Die Casting

L. D. Clark M. T. Alonso Rasgado K. Davey S. Hinduja School of Mechanical, Aerospace and Civil Engineering, The University of Manchester, Sackville Street, Manchester M60 1QD, UK

1

Experimental Investigation Into the Thermal Behavior of Copper-Alloyed Dies in Pressure Die Casting The rate of heat extraction during the pressure die casting process is central to both the quality and the cost of finished castings. Recent efforts to reduce the thermal resistance of dies by optimizing the effectiveness of the cooling channels have shown the potential for improvement. Reducing the thermal resistance of the coolant boundary layer means that a significant proportion of the total thermal resistance becomes attributable to the die steel. Further significant reductions in die thermal resistance can be obtained by replacing the steel with copper. This paper investigates the feasibility of using copper dies, reinforced with steel inserts and coated with a thin layer of wear resistant material, which is deposited using the thermal arc spray process. Experimental work relating to the thermal spray process has been undertaken to establish bond strengths and thermal conductivities for various process parameters. Moreover, experimental investigations have been carried out using two copper coated dies, the first of which was a pseudodie block heated by an infrared heater. The second die was tested on a die casting machine and produced zinc alloy castings at a greatly increased production rate when compared to its steel counterpart. The experimental results from the two dies are compared with those predicted by an in-house thermal-cum-stress model based on the boundary element method. Reasonable agreement between the predicted and experimental results is shown and the feasibility of copper-alloyed dies for pressure die casting is established. 关DOI: 10.1115/1.2280586兴

Introduction

The pressure die casting process involves the repeated injection of a molten metal into a reusable die 关1兴. The dies used in pressure die casting not only contain the melt and impose the final shape of the castings, but are also a means of extracting heat effectively. This requires removing heat quickly so that the solidification time of the casting is reduced, thus making the production rate economically viable. Effectiveness is also determined by the uniformity of the cavity surface temperature. Removing heat too rapidly from one part of the casting may result in poor surface finish, cold shuts, and other casting defects. Conversely, too high a temperature may delay solidification resulting in an unnecessarily long cycle time or bursting of the casting due to premature ejection. Thermal considerations are therefore crucial in the design of the die blocks. This requires each component of the thermal resistance path to be considered. The interfacial heat transfer coefficient of the die/casting interface has been investigated experimentally by a number of researchers 关2–5兴 in order to determine its magnitude and transient behavior. Researchers have also examined the die/ coolant interface and found that it contributes significantly to the overall thermal resistance as shown in column four of the table in Fig. 1. The die/coolant interfacial resistance is mainly due to the insulating effect of the thermal boundary layer composed of a thin layer of coolant moving slowly along the cooling channel wall. It has been shown that this resistance can be greatly reduced by effective design of the cooling channels 关6兴. This may include cooling channels that are designed to exploit boiling mechanisms 关7兴, which disturb the boundary layer thereby reducing the thermal resistance where the coolant meets the die. The die cooling chanContributed by the Manufacturing Engineering Division of ASME for publication in the JOURNAL OF MANUFACTURING SCIENCE AND ENGINEERING. Manuscript received September 9, 2003; final manuscript received February 3, 2006. Review conducted by S. R. Schmid.

844 / Vol. 128, NOVEMBER 2006

nels can be optimized to exploit boiling and extract heat efficiently and effectively. If the cooling channels are well designed and the die/coolant interfacial resistance reduced then the thermal resistance due to the die material becomes significant as shown in column six of the table in Fig. 1. The die material is usually hardened steel which can withstand the rigors of the pressure die casting process. The conductivity of copper is commonly 10–15 times greater than that of steel. Using copper instead of steel would theoretically reduce the total resistance by a factor of about four as shown in column eight, Fig. 1. However, the pressure die casting process imposes severe physical conditions on the dies. The most onerous of these are heat checking, clamping loads, and the abrasion forces on the cavity surfaces as the melt is injected into the die. Copper alloys do not have the strength or hardness to withstand the pressure die casting process. The tool steel commonly used has a hardness of about 50 Rockwell C and a tensile yield strength of 1650 MPa compared to 70 Rockwell B and 90– 600 MPa for copper alloys. This paper investigates the feasibility of using copper die blocks which are reinforced with steel where necessary. An experimental copper die was manufactured to run on a hot chamber die casting machine and a copper pseudodie block was used on a die casting simulation rig. The copper blocks of the experimental die were strengthened to withstand the clamping forces by means of steel inserts. These were attached to the copper die blocks using the diffusion bonding process known as hot isostatic pressing 关8兴. The cavity surfaces were sprayed using the thermal arc process 关9兴 to deposit a hard, wear resistant surface coating on the copper which could better withstand the abrasive filling process. Two copper alloys were tested to ascertain how well they bonded to the spray coating and three types of bonding agent were also tested. The conductivity of the postsprayed coating material was determined using 1D assumptions from results obtained using a hot/ cold plate rig.

Copyright © 2006 by ASME

Transactions of the ASME

Downloaded 12 Jun 2008 to 202.63.117.72. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm

Fig. 1 Thermal resistance path of three types of die

The die blocks were modeled using a computer program developed specifically for the die casting process 关10,11兴. The computer simulation uses the boundary element method 共BEM兲 to carry out thermal and stress analyses of the die blocks. Previously this method has been used to model multiblock dies. The coating material is represented as a separate boundary element block which can cause numerical problems due to it being strongly linked thermally to the copper to which it is attached. For this reason a fine-coarse preconditioner has been developed to deal with this problem 关12兴. The experimental die was instrumented with thermocouples and data collected for several sets of operating conditions on a hot chamber die casting machine. The pseudodie block was instrumented with thermocouples and strain gauges and data collected.

