Parametric optimization of fused deposition modeling and vapour smoothing processes for surface finishing of biomedical implant replicas

Accepted Manuscript Parametric optimization of fused deposition modeling and vapour smoothing processes for surface finishing of biomedical implant replicas Jasgurpreet Singh Chohan, Rupinder Singh, Kamaljit Singh Boparai PII: DOI: Reference:

S0263-2241(16)30501-2 http://dx.doi.org/10.1016/j.measurement.2016.09.001 MEASUR 4318

To appear in:

Measurement

Received Date: Revised Date: Accepted Date:

9 November 2015 24 June 2016 1 September 2016

Please cite this article as: J.S. Chohan, R. Singh, K.S. Boparai, Parametric optimization of fused deposition modeling and vapour smoothing processes for surface finishing of biomedical implant replicas, Measurement (2016), doi: http://dx.doi.org/10.1016/j.measurement.2016.09.001

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

PARAMETRIC OPTIMIZATION OF FUSED DEPOSITION MODELING AND VAPOUR SMOOTHING PROCESSES FOR SURFACE FINISHING OF BIOMEDICAL IMPLANT REPLICAS Jasgurpreet Singh Chohan Ph. D. Research Scholar, I.K.G. Punjab Technical University, Kapurthala, 144601, India E-mail: [email protected] Rupinder Singh Professor, Production Engineering Department, GNDEC, Ludhiana, 141006, India Email: [email protected] Kamaljit Singh Boparai Assistant Professor, Mechanical Engineering Department, RIMT-IET, Mandi Gobindgarh, 147301, India Email: [email protected]

ABSTRACT This study focuses on formulation of robust design for vapour smoothing, an advanced surface finishing technique for finishing ABS replicas where hot vapours tend to level the uneven surface asperities. The process parameters of combined Fused Deposition Modeling (FDM) and Vapour smoothing (VS) process are optimized for sustainability of ABS replicas for biomedical applications. Six input parameters have been investigated, two of FDM and four of VS processes while surface roughness and hardness of ABS part is taken as response. The vapour smoothing process ensue ultra smooth finish with negligible deterioration of upper surface deducing maximum contribution of smoothing time (51.07%) and number of cycles (40.08%) on surface roughness. Hardness of replica has been slightly increased by maximum impact of orientation angle (34.69%) and postcooling time (44.46%) of ABS replicas which endorsed the use of FDM replicas for investment casting of biomedical implants. Keywords: Vapour smoothing, fused deposition modeling, shore d hardness, ABS replicas

1

1. Introduction Additive Manufacturing (AM) technology is a set of diverse Rapid Tooling techniques which emphasized on production of complex geometries precisely; simultaneously reducing the manufacturing time and cost. Alternately called as Layer Manufacturing or freeform fabrication, this new race of production techniques eliminates the use of traditional tools, jigs, fixtures, dies with minimum human intervention [1]. Fused Deposition Modeling (FDM) is gaining widespread use as it integrates Rapid Prototyping and Computer Aided Design technology while providing flexibility of using different materials, shapes to achieve desired properties. Being the simplest, adaptable and cost effective technique leading to minimum wastage of material [2], FDM has become matter of interest of many researchers as it bridges the gap between product conceptualization and realization [3]. FDM has nozzle head which moves in X and Y direction whereas semi molten plastic bead is extruded from nozzle and ultra-thin layers got precisely settled on fixtureless base [4] as shown in Fig. 1.

Fig. 1. Schematic diagram of Fused Deposition Modeling apparatus

In recent years, the demand of high quality products in lesser time and cost has intensified the Rapid Tooling application of Fused Deposition Modeling (FDM). The possibility of manufacturing plastic replicas through FDM has opened a new field of Rapid Tooling where low cost castings are directly manufactured via Investment Casting (IC) Process [5]. Rapid Tooling has been applied to manufacture patterns for casting of jewellery, sports equipments, ornaments, toys, automobiles aerospace parts and medical implants [6, 7]. The production of replicas of biomedical applications via FDM has opened the wide scope which facilitates to customize the implants as per requirements. Recent efforts have been focused on manufacturing sacrificial replicas via FDM for preparing Investment Castings (IC) of biomedical implants which would revolutionize the field of medical science for producing patient specific artificial bones at low production cost and time [8]. Although castings produced via IC route have excellent surface finish and dimensional accuracy [9] but surface asperities of replicas are emulated on surface of castings which could not removed even by controlling various process parameters of IC. The poor surface finish restricts the use of FDM parts as replicas for investment casting as roughness of surface is inherited by the final casting which would further required costly and time consuming surface finishing processes. This would add to the manufacturing cost and moreover original dimensions of product cannot be retained [10] which could bring complexities as implants being inserted into human body. The researchers have found that surface finish of casting could not be improved beyond surface finish of replicas. The surface roughness of castings is directly associated with the surface roughness of replicas and former can only be reduced by adopting suitable surface finishing of latter [11]. Poor surface finish of FDM parts is the major obstruction against fabrication of replicas of biomedical implants. Poor surface quality is an intrinsic defect of fused deposition 2

