Failure Analysis Case Studies II Episode 12 pdf

Jones Editor UNUSUAL CASES OF WELD-ASSOCIATED CRACKING EXPERIENCED IN A HIGH TEMPERATURE CATALYST REDUCTION REACTOR M.. HOLLAND Metallurgical & Inspection Services, Mossgas, Private B

Failure Analysis Case Studies II

D.R.H Jones (Editor)

UNUSUAL CASES OF WELD-ASSOCIATED CRACKING

EXPERIENCED IN A HIGH TEMPERATURE CATALYST

REDUCTION REACTOR

M L HOLLAND

Metallurgical & Inspection Services, Mossgas, Private Bag X14, Mossel Bay 6500,

Republic of South Africa

(Received 3 February 1998)

Abstract-Two case studies are described which concern instances of weld-associated cracking discovered in

a high temperature Cr-Mo catalyst reduction reactor soon after commissioning One of the defects was diagnosed as re-heat cracking a t a heavy section nozzle-to-shell weld, which was attributed largely to the high stress concentration at the toe of the weld in conjunction with tri-axial stress, resulting from the thick section geometry Cracking was believed to have initiated during post-weld heat treatment which was only carried out

2 months after completion of welding

The other defect described is a classic case of HA2 cracking a t the external support legs of the reactor, again attributed largely to the delay in conducting post-weld heat treatment after fabrication

In situ replication metallography was instrumental in establishing the failure modes in both instances, and was also able to demonstrate that the HAZ cracks were present before PWHT was carried out 0 1998 Elsevier Science Ltd All rights reserved

Keywords: Hydrogen-assisted cracking metahrgical examination, process-plant failures, reheat cracking,

The catalyst reduction reactor operates in a hydrogen-rich environment at a temperature of 385°C and a pressure of 1770 kPa and is therefore fabricated in ICr-l/2Mo steel During routine inspection

of the internals of the reactor after approximately one month's operation, a crack was detected in the hydrogen inlet nozzle C1 A more comprehensive inspection of all weld seams was therefore undertaken and revealed the existence of numerous cracks in the external support ring of the vessel The failure investigations undertaken into the causes of these cracks are summarised in the following sections

2.1 Visual examination

The nozzle was a heavy wall set-in forging to SA182 gr F12, having a maximum section thickness

of 133 mm at the shell/nozzle weld where it was welded to SA 387 gr 12 cl 2 plate (see Fig 1)

Magnetic particle testing revealed that the major crack was on the top of the nozzle and extended for a distance of 240 mm along the toe of the weld, in the heat affected zone of the forging The defect was initially assumed to be a fatigue crack It was noted that there was a relatively sharp transition at the toe of the weld joining the nozzle to the shell wall

Reprinted from Engineering Failure Analysis 5 (2), I7 I - 180 ( 1998)

374

Fig 1 Schematic view of catalyst reduction reactor (bottom section)

2.2 Metallographic examination

The two extremities of the crack were prepared for microstructural replication, polished to a 1

pm diamond finish and etched with 2% nital Examination of the replicated micrograph in the laboratory showed the microstructure of the HAZ to consist of tempered martensite/bainite The

crack path was inter-granular in form, following the prior austenite grain boundaries (see Figs 2 and 3) The nature and location of the crack was typical of re-heat cracking of a low-alloy Cr-Mo steel, rather than a fatigue crack

2.3 Hardness

Hardness measurements were taken across the cracked zone on one of the polished replicated areas using a “Microdur” portable hardness tester The results showed generally acceptable hardness

(a) a susceptible alloy composition;

(b) a susceptible HAZ microstructure;

(c) a high level of residual strain combined with some degree of triaxiality;

(d) temperature in the strain relaxation (creep) range

When the value of the parameter P is equal to or greater than zero, the steel may be susceptible to reheat cracking The cracks are intergranular relative to prior austenitic grains and occur pref- erentially in the coarse-grained HAZ of the weld, usually in the parent metal but also sometimes in the weld metal There are two distinct fracture morphologies: low-ductility intergranular fracture (as shown in Figs 2 and 3) and intergranular microvoid coalescence The former is characterised by relatively smooth intergranular facets with some associated particles, and occurs during heating between 450 and 600°C The brittle intergranular mode is initiated by stress concentrators such as pre-existing cracks or unfavourable surface geometry

Compositions that have suffered reheat cracking in practice are Mo or Cr-Mo steels with more than 0.18% V, all of which have parameter values greater than zero ASTM steels that are known

to be subject to reheat cracking in thick sections are A508 Class 2, A517 Grades E and F, A533B, A542 and A387 Grade B Time constraints precluded a detailed chemical analysis of the nozzle forging, and the material certificate does not quote residual values of V, Cu, Nb or Ti According

to eqn (1) above, however, the value of parameter P is 0.571 which would indicate susceptibility to reheat cracking

The cracks generally occur during the PWHT heating cycle before reaching soaking temperature, probably in the 450-700°C range The heating and cooling rates do not appear to have any significant effect on the result Reheat cracks may also form or extend in service if the welded component is operating at elevated temperature and if joints are exposed to tensile stress, due to either inadequate PWHT or service loads

(a) Material selection: for heavy sections, alloy content should be limited as indicated by the

(b) Design to minimise restraint: where restraint is unavoidable, consideration should be given to

(c) Use of a higher preheat temperature: dressing the toes of fillet and nozzle attachment welds; use

The literature indicates that reheat cracking may be avoided by the following means:

Japanese formulae and vanadium should be limited to 0.10% maximum

intermediate PWHT after the vessel is part welded

of a lower-strength weld metal

2.4.1 Fabrication considerations Consideration of the foregoing discussion in relation to the cracking experienced in the nozzle C1 of the catalyst reduction reactor indicates that many of the factors contributing to reheat cracking were present during fabrication The following factors are believed to be particularly pertinent:

(a) Very heavy section thickness (133 mm) of forged nozzle in susceptible material

(b) High stress concentration at toe of shell-to-nozzle attachment weld

(c) Over-matched high yield strength filler metal E801 5-B2 (UTS 567 MPa/+83,000 p.s.i.) com- pared with parent metal specification requirements of A182 gr F12 (min UTS 485 MPa/70,000 p.s.i.) and A387 gr 12 cl 2 (min UTS 450 MPa/65,000 p.s.i.)

(d) Significant delay between completion of welding and subsequent PWHT The vessel code data book indicates that the completed weld was subjected to NDE examination by MT and UT on

15 November 1990 but the vessel was eventually only post weld heat treated some 2 months later on 11 January 1991

2.5 Conclusions and recommendations

Metallographic evidence, supported by a review of the circumstantial evidence leads to the conclusion that the crack in the hydrogen inlet nozzle C1 was a reheat crack which probably initiated

377

during the PWHT of the very heavy section nozzle forging Contributory factors would have been the high stress concentration at the toe of the weld, a degree of tri-axial stress resulting from the thick section geometry in this area, and an over-matching of mechanical properties of the filler metal It is possible that the re-heat cracking may have occurred during service, but it is considered

to be unlikely as the normal operating temperature of 385°C is believed to be too low to initiate the mechanism

The possibility of some delayed cold-cracking in the HAZ cannot be ruled out in view of the protracted delay between welding and PWHT, and if present, would also have assisted in the nucleation of the reheat cracking

Unfortunately it was impractical to weld these materials with a lower-strength filler metal since the lowest strength Cr-Mo filler specified under ASME I1 SFA 5.5 has a minimum UTS of 80,000 p.s.i The following measures were included in the repair procedure for the nozzle:

(a) Pre-heat temperature was increased from 150°C as used in the original fabrication, to 200-

(b) Pre-heat was maintained until PWHT was carried out

(c) The weld toe was dressed to a generous transition radius

It was noted that much of the welding contour and weld surface finish was rather rough, as evidenced in the photographs Particularly noteworthy, however, was the complete absence of any

“rat holes” at the tri-axial joints between gussets, shell and horizontal ring (Fig 7 shows typical detail of this area) The presence of “rat holes” at such intersections is considered to be normal, good fabrication practice in order to reduce or eliminate the complex tri-axial stresses that are otherwise imposed on these members It was ascertained that this detail had indeed been clearly specified on the relevant fabrication drawings

1‘ L - x

Fig 6 Typical cracks found a t external support welds

3.2 Metallurgical examination

Three of the more easily accessible cracks were chosen for microstructural replication, polished

to a 1 pm diamond finish and etched with 2% nital

Examination of the replicated micrographs in the laboratory showed that in each case the cracks followed the martensitic region of the heat affected zone Figures 8 and 9 are “panorama” micrographs across the HAZ from parent metal to weld metal in order to illustrate more clearly the

location of the cracks in relation to the microstructure

A significant feature is that all three of the cracks were filled with an oxide phase, which is shown

particularly clearly in Fig 10 This is indicative that the cracks were exposed to an elevated

temperature oxidizing environment after initiation [2, 31

3.3 Discussion

All three of the cracks examined above are typical of heat-affected-zone cracking which is also

referred to as “hydrogen-induced cracking”, “weld cracking”, “delayed cracking” or “underbead cracking”

379

Fig 7 Typical detail of gusset welds showing complete absence of “rat hole” at the junction of the three members

Cracks in the HAZ are usually sited either at the weld toe, the weld root, or in an underbead position The interaction between the factors responsible for cracking and the ways in which control over them may be achieved are discussed below [4]

3.3.1 Hydrogen level During welding, hydrogen is absorbed by the weld pool from the arc atmosphere During cooling, much of this hydrogen escapes from the solidified bead by diffusion but some also diffuses into the HAZ and the parent metal The amount which does so depends on several factors such as the original amount absorbed, the size of the weld, the decreasing solubility, and the time-temperature conditions of cooling

In general the more hydrogen present in the metal the greater the risk of cracking Control over this hydrogen level may be achieved either by minimising the amount initially absorbed or by ensuring that sufficient is allowed to escape by diffusion before the weld cools Frequently a combination of both measures provides the best practical solution

3.3.2 Stress level Stresses are developed by thermal contraction of the cooling weld and these stresses must be accommodated by strain in the weld metal The presence of the hydrogen appears

to lower the stress level at which cracking will occur In rigid structures the natural contraction stresses are intensified because of the restraint imposed on them by the different parts of the joint These stresses will be concentrated at the toe and root of the weld and also at notches constituted

by inclusions and other defects The higher degrees of strain which result produce higher risks of cracking for a given microstructural hardness

The stress acting upon a weld is a function of weld size, joint geometry, fitup, external restraint, and the yield strengths of the plate and weld metal

3.3.3 Microstructure The heat affected zone (HAZ) of the parent metal adjacent to the weld is raised to a high temperature during welding and subsequent rapid cooling (quenching) by the surrounding parent metal causes hardening by transformation to martensite Close to the fusion boundary the HAZ is raised to a sufficiently high temperature to produce a coarse grain size This high temperature region, because of its coarse grain size is not only more hardenable but also less ductile than regions further from the fusion boundary It is thus the region in which the greatest risk of cracking exists As a general rule, for both carbon-manganese and low alloy steels, the harder the microstructure the greater is the risk of cracking Soft microstructures can tolerate more hydrogen than hard before cracking occurs

3.3.4 Temperature Hydrogen embrittlement of ferritic steels occurs only at low temperatures,

380

381

I,

Fig 10 Detail of oxide filled crack in martensite area of HAZ, support G1 (magnification x 100)

close to ambient It is therefore possible to avoid cracking in a hard, i.e susceptible, microstructure

by maintaining it at a sufficiently high temperature, either until hydrogen has diffused away or until the microstructure is softened by tempering, to render it less susceptible This principle is employed

in multipass welding and in post-weld heat treatments

An increase in temperature increases the rate of diffusion of hydrogen and thus accelerates its removal from the weld Any measure which slows down the weld cooling rate is therefore helpful

in reducing the hydrogen level Preheat, for example, by slowing the cooling rate, not only softens the microstructure but also helps hydrogen to escape As a result, higher hardness levels can be tolerated without cracking than if preheat had not been used

For welds in those steels with hardenability so high that soft microstructures cannot be produced

at all, and where preheat cannot remove sufficient hydrogen, (such as the Cr-Mo steels) a weld interpass temperature, or a post-weld heating temperature, high enough to avoid cracking must be held for a sufficiently long time to allow hydrogen to diffuse away before the weld cools

3.4 Fabrication considerations

During fabrication of the supports of the catalyst reduction reactor, details of possible hydrogen sources and measures taken to effectively diffuse out the hydrogen are not known Factors which would exacerbate the tendency to heat affected zone cracking, however would be the complex tri- axial stresses acting on the gusset/shell supporting welds due to the absence of the specified “rat holes”, and the suspected delay between welding and subsequent post-weld heat treatment as reported in Section 2.4.l(d) If this delay was also applicable to the fabrication of the support ring

it would have been essential to carry out an intermediate heat treatment to diffuse out hydrogen, as discussed above

The oxidation of the cracks as shown in Fig 10 can only have occurred in the Cr-Mo steel by exposure to an oxidising environment at a temperature of at least 550°C [2, 31 This is well above the vessel operating temperature of 385°C and therefore indicates that oxidation of the cracks most probably occurred during post-weld heat treatment, or possibly during some subsequent undocumented welding operation

3.5 Conclusions

It is concluded that the cracking found in the catalyst reduction reactor supports is due to heat affected zone cracking, otherwise known as “hydrogen-induced cracking” Significant contributory factors are believed to be the complex tri-axial stresses imposed due to the absence of the specified

382

“rat holes”, and a suspected delay between the completion of welding and subsequent post-weld heat treatment Oxidation of the cracks leads to the conclusion that they were most probably present before post-weld heat treatment was camed out [Z, 31

4 POSTSCRIPT

In both of the above cases it is evident that there was considerable debate over whether the defects were original fabrication defects, or whether they occurred in service These are real, practical issues which often have contractual as well as technical implications, and for the failure investigator it is sometimes difficult to be dogmatic either way, as he is rarely in possession of all the facts at the time the investigation is carried out If the above cases had been entirely fabrication related then the question arises as to why they were not detected during the final release inspection or during the pre-commissioning inspection?

