The Forge Phase of Friction Welding

Transcription

1 The Forge Phase of Frcton Weldng A plastc model s desgned to theoretcally descrbe the movement of nterfacal materal durng the forgng phase of frcton weldng BYT. RICH AND R. ROBERTS ABSTRACT. Even tugh the forgng phase has been prevously recognzed as the tme durng whch the majorty of actual bondng occurs durng frcton weldng, ths current work represents the frst theoretcal nvestgaton of the forgng process. A model s constructed to represent the dstnct rgd and plastc regons of the weld process whch are present durng forgng. From ths model wth the applcaton of the upper bound theory of plastcty, expressons for the velocty and dsplacement felds for materal wthn the nterfacal plastc zone are obtaned. They requre an upper bound forgng stress equal to the average compressve yeld stress of the plastc materal at the weld nterface. The most sgnfcant result of the analyss s the dervaton of a parameter whch s a theoretcal measure of the orgnal abuttng surface materal to be expelled nto the weld flash. Ths nterfacal dsperson s postulated to be a potental parameter of weld qualty optmzaton. Note that t s found to be proportonal not only to the conventonal upset but also to the ntal characterstc length of the nterfacal plastc zone at the onset of forgng. Introducton Basc frcton weldng possesses dstnct heatng and forgng phases. In the ntal phase of the process, one weld pece s rotated whle another axally algned pece s held statonary. The peces are brought together and an axal pressure s appled whch s suffcent to cause frctonal heat generaton at the abuttng surfaces. Ths ntal phase consttutes the heatng phase. When the weld juncture s adequately heated and softened, the relatve surface speed s rapdly brought to zero. At the same moment the axal pressure s rased to the forgng level. Ths s the T. RICH, formerly wth the Mechancal Engneerng Dept. of Lehgh Unversty, s wth Theoretcal and Appled Mechancs Research Laboratory, Army Materals and Mechancs Research Center, Watertown, Mass.; R. ROBERTS s wth the Mechancal Engneerng Dept., Lehgh Unversty, Bethlehem, Pa. Part I Theoretcal Analyses onset of the forgng phase. The softened plastc nterfacal zone of weld materal s forced nto the characterstc flared flash of a frcton weld under a constant forgng load. The weld s set. Gelman's hypothess, whch classfes frcton weldng as a form of pressure weldng as opposed to a fuson process where meltng plays the prmary role, s useful n approachng the parts played by the process parameters durng weldng. The man object of the heatng phase becomes that of preparng an nterfacal regon wse temperature s hghly elevated and resstance to yeldng s very low. The forgng phase then proceeds to squeeze the softened plastc nterfacal zone, expellng much of the trapped oxdes, vods, and contamnaton nto the flash whle pushng clean, pure surfaces nto ntmate contact for a resultant atomc bondng. Thus, the actual bondng takes place prmarly durng the forgng phase. For some dssmlar welds the hypothess s modfed to take nto account the formaton of welded areas durng heatng caused by surface ploughng. To date no expermental evdence has been presented to dsclam ths hypothess as a useful tool n analyzng the bondng mechansm of frcton weldng. The current lterature sws an enormous lack of nformaton concernng the forgng phase, ts dependence upon the heatng phase, and ts effect on subsequent weld qualty. Even tugh the concept of removal of contamnaton, vods, and oxdes from the nterface s nterlocked wth forgng and weld qualty, the shape of the plastc zone and the resultng flow of materal from the zone has not been nvestgated. However, note that expermental nvestgatons ' 3 concerned wth parameter optmzaton have swn best results at hgh ratos of forgng to heatng loads. Also, Hodge 4 ponted to the possblty that mproper weldng condtons can result n forges whch produce lttle or no upset and very poor jonts. The followng analyss presents a model specfcally applyng to the forgng phase of a frcton weld between two cylndrcal weld peces of smlar materal. An analyss s developed for the deformng plastc nterfacal regon whch results n correspondng velocty and dsplacement felds. From these felds s derved a relatonshp for the amount of orgnal nterfacal surface area at the onset of forgng whch becomes expelled nto the flash durng forgng. The rato of ths quantty to the orgnal nterfacal cross sectonal area shall be referred to as the nterfacal dsperson, r\. It ranges from a value of zero where no dsperson has taken place to a value of one where all of the orgnal nterfacal area has been dspersed nto the flash. The nterfacal dsperson s postulated to be a measure of the degree of removal of contamnaton, rough aspertes, and vods from the weld nterface and, therefore, provdes a quanttatve parameter whch could be used to optmze weld qualty. Fnally, the velocty feld s employed n the development of a requred forgng load n the typcal upper bound sense of plastcty and s seen to result n a forgng load equal to the average yeld stress n compresson of the materal n the plastc nterfacal regon. Ths result s qute reasonable and lends credblty to the conceved flow pattern. Theoretcal Analyss Velocty Feld The analyss of the forgng phase begns wth the constructon of a model to represent the deformaton of the weld peces durng the forgng. As represented n Fg., plastc zones that have been prepared durng the heatng phase extend a dstance, h, on each sde of the nterface. Durng forgng the rgd sectons of the weld peces act as platens to compress the plastc zones and cause the subsequent WELDING RESEARCH SUPPLEMENT 37-s

