with a reference to U5 Charpy notch tests requiring a minimum of 15ft-lbs.



Removing a material sample for analysis will be a last resort, since we're reluctant to modify the pressure envelope. We're more likely to go with an FEA to determine the critical crack size and follow up with a detailed defect survey.

- - - - - - - - - - - - - - -
"Life! No one get's out of it alive."
"The trick is to grow up without growing old..."

 

[metengr]

Just to clarify, the correlation is simply some generic information that generally relates Charpy impacts of one kind to another kind. It wasn't a correlation specifically developed for this vessel or this heat of steel.

Then this becomes a non-issue with your vessel and with API 579.

 

[metengr]

Also, failed to add to the above....In other words, you cannot use this general correlation for any pressure vessel Charpy impact evaluation. Also, regarding your industry concerns related to this correlation, this is a non-issue.

 

[robsalv]

So philosophically speaking, your view is that a low temperature vessel design that's inherently consistent and acceptable on a charpy keyhole basis, would be considered acceptable even though it might fail on a theoretical Charpy V basis?

And presumably that the same vessel would be considered acceptable until proven to fail on an actual charpy V testing basis?



API579 does not mention keyhole or U charpy's once. It assumes Charpy V's in all charpy results and in reviewing equipment for brittle failure (section 3). From where I stand at the moment, API579 seems to have a blind spot.


- - - - - - - - - - - - - - -
"Life! No one get's out of it alive."
"The trick is to grow up without growing old..."

 

[metengr]

So philosophically speaking, your view is that a low temperature vessel design that's inherently consistent and acceptable on a Charpy keyhole basis, would be considered acceptable even though it might fail on a theoretical Charpy V basis?

Only if the code of construction recognizes the difference bewteeen the two tests and accounts for this difference by specific rules for design. Why? Because the Charpy U-notch test is a fundamentally different test in comparison to the V-notch impact test (the U-notch test is less severe in comparison to the v-notch). So, without a unique correlation to show the effects of notch sensitivity of impact behavior for a specific grade or heat of steel, in my opinion, the data from the u-notch should not be used in rules specific for ASME B&PV Code design (it might be acceptable for other codes of construction that perhaps have more conservative temperature requirements for notch impact behavior). So, this means find CVN data that is pertinent to your grade of steel, remove samples for v-notch tests or re-rate.

And presumably that the same vessel would be considered acceptable until proven to fail on an actual charpy V testing basis?

Only if the code of construction requirements have accounted for the U-notch fracture behavior being less conservative in comparison to v-notch testing.

 

[robsalv]

That's a fair enough response. Thanks Metengr.

- - - - - - - - - - - - - - -
"Life! No one get's out of it alive."
"The trick is to grow up without growing old..."

 

[unclesyd]

If you have some CVN data you might like to compare it with some Charpy Data.


Charpy Keyhole Notch impact values for 3 1/2% Nickle steel (Normalized, tempered, 1/2" plate}
@-150F 22 ft lbs with a scatter from 13 to 32 ft lbs.

@-100F 31 ft lbs with a scatter from 23 to 39 ft lbs.

Charpy V-Notch impact values for 3 1/2% Nickle steel (Normalized, tempered, 1/2" plate}
@-150F 22 ft lbs with a scatter from 18 to 40 ft lbs.

@-100F 45 ft lbs with a scatter from 34 to 62 ft lbs.


An interesting table.
Notch toughness properties,Keyhole Charpy, of 3 1/2% Nickel steel at -150F (1/2" welded plate}
As welded: Stress Relieved @ 1150F

Base Plate 20 28
Edge HAZ 14 22
Middle HAZ 16 36
Fusion line 17 36
Weld metal 23 21

Both of my references don't specifically pick one test or the other it only states that the preference now (1965) is V-Notch data over the Keyhole data as the V-Notch data conforms the idealized (S) curve that gives the different transition temperatures.