2

Numerical models have been established that simulate the behavior of copper-alloyed dies in the pressure die casting process. The thermal properties of the die are assumed to be temperature independent and isotropic. The die temperature field T共x , t兲 is thus governed by the transient heat equation 1 ⳵T ␣ ⳵t

on ⍀d ⫻ ⌸t

=0

冉冕 冊 冕 1 ␶

␶0+␶

Tdt =

␶0

1 ␶

␶0+␶

␶0

1 ⳵T 1 dt = 关T兩␶0+␶ − T兩␶0兴 ␣ ⳵t ␶␣

on ⍀d

共2兲

where ␶0 is an arbitrary time. The boundary conditions for the die are of the Robin type and satisfy ¯q共x兲 =

=

1 ␶ 1 ␶

冕 冕

␶0+␶

q⬙共x,t兲dt

␶0 ␶0+␶

␶0

共4兲

The explicit form of the boundary conditions on the component/ die; component/air; and component/component surfaces together with the relevant assumptions are described by Davey et al. 关10兴. The harmonic boundary value problem constituted by 共2兲 and 共3兲 can be expressed as a boundary integral equation 关13兴 of the form ¯ 共x兲 + 4␲C共x兲T d

冕 冉冊 kd

⌫d

⳵ 1 ¯ Td共x⬘兲d⌫共x⬘兲 = ⳵n r

冕

⌫d

¯qd共x⬘兲 d⌫共x⬘兲 r

h共x,t兲Td共x,t兲dt −

1 ␶

冕

where 4␲C共x兲 is the solid angle at a source point, x, x⬘ is a field point, and r is the Euclidean distance from the source point to the field point. The determination of coolant temperature variation over a length L of the channel is achieved by considering an energy balance equation of the form ˙ c兲−1 ⌬TlL = − 共m

共1兲

where ␣ is the thermal diffusivity of the die and ⌸t is the time domain. Integrating 共1兲 over the casting cycle time ␶, gives the governing equation for the time-averaged temperature field as ⵜ2¯T = ⵜ2

¯qd共x兲 = h共x兲关Tl共x兲 − ¯Td共x兲兴

共5兲

Numerical Model

ⵜ 2T =

cycle, as is the heat transfer coefficient, and so it is assumed that Tl共x , t兲 ⬇ Tl共x兲 and h共x , t兲 ⬇ h共x兲. Thus the steady state boundary condition on the channel can be written as

where ⌫L represents the surface of the section of cooling channel ˙ is the mass flow rate, and c is the coolant’s whose length is L, m specific heat capacity. For irregularly shaped channels, L and 兰⌫Ld⌫ can be identified using a mesh partitioning strategy 关14兴. Traditionally, cooling channels have consisted of drilled holes which can be represented by linear pipe elements which lie on the axis of the channel. The distance along the channel axis is thus found by summing the lengths of the pipe elements. For such channels, the surface area of a length of channel is evaluated by summing the areas of the associated surface elements. The thermal stress model is based on the well known 关15兴 thermoelastic boundary integral formulation given as

␶0+␶

h共x,t兲Tl共x,t兲dt

on ⌫d 共3兲

where ⌫d is the boundary for ⍀d. On the cooling channel surfaces the transient temperature variation is small, so the die temperature Td共x , t兲 ⬇ ¯Td共x兲. Also the bulk water temperature Tl is approximately constant over a casting Journal of Manufacturing Science and Engineering

共6兲

¯qdd⌫

⌫L

4␲Clk共x兲uk共x兲 +

␶0

冕

=

冕

冕

˜plk共x,x⬘兲uk共x⬘兲d⌫共x⬘兲

˜ulk共x,x⬘兲pk共x⬘兲d⌫共x⬘兲 +␥

冕冋

册

⳵T ⳵ r,l T共x⬘兲 − r,l 共x⬘兲 d⌫共x⬘兲 ⳵n ⳵n

where ␥ = ␣共1 + ␯兲 / 关8␲共1 − ␯兲兴, ␯ being Poisons ratio, and ␣ is the coefficient of thermal expansion. Also plk and ulk are traction and NOVEMBER 2006, Vol. 128 / 845

Downloaded 12 Jun 2008 to 202.63.117.72. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm

Table 1 Material properties Copper die

Chemical composition Thermal conductivity at y100° C 共W m−1 K−1兲 Microhardness Liquidus 共°C兲 Tensile yield strength 共MPa兲 Coefficient of thermal expansion at 200° C 共␮m m−1 ° C−1兲

95MXC coating

C18200

Ampcolloy

H13 Steel die and inserts

Wire form

Sprayed form

0.6-1.2Cr, 99.1Cu, 0.1Fe, 0.5 other, 0.05Pb, 0.1 Si 171

2.5 Ni, 0.7 Si, 0.4 Cr, 96.4 Cu

0.4C, 5.25Cr, 90.6Fe, 0.4Mn, 1.35Mo, 0.001S, 1Si, 1V

1.6%Si, 29%Cr, 1.65%Mn, 3.75%Bo, 64%Fe

–

259

28.6

26.4

14–15

68 Rockwell B 1075 295

94 Rockwell B 1116 517

54 Rockwell C 1250 1410

– 1204 –

53 Rockwell C – –

17.6

17.5

11.3

–

–

displacement, respectively, with k summed from 1 to 3 and l = 1 , 2, or 3. In addition ˜plk, ˜ulk are the fundamental solutions for traction and displacement. The necessary surface integrations are carried out using analytical and numerical methods where each is appropriate 关16兴. Previously developed boundary element software has been extended to cater for the inclusion of sprayed and copper-alloyed domains. One of the features of the in-house software is the utilization of iterative methods for efficient computation. The inclusion of copper-alloyed and sprayed domains resulted in convergence problems, which were overcome with the development and employment of a coarse preconditioner 关12兴. Another important feature is the use of meshless die domains, which facilitates cooling channel repositioning. The need to perform cyclic stress analysis meant that domain integrals required evaluation and raised the possible need for domain meshing. This problem was solved by meshing the sprayed layer only, as transient variations in the copper-alloy domains proved to be small. This preserved a mesh free environment for the cooling channels. Existing software for shape optimization suffered from convergence difficulties arising from the need to converge on nonlinear boiling heat transfer coefficients for each cooling channel shape configuration. This problem has been overcome by including heat transfer coefficient determination as part of the optimization process. This means that the boiling heat transfer coefficients are determined on the final configuration only resulting in a very efficient and stable approach 关17兴. It was not necessary to optimize the thickness of the sprayed layer as it was found that the induced thermal stresses arising from the thermal spray process limited the thickness of the layer to around 1 mm. The inclusion of the alloyed steel layer in the steady-state thermal model resulted in convergence problems with the multidomain generalized minimal residual 共GMRES兲 iterative solver. This solver, which involves parameter matrix accelerated GMRES in combination with a multiplicative Schwartz method for nonoverlapping domains, works well for weakly connected systems 共high thermal resistance between domains兲. However, the introduction of the extra domains 共alloyed-steel layers兲 to the original model results in a more strongly connected system with low thermal resistance between domains. In order to overcome this problem a new form of coarse preconditioning has been developed. The coarse preconditioner is obtained from a crude representation of the global system of equations. The new scheme 关18兴 has been shown to enhance the convergence of multidomain systems. Problems with more strongly connected systems are overcome by increasing the number of interfacial equations used by the coarse preconditioner. The transient thermal consists of the BEM being used for the die block and the finite element method for the casting. This hy846 / Vol. 128, NOVEMBER 2006

brid model has been enhanced to make the modeling of the alloyed steel layer possible. The model now is capable of predicting transient temperatures for the dies, alloyed steel layer, and casting throughout the casting cycle, enabling energy transfers and the state of the casting at ejection to be investigated with the aim of optimizing the production rate.

3

Experimental Work

Experimental tests have been carried out on copper dies sprayed with a layer of steel using an existing lab-scale rig which simulates the pressure die casting and a proprietary hot chamber die casting machine under industrial conditions. In order to determine suitable operating parameters, investigations were first carried out to determine the influence of the different process parameters and the thickness of the sprayed layer on the bond strength 共the maximum averaged normal stress兲 of the sprayed coating. 3.1 Experimental Equipment and Materials. An 8850MHU thermal arc spray system supplied by Praxair was used; this system incorporates an ArcJet feature that creates a more concentrated spray stream, with a reduced cone angle of approximately 15 deg and delivering particles in the spray stream with a much higher particle velocity. A purpose built booth and extractor system were used to control noise and emissions. The spray coating chosen was 95MXC. Three bonding agents were used: 75B nickel and 10T and 11T aluminum bronze all supplied by Praxair. Two copper alloys were used: chromium copper C18200 and ampcolloy 940. Brown alumina grade 46 was used for shot blasting. Table 1 lists some of the salient material properties. 3.2 Bond Tests. Test specimens were prepared by first shot blasting a cylindrical copper alloy bar of 25 mm diameter and then spraying with a thin deposit 共⬇0.1 mm兲 of bonding agent. The specimens were then spray coated with a layer of chrome steel 共95MXC兲 to the required thickness. Several specimens were prepared and tested and the results are summarized in Table 2. Apart from tests 1, 2, 3, 6, 11, 12, and 13, the tensile adhesion tests were carried out using 25 mm diameter test specimens in accordance with the ASTM C633-79 standard, as shown in Fig. 2, on a tensile testing machine. The main variables in these tests were coating material and thickness of the sprayed layer. Aluminium oxide was used for grit blasting for all the tests except for the first three when glass bead was used. Instead of creating a sufficiently keyed surface, glass was found to polish the surface. A couple of the spray samples were also machined before being pulled. In tests 17–21 and 25, the Arcjet facility was used to create a more focused spray pattern. In tests 5–10, the sprayed specimens were cooled by the air from the spray gun. In all the tests, the same adhesive was used for preparing the samples. Transactions of the ASME

Downloaded 12 Jun 2008 to 202.63.117.72. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm

Table 2 Bond strength test results Test No.