2 Onset of Forgng.P End of Forgng P Rg'd Platen Rgd Platen <V 4/ ^ y y vf V y Plastc Zone Interface Plastc Zone Rgd Platen Fg. Appearance of weld peces durng forgng plastc flow. It s assumed that througut the forgng phase the boundary between the rgd and plastc sectors of each weld pece remans flat and parallel to the nterface. As forgng ensues the rgd platens move towards the nterface at some unknown constant velocty, V. However, V can be related to the axal dsplacements of each weld pece, the upset: and u v = V ' h = h - Up () () where u p s the upset of each weld pece; t' s the tme begnnng at the onset of forgng; h s the nstantaneous plastc zone sze; and h, the plastc zone sze at forgng onset. For the present model the plastc zone sze at the onset of forgng s postulated to be the axal dstance from the nterface where the average temperature equals the mnmum necessary for compressve yeldng at the forgng stress level. The temperature dstrbuton n the weld peces at the tme correspondng to forge ntaton as predcted by a one dmensonal temperature soluton 5 s suggested as a sutable ad n the theoretcal estmaton of h. Ths coupled wth the temperature-compressve yeld stress property of the weld materal provdes an analytcal lnk between the heatng and forgng phases. Fgure depcts a typcal flash of two weld peces joned by frcton weldng. At the onset of forgng, lne elements OA and OA' lay wthn the abuttng nterface whle elements AB and A'B' lay along the lateral surfaces of the rods. The curved profle of the resultng flash, AB and A'B', ndcates that the radal plastc flow velocty decreased from a maxmum value along the abuttng nterfacal surfaces to essentally zero at the rgd-plastc boundares, y ± h. Thus, the radal velocty at the boundary whch could produce the typcal curved flash s assumed to be lnearly dstrbuted from zero at > = ± h to U at y = o as swn n Fg.. Callng the veloctes n the r, y, and 8 drectons «, v, and w respectvely, the lnear expresson for u at any r becomes: (3) -<-) Applcaton of contnuty to the volume of materal enclosed by y = o to h and a radus r results n a relatonshp between the radal velocty at any pont on the nterface, «, and the platen velocty, V : " = V h (4) Thus the radal component of velocty can be wrtten as: "-H -*) (5) Next applcaton of the volume consstency condton of plastcty yelds the axal component: r + e + y (6) where the expressons for the correspondng stran rates n cylndrcal coordnates are: T _ du ~ dr u \ dw 7 + ~r ~d~8 dv " ~ dy (7) (8) (9) Substtuton of eq (5) nto (6) through (7) to (9) results n: Fg. Assumed velocty dstrbuton wthn plastc zone -^- ) () notng that at y = o, v = o and for forgng wtut relatve rotaton between the weld peces, w = o. Thus, eq (5) and () provde a knematcally admssble velocty feld wthn the deformng plastc zones. Dsplacement Feld The next step s to fnd the fnal poston of a materal partcle orgnally at the general poston (r a, y ) wthn the plastc zone after a forge to a gven upset. Refer to Fg. 3. Frst, consder the axal dsplacements. Employng eqs (), (), and () n dfferental form results n a frst order dfferental equaton for y n terms of «,,: du -M-Y \h - Up/ h Up () Equaton () s solvable by the change of varable technque as demonstrated n the Appendx. Employng the ntal condton that at u P = o, y = y, the axal dsplacements wthn the plastc zones durng forgng become y ( - " u p y Vo Up () Ths equaton can be wrtten n dmensonless form as y_ h h "(- )' l y_o Up h. (3) The above dsplacement soluton can become ndetermnate when the upset rato,, becomes smultaneously wth Vo the ntal partcle poston rato, Ths s the case where all of the plastc zone s expelled nto the flash. The same stuaton s seen to arse for the radal dsplacements as derved from 38-s MARCH 97

3 eq (5). Note the dependence of the radal velocty component on the axal coordnate. Substtuton from eq () for y results n the followng frst order dfferental equaton whch s solvable by separaton of varables and ntegraton as swn n the Appendx. dr / V - y h \ d Up h u\h a - y u v ) (4) Employng the ntal condton that at Up = o, r = r D, the radal dsplacements wthn the plastc zone durng forgng become:. _youp h h (5) The radal dsplacements also result n a form representable n a general dmensonless expresson: r R (6) where R equals the orgnal radus of the weld peces. Interfacal Dsperson When y =, the radal dsplacements of eq (5) refer to the movement of partcles at the weld nterface durng forgng. The fnal nterfacal poston of partcles comprsng a rng orgnally at r can be employed to estmate the amount of orgnal nterfacal materal whch becomes dspersed nto the flash. For the rng of partcles whch at the completon of forgng comes to rest on the orgnal outer crcumference of the nterface, r = R and eq (5) becomes R = r R Up^ (7) where r or s the orgnal poston of the rng moved to R. Thus nversely: r ok (8) *('-) It then follows that the orgnal nterfacal materal remanng dspersed througut the nterface after the forge wll be that whch was ntally at r < r or. Ths represents an area of: Aon = *,,» = rrr (l -5EJ (9) or n terms of the fracton of orgnal nterfacal materal expelled nto the flash,.e., the nterfacal dsperson: A R A or v = = ('- () 4 Interface P I o.n (-^ Intal Poston SfcK I -fetensung Poston Intal fnal Fg. 3 Coordnate system for partcle dsplacements wthn the plastc zone durng forgng where A R equals the orgnal cross sectonal area, tr R. Prelmnary observatons of the expresson for T) reveals that the parameter, upset, alone s nsuffcent to predct the amounts of nterfacal materal expelled to the flash. Rather the rato of upset to ntal plastc zone sze s requred accordng to the theoretcal model. Note that u p s the upset of each weld pece and for smlar materals equals y the total upset conventonally measured durng frcton weldng. Also, n order to remove % of the orgnal nterfacal materal, the entre plastc zone would need to be forged nto the flash. Upper Bound Forgng Load By employng the upper bound theorem of plastcty, an evaluaton s made of the conceved knematcally admssble velocty feld to determne the requred forgng load. Through ths theorem the power expended by the unknown forgng load s set equal to that requred to produce the rate of energy dsspaton reflected by the velocty feld. If the actual velocty feld for the forge has been selected, the actual forgng load wll be derved. In general, wever, a velocty feld whch does not exactly match that taken by nature wll be conceved. Ths feld requres more power than actually needed and results n a forgng load greater than demanded; thus, an upper bound. For the purpose of ths analyss, the entre plastc zone s dealzed to be a perfectly plastc, Mses materal regon. Furthermore, the effectve compressve yeld stress of the plastc materal s assumed constant, <r, througut the zone. In the actual case, the yeld stress would vary wthn the plastc zone accordng to the ntal temperature dstrbuton as the forge s frst appled and then to whatever dstrbuton develops as the materal begns to deform plastcally. Here, a s assumed to be some weghted averaged value of the hgh temperaturecompressve yeld property. The power expended by the external forgng force s: W PV () rememberng that P s advancng at a velocty V across both plastc zone boundares, ± h. Ths power s deally spent as the nternal work of deformaton whch for a Mses materal s: W = -= <r. LA V3 [6 +en + hu ] ll dvol. () where e, = the prncpal stran rates; vol. = represents a volume ntegraton. For ths case the prncpal stran rates are gven by eq (7), (8), and (9). Employng the velocty expressons (5) and () the nternal work of deformaton becomes W do I Jvol. e r d vol. (3) By employng the symmetry n 9, an elemental volume can be wrtten as: d vol. = IT r dr dy (4) Hence, consderng ntegraton of half the total plastc volume: W = o-o TT R Vo (6) Equatng the power suppled to the nternal work of deformaton results n the average pressure requred to produce the conceved velocty dstrbuton. rrr (7) Because the upper bound forgng load results n the average compressve yeld stress of the plastc zone, a very reasonable concluson, support s gven to the postulated flow feld as a representaton of the actual materal movement n the nterface regon durng the forgng phase. Conclusons Ths analyss represents the frst attempt to theoretcally descrbe the movement of nterfacal materal durng the forgng phase of a frcton weld. An nstantaneous velocty feld has been presented whch s compatble wth the observed shape of the expelled flash of typcal frcton welded peces. From ths velocty feld, the dsplacement feld has been derved and presented along wth a relatonshp gvng the fracton of orgnal nterfacal WELDING RESEARCH SUPPLEMENT I 39-s

4 materal expelled nto the flash durng the forgng phase. It s upon ths nterfacal dsperson whch an optmzaton of the frcton weldng process may ultmately be based. Fnally, the velocty feld when appled to the upper bound theorem of plastcty resulted n a requred average forgng pressure whch s reasonable n lght of standard unaxal compresson tests. It suld be observed that the nterfacal dsperson proves not to be a functon of axal upset alone as swn by eq (), but rather a functon of the rato of upset to the ntal plastc zone sze. Therefore, n a general sense, upset alone s not suffcent for controllng the removal of oxdes, contamnaton and vods from the weld nterface as earler suggested by Vll. 6 In other words, dependng upon the sze of the plastc zone, and thus the amount of deformable materal, a gven upset can remove a varyng percentage of avalable plastc materal nto the flash. When appled to the materal at the nterfacal surface, the result s a dfferent dsperson of contamnaton tor the same upset. Ths case arses even ABSTRACT. In conjuncton wth the theoretcal analyss of Part I, an expermental study was undertaken to gan knowledge of the forgng phase and ts nfluence on weld qualty. Specfcally, three avenues of nterest were pursued. Frst, temperature ndcatng, grdded, plexglass specmens were frcton welded and swed the dstnct correlaton between the sotherms at the onset of forgng and the nterfacal plastc zone shape. Second, these same specmens revealed the sgnfcant effect whch the plastc zone shape has on the plastc flow feld durng the forge. Thrd, other laboratory cast plexglass rods contanng nternal nylon grds verfed the theoretcal flow feld as a far means of estmatng dsplacements and nterfacal dsperson durng forgng. Introducton An expermental nvestgaton was undertaken to gan nformaton about the nature of the forgng phase of basc frcton weldng. Whle prevous researchers have alluded to the mportance of forgng durng frcton weldng wth respect to weld formaton and qualty, vrtually no results of related studes have been publshed. Thus ths work along wth the theoretcal analyss of Part I provdes the frst steps toward the understandng of forgng's role n the movement of materal at the weld nterface and subsequent weld qualty. As mentoned n Part I, the heatng phase occurs n frcton weldng whle relatve rotaton exsts between the weld peces. At ths tme frctonal Part II Expermental Investgaton when the controllng parameters of speed, heatng and forgng pressures are dentcal, but the heatng tme vares producng dfferent plastc zone szes. Fnally, specal consderaton suld be made notng the assumpton of a flat cylndrcal shape for the plastc zone n the theoretcal model. Ths assumpton has a defnte bearng on the form of an allowable velocty feld for forgng. In Part II of ths nvestgaton, expermental evdence s presented evaluatng the nfluence of the plastc zone shape on materal flow durng forgng, and therefore, assessng the shape assumed n the theoretcal model. The expermental study also presents nformaton on the relatonshp between the heatng phase and the forgng phase as reflected through the temperature dstrbuton at the onset of forgng and the resultng plastc zone shape. In a thrd major segment of the expermental work, verfcaton s obtaned on the applcablty of the derved dsplacement eq () and (5) n predctng movement of nterfacal materal durng forgng. heat generaton causes elevated temperatures adjacent to the abuttng weld surfaces. The hgher temperatures n ths nterfacal regon are reflected by a lower resstance to plastc yeldng. Therefore, at the onset of the forgng phase the hgher axal forgng pressure creates a zone of plastcally deformng nterfacal materal n each weld pece. Of specal nterest s the manner n whch ths softened materal flows nto the flash durng forgng. Specfcally, ths expermental nvestgaton was concerned wth the governng factors whch nfluence the plastc flow patterns and the subsequent removal of contamnaton, oxdes and vods from the abuttng surfaces. The overrdng dea s that exact determnaton and control of these factors would lead to the control of weld qualty. Three man aspects of expermental study were establshed n conjuncton wth the theoretcal analyss of Part I. Frst, a correlaton was sought between the temperature dstrbuton at the onset of forgng and the ensung profle of the rgd-plastc boundary between the rgd and softened sectors of the weld peces. Second, the nfluence of the shape of the resultng plastc zone as determned by the rgdplastc profle was nvestgated wth respect to materal flow patterns durng forgng. Fnally, the actual dsplacement feld for the plastc zone after forgng to a gven upset was determned for comparson wth that predcted by the theoretcal analyss. Expermental Investgaton Expermental Weldng Unt and Specmens Three-quarter nch polymethyl methacrylate rod was csen for test specmen materal because t was easly welded at low axal loads and because ts transparency enabled vsualzaton of the fnal nterfacal flow characterstcs. To weld ths materal, a lghtweght expermental frcton weldng machne was bult 8 and employed as swn n Fg. 4. It conssted of a combned modfed drll press and a loadng frame wth a releasable base chuck. The weldng unt operated n the conventonal frcton weldng manner wth speeds up to 5 rpm and axal loads up to ps. Prevous work 9 - suggested that ths suld be a reasonable range for the frcton weldng of plexglass-lke polymers. A unque and mportant feature of ths expermental weldng machne was the upset control. Ths postve stop on the loadng frame permtted the forge to be halted at any desred degree of axal upset. Thus deformatons n the nterfacal plastc zone were attanable for observaton at ntermttent stages of the forgng phase. The three control parameters contnuously montored durng weldng were the axal load (Baldwn lb load cell), the axal