外文原文-徐正松2008.pdf

International Journal of Machine Tools & Manufacture 43 (2003) 879–887Part shrinkage anomilies from stereolithography injection mouldtoolingR.A. Harris∗, H.A. Newlyn, R.J.M. Hague, P.M. DickensWolfson School of Mechanical & Manufacturing Engineering, Loughborough University, Leicestershire, LE11 3TU, UKReceived 6 January 2003; accepted 4 March 2003AbstractThe use of stereolithography (SL) tooling allows plastic parts to be produced by injection moulding in a very short time due tothe speed of mould production. One of the supposed advantages of the process is that it provides a low volume of parts that arethe same as parts that would be produced by the conventional hard tooling in a fraction of the time and cost.However, this work demonstrates different rates of polymer shrinkage are developed by parts produced by SL and conventionaltooling methods. These revelations may counter the greatest advantages of the SL injection moulding tooling process as the partsdo not replicate those that would be produced by conventional hard tooling.This work identifies the different shrinkage that occurs in mouldings produced by an SL mould as compared to those producedfrom an aluminium mould. The experiments utilise two very different types of polymers and two mould geometries, which areprocessed in the same manner so that the heat transfer characteristics of the moulds are isolated as the only experimental variable.The work demonstrates how the two mould materials exhibit very different rates of expansion due to the temperature profilesexperienced during moulding. This expansion must be compensated for to establish the total amount of shrinkage incurred bymoulded parts. The compensation is derived by a mathematical approach and by modelling using finite element analysis. Bothtechniques depend upon knowledge of the thermal conditions during moulding. Knowledge of these thermal conditions are obtainedby real-time data acquisition and simulated by FEA modeling. The application of the findings provide knowledge of the completeshrinkage values relating to the mould material and polymer used which would enable the production of geometrically accurate parts.Keywords: Finite element analysis; Plastic injection moulding; Polymer shrinkage; Rapid tooling & stereolithography1. IntroductionStereolithography (SL) has shown itself to be capableof directly producing tooling cavities (inserts) for injec-tion moulding. The accuracy of the SL process resultsin epoxy inserts that require few further operations priorto their use in injection moulding. Thus, the process pro-vides a quick route to tooling that, depending on geo-metric complexity and the moulding polymer, can pro-duce up to approximately 50 parts [1]. Various polymershave been successfully moulded by SL injection mould-ing. These include polyester (PE), polypropylene (PP),polystyrene (PS), polyamide (PA), polycarbonate (PC),polyether-ether-ketone (PEEK), acrylonitrile-styrene-∗Corresponding author. Fax: +44-0-1509-227549.E-mail address: r.a.harris@lboro.ac.uk (R.A. Harris).acrylate (ASA) and acrylonitrile-butadiene-styrene(ABS) [2–5].SL produces an epoxy mould cavity which possessesvery different thermal properties in comparison to metalmoulds. Without the use of mould heating or cooling thedifferent mould materials impose very different rates ofpart cooling due to the inherent heat transfer character-istics of the mould material. Many polymers exhibit dif-ferent shrinkage according to the cooling conditions ofthe part during moulding [6–11]. This work intends toestablish the effects of the thermal conditions caused bythe mould material on part shrinkage.2. AimsThis research aims to evaluate the shrinkage of mould-ings from SL tools in comparison with those from a con-880 R.A. Harris et al. / International Journal of Machine Tools & Manufacture 43 (2003) 879–887ventional metal rapid tooling method for injectionmoulding.3. Methodology3.1. Experimental designThe aim of the experiments was to establish theshrinkage that occurs within 48 h of the moulding of twopolymers of very different characteristics (Polyamide 66[PA66, crystalline] and Acrylonitrile-Butadiene-Styrene[ABS, amorphous]) when produced by injection mould-ing in cavities of differing materials (Stereolithography[SL] and Aluminium [AL]). This would be by a directcomparison of the dimensions of the moulding cavityand the moulded parts.Besides the choice of thermoplastic polymer and toolmaterial the shrinkage of an injection moulded part canalso be dependent upon several process variables i.e. fol-low up pressure, machine type etc. However, the onlyvariables in the two experimental sets (PA66 & ABS)were the tool material type (SL & AL) and tool cavitygeometry (bar & disc—these were used to identify dif-ferent shrinkage according to the polymer flow directionduring moulding and are further explained in section3.2). With respect to the injection moulding to be con-ducted, the only influential variable