Documents

TQM

Description
MANAGEMENT
Categories
Published
of 8
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Related Documents
Share
Transcript
  Journal of Materials Processing Technology 164–165 (2005) 978–985 Effect of laser welding parameters on the heat input and weld-bead profile K.Y. Benyounis ∗ , A.G. Olabi, M.S.J. Hashmi School of Mechanical and Manufacturing Engineering, Dublin City University, Dublin 9, Ireland  Abstract Laser butt-welding of medium carbon steel was investigated using CW 1.5kW CO 2  laser. The effect of laser power (1.2–1.43kW), weldingspeed (30–70cm/min) and focal point position ( − 2.5 to 0mm) on the heat input and the weld-bead geometry (i.e. penetration ( P ), weldedzone width ( W  ) and heat affected zone width ( W  HAZ )) was investigated using response surface methodology (RSM). The experimental planwas based on Box–Behnken design. Linear and quadratic polynomial equations for predicting the heat input and the weld-bead geometrywere developed. The results indicate that the proposed models predict the responses adequately within the limits of welding parameters beingused. It is suggested that regression equations can be used to find optimum welding conditions for the desired criteria.© 2005 Elsevier B.V. All rights reserved. Keywords:  Laser welding; RSM; Weld-bead profile 1. Introduction Laserweldinghasbecomeanimportantindustrialprocessbecause of its advantages as a bonding process over the otherwidely used joining techniques. Laser welding characterizewith parallel-sided fusion zone, narrow weld width and highpenetration.Theseadvantagescamefromitshighpowerden-sity,whichmakethelaserweldingoneofthekeyholeweldingprocesses [1]. The laser welding input parameters determinethe shape of laser weld-bead, due to the combination of theseparameters control the heat input [2]. For a good weld qual-itythecombinationoftheoutputpower,weldingspeed,focalposition, shielding gas and position accuracy should be cor-rectly selected [3]. RSM is widely used to predict the weld- bead geometry and mechanical properties in many weldingprocess [4–8]. In this work RSM is used to develop models to predict the heat input and to describe the laser weld-beadprofile (i.e. weld penetration, welded zone width and HAZwidth) for CW CO 2  laser butt-welding of medium carbonsteel. The laser input parameters taken into consideration arelaserpower(LP),weldingspeed( S  )andfocusedposition( F  ). ∗ Corresponding author.  E-mail address:  khaled.benyounis2@mail.dcu.ie (K.Y. Benyounis). 2. Experimental design The experiment was designed based on a three levelBox–Behnken design with full replication [9]. Laser power(1.2–1.43kW), welding speed (30–70cm/min) and focalpoint position ( − 2.5 to 0mm) being the laser independentinput variables. Table 1 shows laser input variables and ex-perimental design levels used. RSM was applied to the ex-perimental data using statistical software, Design-expert V6.Linearandsecondorderpolynomialswerefittedtotheexperi-mentaldatatoobtaintheregressionequations.Thesequential F  -test,lack-of-fittestandotheradequacymeasureswereusedin selecting the best models. A step-wise regression methodwas used to fit the second order polynomial equation (1) to theexperimentaldataandtoidentifytherelevantmodelterms[10,11]. The same statistical software was used to generatethe statistical and response plots. Y   = b 0 +  b i χ i +  b ii χ 2 ii +  b ij  χ i χ j   (1) 3. Experimental work Mediumcarbonsteelwithchemicalcompositioninweightpercent of 0.46% C, 0.2% Si, 0.7% Mn and Fe Balancewas used as work piece material. The size of each platewas 180mm long × 80mm width with thickness of 5mm. 0924-0136/$ – see front matter © 2005 Elsevier B.V. All rights reserved.doi:10.1016/j.jmatprotec.2005.02.060  K.Y. Benyounis et al. / Journal of Materials Processing Technology 164–165 (2005) 978–985  979Table 1Process variables and experimental design levels usedVariables  − 1 0 +1Laser power, LP (kW) 1 . 