Evlution of FDM Process Prmeter for PLA Mteril by Using MOORA-TOPSIS Method 1 Priynk B Ptel, 2 Jksn D. Ptel, 3 Klpesh D. Mniy 1 M.E. Student, Merchnt Engineering College, Mehsn-384315, Gujrt, Indi. 2 Assistnt Professor,Deprtment of Mechnicl Engineering, Merchnt Polytechnic College, Bsn, Mehsn-384315 Gujrt, Indi. 3 Assistnt Professor,Deprtment of Mechnicl Engineering, C.K. Pithwll College of Engg. And Tech., Surt-395007 Gujrt, Indi. Abstrct: Fused Deposition Modelling (FDM) is one of the rpid prototype process tht produce prototypes from plstic mterils such s ABS, PLA, Nylon, etc. It is process tht cretes prts in n dditive lyer by lyer mnner. In FDM process, the criticl fctors re selected for mking component to mesure different properties. The design investigtes the effect of the process prmeters lyer thickness, orienttion nd infill on the tensile strength, tensile module, compressive strength, compressive module, nd surfce roughness. Experiments re conducted using Tguchi s design of experiments with three levels for ech fctor. Experiments were crried out on FDM replictor 2 mchines coupled with Mker Wre TM softwre nd PLA s min mteril. Tensile nd compressive specimens were prepred s per the ASTM stndrd. Multi-objective optimiztion on the bsis of rtio nlysis (MOORA) nd technique for order preferences by similrity to n idel solution (TOPSIS) method re used to find the rnking of FDM process prmeters nd lso compre the results of MOORA nd TOPSIS Method. Keywords: FDM, Lyer Thickness, Orienttion, Infill, MOORA, TOPSIS Method. 1. INTRODUCTION Reduce the product development cycle time is mjor concern in industries to remin competitive in the mrket nd hence, focus hs shifted from trditionl product development methodology to rpid fbriction techniques like rpid prototyping.[4] The Fused Deposition Modelling (FDM) is typicl exmple of RP process, leding to the forementioned chrcteristics. The FDM is ble to produce prototypes from plstic mterils, such s crylonitrile butdiene styrene (ABS) or polylctic cid (PLA), nd the process consists in the deposition of filments of the mteril t the semi-molten stte.[6] The filment is feed through nozzle nd locted t the output of heting device, nd is deposited on to the prtilly constructed prt. Since the mteril is extruded nd lid in trcks t semi-molten stte, the newly deposited mteril fuses with djcent mteril tht hs lredy been deposited. Afterwrds, other mteril trcks re deposited, upon the completion of the current lyer, nd then the deposition of new lyer is strted. Reserch community lredy benefits from the vilbility of low-cost 3D printers in tht the mchines such s the mkerbot replictor llow experimenttion with vriety of esily progrmmble technologicl prmeters.[7] The study presented in this pper differs from the discussed investigtions in two key points: (i) the mteril used in this study is polylctic cid (PLA), which, contrry to ABS, hs not been extensively used in experiments of this kind, nd (ii) the infill used to produce specimens rnges between 100 nd 98%. 2. LITERATURE R EVIEW Ryegni et l. (2014) found tht both process prmeters ffect tensile strength. Negtive ir gp nd smller rster widths improve tensile strength. The zero prt orienttion mximum tensile strength is obtined. Incresed rster ngle lso improves tensile strength. Mrcincinov et l. (2012) presented different types of testing in the mterils properties of selected methods of rpid prototyping technologies. Sood et l. (2011) hve studied the effect of five importnt FDM Pge 84