dsplacements, specfcally the total burn-off and upset (dal ndcator), and the heatng tme (electrc clock). The rotatonal speed was pre-set on the machne. The weldng machne was employed to jon plexglass specmens whch were desgned n two specfc confguratons as swn n Fg. 5. The frst confguraton conssted of commercally cast clear plexglass rods nto whch a temperature ndcatng grd had been placed. These specmens were desgned to provde nformaton on the thermal aspects and nterfacal plastc zone shapes of frcton welds. The second confguraton nvolved a laboratory cast clear plexglass set of rods whch contaned a fne permanent nternal nylon grdwork. These specmens were prepared to study the detaled movement of materal wthn the plastc zone durng forgng. The temperature ndcatng grds were made from ground "Templstks" whch were tamped nto small les drlled nto one end surface of each specmen. Several temperature sequences were employed to obtan a vared range of temperature vs. coordnate poston data. The ndcatng temperature of each grd was determned from the characterstc color of the Templ powder comprsng t. Durng the weldng of these specmens, the Templ powders melted at ponts adjacent to 4-S MARCH 97

5 Fg. 4 Expermental frcton machne weldng plexglass whch had reached the ndcatng temperatures. Thus, at the completon of weldng, each grd conssted of a porton whch, as ntally, appeared dull and powdered. In the porton where the ndcatng temperature had been reached, each grd appeared melted, brght and glossy. A sharp lne usually dvded these regons. Therefore, the measurement of the maxmum dstance from the nterface to a pont where the ndcatng temperature was reached for a gven grd's radal poston was attanable. Also durng forgng, the temperature grds deformed n the plastc nterfacal zone whle remanng undeformed n the rgd porton of the weld pece. The demarcaton between the deformed and the undeformed grd sectons provded a profle of the fnal plastc zone shape. Fgure 6 exhbts the defnton of both the ndcatng temperature demarcaton and the dstnct plastc zone shape, rgd-plastc boundary, as observed n the welded specmens. Concurrently, the nylon grd confguraton enabled accurate measurements to be made of the dsplacement feld wthn the plastc zone due to forgng. In order to obtan the nylon grdded specmens, a glass tubed moldng rg 8 was employed n whch sx lb test nylon monoflament was prestrung axally along a sngle dametrc axs. The cylndrcal specmens were then cast from polymethyl methacrylate n ts monomer state, machned, cut, and faced to the fnal desred dmensons. The resultng radal poston of any one grd strand was not found to vary more than. n. per nch length of cast rod. A sold grd-free rod was also cast to provde a smlar materal upon whch to weld the nylon grdded specmens. Upon completon of both templ and nylon grdded weldments, the flash was machned away, and each specmen was polshed. To cut out reflectons from the cylndrcal specmens' surfaces, they were ndvdually placed n a clear plastc box flled wth mneral ol. A travelng mcroscope was then employed to take measurements of the deformed grdwork. The mcroscope was graduated down to. n. Isotherms and Plastc Zone Shapes Templ grd specmens, whch were welded at 355 rpm and under the condtons descrbed n Table, ndcated an apparent lnk between the contours of the sotherms at forgng and the resultng plastc zone shape. Ths can be observed n the ptos of Fg. 6 " Nylon -Grd.-5 O.OIO J^da. «-O.I" Templ-Grd -3/4" a=.6 b =.5" c =.5 ' d =.3IO" Fg. 5 Plexglass specmen confguratons for two specfc cases where the nddcatng temperature was constant througut the grd pattern. For these welds the sotherms ndcated by the melted-powdered demarcatons possess the same general profle as the rgdplastc boundary. The plastc zone shapes n Fg. 6 are termed concave snce, f removed from the weld peces, they would appear smlar n shape to a concave lens. In contrast, a plastc zone wse rgd-plastc boundary s dshed away from the nterface s termed convex. Table --Templ Grd Specmens Speed rpm Heatng load, ps Forgng load, ps Heatng tme, sec Burn-off Ub (X. n.) Total upset Ub + u p (X. n.) Number of specmens ± 5 constant 9 ± ±.5 constant ± ±.4 decreasng ± Balks Fg. 6 Frcton welded specmens wth templ grds. Top 5 F templ grd; bottom F templ grd WELDING RESEARCH SUPPLEMENT 4-s