modeling as “STL” format approximates the part surface as set of triangles which represent the product’s outline in place of curves and circular shapes [12]. The layer by layer deposition behaviour of rasters (roads) tend to create peaks and valleys on upper surface resulting in surface roughness. A broader framework influences the surface characteristics of parts including selection of machine, part material and processing parameters but there is more flexibility to change the pre and postprocessing parameters at production stage as compared to machine and material. The preprocessing parameters are various input parameters used to manufacture FDM parts and they play an important role in controlling the dimensional accuracy, production time, surface roughness and other mechanical properties [13]. In context of preprocessing parameters, the decrease in layer thickness and road width reduces the surface roughness but this increased production time which further adds to the production cost [12, 14]. The extrusion and envelop temperatures of FDM have significant effect on structural homogeneity and density of parts which further affects surface finish and hardness [15]. Further different base materials and strengthening fluids resulted in variable surface roughness [16]. The orientation angles of 0° and 90° are found to have yielded maximum surface roughness but practical problem arises when all the surfaces manufactured at different angle are equally important [14]. Moreover all the preprocessing optimization studies have been performed on standard test parts while geometry and orientation of actual parts is somehow complex which would deviate from the optimized values. Broadly visualizing, all the additional processes involved to achieve smooth surface finish of raw FDM parts comes under postprocessing techniques which are further divided into two parts i.e. Mechanical finishing and chemical finishing. In Mechanical finishing, the conventional finishing techniques are utilized where tools cut-away to peaks of surface roughness profile via micro cutting. Initially, Kulkarni and Dutta [17] have utilized CNC machining with ball end mill cutter followed by adaptive slicing of CAD model to smoothen the different locations of FDM parts. Pandey et al. [18] developed, modeled and optimized the parameters of hot cutter machining process and found orientation angle and cutting direction to be significant parameters. Although barrel finishing and vibratory grinding are highly efficient mass finishing techniques but high surface finish is achieved at cost of higher material removal rates which disturbs dimensional tolerances [1, 19]. The defects like edge cutting, rounding of corners are experienced which limits the functionality of FDM parts. The chemical finishing techniques involves the use of specific chemicals which tend to smoothen the parts either by eroding or by partially melting the upper surface of ABS parts. These processes induced very minute dimensional changes in FDM parts as compared to mechanical finishing and resulted in ultra smooth surface. Researchers [20] have experimented chemical finishing of FDM parts by dipping in acetone solution which for few minutes which erodes away the surface asperities. The acetone which acts as solvent and thinning agent in industry, improves the surface finish. The acetone also enhances flexural strength and ductility of parts but there was shrinkage by less than 1% of their original dimensions [21]. The acetone vapour polishing has been successfully implemented by many researchers and reported good surface improvement with 3

increase in compressive and flexural strengths [22] while reduction in tensile strength [23]. The acetone being strong chemical reacts to upper surface and erodes away the thin sections which resulted in rounded corners, thus reduces the aesthetic and functional value of parts [24]. An advanced chemical finishing technique introduced as Vapour Smoothing (VS) has been developed by Stratasys which utilizes the hot chemical fumes to smoothen the upper surface of FDM parts [25]. The preliminary research regarding dimensional accuracy of ABS test parts has been performed by Espalin et al. [26] which showed minimal dimensional variations before and after vapour exposure. Hitherto, no research work has been reported regarding surface roughness measurement and its variation after vapour exposure. Moreover, no efforts have been made to study the different process parameters of vapour smoothing apparatus and their impact on surface roughness, hardness. The aforesaid process could prove a breakthrough for commercialization of IC process for production of biomedical implants. But, it is feared that penetration of vapours on upper surface could degrade the surface hardness which would further result in complexities like shrinkage, surface deformity and low compressive strength of replicas during IC process. Thus to thoroughly approve the combined FDM-VS process for biomedical applications, it is required to investigate their suitability in terms of surface finish and hardness which is mandatory for IC sacrificial replicas. The present study has been carried out of scrutinize the impact of various factors of coupled FDM-VS process on surface roughness and hardness of ABS plastic replicas.

2. Methodology An optimization study has been performed to chalk out impact of different input parameters and their interactions on responses and selection of most significant parameters using Taguchi and ANOVA techniques. The numeric controlled extrusion head of FDM follows different tool path and deposition strategies [27] based on part design which could bring variation in surface roughness and hardness of parts. Thus, it is required to perform vapour smoothing on actual replicas of biomedical implant instead of standard test specimen used by previous researchers [14, 20, 21, 26]. So in present case, the experiments have been executed using original geometry of hip implant (Fig. 2) to thoroughly validate the affectivity of VS process for biomedical applications.

Fig. 2. Specimen of hip implant replica

2.1 Experimental Setup The commercial “CPT 12/14” hip implant used as benchmark for present study has been acquired from local orthopaedician supplied by Zimmer Biomet meant for hip replacement surgery of 63 year old female with B.M.I. 27.3. The variation of surface 4

roughness due to sloping profiles and complex geometrical features has been driving force behind using actual commercial implant geometry as a benchmark. The Coordinate Measuring Machine (CMM) supplied by Mitutoyo (Crysta-Apex C163012) having 0.1µm resolution has been configured for reverse engineering the part geometry as per ISO 103602:2009 regulations [28]. The length, diameter, thickness and elevation of various features have been surveyed by 113 touch points to sketch the CAD model of hip implant. Commercial ‘Solidworks’ software has been utilized for dimensioning and processing the scanned data, while slicing and tool path generation has been done by ‘CatalystEx’. The thickness and width of stem varies along the length while head diameter increases from top to bottom (Fig. 2). The curvilinear profile of neck section raised the intricacy of the part geometry which highlights importance of accurate design to minimize post operative complications [29]. The ABS-P400 material has been used to manufacture replicas through Stratasys “u-Print SE” FDM apparatus. Initially, the parts are hanged in cooling chamber for precooling of parts for few minutes as recommended by manufacturer [30]. Afterwards, the parts are hanged in smoothing chamber for few seconds followed by postcooling in cooling chamber (Fig. 3). The cooling coils are connected to refrigerator unit and maintained at 0°C temperature in order to trap the vapours arising for recirculation. The smoothing fluid (supplied by Microcare) is a highly volatile chemical (boiling point 43°C) with composition of Decafluoropentane (30%) and Trans-Dichloroethylene (70%) [31].

Fig. 3. Schematic of Vapour smoothing process

The heaters are maintained at 65°C in smoothing chamber where smoothing fluid is heated to evolve vapours which enters the surface of ABS plastic parts. The vapours present inside and on the part surface are exhausted out by a fan inside the cooling chamber. 2.2 Selection of Process Parameters and their limits In present study, the two preprocessing parameters i.e. orientation angle (A) and part density (B) are selected as preprocessing parameters of FDM apparatus which may affect the initial and final surface roughness and hardness [3, 12]. Layer thickness has been kept constant (0.254 mm) due to machine limitations. The parts produced at different orientation angles would yield different the surface roughness values while different densities may vary the surface hardness. Moreover, effect of cooling and smoothing cycles can be different for different orientation angles and densities which are required to be studied. The two orientation angles i.e. 0° and 90° are selected based on previous literature [18] as optimum angles to achieve best surface finish. The FDM apparatus has capability to manufacture parts through three different densities i.e. low, high and solid based on stacking of the interior rasters.