mchine process prmeters nd found tht fibre-fibre bond strength must be strong which cn be chieving by controlling the distortions rising during prt build stge. Optimiztion of process prmeters gives the mximum compressive stress of 17.4751 MP nd the optimum vlue of lyer thickness, orienttion, rster ngle, rster width nd ir gp s 0.254 mm, 0.036 degree, 59.44 degree, 0.422 mm nd 0.00026 mm respectively. Nnchrih et l. (2010) they were found tht the lyer thickness nd rod width ffect the surfce qulity nd prt ccurcy gretly. Rster ngle hs little effect. But ir gp hs more effect on dimensionl ccurcy nd little effect on surfce qulity. Sood et l. (2009) studied the influence of importnt process prmeter nd they conclude tht mximiztion of grey reltionl grde shows tht lyer thickness of 0.178 mm, prt orienttion of 0 degree, rster ngle of 0 degree, rod width of 0.4564 mm nd ir gp of 0.008 mm will produced overll improvement in prt dimensions. Glntucci et l. (2009) found tht the slice height nd rster width re importnt prmeters while the tip dimeter hs little importnt for surfce running either prllel or perpendiculr to the build direction. Pnd et l. (2009) hve used ltest evolution ny bcteril forging lgorithm to predict optiml prmeters setting of FDM process. After the experimentl work they hve find out tht the lyer thickness nd orienttion ngle is highly significnt prmeters for FDM fbricted prts wheres remining prmeter hve little effect. Aim of this present study is selection of process prmeter of FDM mchine for polylctic cid mteril using MADM method. The responses considered in this study re mechnicl property of FDM produced prts such s tensile strength (T s ), tensile module (T m ), compressive strength (C s ), compressive module (C m ) nd surfce roughness (SR). The specimens re prepred s per the ASTM stndrd t three different prmeter nd level such s lyer thickness (100, 200, 300) (micron), orienttions (0, 45, 90 ) nd infill (100%, 99%, 98%). 3. EXPERIMENTAL PROCEDURE Specimens re fbricted using the FDM replictor 2 mchine. The prts re modelled in modelling softwre nd exported s STL file. STL file is imported to FDM softwre. The mteril used for specimen preprtion is polylctic cid (PLA). For mesuring tensile (ASTM D638) nd compressive (ASTM D695) test respective stndrd specimens hving respective dimensions 115mm X 19mm x 4mm for tensile nd 12.7mm in dimeter nd 25.4mm length for Compressive re prepred. Experimentl run re crete in minitb16. Orthogonl rry L9 re develop in the tguchi shows in tble 1. After fbricting the specimens, these specimens were tested. Tensile nd compressive test is conducted on INSTRON 5965 nd 5982 mchine. And surfce roughness mesure by using surfce roughness tester SJ210. The specimens fter testing re depicted in fig. 1 nd 2. And testing results re shown in Tble 1. Fig.1 Tensile specimen fter test Fig.2 Compressive specimen fter test Tble 1: Experimentl dt obtined from the L9 orthogonl rry Exp. Lter thickness Orienttion (T Infill (%) s ) (T m ) (C s ) (C m ) (SR) No (micron) (degree) (µm) 1 100 0 100 49.09 2246.51 57.68 1621.85 2.82 2 100 45 99 53.13 2849.02 30.53 570.82 3.03 3 100 90 98 55.71 3460.73 39.43 1413.65 2.13 4 200 0 99 39.79 2082.42 56.72 1701.20 4.30 5 200 45 98 54.27 2931.98 35.56 1087.45 2.68 6 200 90 100 51.49 2997.07 54.43 1892.66 2.01 7 300 0 98 36.50 1802.38 54.37 1353.65 2.44 8 300 45 100 47.60 2803.67 47.73 1015.89 2.55 9 300 90 99 49.09 2977.42 52.12 1800.29 2.47 Pge 85
4. MULTIPLE ATTRIBUTE DECISION MAKING METHODS 4.1 Anlytic Hierrchy Process / Multi-Objective Optimiztion on the Bsis of Rtio Anlysis (AHP/MOORA Method) This section describes the proposed integrted AHP/MOORA method for selection of pproprite FDM mchine. The AHP method is potentil decision mking tool developed by Sty (1980) while the MOORA method, is introduced by Bruers (2004) In the pst mny decision mking pplictions were reported using MOORA method. The min steps of the proposed model re described below. Steps of the AHP method s follows: [13] Step 1: Define the problem. This step is ssocited with to define the objective nd identifiction of ll the possible lterntives nd its ttributes. Let A = {A i for i = 1,2,3, m} be set of FDM mchine lterntive, B = {B j for j =1,2,3,,n } be set of decision criteri or ttributes of FDM mchine lterntive selection problem, nd x ij is the performnce of lterntive A i when it exmined with criteri B j. Step 2: Developing the hierrchicl structure. A decision problem is structured s hierrchy structure With the AHP, the gol, decision criteri nd lterntives re rrnged in hierrchicl structure similr to fmily trees shown in fig. 3. Gol Selection Criteri B1 B2 B M Alterntive A1 A2 An Fig.3 A hierrchy of the decision mking problem [13] Step 3: Generte pir wise mtrices. A pir wise comprison mtrix is constructed using scle of reltive importnce s shown in Tble 2. Let, there re M ttributes re involved in the decision mking, the pir wise comprison of ttribute i with ttribute j yields squre mtrix A1 =M x M =[ ij ] M x M. Where ij denotes the comprtive importnce of ttribute i with respect to ttribute j. In the mtrix, ij = 1 when i = j nd ji = 1/ ij. A1 MxM B1 1 B2 21 B3 31 BM M1 12 1 32 M2 13 23 1 M3 1 1 1M 2M 3M 1 Tble 2: Scle of Reltive importnce [13] Scle Importnce Mening of ttributes 1 equl importnce Two ttributes re eqully importnt 3 moderte importnce One ttribute is modertely importnt over the other 5 strong importnce One ttribute is strongly importnt over the other 7 very importnce One ttribute is very importnt over the other 9 Absolute importnce One ttribute is bsolutely importnt over the other 2,4,6,8, compromise importnce between 1,3,5,7 nd 9 Pge 86
Step 4: Determintion of reltive normlized weight. A reltive normlized weight t ech level of hierrchy structure is clculted using Eqution (1) nd Eqution (2). GM M j ij j1 GM j Wj M GM j1 j 1 M (1) (2) If the judgment mtrix or comprison mtrix is inconsistent then judgment should be reviewed nd improved it to obtin the consistent mtrix. Hence, consistency test will be crried out using following steps. Clculte mtrices; A 3 = A 1 x A 2 nd A 4 = A 3 / A 2, Where; A 1 = [r ij ] m m, A 2 =[W 1,W 2,.,W j ] T Clculte Eigen vlue mx (verge of mtrix A 4 ) Clculte the consistency index: CI = ( mx - m) / (m - 1) Clculte the consistency rtio: CR = CI/RI, select vlue of rndom index (RI) Tble 3 ccording to number of ttributes used in decision-mking. If CR < 0.1, considered s cceptble decision, otherwise judgment of the nlyst bout the problem under study. Steps of the MOORA method s follows: [2], [3] Tble 3: Rndom Index (RI) for different mtrix order [13] Attributes 1 2 3 4 5 6 7 8 9 10 RI 0 0 0.52 0.89 1.11 1.25 1.35 1.4 1.45 1.49 Step 5: Construct the decision mtrix. Here 9 (lterntives A 1 to A 9 ) process prmeters of FDM. Response process prmeters of the FDM mchine such s tensile strength, tensile module, compressive strength, compressive module nd surfce roughness. Step 6: Find the dimensionless number or normliztion vlue. Let R ij is dimensionless number which belongs to the intervl zero to one representing the normlized performnce of i th lterntive on j th ttribute. This R ij vlue is clculted s suggested by Bruers. It cn be expressed s below: (3) Step 7: Determine the normlized performnce of lterntive. In this step, the normlized performnce of lterntives is determined with considering weightge of selection criteri involved in the decision mking process. For multi-objective optimiztion, these normlized performnces re dded in cse of mximiztion (for beneficil ttributes) nd subtrcted in cse of minimiztion (for non beneficil ttributes). ( ) (4) Where, g is the number of ttributes to be mximized, (n-g) is the number of ttributes to be minimized, w j is the weight of j th ttribute, which cn be determined pplying nlytic hierrchy process method s described in step3 nd step 4, nd y is the normlized performnce vlue of i th lterntive with respect to ll the ttributes. Step 8: Rnking nd selection of lterntive. The vlue of y vlue cn be positive or negtive depending of the totls of its mxim (beneficil ttributes) nd minim (non-beneficil ttributes), A rnking of lterntive will be crried out bsed on vlue of y nd finlly, the best lterntive is considered who hs the highest y vlue or rnked first while the worst lterntive hs the lowest y vlue or rnked lst. Pge 87
4.1.1 Illustrtion of Exmple Using AHP/MOORA Method Step 1: Decide the ll the possible lterntive for given ppliction, its selection criteri, nd its vlues. In present study, nine experiments is lterntives with five ttributes, the ttributes re tensile strength, tensile module, compressive strength, compressive module nd surfce roughness. Step 2: A FDM process prmeters selection problem cn be decomposed procedure described in the hierrchy structure shown in fig. 4. Fig.4 A hierrchy of FDM process prmeters selection problem Step 3: A reltive importnce of between ttributes is ssigned with respect to the gol. The judgments re entered using scle of reltive importnce of the AHP method s shown in Tble 4. Tble 4: Pir Wise Comprison Mtrix for Different Criteri Attribute B 1 B 2 B 3 B 4 B 5 B 1 1 1 3 3 4 B 2 1 1 2 3 3 B 3 1/3 1/2 1 1 4 B 4 1/3 1/3 1 1 2 B 5 1/4 1/3 1/4 1/2 1 Step 4: A reltive normlized weight of ttributes is clculted using Eq. (1) nd Eq. (2). Here determined the criteri weights s: W Ts = 0.3475, W Tm = 0.3025, W Cs = 0.1566, W Cm = 0.1253, W SR = 0.0681. Further, the vlue of CR is 0.0374. Therefore CR vlue less thn 0.1, the judgments re cceptble. These criteri weights were used for the MOORA method-bsed nlysis. Step 5: Present study totl 9 experiments (Alterntives A1 Up to A9) re considered using Tguchi concept nd the response process prmeters of the FDM such s tensile strength, tensile module, compressive strength, compressive module, nd surfce roughness re s shown in Tble 5 s decision mtrix. Tble 5: Decision mtrix tble Alterntive (T s ) (T m ) (C s ) (C m ) (SR) (µm) A 1 49.09 2246.51 57.68 1621.85 2.82 A 2 53.14 2849.02 30.53 570.82 3.03 A 3 55.71 3460.73 39.43 1413.65 2.13 A 4 39.79 2082.43 56.72 1701.20 4.30 A 5 54.27 2931.98 35.56 1087.45 2.68 A 6 51.49 2997.07 54.43 1892.67 2.01 A 7 36.50 1802.38 54.37 1353.65 2.44 A 8 47.60 2803.67 47.73 1015.90 2.55 A 9 49.10 2977.42 52.12 1800.29 2.47 Pge 88
