The development ofmatching compositionsupermartensitic weld-ing consumables is pre-sented. It is shown thatthe newly developedsupermartensitic metal-cored wires OK Tubrod15.53 and OK Tubrod15.55 produce high-quality welds with prop-erties which match re-quirements when usedwith realistic fabricationwelding procedures.

IntroductionDuplex ferritic-austenitic stain-less steels have been used forsome years to combat corrosionproblems in the oil and gas in-dustry. New weldable supermar-tensitic stainless steels are, how-ever, finding increasing applica-tion as an economical option, of-fering corrosion performancebetween that of carbon steel andduplex stainless steel (refs. 1-5).These steels offer sufficient cor-rosion resistance for sweet andmildly sour environments in com-bination with high strength andgood low-temperature toughnessand are specifically designed forfield welding where the use oflong-term post-weld heat treat-ment (PWHT) is impracticable(ref. 6). The carbon content is re-

duced to extra low levels in orderto obtain the necessary propertiesin the as-welded condition andelements such as Mo and Cu areadded for improved corrosion re-sistance.

The choice of filler materialand welding procedure is largelygoverned by the need 1 to match parent material

strength,2 to achieve sufficient toughness,3 to keep hardness at an accept-

able level,4 to match corrosion resistance

and 5 to avoid PWHT.

Although good results havebeen obtained with soft marten-sitic 13%Cr 4%Ni + Mo alloysand duplex or superduplex con-sumables (refs. 2-4), supermarten-sitic consumables offer a numberof advantages. Not only are theweld metal strength, hardness andcorrosion resistance similar tothose of the parent material. Theweld metal and parent materialalso respond similarly to PWHTand thereby eliminate the risk ofweld metal embrittlement whichcan occur in the case of superdu-plex weld metals. A further ad-vantage is the decreased risk ofdistortion thanks to the lowerthermal expansion coefficient ofsupermartensitic weld metals ascompared to duplex weld metals.Complications related to the dilu-tion of filler metal with parentmaterial are also avoided ifsupermartensitic consumables areused.

The aim of the present study isto show that “matching” composi-tion supermartensitic consum-ables can be formulated and canproduce largely martensitic weldmetals. It is also demonstratedthat supermartensitic consum-ables can be used with realisticfabrication welding procedures toproduce high-quality welds withsatisfactory properties.

Chemical composition,microstructure and mechanical propertiesExperimental weldsIn the very early stages of devel-opment, it became evident thatthe weld metal chemical composi-tion strongly influenced the mi-crostructure and mechanicalproperties. As a first step, chemi-cal compositions and mechanicalproperties for a large number ofexperimental weld metals weretherefore determined and corre-lated to microstructure, transfor-mation characteristics and weld-ing procedure (see also refs. 7 and 8).

An evaluation showed that the very low C and N content(<0.01%), essential to obtaingood toughness and low hardness,could be obtained most reliablywith metal-cored wires. Further-more, metal-cored wires are idealfor submerged arc (SAW), tung-sten inert gas (TIG) and metal in-ert/active gas welding (MIG/MAG), thereby offering flexibil-ity and high productivity.

Svetsaren No.3 1999 3

Development of matching composition supermartensitic

stainless steel welding consumables

Leif Karlsson and Solveig Rigdal, ESAB AB, Göteborg, Sweden and Wouter Bruins and Michael Goldschmitz, FILARC Welding Industries B.V., Utrecht, the Netherlands

Mechanical properties Weld metal yield and tensilestrength were generally well abo-ve the typical values for super-martensitic parent materials (refs.3, 4 and 6). The exception wasweld metals with a large volume

fraction of retained austenite(>25%) and with a yield strengthbelow the minimum level of 550MPa required for X80 gradesteels. Both the yield and ultimatetensile strength were lower trans-verse to welds in supermartensiticplate material than along weldsand fracture always occurred inthe parent material. Elongationwas typically in the range of 15-20%, regardless of compositionand welding method.

Maximum hardness correlatedwell to the C content and PWHT(5 min/620°C) was effective in re-ducing hardness (cf. refs. 1, 3, 4and 6, for example). A maximumas-welded hardness of approxi-mately 350 HV10 could be ob-tained if the C content was keptbelow 0.013%C.