2 1 . 3125 1 . 425Welding speed,  S   (cm/min) 30 50 70Focused position,  F   (mm)  − 2 . 5  − 1 . 25 0 Trial samples of butt-welding were performed by varyingone of the process variables to determine the working rangeof each variable. Absence of visible welding defects and atleast half depth penetration were the criteria of choosing theworking ranges. The experiment was carried out accordingto the design matrix in a random order to avoid any system-atic error using a CW 1.5kW CO 2  Rofin laser provided byMechtronic Industries Ltd. Argon gas was used as shield-ing gas with constant flow rate of 5l/min. Two transversespecimens were cut from each weldment. Standard metallo-graphic was made for each transverse specimen. The beadprofile parameters ‘responses’ were measured using an opti-cal microscope with digital micrometers attached to it withan accuracy of 0.001mm, which allow to measure in  X  -axesand  y -axes.Theaverageoftwomeasuredweldprofileparameterswasrecorded for each response. The design matrix and the aver-age measured responses are shown below in Tables 2 and 3. 4. Results and discussion The results of the weld-bead profile were measured ac-cording to design matrix Table 2 using the transverse sec-tioned specimens and the optical microscope mentioned ear-lier, the measured responses are listed in Table 3. Analysing the measured responses by the design expert software. Thefit summary output indicates that the linear model is statis-tically significant for the penetration ‘the second response’ Table 2Design matrix with code independent process variablesExp. No. RunorderLaserpower (kW)Welding speed(cm/min)Focusedposition (mm)1 1  − 1  − 1 02 8 1  − 1 03 13  − 1 1 04 14 1 1 05 4  − 1 0  − 16 16 1 0  − 17 10  − 1 0 18 3 1 0 19 5 0  − 1  − 110 7 0 1  − 111 9 0  − 1 112 6 0 1 113 11 0 0 014 17 0 0 015 2 0 0 016 15 0 0 017 12 0 0 0Table 3Experimental measured responsesExp. no. Heat input (J/cm)  P  (mm)  W   (mm)  W  HAZ  (mm)1 1920 3.572 2.358 0.5612 2280 4.322 2.805 0.8723 823 2.705 1.342 0.3924 977 3.651 1.852 0.3845 1152 2.655 2.761 0.4536 1368 3.888 3.381 0.5697 1152 3.813 2.087 0.5118 1368 4.539 2.572 0.5749 2100 3.905 3.681 0.62510 900 2.367 1.982 0.37511 2100 4.987 2.423 0.76212 900 3.824 1.649 0.41313 1260 3.712 2.625 0.53114 1260 3.872 2.282 0.56215 1260 3.586 2.567 0.46616 1260 3.505 2.413 0.47817 1260 3.626 2.293 0.506Table 4ANOVA table for heat input reduced quadratic modelSource Sum of squaresd.f. Mean square  F   value Prob> F  Model 3246465 4 811616 11507 <0.0001LP 111932.1 1 111932 1587 <0.0001 S   2880000 1 2880000 40833 <0.0001 S  2 243952.9 1 243952 3459 <0.0001LP × S   10579.56 1 10579 150 <0.0001Residual 846.3732 12 70.53Correctedtotal3247311 16  R 2 =0.9997; predicted  R 2 =0.9989.Table 5ANOVA table for penetration reduced linear modelSource Sum of squaresd.f. Mean square  F   value Prob>FModel 6.279 3 2.093 51 . 399 <0 . 0001LP 1.670 1 1.670 41 . 007 <0 . 0001 S   2.246 1 2.246 55 . 158 <0 . 0001 F   2.363 1 2.363 58 . 031 <0 . 0001Residual 0.529 13 0.041Lack-of-fit 0.451 9 0.050 2 . 560 0 . 190Pure error 0.078 4 0.020Corrected total 6.809 16  R 2 =0.922, predicted  R 2 =0.849; adjusted  R 2 =0.904, adequate preci-sion=21.931. therefore it will be used for further analysis. While for theother responses the quadratic models are statistically recom-mended for further analysis. 4.1. Analysis of variance (ANOVA) The test for significance of the regression models, the testforsignificanceonindividualmodelcoefficientsandthelack-of-fit test were performed using the same statistical package.By selecting the step-wise regression method, which elim-inates the insignificant model terms automatically, the re-  980  K.Y. Benyounis et al. / Journal of Materials Processing Technology 164–165 (2005) 978–985 Table 6ANOVA table for WZ width reduced quadratic modelSource Sum of squaresd.f. Mean square  F   value Prob> F  Model 5.140 6 0.857 58 . 