6 A complaton of measurements taken from all the templ data of the 355 rpm welds s swn on Fg. 7. At 355 rpm a state of unform wear was observed at the nterface durng the heatng phase. Also n the plexglass specmens, the rate at whch the sotherms traveled down the specmens matched the rate at whch burn-off progressed. Therefore, over the entre range of burn-offs employed,.5-.5 n., dentcal fnal plastc zone shapes and szes were observed. For ths reason the sotherm profles mea- Radal Dstance ( ) Fg. 7 Contour smlarty of plastc zone boundary and sotherms for 355 rpm welds.3 Normal Wear Mnmum Wear O Dstance From Interface (") Fg. 8 Intal plastc zone shapes at the onset of forgng at 65 rpm sured for these varous burn-offs are comparable. Wth an observed typcal scatter of ±.8 n., the plotted average sotherms reveal the same dstnct concave shape as the fnal plastc zone. Ths helps to substantate a postulated lnk between the heatng and forgng phases of frcton weldng namely, the rgd-plastc boundary of the nterfacal plastc zone s determned by that sotherm or band of sotherms whch corresponds to the mnmum temperature at whch weldng occurs for the weld materal employed and at the gven forgng level and rate of stranng. As swn n the ensung paragraphs, the shape of the rgdplastc boundary greatly effects the plastc flow pattern durng forgng and consequently the resultng weld qualty. The theoretcal nvestgaton of Part I was based on an assumed flat rgdplastc boundary. Welds made at 65 rpm revealed that when the wearng of materal at the nterface was at a low enough rate to permt the ensung temperature dstrbutons to equalze durng the heatng phase, the plastc zone shapes progressed from concave to flat to convex shapes. However, as soon as the wearng rate ncreased as reflected by rapd rses n burn-off, the sotherms and plastc zone stablzed to the concave shape as was exhbted by the 355 rpm welds. Ths phenomenon can be seen n Fg. 8 from frcton welds made under the condtons ndcated n Table at 65 rpm. Note that for the hghly nsulatng materal, plexglass, the mnmzed wearng could only be acheved by a decreasng loadng arrangement whch ntally restrcted burn-off. For more conductve materals, such as metals, temperature equalzaton and nverson has been expermentally observed under normal weldng condtons. A levelng off of plastc zone shapes 3 has also been observed n metal welds for ncreased heatng tmes. Therefore, the patterns of plastc flow as nfluenced by the plastc zone shape for the plexglass specmens can be extended n consderng other materals. As swn n Table, under condtons of mnmum wear an occasonal balk occurred n the forgng of the weld peces. Ths s reflected by the relatvely low degree of upset n comparson to that acheved for equvalent condtons and heatng tmes for normal welds. These balks were assocated wth lttle plastc deformaton at the nterface and severely convex plastc zone shapes (see Fg. 9). Perhaps the acute convexty of the plastc zones can be attrbuted to an unusually hgh rate of heat generaton at the central porton of the nterface. Ths case could have been caused by a slght hgh spot at the center of one of the abuttng surfaces, whch because of the mnmum wear con- 4-s MARCH 97

7 Heatng Tme 3 *4 \ (sec) 5 J Dstance From Interface (") Fg. 9 Acute convex plastc zones of balked frcton welds dtons was slow to wear down. Durng forgng, the rgd sectons of the weld peces acted as cusped punches to compress the plastc materal radally nward rather than outward as desred. The net result was a "sck-absorber" effect wth reducton n upset, the entrapment of contamnaton wthn the nterface, and an nherent drop n weld qualty. Ths can be observed n Fg. where whte plexglass partcles produced at the nterface durng the wearng acton of the heatng phase were forged nward nto the center of the nterface for the balked specmen. Ths s n contrast to the specmen forged wth a concave zone where the partcles were forced out nto the flash. Ths connecton between plastc zone shape and the concept of weld qualty and nterfacal dsperson of contamnaton, oxdes and vods s of consderable practcal mportance. Concave plastc zone shapes enhance dsperson, whereas convex shapes nhbt t. The shape of the resultant flash, or collar, was observed to drectly reflect the nternal plastc zone shape and correspondng plastc flow pattern. Thus, a means of qualtatvely estmatng flow characterstcs for frcton welds of opaque materals s avalable. As llustrated n Fg., concave plastc zones forged nto gradually flarng collars. Convex zones forged nto flat edged, pnched collars. Excessve heatng and/or hgher forgng loads on convex zones leads to the characterstc splt collar. Here the forgng level s suffcent to break down the outer rm of cooler materal and force plastc materal outward. In any case, the radally nward veloctes mposed at ;he rgd plastc boundary are detrmental to fnal weld qualty. Plastc Flow Feld Dsplacements Resultant measurements of deformed nylon grdwork for frcton welded specmens verfed the applcablty of the theoretcal plastc flow analyss to estmate dsplacements wthn the plastc zone. The nylon grdded specmens were all welded under dentcal speeds, loads, and burn-offs as ndcated n Table. The upset control was employed to stop the forge at specfc ntermedate upset postons for each specmen. Thus by vewng an entre set of specmens a verfcaton was obtaned of the sequental progresson of dsplacements durng forgng. Fgures -5 sw ths progresson from the expermental measurements made on the ndvdual grd lnes for the 355 rpm set of welds. All measurements Fg. Interfacal entrapment for balked specmens were made to varous ponts along the centerlne of each grd flament from the lateral surface and free end surface of each welded specmen. Ths data was then converted to grd postons