ML20127K665

Provides Comments from Review of Rept of British Ammonia Vessel Failure

ML20127K665
Person / Time
Site:	Monticello
Issue date:	02/01/1967
From:	Shaw M US ATOMIC ENERGY COMMISSION (AEC)
To:	Morris P US ATOMIC ENERGY COMMISSION (AEC)
References
NUDOCS 9211200465
Download: ML20127K665 (3)

Text

<

g. >

• "..,T.

b'E$}

UNITED STATES GOVERNMENT Memorandum PNL To  : P. A. Morris, Director DAT r.: g jgy/ l Division of Reactor Licensing

^ '

FROM : Milton Shav L c'toh Division of Reactor Levelopment and Technology SUBjtcT: BRITILE FRACTURE OF BRITIS11 AW.ONIA VESSEL - RDI EVAU.lATION OF FAIIDRE REPORT RDP:RE V:006

Reference:

Report entitled " Brittle Fracture of a Thick Waned Pressure Vessel", from special issue of BWRA Bunetin, Vol. 7, No. 6, June 1966 (Reprinted in " Welding Research Abroad", Oct.1966).

In accordance with the telephone request of your Mr. J. J. Shea to our Mr.

P. A. Halpine, RDI staff members have reviewed the referenced report of the British ammonia vessel failure and have the fonowing coments to

. offer concerning its implications or applicability to the fabrication and test of nuclear vessels in this country: -

1. The report appears to be a comprehensive and ob,$ective evaluation of the factors contributing to the failure.

2. The primary cause of the vessel failure can be attributed to inadequate control of the furnace temperature during the tempering operation (stress relief), coupled with failure to maintain an (NDP + 600F) rela-tionship for selection of a suitable test temperature. Because of improper temperature control, the transition temperature of the veld metal was in excess of 10000, whereas the transition temperature of the base materials was in the order of 0-1000. With proper tempera-ture control, the transition temperature of the veld metal vould have been in the order of 10-1500. Therefore, using our (NDT + 600F) criterion, the test temperature should have been at least 40-4500 in lieu of the 1000 temperature actuany used. If a 40-4500 temperature had been used, catastrophic failure probably would not have occurred because the still-relatively-sman crack in the veld metal would have beeA arrested by the relatively good properties of the base materials.

Further, if the veld metal had been properly stress-relieved, failure probably would not have occurred even at the 1000 test temperature (because the existing flaws were very small and the stress level at test pressure was no more than 85% of minimum yield). Nevertheless, it is emphasized that these are probabilities and that a proper criterion vould have required an (IDI + tXFF) test temperature in rela-tiontothepoorestpropertiesinthevessel(highestNDIofany component in the vessel). This is recognized in the report, although the report does not recommend or suggest a specific criterion.

?

9211200465 670201 PDR ADOCK 05000263 A PDR Buy U.S. Savings Bonds Regularl,y on the Payroll Savings Plan I

.. u P. A. Morris 3 In addition to the foregoing, there were other questionable practices used in the fabrication and inspection of the vessel, including the following:

a. Preheat was not maintained on the vessel until stress relief, which the report recognizes as the cause of the = mil cracks in the heat affeeted zone of the manual welos.

b. The velds were non-destructively tested prior to stress relief.

The first of these items is not covered by the A!EE Code but is recog-nized as being highly important for materials ol' this type; "best American practice" does require that preheat be maintained until stress relief. A recent change to the Code vill require that radiography be accomplished after some stage of stress relief. However, it is questionable that this requirement is entirely adequate; further, it is doubtful that the best radiography would have detected cracks as small as those existing in this vessel. The importance of performing non-destructive testing after stress relief is increased by recent evidence

. that veld cracking can sometimes develop during stress relief.

4. The implications of this report with respect to our own requirements and practices can perhaps be summarize 6 as follows:

a. Failure to maintain proper furnace temperature control is basically afailureofinstrumentationand/orpersonnel. In this case, the furnace was top-fired; and apparently the large, long vessel restricted gas circulation to the degree that the furnace tempera-ture was no longer uniform. This vould seem to say that assurance of proper stress relieving temperatures is best provided by a number of thermocouples attached directly to the vessel. It should te noted that this may constitute a particular problem in the stress relieving of field erected vessels where facilities may be less than optimum.

b. Paragraph N-532 3(3) of the Code pemits a variation of up to 1000F in the temperature of a vessel during stress relief. -It is not clear that the requirements of Table N-532, Note (3), or of N-541.2 covering the heat treatment of the velding procedure '

qualification test plates, adequately insure that the test coupons and test plates are really representative of the vessel. In this regard it should be noted that the test plates probably should be tested at the maximum temperature limit for measurement of tensile strength properties, but at the minimum temperature in order to be representative of the poorest impact properties. This problem could be further accentuated through use of the lower stress relieving temperatures permitted by Table N-532, Note (1).

fl

.. . n P. A. Morris , c. The referenced report incicates that the low furnace temperatures resultedinaninadequatestressrelief(intermsofthephysical propertiesoftheveldnetal)inadditiontoprylucin6Poorim-pact properties. ReferringagaintoTableN-532, Note (1)ofthe Code, it is understood from discussion with Mr. R. D- Wylie that the temperature-time relationships of this note are predicated on old stress relaxation data for carbon steel and that its validity has never been properly checked for lov alloy steel.a.

d. The referenced report indicates some confidence in the presumed benefits of " vara prestressing" provided that its beneficial effects are act " invalidated by additional growth of defects, or by changes at defect tips causing significant local embrittle-ment". In this regard, it should be noted that a boiler drum fabricated by Babcock & Wilcox, Ltd. (from material of this same n=tnn composition) failed recently (May 1966) during its third hydrostatic test at a pressure materially less than that used for the two prim. tests. Very few details of this failure are car-rently available,"but it is understood that the vessel contained

, a flaw perhaps 2{ deep x 4-5" lon6 in a 4" thick wall. The vessel withstood two normal hydrostatic tests, presumably to about150%ofdesignpressure. About a month later, the vessel failed during its third hydrostatic test at approximately its design pressure. The failure has been attributed to strain aging of the material at the tip of the defect, which permitted initia-tion of fast fracture. This emphasizes the need for conclusive information concerning such phenomena as strain a61us, temper embrittlement and irradiation dama6e, which can cause a dateriora-tion of the impact properties of a material. These effects are believed fairly well understood in our presently used materials (represented by A-302, nickel modified steel), but should (and vill) be given particular attention in any new materials.

e. The referenced report states "The Charpy V-Notch impact test is not entirely satisfactory for indicating the fracture performance of 6" thich high strength steel. Pather work based on fracture mechanics principles is necessary k vovide criteria for thich high strength steels". We concur a this statement, the study of which vill be a major objective of :he Heavy Section Steel Technology program being assigned Oak Rid 6 e National Iaboratory.

In conclusion, we do not believe that such a failure is a siE nificant risk in nuclear vessels built and tested in accordance with " recognized good practice" in this country. Nevertheless, it emphasizes potential limitations in some of our Code requirements (as differentiated from "Soodpractice")andtheneedforemphasisoncertainproblemsunderour Heavy Section Steel pro 6 ram, cc: J. J. Shea, DRL(20)

J. A. Liebenman, RDF

__ __ _ -- - -