732 <0 . 0001LP 0.531 1 0.531 36 . 440 0 . 0001 S   2.466 1 2.466 169 . 105 <0 . 0001 F   1.181 1 1.181 80 . 985 <0 . 0001 S  2 0.361 1 0.361 24 . 750 0 . 001 F  2 0.386 1 0.386 26 . 448 0 . 0004 S  × F   0.214 1 0.214 14 . 666 0 . 003Residual 0.146 10 0.015Lack-of-fit 0.048 6 0.008 0 . 330 0 . 891Pure error 0.098 4 0.024Correctedtotal5.286 16  R 2 =0.972, predicted  R 2 =0.922; adjusted  R 2 =0.956, adequate preci-sion=29.498.Table 7ANOVA table for HAZ width reduced quadratic modelSource Sum of squaresd.f. Mean square  F   value Prob>FModel 0.259 4 0.065 42 . 631 <0 . 0001LP 0.029 1 0.029 19 . 138 0 . 0009 S   0.197 1 0.197 129 . 953 <0 . 0001 F   0.007 1 0.007 4 . 666 0 . 0517LP × S   0.025 1 0.025 16 . 766 0 . 0015Residual 0.018 12 0.002Lack-of-fit 0.012 8 0.002 0 . 990 0 . 5436Pure error 0.006 4 0.002Correctedtotal0.277 16  R 2 =0.934, predicted  R 2 =0.861; adjusted  R 2 =0.912, adequate preci-sion=22.899.Fig. 1. Scatter diagram of heat input.Fig. 2. Scatter diagram of penetration. sulting ANOVA Tables 4–7 f or the reduced quadratic mod- els summarise the analysis of variance of each response andshow the significant model terms. The same tables show alsothe other adequacy measures  R 2 , adjusted  R 2 and predicted  R 2 . The entire adequacy measures are close to 1, which is inreasonable agreement and indicate adequate models. The ad-equate precision compares the range of the predicted value atthe design points to the average prediction error. In all casesthe value of adequate precision are dramatically greater than4. The adequate precision ratio above 4 indicates adequatemodel discrimination.The analysis of variance indicates that for the heat in-put model. The main effect of the laser power (LP), weldingspeed ( S  ), the second order effect of welding speed ( S  2 ) andthe two level interaction of laser welding and welding speed(LP × S  )arethemostsignificantmodeltermsassociatedwith Fig. 3. Scatter diagram of WZ width.  K.Y. Benyounis et al. / Journal of Materials Processing Technology 164–165 (2005) 978–985  981Fig. 4. Scatter diagram of HAZ width. heat input. Secondly for the penetration model, the analysisindicated that there is a linear relationship between the maineffects of the three parameters. Also, in case of welded zonewidth model the main effect of laser power (LP), weldingspeed ( S  ), focused position ( F  ), the second order effect of welding speed ( S  2 ), the second order effect of the focusedposition ( F  2 ) and the two level interaction of welding speedand focused position (SF) are significant model terms.However, the main effect of welding speed ( S  ) and themain effect of focused position ( F  ) are the most significantfactors associated with the welded zone width. Finally, forHAZ width model it is evident that the main effect of laserpower (LP), welding speed ( S  ), focused position ( F  ) and thetwo level interaction of the laser power and welding speed(LP × S  ) are significant model terms. However, the main ef-fectofweldingspeed( S  )isthemostimportantfactorinfluent Fig. 5. 3D graph show the effect of LP and  S   on the heat input.Fig. 6. Contours graph show the effect of LP and  S   on the heat input. the HAZ width. The final mathematical models in terms of coded factors as determined by design expert software areshown below:heat input = 1260 + 118 . 29 × LP − 600 × S  + 240 × S  2 − 51 . 43 × LP × S   (2) P   = 3 . 68 + 0 . 46 × LP − 0 . 53 × S  + 0 . 54 × F   (3) W   = 2 . 42 + 0 . 26 × LP − 0 . 56 × S  − 0 . 38 × F   − 0 . 31 × S  2 + 0 . 30 × F  2 + 0 . 23 × S  × F   (4) Fig. 7. 3D graph shows the effect of LP and  S   on penetration at F  = − 1.25mm.
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks