Weld Pl scillatin during GT Welding f Mild Steel The scillatin behavir f the GT weld pl depends n the welding cnditins and can be used fr in-prcess cntrl f weld penetratin BY Y. H. XI ND G. DEN UDEN BSTRCT. In this paper the results are reprted f a study dealing with the scillatin behavir f weld pls in the case f GT bead-n-plate welding f mild steel, Fe 30. During welding, the weld pl was brught int scillatin by applying shrt current pulses, and the scillatin frequency and amplitude were measured by mnitring the arc vltage. It was fund that the scillatin f the partially penetrated weld pl is dminated by ne f tw different scillatin mdes (Mde 1 and Mde 2) depending n the welding cnditins, whereas the scillatin f the fully penetrated weld pl is characterized by a third scillatin mde (Mde 3). It is pssible t maintain partially penetrated weld pl scillatin in Mde 1 by chsing apprpriate welding cnditins. Under these cnditins, an abrupt decrease in scillatin frequency ccurs when the weld pl transfers frm partial penetratin t full penetratin. Thus, weld penetratin can be in-prcess cntrlled by mnitring the scillatin frequency during welding. Intrductin It has been shwn in previus papers (Refs. 1-3) that the scillatin frequency and amplitude f a statinary GT weld pl, excited int scillatin by shrt current pulses, can be measured by mnitring the arc vltage. It appears that Y. H. XI and G. den UDEN are with the Department f Materials Science and Engineering Delft University f Technlgy, Delft, The Netherlands. the scillatin behavir f a partially penetrated weld pl is cnsiderably different frm that f a fully penetrated weld pl; the frequency f the partially penetrated weld pl being much higher and the amplitude being much lwer than thse f the fully penetrated weld pl. It is expected that these characteristics can be used as a pssible tl fr in-prcess cntrl f weld penetratin. The feasibility f such in-prcess cntrl is especially imprtant in the case f practical arc welding using a traveling arc. Fr this reasn it is interesting t find ut whether the weld pl in the case f traveling arc welding behaves in a similar way as in the case f statinary arc welding. s during traveling arc welding, the arc mves relative t the wrkpiece, the weld pl will becme elngated and the welding arc will shift frm the gemetric center f the weld pl twardthe frnt edge f the weld pl. Thus, it is evident that the scillatin behavir f the weld pl i n the case f travel i ng arc KEY WRDS GTW Mild Steel Weld Pl scillatin scillatin Mde Weld Pl Gemetry Weld Penetratin Partial Penetratin Full Penetratin In-Prcess Cntrl Real-Time Cntrl welding is cnsiderably mre cmplex than that in the case f statinary arc welding. Srensen and Eagar (Ref. 4) prpsed an scillatin mdel fr the partially penetrated weld pl bth fr statinary and traveling arc welding and als measured the scillatin frequency by mnitring the arc vltage variatin. Their results shw that the measured scillatin frequency in the case f statinary arc welding is higher than the scillatin frequency which is calculated, whereas in the case f traveling arc welding the measured frequency is lwer than that calculated. Maru and Hirata (Ref. 5) measured the scillatin frequency f partially penetrated weld pls by means f highspeed phtgraphy, and fund that the scillatin frequency varied between 20 and 80 Hz, values which lie within the range bserved by Srensen and Eagar. Madigan, etal. (Ref. ), measured the scillatin frequency by mnitring the variatin f arc vltage using a cmputer system and reprted that the scillatin frequency ranged frm 1 50 t 400 Hz fr partially penetrated weld pls and frm 38 t 1 88 Hz fr fully penetrated weld pls. These results indicate that there is an verlap range fr the scillatin frequencies f the partially penetrated weld pls and the fully penetrated weld pls. Deam (Ref. 7) measured the scillatin frequency f weld pls by mnitring the arc radiatin, making use f the phenmenn that the intensity f the arc radiatin is prprtinal t the arc length, and fund that the scillatin frequency f partially penetrated weld 428-s I UGUST 1993
pls is cnsiderably higher than that f fully penetrated weld pls. In this paper the results are presented f a study dealing with the scillatin behavir f weld pls in the case f traveling GT welding, using a methd based n the measurement f arc vltage variatins. Special attentin is given t the different scillatin mdes which can ccur and t the transitin frm partial penetratin t full penetratin. Experimental Prcedure Bead-n-plate weld pls were prduced by GT welding in -mm-thick plates (partial penetratin) and 4-mmthick plates (partial penetratin and full penetratin) f mild steel Fe 30. Welding was carried ut using a 2% thriated tungsten electrde with a diameter f 2.4 mm and a tip angle f 0 deg at straight plarity (electrde negative) with helium r argn as shielding gas. In the case f full penetratin, argn was used as backing gas. During welding, the trch was fixed while the test plate mved with respect t the trch. The weld pls were excited int scillatin by applying shrt current pulses, and the scillatin frequency and amplitude were measured during welding by mnitring the arc vltage in the same way as described in a previus paper (Ref. 2). The experimental setup is schematically given in Fig. 1. The welding pwer surce used was a three-phase full-wave rectifier machine (Philips PZ 2351) prvided with a pulse generatin unit. ll pulse parameters (base current \^, peak current L, base current duratin t^, and peak current duratin t p (Fig. 2) were digitally cntrlled by the pulse generatin unit. The standard welding cnditins used are given in Table 1. fter welding, crss-sectins were Table 1 Standard Welding Cnditins peak current () base current () effective current () pulse frequency (Hz) peak current duratin (ms) arc length (mm) travel speed (cm/min) shielding gas backing gas plate thickness (mm) Partial Penetratin 80-0 127 5-25 1.2-22 1.5 He, 24 L/min made f each weld. These crss-sectins were grund, plished and etched and subsequently used t determine the gemetry f the welds. Results and Discussin Partial Penetratin scillatin Mdes Full Penetratin 50-5 75-85 12.5 2.4 1.5 He, 24 L/min r, 5 L/min 4 Transitin frm Partial t Full Penetratin 30-5 0-85 12.5 2.4 1.5 He, 24 L/min r, 5 L/min 4 Using the prcedure described abve, a large number f experiments were carried ut. It appears that in the case f the partially penetrated weld pl, tw different scillatin mdes (Mdes 1 and 2) can ccur depending n the experimental cnditins. Mde 1 is characterized by a high scillatin frequency and a lw scillatin amplitude, Mde 2 by a lwer scillatin frequency and a higher scillatin amplitude. Under certain cnditins the develpment f Mde 2 may give rise t the frmatin f the s-called humping bead (Ref. 5). The bserved scillatin behavir can be described mathematically by applying the principles f classical hydrdynamics t the liquid metal in the weld pl (Ref. 8). ssuming the liquid metal in the weld pl t be incmpressible and inviscid, the mtin f the liquid beys Laplace's equatin = 0 (1) with < > the velcity ptential f the liquid mtin and 0 = 0 (2) at the weld pl bundary. Equatin 1 can be slved fr a statinary (circular) weld pl, taking int the accunt the pressure balance n the weld pl surface. This leads t the fllwing expressin fr the frequency f f the weld pl scillatin f (2K)- k + ^-k- \tanh(2kh) P, ) (3) where y is the surface tensin f the liquid metal, p^ the density f the liquid metal, g the gravitatinal cnstant, h the depth f the weld pl and k the wave number f the surface wave. The value f k is determined by the bundary cnditin expressed by Equatin 2. This bundary cnditin can be rewritten as k l = 7) (4) pltter data acquisitin system scillscpe welding mnitr time Fig. 1 Schematic illustratin f experimental setup. Fig. 2 Wavefrm f current pulse. WELDING RESERCH SUPPLEMENT I 429-s
kd/2 = 4.09 RZT" kd/2 = 11.04 CS) f^7 kd/2 = 17.28 m K^r kd/2-7. kd/2 = 14.03 (D C^ XZ7 kd/2 = 20.34 m "V^WN^TH welding directin [*) (b) (c) (d) (e) (0 (a) mde 1 (b) mde 2 F/g. 3 scillatin mdes f the statinary weld pl (+: psitive displacement f the weld pl surface in the directin f arc pressure; : negative displacement f the weld pl surface in the directin f arc pressure). Fig. 4 scillatin mdes f the traveling weld pl. Mde 2. -Mde I; B where J n is the nth rder Bessel functin f the first kind, and D the diameter f the weld pl. The different rts f Equatin 4 crrespnd with different mdes f scillatin. In Fig. 4 a number f scillatin mdes is presented tgether with the crrespnding values f k D/2. It appears that under practical welding cnditins the tw mdes depicted in Fig. 3B and 3D are dminant. These mdes, dented as Mde 1 and Mde 2, respectively, can be easily bserved by means f high-speed filming (Ref. 3) and are presented in Fig. 4 fr the case f the traveling weld pl. The frequency f the tw scillatin mdes can be calculated frm Equatin 3 by substitutin f apprpriate values f k D/2 (11.04 fr Mde 1 and 7. fr Mde 2) and taking fr the surface tensin and the density f the liquid metal 1 N/m and 7 x 3 kg/m 3, respectively (Ref. 2). This leads, after neglecting the term gk with respect t the term y/pi, k 3 and taking tanh (2kh) = 1 (2kh > 2K), t the fllwing tw equatins scillatin Mde 1: / = 0.07 Dy (5) Fig. 5 The measured scillatin frequency f f the partially penetrated weld pl as a functin f D{ 3/2 (peak current, arc length 1.5 mm, travel speed cm/min and helium as shielding gas). The straight lines represent Equatins 5 and. >, 200 150 0 50 / scillatin Mde 2: / = 0.04Dr v2 () with D, the diameter f the circle having a surface area equal t the surface area f the weld pl. lthugh, strictly speaking, Equatin 3 is nly valid fr the case f a statinary (circular) weld pl, it can als be used fr an apprximate descriptin f the situatin f the traveling (elngated) weld pl. In Fig. 5 the measured scillatin frequency is pltted as a functin f D," 3/2 fr the tw scillatin mdes, taking fr D-^ the width f the weld pl. The straight lines in the figure represent Equatins 5 and. lthugh the experimental pints are characterized by cnsiderable scatter, the results presented in Fig. 5 clearly shw that there is gd agreement between experiment and thery. The fact that the experimental pints lie smewhat belw the straight lines predicted by thery, can be understd by realizing that the equivalent diameter D} is larger than the real width f the weld pl, because in the case f traveling arc welding the weld pl is elngated in the travel directin...- ^ mde 1 mde 2 0.00 0.02 0.04 0.0 0.0B 0. 0.12 0.14 D- 3/2 (mm- 3/2 ) Effect f Welding Parameters n the scillatin Behavir s was mentined in the previus sectin, the scillatin f the weld pl is excited by a pressure pulse (current pulse). In rder t btain a better insight int the mechanism f excitatin, it is useful t cnsider the effect f the pressure pulse n the surface f the weld pl in mre detail. In the case f statinary GT welding, the arc is lcated exactly abve the center f the weld pl. During the perid f high arc pressure, the pl surface is symmetrically depressed and liquid metal is pushed in radial directin. This leads t the surface shape shwn in Fig. 3B. fter the peak current is shut ff, the weld pl will start t scillate in Mde 1. In the case f traveling GT welding, the weld pl is elngated, and the arc has an asymmetric psitin with respect t the weld pl center. If the duratin f the pressure pulse is shrt, the depressin f the weld pl surface will be small and the liquid metal in the weld pl will be pushed predminantly in radial directin as indicated in Fig. 4, which will result in scillatin Mde 1. When the pressure pulse duratin is increased, the weld pl surface belw the arc is pushed dwn ver a larger area, and a cnsiderable amunt f liquid metal is driven frm the frnt t the rear f the weld pl. s a result f this frnt-t-back flw, the weld pl surface takes the shape shwn in Fig. 4B as bserved by high-speed filming (Ref. 3). This will eventually lead t the generatin f scillatin Mde 2. The freging implies that the decisive factr, determining the scillatin behavir f the weld pl, is the pressure pulse duratin, i.e., the peak current duratin t. In fact, the scillatin behavir can be characterized by a critical peak current duratin t, /hen P c.',, t p < t pc the weld pl will scillate in 430-s I UGUST 1993
200 150 0 50 [> "\ i I 1 0 s. - -- -- a Ip = lp = tf 350 0 7 -? E 5 JZ ** 4 Q Q D d 2 W h - mde Xs^^ mde 2 i.. «2 " 0 f! 12 20 0 250 :l:')0 1 peak current duratin (ms) Fig. scillatin frequency f and weld pl gemetry f the partially penetrated weld pl as a functin f peak current duratin t p fr peak currents f and 350 (arc length 1.5 mm, travel speed cm/min, helium as shielding gas). peak current Fig. 7 Effect f peak current l p n t pc (arc length 1.5 mm, travel speed cm/min, helium as shielding gas). () Mde 1, when t p > t pc scillatin in Mde 2 will ccur. It is evident that the value f t pc depends n the peak current and als n ther welding parameters. In rder t find ut under which experimental