[AVESTA JERNVERKS AKTIEBOLAG INFORMATION Bt AVESTA/SWEDEN TBatum 28.9.66 Ferrite- a structure clement with favourable and harmful effects n austenitio stainless welds. Papor prosented oy Jarl Bjérkroth and Sturo Honrikson, Avesta Jernverks AB, at tho autumn meeting of the Swodish Welding Socioty on Novombor 23rd, 1965. ‘>, Extract from "Svotsen" 1966, No 2. Ferrite or not ferrito, that's the question, Had Shakespeare been occupied with the welding of stainless stecl, he might casily have had Hamlet uttor his immortal words on tho properties of ferrite. The necessity of ferrite in austenitic stainless wolds has booome ho subject of increasing study in recent years. Its importance as a factor ,in provonting crack formation was established as carly as the 1940's, Different investigators havo advanced cxplana- tions for this favourable effect but noyertholess the relationship potwoon ferrite and hot cracking romains\ shrouded in mystory in many rospocts. : At a relatively carly stage in the history of stainless steel 44 was found that certain typos of steel wore subject to,em- brittlomong or aging on prolonged heating in the 550-900°C (1020-1650°F) tomperature range. The ombrittloment phenomenon is espocially pronounced when the structure contains ferrite, sineo this is transformed easily to sigma phaso, Therefore during the last 15 years or so limits havo beon imposed on the rite contont in certain steels for high tomperature appli- cations. tho other hand the role of ferrite in corrosive media has not received attontion until during the last few years. Cor- tain chemicsl substances are know to attack the forrite itself or its boundaries with the austenitic matrix. Bven where prosent in contents as low as 2 to 3 % the forrite is subject to selective attack under certain operating conditions. Since the problems associated with this appoar to be unknom to a large number of welding engineers, tho main object of this paper is to shed sono light on tho behaviour of forrite under corrosive conditions.‘aro generally found in tho boundaries botweon th AVESTA JERNVERKS AKTIESOLAG INFORMATION |) 5 AVESTA/SWEDEN Batu 18.966. he nature and formation of ferrite Austenite forms tho matrix in an wold, Secondary phases such as ferrite, carbides, signa ote. are also formed under certain conditions. Thoso socondary phases austenite grains. ‘A condition for the occurrence of dotectablo amounts of ferrite in the wold structure is that ferrito-forming oloments be present in the molt in sufficient quantities. Tho most pro~ Rinent of these cloments are chromium, molybdenum, silicon end niobium, Other ferrite-forming elenonts ero aluminiun, titaniun, vanadium, tungsten and zirkonium glthough those aro not normal alloying clemonts of the more conmon grades of stainless steel. Nickel, manganese, carbon and nitrogen con- tribute to the formation of austenite. As a general rule he ferrite content of tho weld can be controlled by varying the rolationship betwoen the ferrite- and austenite-forming elonents. Howover, the heat of welding and tho cooling con- ditions exer tively largo influcnce on the final composition of the structure. Fig. 1 shows a phase diagram for a 18 Or, 8 % Mi stool carbon contents between 0 and 0,4 % (1). At the carbon contents found in welding stainless stools, ferrite is the first solid Fn eee es ee Lemporeturo faiie below the Ligicus.f forrite-forming eloments are concentrated in the forri and the austenite-forming clemonts in the eustenite, Invosti- gations using an olectron microprobe analysor have established = considerable difference in analysis betwoon the forrite and tho ustenite, In weld metel with an average composition of 18 # Cr/ 9 %Mi the forrite phase may consist of 22-23 % Cr and 6-8 ~ Ni. She amount of ferrite thus influences tho composition of tho austenite, A ferrite content of appr 15 % would thus cor- respond to an austonite containing 17 % Cr and 10 % Mi. However, tho results obtained in the olectzon microprobe analysis should only be considered as scmiquantitativo since the ferrite grains wero just slightly lergor than the diamcter of tho, oleetron boam, Groator differonces in composition botwoon the forrite and the austenite phases are to be expected whon largor forrite grains are analysed.Be iF 7 AVESTA JERNVERKS AKTIEBOLAG INFORMATION eet 5 AVESTA/SWEDEN TBotem Continued coolin ving soli ation of the austenite esults ina cortain amount of the ferrite being ro~tranaforned nto austenite, ‘The influonce of ferrite in proventing crack formation has been nvestigated by many researchers. Various theories havo beon Le it is thought te reduces in- or that its prosonco ii ain donéritic utes 2 reinforcing cffcct, blocking the path of inoipiont oracke, Othore claim that ‘tho onco of ferrite refines tho structure or that the tomporature interval botweon tho liquid end solid phasos becomes shortor (2, 3, 4). no fact that tho above mentioned factors have a n significance, the hypothosis (5) that the ability of te to dissolve or incorporate ost of tho clenonts tond to assist crack formation, particularly silicon and phosphor, appoars to bo gonerally accepted. austenite on the er hand has an oxtromely limited ability to dissolve those caces where ferrite is not prosent or is pro~ y am nogligible quantities, the silicon and phosphor a % which solidify last and ow molting points. The result is the pro~ ions or slag films in the grain bownd~ croseopic cracks Drawing an analogy, which absorbs larger cause cracking. olononta the grain structure may - onsile tests on pceimens of a 19 Or/ AST) weld. Spocinon A unber ef fino fiasures vhercas spocimen Determination of tho forrite en Aand 5.5 for spocimon B. Tho yearances of the fractures aro shown in Pig. 3, Fracture B vits e fine-graih, homogenous surface while fracture & has a coarso crystalline fibrous appoaranco,4 INFORMATION 660d ! AVESTA JERNVERKS AKTIEBOLAG AVESTA/SWEDEN Toe 18.9.66 Usually, in the case of all-weld tonsion test specimons tho fissures if any aro concentrated on one sido, i.c. that facing the welding bead. Fissures on the lowor side aro usually few or non-existent. When the ocourrence of microsco— pic cracks in a thick plate weld is examined more closcly, it s found that the froquoncy of dofects is groator in the final han in the initial passes. 4 possible reason for this is that the phosphor and silicon graduslly become concentrated in the final passes duc to tho melting action of tho provious passes. Detormination of forrite Cy The ferrite content of an austenitic wold can bo determined by sovoral methods. The following are of particular importance: 1, Calculation based on tho chonical composition 2, Metallographic dotormination 3 « Magnotic determination 4, eray diffraction determination With a mmowlodge of the composition of tho wold the forrite content may be deto#minod approximately by moans of tho Schacffler-DeLong diagram. Figure 4 shows tho portion of the Schaeffler diagram modified by DeLong (5), Tae consituent \ percentages are substituted in tho equations for the chromiun and nicke) oquivalonts. The ferrito content is read from the diagram vy plotting the values thus obtained. Use of tho Schacfflor diagran is x acourate method of determining the forrite content. yw ferrite contents the error in tho plotted valuo s 20 groator than the true value, However, the diagran nay useful when con= dined wi For instance where the results of an oxtonsive sorios of tests aro available, the diagram may be medificd in individual oasce for a cortain type or size of cloctrode and thus be nade to provide more accurate results.E Din AVESTA JERNVERKS AKTIEBOLAG INFORMATION 66282 AVESTA/SWEDEN TBotwm 18.9.6 Hiei Motellographic determination ‘Tho mothod Imown as linear analysis is gonorally used in netallograshic dotermination. In this method the surface area consisting of ferrite is measured under the microscope on 4 polished ané ctched spocinen of the materiel, Ferrite and any other phases present can bo developed by a judicious choice of etching procedure, The nicroacope is provided with a graduated glass plate and the percentage of ferrite is dotermined by counting tho nunber of units of ferrite line by lino, This method is time-consuming and requires a good deal of oxperience on tho part of tho individual porforning the cxanination, Less accurate results nay vo czpocted when the forrite content is small, since otching offects produced by dondritic sogrogation may aplicato the idontification of ferrite. ‘A quicker nethod of dotermination based on this principle now cxists. The microscope image is projected into a TV-sercen from waich an electronic scanner and counter determine the content. 3 gnetic dotermination Since ferrite is magnetic and austenite is non-magnetic the forrite content may be determined by noasuring tho magnetic properties of the material. However, it nust be pointed out that martensite is also magnetic and in cases where this is present in the structure, the result will ropresent the sun of the ferrite and nartonsite contents. ‘The most widely used instruments for maguotic determination are the following: a) Magnetic neasuring instrunont know as tho Juince-Bronner Magne-Cago (see Fig. 5) is manufactured by the Amorican Instrument Co. The comparator consists of a small pormanont magnct, the attractive force of which may be balanced against that of the ferrite-containing cpocinon, This instrunont has aoquired wide popularity in laboratories but is not suited for use on uoldea components on the shop floor, Waore maximum accuracy is required the instrument should be calibrated for various types of welding naterials. compare tornin 108 6 AVESTA JERNVERKS AKTIEBOLAG INFORMATION ¢¢5 ) Eoprascope This instrument (see Fig. 6) provides a convenient, simple and dircet moans of moasuring tho ferrite content in the welded assembly. The permascope which is manufactured in Gormany by Helmuth Pischor is of cloctro-negnotic typo with stabilized voltage and curront, Moasuroment ic carried out by placing a probe on tho surface to bo oxemined, The probe forms nag- netic circuit with tho component and the gouge indicates tho Forrito content as 2 function of magnetic curront intensity. To ensure accurate readings of the ferrite content in stein- less natorials calitration must first be oarricd out by other thods, Otherwise the values indicated will normally be too gh. c)” Magnetic induction coll Curtis and Sherwin (7) have described a method in which an in- dscation of tho ferrite content may be obtained by measuring the saturation magnetization of a test specimen, Thoir first attempts wore nade using a Magnaflux crack detector connected Yo a ferronetor, achieving field strengths up to 2200 Oorsted in this way. Similar instruments have been developed by others. As an example field strengths of up to approx. 