Step 6: Using Eq. (3) determine the x i is dimensionless number which belongs to the intervl [0, 1] representing the normlized performnce of response process prmeters of FDM s show in Tble 6. Alterntive (T s ) Tble 6: Dimensionless number (xi) for ech lterntive (T m ) (C s ) (C m ) (SR) (µm) A 1 0.1163 0.0830 0.0620 0.0470 0.0230 A 2 0.1259 0.1053 0.0328 0.0165 0.0247 A 3 0.1319 0.1279 0.0424 0.0409 0.0174 A 4 0.0942 0.0769 0.0610 0.0493 0.0350 A 5 0.1285 0.1083 0.0382 0.0315 0.0218 A 6 0.1220 0.1107 0.0585 0.0548 0.0163 A 7 0.0865 0.0666 0.0585 0.0392 0.0199 A 8 0.1127 0.1036 0.0513 0.0294 0.0208 A 9 0.1163 0.1100 0.0560 0.0521 0.0201 Step 7 nd 8: For multi objective optimiztion, these normlized performnces re dded in cse of mximiztion (for beneficil ttributes) nd subtrcted in cse of minimiztion (for non-beneficil ttributes). tensile strength, tensile module, compressive strength nd compressive module re considered s beneficil ttribute (i.e. higher vlues re desirble), surfce roughness is considered s non-beneficil ttribute (i.e. lower vlues re desirble).using Eq. (4) clculte the weighted ssessment vlue. The best lterntive hs the highest y i vlue, while the worst lterntive hs the lowest y i vlue s shown in Tble 7. Tble 7: Weighted ssessment vlues (yi) nd rnking for selection of the process prmeters of FDM Alterntive (T s ) (T m ) (C s ) (C m ) (SR) (µm) Weight 0.3475 0.3025 0.1566 0.1253 0.0681 - - A 1 0.1163 0.0830 0.0620 0.0470 0.0230 0.2853 4 A 2 0.1259 0.1053 0.0328 0.0165 0.0247 0.2558 7 A 3 0.1319 0.1279 0.0424 0.0409 0.0174 0.3258 2 A 4 0.0942 0.0769 0.0610 0.0493 0.0350 0.2465 8 A 5 0.1285 0.1083 0.0382 0.0315 0.0218 0.2848 5 A 6 0.1220 0.1107 0.0585 0.0548 0.0163 0.3297 1 A 7 0.0865 0.0666 0.0585 0.0392 0.0199 0.2308 9 A 8 0.1127 0.1036 0.0513 0.0294 0.0208 0.2763 6 A 9 0.1163 0.1100 0.0560 0.0521 0.0201 0.3144 3 4.2 TOPSIS METHOD Technique for order preferences by similrity to n idel solution (TOPSIS), known s clssicl multiple ttribute decision-mking (MADM) method, hs been developed in 1981. In TOPSIS method, the optiml lterntive selected should hve the shortest distnce from the positive idel solution nd the frthest distnce from the negtive idel solution. The procedure cn be ctegorized in six steps: [17] Step 1: Creting the decision mtrix. The method strts with decision mtrix of responses of different lterntives to evlution criteri. Step 2: Construct normlized decision mtrix. This step trnsforms vrious ttribute dimensions into non-dimensionl ttributes, which llows comprisons cross criteri. Normlize scores or dt s follows: y i Rnk (5) Pge 89
Step 3: Construct the weighted normlized decision mtrix by multiplying the normlized decision mtrix by its ssocited weights. Here weightge of ech output prmeters re clculted using Anlyticl hierrchy process.the weighted normlized vlue v ij is clculted s: Step 4: Determine the positive idel solution nd negtive idel so (6) { } ( ) { } ( ) Where J is ssocited with the benefit criteri, J = 1, 2, 3 n Where J is ssocited with the cost criteri, J = 1, 2, 3 n Determine Idel Solution Vj*. Vj* = {V1*, V2* Vn*} Determine Negtive Idel Solution Vj. Vj = {V1, V2 Vn} Step 5: Clculte the seprtion mesures for ech lterntive. The seprtion of ech lterntive from the positive idel one is given by: ( ) (9) Where i = 1, 2 m Similrly, the seprtion of ech lterntive from the negtive idel one is given by: ( ) (10) Where i = 1, 2 m Step 6: Clculte the reltive closeness to the idel solution Ci* nd rnk the preference order. 