Fracture mechanics

[glardner]

I am looking at techniques for calculating crack growth in pipe/nozzle attachments to pressure vessels. While there are engineers in the company who have some knowledge of fracture mechanics, they are still quite inexperienced; I have a lot more experience, but do not have the knowledge of fracture mechanics.

Has anyone done any sample worksheets where I could see typical techniques; or could anyone point me towards a decent source of information.

Gerard Lardner

[glardner]

The immediate requirement is to predict the possibility of crack propagation from a marginally inadequate site weld on a boiler. The weld is for a nozzle onto the tube plate of an existing, old shell boiler. The position of the weld made it very difficult to do properly; it had to be done from one side, through a 20 mm thick tube plate and very close to the furnace tube. At one point, for about 10% of the weld length, there is a minor lack of complete penetration - say 1.5 mm over an arc of 100 mm. Strength wise and from NDT testing, the weld is adequate; but it does not meet the strictest interpretation of the specified code. Unfortunately, the insurance inspector on the job is very strict and "by the book"; other insurance inspectors we have consulted have indicated that they would not have rejected the weld.

The contractor has been trying to see how he could cut out the weld and re-make it; they have tried on a test rig because of the inaccessibility of the site weld; they want to get it right first time on the boiler. But I understand that all attempts on the test rig so far have resulted in cutting out too much metal and leaving an unacceptable weld gap between the tube plate and the nozzle.

My group has been asked to advise informally if an analysis suggests the lack of penetration is or is not likely to result in any crack propagation. In other words, can an analytical argument be made that the weld is safe as it is.

Gerard Lardner

[AlvaroDíaz]

Resume the needed information:

First, the material. Needed because materials are who fracture. Which kind of steel? Assuming that in veesel manufacturing are not under special processing, it's generic and standard.

Second, conditions: materials fails under some conditions. Corrosion exposition, temp & press (stationary but transitorial must to be specified), maintenance periods and type (like inspections, recubriment treatment), dead hours by mounth, external isolation, etc.

If have internal recubriment, are two material and two interface surfaces (fluid-recubriment, recubriment-steel).

Next, collecting information. Source are: books, manufactury papers, ISO standards (or ASME or API) and guys knowdoladge.

For better references search for science materials books, like Askeland. Resistence of materials too, but think that this calculations aren't in all of this books.

In theory all factory equipament came with papers with statistcal information for failures, email where you buy vessels asking for it's. Also ask about under which ISO normative is make this vessel.

Search for your self in ISO standards books for vessels or for equipament specifically from your industry, I guess that your enterprise have a printed copy anywhere in administration, or (maybe better for quickly search) an electronic copy in pdf format.

Those ISO's if not have the calculus must to have a reference from where are they (an other ISO standard also).

If there are things in calculus procedure from manufacture's manuals or ISO standards can post here it and sure someone tell us how to make to work this into mathcad.

Regards. Alvaro.

[glardner]

Thank you. That is pretty much the route of my investigations to date. I asked here in the hope that someone might be able to point me towards the appropriate fracture calculation methods.

If I do follow the calculation route - I am beginning to think I might decline, as this is sufficiently far from my expertise - I will offer the worksheet here for comment.

Gerard Lardner

[glardner]

My own opinion, independently of being involved and having seen photos and NDT reports of the weld and having checked the structural calc for the tube-plate material thickness, is that it's probably OK. The applicable weld thickness is adequate, but I am not absolutely certain how the lack of penetration will behave - will it allow a crack to form and propagate into the weld?

[schneidrax]

On 11/24/2009 12:18:56 PM, Gerard_Lardner wrote:
>Re: Fracture mechanics.
>
>... could
>anyone point me towards a
>decent source of information.
>
Professor John W. Hutchinson of Harvard is a world reknown expert on fracture mechanics.

[DavidMatthew]

Gerard,
I am a specialist in fracture mechanics within the aerospace industry. I have used Mathcad for simple linear elastic fracture mechanics (LEFM) in the past but it's not ideally suited to this. My suggestion would be to check some freely available LEFM software packages. I suggest AFGROW which is a very user friendly application with built in materials database. Have a look at:
http://www.afgrow.net/

AFGROW was developed by the US Air Force, but it is used widely in all industries.

If you did want to do the analysis in Mathcad, the basic theory is fairly easy to code. For the theory, a good text book is "The Practical Use of Fracture Mechanics", David Broek. I'll see if I can find a simple Mathcad example of a basic analysis.