The metallographic evidence which was presented builds a strong case for the defects having originated at the fabrication stage The absolute truth of the matter, however, probably lies some- where between the two extremes, in that the initial defects may not have been easily detectable

during routine inspection but probably propagated during service to a size that only subsequently became readily detectable

An interesting corollary to the saga of the external support leg cracking became apparent shortly after the above investigation was undertaken A design review of the reactor installation revealed that radial thermal expansion of the support brackets had been restricted by grouting in the hold- down bolts in the slotted holes, which would probably have exerted considerable stress on the existing HAZ cracks, and caused crack propagation The problem was rectified by removing the grouting, and installing a stainless steel foot plate under the bracket, resting on a bronze support plate to reduce friction

In the final analysis, it must be recorded that weld repair of all of the defects was carried out meticulously under close supervision during August/September 1992, and that numerous subsequent inspections carried out on the vessel have failed to reveal any re-occurrence of the cracking

’

REFERENCES

1 Laneaster, J F., Merdfurgy of Welding, 1987, Allen & Unwin, pp 206-209

2 ASM Metals Hundbmk, Vol I, 10th edn, “Properties & Selection: Irons, Steels & High-Pdomance Alloys”, 1990,

3 Smithens, Meids Reference Book, Butteworth, 1983, p 31-13

4 Coe, F R., We/dhg Steels Withour Hydrogen Cracking, The Welding Institute, 1973

p 617

Failure Analysis Case Studies I1

D.R.H Jones (Editor)

HYDROGEN CRACKING OF FERRITIC STAINLESS

SHINJI KONOSU* and T S W O S H I NAKANIWA

Department of Mechanical Engineering, Ibaraki University, 4-12-1 Nakanarusawa, Hitachi 3 16, Japan

(Receiced 28 January 1998)

Abstract-A ferritic stainless steel (SUS436L), which was subjected to various kinds of reduction ratio was precharged with hydrogen at 40°C in 15% HCI solution by employing galvanic reaction with zinc Tensile tests were performed in air at room temperature on both uncharged and charge specimens Finite element method (FEM) analyses were carried out to obtain strain a t panel comers under various different internal radii in a thermal storage tank when it was subjected to internal pressure As a result, it was found that the

value of internal comer radius/thickness of the panel ( R / t ) should be more than about 2 in order to prevent hydrogen embrittlement cracking Q 1998 Published by Elsevier Science Ltd All rights reserved

Keywords: Cleavage fracture, embrittlement, heat-exchanger failures, hydrogen-assisted cracking tanks (fail- ures)

I INTRODUCTION

Ferritic stainless steel is frequently used in the manufacture of waterheating appliances due to its excellent formability and its extremely high resistance to such shortcomings as pitting and stress corrosion cracking associated with austenitic stainless steel However, because ferritic stainless steel possesses high hydrogen embrittlement susceptibility, investigations are being conductcd [I] on the effect of hydrogen on the mechanical properties of the material Meanwhile, numerous accidents thought to be due to hydrogen embrittlement are occurring at the corner portions of cold-bent ferritic stainless steel (SUS436L) panels used in thermal storage tanks It is believed that fracture elongation in hydrogen-charged ferritic stainless steel is largely due to the influence of cold working and, further, that the strain occurring on the inside of bent portions during water proof tests is attributable to the influence of the inner radius of the bent portion

Hence, using ferritic stainless steel in the current series of investigations, the influence of cold working on fracture elongation in hydrogen-charged specimens was determined experimentally and the limits of hydrogen embrittlement cracking on the inside of cold-bent portions were studied and clarified by analyses employing the finite element method

2 FAILURE OF THERMAL STORAGE TANK

Figure 1 shows a portion of a thermal storage tank assembly measuring 4 m in height, 3 m in width and 10 m in length It consists of 4 banks of tanks stacked vertically, with three rows arranged

in the longitudinal direction The tank panel is made from ferritic stainless steel (SUS436L), with the corner portions being bent by cold forming, as shown in Fig 2 Hot-dip zinc-coated steel tubes are laid inside the tank Coolant is passed through these tubes to freeze the water in the tank during the night and the heat of melting is utilized during the day by means of air conditioners