4.2.1 Illustrtion of Exmple Using TOPSIS Method Step 1: Construct the decision mtrix s shown in Tble 5. ( ) (11) Where i = 1, 2 m Step 2 Normlize the decision mtrix D by using the Eq. (5) nd shown in Tble 6. Step 3: Construct the weighted normlized decision mtrix using Eq. (6) by multiplying the normlized decision mtrix by its ssocited weights. Here weightge of ech output prmeters re clculted using Anlyticl hierrchy process. The weighted normlized vlue v ij is s shown in Tble 8. Tble 8: Weighted normlized decision mtrix Alterntive (T s ) (T m ) (C s ) (C m ) (SR) (µm) A 1 0.1163 0.0830 0.0620 0.0470 0.0230 A 2 0.1259 0.1053 0.0328 0.0165 0.0247 A 3 0.1319 0.1279 0.0424 0.0409 0.0174 A 4 0.0942 0.0769 0.0610 0.0493 0.0350 A 5 0.1285 0.1083 0.0382 0.0315 0.0218 A 6 0.1220 0.1107 0.0585 0.0548 0.0163 A 7 0.0865 0.0666 0.0585 0.0392 0.0199 A 8 0.1127 0.1036 0.0513 0.0294 0.0208 A 9 0.1163 0.1100 0.0560 0.0521 0.0201 Pge 90
Step 4: Determine the positive idel solution nd negtive idel. Determine Idel Solution V j * using Eq. (7). Vj* = {0.1319, 0.1279, 0.0620, 0.0548, 0.0163} Determine Negtive Idel Solution V j ' using Eq. (8). Vj' = {0.0865, 0.0666, 0.0328, 0.0165, 0.0350} Step 5: Clculte the seprtion mesure using Eq. (9) the seprtion of ech lterntive from the positive idel one is given by: Tble 9: Positive idel solution Vj* 0.1319 0.1279 0.0620 0.0548 0.0163 A * Idel Solution S i * A 1 0.00024439 0.00201564 0.00000000 0.00006131 0.00004459 0.0486 A 2 0.00003657 0.00051232 0.00085100 0.00146448 0.00007048 0.0542 A 3 0.00000000 0.00000000 0.00038418 0.00019209 0.00000116 0.0240 A 4 0.00141777 0.00259677 0.00000101 0.00003060 0.00034921 0.0663 A 5 0.00001128 0.00038296 0.00056484 0.00054320 0.00003013 0.0391 A 6 0.00009869 0.00029462 0.00001205 0.00000000 0.00000000 0.0201 A 7 0.00206467 0.00375839 0.00001249 0.00024327 0.00001290 0.0780 A 8 0.00036736 0.00059097 0.00011392 0.00064409 0.00002006 0.0417 A 9 0.00024357 0.00032006 0.00003546 0.00000709 0.00001459 0.0249 Similrly, the seprtion of ech lterntive from the negtive idel one is given by: using Eq. (10) nd shown in Tble 10. Tble 10: Negtive idel solution Vj 0.0865 0.0666 0.0328 0.0165 0.035 A * Idel Solution S i A 1 0.00088608 0.00026910 0.00085404 0.00092842 0.00014454 0.0555 A 2 0.00154863 0.00149502 0.00000000 0.00000000 0.00010618 0.0561 A 3 0.00206542 0.00375366 0.00009215 0.00059733 0.00031062 0.0826 A 4 0.00006001 0.00010695 0.00079483 0.00107375 0.00000000 0.0451 A 5 0.00176744 0.00174145 0.00002953 0.00022480 0.00017454 0.0628 A 6 0.00125783 0.00194795 0.00066193 0.00146790 0.00034844 0.0754 A 7 0.00000000 0.00000000 0.00065873 0.00051542 0.00022827 0.0374 A 8 0.00068818 0.00136827 0.00034324 0.00016696 0.00020224 0.0526 A 9 0.00088763 0.00188440 0.00054034 0.00127006 0.00022143 0.0693 Step 6: Clculte the reltive closeness to the idel solution by using Eq. (11) nd rnk the preference order s shown in Tble 11. Tble 11: Reltive closeness to the idel solution Alterntive * C i Rnk A 1 0.5330 6 A 2 0.5088 7 A 3 0.7746 2 A 4 0.4049 8 A 5 0.6158 4 A 6 0.7892 1 A 7 0.3242 9 A 8 0.5581 5 A 9 0.7356 3 Pge 91
5. RESULT & DISCUSSION Here, bsed on evlution criteri weights obtined by AHP, the rnking for selection of the process prmeters of FDM using MOORA nd TOPSIS method, s present in Tble No 12. MOORA nd TOPSIS rnking results show tht lterntive 6-3-9 is the best three choices mong the 9 lterntives. Results we found tht 100micron lyer thickness, 90 orienttion nd 98% infill get optimum result of ll response. Tble 12: A result comprison of MOORA nd TOPSIS Alterntive Rnking result MOORA method TOPSIS method A 1 4 6 A 2 7 7 A 3 2 2 A 4 8 8 A 5 5 4 A 6 1 1 A 7 9 9 A 8 6 5 A 9 3 3 6. CONCLUDING REMARKS The present work concluded tht in order to get effective selections