Your main challenge will be finding crack growth material data as it's generally not as widely available in the public domain. You then need to be able to model the geometry, which can be done with Finite Elements, or using a standard solution from a reference book such as "Compendium of Stress Intensity Factors, Rooke and Cartwright". It would be easier to use AFGROW as it includes a large material database and many standard geometry models.

David

[glardner]

Thank you. I'll take a look for the books you suggest, and at the web site.

Even if I do not use it on this occasion, I think it would be useful for our office to have a better understanding, if not expertise, of this subject.

Gerard Lardner

[AlvaroDíaz]

Here a good starting point about what to do with a vessel.

Regards. Alvaro.

[AlvaroDíaz]

On 11/25/2009 4:44:44 PM, jmG wrote:

>Click on the link = nothing comes ?

I can open the uploaded pdf file in the browser, try to save to your desktop and open from there.

Regards. Alvaro.

[AlvaroDíaz]

What's the problem Jean? It's only a one page pdf file, with a flux diagram, nothing else. Open well in three machines that I check.

Alvaro.

[AlvaroDíaz]

What???

If something is wrong, sure Mona can correct this. Can other collab check what Jean says? Please, this issue turns offensive.

Alvaro.

[AlvaroDíaz]

Why do you make me responsable about which webs do you visit? Why do you join my name with that offensive words in yours posts? What game are you playing?

This is my last post answering anything yours.

Alvaro.

[alnstevens]

On 11/26/2009 2:39:35 AM, adiaz wrote:
>What???
>
>If something is wrong, sure
>Mona can correct this. Can
>other collab check what Jean
>says? Please, this issue turns
>offensive.
>
>Alvaro.

I can open the pdf ok. I see a single page flow diagram.

stv

[AlvaroDíaz]

Thanks very much stv.

Alvaro.

[RichardJ]

On 11/26/2009 2:39:35 AM, adiaz wrote:
>What???
>
>If something is wrong, sure
>Mona can correct this. Can
>other collab check what Jean
>says?

I don't have the faintest idea what he is talking about. If I click on your attachment it opens: a 1 page pdf with a flow chart. Maybe Jean is referring to something else, but if he is then I don't have a clue what it would be.

Richard

[ptc-1368288]

On 11/26/2009 7:59:41 AM, rijackson wrote:
...
>Maybe Jean is referring to
>something else, ...
>
>Richard
___________________________

Not something else, just the applicable API standards.



jmG

[AlvaroDíaz]

On 11/26/2009 7:59:41 AM, rijackson wrote:
>I don't have the faintest idea what he is talking about.

I have not too. API it's an acronim, and 510 the number for a publication: API 510 (8Th 1997 52P) Pressure Vessel Inspection Code. Thanks Jackson.

Regards. Alvaro.

[ptc-1368288]

On 11/26/2009 4:06:46 PM, adiaz wrote:
...
>API it's an acronym<

==> nobody said it was not, all Engineers know API.
==> especially myself as a retired Process Control & Instrumentation Consultant.


>510 the number for a publication: API
>510 (8Th 1997 52P) Pressure Vessel
>Inspection Code.
=======================================

The publication you are referring too is outdated by 10 years:

"
API 510
Pressure Vessel Inspection Code: Maintenance Inspection, Rating,
Repair, and Alteration
(ANSI/API 510-2006)
(Purchase includes addenda to the current edition of the code)
Covers the in-service inspection, repair, alteration, and rerating activities for
pressure vessels and the pressure- relieving devices protecting these vessels.
This inspection code applies to most refining and chemical process vessels
that have been placed in service. This includes:
� vessels constructed in accordance with an applicable construction code;
� vessels constructed without a construction code (non-code)-A vessel not
fabricated to a recognized construction code and meeting no known recognized
standard;
� vessels constructed and approved as jurisdictional special based upon
jurisdiction acceptance of particular design, fabrication, inspection, testing,
and installation;
� nonstandard vessels-A vessel fabricated to a recognized construction
code but has lost it's nameplate or stamping. Pages: 56
9th Edition | June 2006 | Product Number: C51009 | Price: $146.00"

......... API is now at the age of electronic books.
Browse their web site(ssss) for update and +.



jmG

[Fred_Kohlhepp]

The way I have done this previously is to find the crack growth rate vs K curves (da/dN vs K) for the material you have. From the geometry of the crack and the stress levels you get the current K value. That translates into a crack growth rate (da/dN is crack length change per cycle.) The new crack length "a" results in a new K, which results in a new "a" and continues until K reaches K sub Ic (or your appropriate failure criteria.)

Usually the hardest part is the material properties graph. The nuclear industry is probably your best source for pipe steel.

Fred Kohlhepp
fkohlhepp@sikorsky.com