Significant amounts of ferritewere found to be detrimental to

impact toughness and tended toresult in an unacceptably highductile to brittle transition tem-perature. However, the most sig-nificant effect on impact tough-ness was clearly the strong effectof weld metal oxygen content. Byplotting impact toughness againstoxygen content (Fig. 1), it can beseen that impact toughness in-creases rapidly for oxygen levelsbelow approximately 300 ppm.For example, TIG welds with anoxygen content of about 100 ppmhad a toughness of up to 150 J at–40° C, whereas SAW welds withabout 600 ppm oxygen had thelowest toughness, typically 30–40J at –40° C. However, a shortPWHT could be used to improvethe impact toughness of SAWwelds significantly (Fig. 1).

MicrostructureVirtually fully martensitic weldmetals could be obtained for anMo content of up to 2.5% (Fig.2). However, the compositionalrange for “fully“ martensitic weldmetals was somewhat limited andbecame narrower as the alloyingcontent was increased.

Ferrite (Fig. 3), with a mor-phology very similar to that ofthe ferrite found in duplex stain-less steel weld metals, was presentwhen the relative amount of fer-rite stabilising elements was toohigh.. Another ferrite morpholo-gy, similar to that of common aus-tenitic stainless steel weld metals,was found in weld metals solidify-ing as a mixture of ferrite andaustenite (Fig. 4). One complica-tion, which was dependent on

4 Svetsaren No.3 1999

–40°C

–20°CIm

pact

toug

hnes

s (J

)

Weld metal O-content (ppm)

Figure 1.Influence of oxygen content on Charpy-V impact toughness at –20°C and–40° C in Mo-alloyed supermartensitic weld metals.

Figure 2. Fully martensitic microstruc-ture of a 2.5% Mo supermartensiticweld metal deposited with the metal-cored wire OK Tubrod 15.55.

Weld Consumables Parent material Joint/ Interpass Heat input Commentsposition temp. (°C) (kJ/mm)

OK Tubrod 15.53:1.5 % Mo, B 1.6 mm metal cored wire:TIG/MIG-1.5Mo Ar/ Ar + 0.5%CO2 12Cr 4.5Ni 1.5Mo 0.5Cu/ 60°V, PA <100 1.0/1.5 TIG/pulsed MIG,

20 mm plate 3+11 beadsOK Tubrod 15.55:2.5 % Mo, B 1.6 mm metal cored wire:TIG/MIG-2.5Mo Ar/ Ar + 0.5%CO2 12Cr 6.5Ni 2.5Mo 0.5Cu/ 60°V, PA <100 1.0/1.5 TIG/pulsed MIG,

20 mm plate 3+11 beadsSAW-2.5Mo OK Flux 10.93 2Cr 6.5Ni 2.5Mo 0.5Cu/ 60°X, PA <100 0.7 side 1: 11 beads

20 mm plate side 2: 3 beadsTIG/MIG-2.5Mo-pipe Ar/ Ar + 0.5%CO2 12Cr 6.5Ni 2.5Mo 0Cu/ 60°V, PA <100 1.5/ 2.8 TIG/pulsed MIG,

pipe B 255 mm, t= 13 mm 1+2 beads

Table 1. Welding conditions for pipe girth weld and plate butt welds in supermartensitic steels.

the C content, was the sometimesunacceptably high amount of re-tained austenite in some of themost highly-alloyed experimentalgrades.

Supermartensitic weld metalconstitution diagram

The need for a tool capable ofpredicting weld metal microstruc-ture from chemical compositionwas recognised at an early stage

of consumable development. Thenegative effects of ferrite, auste-nitic solidification and excessiveresidual austenite content have tobe avoided to obtain the opti-mum properties. However, no di-agram or equation covering therange of compositions studied inthe present investigation could befound in the literature. A newconstitutional diagram for super-martensitic weld metals thereforehad to be constructed.

Like earlier diagrams proposedby Balmforth and Lippold (refs. 9and 10), the new diagram wasbased on the Kaltenhauser Crand Ni equivalents (ref. 11), butthe compositional range was ex-tended and Nb and Cu were in-cluded in the compositional fac-tors. Limiting boundaries bet-ween different microstructural re-gions (Fig. 5) were defined frommicrostructural information for alarge number of experimentalwelds, plotted on the diagram.