cnditins which f the tw scillatin mdes is generated by the current pulse, t pc was measured as a functin f peak current, arc length and travel speed with helium and argn as shielding gas. This was dne by measuring the scillatin frequency as a functin f peak current duratin. In rder t eliminate pssible effects f weld pl gemetry variatins n the scillatin frequency, the heat input was kept as cnstant as pssible thrughut the measurements by adapting the level f the effective arc current l e t the value f the travel speed and by keeping l e cnstant during each measurement. s an illustratin, the results f tw series f experiments are given in Fig.. In this figure bth scillatin frequency and weld gemetry (width and depth) are pltted as a functin f t p fr peak currents f and 350, using an arc length f 1.5 mm, a travel speed f cm/min and helium as shielding gas. s can be seen in this figure, the scillatin frequency drps at a certain value f t p (defined as the critical peak current duratin t pc ), while the weld pl size remains almst cnstant. The results f the measurement f t pc as a functin f experimental cnditins (peak current, arc length and travel speed) with helium as shielding gas are presented in Figs. 7 t 9. The results shw that t pc decreases with increasing peak current, arc length and travel speed. Similar results were btained with argn as shielding gas. In the case f argn, hwever, the values f t pc are cnsiderably lwer than in the case f helium. pparently, the generatin f Mde 1 scillatin is favred by shrt peak current time, lw peak current, shrt arc length and lw travel speed and als by using helium as shielding gas. These findings can be easily understd by cnsidering the surface depressin f the weld pl due t the arc pressure. Increasing the peak current and using argn instead f helium as shielding gas will result in a higher arc pressure, whereas with increasing arc length, a larger part f the pl surface will be affected by the arc pressure. Furthermre, with increasing travel speed the weld pl will becme mre elngated and the arc will mve further away frm the pl center. It is evident that all these effects, separately r in cmbinatin, will result in an increase f the pressure difference between the frnt and the rear f the weld pl and thus will prmte the frnt-t-back flw, leading t Mde 2 scillatin. The fact that t pc decreases with increasing arc length is cnsistent with the results btained by Savage, et al. (Ref. 9), wh fund that the travel speed limit t avid the ccurrence f humping decreases with increasing arc length. Full Penetratin It has been shwn (Ref. 2) that in the case f statinary GT welding, the scillatin behavir f a partially penetrated weld pl differs significantly frm that f a fu I ly penetrated weld pl. In fact, the fully penetrated weld pl can be regarded as a stretched membrane, the mtin f which is cntrlled by the surface tensin f the tw surfaces. scillatin ccurs in a mde (Mde 3), which is schematically illustrated in Fig. and which is cnsiderably different frm thse f a partially penetrated weld pl Fig. 4. The frequency f f this scillatin mde is given WELDING RESERCH SUPPLEMENT I 431-s
3 \ mde 2 2-4 - 0 15 mde 2 mde 0. mde 1 0 1 2 3 4 arc length (mm) Fig. 8 Effect f arc length L n t pc (peak current, travel speed cm/min, helium as shielding gas). travel speed (cm/min) 15 20 Fig. 9 Effect f travel speed v n t pc (peak current, arc length 1.5 mm, helium as shielding gas). by the equatin 7 = 1.08 up* D;> (7) in which y is the surface tensin f the liquid metal, p s the density f the slid metal, H the plate thickness and D 2 the diameter f the equivalent cylinder, the vlume f which equals the vlume f the fully penetrated weld pl. By substituting fr y, p s and H, the values 1 N/m, 7.9 x 3 kg/m 3 and 4 mm respectively, the fllwing simple equatin is btained fr the scillatin frequency f the fully penetrated weld pl scillatin Mde 3: / = 0.2D7 1 (8) It must be nted that Equatins 7 and 8 were develped fr the case f statinary arc welding. T find ut whether these equatins are als valid fr the case f traveling arc welding, the measured scillatin frequency was pltted as a functin f D 2 _1. The results are shwn in Fig. 11. The straight line in this figure represents Equatin 8. It can be seen that nly fr relatively large values f D 2 reasnable agreement exists between calculatin and