in each experimentalset were the thermal properties of the SL and AL mould-ing cavities. This dictated the rate at which the heat fromthe injected polymer would be conducted away from themoulded part through the cavity material. The value ofthermal conductivity of the mould materials differ gre-atly: SL = 0.2 W/m/K, AL = 200 W/m/K.3.2. Mould designThe mould design was based upon BS EN ISO 294—1 and 4 [12,13] and ASTM D955 [14] standards forestablishing shrinkage of injection moulded polymers.Specimens of two differing geometries were mouldedin order to provide shrinkage measurements both parallel(bar shape) and perpendicular (disc shape) to the direc-tion of polymer flow. Dimensioned images of thesemould cavities can be viewed in Fig. 1.The draft angle used to ease part removal from themould was 1.5°. This value has previously been shownto be an optimum value for reducing potential damageto SL tools upon part ejection [15].An open gate design was used on both moulds toavoid areas of heat and pressure concentration (such aswith fan gating) that can damage an SL tool. This gatingsystem was employed to ensure no great pressure differ-ences in the mould that may have led to unequal stressesin the part and also prevents an interruption to followup pressure caused by premature freeze-off of the gate.The depth of the gate was the same as the cross-sectionalthickness of the part (3.2 mm in both disc & bar). Thewidth of the gate size is proportional to the cavity sizein each case.The moulds did not include a part ejection system.The inclusion of steel ejector pins would have providedan area in the moulding cavity which would have cooledthe polymer at a different rate than other areas in themould due the different rates of heat transfer in theimmediate areas. Steel is of a much greater thermal con-ductivity than SL and considerably less than AL. Thiswould not have enabled the experiment to assess thecomplete effects of the heat transfer provided by SL orAL moulds on part shrinkage. The absence of an ejectorsystem caused no problems as the moulded parts werevery simple in shape and were easily removed by handfrom the mould.The SL moulds were manufactured by a 3D SystemsSLA350 machine, using Vantico 5190 resin. The buildlayer thickness was 0.05 mm, as this has previously beendemonstrated as an optimal value in extending the work-ing life of SL moulds [16].The inserts were contained within a steel bolster formoulding. These bolsters facilitated alignment on themachine platen, provided material entry into the mouldvia a tapered sprue bush and protected the inserts fromany excessive application of pressure.3.3. Tool temperature recordingIn order to establish the heat transfer characteristicsthat occured in each of the moulds, the temperature wasrecorded throughout the moulding cycle by the insertionof k-type thermocouples. The tip of the thermocoupleswere situated such that they were located 0.5 mm belowthe surface of the moulding face. The calibration of eachindividual thermocouple was checked prior to insertion.After insertion into the mould the thermocouples werecalibrated to ensure that the temperature 0.5 mm belowthe cavity surface was an accurate reflection of the actualcavity surface by comparison of their simultaneousvalues and their response to temperature change. Thedifference between the temperatures measured by thethermocouple and the actual surface temperature wasnever greater than ± 1 °C. A data acquisition programrecorded and logged the temperature profiles over a 10minute period during the moulding of a part.3.4. Injection mouldingIn order to eliminate extra experimental variables, itwas important to find universal parameter values thatwould work with both the polymers, both the mould geo-metries and both mould material types. Other than tocompensate for the different part volumes of the bar &881R.A. Harris et al. / International Journal of Machine Tools & Manufacture 43 (2003) 879–887Fig. 1. CAD images of the mould cavity designs.disc geometries, the parameters were identical in all theexperiments conducted.The PA66 used was Bergamid A70NAT produced byPolyOne. The ABS used was Lustran Ultra 2373 pro-duced by Bayer. Both polymers were hygroscopic andwere dried immediately prior to processing.The injection moulding machine used was a Bat-tenfeld 600/125 CDC model with a Unilog 4000 controlunit. This machine consisted of a 60 tonne hydraulicclamping unit and a 125 × 35 mm reciprocating screwinjection unit with a conventional tapered nozzle.The following process parameters were used:L50539 Melt temperature set at 270 °C in each of the fivebarrel temperature zones.L50539 Injection speed set at 100 mm/sec.L50539 Injection pressure of 150 bar.L50539 Follow-up pressure of 150 bar, held for 3 sec, on a100 mm cushion.L50539 The ambient temperature of the mould prior to injec-tion was 23.5 °C.L50539 A cooling time of 40 sec prior to part removal. Thiswas established as the time at which parts could beremoved without distortion.L50539 A clamping force of 15 tonnes.3.5. Shrinkage measurementBS EN ISO 291 [17] was used as a standard for theenvironment in which the parts were conditioned andmeasured 48 h after moulding. Prior to injection mould-ing, the mould cavity and later the moulded specimenswere measured in the laboratory atmosphere to the near-est 0.01 mm. Twenty parts were moulded from eachexperimental set and measurements made of each. Theparts and cavities were measured across their gaugelength. Since the gate for the bar mould was centrallyplaced at the top of the gauge length, two measurementswere taken either side of the gate for each part and thecavity. Gauge length measurement for both part &mould geometries are illustrated in Fig. 2. The measure-ments taken from the specimens were compared to themeasurements of the cavities and expressed as a percent-age difference of the moulding cavity and the parts ineach experimental set after compensation for thermalexpansion of the mould.4. Results4.1. Mould temperature profilesThe data showed a strong similarity (no greater than± 5% difference) between the temperature profiles forboth polymer types (ABS & PA66) and both cavity geo-metries (bar & disc). Very different temperature profileswere experienced by the SL and AL mould types. TheFig. 2. Measurement positions for disc & bar geometries.882 R.A. Harris et al. / International Journal of Machine Tools & Manufacture 43 (2003) 879–887Fig. 3. Average temperature profiles for AL & SL moulds.similarity between the readings allowed an average plotto be generated that represents the temperature profilefor each tool material type (AL & SL) as shown in Fig.3. The profiles illustrate the vastly different temperatureconditions experienced in the SL and AL moulds. Thetemperature activity in the AL moulds occurred in a veryshort period of time due to the materials high thermalconductivity. The temperature profile in the SL mouldwas more extreme and protracted; without externalassistance (i.e. cooling by compressed air) the SL mouldwould take 15 minutes to return to its ambient tempera-ture.4.2. Initial shrinkage resultsThe initial shrinkage values, calculated from thepart/mould measurements, can be viewed in Table 1.However, these figures need further consideration toestablish the total shrinkage that occurs. This isexplained in the following section.4.3. Compensation for thermal expansion—bycalculationBoth of the mould materials used expand whenheated, albeit in differing amounts. The measurementsTable 1The % size difference of part/mould, including compensation for thermal expansion by calculation & FEAPA66 Part measurement ABS Part measurementMould type AL bar SL bar AL disc SL disc AL bar SL bar AL disc SL disc% part/mould difference H110021.3272 H110022.611 H110021.2177 H110022.4149 H110020.7524 H110020.6474 H110020.7351 H110020.5611% part/mould difference H110021.3434 H110022.7308 H110021.2339 H110022.6101 H110020.7687 H110020.7696 H110020.7514 H110020.7599including compensationfor thermal expansionby calculation% part/mould difference H110021.3841 H110022.7447 H110021.2712 H110022.5649 H110020.8097 H110020.7837 H110020.7887 H110020.7139including compensationfor thermal expansionby FEAtaken for shrinkage must be compensated for by theamount of cavity expansion to establish the true amountof difference between mould and part measurements.Not compensating for thermal expansion could lead toan underestimation of shrinkage that occurs duringmoulding. The value obtained for shrinkage had to becorrected by an amount corresponding to the thermalexpansion of the mould.The expansion of the mould resulted in an expansionof the cavity, not contraction. This was verified by plac-ing the mould inserts in an oven at a nominal tempera-ture of 50 °C for 10 min and then measuring the cavityin the appropriate measurement axis (see Fig. 2). Allmoulds showed an increase in cavity size, demonstratingan expansion of the cavity. This was authenticatedfurther in later work (section 4.4.5).This correction factor H9004SMto the part shrinkage duethe mould expansion, is given by: H9004SM= am(TmH11002Ta)[13] where: am=linear coefficient of expansion of themould material in 10H110026m/m/K Tm=maximum mouldtemperature during injection cycle in °CTa=ambienttemperature of the mould on the machine in °C. Thevalues used were:L50539 amL50539 SL = 59 × 10H110026m/m/KL50539 AL = 23.8 × 10H110026m/m/K [18]Ta:L50539 SL = 23.5 °CL50539 AL = 23.5 °C.An explanation of these values is given in section 3.4.Tm:These values were derived from the thermal historyprofiles of the moulds and the injection parameters. Thetemperature at the end of the injection cycle was usedin the calculations as this was the point at which pressureapplication to the injected polymer ceased. When press-883R.A. Harris et al. / International Journal of Machine Tools & Manufacture 43 (2003) 879–887ure application ended, so did the period at which theshrinkage could be influenced. Any further expansion ofthe mould after this period would be unable to affect thepart size. The derivation of the values is illustrated inFig. 