relatve to the axal centerlne and nterface for the comparatve plots. Also swn on the fgures are theoretcal dsplacements wthn the plastc zone, whch were based upon averaged ntal radal postons (r ), averaged ntal plastc zone sze (h ), and averaged upset (u p ) of each weld pece durng forgng. These averaged values were made for the two expermental welds at each upset level. For the concave zones the ntal characterstc plastc zone sze at the onset of forgng was establshed as the average heght of the fnal plastc profle plus the total weldng upset mnus the burnoff. All were taken on a per weld pece bass. The average heght of the fnal plastc zone was that heght for a flat zone whch would encompass the same amount of plastc materal as found under the concave profle Table. Note that the theoretcal dsplacements were prevously derved as: axal: y yo(h - u p y- - y Up () Table Nylon Grd Specmens" Burn-off, Ub (X. n.) Heatng tme, sec Upset u p (X. n.) Up (avg) (X. n.) (X. n.) u P (avg) (avg) Gradual Collar (Concave Zone) Splt Collar Flat Collar (Convex Zone) Fg. Plastc flow patterns and concdng resultant collar shapes h a (avg) =.83 n Speed 355 rpm; heatng load 6 ± 4.5 ps (constant); forgng load 5 ±.5 ps. WELDING RESEARCH SUPPLEMENT 43-s

8 E o OR 6 4 Expermen -al N \ s j * T j) ; d - \ v I! >' t I H I! I [ r ' ra f ' * ' ;! ' J'?! ff T w t k k * T /Theor etcal! T ll! Radal Dstance (") Fg. Plastc zone dsplacements durng forgng of grdded specmens, 355 rpm and u,,/h =.6! K '< U n * 4 t nylon O D.8 *.6.4 ) o o. Expermental '-+- A-."... Radal Dstance (") Fg. 3 Plastc zone dsplacements durng forgng of nylon grdded specmens, 355 rpm and u P /h =.566 Expermental Theoretcal Expermental. \ Radal Dstance (") Fg. 4 Plastc zone dsplacements durng forgng of nylon grdded specmens, 355 rpm and u,,/h =.87 w~. "P" -!. Radal Dstance (") Fg 5 Plastc zone dsplacements durng forgng of nylon dded specmens, 355 rpm and u /h =.4455 radal: r y_o Up h h (?) Convenent values of ntal axal postons wth respect to the weld nterface (y ) were csen for each computed grd lne. The results sw general agreement between the analytcal dsplacement felds and the measured ones, partcularly at the lower upset ratos where the concave zone profle s not fully developed. For hgher ratos the analytcal flow felds predcted greater movement of materal n the outer regons of the nterface than was actually observed. Thus the nterfacal dsperson (77) as derved n Part I from eq () may be an over-estmaton of the removal of oxdes, vods and contamnaton from the nterface: -o-tr (3) Note that a set of smlar nylon grdded specmens welded at 65 rpm and reported n a prevous paper 8 revealed the same correlaton and 44-s MARCH 97 trends as ndcated here for the 355 rpm welds. They also possessed dstnct concave plastc zone shapes. The observed dfferences n the theoretcal and expermental dsplacements pont out the sgnfcance of the plastc zone shape to the forgng phase.and to a future analytcal model of t. Conclusons The expermental study presented here revealed several mportant facts about the nature of forgng durng frcton weldng. The patterns of plastc flow durng the forge were seen to be greatly nfluenced by the shape of the rgd-plastc boundary at the weld nterface. In turn ths rgdplastc boundary shape appeared to correlate well wth the sotherms developed durng the heatng phase of the process. Fnally n affectng the plastc flow characterstcs, both the sotherms and resultng plastc zone shape had a prmary nfluence on the dsperson of orgnal nterfacal surface materal. Thus, they are a key to the control of both weld qualty and the resultng flash formaton. The value of concave and flat plastc zone shapes n deference to convex zones to the dsperson of nterfacal materal s of practcal mportance. Ths s partcularly useful snce t was swn to be qualtatvely ndcated by the shape of the resultng collar. Also of practcal mportance s the knowledge that control of the sotherms durng the heatng phase could lead to control of the plastc flow felds to produce a desred flash pattern and modes of dsperson consdered most useful for a partcular jont confguraton. The expermental study verfed the potental applcablty of the flat zone analytcal plastc flow model of Part I for estmatng dsplacements and the nterfacal dsperson. It also ponted out the need for a theoretcal heat transfer soluton enablng predcton of concave or convex sotherms. Ths n turn coupled wth a velocty feld compatble wth the correspondng contoured rgd-plastc boundary would provde refnement to the present soluton. Perhaps of equal mportance s the need for theoretcal and expermental nformaton to further evaluate the nterfacal dsperson n terms of optmzaton of weld qualty. Greater under-