5

Table 1 Control parameters of FDM and VS apparatus

Table 1 depicts the various control used for experimentation. The pressure, concentration of smoothing fluid, temperature of smoothing chamber heaters and refrigeration system for cooling coils could not be altered [30]. The lower and upper limits of precooling (C), smoothing (D) and postcooling (E) times has been selected based on the operation manual [24]. The precooling and postcooling times has been recommended between 10 to 20 minutes by the manufacturer while smoothing time has been suggested to vary within 10 to 40 seconds depending upon size of the part. The pilot experiments have been performed on FDM replicas so as the selected DOE would yield best results. In pilot experiments, the effect of three smoothing times i.e. 15, 30 and 45 seconds has been studied on surface quality of FDM parts. It was noticed that parts smoothed above 30 seconds acquired blisters due to excessive heating which damaged the surface texture and part dimensions. Thus, the range of smoothing time has been kept less than 30 seconds to avoid any deterioration of part surface. The manufacturer recommends repeating the whole precooling-smoothing-postcooling process until desired surface finish is achieved. Espalin et al. [26] reported improves finish with negligible dimensional changes and deterioration in surface quality after three smoothing cycles. Thus, number of cycles (F) is taken as sixth input parameter with three levels (Table 1). The successive cooling and smoothing of ABS parts i.e. precooling and smoothing may interact with each which can produce errors in estimation of parameters. Thus the interaction effects are considered while investigating the process parameters. The initial surface roughness and hardness of FDM parts may vary insignificantly and may lead to an error during analysis. So the percentage change in surface roughness (%∆Ra) and surface hardness (%∆HD) is selected as output parameters using formula (1). Percentage change = [(Initial value-Final value) / Initial value] x100

(1)

There are few environmental factors like temperature, humidity which cannot be controlled during experimentation and could bring unwanted variation (noise) in results. In case of uncontrolled noise factors Signal to Noise ratio (SN ratio) is considered and for selecting the robust experimental design, the parameters yielding maximum signal to noise ratio are considered. 2.3 Experimental Design Matrix To derive the clear and accurate conclusions from observations, the proper planning and execution of experiment is of utmost importance. The study has been focused on development of robust design for surface finishing of ABS replicas of biomedical implants used for investment casting. Further the work aims to establish the relationship between various pre and postprocessing parameters on percentage change in surface roughness (%∆Ra) and hardness (%∆HD). The Taguchi design of experiment technique has been selected to perform experiments for evaluating the performance of VS and minimizing impact 6

of noise factors to achieve consistent surface finish and hardness. Taguchi Design Orthogonal arrays (OA) are most commonly used statistical technique for designing effective systems through optimization [32]. In present study, total six parameters are selected out of which one has two levels while other five has three levels each. The minimum number of experiments (N) required for Taguchi OA can be calculated using the formula [32]: N = [(L-1)*P] + 1

(2)

Where L is number of levels and P is number of parameters. Considering six parameters with three levels (maximum), we get N=13 using the formula (2). But number of experiments must be multiple of 2 and 3 which indicated minimum 18 experiments. Minitab 17 also suggested similar output when data regarding number of parameters and levels was given as input. Using full factorial design, the number of experiments would have increased to 486 leading to wastage of time, material and cost. The Taguchi’s experimental matrix was analyzed with Minitab 17 to investigate the impact of input parameters and their interaction on responses in all the 18 experiments. Signal to Noise ratio measures the sensitivity of response being investigated in controlled manner with respect to the external noise factors which are uncontrolled. The main effects plot of SN ratios have been used to select optimum combinations of input parameters for maximum output response. The rank of influencing parameters on output parameters has been determined using signal to noise ratios. It has been considered as tool to sort out most and least influencing parameters among those investigated [33]. In present study, it is required to increase the response (change in surface roughness and hardness) and thus, “Larger is better” characteristic is required. Formula (3) has been used to calculate SN ratio in conditions where maximum target value of quality characteristic (response) is required [34]:

S/N ratio =

(3)

where is output response and its value should be non negative and non zero. The number of observations and for present study, =3.

is

ANOVA (Analysis of Variance) test results were used to determine significance of input parameters in terms of their F-value at 95% confidence level. 2.4 Measurement Instruments and Techniques To analyze the surface roughness, the average roughness (Ra) is considered and measured perpendicular to lay direction. The measurements have been taken using Mitutoyo Surface Roughness Tester “SJ-210” having stylus tip radius 2μm and tip angle 60°C with measuring force of 0.75mN. The measurements have been recorded employing Gaussian filter, cut-off length 0.25 mm and 2.5 mm exploratory length as per ISO 4287 regulations 7

[35]. The digital Durometer having least count 0.5 HD has been used with 0.1mm indenter tip, 30°angle and 2.5 mm indenter length as per ASTM D 2240 [36] standards to measure shore D scale hardness of plastic ABS replicas. The D scale hardness is used to measure hardness of thermoplastic elastomers, vulcanized rubber, cellular materials and plastics. The value of shore D hardness is unitless and ranges from 0 to 100. The surface roughness and hardness have been measured before and after vapour smoothing operations. The experiments are executed in random order so as to minimize the effect of increased temperature of vapours and heaters as well as fall in temperature of cooling coils with an increase in time. The run order for vapour treatment of different replicas has been shown in table 2 along with design matrix. 3. Results and Discussions 3.1 Data Analysis: The Taguchi L18 Orthogonal array with SN ratios of initial and final measurements of surface roughness and hardness has been displayed in Table 2. There has been appreciable improvement of surface finish with vapour smoothing while the process furnishes positive impact on surface hardness of ABS replicas. The visual inspection ruled out the possibility of overheating of parts while intricate details of design are retained even after VS. There is significant difference between initial surface roughness and hardness of FDM replicas fabricated at different orientation angle. The different densities obviously affected initial surface hardness of FDM parts but have negligible impact on initial surface roughness. Table 2 Experimental control log and initial and final measurements