Material

Geometry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43

Steel Steel Cu Cu Cu CrCu Cu Cu Cu Cu CrCu CrCu CrCu CrCu CrCu CrCu CrCu CrCu CrCu CrCu CrCu CrCu CrCu CrCu CrCu CrCu CrCu CrCu CrCu CrCu Ampco Ampco Ampco Ampco Ampco Ampco Ampco CrCu CrCu CrCu CrCu CrCu Ampco

Flat Flat Flat Round Round Flat Round Round Round Round Flat Flat Flat ’Round Round Round Round Round Round Round Round Round Round Round Round Round Round Round Round Round Round Round Round Round Round Round Round Round Round Round Round Round Round

a

Coating

Coating thickness 共mm兲

Preheat

Grit grade

Machining

ArcJet

Cooling

Failure modela

Bond strength 共MPa兲

Adhesive

75B 75B 75B 75B/95MXC 75B/95MXC 75B/95MXC 75B/95MXC 75B/95MXC 75B/95MXC 75B/95MXC 75B/95MXC 75B/95MXC 75B/95MXC 75B/95MXC 75B/95MXC 75B/95MXC 75B/95MXC 75B/95MXC 75B/95MXC 75B/95MXC 75B/95MXC 75B/95MXC 75B/95MXC 75B/95MXC 75B/95MXC 75B/95MXC 75B/95MXC 75B/95MXC 75B/95MXC 75B/95MXC 10T/95MXC 11T/95MXC 10T/95MXC 11T/95MXC 10T/95MXC 11T/95MXC 10T/75B/95 MXC 10T 11T 11T/75B/95MXC 11T 10T/95MXC 11T/95MXC

0.5 0.5 0.1 0.5 1.0 1.0 1.5 2.0 2.5 3.0 0.5 1.0 1.5 0.5 0.7 0.9 0.5 0.7 0.9 0.9 0.7 0.7 0.7 0.7 0.7 0.7 0.7 0.9 0.7 1.5 0.5 0.5 1.0 1.0 1.5 1.5 1.0 0.2 0.2 1 1 0.5 0.5

N N N N N N N N N N N N N N N N N N Y N Y N N N N N N N N N N N N N N N N N N N N N N

Glass Glass Glass Al2O3 Al2O3 Al2O3 Al2O3 Al2O3 Al2O3 Al2O3 Al2O3 Al2O3 Al2O3 Al2O3 Al2O3 Al2O3 Al2O3 Al2O3 Al2O3 Al2O3 Al2O3 Al2O3 Al2O3 Al2O3 Al2O3 Al2O3 Al2O3 Al2O3 Al2O3 Al2O3 Al2O3 Al2O3 Al2O3 Al2O3 Al2O3 Al2O3 Al2O3 Al2O3 Al2O3 Al2O3 Al2O3 Al2O3 Al2O3

None None None None None None None None None None None None None None None None None None None None None None None None None None None Ground Ground None None None None None None None None None None None None None None

N N N N N N N N N N N N N N N N Y Y Y Y Y N N N N N N Y N N N N N N N N N N N N N N N

Natural Natural Natural Natural Air jet Air jet Air jet Air jet Air jet Air jet Natural Natural Natural Natural Natural Natural Natural Natural Natural Natural Natural Natural Natural Natural Natural Natural Natural Natural Natural Natural Natural Natural Natural Natural Natural Natural Natural Natural Natural Natural Natural Natural Natural

G I M I I I – – – – I I I I I I I I – I – I I I I I I I I I M M I I I I I C C I I M M

– 12.44 3.80 8.01 6.03 27.51 Lifted Lifted Lifted Lifted 34.7 26.2 19.1 33.63 22.00 13.22 13.14 11.00 Lifted 5.25 Lifted 19.5 18.6 18.1 19.1 19.0 18.7 4.77 12.72 3.33 16.29 18.77 13.58 8.29 3.61 3.96 2.32 50.87 46.67 16.18 5.08 38.81 37.21

Araldite 3M-2214 3M-2214 3M-2214 3M-2214 3M-2214 3M-2214 3M-2214 3M-2214 3M-2214 3M-2214 3M-2214 3M-2214 3M-2214 3M-2214 3M-2214 3M-2214 3M-2214 3M-2214 3M-2214 3M-2214 3M-2214 3M-2214 3M-2214 3M-2214 3M-2214 3M-2214 3M-2214 3M-2214 3M-2214 3M-2214 3M-2214 3M-2214 3M-2214 3M-2214 3M-2214 3M-2214 3M-2214 3M-2214 3M-2214 3M-2214 3M-2214 3M-2214

G⫽adhesive failure; I ⫽coating/substrata interface failure, C ⫽coating/coating failure, and M = I and C.

3.3 Thermal Conductivity Tests. The 95 MXC chrome steel material was supplied in wire form and a value for the thermal conductivity was quoted by the manufacturer. However, the thermal conductivity of the sprayed material was required and this has been determined experimentally. For this, two cylindrical samples of the 95MXC material were manufactured by repeatedly spraying into a blanked off section of a steel pipe which was greased on its internal wall to stop the spray from adhering. The samples were 25 mm in diameter and 25 mm long. These were drilled to accept thermocouples and then attached to the rig as shown in Fig. 3. The two samples were placed in a box insulated with rockwool and two small thermostatically controlled electric heaters were placed between them. Cold plates were clamped to the samples and to a heat sink, which was air cooled. When the heaters reached a preset temperature of 100° C, the rig was allowed to run until readings at the thermocouples were reasonably stable indicating a steady heat flow through the two samples. Temperature data were then colJournal of Manufacturing Science and Engineering

Fig. 2 Tensile adhesion test

NOVEMBER 2006, Vol. 128 / 847

Downloaded 12 Jun 2008 to 202.63.117.72. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm

Fig. 3 Thermal conductivity rig „shown with top insulation removed…

lected for about 3 min. The thermal conductivity of the sprayed 95MXC material was then derived from the temperature measurements using Fourier’s equation. 3.4 Casting Simulation Rig. An experimental rig was designed and manufactured to simulate the thermal conditions of the pressure die casting process. The main components of the rig are: a pseudodie block, a heater, an air compressor 共with drier and filter兲, a combined water temperature controller and pump, a pneumatically controlled shutter, and a thermal shield. The layout of the rig is shown in Fig. 4. A pseudodie block was manufactured from C18200 copper and

Fig. 4 Casting simulation rig

848 / Vol. 128, NOVEMBER 2006

sprayed with 95MXC on a thin coating of bonding agent. The rig was designed to reproduce the heat transfer mechanisms involved in pressure die casting but avoiding at the same time the hazardous features of the actual process. The rig has two modes of operation. In the noncyclic mode, the heat is applied to the die cavity at a constant rate and this simulates the steady-state condition. In the cyclic mode, heat applied to the cavity is applied for part of the cycle time and for the other part, a reciprocating thermal shield is placed between the infrared heater and the die block. This simulates the transient behavior of the pressure die casting process. Thermocouples were inserted at strategic points in the die block and the results compared with numerical predictions. The pseudodie block contained a single cooling channel, as shown in Fig. 5. The die blocks, made of chromium copper C18200, consisted of three main parts: a die block with the cavity and cooling channel machined in and with two glass windows to observe if any boiling takes place in the coolant; the backplate containing threaded apertures for cooling channel hose connectors and a rectangular aperture for an optional glass observation window; and a retaining plate to hold the three observation windows in place. Figure 5 also shows the assembled and sprayed pseudodie block prior to final grinding. An infrared heater was used to heat the die cavity. The heater comprised six quartz infrared heating bulbs set in front of a parabolic reflector. At the rated voltage of 144 V, the net heat flux at the heater window was 0.38 W / mm2. The heater itself was cooled by water at a flow rate of 180 l / h and dry, refrigerated, filtered air from the compressor at 44,200 l / h. Cooling water was supplied to the die block by an integrated temperature control unit containing a pump and heat exchanger. The desired inlet water temperature could be preset to within ±2 ° C. A water-cooled stainless steel thermal shield was inter-

Fig. 5 Pseudodie block

Transactions of the ASME

Downloaded 12 Jun 2008 to 202.63.117.72. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm

measure the inlet and outlet water temperatures. Portions of the six thermocouples exposed to the heater were insulated with thermal cement. The cooling circuit included two pressure regulators close to die inlet and outlet and a combined flow meter/regulator near the die inlet. The experimental rig was operated in steady-state and transient modes. In the steady-state mode, the shutter was redundant and was left open, while in transient mode an adjustable pneumatically controlled time delay valve was set to operate the shutter at the desired cycle time. Figure 4 shows the complete rig as well as its schematic layout. The rest of the experimental procedure was applied to both operating modes as follows: 共i兲 共ii兲 共iii兲 共iv兲 共v兲 共vi兲 共vii兲 Fig. 6 Copper die block assembly

posed between the die block and heater so that only the cavity was exposed to the radiant heat. A pneumatically controlled shutter was used to periodically shield the cavity from the heater thus inducing transient temperature fields in the block when required. Six sheathed 0.5 mm iron-copper/nickel thermocouples were used to measure the die temperatures at various distances between the cavity and cooling channel; two thermocouples were used to

Fig. 7

Journal of Manufacturing Science and Engineering

Water was circulated through the test die at ambient temperature. The water was then heated to the required inlet temperature. The inlet flow rate and pressure were adjusted in tandem. The cooling water and air were switched on and circulated through the infrared heater and the thermal shield. The infrared heater was then switched on and adjusted up to a pre-set power output level, identical for all runs. Data were recorded once readings had stabilized. The nonoccurrence of boiling was confirmed by visual inspection through the die block windows.

3.5 Die Casting Trials. To further verify the model, a copperalloy 共C18200兲 test die was designed and manufactured comprising two die blocks and a stepped cavity, which included core pin holes, a gate, and a runner. This die was used on a production hot chamber die casting machine. Steel inserts were incorporated in the die to provide structural support and provision for ejection pins and cores as shown in Figs. 6 and 7共a兲. The die was spray coated with alloyed steel on the cavity and interface surfaces and subsequently machined 共Figs. 7共b兲 and 7共c兲兲. Figure 7共d兲 shows

Copper die block

NOVEMBER 2006, Vol. 128 / 849

Downloaded 12 Jun 2008 to 202.63.117.72. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm

Fig. 8 Thermocouple arrangement

the cavity after it was sprayed; the figure shows cracks whose positions correspond to those of the interface between steel inserts and copper, suggesting they are caused by differential cooling rates. As many as 21 thermocouples were placed in this die and their position is illustrated in Fig. 8. Prior to spraying, seven blind holes of 2 mm diameter were drilled from the backface of the right die block, such that they terminated at 2.5 mm from the front face of the die. Next seven holes of 1 mm diameter were drilled through the cavity face of the die block such that they were concentric with the 2.5 mm holes. These holes were temporarily filled with glass bead masking tape and the die block was sprayed. The sprayed coat was then machined before a final seven holes of 0.4 mm diameter were laser cut through the coating material to meet the holes in the copper block. Seven J type 共iron-constantan兲 thermocouples, with a sheath diameter of 0.25 mm, were fed through the holes from the back of the die block until they protruded from the front face. The protruding stems of the thermocouples were coated with epoxy resin, rated to 260° C for continuous use, and then pulled back into the die so that the tips were flush with the front face of the die half. This ensured that they would be in contact with the component surface during casting. A further seven thermocouples of type J type and diameter 0.25 mm diameter were inserted so that their tips were in contact with the back surface of the spray coating, i.e., 1 mm back from the cavity surface. Finally, seven larger J type thermocouples, with a sheath diameter of 0.5 mm, were positioned in the 2 mm holes so that their tips were in contact with the die and were at a distance of

2.5 mm from the front face of the cavity. These holes were then backfilled with more epoxy resin to hold the thermocouples in position. The inlet water temperature for all the three channels was monitored using a combined pump/temperature controller unit with a digital readout. The outlet temperatures of the three channels was measured using 1 mm diameter thermocouples.