The new diagram was found to bevery useful in defining the opti-mum consumable compositionssuitable for steels with a varyingMo content.

Representative micrographsfor weld metals belonging to themartensite (M), the martensiteand ferrite (M+F) and the mixedsolidification microstructural re-gions of the constitution diagram(Fig. 5) are shown in Figures 2, 3and 4 respectively.

Simulated productionweldsWelding conditionsConsumables producing the de-sired weld metal microstructurewere selected for welding trials inpipe and plate material using sim-ulated production welding proce-dures. Metal-cored wires with1.5%Mo (OK Tubrod 15.53) and2.5%Mo (OK Tubrod 15.55) wereused for the butt welding of 20

Svetsaren No.3 1999 5

Figure 3. Ferrite (white phase) in amainly martensitic 1.5% Mo weldmetal solidifying as ferrite.

Figure 4. Ferrite (white phase) formedduring ferritic/austenitic solidificationof a mainly martensitic 2.5%Mo weldmetal.

Figure 5. Constitution diagram for supermartensitic weld metals. Information isprovided about the solidification mode and microstructural constituents, includingthe presence of significant amounts of retained austenite at two different C levels.

Weld/steel C N (ppm) Si Mn Cr Ni Mo Cu O (ppm)

Steel:Plate: 12Cr 4.5Ni 1.5Mo 0.5Cu 0.023 112 0.19 1.96 11.5 4.5 1.3 0.48 51Pipe: 12Cr 6.5Ni 2.5Mo 0Cu 0.010 87 0.17 0.50 12.3 6.6 2.5 0.02 63Plate: 12Cr 6.5Ni 2.5Mo 0.5Cu 0.020 130 0.10 1.76 12.4 6.5 2.3 0.49 110

Weld:MIG/TIG-1.5Mo 0.017 80 0.63 1.32 12.4 6.7 1.5 0.55 230TIG/MIG-2.5Mo 0.012 90 0.33 1.97 12.5 7.0 2.3 0.50 270SAW-2.5Mo 0.008 220 0.35 1.78 12.6 6.9 2.4 0.50 670TIG/MIG-2.5Mo-pipe 0.009 135 0.31 2.0 12.5 7.0 2.2 0.44 284

Table 2. Chemical composition (wt.%) of steels and weld metals.

Cr =Cr+6Si+8Ti+4Mo+4Nb+2Al eq

Ni

=40(

C+N

)+2M

n+4N

i+C

ueq

(Cu)

F + FA solidification

> 0.02 %C < 0.01 %C

M

M+F

M+A

M+F+A

mm 1.5%Mo and 2.5%Mo super-martensitic plates and for thegirth welding of a 2.5%Mo pipewith an outer diameter of 255mm and a wall thickness of 13mm. As shown in Table 1, TIGwas used for root passes in MIGwelds, whereas the SAW weldingof plate material was performedfrom both sides in an asymmetri-cal X joint. The pipe girth weldwas completed in only three pass-es, one TIG root pass and twoMIG passes (Fig. 6), with a heatinput of approximately 2.8kJ/mm.

Microstructure and chemicalcompositionAll the plate and pipe welds weredefect free and had a largely mar-tensitic microstructure. The chem-ical compositions of weld metalsand parent materials of simulatedproduction welds are presentedin Table 2. It can be seen thatTIG/MIG welds in a high-C par-ent material have a C content ofabove 0.010%, whereas the SAWand the TIG/MIG welds in thelow-C pipe material have0.008%C and 0.009%C respec-tively.

Mechanical propertiesThe mechanical properties of thesimulated production welds were

in good agreement with the expe-rience acquired from the experi-mental welds. The strength wasclearly overmatched, as is evi-denced by cross-weld tensile testfractures appearing in the parentmaterial, and the ductility wassufficient to pass bending tests(Tables 3 and 4). The maximumhardness was approximately 350HV10 or below and, as expected,the impact toughness was strong-ly dependent on the oxygen con-tent (Table 2). However, a com-parison of the impact toughnessof the TIG/MIG-2.5Mo pipe weldand the TIG/MIG-2.5Mo platebutt weld produced the somewhatunexpected result that a higherheat input and larger weld beadswere beneficial (Tables 3 and 4).This result is contrary to what hasbeen seen for experimental weldsand further studies are clearlyneeded to establish the effect ofheat input on impact toughness. Itis encouraging, however, that highproductivity welding procedurescould also offer advantages interms of improved toughness.

Corrosion resistanceThe preliminary results from SSCtesting in formation water (20°C,20 bar pCO2, 100,000 ppm Cl–,4 mbar or 40 mbar pH2S,4.5<pH<5) show that welds pro-

duced with supermartensitic con-sumables have a corrosion resis-tance comparable to that of theparent material. The tests wereperformed on specimens ma-chined from welds in the as-wel-ded condition. In this particularenvironment, a PWHT, decreas-ing hardness and lowering residu-al stresses, does not therefore ap-pear to be necessary to ensuresufficient corrosion resistance.

Future supermartensiticwelding consumablesAlthough the development ofmatching composition supermar-tensitic welding consumables isstill progressing rapidly, it is al-

6 Svetsaren No.3 1999

Figure 6. Cross-section of TIG/MIG-2.5 Mo pipe girth weld in 2.5% Mopipe with a wall thickness of 13 mm.(Consumable: OK Tubrod 15.55)

Impact toughness Cross weld tensile strength Maximum hardness Face bend Side bend(J) (MPa) (HV10) (t=10 mm)

–60°C –40°C –20°C +20°C Rm Weld metal HAZ B 3 x t, 120° B 4 x t, 180°

58 66 75 80 895* 323 326 OK OK

* Fracture in parent material

Weld Impact toughness (J) Cross weld tensile Maximum hardness Bend testingstrength (HV10)(MPa)

Weld centre Fusion line Weld HAZ Face Root Side (t=10 mm)–40°C –20°C –40°C –20°C Rm metal B 3 x t, 120° B 4 x t, 180°

TIG/MIG-1.5Mo 63 64 184 175 898* 348 359 OK Fissures OKTIG/MIG-2.5Mo 50 50 77 65 905* 351 363 OK OK OKSAW-2.5Mo 38 40 66 76 – – – – – –

Table 3. Mechanical properties of TIG/MIG-2.5Mo-pipe girth weld (consumable: OK Tubrod 15.55) in a 2.5 % Mo super-martensitic pipe (B 255 mm, t= 13 mm).

Table 4. Mechanical properties of welds produced with OK Tubrod 15.53 and 15.55 metal-cored wires in 20 mm super-martensitic plate material.

ready obvious that this conceptoffers a number of advantages interms of properties, productivityand the chance to perform aPWHT when desired.One furtheradvantage that is often over-looked is the fact that a marten-sitic weld metal microstructure isexpected for all levels of dilutionwith parent material when weld-ing with a supermartensitic con-sumable. This is clearly an advan-tage compared with what hap-pens when using duplex or super-duplex consumables, as the result-ing microstructure in this case isdirectly related to the degree ofdilution.

In conclusion, although the fur-ther fine tuning of matching com-position supermartensitic con-sumables is expected, it is nowpossible to predict the micro-structure from the new supermar-tensitic weld metal constitutiondiagram and the relationship bet-ween microstructure, compositionand properties is fairly wellunderstood. It is therefore mostprobably only a matter of timebefore supermartensitic consum-ables replace duplex and super-duplex in the welding of super-martensitic stainless steel.

ConclusionsMetal-cored wires have been de-veloped for supermartensiticsteels; OK Tubrod 15.53 with1.5%Mo is suitable for steels with0–1.5%Mo and OK Tubrod 15.55with 2.5%Mo is intended forsteels with an Mo content of upto 2.5%Mo. A very low C and Ncontent (<0.01%) was consistent-ly obtained in SAW, MIG/MAGand TIG welding.

A constitution diagram wasdesigned for supermartensiticweld metals defining the compo-sitional ranges that produce amartensitic microstructure andproviding information on the solidification mode and retainedaustenite.

Weld metal strength over-matched parent material strength,apart from excessive weld metalresidual austenite content. Themaximum hardness was approxi-mately 350 HV10 or lower at0.010%C. The impact toughnesswas critically dependent on the

weld metal oxygen content andwas reduced by residual ferrite.

A short (5 min/620°C) PWHTincreased impact toughness andreduced hardness.

SSC testing shows that weldsproduced with supermartensiticconsumables have a corrosion re-sistance comparable to that of theparent material.

It is shown that supermarten-sitic consumables can be usedwith realistic fabrication weldingprocedures to produce high-qua-lity welds with mechanical andcorrosion properties that matchrequirements.

AcknowledgementsL. Coudreuse (CLI, France) isgratefully acknowledged for per-forming the corrosion testing.

References1. J. Enerhaug and P. E. Kvaale, 1996,

Qualification of welded super 13%Crmartensitic stainless steels for sourservice applications. Proc. Materiald-agen, 13 November 1996, Stavanger,Norway.

2. E. Perteneder, J. Tösch and G. Raben-steiner,1997, Zur Metallurgie derSchweissung weich martensitischer13% Cr Chromstähle für Öl- undGaspipelines, Proc. Int Conf. Weldingtechnology, materials and materialstesting, fracture mechanics and qual-ity management, September 1997,Vienna, Austria, Vol. 1, pp. 43-56.

3. A.W. Marshall and J.C.M. Farrar,1998, Welding of ferritic and marten-sitic 13%Cr steels. IIW Doc. IX-H452-99.

4. J.C.M. Farrar and A.W. Marshall,1998, Supermartensitic stainless steels– overview and weldability. IIW Doc.IX-H 423-98.

5. L. M. Smith and M. Celant, Martensit-ic stainless flowlines – Do they pay?,Proc. Supermartensitic StainlessSteels’99, May 1999, Brussels, Bel-gium, pp. 66-73.

6. J.J. Dufrane, 1998, Characterisation ofa new family of martensitic-supermar-tensitic plate material for gas trans-port and processing. Proc. Euro-corr’98, September 1998, Utrecht, theNetherlands.

7. L. Karlsson, W. Bruins, C. Gillenius, S.Rigdal and M. Goldschmitz, Matchingcomposition supermartensitic stain-less steel welding consumables. Proc.Supermartensitic Stainless Steels’99,May 1999, Brussels, Belgium, pp. 172-179.

8. L. Karlsson, W. Bruins, S. Rigdal andM. Goldschmitz, Welding supermar-tensitic stainless steels with matching composition consumables. Proc. Stain-less Steel’99 – Science and Market,June 1999, Chia Laguna, Sardinia, Ita-ly, Vol. 3, pp. 405-414.

9. J.C. Lippold, 1991, A review of thewelding metallurgy and weldability offerritic stainless steels. EWI ResearchBrief B9101.

10. M.C. Balmforth and J.C. Lippold,1998, A preliminary ferritic-martensi-tic stainless steel constitution dia-gram. Welding Journal, Vol. 77, No. 1,pp. 1s-7s.

11. R.H. Kaltenhauser, 1971, Improvingthe engineering properties of ferriticstainless steels. Metals EngineeringQuarterly, Vol 11, No. 2, pp. 41-47.

Svetsaren No.3 1999 7

About the authorsWouter Bruins is developmentengineer for the ESAB group,located in Utrecht, The Nether-lands. Currently he is involved inthe development of stainless steelcored wires for supermartensiticsteels.

Michael Goldschmitz is develop-ment engineer cored wires for theESAB group, located in Utrecht,The Netherlands. He was nomina-ted senior expert in 1995. He ispresently involved in several deve-lopment projects concerning stain-less steel.

Leif Karlsson, Ph.D, Senior Expertin welding of stainless steels at theESAB Central Laboratories inGöteborg, Sweden. He JoinedESAB in 1986 after graduatingfrom Chalmers University of Tech-nology, Göteborg with a Mastersdegree in Engineering Physics in1981 and finishing his Ph D. inMaterials Science in 1986. AtESAB he has been working withR&D on highly alloyed weldmetals devoting much time toduplex stainless weld metals.

Solveig Rigdal, M.Sc. and EWE, isworking as a development engi-neer in the department for Auto-matic Welding Consumables withinR&D in ESAB. After graduatingin chemistry from the University inGöteborg in 1974 she has beenworking with consumable develop-ment at Elga in Lerum and at AirLiquide in Malmö before joiningESAB in 1982.