fatigue crack growth assessment in underwater …...the fatigue crack growth rate data, da/dn, were...

8
WELDING RESEARCH AUGUST 2017 / WELDING JOURNAL 287-s Introduction The increasing demands for global energy continue to push oil companies into deeper waters in search of eco- nomically recoverable oil and gas re- serves (Refs. 1, 2). Marine structures may be damaged by environmental and mechanical factors resulting in unsafe conditions. In general, it is too expensive or even impossible to move the damaged structures to land (Ref. 3). Such structures must then be re- paired. Additionally, if structural in- tegrity is to be maintained, repair welds must be designed and installed with the same rigor applied to initial construction. These welds must be made underwater. Underwater welding techniques can be subdivided into two main types: wet and dry. In contrast to dry weld- ing, the wet weld requires no compli- cated setup. The process simply makes it possible to weld even the most geo- metrically complex structures (Refs. 2, 4). Thus, wet welding is much more cost effective, which makes it the obvi- ous choice whenever possible. Howev- er, since wet welding is performed at ambient pressure with no physical bar- rier between the water and the weld- ing arc, the increased pressure makes the welding arc unstable. Moreover, as the contents of oxygen and hydrogen present in the weld pool are higher due to water dissociation in the elec- tric arc, more porosity and hydrogen- induced cracks are produced during underwater wet welding (Refs. 5, 6). Porosity is one of the main defects in underwater welding, along with the loss of alloying elements and the presence of large amounts of nonmetallic inclusions (Refs. 1, 5). As a result, the yield stress, strength, ductility, and toughness of the weld are impaired (Refs. 2, 7). The in- creased depth of welding increases the porosity of the weld metal to levels that may be unacceptable for certain applica- tions (Refs. 1, 8). Limited systematic data on me- chanical properties have been ob- tained to date. However, it has been recognized that welds made with un- derwater shielded metal arc welding (SMAW) have mechanical properties that are generally inferior to those of surface welds made in air (Refs. 7, 9). Mechanical properties of underwa- ter wet welds have a strong relation- ship with water depth due to the in- crease in the oxidation of alloying ele- ments and porosity (Ref. 10). Table 1 presents an overview of weld metal mechanical properties as a function of water depth. On the other hand, on cycling load- ing, the behavior of underwater wet weld metals has received less attention and research. In this case, evaluation of fatigue and corrosion fatigue behav- ior becomes of vital importance since this phenomenon of structural degra- dation is present in about 80% of the fails found in offshore installations. Yara et al. (Ref. 14) performed fatigue studies using classical mechanics to determine the fatigue limit of under- water wet welds by obtaining the S-N curve. In this study, it was noticed that the reduction of the fatigue limit for welds was made at a depth of 10 m when compared with welds made in a dry environment. Fatigue Crack Growth Assessment in Underwater Wet Welds The characteristics of fatigue crack propagation of welds produced with underwater wet welding were evaluated BY A. R. ARIAS AND A. Q. BRACARENSE ABSTRACT The characteristics of fatigue crack propagation of welds produced with underwater wet welding were evaluated out of water. Butt joints were produced in a hyperbaric chamber at simulated depths of 10, 60, and 90 m. A gravity welding system was used to deposit a AWS E6013 commercial electrode coated with vinylic varnish. Welding joints were prepared from an A-36 plate, 19 mm thick, with 45-deg V-grooves filled with an av- erage of 18 passes. The fatigue crack growth rate properties in the near threshold and Paris regimes for the weld metal were determined by using compression precracking fol- lowed by load reduction and constant amplitude test procedures. Mechanisms of crack propagation were investigated on the fracture surface by means of a scanning electron microscope and confocal laser scanning microscopy. Lateral surface observations, consid- ering different loading conditions, were made using optical microscopy. The resulting fa- tigue crack growth rates were shown to depend on pore density that varies with the depth at which underwater wet welding was performed. The fatigue crack growth path showed a branched morphology, which is a consequence of the interaction between cracks and pores. The results of this study show that underwater wet welding procedures produced fatigue-resistant weld metal that is adequate for use at low applied stresses in structures, in agreement within design codes. KEYWORDS • Fatigue Crack Growth • Underwater Wet Welding • Shielded Metal Arc Welding (SMAW) • Porosity

Upload: others

Post on 21-Feb-2020

6 views

Category:

Documents


1 download

TRANSCRIPT

Page 1: Fatigue Crack Growth Assessment in Underwater …...The fatigue crack growth rate data, da/dN, were correlated with the im-posed stress intensity factor incre-ment, K, and analyzed

