ML20234B828
| ML20234B828 | |
| Person / Time | |
|---|---|
| Site: | 05000000, North Anna |
| Issue date: | 06/26/1976 |
| From: | Pellini W SUN SHIPBUILDING & DRY DOCK CO. (SUBS. SUN CO., INC.) |
| To: | |
| Shared Package | |
| ML20234A777 | List:
|
| References | |
| FOIA-87-40 NUDOCS 8707060221 | |
| Download: ML20234B828 (23) | |
Text
_ _ _ _ _ _
f REVIEW OF VLFC0 CCill.MNTS ON SUN SH7P PRESENTATIONS by l
WIL1,I A!.i S. NLLINI June 26,1976 (f) INTRODUCTION The VEPCO comments on the Sun Ship presentations involve major issues regarding the basis for certification of structural reliability. The review is directed to these issues.
The first issue is that the characterization of fracture properties, with respect to section size effects, is not represented properly by V3FCO. In discussing the Pellini presentation to NRC, VEPCO cites a sharp transition (60 F above the NDT) as providing elastic-plastic arrest properties. This is the case for steels of 1.0 in. section size. It is agi the case for steels of 2 and 3 in. sec tion size. Section cize increases above 1.0 in. have tne effect of expanding the plane str.ain region in the order of 3d and50'F for
- in. and 3 in. sizes respectively, as compared to a 1.0 in. section size.
A conservative estimate for the plane strain limit (L) for 1.0 in. is NDT + 10 to 20 F and for 3 in. it is NDT + 40'to 60 F.
curve (and the K The effects are represented by the kid 7g curve) which define the temperature range above DDT that is appropriate for Ky analyces that apply to the plane strain state. In brief, desirable arrest protection is not developed in the 6d'F range above NDT for the thick section sizes of the structures. It should be noted also that arrest protection temperature range is not defined by an intermediate loading rate K7 curve.
(1) 8707060221 870610 PDR FOIA THOMAS 87-40 PDR
(
l.
The second isnue, is that unique design featuren (redundant)
(DP N16) are used as an argument by \\EPCO againct the use of a did curve analysic. Differences between pressure vesocla. shipn and trussed structures may decide if total fracture occurs, or not, following initiation. These differences have nothing to do with analys'es of fracture initiation, because it is a purely localized event.
Thefthird issue is the VEPCO assertions that meeting of a value, that corresponds to plane strain properties, specific Cy provides certification of structural reliability. No such guarantee is provided; it is a matter of analysis based on an appropriate Ky curve and other proper engineering considerations that decide if fracture initiation is possible. Thus, meeting of 23, 24 or 25 mil L.E. for the C specimen, at 80'F does not define per se y
that the structure is fracture cafe. Tha AST.IE Code (NF) does not define what type of fracture reliability evolves from meeting minimum cited values of L.E. by the 0 test-- it is left to y
appropriate analyses methods to do this.
The fourth issue is that the VEPC0 case for structural reliability is primarily based on the argument of redundancy.
However, the discussions of fracture properties do not provide an adequate definition as to the real importance of emphasizing redundancy. In our view,, this is because the fracture properties are not adequate to preclude fracture initiation with the assurance required for a critical structure.
(2)
(
(ZI) DISCUSSION
~
There are three routes by which the structural reliability with respect'to failu-
- fracture may be examined for any structure. These are (1) arrest critoria (2) initiation criteria and (3) structural redundancy.
Arrest criteria are the mont ponitive Neauce they provido assurance that fracture cannot develop. Initiation critoria must include $any factors other than the fracture properties of the base steel, and thus are generally complex and difficult to validate in practice for structures that are not stress relieved.
Structural redundancy is generally considered as a factor in relation to the applicability of either arrest or initiation criteria. For example-- if arrest criteria are assured then there is no dependence on protection by redundancy. If initiation criteria are involved (by necessity due to fracture properties) then the relative structural redundancy is an important factor.
If structural redundance is positively assured, there is no need to discuss fracture properties; excent as to the possible develop-ment of partial fractures, which do not cause immediate catastrophic failure of the structure as a whole.
The first point that should be resolved in assessing the fracture reliability of any stucture is if the steel used features arrest properties or not.
All available data indicates that the A36 and A 572 steel of the section size used, does not have fracture arrest properties.