6000 Oersted can bo achieved with an instrument developed by tho Avesta Jornvork laboratories. Calibration for the size or shape of tho ferrite grain is not required when the ficld strength is sufficiently gh, Measurements can be carricd out on a cylindrical specimen 12-15 mm in diametor. 4 ay diffraction detersination By using an F-ray diffractonotor it is possible to identity unknown phases and to distinguish difforont structures, Ferrite has a body-centered cubic lattice whilo that of austenite is Pace-centered cubic. Martensite has a body-contered tetragonal lattice. Yon tho sample is exposed to monochromatic X-radiation, rofloction poaks are obtained from tho ferrite and the austenite lattices Pospectively, The intensity of the peaks is proportional to the concentration of cach phasc. By measuring tho intonsity of # strongest forrite peaks and the strongost austenite poaks and by carrying out comparative calculations the ferrite concentration can thon be evaluated.INFORMATION ‘TBatum 18.9.66. AVESTA JERNVERKS AKTIEBOLAG Influence of welding conditio As mentioned proviously the thermal balance of wolding exerts an influence on the final ferrite content of the weld. The following nain factors dotermine the amount of ferrito which precipitates in notal arc wolding: 1. Hloctrode dianeter + Value of welding current + Speed of welding . Interpass temporature . Plate thickness aupun . Cooling conditions Sono oarlier investigations have shown that tho forrite con— tent varies with the electrode dismeter under otherwise constant welding conditions (8,9). Tho following report deals with the results of tests for ferrite determination carricd out on wold analysis samples and on tonsile test specimens takon fron welds in 14 mm stainless plate. Both the parent material and the electrodes were of a 19 Cr/10 Wi/0.03 type (type 308 L, ASTM). The electrodes wore gon so that each size usod was as nearly as possible of tho same composition. The weld analysis samples were woldod in the following way to obtain some indication of the effect of tho different variables: 1, Hlectrodes of 2, 2.5, 3.25, 4 and 5 mm diameter wore used. 2, Three thickness of baseplate, 3, 8 and 19 mm, were uscd. 3. A sorios of wold analysis samplos was made with water cooling after cach pass. 4, A sories gf wold analysis samples was nade with air cooling tc velow 100°C (212°F) botwoon passes. Fig. 7 shows the appearance of the weld anslysis sanplo which was nade somewhat larger than proscribed vy existing stendarcs, The ferrite content was moasured by moans of 2 magnotic comparator at a largo number of points in sections 4 and B, 10 ma above the asoplate. The section betwoon these two cross-sections was used for chemical analysis.oO g INFORMATION AVESTA JERNVERKS AKTIEGOLAG AVESTA/SWEDEN TBatem 18.9.66 The tost results are tabulated in Table 1, Only the results obtained from the water-cooled sample are shown, since the difforence duc to the type of cooling was shown to be nogligible. As the table shows, the thickness of the basc— plato had ‘no appreciable cffcot on the ferrite content, Figure 8 shows tho tonsile test spocimen. This was mado form 14 ma plate in accordance with accepted standards, Tho wold was allowed to cool below 100°C after cach pass, The nechani- cal propertios are illustrated by the curves in Fig. 9 which shows two characteristics for forrite. Curve F (P) shows the ferrite content in the tensilo tost specimens while curvo F (A) ropresents tho average ferrite content of the analysis samples. Table 2 gives the chemical analyses of the tensile spocimons manufactured from differont sizos of cloctrode, The analyses of all wold analysis samples were also checked. Tho nitrogen con~ tent is that which distinguishos cach analysis. This is consider~ ably higher in the prepared samples than in the spocimens taken froma weld, In the former caso the gas normally liberated from the electrode cover is unable to protect the wold from the nitrogen in tho air whon woléing is carried out on e "ridge". Tho column to the extreme right givos tho nitrogen contents of tho analysis samples, ‘These were an average of 0.03 % greater than tho corresponding values for the test specimens. 