Structural Integrity Assessment of Ferritic-Austenitic Welded Joints

Transcription

1 Radomir Jovičić Faculty of Mecanical ngineering Innovation Center Aleksandar Sedmak Full Professor Faculty of Mecanical ngineering Zija Burzić Military Institute, Zarkovo Vencislav Grabulov Military Institute, Zarkovo Jasmina Lozanović Faculty of Mecanical ngineering Innovation Center Structural Integrity Assessment of Ferritic-Austenitic Welded Joints In tis paper te influence of cracks on structural integrity of eterogeneos ferrite-austenite welded joints applying te basic fracture mecanics concept was analysed. Te crack driving force was obtained numerically using King s metod, wile materials resistance to crack growt was determined experimentally via J-integral and was given as te J-R curve. In order to find out properties of suc welded joints, plates made of ig alloyed austenite steel were welded to plates made of microalloyed ferrite steel. Te experimental investigation, in combination wit te numerically obtained results, enabled reliable estimation of structural integrity of ferrite-austenite welded joints. Keywords: ferrite-austenite welded joints, fracture mecanics, structural integrity, cracking penomena, weldment eterogeneity. 1. INTRODUCTION Hig strengt low alloyed (HSLA) ferrite steels are increasingly used for welded constructions due to teir ig strengt and tougness, as well as good weldability. Teir typical application is for low temperature pressure vessels, as in te case of CO storage tanks, made by welding of rolled and bended plates. At te same time, since attacments like small pipes are not produced of tese steels, anoter type of steel as to be used, like ig alloyed ausenite steels, also suitable for te low temperature applications. Anyow, suc a combination of steels causes anoter problem, related to te welding of different types of steels. Namely, bot ferrite and austenite steels are prone to different types of cracking, like cold, lamellar, ot cracking, and especially due to teir different pysical and mecanical properties teir eterogeneous welded joints are even more susceptible to all types of cracking. Terefore, in tis paper te influence of cracks on structural integrity of eterogeneous welded joints applying te concept of fittness-for-purpose was analysed, as described by Jovicic in [1,]. To tis end, te fracture mecanics basic concept as been applied, namely te crack driving force as been compared wit te material resistance to te crack growt. Te crack driving force as been obtained numerically using King s metod [3], wile materials resistance to crack growt as been determined experimentally via J- integral and given as te J-R curve.. CRACKING PHNOMNA IN FRRIT- AUSTNIT JOINTS Typical problems wit welded joints made of HSLA ferrite steels are tougness reduction, nil ductility temperature increase, cold cracking and brittle fracture Received: Marc 008, Accepted: May 008 Correspondence to: Dr Aleksandar Sedmak Faculty of Mecanical ngineering, Kraljice Marije 16, 1110 Belgrade 35, Serbia -mail: asedmak@mas.bg.ac.yu can occur. On te oter and, typical problem wit welded joints made of ig alloyed austenite steels are coarse grain structure, σ-pase, ot cracking, intercrystal corrosion and strengt reduction. Furtermore, if eterogeneous joints are to be made of tese two types of steels, it as to be take into account tat austenite steel as 100 ºC lower melting temperature, 0 % iger specific eat, 3 times lower eat conductivity, 5 times iger electrical resistance, 50 % iger termal linear coefficient, and significantly iger solubility of ydrogen. In order to avoid te mentioned problems, te Sceffler s diagram can be used. Basically, tis diagram enables te rigt coice of te filler metal in order to avoid all problems typical for ferrite-austenite welded joints: cold cracking (martensite), ot cracking (austenite), embrittlement (-ferrite) and its combination (A+M), Fig. 1. Practical application is illustrated by points A-D, as follows: point A corresponds to te ferrite base metal HSLA steel NIOVAL 47-ČRN 460 (JUS), point B to te austenite base metal X7CrNiNb18.10 (DIN), so tat point C presents te average content of teir % mixture, being te input value for filler metal coice. Filler metal INOX R 9/9, steelworks Jesenice, was cosen (point D) in suc a way tat te point, corresponding to te weld metal (WM) composition, falls into te wite region, being outside of all critical regions in diagram, as sown in Figure 1..1 Storage tank for CO cracking penomena Te storage tank for CO as been tested recently, during regular inspection period, as described in more details in []. Te storage tank is cylindrical, orizontal, diameter 1600 mm, total lengt 7180 mm and volume 1.5 m 3, Fig.. Its body as been made of HSLA ferrite steel NIOVAL 47-ČRN 460 (JUS) of tickness 14 mm, wile te attacments and revision opening flange ave been made of ig alloyed austenite steel X7CrNiNb18.10 (DIN). Te lowest operating temperature is 55 ºC, te igest pressure 30 bars, testing pressure 39 bars. Faculty of Mecanical ngineering, Belgrade. All rigts reserved FM Transactions (008) 36,

2 Figure 1. Sceffler diagram Figure 3. Pores at te inner side of flange Figure. Te storage tank for CO Different non-destructive-testing (NDT) metods ave been applied, revealing pores at te inner side of flange, Fig. 3, and cracks at te outer side of flange, Fig. 4. Replicas revealed microcracks in eat-affected-zone (HAZ) of microalloyed steel, lengt 1.8 mm, Fig. 5, and cold cracks (longitudinal: lengt 60, 46 & 9 mm, dept cca 3.5 mm; radial: lengt 10, 9 & 5 mm, dept cca 6.5 mm), Fig. 6. Te coarse grain structure and traces of martensite in beinite structure ave been identified as well. Microstructure of WM is austenite wit cca 35 % -ferrite. Te ydro test revealed two troug-tetickness cracks at te outer side of flange. Figure 4. Cracks in a flange welded joint Figure 5. Microcracks in HAZ of microalloyed steel 3. SIMPLIFID KING S SPRING LIN MODL Basic assumptions of King s spring line model are: surface crack is positioned in flat infinite plate loaded by remote tension, real crack front is replaced wit quadrangle, wit constant dept, c = const., spring is elastic-ideal plastic, Dagdale model is applied to take into account plasticity at te crack tip. Figure 6. Cold cracking 34 VOL. 36, No 1, 008 FM Transactions

3 Tus, denoting by σ c = N/ membrane loading and by m = M/( /6) bending loading, one can get: wit 4 a ( c ) = σ σ, (1) 8(1 + ν ) a θ = m. () (3 + ν ) On te oter and side it is obvious tat: (1 ν ) C 11 c C 1 m ( σ ) = +, (3) 1(1 ν ) C 1 c C m ( σ ) θ = +. (4) By eliminating and θ from (1) (4), it follows: σ a a c = C 1 D + ν 3(1 ν)(3 ν) σ = + ασ, (5) a m = C 1σ = 1 ν D βσ, (6) a a D = C + C + C 1 ν 3(1 + ν)(3 + ν) (7) Spring remains elastic until yielding point. As te yielding criterion a simple expression is used: c σc = σ F. (8) Tus, te yielding appears wen te average stress in te ligament reaces stress, σ F, defined as: σ y + σ m σ F =. (9) From (7), te expression for te ligament yield stress, σ, follows: 1 c σ = 1 σ F. (10) α After yielding plasticity at te crack tips is taken into account by using te effective crack lengt, a p = a + r v, were r v stands for te plastic zone size. Based on tat and by applying Dagdale ligament yield model, Fig. 7, one gets: π a π a π a sin sin p sin = w = w, (11) w π σ σ cos c π σ cos 1 σf σ c c σ F c were w stands for te plate widt. Te plate can sustain loading until yielding reaces its ends, defined by net yield stress, σ NSY, as follows from (9) wit a p = w/: σ ac = 1 NSY σ F w. (1) One sould notice tat King s model is valid for a finite widt plate. Anyow, if w, sin (σa p /w) σa p /w and sin (σa/w) σa/w one gets te approximate solution for an infinite plate: a ap = and σ NSY σ F, (13) π σ cos 1 σ F c wic can be applied for te tin sell wit small curvature, like te large spere. Figure 7. Ligament yield model Since σ c and m are now defined for te wole range of te remote stress σ, COD and J can be evaluated as follows (see also Figure 8): CTOD = + θ ( C), (14) CMOD = + θ. (15) Figure 8. Calculation of CMOD and CTOD By substituting expressions for, θ one gets (σ σ ): 4aσ (1 ) CTOD C 1 + ν = β α ν, (16) 4aσ (1 + ν) CMOD = 1 β α ν, (17) i.e. (σ < σ < σ NSY ), ( y ) ( σ σ ) 4 a+ r a+ ry CTOD = + CTOD a, (18) ( y ) ( σ σ ) 4 a+ r a+ ry CMOD = + CMOD a. (19) To calculate J, for σ < σ one can use te linear elastic fracture mecanics relation: FM Transactions VOL. 36, No 1,

4 1 ν 1 ν c I ( σc 1 ) J = K = F + mf, (0) were F 1 and F are defined in [3]. For σ < σ < σ NSY, J p can be calculated by using load line displacement: N M Jp = d dθ c c, (1) Approximately 50 % of te final fatigue precracking was done by maximal force F max = 0.4 F L, wereas te minimal force was F min = 0.1 F max. Te ig-frequency pulsator CRACKTRONIC was used, Fig. 10. were for N/ c = σ F, M/ c = 0 σ J ( ) ( ) c 4 F p = σ F = a+ ry σ aσ ryσf. () 4. FRACTUR MCHANICS AND STRUCTURAL INTGRITY OF WLDMNTS Basic fracture mecanics parameters, applicable to weldments, are te crack tip opening displacement (CTOD) and J integral [3]. By using tese parameters, te structural integrity of weldments, as described in [3], can be assessed by comparing te crack driving forces (CDF), vs J-R curve. Te most suitable (simple) metod for CDF evaluation in te case analysed in tis paper is te line spring King s metod, wic is based on CTOD analysis of large flat plates wit a surface crack [3], wereas J-R curves can be obtained in standard experimental way [4]. By applying King s model, one can get te set of CDFs, as described in [3]. Standard fracture mecanics testing includes J-R curve evaluation by using partial unloading tecnique and applying it to te welded joint, as described in ASTM 180 [4], including specimen geometry, Fig. 9. Figure 10. Te ig-frequency pulsator CRACKTRONIC Results of fracture mecanics testings are sown in Figure 11 for a typical case, sample (lower eat input), specimen wit crack in WM, tested at 0 C. Wen J-R curve is positioned in te same diagram as CDFs, obtained by King s metod, Fig. 1, bot te stress for stable crack initiation (σ inic ) and te initial crack lengt (a un ) for unstable crack growt can be estimated. Crack Driving Forces, as sown in Figure 1, are expressed via J integral and given for different levels of loading, defined as a ratio of remote tensile stress and yield stress (0. 1.0). Diagrams for oter samples are given in [1]. Figure 9. Specimen for J-R curve testing a) 5. RSULTS OF FRACTUR MCHANICS TSTING AND STRUCTURAL INTGRITY ASSSSMNT Two different samples were made by manual arc welding of te plates (dimensions 500 x 00 mm, one plate HSLA steel NIOVAL 47 te oter one ig alloyed austenite steel X7CrNiNb18.10, te filler metal INOX R 9/9, Jesenice steelworks) wit te iger and lower eat input (samples 4 and, respectively). Specimens wit two different crack positions, in HAZ and WM, were made and tested at two different temperatures, 0 C and 60 C. Te electromecanical testing macine SCHNCK TRBL RM 400 was used, including special inductive measurement unit for CTOD, wit ± 0.01 mm accuracy. b) Figure 11. a) F- curve, b) J-R curve for sample, WM, 0 C 36 VOL. 36, No 1, 008 FM Transactions

5 Te integrity of eterogeneous welded joints is not affected by te presence of surface cracks because overmatcing plays a protecting role, wic consists of a small plastic deformation of weld metal even at ig loads causing fracture of parent metal. Te latest conclusion olds at low temperatures, as well. ACKNOWLDGMNTS We acknowledge te support for tis investigation by MSSR Tenogas, AD, Beograd. RFRNCS Figure 1. CDFs vs. J-R curve for sample, WM, 0 C In te same way, all oter samples were tested and te results sown in Table 1, were te stresses for te stable crack initiation (σ inic ) and te crack lengts (a un ) for te unstable crack growt are given for bot samples, bot crack tip positions (WM and HAZ) and bot testing temperatures (0 C and 60 C). Table 1. Results of CDFs vs J-R curve analysis σ inic [MPa] WM WM 4 HAZ + 0 C HAZ C HAZ HAZ 4 60 C 60 C σ inic /σ a un [mm] CONCLUSIONS Based on results presented ere and in more details in [1], one can conclude te following: For te iger eat input σ inic is always iger, wile a un is iger in WM, but smaller in HAZ, so te iger eat input is somewat better; Te HAZ of microalloyed steel as greater resistance against cracks tan te WM, being quite different comparing e.g. to te beaviour of microalloyed steels welded joints; Hig stress levels for initiation of stable crack growt suggest te possibility tat te welded structure can operate safely even in te presence of relatively large surface cracks; [1] Jovicic, R.: An analysis of te crack effect on te integrity ferrite-austenite welded joints, PD tesis, Faculty of Mecanical ngineering, University of Belgrade, Belgrade, 007, (in Serbian). [] Jovicic, R. et al.: Report for te experimental investigation at Messer tenogas, Faculty of Mecanical ngineering, Belgrade, 005, (in Serbian). [3] Sedmak, A.: Application of fracture mecanics to structural integrity, Faculty of Mecanical ngineering, Belgrade, 003, (in Serbian). [4] ASTM a Standard Test Metod for Measurement of Fracture Tougness, Annual Book of ASTM Standards, Vol , ПРОЦЕНА ИНТЕГРИТЕТА ФЕРИТНО- АУСТЕНИТНИХ ЗАВАРЕНИХ СПОЈЕВА Радомир Јовичић, Александар Седмак, Зијах Бурзић, Венцислав Грабулов, Јасмина Лозановић Овај рад приказује утицај прслина на интегритет хетерогених феритно-аустенитних заварених спојева применом основих принципа механике лома. Силе раста прслине су одређене нумерички применом Кингове методе, док је отпорност материјала на раст прслине одређена експериментално помоћу Ј-R криве. Да би се одредила својства оваквог завареног споја, заварене су плоче високолегираног аустенитног челика и микролегираног феритног челика. Експериментално истраживање, у комбинацији са нумерички добијеним резултатима, омогућило је процену интегритета феритноаустенитних заварених спојева. FM Transactions VOL. 36, No 1,