After water proof tests were conducted at the respective pressures concerned (Case 1: hydraulic pressure 3.52 x IO-’ MPa, Case 2: 6.36 x IO-’ MPa), cracks were found in the corner portions (inner radius R = I 62 mm) The appearance of the cracked portion in Case 2 is shown in Fig 3 It can be seen that the crack has propagated from the inside of the panel toward the outside

*Author to whom correspondence should bc addressed

Reprinted from Engineering Failure Analysis 5 (4), 323-33 1 (1998)

crack, which had propagated transgranularly in a brittle manner, was a cleavage fracture exhibiting features typical of hydrogen embrittlement cracking Accordingly, it is considered that the following

reactions (galvanic corrosion [ 2 ] ) between zinc and ferritic stainless steel occurred in the tank during acid washing at the fabrication stage, causing the stainless steel to become hydrogen-charged

anode reaction: M + M”+ +ne cathode reaction: nH+ +ne- -+ nH The cracking is believed to have occurred in the subsequent water proof testing when the material was subjected to tensile strain

Cold working becomes more pronounced as the inner radius of the corner portion shown in Fig

2 becomes smaller and, consequently, hydrogen embrittlement susceptibility is thought to increase Further, it is conceivable that the amount of tensile strain occurring on the inside surface during water proof testing also increases with smaller inner radii

385

N

-L Fig 2 Enlarged sketch of corner of thermal storage tank panel

387

ranging from 10% to 40% Then JIS 14 flat test pieces (JIS 22201) with a parallel portion of 35

mm were prepared and used in the tests

The test pieces were precharged with hydrogen by immersing the test pieces, together with zinc (2 : 1 surface area ratio), in a 15% HCI aqueous solution for 24 h at 4 0 T , thereby forming a galvanic

coupling between the test piece and the zinc As a result of this procedure, the charged hydrogen

content of the steel of the 30% reduction ratio test piece was 2.3 ppm, which is virtual2y the same

as that of the bent portion of the actual equipment The test pieces were subjected to tensile tests at

ambient temperature at constant cross-head speed (a strain rate of 4.8 x sec-', which is a sufficiently low speed to evaluate hydrogen embrittlement [ 11) immediately after being hydrogen-

charged and the surfaces and fractures of the test pieces were then examined

4 EXPERIMENTAL RESULTS AND DISCUSSIONS 4.1 Strain at bent portion

applying the following formula [3]:

Strain ci on the inner side of a panel which has been subjected to bending was approximated by

where R = inner radius of bent portion, t = plate thickness and AI = distance from inner surface to neutral axis

Then, by substituting the values applicable to the bent portion of the actual equipment (R = 1.62

mm, t = 2 mm) in the above formula, it is estimated that the compressive strain occumng at the

inner surface of the bent portion was about 31% The results of hardness tests carried out on

specimens possessing the various reduction ratios due to rolling are given in Fig 6 Here, the hardnesses of specimens whose reduction ratio is of the same magnitude as the compressive strain

(about 31%) on the inside surface resulting from the bending process are Hv = 24&280 This is

about the same as the hardness (Hv = 250-300) of the inner surface attributable to the bending

process

4.2 Tensile properties

Using material that had been cold-rolled at various reduction ratios, tensile tests were conducted

on hydrogen-charged test specimens to study the effect of reduction ratio on rupture elongation

Hardness at inner comer of panel

388

Reduction ratio (9%)

Fig 7 Influence of reduction ratio on rupture elongation of specimens

The results are given in Fig 7 With both the hydrogen-charged and hydrogen-uncharged specimens,

it was found that rupture ductility decreased with increased reduction ratios (increased magnitude

of cold working) Furthermore, rupture elongation decreased still further in the hydrogen-charged specimens, as compared with the hydrogen-uncharged specimens

A chart showing the stress-strain relationship in the case of a reduction ratio of 30% is given in Fig 8 From this, it can be seen that the rupture elongation of hydrogen-charged specimens is very markedly reduced The results of obscrvations of thc cross scctional microscopic structure in the vicinity of ruptures in a 30% reduction ratio specimen and related scanning electron micrographs are shown in Fig 9 The hydrogen-uncharged specimen had ruptured in a ductile manner and the entire surface of the fracture is dimpled (Fig 9(a)) On the other hand, cleavage fractures had occurred in the hydrogen-charged specimen (Fig 9(b)), indicating the large influence of hydrogen