4

Experimental Results

4.1 Bond Strength Results. The effect of the different process parameters on the bond strength has been studied by carrying out several tests. In each test, a coated copper specimen was glued to a cylindrical pull bar and then the glued assembly was pulled to failure. Table 1 shows the details of the tests performed and the resulting bond strengths. The word “lifted,” in the table, indicates that the coating separated from the substrata before the pull test was carried out. Although identical process conditions were used in tests 22–27, differing results were obtained thus exhibiting the natural variability of the spray process due to its stochastic nature. This sample of results shows a standard deviation of 0.48 MPa and an absolute variation of about 8% which must be taken into consideration when making comparisons with other tests. The measured bond strengths varied from 2.32 to 50.8 MPa over all the tests. A typical load/displacement graph is shown in Fig. 9. Tests 7–10 suggest that the thickness of the coating on copper should be limited to 1 mm or less. If the thickness is greater, the spray coat was found to separate from the copper. As

Fig. 9 Typical load/displacement graphs

850 / Vol. 128, NOVEMBER 2006

Transactions of the ASME

Downloaded 12 Jun 2008 to 202.63.117.72. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm

Fig. 10 Strength/coating thickness graphs using 95MXC on 10T and 11T bonding spray

mentioned above, a cylindrical specimen of coated copper was glued to a cylindrical pull bar. This was true for all the tests except 12 and 13. In these two tests, a cylindrical pull bar was glued to a flat rectangular sample of coated copper and this resulted in essentially pulling a round portion from the rectangular area of coating. The registered strength for these tests is not solely bond strength but also includes the shear strength of the coating. This is confirmed by comparing the bond strengths registered in tests 13 and 30. The first plot in Fig. 10 is based on tests 28, 30, and 32 and shows the relationship between coating thickness and failure strength using 10T as the bonding agent. A similar relationship is shown using 11T as the bonding agent. In both cases, as the thickness increases, bond strength decreases. The high thermal conductivity of copper means that impinging droplets of spray are frozen quickly 共in the order of 1 ␮s兲 resulting in little or no fusion layer. Ideally the copper should be hot when receiving the spray. Tests 19 and 21 highlight the problems of preheating the copper. Heating the copper accelerates oxidization on the surface to be sprayed thus reducing the adhesion strength of the coating/copper interface. Comparing test 14 with 17, and 15 with 18 suggests that using the ArcJet facility available on the spray gun has a detrimental effect on the bond strength. When compared with Ampcolloy, the use of chrome-copper almost doubles the bond strength as can be seen by comparing test 31 with 42, and 32 with 43. There was no significant change in the bond strength if the bonding agent is changed from 10T to 11T as is evident from tests 31 and 32, 35 and 36, and 42 and 43. However, both these bonding agents resulted in a slightly higher bond strength than 75B as can be seen by comparing test 14 with tests 42 and 43. Also, using a combination of two bonding agents, 10T and 75B, instead of a single bonding agent 共10T兲, lowers the bonding strength drastically. For example, comparing test 33 with 37, the bond strength is reduced from 13.58 to 2.32 MPa. 4.2 Thermal Conductivity Results. The conductivity rig was first tested using materials whose conductivity was known. The Journal of Manufacturing Science and Engineering

disparity between the manufacturers data sheet figures and the experimental results was assumed to be a measure of the error of the experimental procedure. This error was found to be a maximum of 15% using copper and 8% using steel. Figure 11 shows the positions of thermocouples T1 to T6. The graphs in Fig. 11 show the average measured temperatures over a 3 min data collection period. The sprayed material was found to have a thermal conductivity of 14– 15 W mm−1 ° C−1 which was approximately half that of the material in its wire form. 4.3 Casting Simulation Rig Results. Two series of results were obtained using the die casting simulation rig: the first with the steel coated pseudodie block running in steady state mode and the second using the same block in transient mode. 4.3.1 Pseudodie Block—Steady State Results. Twelve runs were made using a half factorial combination of three pressures 共0.15, 0.2, and 0.275 MPa兲, four volumetric flow rates 共40, 80, 140, and 200 l / h兲, and two inlet water temperatures 共25 and 75° C兲. The initial flow rate used was 200 l / h, at a pressure of 0.2 MPa and an inlet temperature of 75° C. During this first run, the power to the heater was adjusted until the thermocouple, located 1.5 mm from the cavity surface, registered the same reading as the predicted temperature. This heater power setting was then noted and used as the power output for all other runs. The temperature readings at depths of 1, 2, 3, 4, and 5 mm from the cavity surface were recorded. As predicted, these operating conditions produced no visibly discernible boiling and the results are shown in Fig. 12. It may be mentioned that, in the case of the pseudo steel die block, reducing the flow rate from 200 to 80 l / h established vigorous boiling on the channel surface 关1兴. Altering the coolant pressure did not affect die temperatures significantly. The predicted temperatures for this die are compared with the experimental temperatures recorded by all the 36 thermocouples and the errors are shown in Fig. 13. The errors lie within a band of −17% to +9%. This compares with a band of −6%to +8% for the pseudosteel die block. The increased error is probably caused by the greater thermal variation due to the absence of boiling which tends to dampen temperature fluctuations. NOVEMBER 2006, Vol. 128 / 851

Downloaded 12 Jun 2008 to 202.63.117.72. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm

Fig. 11 Schematic of 1D conductivity rig and thermocouple results for 95MXC samples

4.3.2 Pseudodie Block—Transient Results. The transient tests were carried out at 0.125 MPa pressure, 80 l / h volumetric flow rate and an inlet temperature of 75° C. Two cycle times were used: 2 and 10 s. The time ratio of shutter open/closed was maintained at approximately 1 for both the cycle times. Figure 14 compares the transient results obtained from the copper and steel dies for a 2 s cycle time 关1兴. The temperature amplitudes with the copper coated die are much greater. This is probably due to the absence of boiling which tends to have a dampening effect. As expected, the amplitude of the temperature variations decreases with increased depth from the cavity surface. The amplitudes with a 10 s cycle time 共see Fig. 15兲 are much greater than the 2 s cycle time. However, as can be seen from Fig. 16, there is little variation in the mean temperatures; any apparent variation could be due, in part, to the inexactitude of the shutter open/closed ratio. The mean values of the transient variations are slightly lower than the corresponding value obtained from the steady state test. This could be 852 / Vol. 128, NOVEMBER 2006

due to the thermal feedback from the shutter causing the heater output to vary. Even though the shutter surface was painted black its absorptivity is less than 1. 4.4 Stepped Cavity Test Die—Thermal Results. The second phase of experimental results were carried out on a Dynacast machine using the copper die blocks shown in Fig. 7. While setting the die on the machine and during the first few minutes of operation thermocouples 4, 5, 6, 12, 14, 19, 20, and 21 ceased to function. Also, after a few dozen castings had been produced, the area of the die that contacts the nozzle was found to be damaged. This rendered the die unusable and a steel insert had to be attached to the die so that the injection nozzle could be seated correctly 共Fig. 17兲. The experiments were conducted at speeds of 31, 40, 50, and 60 cycles per minute 共cpm兲. When steady state conditions were reached for each speed, thermocouple readings were taken at a sampling rate of 200 Hz for Transactions of the ASME

Downloaded 12 Jun 2008 to 202.63.117.72. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm

Fig. 12 Measured and predicted temperatures on pseudodie block

about 30 s. The machine speed was then increased. Castings were produced up to a running speed of 64 cpm. The die showed numerous signs of damage particularly on the gate and runner, and around the core pins as shown in Fig. 18. The castings produced were of poor quality due to the many surface defects, which are clearly shown in Fig. 19. This was not unexpected, as porosity of up to 10% was estimated prior to casting, as shown in Fig. 20. The results presented in Fig. 21 represent the readings taken over two casting cycles of 1 s. The moment of injection is represented by values of approximately 0.8 and 1.8 s on the x-axis. Figure 21 also shows the comparison between the thermocouple readings and the predicted transient temperatures. Thermocouples 1–7 were in contact with the component surface. Of these only 1, 2, 3, and 7 produced results. Their readings are represented by one of the continuous curves in plots a, b, c, and g. Thermocouples 8–14 were set 1 mm back from the front face of the right die half measuring the temperature at the back of the spray coating. Of these seven thermocouples, only 8, 9, 10, 11, and 13 produced results. The temperature recorded by them are shown in plots a, b, c, d, and f. Thermocouples 15–21 were 2.5 mm back from the Journal of Manufacturing Science and Engineering

Fig. 13 Error of model predictions for the cavity wall of the pseudodie running in steady state mode

NOVEMBER 2006, Vol. 128 / 853

Downloaded 12 Jun 2008 to 202.63.117.72. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm

Fig. 14 Cyclic temperatures for steel and copper pseudodie blocks „30 cpm…

casting surface and, hence, measured the temperatures in the copper die block. Of these seven, results from four of them are shown in plots a, b, c, and d. Predictions for the casting surface temperatures from the model are shown at 0.04 s intervals until the casting is ejected. Predictions of the die temperatures are also shown at 0.04 s intervals throughout the casting cycle.

The melt was injected at a pot temperature of 410° C. It can be seen that all the thermocouples on the casting surface could not respond to the rapid temperature variation at injection. After 0.1 s into the cycle, however, all the thermocouples give reasonably good results. From the temperatures recorded by thermocouples 1, 2, 3, and 7, it appears that the surface temperature of the casting

Fig. 15 Cyclic temperatures for copper pseudodie block „6 cpm…

854 / Vol. 128, NOVEMBER 2006

Transactions of the ASME

Downloaded 12 Jun 2008 to 202.63.117.72. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm

Fig. 16 Mean cyclic temperatures

does not decay after injection as rapidly as predicted. This suggests a higher thermal resistance at the copper/spray interface than used in the computer simulation. The results also show that the central portion of the 2.5 mm thick section of the casting is the hottest area and probably the last to solidify. This is probably due to the flow pattern resulting in this region of the cavity being the last to be filled. Results from thermocouples 5 and 6 show that the thin 0.5 mm section of the casting is the coolest, resulting in almost instantaneous solidification in this region and inhibiting further feeding of the melt, which in turn could be the reason for the poor surface finish on the thin section. There is good agreement