Even if it is assumed (not proven at this point) that the UDT is 40 to 60 F maximum for the steel population in the structure.
a service temperature of 60 F higher level (that is 100to 120' F) does not result in desirable fractur f arrest properties for 3 in.
thickness. In fact the service temperature may be as low as 60*to 80 F, as cited by VEFCO. Accordingly. the properties are close to plane strain for which K could be neasured, at the lower end Id of the service range s or at best of Low leve.1 clastic-plastic properties at the high end. If the nnximum HDT temperature exceed:
40 to 60 F, plane strain properties (kid) should be expected over the entire service temperature rance. In brief, a dynamica]ly loaded KIC specimen could be made to fracture in a typically flat, brittle fracture mode at service temperature.
(3)
7-_-_-
(
The difference in fracture appearance."or steels of 40', 60"or 80 F t
NDT, and 2 to 3 in. thickness, would primarily involve the presence or absence of small shear lips in the nervice temperature range. It would be informative to conduct such tests, at leant for steel of 40*F HDT.
VETCO has avoided any direct clain that the nteel is of arrest properties at service temperatures of JO*to 120"F because this is obviously not the case. Since the stee '. can fracture in a brittle mode, with'little or no shear lips, it is then necessary to examine very closely the conditions that may cause fracture initiation.
Fracture initiation is a very localized event-- it can develop in-weld or HAZ regions and in backing baru, clipc, etc. In attempting to establish initiation prevention criteria, the critical issue is no longer the base metal alone. It includes all metal sites that can serve as points of initiation.
No credible case for fracture prevention can be made unless it is proven that initiation is not ponsible for normal and other conditions of service. For structures that are not stresa relieved, there is no practical way to prove prevention of fracture initiation.
This is particularly true for large we'ded structures containing a large number of complex weld connections. If the base steel is not of fracture arrest properties, the only credible alternative is to invoke protection of the structure as a whole by virtue of i
redundancy, if justified.
However, redundant design does not preclude the development of local fractures in regions that do not cause catastrophic failure l
of the entire structure. VEPCO appears to accept this fact by citing (page 2 of enclosure 1, letter to NRC, June 11 1976) quote:
"the wide margin of safety of the bridge (trussed) structure, where, under rare circumstances, critical members have failed long i
before any further problem with the structures."
These events have not been entire 3y rare and when they developed, the continued use of the bridge structures was due to not knowing that a local fracture event had occurred. When detected for a critical member, it has been general practice to repair as a reasonable precautionary measure. Moreover, these events have led to case inquiry, removal of fracture test npecimens and other measures deemed appropriate by responsible authority.
(4)
Ic it inferreo by V%PCO t ia: teceptance o C localiced frec;ure f
chould be the case for the curoort stri eturen': Tu /ICC preparad to guarantee that no " rare" events of localize <1 frac +ure can develop in the structures durinc, their cervice life end considering normal and other conditionc of nervice*
Juch a guarantee doen noi evolve iy &c use o ' :.tod i 'i ni (45 P shift).usHTO analynic. Jinilarl, auch n ataran tee iioe:.
iubsection J f rac ture not evolve,by barely meeting the A31F test at cervice temperatures.
properties' ef 25 mil L.E. by the C7 In fact, consideration of such low -fracture properties should lead to reasonable inquiry that calls for the prudent application of AUFI Appendix G analysic methods-- includinc weld, HaZ and any other potential source region for fracture initiation.
VEPCC objects to the use of Jection 0 enalysis while the subsection HF endorces its uce as valid for support ctructures.
The VEPCO objections to Section G analysis appear to center on the fact that the structure should not be analyzed ac for the case of a precoure vessel. The primary objcction is that the support structure is protected. by redundant design and the pressure vessel is not. However, relative redundant featuren have no direct bearing on the recults of the analysis with respect to initiation of fractures.
Apparently the VEPCO objections includo "Id curve analysic as well ao KIR curve analysis. Prof. Corten hac uced an intermediate loading rate Ky curve type of analysic. Fellini's presentation was nicunderstood by VEPCC. 2he real point that he made was for the uce of h curve for analycic of initiation from base r.ctal, and Id a considerably lower curve for the case of anL initiation for weldc l
that are not stress relieved. The real issue in this respect is the credibility of the specific.7 cucvc that is used for analycic purpoces. Objectionc to the rIA curve raiced by V2rCO, do not recolve thic question and do not in any way ancwor why a proper h
curve analycic should not be made.
Id The VZPCO objectione to conciderin; failure experience for chipc, all types of preocuro vescelo 3tc., ic purely baced on the argument of structural redun6ancy for the cupport structurec.
Carried to logical conclusions, thic point of view sayc that fracture properties are not of primary consequence. '2he same claim for safety would evolve if no c iccuncion was nade of fracture properties.
(5)
level
(
would hold for an:r
- actepted, of The VEPCC claim for safe y, i:
na for low levelu i
t wn])
of plane ctrain fracture proport es a elactic-plactic fracture proportion.
- bir enrmultan to.
i oint b:r the ;nn Yhere in no confucion on th s p advanced a fracture VEFCC has noo They collectively recognize that i n or arrrot criteria. It hac f
control case based on either initiat oof the structurec.
Phe appear-cimply invoked redundancy in defensebased on initiation criteria is R.
ances of fracture provention case If these calculations are l
i provided by Prof. Corten'c calculat ons. sider the differences in accepted, then there ic no need to contypes of structures au empha-structural redundancy between variouse initiation is not expected.
sized by V2PCO. His case is that fracturIt is repe are equivalent to i
l l
proving, by mathematics, that notab ed urder static load conditions.
structures should not have occurre tc Sun Ship eptabla Accordingly, Prof. Corten's case is net accs that are not stresc relieved consultants, particularly for structure Frof. Corten's calculations.
Redundancy has nothing to do witi t be credibic for any In brief, prof. Corten's calculations musd ncy. Since they are concerned condition of relative structural redun a
,e i. ciend the test of he.
with local regions of the sa acture, experience involving fracture.
i examination in terms of general serv ce annot be used for these The argument of unique redundant features c addresced the Collectively, the Sun Ship consultants havelishing structural calculations.
question of requirements for estabi ions clearly are thr.t it based on frac ture properticc. Their op n d on what war, or is ic not pocsible to certify the structure basef the cteelc involved.
now known of the fracture properties o d
y, VJPCL has By basing itc cace so strongly on redun anc l
ion.
apparently arrived at the sarae conc u :
Closine Hote_
jor iccues, so ac not to This review is addresset only to the ma by details, complica'te this important aspect e
p L
L-
.111111 S. Fellini (6)
l o,&
i 9
I I
8 1
THE WELDING INSTITUTE O
l r
i 1
o i
L-..
T LD 22955/2 June 1976 FURTHER COm!ENTS ON THE SAFETY OF THE NORTH ANNA SUPPORT STRUCTURES.
For : Sun Shipbuilding and Dry Doch Ltd.
By : J.D. Harrison and R.E. Dolby.
(
THE WELOING INSTITUTE i
i PlEASE REPLY TO RESEARCH LABORATORY ABINGTON H ALL ABINGTON CAMDRIDGE CGI 6AL Telepnone CAMBRIDGE 0223 891162 Telegrams WELDASERCH CAMBRIDGE Teles 81183 LONDON OFFICE 54 PRINCES GATE EXHIBITION ROAD LONDON SW7 2PG Telephone 01 584 8556 Telegrams WELDINST LONDON 5.W.7 LD 22055/2 Jtpe 1976.
FURTHER COMMENTS ON THE SAFETY OF THE NORTH ANNA SUPPORT STRUCTURES.
For : Sun Shipbuilding and Dry Dock Ltd.
By : J.D. Harrison and R.E. Dolby 1.
INTRODUCTION Our views concerning the safety of the North Anna steam generator Vepco(
supports, Units 1 and 2 were contained in Ref. 1.
in their submission to the NRC commented on a number of points in our report and the present report discusses some of these matters further.
The points are indexed to the enclosures in Ref. 2.
l 2.
APPLICABLE CODES AND STANDARDS (Enclosure 1, Section A, Page 1)
Yepco refer to the code situation.
At the time of design there were no ASME criteria stipulated for support structures. Does this remove from the designer the onus of ensuring that these structures can be certified as safe? Whilst not wishing to set
.ourselves up as a code writing committee, we would certainly question whether the toughness requirements of ASME Code Section III, Sub-section NF are adequate to ensure fracture safety of every support structure, however designed.
As stated in Vepco's analysis of Mr.
Pellini's presentation, Sub-section NT states that the designer may use Appendix G.
A prudent designer of a struccure as complex as these supports might now in our view use that Appendix.
He
would certainly require more assurance concerning the toughness than is acquired from the 25 mils lateral expansion criterion.
It is perhaps worth examining further what vendors in other countries using Pn'R's specify for support structures. These vendors are also for the most part working without, as yet, a national code for support structures.
We have contacted the following :
Prof. Kussmaul
- Materialprufungsanstalt (MpA), Stuttgart.
Mr. J.M. Vassal - Framatome, paris.
Mr. K. Tanaka Nippon Steel Corporation, Tokyo.
Mr. M. Satoh Mitsubishi Heavy Industries, Kobe, and hav6 received the following information :-
In Germany vendors' specifications require the material for support structures of primary loop components to meet the minimum Charpy toughness requirements of the Technische Regeln Fuer Dampfkessel and Arbeitsgemanschaft Druckbehaelter Merkblaetter.
The current edition of these rules requires both longitudinal and transverse impact tests, the requirements beinC 28 and 20ft.lbs respectively, as room temperature,(say 68 F).
These rules have changed little since 1970.
In France a variety of steel product forms are used for supports; all are bought to a minimum Charpy requirement, which since 1970 has been in the range 20-29ft.lbs at 32 F.
The current requirement is 29ft.lbs.
1 In Japan steels to SM41 or SM50 are used for supports. These are similar to ASTM A36 and A572, but 'when designers consider parts to be rather important they choose grade B or C of SM41 and SM50 for vhich toughness requirements are 20 and 35ft.lbs at 32 F respectively.
Grade A for which no toughness requirement is specified is also used on designer's risk'.
Finally, we have recently had the opportunity to study the NRC's draft regulatory guide 1.104 for overhead cranes. This guide allows exemption from Charpy testing if the crane is given an w______
3.
E..
f
~
overload test, but if not, toughness measurements must be made on each heat of steel and either :-
a.
the steels must have an NDT of 60 F below minimum operating temperature, or b.
they must have Charpy properties at 60 F below the ninimum operating temperature of 13f t.lbs (average of 3),10f t.lbs (minimum), or c.
an alternative method of fracturo analysis achieving the same margin of safety a2 (a) or (b) above must be carried out.
l It would be surprising if support structures for the primary coolant l
loop failed to meet minimum requirements established for overhead
- cranes, 3.
REDUNDANCY j
l (Diclosure 1, Section B, Page 2)
Vepco in their comparisons between the support structures and pressure vessels or ships make claims based on the redundant nature of the supports.
However, they do not answer our criticism of this argument. Fail safe design forms an important part of i
the safety analysis of aircraft structures where the problem is fatigue and the material is neither prone to brittle fracture nor strain rate sensitive. Can Vepco demonstrate that the fast brittle failure of one member would not lead to a high strain rate in an adjacent member? If not and if redundancy is to be made a major part of the argument distinguishing the supports from other struct-ures that have failed, the analysis must be based on a dynamic For these value of the critical stress intensity factor, Kg.
, steels this might be
- 30ksi [in, f
On this question of redundancy, Vcpco appear to suggest that complete brittle fracture leading to collapse is virtually impossible in a redundant structure. This is cicarly incorrect.
A number of failures of bridges occurred in Belgium in the late 1930s and 1940s. Thesc were Vierendeel truss bridges and therefore not
4.
i
=
4 statically determinate.
One of the failures is described in Refs. 3 and 4.
It may be argued that these bridges were not as statically indeterminate as the North Anna supports, but in our opinion it would be quite wrong to certify the supports as safe if the argument depends on niceties concerning th'e degree of redundancy and if the redundancy argument cannot be supported igorously anyway.
~
~
We do not cite these br1dge failures as being in steels comparable with A36 or A572', but simply to show that complete collapse of redundant steel structures can occur by brittle fracture.
4.
COMPAhlSON BETWEEN SUPPORT STRUCTURES AND PRESS 0hE VESSELS (Enclosure 1, Section C, Page 5)
We accept that the support structuree differ from pressure vessels but the comparison is mainly unfavorable :
a.