WELDING RESEARCH

AUGUST 2017 / WELDING JOURNAL 287-s

Introduction The increasing demands for globalenergy continue to push oil companiesinto deeper waters in search of eco-nomically recoverable oil and gas re-serves (Refs. 1, 2). Marine structuresmay be damaged by environmentaland mechanical factors resulting inunsafe conditions. In general, it is tooexpensive or even impossible to movethe damaged structures to land (Ref.3). Such structures must then be re-paired. Additionally, if structural in-tegrity is to be maintained, repairwelds must be designed and installedwith the same rigor applied to initialconstruction. These welds must bemade underwater. Underwater welding techniques can

be subdivided into two main types:wet and dry. In contrast to dry weld-ing, the wet weld requires no compli-cated setup. The process simply makesit possible to weld even the most geo-metrically complex structures (Refs. 2,4). Thus, wet welding is much morecost effective, which makes it the obvi-ous choice whenever possible. Howev-er, since wet welding is performed atambient pressure with no physical bar-rier between the water and the weld-ing arc, the increased pressure makesthe welding arc unstable. Moreover, asthe contents of oxygen and hydrogenpresent in the weld pool are higherdue to water dissociation in the elec-tric arc, more porosity and hydrogen-induced cracks are produced duringunderwater wet welding (Refs. 5, 6).

Porosity is one of the main defects inunderwater welding, along with the lossof alloying elements and the presence oflarge amounts of nonmetallic inclusions(Refs. 1, 5). As a result, the yield stress,strength, ductility, and toughness of theweld are impaired (Refs. 2, 7). The in-creased depth of welding increases theporosity of the weld metal to levels thatmay be unacceptable for certain applica-tions (Refs. 1, 8). Limited systematic data on me-chanical properties have been ob-tained to date. However, it has beenrecognized that welds made with un-derwater shielded metal arc welding(SMAW) have mechanical propertiesthat are generally inferior to those ofsurface welds made in air (Refs. 7, 9). Mechanical properties of underwa-ter wet welds have a strong relation-ship with water depth due to the in-crease in the oxidation of alloying ele-ments and porosity (Ref. 10). Table 1presents an overview of weld metalmechanical properties as a function ofwater depth. On the other hand, on cycling load-ing, the behavior of underwater wetweld metals has received less attentionand research. In this case, evaluationof fatigue and corrosion fatigue behav-ior becomes of vital importance sincethis phenomenon of structural degra-dation is present in about 80% of thefails found in offshore installations.Yara et al. (Ref. 14) performed fatiguestudies using classical mechanics todetermine the fatigue limit of under-water wet welds by obtaining the S-Ncurve. In this study, it was noticed thatthe reduction of the fatigue limit forwelds was made at a depth of 10 mwhen compared with welds made in adry environment.

Fatigue Crack Growth Assessment in Underwater Wet Welds

The characteristics of fatigue crack propagation of welds produced with underwater wet welding were evaluated

BY A. R. ARIAS AND A. Q. BRACARENSE

ABSTRACT The characteristics of fatigue crack propagation of welds produced with underwaterwet welding were evaluated out of water. Butt joints were produced in a hyperbaricchamber at simulated depths of 10, 60, and 90 m. A gravity welding system was used todeposit a AWS E6013 commercial electrode coated with vinylic varnish. Welding jointswere prepared from an A­36 plate, 19 mm thick, with 45­deg V­grooves filled with an av­erage of 18 passes. The fatigue crack growth rate properties in the near threshold andParis regimes for the weld metal were determined by using compression precracking fol­lowed by load reduction and constant amplitude test procedures. Mechanisms of crackpropagation were investigated on the fracture surface by means of a scanning electronmicroscope and confocal laser scanning microscopy. Lateral surface observations, consid­ering different loading conditions, were made using optical microscopy. The resulting fa­tigue crack growth rates were shown to depend on pore density that varies with thedepth at which underwater wet welding was performed. The fatigue crack growth pathshowed a branched morphology, which is a consequence of the interaction betweencracks and pores. The results of this study show that underwater wet welding proceduresproduced fatigue­resistant weld metal that is adequate for use at low applied stresses instructures, in agreement within design codes.

KEYWORDS • Fatigue Crack Growth • Underwater Wet Welding • Shielded Metal Arc Welding (SMAW) • Porosity

ARIAS 2016108.qxp_Layout 1 7/11/17 3:58 PM Page 287

Page 2: Fatigue Crack Growth Assessment in Underwater …...The fatigue crack growth rate data, da/dN, were correlated with the im-posed stress intensity factor incre-ment, K, and analyzed

WELDING RESEARCH

WELDING JOURNAL / AUGUST 2017, VOL. 96288-s

The linear elastic fracture mechanics(LEFM) was applied to characterize thefatigue behavior of steels and theirwelded joints applied in the offshore in-dustry by means of the da/dN – Kcurve. This curve relates the fatiguecrack propagation rate with the stressintensity cycling factor. In the particu-lar case of underwater wet welding, thelarge heterogeneity present in the weldcauses the behavior of the crack growthrate to be different from welds madeout of water (Refs. 9, 15). Table 2 presents an overview on Cand m, empirical constants called pow-er law coefficient and Paris exponentas a function of underwater weldingconditions. In this report, the fatigue crackpropagation behavior in weld metalproduced by underwater wet weldingwas experimentally studied with theconsideration of the effect of waterpressure during welding. Particular at-tention was placed on the influence ofporosity on crack growth behavior. Itis important to understand the stagesof the fatigue crack growth process toallow for the development of a per-formance optimization criteria.

Experimental Procedures

The Wet Welding Procedure

Butt joints in groove weld testpieces were produced in a hyperbaricchamber capable of simulating under-water depths up to 200 m. A gravitywelding system, commonly used tomechanize the SMAW process, wasused to prepare the test pieces insidethe chamber — Fig. 1. The consumable employed in thisresearch was an AWS governing speci-fication E6013 commercial electrodeof 3.25 mm diameter and a 350-mmlength coated with vinylic varnish. Ru-tile-based electrodes are generally pre-ferred due to their good operability.Weld joints were prepared from a 19-mm-thick ASTM A-36 steel plate. V-grooves of 45 deg were prepared on160 mm × 250 mm plates. Backingbars were used for the 5.0-mm rootopening. To fill the groove, an averageof 18 passes was made. All beads be-gan at the same side of the groove.One weld was made for each condi-tion, i.e., current, travel speed, etc.

Table 1 — Mechanical Properties of Underwater Wet Welds as a Function of Depth (Reported in the Literature)

Reference Depth Yield Strength Ultimate Tensile Charpy Jic (m) (MPa) Strength (MPa) (J, 0oC) (kJ/m2)

E6013 E7014 E6013 E7014

Bracarense et al. 0.5 362 – 513 – 40.5 – (Ref. 11)

Dexter 10 507 – 556 – 43 120 (Ref. 7)

Santos et al. 10 – 488 – 506 44 – (Ref. 12) Rowe et al. 10 455 – 510 – 42 – (Ref. 13)

Dexter 60 402 – 451 – 45 35 (Ref. 7)

(A) The table shows values of weld metal mechanical properties.

Fig. 1 — A — Hyperbaric chamber; B — gravity welding system.

Table 2 — Paris Law Constant as a Function of Underwater Welding (Reported in theLiterature)

Reference Welding Condition C m

Arias et al. (Ref. 15) Underwater wet weld made at 10 depth 1.0 x 10–12 5.3 Underwater wet weld made at 60 depth 7.0 x 10–14 6.0 Matlock (Ref. 9) Underwater wet weld 2.7 x 10–15 6.6 Underwater dry weld 3.0 x 10–9 3.3

Table 3 — Welding Condition According to Water Depths (Welding Parameters)

Water Depth (m) Welding Current (A) Welding Voltage (V) Welding Speed (mm/s)

10 160 24–28 3.0–3.4 60 170 23–25 3.2–3.6 90 180 22–26 3.3–3.7

A B

ARIAS 2016108.qxp_Layout 1 7/11/17 3:58 PM Page 288

Page 3: Fatigue Crack Growth Assessment in Underwater …...The fatigue crack growth rate data, da/dN, were correlated with the im-posed stress intensity factor incre-ment, K, and analyzed

WELDING RESEARCH

AUGUST 2017 / WELDING JOURNAL 289-s

Table 3 shows the welding conditionaccording to the simulated waterdepth. The welds were simulated atdepths of 10, 60, and 90 m.

Fatigue Test Procedure

The standard compact tensile (CT)specimens (shown in Fig. 2) were ap-plied to test the fatigue crack growthrate and threshold, and the crack tipwas located in the fusion zone. The po-sition where the test specimen was ex-tracted from the weld joint and the re-sulting CT specimen are also shown inFig. 2. The fatigue crack growth rate data,da/dN, were correlated with the im-posed stress intensity factor incre-ment, K, and analyzed according tothe Paris crack growth expression inEquation 1.

da⁄dN = C(K)m (1)

where C and m are material constants,which are determined by experimentaldata fitting and dependent on the testconditions. The stress intensity factorincrement K for CT specimens couldbe calculated according to Equation 2.

(0.886 + 4.64- 13.322

+ 14.723 - 5.64) (2)

where B is the specimen thickness, Wis the specimen width, P is the ap-plied loading amplitude, is the fa-tigue crack coefficient defined by /W,and is the crack length. The da/dNvs. K plot generally has three regions:I, II, and III. Regions I and III are thenear threshold and the rapid crackpropagation regions, respectively. To generate the fatigue crackgrowth (FCG) rate data under constantamplitude loading in the thresholdand near-threshold regimes, reducingload-history effects, compression-compression precracking methods de-veloped by Newman et al. (Ref. 16)were used. Using this procedure,prenotched specimens were cycled un-der compression-compression loadingto produce an initial fatigue crack,which naturally stops growing. The ap-plied load (Pmax = –892N and Pmin

= –8920N) for precrack growth was cho-sen to generate the stress ratio R = 10 ata frequency of 30 Hz. The average sur-face length of the precracks obtainedfrom all the tested specimens was about1 mm.

All the FCG experiments were conducted using a uniaxial servo-hydraulic load frame. The load capacityof the load cell was ±25 kN. The cracklength was monitored by the compli-ance method, extensometer with agauge length of 9 mm and travel of 4mm was used to measure the cracklength. A load rate (the ratio betweenthe minimum and maximum loads) of0.1 was used for all conditions tested. The FCG experiments were per-formed in the Paris regimen by applyinga sinusoidal waveform with constantload amplitude and a frequency of 20Hz. Threshold and near-threshold test-ing was performed using a method ofstandard load reduction described inASTM E-647 for threshold determina-tion. Initial starting load levels werecarefully selected to ensure that growthrates for cracks from the crack starternotch were less than 10–5 m/cycle at thestart of the load reduction test, as re-quired by the standard. A load reductionrate of C = –0.08 mm–1 was maintainedin all K-decreasing tests.

Microstructure Analysis

Metallographic analysis was carriedout in different regions of the CT spec-imen made after the fatigue crackpropagation test by using load reduc-tion. Figure 3 shows the position ofthe sample extraction for metallo-graphic testing. Plane A, located onthe specimen’s lateral face, was used toanalyze the fatigue crack growth pathwith different levels of the stress in-tensity factor amplitude (K). Crosssection B was located in a region

�K = �PB W

2+�( )1��( )3/2

Fig. 2 — A — Position of sample extraction for tests; B — a photograph showing weld location in a CT specimen.

Fig. 3 — Positions of sample extractionsfor metallographic tests.

A B

ARIAS 2016108.qxp_Layout 1 7/11/17 3:58 PM Page 289

Page 4: Fatigue Crack Growth Assessment in Underwater …...The fatigue crack growth rate data, da/dN, were correlated with the im-posed stress intensity factor incre-ment, K, and analyzed

WELDING RESEARCH

WELDING JOURNAL / AUGUST 2017, VOL. 96290-s

where the crack propagated underhigh levels of K, and in cross sectionC, near-threshold fatigue crack growthconditions were found. Fatigue fractured morphology ofvarious specimens was examined byscanning electron microscopy (SEM)and by confocal laser scanning mi-croscopy (CLSM).

Results and Discussion

Underwater Wet WeldCharacterization

Figure 4 shows macrographs of thejoint cross section of underwater wetwelds made at depths of 10, 60, and 90m. As observed in previous studies,(Refs. 5, 8) an increase in hydrostaticpressure in underwater wet weldingincreases the amount of pores in theweld metal.

Porosity levels were obtained fromimages of the cross section of the weldfor each condition using an image pro-cessing software (ImageJ). Firstly, thetotal area of the weld was measured,followed by the total area of the pores.Results indicated an increase in poros-ity levels from 2.0% in welds made at10 m to 6.8% at 60 m, and to 11.5% at90 m. The tested welds were all in therange accepted by the AWS D3.6:2010standard (Ref. 24). Figure 5 shows a three-dimensionalstereography of the optical microstruc-ture of the multipass underwater wetweld obtained at a 60 m depth. In thisillustration, one can see the possible re-gion to be affected by fatigue crackpropagation generated in the test. Fac-tors such as the microstructural differ-ences, product of multipass, and coldcracking (hydrogen cracks), which gen-erally decrease with depth for rutile-

type electrodes and formed transverseto the weld direction as referenced bySantos (Ref. 12) and Bracarense (Ref.11), can be observed. The size and uni-form distribution of the pores, as wellas the different mechanical propertiesin the singularity zone, make the mech-anisms of fatigue crack propagation inunderwater wet welds more complexthan for homogeneous materials.

Fatigue Crack Propagation

The fatigue crack growth rate dataat Paris regime (Region II) for under-water wet weld metal in different con-ditions studied are shown in Fig. 6.The appropriate crack growth equa-tions for the high K linear data areincluded along with data from theBritish Standard BS 7910 (Ref. 25),which represents the simplified fa-tigue law for welded steels.

Fig. 4 — Macrographs of welds used for the fatigue study: A — Underwater wet weld made at a 10 m depth; B — underwater wet weldmade at a 60 m depth; C — underwater wet weld made at a 90 m depth.

A B C

Fig. 5 — Three­dimensional stereography of the optical mi­crostructure of the multipass underwater wet weld obtainedwithin a 60 m depth.

Fig. 6 — Fatigue crack growth rate data in the Paris regime ofthe underwater wet weld metal performed at 10 and 90 mdepths.