4. The maximum temperature (Tm) at the end of theinjection cycle for each mould type were:L50539 SL disc = 57.46 °CL50539 SL bar = 44.37 °CL50539 AL disc & bar = 30.39 °C.The same maximum temperature was experienced byboth the aluminium moulds. This maximum temperaturewas reached before the end of the injection cycle. Thetemperature in the stereolithography moulds continuedto rise for a long period after completion of the injectioncycle. Different maximum temperatures were experi-enced in the SL disc & bar moulds due to the differenttimes of the injection cycle required by their differentpart volume.Therefore, the calculation for thermal expansion foreach mould type was as follows:SL disc mould = 59 × 10H110026(57.46H1100223.5)= 2.00364 mm/m= 0.200364%SL bar mould= 59 × 10H110026(44.47H1100223.5)= 1.23133 mm/m= 0.123133%AL disc & bar mould= 23.8 × 10-6(30.39H1100223.5)= 0.16422 mm/m= 0.016422%.The extra amount of shrinkage determined by calcu-lating the expansion of the mould was incorporated intothe percentage shrinkage of the measured parts to revealthe compensated total shrinkage. These values are shownin Table 1. However, these calculations were slightlysimplified. An assumption was made that the wholemould was at the temperature determined by the dataacquisition for tool temperatures and as such equalFig. 4. Thermal conditions in the moulds during injection cycle.expansion was experienced throughout. In reality theincrease in temperature was localised around the mould-ing cavity. Further work was required to prove whetherany differing expansion of the mould cavity occurreddue to the temperature distribution and to assess anysuch effects on the compensation values used.4.4. Compensation for thermal expansion—by FiniteElement AnalysisFinite Element Analysis (FEA) was used to model themould expansion due to the localised heating caused bythe hot polymer contained within the cavity. The FEAsoftware package used in this work was Algor. Twoforms of FEA analysis from Algor were used. Firstly,a transient thermal analysis was conducted in order todetermine the temperature distribution in the mould.Then a linear elastic analysis was performed using thetemperature distribution results to determine theresulting expansion of the mould cavity.4.4.1. FEA modelling. Stage 1—creation of modelThe FEA work began with the creation of the requiredmodels and FEA mesh. In order to reduce solution time,one quarter of the full mould was created, due to thequarterly symmetrical nature of each of the specimen’sgauge length. This is shown in Fig. 5.The model was extruded such that the mesh spacingbetween each node was 0.5 mm in the immediate regionbeneath the moulding cavity (~4 mm deep) as these arethe elements of interest when evaluating the cavityexpansion. This was conducted to ensure that a node wasat the equivalent point to the thermocouple in the experi-ments, from which the critical time dependant tempera-tures were derived (see section 3.3). After this the meshspacing for the rest of the model was ~10 mm in orderto achieve a shorter analysis time.4.4.2. FEA modelling. Stage 2—allocation of materialpropertiesThe materials were assumed to be homogeneous andisotropic with constant material properties independentof temperature. The values are listed in Table 2.4.4.3. FEA modelling. Stage 3—transient thermalanalysisTransient thermal analysis refers to a thermal con-dition where temperature is a function of time. Thisanalysis type was relevant to the conditions that occurredwithin the moulds during the experiments. The heat wassupplied by the injected polymer which transferred itsenergy (heat) into the surrounding mould material. Thisenergy was not limitless and the polymer correspondedby reducing in temperature as the heat was transferredinto the lower temperature mould.884 R.A. Harris et al. / International Journal of Machine Tools & Manufacture 43 (2003) 879–887Fig. 5. FEA model section of bar & disc moulds.Table 2Material values used in FEA for epoxy, aluminium & polymeraEpoxy SI units Reference sourcer (density) 1250 kg/m3[22]K (thermal conductivity) 0.19 W/m K [22]Cp(specific heat capacity) 1046.7 J/kg°C [22]E (modulus of elasticity) 2.6×109N/m2[22]n (poissons ratio) 0.35 [22]a (thermal coefficient of expansion) 59×10H110026/°C [18]Aluminiumr 2720 kg/m3[23]K 170 W/moK [23]Cp880 J/kg°C [23]E 68.9×109N/m2[23]n 0.3 [23]a 23×10H110026/°C [23]Polymerr 1145 kg/m3[22]K 0.2962 W/m K [22]Cp1625.7 J/kg°C [22]E 0.26×109N/m2(a)(a)n 0.35 [22]a 59 or 23×10H110026/°C(a) [18] or [23]aThe value of E for the polymer was a nominal low value to reduce restraint. The value of a in the polymer was made the same as the mouldmaterial to prevent interference. The use of such values in both cases was to ensure that the presence of the polymer did not influence the expansionof the cavity in the simulationTo conduct