Pressure vessels are comparatively simple to analyse.
l b.
They have f ew areas of gross stress concentration.
c.
They are easier to insp'ect.
We consider that experience of brittle fractures in pressure vessels in similar steels and at similar temperatures is indeed relevant in the present context. Failure investigations indicate that the brittle fracture problem is not confined to one type of structure or another. The attraction of modern fracture mechanics is its wide range of application.
A further difference between pressure vessels and structures in general is that the former are usually subjected to an overload test.
Surveys by Smith and Warwick
, Phillips and Warwick and by Bush indicate that service failures by brittle fracture of pressure vessels are extremely rare (there were none recorded in any of these three surveys).
There have been, however, a
S.
number cf failures during pressure testing and it is clear that the test induces failure in susceptible structures.
Nichols(
has surveyed the literature on this subject and shows that the survival of a prior overload is almost invariably sufficient to ensure freedom from subsequent brittle fracture. We believe that because bridges and like structures cannot be overloaded, service ailures of such structures are comparatively more frequent.
1 5.
MAGNETIC PARTICLE TESTING (Enclosure 1, Section C, Page 5)
We note that 119 welds in the Unit 2 supports were inspected by MT by Vepco's consultants, and we underaLand from earlier present-ations that all welds were inspected by MT'by Vepco personnel.
Firstly, without a detailed knowledge of the toughness in the untempered HAZ at the toes of welds we cannot comment as to whether this would suffice to detect unacceptable defects.
Under ideal conditions we believe that defects as small as 0.050in can be detected by MT (prof. Corten appears to suggest 0.06251n as the limit of detectability by MT), but in the absence of a knowledge
- o. the toughness we can only make pessimistic assumptions and these lead us to conclude that the acceptable flaw size could be on the borderline of detectability by MT.
6.
TYPES OF CRACKING OBSERVED IN CORE SM!PLES.
(Enclosure 1, Section C, Page 5)
The statement on page 6 of our report was a general one covering core samples taken by Vepco and 'by Sunship, in which examples of both weld metal and HAZ hydrogen induced cracking were observed.
Concerning the Sunship core samples of Vepco repair welds, the extension of the defect in sample 4 is typical, in Welding Institute experience, of cracking due to hydrogen in weld metal.
The zig-zag nature of regions of the crack is characteristic.
In the North Anna support structures there is a strong probability that other lack of fusion or penetration defects will be ' sharpened' by
6.
3 hydrogen induced cracking as in the defect seen in sample 4.
7.
SIGNIFICANCE OF THE FACT THAT CRACKS DO NOT RUN IN WELD ZONE (Enclosure-1, Section C, Page 6)
The fact that propagating cracks do not run in weld regions in as-welded C-Mn steels is well known but is no indication of the-resistance of weld regions to fracture initiation. The vast majority of brittle fractures in are welded carbon manganese steels have initiated in regions of local embrittlement in the weld area with subsequent propagation in the parent steel. We have observed similar effects in wide plate tests to those reported by pellini for explosion bulge tests. The deviation of the crack from the weld region is generally believed to be due to residual stress effects.I )
8.
HAZ AND WELD METAL TOUGHNESS (Enclosure 1 Section C, Page 6)
Vepco point out that the K valueof36ksi/inobtainedbyBanks y
in a sample of A36 steel subjected to HAZ strain age embrittlement was at 32 F.
This is so but it is also true that Banks started with a steel having superior Charpy properties to those in the Nerth Anna support structures, namely 21ft.lbs at 32 F.
The fact that low Charpy properties in the parent steel will be associated with a lower toughness in the suberitical RAZ is seen from published work by Dolby and Scunders
)and supported in a limited way by Banks' work. Furthermore, Figure 4 of Banks' paper does not suggest that there would be any great increase in toughness resulting from an increase in temperature from 32 F to 80 F.
- Whilst we acree that weld metal toughness will probably not be a controlling factor in the present case, it should not be assumed that the toughness of MMA weld metal is always as shown in W.S. Pellini, R.D. Stout and W. Doty 's referenced work, all this having been obtained with welds deposited in the flat position. The toughness is greatly influenced by the electrode size, welding parameters and welding position used, (vertical up
[
7, position usually the worst),
Our original point was that a safety case must consider the tough-ness of welded regions if based on a philosophy of preventing fracture initiation, and we saw no evidence that weld metal or RAZs had been considered in Vepco's presentation to the NRC on pril 13th.