Fig. 4. Main effects for SN ratios of percentage change in (a) Surface roughness (b) Surface hardness The Fig. 4 plots the factor effects on SN ratio against various levels of each input parameter for change in surface roughness and hardness. It is evident that as the parameters D and F are increased from levels 1 to 3, there is noticeable change in SN ratios while for parameters A, B, C and E the change is insignificant for surface roughness. But for surface hardness, parameters A and E highly increased the SN ratios and there was least effect of B, C, D and F. Table 3 Response table for SN ratios for change in surface roughness and hardness

The mean values of %∆Ra and %∆HD at different levels have been calculated using formula (1) as shown in Table 3. The delta value has been tabulated for each parameter from the difference between maximum and minimum values of response at different levels. The 8

rank of individual parameter has been determined from the delta values of all the parameters. The rank signifies the relative importance of input parameters on the response i.e. surface roughness and hardness. Table 4 ANOVA results for percentage change in surface roughness

ANOVA (Analysis of Variance) is a tool numerously adopted by experimenters as it covers the shortcomings of graphical assessment. The ANOVA results for percentage change in surface roughness are shown in Table 4. For surface roughness, the smoothing time (D) and number of cycles (F) are both significant as their F-values are more than table values at 95% confidence levels while other parameters are insignificant. The parameters having negligible contribution (A, B, C, and E) for response are pooled into error and hence degrees of freedom of pooled error are increased. The interaction effect between parameters C and D does not significantly contribute and hence treated as error. Table 5 ANOVA results for percentage change in surface hardness

Similarly, surface hardness is significantly improved by orientation angle (A) and postcooling time (E) at 95% confidence level. The other parameters (B, C, D and F) and interaction effects (CxD) are pooled into error having relatively least contribution in enhancing the surface hardness (Table 5). The ANOVA calculations indicate that the predictors explained 97.56% and 96.51% of the variance in %∆Ra and %∆HD respectively. The R2 (Adjusted) is 79.29% and 70.30% which accounts for fair number of predictors in the model. As depicted by these values, the model fits the data well and ANOVA results are in accordance with Taguchi Analysis for SN ratios.

Fig. 5. Normal Probability plots of residuals of percentage change in (a) Surface roughness (b) Surface hardness. As seen from the normal probability plots of residuals in Fig. 5, the points generally form straight line for both output parameters i.e. surface roughness and surface roughness indicating residuals are normally distributed. The normality assumption can be declared invalid if the points departed from straight line. Thus, normal distributed data indicates efficacy of Taguchi technique to create robust design with minimum errors even performing less number of experiments as found by previous researchers [33]. 3.2 Effect of Process Parameters on Surface Roughness The most significant input parameters for improvement in surface roughness are smoothing time and number of cycles being directly proportional to change in surface 9

roughness. Although different orientation angles resulted in different surface roughness profiles but it only affects the initial surface roughness. The replicas produced at 90° orientation angle have higher initial average roughness values as compared to 0° but former took lesser manufacturing time with least usage of support material as reported by Ali et al. [37]. The conception that 0° and 90° orientation angles yield best finish is case specific and purely based on part geometry. Sometimes, deviation of orientation angle from optimized value may activate systematic error as experienced by Ippolito et al. [38]. This phenomenon is experienced in complex designs having varying thickness and sloping surfaces as hip implant used in present study. The postprocessing with vapour smoothing affects all the surface profiles uniformly and does not relate with orientation angle. The precooling time does not affect significantly and thus recommended to be done for less than 10 minutes as it only prepares the part surface for smoothing process. The precooling and smoothing of parts does not interact with each other as temperature of hot vapours (60°C) is considerably lower than glass transition temperature of ABS material (105°C). The postcooling enhances the settlement and freezing of partially melted upper surface after vapour smoothing but it has no significant effect on surface roughness. The ABS replicas are mostly influenced by smoothing time which has highest contribution (52.10%) to the response. The hot vapours infiltrate the upper surface and upper layers undergo partial melting and swelling for small durations which are later cooled down in postcooling process. The layers ABS thermoplastic layers involve localized viscous mass transport and settles down as smooth surface when cooled. Similar behaviour for polyamide films exposed to dimethyl sulfoxide vapours have been experienced by previous researchers [39]. As the smoothing time is increased, the vapour exposure is increased which resulted in lower surface roughness values. Fig. 6. Effect of number of cycles on profile height (a) before smoothing (b) after first cycle (c) after second cycle (d) after third cycle.

It was observed during pilot experiments that smoothing time beyond 30 seconds deteriorates the part surface. Thus to increase the efficiency of process, the whole precoolingsmoothing-postcooling cycle is repeated for three times which finally gives very smooth surface as compared to initial values. As explained by geometrical representation in Fig. 6 of surface roughness profile, there is considerable difference between mean line and highest peak (H0) which decreases as next cycles are repeated. The material undergoes meltdown and get resettled in valleys uniformly under the effect of surface tension forces resulting decrease in height as following the relation as H0 > H1>H2>H3 Where, H1, H2 and H3 are difference between mean line and maximum height after first, second and third cycle respectively. The same phenomenon has been verified by Fig. 7 which shows the actual surface roughness profiles for replicas (Replica 3 and 9) manufactured at orientation angle 0°. The 10

replica 3 is smoothed for 20 seconds three times and gives average roughness (R a= 0.2154 µm) with 92.98% improvement while replica 9 is treated only two times for 10 sec which gives comparatively higher profile heights (Ra= 0.9021) with 76.38% improvement in surface finish.

Fig. 7. Surface roughness profiles of replicas before and after smoothing. The replicas produced at 90° orientation angle have higher initial average roughness (Ra=8.9575 µm) values as shown in Fig. 8 but it took lesser manufacturing time with least wastage of support material. Similar high peak surface profiles have been experienced by Krolczyk [40] in parts made by FDM technology. The impact of vapour smoothing is same for both angles as percentage change in surface roughness is considered as response. Replica 16 is treated for 20 seconds for three times and gives lower profile height (Ra=0.5411) as compared to replica 12 (Ra=2.2689) which is only once treated for 15 seconds. The geometry of initial surface roughness profiles of ABS replica depicted that more uniform peaks and valleys are achieved in parts manufactured at 90° angle as compared to 0° angle. The replicas manufactured in horizontal direction (0° orientation angle) yielded lower average roughness values but profile geometry is aberrant (Fig. 7a and 7c) because extruder head has to be raised at regular intervals due to slope in the stem length of replica. At vertical layer deposition (90° orientation angle), the extruder is raised after complete deposition of layers which resulted in consistent geometry as shown in Fig. 8a and 8c. The distance between consecutive valleys (road width) is reduced in ABS parts after smoothing as compared to before smoothing. After smoothing, the dimensions of road width and profile height are significantly reduced as considerable volume of material flows down from peaks to valleys.