9 standng of ts role n estmatng weld qualty could provde a powerful analytcal tool n future desgn of frcton welded jonts. A cknowledgement The research descrbed n ths paper was supported by the Materals Research Center of Lehgh Unversty and by the Natonal Aeronautcs and Space Admnstraton. References. Gel'man, A. S., "The Nature ol Frcton Weldng," Automatc Weldng, Vol. 8, March, 965, pp Hollander, M. B., Cheng, C. J., and Wyman, J. C, "Frcton Weldng Parameter Analyss," WELDING JOURNAL, 4(), Research Suppl. 495-s to 5-s (963). 3. Hazlett, T. H., and Gupta, K. K., "Frcton Weldng of Hgh Strength Structural Alumnum Alloys," Ibd., 4(), Research Suppl. 49-s to 494-s (963): Engneer's Dgest, Vol. 5, No pp Hodge, E. S., "Frcton Jonng," Battelle Techncal Revew, Vol. 4, August- Sept. 965, pp Rch, T., and Roberts, R., "Current Thermal Analyses for the Basc Frcton Weldng Process," Brtsh Weldng Journal, accepted for publcaton. 6. Vll, V. I., Frcton Weldng of Metals, translated from Russan, Amercan Weldng Socety, Inc., Renld Publshng Corp., Dwght, H., Tables of Integrals and Other Mathematcal Data, Fourth Edton, Macmllan Co., N. Y Rch, T., Dssertaton n Mechancal Engneerng, "An Exploratory Study of Frcton Weldng," Lehgh Unversty, Bethlehem, Pennsylvana, "Plastc Weldng Technques," Brtsh Plastcs, Vol. 36, No. 9, Sept. 963, pp Smpson, K., "Spn Weldng," Plascs, Vol. 3, No Sept. 965, pp Kamnsky, S. J.. "Desgn Welded Plastc Parts." Plastc Desgn and Processng, Vol. 5, No. 3, March 965, pp Zackson, R. I., and Voznesensky, V. D., "Power and Heat Parameters of Frcton Weldng," Weldng Producton, October 959, pp Vll, V. I., "Energy Dstrbuton n the Frcton Weldng of Steel Bars," Weldng Producton, Oct. 959, pp Appendx Axal Dsplacements The axal dsplacements of partcles wthn the plastc zones adjacent to the nterface durng forgng are dervable from the velocty expresson, eq (): df V ~ h) (Al) Employng relatonshps () and () yelds: d y = d Up Call: = - Up dy \ Up) - h (A) d (?)"] d d u v (A3) Call: Thus: y = x dy = xd + dx (A4) Substtuton back nto eq (A3) results n the followng separable frst order dfferental equaton: Xd\+ldx =(x ~X )d (A5) dx x - x =d Integraton of eq (A6) results n: or: In X X - ln+c c (A6) (A7) (A8) Substtuton back for x an d then results n the fnal form of the soluton: y = c(h - Up) c(h Up) ~ (A9) Applcaton of the ntal condton that at Up = o, y = y, leads to the determnaton of the constant of ntegraton: y a h a (A) Therefore, the soluton for the axal dsplacements wthn the plastc zone proves to be: Radal Dsplacements y ( - Up) y u p (All) The radal dsplacements are dervable from the velocty expresson, eq (5): d r dt' ' K-9 (A) Substtuton of the soluton for axal dsplacements, equaton (All), nto eq (A) results n the followng dfferental equaton: dr r = dut y h ( - u p )(h - Vo u p ) (A3) Note agan that the relatonshps () and () have been employed. Equaton (A3) can then be wrtten as: r ( y h ) d Up y«up (yo h + h ) Up + (A4) The rght hand sde of equaton (A4) upon ntegraton s of the form j dx /ax + bx + c. For ths problem the condtons on the ntegral are: b - 4 ac = (h - yof > o b >4ac (A5) Under condtons (A5) the general soluton of the ntegral eq (A4) s: dx ax + bx + c a(p - q) In (A6) as swn n (7) where p and q are the roots of ax + bx + c. In terms of eq (A4) the roots n terms of dmenson less ratos are: q "fe+'^te)'-'g+ yo (A7 "(g + '-l/fe)'-'h y (A8) Takng the postve root of the radcal,, leads to the followng values of p and q. Note that takng the negatve root,, has the same effect as swtchng the expressons for p and q whch results n the same fnal soluton: P = h yo (A9) Employng the results of (A9) to eq (A4) through the general soluton (Al6) yelds: In r = In yo h yo «pyo _ h h Up + c (A) j (A) Applyng the ntal condton that at Up =, r = r, leads to the determnaton of the constant of ntegraton: c = r yo (A) Thus the fnal soluton for the radal dsplacements wthn the plastc zone becomes: r = ro y_o Up - _ Up^ (A3) WELDING RESEARCH SUPPLEMENT 45-s

10 Symposum Announcement... Thursday, Aprl 9,97 The Hgh Alloys Commttee of the Weldng Research Councl wll sponsor a symposum on the followng topc on Aprl 9: "Characterstcs of Nckel-Base Alloy Weldments" The symposum s beng held n conjuncton wth the 5nd Annual Meetng of the AMERICAN WELDING SOCIETY between 9:3 A.M. and : noon n Room 4 of Cvc Audtorum, San Francsco, Calforna. Included wll be the presentaton of the followng srt papers: "Corroson of Nckel-Base Alloy Weldments n Superheated Steam" by J. P. Hammond, P. Patrarca, and G. M. Slaughter, Oak Rdge Natonal Laboratory, and W. A. Maxwell, Southern Nuclear Engneerng, Inc. "Acoustc Emsson Characterstcs of Postweld Heat-Treat Crackng n Rene 4 Weldments" by A. T. D'Annessa and J. S. Owens, Lockheed-Georga Company "A Mechansm for Crackng Durng Postweldng Heat Treatment of Nckel-Base Alloys" by M. Prager, Copper Development Assocaton, and G. Snes, Unversty of Calforna at Los Angeles "Metds for Improvng the Weldablty of Hgh Strength Super-Alloys" by D. S. Duvall and W. A. Owczarsk, Pratt and Whtney Arcraft "Some Weldng Problems wth Inconel 78" by J. Gordne, Canadan Department of Energy, Mnes and Resources "Progress n the Weldng of 5% Cr 5% N for Elevated Temperature Servce" by E. P. Sadowsk, Internatonal Nckel Co., Inc. All are nvted to attend and to partcpate n the audence dscusson whch s expected to ensue. 46-s MARCH 97