9.
PARENT STEEL CHARpY VALUES (Enclosure 4.)
l prof. Corten alleg'es that we were furnished with wrong information by fun Ship concerning the source of the specimen blanks tested by Dr. Stout.
In fact we were misled by prof. Corten's own l
presentation to the NRC on 13th April 1976. On page 62 we read that the specimens were taken from the quarter thickness location in plates that were two and three inches thick. We only learnt of the location from which the Charpy specimens were taken from Vepco's letter, reference 006/042676, received by the NRC on 19th May 1976.
l Concerning the completeness of the Charpy information supplied by Vepco we note the followiag from Enclosure 1 to the above Vepco~1etter :
a.
It is stated that 20 heats of material were used but these were for rolled secticns. No details for the plate material-are given.
b.
With the additional Charpy data provided as Enclosure 1 to Vepco's letter (006/42670), we now probably have data for flange material from three heats used in rolled beams and from one heat used for plate.
c.
We also note that no Charpy results whatsoever have been i
f presented for A572 material, nor for the A36 heats used in the W14 x 605 beams. We drew attention in our earlier report to the very low Charpy values for AS72 contained in an internal Stone and Webster report. We are particularly
f 8.
concerned about the large W14 x 605 sections because the i
cooling rate in the thicker flanges of these beams will have been slower after rolling and, since the beams were not normalised, the grain size will be coarse and the toughness low. This f act is recognised by Vepco (see page 11, Section E, Sub-section 2(d) of Enclosure 1 with Vepco's letter to the f
NRC, Serial No. 079, dated June lith 1976)
In summary we still find the data provided insufficient to assess the properties of the parent steels used in the supports. The minimum values already presented give us cause for concern, but we consider these more significant than averages quoted by Prof.
Corten from 2 or possibly 3 out of 20 heats used in rolled sections.
In the light of the above we cannot accept Prof. Corten's claim that we may now know more about the toughness of the base metals in these structures than we would have done had they been bought to a toughness specification.
10.
FRACTURE liFCHANTCS ANALYSIS (Enclosure 4)
I 10.1.
Elastic-Plastic Fracture Mechanics Although no generally accepted methods for elastic-plastic fracture mechanics exist, we believe that in termt of applied conditions the parameter of importance is the local strain in the vicinity of the crack. Strains in stress concentration regions may be well above the yield' strain (twice yield would be typical
/3 for a pressure vessel designed at yield with a penetration lead-ing to an elastic SCF of 3).
The fact that such strai;ts lead to pseudo-elastic stresses above the yield stress or above ultimate stress is not in the least surprising. Provided that the material in tough enough such strains can be accommodated because they are well below the ultimate strain.
In the past the main concern has been that the use of such pseudo-elastic strains might be unconservative (not over conservative) when the generally applied stress level approaches yield as it does in the present instance.
Begley, Landes and Wilson in their discussion of the application of the J integral to design
D.
(
indeed suggest that it would be necessary in such instances to e
estimate the strain concentration fractor which would exceed the elastic stress concentration factor.
In the proposed application of COD to design in Britain, it was also believed that it would be necessary to estimate local strains in the vicinity of the crack which might exceed elastica 11y calculated values
- However, Sumpter and Turner by means of clastic plastic work hardening finite element analysis of cracks in the region of a stress concent-ration show that for contained yielding an elastic analysis gives K
J =
p even when the applied remote stress is up to 0.83c. Sumpter and y
Turner's work, however, draws attention to a possible lack of conservatism in both our own and prof. Corten's analyscs. Sumpter and Turnor's results show good correlation between a pseudo-elastic-ally calculated K and J on the basis K5 only when the crack length a is corrected for plastic zone size i.e.
Yo/w (a + r )
K
=
y where Y is the usual LEFM factor depending solely on the geometry, h(
r
= plane strain plastic zone size
)
=
y The use of a value of K uncorrected for plastic zone size becomes more and more unconservative as the ratio of applied stress to yield stress approaches 1.0 (Note : For the North Anna supports the design l
stress is taken as 0.90.j).