of FDM mchine using PLA mteril; it is necessry to consider possible lterntives nd ttributes. The MADM method, the AHP provides opportunity to select the best lterntive of FDM mchine considering with multi ttributes hving different mesures. The priority or rnking of lterntives depends on ttributes weight or reltive importnce ssigned between ttributes nd on the vlues of the selected ttributes. The AHP cn hndle tngible (objective) s well s non-tngible (subjective) ttribute mesures. It hs been observed tht MOORA method is very simple, stble nd robust. It requires minimum Mthemticl clcultions nd computtionl time. REFERENCES [1] Bkr N, Alkhri MR, Boejng H, 2010. Anlysis of Fused Deposition Modeling, Journl of Zhejing University-Science A (Applied physics nd science), 11, 972-977. [2] Bruers W, 2008. Multi objective decision mking for rod design, journl of Trnsporttion, 23, 183 193. [3] Bruers W, 2009. Robustness of the multi objective MOORA method with test for the fcilities sector technologicl nd economic development of economy, Journl on Sustinbility, 15, 352 375. [4] Croccolo D, Agostinis MD, Olmi G, 2013. Experimentl chrcteriztion nd nlyticl modelling of the mechnicl behvior of FDM prts mde of ABS-M30, Computtionl Mterils Science, 79, 506 518. [5] Glntucci LM, Lvecchi F, Percoco G, 2009. Experimentl study iming to enhnce the surfce finish of fused deposition modeled prts, Mnufcturing Technology, 58, 189 192. [6] Hoon AS, Michel M, Dn O, Shd R, Pul WK, 2002. Anisotropic mteril properties of fused deposition modeling ABS, Rpid Prototyping Journl, 8, 248-257. [7] Luznin O, Movrin D, Plnck M, 2014. Effect of lyer thickness, deposition ngle, nd infill on mximum flexurl force in FDM-built specimens, Journl for Technology of Plsticity, 39, 49-57. [8] Mrcincinov LN, Mrcincin JN, 2012. Testing of mterils for rpid prototyping fused deposition modelling technology, World Acdemy of Science, Engineering nd Technology 70, 411-414. [9] Nnchrih T, Rju DR nd Rju VR, 2010. An experimentl investigtion on surfce qulity nd dimensionl ccurcy of FDM components, Interntionl Journl on Emerging Technologies, 2, 106-111. Pge 92
[10] Pnd SK, Pdhee S, Sood A, Mhptr SS, 2009. Optimiztion of fused deposition modeling (FDM) process prmeters using bcteril forging technique, Intelligent Informtion Mngement, 1, 89-97. [11] Ptel VB, Ptel JD, Mniy KD, 2014. Selection of wire cut electric dischrge mchining mchine using Anlyticl hierrchy process method, Ntionl Conference on Therml Fluid nd Mnufcturing Science, 2, 24-25. [12] Ryegni F, Godfrey C, 2014. Fused deposition modeling (FDM) process prmeter prediction nd optimiztion using group method for dt hndling (GMDH) nd differentil evolution (DE), Journl of Engineering Mnufcture,73, 509-519. [13] Sty TL, 1990. How to mke decision: The Anlytic Hierrchy Process, Europen journl of opertionl reserch, 48, 9-26. [14] Sood A, Phdr RK, Mhptr SS, 2009. Improving dimensionl ccurcy of fused deposition modeling processed prt using grey tguchi method, Mteril nd Design, 30, 4243-4252. [15] Sood A, Phdr RK, Mhptr SS, 2011. Experimentl investigtion nd empiricl modelling of FDM process for compressive strength improvement, Journl of Advnce Reserch, 3, 81-90. [16] Tymrk BM, Kreiger M, Perce JM, 2014. Mechnicl properties of components fbricted with open-source 3-D printers under relistic environmentl conditions, Mterils nd Design, 58, 242 246. [17] Yoon K, Hwng CL, 1981. Multiple ttributes decision mking-methods nd pplictions. Berlin: Springer Verlg. Pge 93