4 comparison betwoon the forrite contents as determined by tho Schaeffler-DeLong diagram and the rosults of magnetic determination is given in Table 3. Account has been taken of tho nitrogen pick-up of the analysis samples in the extreme right column of the table, 4 certain tendency towards a lower forrite content is discernible when comparing columns 6 ané 10. The nitrogen pick-up remains relatively constant for the @ifferent electrode diameters and therefore doos not contribute to the slope of the ferrite curve F (st) in Fig. 9. ‘The principle reason for the variation in forrite content with the size of electrode is tho quantity of heat input, which varies with the diameter. Tho fact that tho slope of the Serrite curve is stooper for the analysis samples than for the Geneile specimens shows that cooling end heat dissipation are also of importance. The heat is dissipated more quickly in + 2 weld dotween two plates than in tho caso whore @ sample is puilt up on a baseplate.oO ) INFORMATION 6628rE AVESTA JERNVERKS AKTIEBOLAG AVESTA/SWEDEN TBotum 18,9.66 Influones of parent material ordinary weldable 16/8 plate materials o.g. typos AIST 304, 3165 S04 Land 316 L vory often contain botweon 0 and approx. $ # ferrite in melt condition, which is to be found in the solidified ingot. However, tho ferrite is transformed to austonize Gusing the various heating processes carricd out during rolling ané especially during tse fine) ennoaling carried out prior +o Gelivery from tho mill, Determination of forrite carried out gn an annealed and pickled sheot will give a result of around 0 On the other hand if the shoot is molted locally, or if Quo plates from the materiel aro welded without a filler, the fesrite content of tho molten material will be aore or loss oqual 4o that of the original ingot. Duo consideration must also be given to tho effects of molting parent metal when the ferrite content of the weld is to be Kept low, Where the design of a wolded component imposos a linit on the forrite content, a parent material having an Qutroncly low ferrite contont in the melt condition should be Ghoson, On the other hand it is poscible to reduce tho forrite Sontont of the weld by post-wold annealing. This is sonctinos Garniod out on welded components whore annealing is feasible. Table 4 illustrates how the ferrite in e weld of 19 cr/20 wi/ 0.03 © (typo 308 L, ASTM) stocl varies with sonealing ‘temperature. Ti is evident from the tablo that extremely high tonperaturcs dro required to onsure rapid transformation of the forgite. iis hentioned previously, tomperatures between 500 and go0°c (930 and 1650°F) are unsuitablo duo to the risk of carbide precipitation. Goulding tines in this range are of necessity extrenely long 1? the Serrite content is to be reduced to a low value. Optinun forrite content Opinione vary sonowhat regarding the optimum ferrite contest Sea weld, The majority of investigators who havo dealt with tho problens associated with this appear to agro on o nininup value of 5 % to ensure resistance to cracking. Investigations carried out at avosta have revealed that a distinction muss iorthdo in this context between alloy steels containing nolyb- denun and those which do not. We considor that a minimun forrite contont of approx, 3 is required to prevont cracking in welds of the 18 Cr/ 9 Ni type. On the othor hand good resis- tance to cracking is obtained with ferrite contents as low as20. INFORMATION Nf626 B AVESTA JERNVERKS AKTIEBOLAG: AVESTA/SWEDEN TBetum 18.9.66 pacer 0.5-1% in 18 Cr/ 12 Ni/ 2.7 Mo welds, ‘The factors which govern tho absolute minimum value, apart from the composition of tho paront material, are tho heat of welding (sizo of clectrodc), intorpass tomporature and the stross conditions provailing while the weld is cooling. However, a forrite content of 5 %as not always permissible undor cortain operating conditions. In applicagions whoro the working temperature is greater than 550°C (1020°F) (applying mainly to type 18/9 or 18/9/ib welding materials) it is maintained that tho optimum forrite contont should lie betweon 1 and 4 g (10), Other sources suggest 2-5 # (11). The conclusion to bo drawn is that 2-4 % would scom to be ideal. Tt has been found necessary to limit tho ferrite content to 2 %undcr cortain corrosive conditions. In lator years there has thorefore beon a sporadic donand for 18/12 Mo electrodes with a maximum of 2 %forrite. 4 maxinum of 1 % is spocifiod in certain applications to provide an oxtra margin of oafoty. ‘Those requirenonts ontail cortein complications for tho electrode nanufacturor, Extremely closc analysis tolerances must bo kopt and difforont material analysos are required for 2 mm and 5 ma cloctrodes in order to maintain exactly the samo ferrite content. Influence of ferrite on the corrosion resistance of wold notals Bffoct_of corrosive onvironnent Tho influence of ferrite on the corrosion resistant proporties of the weld material is completely depondent on the corrosive medium to which the steel is exposed, The primary reason for the corrosion resistance of stainless steel is its ability to passivate in the presonce of an oxidizing agent. This property is generally utilized in choosing 4 matorial and it is usual to select a grade which moots the requirements with an adequate margin, No complications are encountered in such cases, Tho forrite-containing weld is passivated equally well as the comple- toly austenitic paront material and withstands corrosion, pro- vided that welding is carriod out using an olectrode which gives the weld the same chemical composition as the parent material. In cxceptional cases it is necessary to use a steel under con- ditions whore the corrosion resistant properties aro barely adequate or even inadequate. This may even occur accidentally Guo to a faulty choice of material or to more aggressive con- @itions than expected, In theso borderline casos especially8 ne AVESTA JERNVERKS AKTIEBOLAG INFGRMATION — Nr 6626 2 ‘AVESTA/SWEDEN Baten 18.9.6 the wold is subject to more severe corrosion than the parent material, Ferrite may bo responsible for this type of corro~ sion but where strong oxidizing egonts aro prosont, the attack will be more severe if part of tho forrite is transformed to tho signa phase. Practical experionce Sone examples describing cases of selective weld corrosion are given below. Those are tekon both from the literature (12, 13,14) and from the authors! experience and aro grouped in order of the type of attack: Oxidising agents . 1, Corrosion occurs in hot concentrated HNO,, primarily in Movalioy wolde. Molyddonum steole are séidon intentionally used for nitric acid. Fig. 10 shows selective attack in a type 316 weld. The attack is concentrated to tho ferrite which has been partly transformed to the sigma phase. Th agont was boiling 65% ENO, (ASTM-A 262-64 T or Huoyts test). 2, Uolybéenun steols oxhibit a somewhat better resistance than normal Cz/¥i-steels ina pickling bath containing HNO, and HF, probably due to tho high nickel content. Howover; attack may occur quite easily ospecially in welds conteining forrito, Fig. 11 shows a sevorcly corroded wold in a type 316 steol fron a pickling bath vat. The attack is concen~ trated to the ferrite, [ven though signa phase wes not dotected in this caso it is not impossible thet small amounts wore present and contributed to the corrosion. Nonmoxidizing agents 1, Corrosion oceurs in hydrochloric acid vapours and in solu- tions containing dilute hydrochloric acid, as for example in the manufacture of paint ingredients, paarmaccuticals and starches as well as in acetic acid-porchloric acid and water-fornaldehyde solutions, Stainless steel is used since nobler materials such as titaniun, zirconium and tantalun are prohibitively expensive. Figures 12 and 13 show solootive ferrite attack in q laboratory specimen exposed to 5 /# hydro~ chloric acid at 50°C, 4 cortain amount of signa phase associated with the ferrite is also evident in this case.8 AVESTA JERNVERKS AKTIESOLAG AVESTA/SWEDEN TBetum 12, INFORMATION 6626 B 18.9.6! 2, Corrosign gcours ip proparing urea fron GO, + NH, ot seerogs8o (ae00365¢F). tow carbon molybdofun stéols with a maximum of 1 % ferrite in the weld netel are usually specified for this application, since wolds with highor ferrite contents are subject to oxtremely severe attack, Figure 14 is a microphoto of selective forrite corrosion in a type 316 steel following a 7O-day test in an urea reactor. the wold contained approx, 2 % forrite and 1 @ sigma phase. im interosting observation is that welds with highor sigua phase ané lower ferrite contents tested simultancously oxhibi~ ted considerably less corrosion. 3, Selective attack has boon cbsorved in chromium nickol stools congaining 2.553 Mo wnon distilling pino oil at 250 — 200°C (480-570°F) but not in spocimens containing 4,5 flo. The reason for this is that a steel with the lowor molybdonum Gontont is nore active cloctrochenically, while a highor content rendera tho stecl passive. tral chloride solutions Experiments carricd out at Avests have show that molybdensn siecle conteining high amounts of austenite-forming elements have @ somowhat botter resistance to pitting than materials conteining a fcw percent forrite according to tho Schac?lor—~ JeLong diagram, Hovever, the authors are not for tho prosont gequainted with any practical oxanples inarhich ferrite has peon proved to bo responsible for the pitting vhich commonly goeurs widor tho action of corrosive agents of this type. ‘The meckanisn of corrosion The selootive corrosion which ocossicnally occurs in austenitic elas containing ferrit? er @ coubination of ferrito and sigma phase appears to be related to the difforence in chemical Gomposition between those yaases and the austenitic matrix: fe montionea previously, ferrite contains nore chromium and nolybienun than austenite. Tho signa phase produced by Gorrite when tho wold ig cooled throygh the signa phaso range of approx. 900°C to 550 ¢ (1650~1020 -F) contains a still greater peroontage of chromium, Austenite for its part is richer on nickel veen the ferrite and signe phases. Tvo theories can explain “he nechanisa of selective corrosion:ANESTAISWS 18.9.66 1, ‘The classic theory assunos that corrosion tales placo in those sections with the lowest chromiun content, ‘The truth i of this assumption, at least in cases where sigma phaso is present, is demonstrated by tho fact that short heating times in the sonsitive tomporature rango result in nore severe attack than longer tines which allow chromium equalization to tako place through diffusion (9, 15, 16). Sice newly formed signa phaso is nearly always present ia molybdenum wola metal containing ferrite, it is oxtromo~ ly difficult to distinguish between the effects of this eo) phase and that of the ferrite. However, the authors! view basod on experience is that the ferrite is not responsible for the sane severe dogreo of selective corrosion in strong oxidizing agonts as the sigma phaso. On tho other hand the occurrence of the forrito phase appears to be V decisive when corrosion attack takes place in non-oxidizing agonts. 2, kecording to anothor thoory it is the precipitated phase which is subject to selective attack and not the noigh- pouring austenite which is poor in chromium, This theory has boon confirmed by soloctive etching of both forrite and signa phase et cortain characteristic, constant potentials. The work of Cihal and Prazak (17), Baelenau (18) and Réschenbleck (19) in this field is worthy of mention. Tt appears probable that sclective corrosion follows this mechanism in reducing corrosive agents, especially at the ; strong active potentials assumed by the stocl in hydro- chloric acid, In this condition tho rate of corrosion is accelerated by an increased chromium content and 2 reduced nickel content. ~ ‘ow can selective corrosion be provented? - The most practical mothod of preventing selective corrosion in welds is obviously to use a filler which maintains tho ferrite content of the wold as low as possible. This has boon suggested by Jesper (14) among othors although this writer advances no detailed specification of the wold compo- sition.AYESTA JERNVERKS AKTIEBOLAG AVESTA/SWEDEN TBotm INFORMATION —Ntggo6 5 28.9.66 Corrosion in different media is dealt with by Espy (13) for AISI 316 L steel welded with a largo nunbor of different st less electrodes. The aim of this writer is to devolve a un! electrode for oxidizing and reducing as well as pitting agen? Despite the wiformly good corrosion resistance obteined with a forrite-froe type 16 Cr/18 Ni/2.5 Mo wold, the writer doos not recommend this in view of the rick of hot cracking. He recor~ + 316 L be wolded with a universal electrode giving the following weld analysis: ©<0.03 % Si .0,60 %, Un 1-2,5 % P<0,03 %, S<0,03 % Cr2c. 22.5 %, Mi 13.5-14.5 % Mo 2-2.5 and forrite 4 % rersal Corrosion tests have been carriod out at Avesta on welds of a sinilar composition but with slightly higher chromiun and nickel contents, specifically 23 on/ 15 wi/2.2 Mo and¢.<0.03¢. This composition provided extremely good resistance to pitting in neutral chloride solutions and to general corrosion in oxidising agents. Severe attack on the weld itself yes exporienocd in reducing agonts such as 5 %HC1 at 50°C due tc the relatively high ferrite content. During the sane tests a weld with a composition of 18 % Gr, 14 %Mi, 3.0 % Mo and ferrite exhibited extrenely good resistance. Sinilar rosul have deen obtained in other non-oxidizing corrosive atmosp! such as in urea reactors. is evident fron the foregoing that the choice of fillor naterial ig peculiar to each individual application, dopendirg on the type of corrosive agent in question. The following rules should be applied in choosing a filler for welding nolybdenun steels in which solective corrosion of the weld mainly occurs: 1, In normal cases, that is in cases where the parent netal has acceptable corrosio: nce to the agent in quos~ tion and remains ontirely passive, a filler of the sanc composition as the parent material and with a normal forrite content may be used as recommended by the manufacturer. agonts such as urea, dilute hydrochloric acids or other { reducing agents which are Imow to assist selective corrosion due to ferrite, @ spociel electrode giving e low | forrito content in tho weld should be used, The chroaiua | and molybdenum contonts should be the sane es in the oq valent grade of sheet. Bloctrodes of this typo also vido somewhat ‘better resistance to pitting than similar electrodes with a high ferrite content. om15. INFORMATION 9626 E AVESTA JERNVERKS AKTIEDOLAG AVESTA/SWEDEN TBotm O oO 18.9.66 Signa phase in.weld netal of alloys containing molybdenun nay cause selective corrosion ospocially in strong oxidizing agonts, Hot concentrated nitric acid is a typical cxanplo, However, sinco molybdenum stools should not normally be used in this application, tho risk of the associated corro~ sion probloms is not ospocially groat, On the othor hand if circunstances do roquire the usc of a molyddenun stool in a strong oxidizing mediua the formation of sigma phase should be reduced firstly by judicious choice of material and secondly by applying suitable welding techni- auos. It should be mentioned that tho offect of ferrite on corrosion rosistance is not always negativo. It is known that the risk of stross corrosion cracking duc to chlorides diminichos with inereasing ferrite content (20), Susceptibility to inter-crystallinc corrosion in welds also appoars to be less in welds containing forrite as compared to those poor in this phase (21). Standards for forrite determination The above discussion illustrates the desirability of a goncral~ ly accepted directive or standard for determination of ferrite in austonitic stainless welds. Tho practice somotinos usod of prescribing determination on an analysis sample is completely faulty. There is every chance that the values in actual wolded piccos will be different, Since the dosirability or othorwiso of the ferrite is dictated ontirely by tho finished product this must naturally form the basis for determination. The only question is whether a single plate thickness is adequate for the purposes of a standard tost. It is possible that several tests for difforent heating and cooling conditions would be required. For the present we cannot make any suggestion as to how a standard test for forrite doter~ mination should be formulated, AVESTA JORNVERKS AKTIEBOL Information Department Hf re ‘é H Morell siINFORMATION WE26 BD AVESTA JERNVERKS AKTIEBOLAG AVESTA/SWEDEN TBetm 5 18.9.86 Roferonces 1, Keivobok, VN, The dook of stainloss stools. Ancrican Socicty for tals, Cleveland (1935). 2, lashko-Avakyan: duto, Svarka, vol. 11/1958, pp 99-110. 3. Thomas, R Ds Crack sonsitivity of chromiun-nickel stain loss wold motal, Metel Progress, Sopt. 1946, pp 474-479. 4. apblott, WR - Pollini, W S: Factors which influence wold hot cracking, The Wolding Journal, Fer. 1954, pp 838-905. 5. Parks, J MsDiscussion - Welding stainless stool. Tho Welding Journal, June 1955, pp. 568-570. 6. DeLong, WT: 4 modified phase diagram for stainless stool wold metals. Metal Progress, Fotr. 1960, pp. 98-101. a 7. Gurtis, GC - Sherwin, J: Magnetic’ mothod for the osti- mation of ferrite in Gtainloss stecl welds. British Jounal of Appliod Physics, July 1961, pp. 344-345. 8. Bjérlwoth, J: Worketoffproblone beim Schweisson von austoni- tischen nichtrostendon Stéhlon, Technica, 1958/o. 25, pp. 1467-1470. 9, Bjdrkroth, J: Die Bigonschaften des Schweissgutes von niodriggokchlten sdurobostdndigon Stihlon. Zeitschrift fir Schweisstechnik 1962/No. 3, pp. 81-88. 10. Curran, RM - Rankin, A W: Welding type 347 stainless stecl for 1160°F turbine operation. The Welding Journal, March 1955, pp. 205-213. 11, Thomas, R Ds Soloction of stainless stecl cloctrodes for troublefree welds, Motal Progress, July 1956, pp. 73-76 12, Carruthers, 1f E: Weld corrosion in typo 326 and 316 L stainloss steels and roletod provlems, The Welding Journal, Juno 1959, pp. 2595-2678. 13. Eepy, RH: What wold motal for type 3161 stainless stool. otal Progress, Sept. 1964, pp. 109-115. 14, Jesper, H- Griitenor, G: Korrosionserscheinungen an rost- tna sdurebestéindigen Stuhlon, Wickel-Borichte 1965/2 pp. 39-46. 15. Wachneister, I: Signa phase in typo 19 Gr-9 Mi-type stain- joss wolds, "Svetsen" No, 5/1959, pages 107-115 (Swedish). 3§ a AVESTA JERNVERKS AKTIEZOLAG 1. INFORMATION NIS626 E AVESTA/SWEDEN Tot ee 18.956 16, Warron, D: The effect of signa phase vs chromium carbides 1b 18. 19. 20. ai. on the intergranular corrosion of 316 and 316 L stainless Stool. Corrosion, April 1959, pp. 213t-220t. cial, V- Prazak, ll: Gorrosion and notallograrhic study of stiinloss stools using potontiostat techniques. Journal Seon and Stool Institute. Doc. 1959, pp. 360-367. ioloanu, U: Tho potentiostat as a motallographic too}. Journal tron and Steel Inst., April 1957, pp. 482-486, Réschonbleck, B - Buss, B K: Potontiostatischos aifferont~ jolles Atzon dor Gofiigobostandtcile in 16/8 Or-Ni-Stéhlon, Workstoff wd Korrosion, 1963/4, pp. 261-269. Ferrite content cuts stress corrosion cracks, Steel No. 23/1961, p. 95+ Honrikson, $1 Unpublished works, Aveste Jornverks dictio~ polag, Avesta. (Swedish).AVESTA JERRVERKS ANTIESOLAG REC eTIOe ‘AVESTA/SWEDEN, Bt 18.9.6 Forrite content of analysis samples as deter ned by nagnotic comparator (iiagne=Gage). Bloctrode ontggo feerite in es diancter ‘EvOTaGO oS nm 3 2 5.1 5 49 2.5 3.9 1 4.2 © 3.25 Be ius = 4 1.3 1.5 0.5 0.9 LC 2 4.6 | 408 2.5 42 44 3.25 2.1 12.2 4 2.0 19 5 0.4 i1.0 2 43 1 4.5 4 2.5 42 1 4.0 3.25 1g 1.8 u 4 13 od 2.2 11.2 Table 2 Chemical analysis of weld, “ Pleotrode! Composition of test specimens ‘analysis Dianeter - am ~-fsamples ‘bm ¢ Si iin Mo oy 2 10.034 0.84 2.22 0.03 0.082 !0.107 to.5 '0,032 0.78 0.04 0.080 !0.104 13.25 10.030 0.80 1.22 19.2 10,0 0,04 0,077/0.121 4 0.028 0.76 1.05 19.3 9.9 0,03 0.07310.303 5 0.029 0.75 1.10 19.2 10.1 0.07 _0.077'0.123 \ laverago '0.031 0.76 1.15 19.3. 10.1 0.04 0.078/0.110INFORMATION Botan AVESTA JERNVERKS AKTIEBOLAG AVESTA/SWEDEN 16.9.66 Lee vy ous t ee ay 069 : ve et OSL. ; eel sig! : ' $0} 090’ ‘eanoy oot! mou 6| 29! g ee emnypocoduay, poy vowue-uor : | Jo ott Surprow tox ogtzzos gf | Supreouay Up og tao! Burtusuue ypFA yuoquos ogqLL0y UT uoFFEEIEA FOTAL T T i T i ae os | ee Ls \ ore! Ltytt uy uy i atv! Sti Ly Ls i Sb orvt. SG og TEES t - 4 | erooay” : : | poururot 1 : i eyttzog toxore | STSATeUy pat po Bao og 9urps0goe ee orduts siekpeue | ~eteduoo oroutau £q Geaiepp Tad-§ 0} Fappsoooe. "werd op i Uy ea Tend froryourmtorop oyFttod | suoripoods 4604 ur 04 f4tod! 0240013 sxoqvateduios ofjoudet puv uosurp Juoped-coTzFoTyOS Aq uoTwuTUAOZOp O4TAsOS Jo UOsTHEduog ESTEE ry fy ) ot vevInf, 6626 Temp. Imes (s) c Schmelze (s) 1500] | 1400 Austenit (hr) | 1300) | © i ae || 1200] 4 | | j ifr /rarbid oO one 1 020s 04 C 18% Cr aN! 76% Fe Fig. 1 Phase diagram for steels with 18 % cr (D250) and 8 %Mi according to (1). Phasendiagramm fiir Stahl mit 18 % Cr und 8 % Mi gemiss (1) { | ( i ~ | si i Sabie od wet i 1 7 {3} i u am | Fig. 2 All-weld-metel tensile test bars, type 308 L (D251) (ASTM). A) ferrite = 0 % very many cracks caused by hot cracking (micro cracks) during welding, B) ferrite = 5.5 %: no cracks or defects. Zugstibe aus Schweissgut des Werkstoffs 1.4306 (type 308 L). A) Ferrit = 0 % sehr grosse amzahl Risse, durch Warmrisse (Mikrorisse) beim Sohweissen verursacht. 3B) Ferrit = 5,5 %, Keine Risse oder Fehler.Fig. 4 (D 254) Inf, 6626 ‘The surfaces of fracture of the test bars in fig 2. A) radiated coarse-crystalline fracture. B) fine-grained homogeneous surface of fracture. (ilagn. 3X). Aussohen der Bruchfluchen der Zugstibe aus Bild 2. A) grobkristalline, strahlige Bruchfluche. 3B) feinkérnige, homogene Bruchfliche (V= 3:1) en en | Lb 3s romeivivalent + % Cr+ %MOC1S xHSi¢ OSKKNE trois Modified Schaeffler diagram according to (6). Abgedndertes Schaeffler—Diegramm gemiss (6).S C Fig. 5 (20813) Fig. 6 (D 253) Inf. 6626 Magnetic balance for the determination of ferrite, e.g. in weld metal. Magnetische Waage sur Ferritbestimmung von u.a. Sohweissgut. Permascop allowing direct ferrite determination on finished welds, Permascop, mit dessem Hilfe direkte Ferritmessungen auf fertigen Schweissverbindungen vorgenommen werden kinnen,Inf. 6626 [min 4 layers min 4 Tagen Fig. 7. Welded specimen for chemical analysis of the weld (D247) metal. Schweissgutprobe fiir die chemische Analyse. SY ms f iE x y x2, = (min 80) (mint) min 80 (min 12) Fig. 8 Welded specimen for the preparation of weld (D248) metal test bars. Intnahme der Zugstbe einer Schweissgutprobe.Fig. 9 (D 249) Inf. 6626 1 0 - 60 i B 50 110 (e 40 = ae G2 8 30 |__| _ 6 + 1 + I | c 20 > Sie i F(P) a 10 + | 2 | Toe ° | 2 28 325 4 5 Electrode diam. ma, 4, = Tensile strength in kep/mn®, Bruchgrenze in kp/mm”. 5 = Blongation Sad in % Dehmng Sxa in %. Oc2e Yield point in kp/m”. Streckerenze in xp/nm, Mechanical: properties and ferrite content of the weld metal for different electrode diameters, type 308 L (ASTM). F (P) = ferrite content measured on test bars. F (A) = ferrite content measured on the analysis specimen. Diagramm der Festigkeitseigenschaften und Ferrit- gehalte von Schweissgut aus verschiedenen Blektroden— durchmessern, Werkstoff 1.4306 (Type 308 L). F (P) = gomessene Ferritmenge der Zugstabe, F (a) = gemessene Ferritmenge der Analysenproben.Inf, 6626 Fig. 10 Corrosion attack in weld metal, type 316 (ASTM). (if 10801) ‘The attack follows the ferrite. Ferrite content, 4 Traces of signa phase, : (Magn. 150 X). Korrosionsangriff im Schweissgut vom Typ 1.4436 (type 326). Der Angriff folgt dem Ferrit, Ferrit- gehalt 4 # sowie Spuren von Sigma-Phase (V = 150:1) Fig. 11 Corroded type 316-weld from pickling vat for (U 4530) HNO, + HF, 60°C. Ferrite content, 6 % Yo igna phase, Aogetragene Schweisse vom Typ 1.4436 (Tyge 316) aus einem Beizbehlter fiir HNO, + HP, 60°C. Ferritgehalt 6 % Keine Sigma=phase.Inf. 6626 Fig. 12 Type 316—weld, corroded at the testing in (U 4661) 5 #HC1 at 50°C. (Magn. 5 X).° Avgetragene Schweisse 1.4436; (Type 316) nach Priifung in 5 % HC1 bei 50°C. (V = 5:1). Fig. [3 Photomicrograph of the weld metal in fig. 12. (if 10799) The attack follows the ferrite. Ferrite content, 4% Traces of sigma phase. (Magn. 150 X). Mikrofoto des Schweissgutes aus Bild 12, Der Angriff folgt dem Ferrit, Ferritgehalt 4 % sowie Spuren von Sigma, (V = 150:1).. Big. 14 (ae's730) Inf. 6626 Selective corrosion attack in type 316-weld after 70 days of testing in a urea reactor, The weld meter | contains 2 % ferrite end approx. 1 % sigma phase. (uagn. 300 X). Selektiver Korrosionsangriff einer Schweisse aus 1.4436 (Type 316) nach 70-taégiger Prifung im Harnstoffreaktor. Das Schweissgut onthdlt 2 % Ferrit und ca 1 % Sigma. (V = 300:1).es a: . Spec. No. + te : Revision + Page : of ABBREVIATION 1. Welding Process we MA-: Manual Welding ~SNAWt Shielded’ Metal-Arc Welding SAW Submerged: Are Welding TAU + Autonatie Welding GTAW + Gas Tungsten-Are Welding - -SA ; Sen{-Automatic Welding ‘GMAW Gas Metal-Arc Welding -BA : Band Arc Overlay Welding Foaw + Flux Cored Are Welding OA + Open Are Welding 2. Type of Joint 3. Melding Position bu + Buce Weld Flate™ Pipe plate Fe: Full Penetration Weld ne a. ow i PP : Partial Penetration Weld (Rotated) (Rotated) FW : Filled Weld ' Horizontal 26 2G 2F 27R : (rotated) Vertical 36 -— 3F ~ Over-Head 4G 0 == sae : Maltiples -- sc 4, Polartey Multiple == 6G ee AC: Alternating Current DC: Direct Current Dc/sP : Direct Current/Straight Polarity DC/RP : Direct Current/Reverse Polarity 5. Method of Examtnation RT (Radiography) UT (Ultrasonic examination) _* MT _\_(lagnetic-particle examination) aE e . DT — (Dimensional inspection) VE (Visual inspection) 6. Others PH: Preheating IPWHT + Intermediate: Postweld Heat Treatment PWHT : Final Postweld Heat Treatment per Code Requirements ,‘WHA 4 fe 2 ORF) pop BPSIPE SE2, 7X (woes) C rg7zencco6 ) We ERTS - FEW, a -sss gpd bees | (FEW) QW - 252 SHIELDED PUETAL-PRC WELDIWG ( smaw) (caismuen Mame cov Bib 7re0n0 fivEs7/70, erro (yn IE Maaco wiarnereo) QW-OS4 SUBPERGED EC WEcoWwG Co va) (srevareea AtToMaTICA 1 Agca soramense ) GW - ISS GAs poETAe ARE wero (vw) (SALDBIOOR (0 BUD ERA ee ee Ss LORS g v6. J ) (TWS WeLdDES ALdK CORED Bee wheoule) C (For w) GW ~25C GAS TWGSTEM-ARC WELDING 0 GTAn ) (sALaaTve? W) ATIDOSFERA BI GAS Coo EkETF20DC ) Dr Gouge srewe QW 25% PLasrIp-ARC WELDING (PA Ww) (sataniven 9 Persea J QU 258 ELECTROSLAG WELDING aS Ww) J (umsoarven gurorearieg VERTICAL Sarro ROLF Llaw QW-OSG ELECTRGAS WrELONG C2Gw) gee (RUTOPFBTED VERTICAL GAS (287A PRC ° SO WELD G For yEeriere Posirred! OWb en ceo steczeo,/ pea were (EB Ww) (SiLdnTUL0 Pare BoRARDAIDETO Boge Boles \QW- 267 STUD WELIWE (S7L0_) \ (SPLDBWED BEC PRI GIO ERE \ PUP PST oLe NO - S Opetistin wu - Co . . Sf APrhrce mitetien Gx gee verte, sub motion Co > . 0 foe prvvtome vertenee / ee SPECHNIEMS OF coun oO ATI FB wancraa nea. O6.14 WONTEDISON B.0'A