Figure 1 is an abstract from Sumpter and Turner's paper showing this lack of conservatism for various geometries.
The effect of incorporating a plastic zone size correction would be to reduce the critical flaw sizes calculated l
in our report LD 22055/1 to still lower levels. Ilowever, we do not consider it fruitful to pursue this question further since we believe that the values quoted in our report indicate that there l
is already sufficient cause for concern.
10.
(
i Vepco state that they have determined that our calculations were based on unrealistic extrapolations of the principles of linear fracture mechanics yet the method used was similar to that adopted by Vepco's consultant Prof. Corten in their original submission on April 13th 1976 and Prof. Corten has used an identical analysis in the submission of June lith 1976.
In fact, Prof. Corten re-calculated one result from the table on page 16 of our report and arrives at virtually the same critical flaw size ( O'31n. compared with our OY o
69k'si/in., 0 g
7 = 18ksi)
= 31 and o
=
0.271n. for K
=
One might, of course, question the application of linear clastic fracture mechanics at all to a structure containing yield or half l
yield residual stresses and loaded to 0.Da and as stated in our y
report thero is no generally agreed method of analysis for such l
situations.
Nevertheless, the method used by Corten and ourselves seems to be in agreement with an emerging concensus.
We are not clear whether Vepco or Corten are questioning the values l
of li in our report and the application of hi in W present shadon.
g g
h!g was computed by Gurney (15) using a linear elastic fracture mechanies finite element analysis.
Very similar values were obtained by Frank at Lehigh University.
Furthermore, the accuracy of the analysis has been confirmed by Gurney in that the S-N curve for a fillet wcld loaded in fatigue has been accurately predicted with a fracture mechanics analysis using M a d a Paris type K
relationship for fatigue crack growth. These facts give us some confidence in the value of M for the quoted weld angles. We do g
not know what the actual weld toe geometries are on the North Anna supports, but would expect that the acgles would be between 30 and 60, the range covered by our analysis.
10.2, Residual Stresses.
We cannot agree with Prof. Corten that it can be assumed that residual stress in the transverse direction at the weld toe will only be half yicid.
Under normal circumstances the residual stress transverse to butt welds will be less than yield, but Nordell and Hall censured transverse residual stresses over 0.75 of yield.
However, these results were obtained on plate lin thich in which
11.
(
1 a normal welding procedure was used. For the welds on the North Anna Unit 1 supports up to 41n thick, all the passes prior to the final pass were allowed to cool out for the peoning operation, the interpass temperature only being maintained.
With all of the 41n to resist the lateral contraction of the final pass we feel that it is quite possible for the transverse residual stress in this In the absence of direct evidence to
}egiontobeuptoyield.
the contrary we would question whether it is prudent to make any other assumption.
10.3.
Backing Bars We note that Vepco do not address themselves to the question of the use'of backing bar's in the North Anna structures. These could be the most damaging feature of all.
The effect of backing bars will depend on the detailed geometry but the crack like defect between backing bar and parent steel could clearly exceed in effective size the value of 0.4131n given in Table 3 of Corten's original report in the Vepco presentation of 13th April and that value is based on aK of 69ksis/in, possibly too high for the HAZ.
y 10.4.
Factors of Safety Also Vepco do not address themselves to the question of safety factors.
Corten's analyses all give a critical defcet size, 11.
CONCLUDING REMARKS The burden of our argument is that in order to certify the safety from brittle fracture of any welded steel structure it must either l
l be shown that no crack can initiate in any embrittled region or otherwise that a running crack would arrest.
The initiation approach implies a detailed knowledge of the mininum initiation toughness for all regions of the structure.
For the North Anna structures insufficient data is available for the parent steel (none for A572 or for A36 W14 x605 beams) and none has been presented j
l The arrest approach implies the use of dynamic fracture toughness values appropriate to the tip of a running crack. These would be
-/,
12.
A e
about 32ksi h for the steels in question as stated by Corten.
- ~
This is lower than the 36ksi d which we used in example cale-ulations in our earlier report ar.d which gave unacceptably small critical defect size indications.
Thus, these materials are not tough enough to enable an arrest approach to be argued.