ARIAS 2016108.qxp_Layout 1 7/11/17 3:58 PM Page 290

Page 5: Fatigue Crack Growth Assessment in Underwater …...The fatigue crack growth rate data, da/dN, were correlated with the im-posed stress intensity factor incre-ment, K, and analyzed

WELDING RESEARCH

AUGUST 2017 / WELDING JOURNAL 291-s

Several important observations areobtained from an analysis of the fa-tigue crack growth rate data presentedin Fig. 6. First, in comparison to datafrom the British Standard BS 7910and underwater dry weld metal datareported in literature (Ref. 9), the twoexperimental welds (underwater wetwelds made at a 10 and 90 m depth)exhibited lower growth rates for lowvalues ofK. This observation is con-sistent with previous studies on fa-tigue in wet welds (Ref. 9). In addition, all wet welds exhibiteda higher sensitivity of the fatigue crackgrowth rate to K, as evidenced by themagnitude of the exponent m in Equa-tion 1 (m = 2.88 for British StandardBS 7910; m = 3.3 for an underwaterdry weld; m = 4.1 for an underwaterwet weld made at a 10 m depth; and m= 5.6 for an underwater wet weldmade at a 90 m depth). The presenta-tion in terms of log C and m calculatedfrom Equation 1 are presented in Fig.6. The m values for an underwater wetweld increased with increasing porosi-ty in the weld metal. Figure 7 summarized the experi-mentally obtained relationship be-tween the crack growth rate, da/dN,and the stress intensity factor range,K, for all testing conditions. An ar-row in the figure denotes the apparentthreshold stress intensity factor. Thedata from the British Standard BS

7910 for sim-plified fatiguelaw for weldedsteels recom-mended fatiguecrack growththreshold val-ues for assess-ing weldedjoints are in-cluded alongwith the fig-ures. The fatiguecrack propaga-tion behaviorof underwaterwet weldsshows goodagreementwith designcodes based onfitness-for-service and flaw acceptabil-ity concepts, according to BS 7910,that define guidelines for the analysisof components that have crack-like de-fects and the expected material prop-erties for several applications. It can be identified in Fig. 7 that thefatigue crack propagation rate exhibitsthree featured stages, namely Stage I(low value K), Stage II (intermediatevalue K), and Stage III (high value K). The results presented in Figs. 6 and7 demonstrate that the porosity pres-ent in the weld metal decreased fa-

tigue crack growth at low and interme-diate stress intensities relative to itsexpected values in low, porosity-freeweld metals. This behavior changedfor stress intensity ranges above ap-proximately 40 MPa√m, in which crackgrowth rate tended to be higher poros-ity underwater wet weld metals. Exam-ination of optical images of the fatiguecrack propagation path of underwaterwet welds showed samples containingwelds made at 90 m have more tortur-ous crack paths than the samples con-taining welds made at 10 m.

Fig. 7 — Fatigue crack growth rate data in the Paris and near­threshold regimes with stress ratios of 0.1.

Fig. 8 — Fatigue crack path deviation due to porosity in underwaterwet welds made at a 90 m depth of the specimen’s lateral surface (R= 0.1, two different magnifications). (Etching with 2% Nital reagent.)

Fig. 9 — Transversal crack path analysis obtained by sectioning thespecimens with underwater wet welds made at a 90 m depth. (Etch­ing with 2% Nital reagent.) A — Crack propagated high levels of K;B — near­threshold fatigue crack growth conditions.

A

B

ARIAS 2016108.qxp_Layout 1 7/11/17 3:58 PM Page 291

Page 6: Fatigue Crack Growth Assessment in Underwater …...The fatigue crack growth rate data, da/dN, were correlated with the im-posed stress intensity factor incre-ment, K, and analyzed

WELDING RESEARCH

WELDING JOURNAL / AUGUST 2017, VOL. 96292-s

A crack propagation mechanismthat can be inferred from detailedanalysis of many metallographs of thetype shown in Figs. 8 and 9 is given bythe small cracks emanating from thepore and growing toward the principalcrack. In other words, propagation ofthe main crack was partly due to theinitiation and backward growth ofsmall cracks that started at some sur-face irregularity of the pore. This prop-agation mechanism is consistent withthe suggestions in literature for mech-anisms of fatigue crack growth inaustempered ductile iron (Ref. 17). Evidence of this initiation mecha-nism is shown in Fig. 10 through thepore surfaces scanning electron micro-scope analysis. Figure 10A shows irregu-larities of the pore surface due to bub-bles formed in molten metal, which istrapped by the solidification front pro-gressing during the underwater wetwelding process. To guarantee that thepore was not affected by the loading ofthe test, showing that the discontinu-ities found within the pore were theproduct of the weld, the sample temper-ature was lowered to cause a brittle frac-ture generating less deformation. Figure 10B shows a pore that wasintercepted by the dominant fatiguecrack. One can see secondary micro-cracks nucleated on the pore surfacethat has not coalesced with the maincrack front.

Initiation of these cracks is appar-ently activated by high stress levelsproduced when the tip of the maincrack is sufficiently close to the pore.These small cracks eventually coalescewith the main crack front that contin-ues to grow until a new pore isreached. It is important to take intoaccount that several pores can be in-volved in the growth process at differ-ent portions of the crack front, so thatthe average growth rate is affected bythe pore’s size, shape, and distribu-tion. Studies developed by Bouafia etal. (Ref. 18) using the finite elementmodel show the presence of pores inthe weld joint leads to a stress concen-tration. These stresses are more signif-icant when the pores are in the vicinityor very close to each other. Photographic analysis of the FCGtested specimens, shown in Fig. 11,suggested that the number of pores in-tercepted by a dominant fatigue crackwere controlled by K. In other words,increasing numbers of pore and sur-face roughness increased the values ofK. From Fig. 11, it is possible to ob-serve that at values of K above 20MPa√m, the dominant crack inter-cepts a greater number of pores. How-ever, at values of K lower than the 20MPa√m, fewer events are observed. Itsuggests the increase in size of themonotonic crack tip plastic zone is as-sociated with increasing K. More

porosity will be encompassed by theplastic zone, effectively giving agreater crack tip. The greater extent of the near-tipplastic zone is likely to increase theprobability of pore and microcrack ini-tiation ahead of the crack tip and, inturn, subsequent deflection of thedominant crack by coalescence events.Therefore, it is reasonable to proposethat larger values of K result in largerplastic zone volumes that can affect alarger number of pores and result inan advancing dominant crack to “seekout” pores. Fracture surfaces were investigatedby quantitative confocal laser scanningmicroscopy (CLSM). Analysis of manyimages (Fig. 12) showed increasedcrack deflection and fracture surfaceroughness and consequently may de-crease global crack growth rates byroughness-induced closure mecha-nisms and shielding for low and inter-mediate stress intensity. Crack branching was observedalong the crack path in all the samplestested. Crack branching was more se-vere in samples with the underwaterwet weld made at a 90 m depth, asshown in Figs. 8 and 9. Crack branch-ing takes place due to the crack porosi-ty interaction. The crack front appearsto connect the pore along its path andabsorbs more energy if the crack takesa more zig-zag-shaped morphology. Further, the dominance of rough-ness-induced closure at low and inter-mediate stress intensities will also addto the decrease in fatigue crack growthrate of the weld metal. In addition tothe zig-zag fracture, the secondarycracks associated with the propagatingmain crack, observed in the weld metal(Figs. 8 and 9), would also add to de-creasing the stress intensity at the prop-agating main crack tip. Also, the ob-served zig-zag fracture path in the weldmetal could have resulted from its inho-mogeneous microstructure having den-dritic and interpass boundaries. Manyinvestigators (Refs. 19–21) observed animproved fatigue crack growth resist-ance of weld metals due to rough frac-ture surfaces leading to tortuous crackpaths relative to the base metal. The secondary crack that branchedfrom the main crack and the pore areshown in Figs. 8 and 9. After reachinga certain length, the secondary fatiguecrack propagation stopped. This prob-

Fig. 10 — SEM fractographs showing: A — A pore that was not reached by the main fa­tigue crack (free of load); B — a pore that was divided by the main fatigue crack.

A

B

ARIAS 2016108.qxp_Layout 1 7/11/17 3:58 PM Page 292

Page 7: Fatigue Crack Growth Assessment in Underwater …...The fatigue crack growth rate data, da/dN, were correlated with the im-posed stress intensity factor incre-ment, K, and analyzed

WELDING RESEARCH

AUGUST 2017 / WELDING JOURNAL293-s

ably occurred when these small crackssimultaneously propagated beside themain crack. The available elastic ener-gy for the propagation of the maincrack was reduced because of the cre-ation of a larger cracked surface, thusreducing the general rate of crackpropagation. Additionally, it is sug-gested in the literature that when sec-ondary cracks coalesce with the maincrack, there is a sudden increase in thestress intensity factor, and the crackgrowth rate recovers (Ref. 22). Themechanism represented may explainthe variable behavior of fatigue crackgrowth rate in lower values of K pre-sented in underwater wet weld metalsas shown in Fig. 7. This figure shows amore marked variation in the apparentKth values on the conditions studiedin this work. Wet welds with higherporosity have higher near-thresholdfatigue values product of the deflec-tion of the crack plane out of theMode I plane due to the crack-porosityinteraction mechanism. In addition,the reduction of nucleation events ofsmall cracks from pores can cause apossible block of the main crack. For high values of K, the pores de-creased the total load-bearing area andincreased the local stress at the cracktip. Correspondingly, the growth ratesat high K were greater in porouswelds. The damage processes in theplastic zone were caused by a highamount of microcracks emerging atthe pores in the whole plastic zone.Figure 13 shows the presence of alarge damaged zone characterized by amatrix plasticization with evident slipbands especially near the pores, and

secondary cracksare evident. Fatigue crackpropagation condi-tions in underwaterwet welding metalcan be described us-ing the KI (stress in-tensity factor) onlyby considering thisparameter a roughapproximation ofthe effective stressstate ahead of thecrack tip. In theweld metal, it is thereversed plasticzone that activatesthe fatigue propaga-tion micromech-anisms (e.g., slip bands); the monotonicplastic zone and elastic zone interactionwith pores implies the nucleation of sec-ondary cracks. Consequently, a deeperanalysis of the stress state at the cracktip is necessary. For future work, it issuggested that numerical and fracturemechanics modelling of the mechanismof fatigue crack growth in underwaterwet welding take place to provide fur-ther understanding of the phenomena.The numerical tool for the analysis isbased on the boundary element method(BEM), customized for the accurateevaluation of the interaction effects be-tween main crack, microcracks, andpores. The shielding effect of microc-racking (on the macroscopic) can becrack studied using a continuum me-chanics approach. The British Standards institute,Document BS 7910, provides guidance

on some methods for the derivation ofacceptance levels (fitness for service)for defects in fusion welded joints. Theanalysis methods used to assess the ef-fect of porosity on the fatigue per-formance of weldments generally giveconservative and lower-bound fatigueresistance estimates for weldmentswith porosity. The results obtained for the regionof high values of K are in correspon-dence with toughness fracture values re-ported in literature (Refs. 7, 23), show-ing that toughness fracture of underwa-ter wet welding varies with weld depth.

Conclusions The following conclusions can bedrawn from the experimental study offatigue crack growth of underwaterwet weld metal: 1) The fatigue crack propagation

Fig. 11 — Fracture surface of CT specimens after fatigue test.Fig. 12 — A 3D confocal laser scanning microscopy of isomet­ric images of fatigue fracture surface of an underwater wetweld made at a 10 m depth. The red plane is used to obtainthe surface profile of the CLSM.

Fig. 13 — Ductile fracture initiation in the plastic zone.

ARIAS 2016108.qxp_Layout 1 7/11/17 3:58 PM Page 293

Page 8: Fatigue Crack Growth Assessment in Underwater …...The fatigue crack growth rate data, da/dN, were correlated with the im-posed stress intensity factor incre-ment, K, and analyzed

WELDING RESEARCH

WELDING JOURNAL / AUGUST 2017, VOL. 96294-s

behavior of underwater wet weldsshows agreement within design codesbased on fitness-for-service and flawacceptability concepts, such as BS7910, that define guidelines for theanalysis of components that havecrack-like defects and specify materialproperties for several applications. 2) The surface of the pores have ir-regularities, with sharp corners that,in some cases, can constitute immi-nent microcracks that emanate fromthe pore. The vicinity of the maincrack front increases the values of Kin preferential points of crack initia-tion. This causes the propagation ofmicrocracks from the pore toward themain crack, in a propagation sense in-verse to the general crack growth, un-til joining the main crack. 3) Fatigue crack growth path showsa branched morphology, which is theconsequence of interaction betweencracks and pores. 4) The number of pores interceptedby a dominant crack is a factor of K.At values of K above 20 MPam, thedominant crack intercepts a greaternumber of pores. The greater extent ofthe near-tip plastic zones are likely toincrease the probability of microcrackinitiation from pores at the head ofthe crack tip. 5) For higher values of K, thepores decreased the total load-bearingarea and increased the local stress atthe crack tip. Correspondingly, thegrowth rates at high K were greaterin porous welds.

The authors would like to acknowl-edge Capes and PPGMEC of the FederalUniversity of Minas Gerais, for financialsupport of this graduate research work.The authors also thank Dr. P. J. Moden-esi, of DEMET at the Federal Universityof Minas Gerais, for helpful discussionsand editing of the manuscript. The au-thors also acknowledge the CDTN(CNEN) of Minas Gerais, SENAI-CI-MATEC of Bahia, and IFMG-Congonhasfor technical assistance.

1. Liu, S., Olson, D. L., Smith, C., andSpencer, J. S. 1994. Underwater Welding ofMarine Structure. New Orleans, La. 2. Rowe, M. D., and Liu, S. 2001. Recentdevelopments in underwater wet welding.Science and Technology of Welding and Join-ing 6(6): 387–396. 3. Rodriguez, J. E., Perez, F., Liu, S., etal. 2014. Underwater repair of fatiguecracks by gas tungsten arc welding process.Fatigue Fracture Engineering MaterialsStructure (23)37: 637–644. 4. Fydrych, D., Łabanowski, J., and Ro-galski, G. 2013. Weldability of highstrength steels in wet welding conditions.Polish Maritime Research 20(2): 67–73. 5. Pessoa, E. C. P., Bracarense, A. Q.,Liu, S., Guerrero, F. P., and Zica, E. M.2006. Porosity variation along multipassunderwater wet welds and its influence onmechanical properties. Journal of MaterialsProcessing Technology 179: 239–243. 6. Sun, Q. J., Cheng, W. Q., Liu, Y. B.,Wand, J. F., Cai, C. W., and Feng, J. C.2016. Microstructure and mechanicalproperties of ultrasonic assisted underwa-ter wet welding joints. Materials Design103: 63–70. 7. Dexter, R. Fracture Toughness of Un-derwater Wet Welds. SPT 1058. 1990. WestConshohocken, Pa. ASTM International;256–271. 8. Suga, Y., and Hasui, A. 1986. On for-mation of porosity in underwater weldmetal (The 1st report): Effect of water pres-sure on formation of porosity. Transactionsof the Japan Welding Society 17(1): 58–64. 9. Matlock, D. K., Edwards, G. R., Olson,D. L., and Ibarra, S. 1987. Effect of sea wateron the fatigue crack propagation characteris-tics of welds for offshore structures. Journalof Materials Engineering 9(1): 25–34. 10. Gao, W. B., Wang, D. P., Cheng, F. J.,Deng, C, Y., and Xu, W. 2015. Underwaterwet welding for HSLA steels: Chemicalcomposition, defects, microstructures, andmechanical properties. Acta MetallurgicaSinica 28(9): 1097–1108. 11. Bracarense, A. Q., Pessoa, E. C., San-tos, V. R., Monteiro, M. J., Rizzo, F. C., Pa-ciornik, S., Reppold, R., Domingues, J. R.,and Vieira, L. A. 2010. Estudo comparativode eletrodos comerciais para soldagem sub-aquática molhada. Soldagem Inspeção 15(4):325–335. 12. Santos, V. R., Monteiro, M. J., Rizzo,F. C., Bracarense, A. Q., Pessoa, E. C. P., Mar-inho, R. R., and Vieira, L. A. 2012. Develop-

ment of an oxyrutile electrode for wet weld-ing. Welding Journal 91(12): 319-s to 328-s. 13. Rowe, M. D., Liu, S., and Reynolds, T.J. 2002. The effect of ferro-alloy additionsand depth on the quality of underwater wetwelds. Welding Journal 81(8): 156-s to 166-s. 14. Yara, H., Makishi, Y., Kikuta, Y., andMatsuda, H. 1971. Study on the underwa-ter welding (Report 2). Journal of JapanWelding Society 40(8): 748–54. 15. Arias, A. R., and Bracarense, A. Q.Velocidade de propagação de trinca porfadiga de soldas subaquáticas molhadas:Avaliação fora da agua. 2015. Soldagem In-speção 20(4): 403–11. 16. Newman Jr., J. C., and Yamada, Y.2010. Compression precracking methodsto generate near-threshold fatigue-crack-growth-rate data. International Journal ofFatigue 32(6): 879–885. 17. Greno, G., Otegui, J., and Boeri, R.1999. Mechanisms of fatigue crack growthin austempered ductile iron. InternationalJournal of Fatigue 21(1): 35–43. 18. Bouafia, F., Serier, B., Serier, N., andHayat, S. S. 2014. Effect of density andpointed corner degree of pore on localstress in welded structures: Defect in ma-rine structures. ISRN Mechanical Engineer-ing 2014: 1–7. 19. Kusko, C. S., DuPont, J. N., andMarder, A. R. 2004. The influence of mi-crostructure on fatigue crack propagationbehavior of stainless steel welds. WeldingJournal 84(2): 6-s to 15-s. 20. Deng, X., Lu, F., Cui, H., Tang, X.,and Li, Z. 2016. Microstructure correlationand fatigue crack growth behavior in dis-similar 9Cr/CrMoV welded joint. MaterialsScience and Engineering 651: 1018–1030. 21. Tsay, L., Chung, C., and Chen, C.1997. Fatigue crack propagation of D6AClaser welds. International Journal of Fatigue19(1): 25–31. 22. Soboyejo, W. O., Knott, J. F., andWalsh, M. J. 1990. Fatigue crack propaga-tion of coplanar semi-elliptical cracks inpure bending. Engineering Fracture Mechan-ics 37(2): 323–340. 23. Di Lorenzo, R. F., Soares, W. A., andBracarense, A. Q. 2005. Fracture toughnessof ferritic steel underwater wet welded.18th International Conference on StructuralMechanics in Reactor Technology (SMiRT 1). 24. ANSI/AWS D3.6M:2010, Underwa-ter Welding Code. American Welding Socie-ty, Miami, Fla. 25. BS 7910, Guidance on Methods forAssessing the Acceptability of Flaws in FusionWelded Structures. 2000. British StandardsInstitution, London, UK.

ARIEL ARIAS ([email protected]) and ALEXANDRE BRACARENSE are with the Mechanical Engineering Department, Federal Universityof Minas Gerais, Belo Horizonte, Brazil.

Acknowledgments

References

ARIAS 2016108.qxp_Layout 1 7/11/17 3:58 PM Page 294