Fig. 8. Surface roughness profiles of replicas before and after smoothing.

3.3 Effect of Process Parameters on Surface Hardness The different densities of parts depend upon deposition of internal rasters beneath the surface. The rasters are most densely arranged in “Solid” density, lesser “High” density while least dense in “Low”. It took lesser time to produce parts with low density while maximum time for solid parts. There is slight variation in initial surface hardness for different densities but vapour smoothing process is not significantly controlled by different densities as the response is percentage change in hardness. The vapours enter the upper portion of the surface and do not enter deep inside surface and thus surface hardness of parts after smoothing is not significantly affected by different part densities. The surface hardness is slightly increased by vapour smoothing process as resettled layers efficiently restrict the indenter to pierce into the surface. The indenter enters more easily during initial uneven surface layers before 11

smoothing. This is evident in lower initial hardness values of replicas with 90° orientation angle (higher Ra) as compared those fabricated at 0° (lower Ra). The role of postcooling is highlighted during hardness analysis as higher postcooling efficiently freezes the surface after partial meltdown and thus increases the hardness. The orientation angle 90° also significantly increases the response as initial hardness is lesser in this case which increased the percentage change in surface hardness. 3.4 Confirmation of Optimized Parameters The effect of process parameters on response as discussed in previous sections is based on outcome of statistical analysis of quantitative data. It is required to compare the results of statistical equations with experimental values at optimum conditions to endorse the FDM-VS for commercial production of hip implants. The confirmatory experiments have been performed to validate the optimum parameter levels and finally two replicas have been manufactured to i.e. one each for optimized surface roughness and surface hardness respectively. The optimized parameters are A2B2C3D3E1F3 for maximum change in surface roughness and A2B2C3D2E3F2 for maximum change in surface hardness. The similar procedure of repetitions for measurements has been adopted and experimental values are compared with theoretically predicted values. The respective values of SN ratios for surface roughness and hardness under the optimum conditions can be calculated using equations (4) and (6) Optimum value of SN ratios (ηopt) for surface roughness [41] is given as:

ηopt = M%∆Ra + (MA2- M%∆Ra) + (MB2- M%∆Ra) + (MC3- M%∆Ra) + (MD3- M%∆Ra) + (ME1M%∆Ra) + (MF3- M%∆Ra)

(4)

Here, M%∆Ra= Mean of SN ratios of %∆Ra = 38.35944 (Mean of column no. 12 in Table 2) Put optimum values (MA2, MB2, MC3, MD3, ME1, MF3) from Table 3 for surface roughness in Equation (4). This gives:

ηopt = 38.35944 + (38.39-38.35944) + (38.47-38.35944) + (38.45-38.35944) + (38.0938.35944) + (38.42-38.35944) + (38.96-38.35944)

ηopt = 39.9828 which is optimum value SN ratio. The corresponding values of percentage change can be calculated using formula: yopt2 = 1/10–ηopt/10

(5)

Formula (5) is derived from formula (3) where yopt is optimum response (percentage change). yopt = 99.80% which is optimum value of percentage change in surface roughness achieved at optimized parameter settings A2B2C3D3E1F3.

12

Similarly, Optimum value of SN ratios (ηopt) for surface hardness [42] is given as:

ηopt = M%∆HD + (MA2- M%∆HD) + (MB2- M%∆HD) + (MC3- M%∆HD) + (MD2- M%∆HD) + (ME3M%∆HD) + (MF2- M%∆HD)

(6)

Here, M%∆HD= Mean of SN ratios of %∆HD = 18.35663 (Mean of column no. 16 in Table 2) Put optimum values (MA2, MB2, MC3, MD2, ME3, MF2) from Table 3 for surface hardness in Equation (6). This gives:

ηopt = 18.35663 + (19.08-18.35663) + (18.55-18.35663) + (18.57-18.35663) + (18.6418.35663) + (19.46-18.35663) + (18.60-18.35663)

ηopt = 21.11685 which is optimum value SN ratio. Using formula (4), yopt = 11.37% which is optimum value of percentage change in surface hardness achieved at optimized parameter settings A2B2C3D2E3F2. The theoretical calculations predicted 99.80% and 11.37% increase in percentage change in surface roughness and hardness respectively. On the other hand, the replicas actually manufactured at optimized pre and postprocessing parameters recorded 99.62% decrease (8.803 µm to 0.033µm = 33nm) in surface roughness while 11.32% increase (53 to 59) in surface hardness. The negligible error (0.43%) in predicted and experimental values has been noted which validate the statistical equations and consistency of combined FDM-VS process. The surface roughness of 33 nm and hardness value of 59 is acceptable as per requirement of manufacturing sound investment castings [42]. 3.5 SEM Micrograph Analyses The SEM micrographs have been acquired to study the phenomenon of vapour smoothing process performed at optimum parameters as discussed in previous section. Fig. 9 and 10 shows the SEM micrograph of ABS replicas showing top and transverse view of part surface respectively before and after vapour smoothing process. The SEM images have potential to measure even minute changes in dimensions at micro scale [43] and thus, it was preferred instead of optical microscope. The peaks and valleys are visible and distinct lines can be seen where two roads join to make valleys while roads are not visible after vapour smoothing process. The semi circles can be clearly seen as side view of upper surface (Fig. 9) before smoothing while after vapour smoothing, the peaks and valleys cannot be discriminated and ultra smooth surface is achieved.

Fig. 9. SEM images (top view) of ABS replica (a) before smoothing (b) after smoothing.

The ABS plastic is heated at 60°C being lower than its glass transition temperature (105°C) which tend to temporarily melt only the uppermost layers comprising peaks and 13

valleys. The viscosity of upper layers is lowered and plastic material starts flowing as parts are hanged in smoothing chamber. During this phase, the surface tension forces triggers the smoothing action as the peaks are lowered to cover the minimum surface area. The plastic material is transferred from peaks to valleys and once the layers are resettled, the part are cooled which freeze the molten plastic. The phenomenon can be seen in Fig. 9 where the road width reduces from 389 µm to 206 µm after vapour smoothing cycles as the material deposited in valleys flats the surface. Similar phenomenon has been discussed in section 3.2 while analysing the surface roughness profiles in Fig. 7 and 8. The parts manufactured by FDM showed traces of material overlapping due to lower air gaps. The overlapping of material is absent after smoothing and thus, much cleaner and even surface can be viewed. Initially, the surface looks as semi-circular profile from side-view while after performing vapour smoothing; the flat surface appears (Fig. 10). The traces of resettled plastic layers are visible in Fig. 10b which freeze immediately as parts are hanged in postcooling chamber. The initial height of semi-circular profile was 35.39 µm which is reduced comparatively much smoother surface whose peaks and valleys cannot be discriminated by SEM images. The vapour smoothing process yields a shiny and glossy surface which demonstrates the reflective nature of surface.

Fig. 10 SEM images (transverse view) of ABS replicas (a) before smoothing (b) after smoothing.





 



4. Conclusions The non-traditional surface finishing technique has been presented to develop hip implant replicas via combination of FDM and VS processes. The results of study suggest that the small smoothing durations (30 sec) and repeated cycles (precooling-smoothing-postcooling) can yield excellent surface finish. Further the surface finish up to nano level has been achieved at optimized parameters resolving major hurdle of surface roughness of FDM parts. The interaction among the process parameters was also checked and was found insignificant. It is recommended to avoid over heating of parts by increasing exposure duration which may deteriorate surface texture and dimensional accuracy. The outcomes of study can be useful for production of patient specific implants via FDMVC-IC route which can reduce production time and cost. The investment castings prepared can be used for in-vitro study to validate practical applicability of the process. Moreover in-vivo studies can be performed on animals to replace damaged bones with low cost implants made through FDM-VS-IC. This may lead to the commercialization of FDM for investment casting applications especially in development of biomedical implants. 5. Acknowledgments

14

The experimental work has been done at Manufacturing Research Lab at Guru Nanak Dev Engineering College, Ludhiana and funded by AICTE, New Delhi. The authors are highly thankful to I.K.G. Punjab Technical University, Kapurthala for providing a platform to carry out research. 6. Nomenclature ABS

Acrylonitrile Butadiene Styrene

AM

Additive Manufacturing

CAD

Computer Aided Designing

CPT

Collarless Polished Tapered

CT

Computerized Tomography

FDM

Fused Deposition Modeling

HD

Hardness

IC

Investment Casting

MRI

Magnetic Resonance Imaging

Ra

Average Surface Roughness

SEM Scanning Electron Microscope STL

Standard Tessellation Language

VS

Vapour Smoothing

%∆Ra Percentage Change in Surface Roughness %∆HD Percentage Change in Surface Hardness mN

Millinewton

nm

Nanometer

µm

Microns

7. References 1. A. Boschetto, L. Bottini, Roughness prediction in coupled operations of fused deposition modeling and barrel finishing, J. Mater. Process. Technol. 219 (2015) 181-192. doi:10.1016/j.jmatprotec.2014.12.021 2. A. Peng, X. Xiao, R. Yue, Process parameter optimization for fused deposition modeling using response surface methodology combined with fuzzy inference system, Int. J. Adv. Manuf. Tech. 73 (2014) 87-100. doi:10.1007/s00170-014-5796-5. 3. A. Boschetto, V. Giordano, F. Veniali, Surface roughness prediction in fused deposition modeling by neural networks, Int. J. Adv. Manuf. Tech. 67 (2013) 2727-2742. doi:10.1007/s00170-012-4687-x. 4. P. Dudek, FDM 3D Printing Technology in Manufacturing Composite Elements, Archives of Metallurgy and Materials 58 (4) (2013) 1415-1418. doi:10.2478/amm-20130186. 15

5. D.K. Pal, B. Ravi, Rapid tooling route selection and evaluation for sand and investment casting, Virtual and Physical Prototyping 2 (4) (2007) 197-207. 6. E. Bassoli, A. Gatto, L. Iuliano, M.G. Violante, 3D printing technique applied to rapid casting, Rapid Prototyp. J. 13 (2007) 148-155. doi:10.1108/13552540710750898. 7. S. Pattnaik, P.K. Jha, D.B. Karunakar, A Review of Rapid Prototyping Integrated Investment Casting Processes, Journal of Materials: Design and Applications. (2013) 129. doi: 10.1177/1464420713479257. 8. R. Singh, S. Singh, V. Mahajan, Investigations for Dimensional Accuracy of Investment Casting Process after Cycle Time Reduction by Advancements in Shell Moulding, Procedia Mater. Sci. 6 (2014) 859-865. doi:10.1016/j.mspro.2014.07.103. 9. M. Macků, M. Horáček, Applying RP-FDM Technology to Produce Prototype Castings Using the Investment Casting Method, Archives of Foundry Engineering 12 (3) (2012) 75-82. doi:10.2478/v10266-012-0085-y. 10. P. Jain, A.M. Kuthe, Feasibility Study of Manufacturing using Rapid Prototyping: FDM Approach, Procedia Engineering 63 (2013) 4-11. 11. A. Hanus, N. Špirutová, J. Beňo, Surface quality of foundry pattern manufactured by FDM method - rapid prototyping, Archives of Foundry Engineering 11 (1) (2011) 1520. 12. B. Vasudevarao, D.P. Natarajan, M. Henderson, Sensitivity of RP surface finish to process parameter variation, in: Proc. of 11th Solid Freeform Fabrication Symposium, Austin, August 8-10, 2000, pp. 252-58. 13. M. S. Hossain, D. Espalin, et al., Improved Mechanical Properties of Fused Deposition Modeling-Manufactured Parts Through Build Parameter Modifications, Journal of Manufacturing Science and Engineering 136 (2014) 1-12. doi:10.1115/1.4028538. 14. P. Sreedhar, C.M. Manikandan, G. Jothi, Experimental Investigation of Surface Roughness for Fused Deposition Modelled Part with Different Angular Orientation, Int J Adv Design Manuf Technol. 5 (3) (2012) 21-28. 15. I. Gajdoš, J. Slota, influence of printing conditions on structure in FDM prototypes, Tech. Gaz. 20 (2) (2013) 231-236. doi: 658.512.2:004.896. 16. T. Galeta, G. Šimunović, M. Mazurek, Impact of strengthening fluids on roughness of 3d printed models, Metalurgy. 54 (1) (2015) 231-234. doi: hrcak.srce.hr/126741 17. P. Kulkarni, D. Dutta, On the Integration of Layered Manufacturing and Material Removal Process, International Journal of Machining Science and Engineering 122 (2000) 100-108. 18. P.M. Pandey, N.V. Reddy, S.G. Dhande, Virtual hybrid-FDM system to enhance surface finish, Virtual and Physical Prototyping 1 (2) (2006) 101-116. doi:10.1080/17452750600763905. 19. M. Fischer, V. Schoppner, Finishing of ABS-M30 Parts Manufactured with Fused Deposition Modeling With Focus on Dimensional Accuracy, in: Proc. of 25th Solid Freeform Fabrication Symposium, Austin, August 4-6, 2014, pp. 923-934. 20. L.M. Galantucci, F. Lavecchia, G. Percoco, Experimental study aiming to enhance the surface finish of fused deposition modeled parts, CIRP Annals - Manufacturing Technology 58 (2009) 189-192. doi:10.1016/j.cirp.2009.03.071. 21. L.M. Galantucci, F. Lavecchia, G. Percoco, Quantitative Analysis of a Chemical 16

Treatment to Reduce Roughness of Parts Fabricated Using Fused Deposition Modeling, CIRP Annals Manufacturing Technology 59 (2010) 247-250 doi:10.1016/j.cirp.2010.03.074. 22. C. Kuo, R. Mao, Development of a Precision Surface Polishing System for Parts Fabricated by Fused Deposition Modeling, Mater. Manuf. Process. (2015). doi:10.1080/10426914.2015.1090594. 23. G. Percoco, F. Lavecchia, L.M. Galantucci, Compressive Properties of FDM Rapid Prototypes Treated with a Low Cost Chemical Finishing, Research Journal of Applied Sciences, Engineering and Technology 4 (19) (2012) 3838-3842. 24. A. Garg, A. Bhattacharya, A. Batish, On Surface Finish and Dimensional Accuracy of FDM Parts after Cold Vapor Treatment, Mater. Manuf. Process. 31 (4) (2016) 522-529. doi:10.1080/10426914.2015.1070425. 25. R.L. Zinniel, Vapor smoothing surface finishing system, Patent No. US 8075300 B2. (2011) Downloaded from http://www.google.com/patents/US8075300 on April 13, 2015. 26. D. Espalin, F. Medina, et al., Effects of Vapour Smoothing on ABS Part Dimensions, in: Proc. of Rapid 2009 Conference & Exposition, Schaumburg, May 12-14, 2009, pp. 1-17. 27. P. Kulkarni, D. Dutta, Deposition Strategies and Resulting Part Stiffnesses in Fused Deposition Modeling, Journal of Manufacturing Science and Engineering 121 (1) (1999) 93-103. 28. ISO 10360-2 (2009), Geometrical product specifications (GPS) - Acceptance and reverification tests for coordinate measuring machines (CMM) - Part 2: CMMs used for measuring linear dimensions, 2009. 29. D. Putzer, A. Pizzini, M. Liebensteiner, J. Luis, M. Nogler, Recognizing the surgical situs in minimally invasive hip arthroplasty : a comparison of different filtering techniques, Biocybern. Biomed. Eng. 36 (2015) 182-192. doi:10.1016/j.bbe.2015.11.006. 30. Stratasys Inc: Finishing Touch™ Smoothing Station - Service Manual, 2010. 31. Microcare™ Safety Data Sheet 2009, http://www.stratasys.com/~/media/Main/Secure/MSDS/Smoothing_Station_Fluid/MSD S_Smoothing_Station_Fluid.pdf (accessed 10.05.15) 32. V. H, Wilson, Udayakumar, Optimization of diesel engine parameters using Taguchi method and design of evolution, Journal of the Brazilian Society of Mechanical Sciences and Engineering 34(4), (2012), 423-428. 33. D.P. Selvaraj, P. Chandramohan, M. Mohanraj, Optimization of surface roughness , cutting force and tool wear of nitrogen alloyed duplex stainless steel in a dry turning process using Taguchi method, Measurement 49 (2014) 205-215. doi:10.1016/j.measurement.2013.11.037. 34. T.A. El-Taweel, M.H. El-Axir, Analysis and optimization of the ball burnishing process through the Taguchi technique, Int. J. Adv. Manuf. Tech. 41 (2009) 301-310. 35. ISO 4287 (1997), Geometrical Product Specification (GPS) – Surface Texture: Profile Method – Terms, Definition and Surface Texture Parameters, International Organization for Standardization (ISO), Geneva, 1997. 36. ASTM D2240 (2010), Standard Test Method for Rubber Property - Durometer Hardness, American Society for Testing and Materials, 2010 17

37. F. Ali, B.V. Chowdary, J. Maharaj, Influence of Some Process Parameters on Build Time, Material Consumption, and Surface Roughness of FDM Processed Parts: Inferences Based on the Taguchi Design of Experiments, in: Proc. of 4 th IAJC/ISAM Joint International Conference, Orlando, September 25-27, 2014. 38. R. Ippolito, L. Iuliano, A. Gatto, Benchmarking of rapid prototyping techniques in terms of dimensional accuracy and surface finish, CIRP Annals - Manufacturing Technology 44 (1) (1995) 157-160. 39. A. Anthamatten, S.A. Letts, R.C. Cook, Controlling Surface Roughness in VapourDeposited Polyamic acid Films by Solvent-Vapour Exposure, Langmuir 20 (2004) 62886296. 40. G. Krolczyk, P. Raos, S. Legutko, Experimental Analysis of Surface Roughness and Surface Texture of Machined and Fused Deposition Modelled Parts, Tech. Gaz. 21 (1) (2014) 217-221. 41. K. Krishnaiah, P. Shahabudeen, Applied Design of Experiments and Taguchi Methods, PHI Learning Private Limited, New Delhi, 2013. 42. J.E. Sopcak, Handbook of Lost Wax or Investment Casting, Gem Guides Book Company, Baldwin Park, 1986. 43. B. Trifkovic, I. Budak et al., Application of Replica Technique and SEM in Accuracy Measurement of Ceramic Crowns, Measurement Science Review 12 (3) (2012) 90-97.

Figure 1

18

Figure 2

19

Figure 3

Figure 4

20

Figure 5

Figure 6

21

Figure 7

Figure 8

22

Figure 9

Figure 10

23

Table 1

Level Apparatus

Parameters

1

2

3

Orientation Angle (°)

A

0°

90°

-

Density

B

Low

High

Solid

Precooling Time (min.)

C

10

15

20

Smoothing Time (sec)

D

10

15

20

Postcooling Time (min.)

E

10

15

20

Number of Cycles

F

1

2

3

FDM

Vapour Smoothing

Symbol

Table 2

Parameters Replica Run A No. Order 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18

3 7 8 16 2 10 12 5 14 17 13 18 1 15 4 6 9 11

1 1 1 1 1 1 1 1 1 2 2 2 2 2 2 2 2 2

Surface Roughness

B C D E F

1 1 1 2 2 2 3 3 3 1 1 1 2 2 2 3 3 3

1 2 3 1 2 3 1 2 3 1 2 3 1 2 3 1 2 3

1 2 3 1 2 3 2 3 1 3 1 2 2 3 1 3 1 2

1 2 3 2 3 1 1 2 3 3 1 2 3 1 2 2 3 1

1 2 3 2 3 1 3 1 2 2 3 1 1 2 3 3 1 2

Initial Final Ra (µm) Ra (µm) 3.011 3.106 3.068 3.393 3.389 3.495 3.788 3.897 3.819 8.680 8.801 8.957 8.806 8.454 8.482 8.525 8.652 8.582

0.899 0.535 0.215 0.845 0.445 0.452 0.466 0.528 0.902 0.926 1.522 2.268 2.130 0.765 0.938 0.541 2.502 1.414

%∆Ra

Surface Hardness Initial

Final

%∆HD

SN Ratio

71.5 72 70.5 76.5 77 77.5 81.5 82 83 46.5 46.5 46.5 52 52.5 52.5 58.5 58.5 60.5

76 77.5 77 82.5 83.5 82.5 87.5 88 90 51 50 50.5 57.5 57 57.5 63 64 66

6.29 7.63 9.22 7.84 8.44 6.45 7.36 7.32 8.43 9.67 7.53 8.60 10.57 8.57 9.52 7.69 10.25 9.09

15.9730 17.6505 19.2946 17.8863 18.5268 16.1912 17.3376 17.2902 18.5166 19.7085 17.5359 18.6900 20.4815 18.6596 19.5727 17.7185 20.2145 19.1713

SN Ratio 70.14 82.76 92.98 75.08 86.87 87.05 87.69 86.46 76.38 89.33 82.70 74.67 75.81 90.96 88.94 93.65 71.08 83.53

24

36.9193 38.3564 39.3678 37.5105 38.7774 38.7954 38.8590 38.7363 37.6596 39.0199 38.3501 37.4629 37.5945 39.1770 38.9819 39.4302 37.0349 38.4368

Table 3 Level

Response Parameters

%∆Ra A B C D 1 38.33 38.25 38.22 37.74 2 38.39* 38.47* 38.41 38.25 3 38.36 38.45* 39.09* Delta 0.06 0.23 0.23 1.35 Rank 6 4 3 1

E 38.42* 38.41 38.24 0.18 5

%∆HD A B C D E F 17.63 18.14 18.18 18.28 17.48 18.14 19.08* 18.55* 18.31 18.64* 18.13 18.60* 18.37 18.57* 18.14 19.46* 18.33 1.45 0.41 0.39 0.50 1.98 0.46 2 4 5 3 1 6

F 37.76 38.36 38.96* 1.20 2

* Indicating the optimum value

Table 4 Source

Degrees Seq SS of freedom A* 1 0.0143 B* 2 0.1542 C* 2 0.1756 D 2 5.5396 E* 2 0.1237 F 2 4.3478 CxD* 4 0.2262 Error 2 0.2643 Pooled Error 13 0.9583 Total 17 10.8456 *Pooled into error F0.05(2,13)= 3.81

Seq MS

F-Value P-Value

0.01425 0.07710 0.08780 2.76979 0.06184 2.17388 0.05656 0.13214 0.0737

25

0.11 0.58 0.66 20.96 0.47 16.45 0.43

0.774 0.632 0.601 0.046 0.681 0.057 0.787

Percentage Significance contribution 0.13 No 1.42 No 1.61 No 51.07 Yes 1.14 No 40.08 Yes 2.08 No 2.43 8.83 100

Table 5 Source

Degrees Seq SS Seq MS of freedom A 1 9.5131 9.5131 B* 2 0.5096 0.2548 C* 2 0.4700 0.2350 D* 2 0.7961 0.3981 E 2 12.1924 6.0962 F* 2 0.6372 0.3186 CxD* 4 2.3448 0.5862 Error 2 0.9580 0.4790 Pooled Error 14 5.7157 0.4082 Total 17 27.4212 *Pooled into error F0.05 (2,14)= 3.74

F-Value

P-Value

19.86 0.53 0.49 0.83 12.73 0.67 1.22

0.047 0.653 0.671 0.546 0.073 0.601 0.496

26

Percentage Significance contribution 34.69 0.764 1.71 2.90 44.46 2.32 8.55 3.49 20.84 100

Yes No No No Yes No No