Finally, it seems to us that the question for the NRC is not y ether the structures complied with the codes existing or not existing at the time of design or whether they comply with a
present codes, particularly when these leave important decisions concerning safety to the discretion of the designer.
It is rather can the structures be certified as safe in the light of current knowledge and on the basis of information presented.
- $$$4******
O i
,'[
(
4 REFERENCES L
1.
Harrison, J.D..and.Dolby, R.E.,
"The safety of steam generator support structures from North Anna Units 1 and 2".
Welding Institute Report L
LD 22055, May 1976 i
l L
2.
Vepgo comments on Sun Ship presentation on May 20th 1976, concerning 1
the safety of North Anna 1 and 2 support structures, June lith 1976, I
i 3.
Anon, "Hasselt Bridge.
Why did it fail 7" The Welding Industry, Vol. 6, No. 6, 1938-39, pp. 109-203.
l 4.
Reeve, L., " Examination of welded steel specimens from the Hasselt Bridgn".
Quarterly Trans., Institute of Welding, Vol. 3, 1940, pp. 3-13.
5.
Smith, T.A. and Warwick, R.G.,
" Survey of defects in pressure vessels built to high standards of construction and its relevance to nuclear primary circuits".
Int. J. of Pressure Vessels and-Piping, Vol. 2, No. 9, October 1974, pp. 283-322.
6.
- Phillips, C.A.G.,
and Warwick, R.G., "A wurvey of defects in pressure vessels built to high standards of construction and its relevance to primary envelopes". UKAEA Henith and Safety Branch Report, AHSB (S)
R 162.
7.
- Bush, S.H.,
" Pressure vessel reliability". Trans. ASME J. of Pressure Vessel Technology, February 1975, pp 54-70.
8.
Nichols, R.W.,
"The use of overstressing techniques to reduce the risk of subsequent brittle fracture".
British Welding J., Vol. 15, Nos.
I and 2, pp. 21-42 and 75-84, January and February 1968.
9.
Ikeda, K. and Kihara, H.,
" Brittle fracture strength of wclded structures".
2nd Int. Conf. on Fracture, Brighton 1969. (Pub.
Chapman and Hall).
f 10.
- Dolby, R.E., and Saunderc, G.G. " Sub-critical HAZ fracturc toughness l
of C-Mn steels".
!Jetal Construction, Vol. 4, No. 5, 1972, pp. 85-90 I
_s
- ./
Begley,J.k.,Lt.ndes,J.D.,andWilson,W.K., "/.n estimation model 11 for the application of the J-integral".
Fracture Analysis, ASTM l
STP 560, 1974 12.
- Durdekin, F.M., and Dawes, M.G.,
" Practical use of yielding and linear elastic fracture mechanics with particular reference to pressure vessels".
j Proc. IMechE Conf., 'The Practical Application of Fracture Mechanics to Pressure Vessel Technology', London, May 1971.
l 13.
Sumpter, J.D.C., and Turner, C.E., Proceedings of 2nd Int. Conf. on Pressure Vessel Technology, Vol. 2, ASME, San Antonio 1973, pp 1095.
14.
Sumpter J.D.G. and Turner, C.E.,
" Design using elastic-plastic frccture i
mechanics".
Submitted Int. J. Fract.
15.
- Gurney, T.R.,
" Finite element analyses of some joints with welds transverso to the direction of stress".
Welding Institute Report E/G2/75, March 1975. (To be published shortly in Welding Research International) 16.
- Frank, K.H.,
"The fatigue strength of fillet welded connections".
PhD Thesis, Lehigh University 1971.
17
- Nordell, W.J.,
and Hall, W.J.,
"Two stage fracturing in welded mild steel pintes".
Welding Journal, Vol. 44, No. 3, March 1905, pp 424-s to 134-s.
e l
[
r m T I <> E m, 5 u,5' i
0 0
0 0
0o 6
5 4
3 0
1 5
TR0 A/-
a0 K
=
~
. A L/
</
CE 8
RO 0
CH i
oy
),
i
(
A E PS
~
6 S A
I
- g T C S
0 E R
T S
)
E i
A (E T
PS 4
O
. T A
~
0 I
M C
=
E
~
R r
2 0
0 i
t l
,lI. l 0
5 0
5 0
G 3
/
2 2
1 1
J
.