experiment and that the difference between the experimental data and the theretical predictin increases with decreasing value f D 2. This can be understd by realizing that the theretical predictin (Equatin 7) is based n the assumptin that the weld pl has a cylindrical shape, whereas the real weld pl has a shape which deviates frm a cylinder. This deviatin is maximum at small weld pl size (when the weld pl is just fully penetrated and the bttm width is very small with respect t the tp width) and decreases with increasing weld pl size. It is t be expected that the descriptin f the scillatin behavir f the fully penetrated weld pl by Equatin 7 will als becme less accurate with increasing travel speed, since with increasing travel speed the weld pl tp surface will shift with respect t the bttm surface f the weld pl, and the bulk mtin f the weld pl will be mre and mre ff the vertical directin. Transitin frm Partial Penetratin t Full Penetratin n the basis f the results presented abve, it may be expected that the transitin frm partial penetratin t full penetratin will give rise t an abrupt change in scillatin behavir. It is be- Fig. Schematic illustratin f scillatin Mde 3. 0.00 0. 0.20 0.30 0.40 0.50 D" 1 (mm" 1 ) Fig. 11 The measured scillatin frequency f f the fully penetrated weld pl (Mde 3) as a functin f D 2 ~ 1 (peak current, arc length 1.5 mm, travel speed cm/min, helium as shielding gas). The straight line represents Equatin 8. 432-s I UGUST 1993
arc pressure 250 m m m m 200 --... """ --0 150 - ': 0 50 ': *-...... *" c ""-... 48 50 52 54 5 50 0 2 4 base currenl () Fig. 12 scillatin frequency f and weld pl gemetry as a functin f base current tu (transitin frm Mde I t Mde 3). arc pressure 200 : 5 150 0. c 3 arc pressure 50 #-- -0 weld pl bltm width (mm) Fig. 13 The scillatin frequency f f a fully penetrated weld pl as a functin f the bttm width W^ f the weld pl (transitin frm Mde 1 t Mde 3). Fig. 14 Schematic illustratin f surface depressin under different weld pl cnditins. lieved that this transitin can be used fr in-prcess cntrl f weld penetratin. Especially, the transitin frm Mde 1 t Mde 3 is f imprtance in this respect. In rder t study the transitin frm partial penetratin t full penetratin in mre detail, welding experiments were carried ut n 4-mm-thick plates f mild steel Fe 30. In these experiments, weld pl size was gradually increased (frm partial penetratin t full penetratin) by increasing the base current, leaving the ther welding parameters unchanged. T preclude the generatin f Mde 2 scillatin in the case f partial penetratin, the welding parameters were chsen in such a way that nly Mde 1 scillatin culd ccur (see abve). It was fund that during the grwth f the weld pl, the scillatin frequency at first decreases gradually frm 33 t 204 Hz in the range f partial penetratin, then drps abruptly after reaching a certain degree f penetratin and finally decreases again slwly frm 1 00 t 37 Hz in the range f full penetratin. These results are cnsistent with the results reprted by Madigan, etal. (Ref. ), wh fund that the scillatin frequency f partially penetrated weld pls ranges frm 400 t 150 Hz and that f fully penetrated weld pls frm 188 t 38 Hz. san illustratin f the transitin frm partial penetratin t full penetratin, the weld pl gemetry and the crrespnding scillatin frequency are given as a functin f base current in Fig. 1 2. The weld pl gemetry (weld bead prfile) was btained frm the crss-sectin f the welds after welding. When cnsidering Fig. 1 2 in mre detail, it appears that the drp in scillatin frequency des nt exactly cincide with the transitin frm partial penetratin t full penetratin. This is illustrated even mre clearly in Fig. 1 3, in which the frequency f the fully penetrated weld pl is pltted as a functin f the width W b f the bttm surface f the weld pl. Figure 1 3 shws that at a relatively small value f W b the weld pl still scillates in scillatin Mde 1 (high frequency, lw amplitude) and that an abrupt decrease in scillatin frequency ccurs at W b ~ 2 mm, a value crrespnding t abut half the plate thickness. T explain why the transitin f scillatin Mde ccurs at a value f Wb > 0, the surface depressin f the weld pl under different cnditins shuld be cnsidered. During the high pressure actin f the arc n the weld pl (peak current), the surface depres- WELDING RESERCH SUPPLEMENT I 433-s
sin f a partially penetrated weld pl is radially symmetric with respect t the axis f the arc, since the liquid metal is supprted by slid material Fig. 14. This leads t Mde 1 scillatin. In the case f a large fully penetrated weld pl, there is n lnger any slid material at the bttm t supprt the liquid metal in the weld pl and, as a result f this, the liquid metal in the weld pl is pushed dwn Fig. 14B. When the pressure is lifted, the liquid metal will start t mve up again due t the actin f the surface tensin, which results in Mde 3 scillatin. In the situatin where the weld pl is fully penetrated but W b is still very small, a large part f the weld pl is backed by the slid material Fig. 14C. During the pressure pulse, nly a small part f the liquid metal is pushed dwnward, the remainder being pushed in a radial directin. Thus, under these cnditins, Mde 1 scillatin is dminant in the weld pl. With increasing W b, less liquid metal is backed by slid material and mre liquid metal is pushed in dwnward directin. When Wb has reached a certain value, dwnward mvement f the liquid metal becmes dminant and the weld pl scillatin transfers frm Mde 1 t Mde 3. It is expected that the abrupt change in scillatin behavir, which accmpanies the transitin frm partial penetratin t full penetratin has practical significance, as it can be used as a means fr in-prcess cntrl f weld penetratin. The results f preliminary tests carried ut under labratry cnditins indicate that in-prcess cntrl f weld penetratin by mnitring the scillatin frequency is indeed pssible. t present, preparatins are being made fr experiments aimed at testing the feasibility f the methd fr ther materials and under real welding cnditins. Cnclusins This paper deals with the scillatin behavir f weld pls during GT welding f mild steel Fe 30. n the basis f the experimental results btained, the fllwing cnclusins can be drawn: 1) During welding, the weld pl can be brught int scillatin by applying shrt current pulses. The natural frequency and the amplitude f the scillatin can be measured by mnitring the arc vltage variatin. 2) In the case f the partial penetratin weld pl, tw different scillatin mdes can ccur. ne f these mdes can give rise t weld bead humping and can be suppressed by selecting apprpriate welding cnditins. 3) The scillatin f the fully penetrated weld pl is gverned by an scillatin mde, which differs significantly frm thse ccurring in the partially penetrated weld pl. 4) The bserved scillatin behavir f bth the partially penetrated weld pl and the fully penetrated weld pl can be explained in terms f classical hydrdynamics, taking int accunt the pressure balance n the weld pl surface 5) The abrupt change in scillatin frequency, accmpanying the transitin frm partial penetratin t full penetratin can be used as a means fr in-prcess cntrl f weld penetratin. cknwledgments The authrs wuld like t thank F. J.. M. Bsman and W.. J. Brabander fr technical assistance during the curse f the experiment. References 1. Renwick, R. J., Farsn, D. F., and Richardsn, R.W. 198. Experimental investigatin f CT weld pl scillatins. Welding Jurnal 2 (2): 29-s t 35-s. 2. Xia, Y. H and uden, G. den. 1990. study f GT weld pl scillatin. Welding Jurnal 9 (8): 289-s t 293-s. 3. Xia, Y. H., and uden, G. den. 1992. Direct bservatin f weld pl scillatin. Prc. 3rd Intern. Cnf. n Trends in Welding Research, Gatlinburg, Tenn., U.S.. 4. Srensen, C. D., and Eagar, T. W. 198. Digital signal prcessing as a diagnstic tl fr gas tungsten arc welding. Prc. 2nd Intern. Cnf. n Trends in Welding Research, Gatlinburg, Tenn., pp. 47^172. 5. Maru, H., and Hirata, Y. 1987. Study n pulsed TIG arc welding. Technlgy Reprts f saka University, 37: 51-3.. Madigan, R. B., Renwick, R. J., Farsn, D. F., and Richardsn, R. W. 198. Cmputer cntrl f full penetratin GT welds using pl scillatin sensing. Prc. 1 st Cnf. n Cmputer Technlgy in Welding, The Welding Institute, Lndn: 15-174. 7. Deam, R. T. 1989. Weld pl frequency: new way t define a weld prcess. Prc. 2nd Intern. Cnf. n Trends in Welding Research, Gatlinburg, Tenn., pp. 97-971. 8. Landau, L. D., and Lifshitz, E. M. 1987. Fluid Mechanics (2nd Editin), Lndn, U.K., Pergamn Press:, pp. 244-245. 9. Savage, W. F., Nippes, E. F., and gusa, K. 1979. Effect f arc frce n defect frmatin in GT welding. Welding Jurnal 58 (7): 212-st224-s. 434-s I UGUST 1993