Fig. 19 Steel runner insert

between the predicted and measured results for thermocouples 9, 11, and 13 but thermocouple 8 has registered a much higher value than predicted. This could be due to the ingress of molten zinc into the thermocouple hole. Readings from thermocouples 16, 17,

Fig. 17 Porosity in the spray coating

Fig. 18 Cavity damage after casting

Journal of Manufacturing Science and Engineering

Fig. 20

Casting defects

NOVEMBER 2006, Vol. 128 / 855

Downloaded 12 Jun 2008 to 202.63.117.72. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm

Fig. 21 Measured and predicted temperatures on stepped die

18, and 19 also show good agreement with the predicted values and confirm that the amplitude of the cyclic temperature variations decay rapidly within a couple of millimeters depth from the cavity surface. However, the flatness of the temperature curve for each of these thermocouples could be due to the poor physical contact between the tip and the die. Thermocouples 12–15 all read slightly lower than predicted over most of the casting cycle. The temperature curve from thermocouple 16, although slightly lower, shows excellent agreement with the predicted variation of temperature. Figure 22 shows a −15% to 32% difference between measured and predicted temperatures. An analysis carried out between the 856 / Vol. 128, NOVEMBER 2006

predictions and data over the range 0.08– 1 s for the measurements taken on the casting surface revealed an error range of −7% to +20% as shown in Fig. 23. Measurements of the die temperature were predominantly lower than predicted. This could be partly due to the difficulty of positioning the thermocouple tips accurately in the die and keeping the tips free from contamination by the epoxy cement. Overall the results show a mean absolute error of 8% on the die and 9% on the casting, and a maximum error of 32% overall. The main sources of error in the numerical model are: 共i兲

the absence of a flow model for the filling of the cavity. The assumption is made that the cavity is filled instanTransactions of the ASME

Downloaded 12 Jun 2008 to 202.63.117.72. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm

Fig. 22 Error of model predictions for the right hand die block of the stepped die

taneously with liquid metal at a uniform temperature; 共ii兲 the heat transfer coefficient between the melt and die is assumed to be uniform over the whole of the cavity surface and constant over the whole casting cycle; and 共iii兲 the boundary element mesh does not include features such as steel reinforcing pins, back plates; ejection pins, assembly screws and interfaces between the copper die and its steel reinforcements. The temperatures of the casting in both copper/alloy and steel dies are compared in Fig. 24. 4.5 Stress Measurements and Predictions. Strain gauges were attached to the coated pseudodie block at the locations shown in Fig. 25. Ideally a temperature compensation of zero is required in order to measure thermal stress. Therefore gauges with a thermal expansion of 1.0 ppm/ ° C were chosen. Gauges 1 and 3 measured strain in the x direction while gauges 2 and 4 measured

Fig. 23 Error of model predictions on the surface of the stepped casting

strain in the y direction. The predicted stress 关16兴 and the stress derived from experimental strain values are compared in Fig. 26. The mean absolute predictive error over all 80 predictions was approximately 10%.

5

Conclusions

The research demonstrated the feasibility of using copper dies, reinforced with steel inserts and coated with a thin layer of wearresistant material. The research demonstrated that copper-backed dies have the following advantages over traditional steel dies: 共1兲 Higher rates of heat extraction leading to shorter casting times. The theoretical models predicted that it should be possible to run the test die at 80 cpm compared to

Fig. 24 Comparison of steel and copper die temperatures

Journal of Manufacturing Science and Engineering

NOVEMBER 2006, Vol. 128 / 857

Downloaded 12 Jun 2008 to 202.63.117.72. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm

共2兲 More uniform temperature distributions were obtained on the die cavity surfaces. The disadvantages are: 共1兲 Copper-alloyed dies have to be strengthened with a steel base and columns to ensure sufficient strength to withstand the forces, thus making the die relatively more expensive to manufacture. 共2兲 Provision has to be made for special inserts to be placed in the die in the runner region to withstand nozzle impact. The wear-resistant material was deposited using the thermal-arc spray process and the following conclusions can be drawn from the experimental work:

Fig. 25 Strain gauge positions

40 cpm for a conventional steel die. This speed was not attainable on the Dynacast die casting machine but cycle times of 60 cpm were easily achieved even without cooling channel optimization.

共1兲 The bond strength of the spray coating is approximately inversely proportional to the thickness of the coating over the range tested, which limited the layer thickness to a maximum of 1 mm. 共2兲 Of the combinations tested, a chrome-copper die, sprayed using a single bonding agent of 10T produced the highest bond strength of 51 MN/ m2. 共3兲 Porosity in the sprayed layer at the die-cavity surface resulted in poor surface finish on both the casting and die. 共4兲 The presence of steel inserts can result in surface cracks appearing due to differential cooling between the two materials necessitating localized repair.

Fig. 26 Predicted and experimental stress values on the pseudodie block

858 / Vol. 128, NOVEMBER 2006

Transactions of the ASME

Downloaded 12 Jun 2008 to 202.63.117.72. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm

References 关1兴 Street, A., 1975, “Developments in Pressure Diecasting,” Int. Metall. Rev., 20, pp. 121–136. 关2兴 Bounds, S., Davey, K., and Hinduja, S., 2000, “An Experimental and Numerical Investigation Into the Thermal Behavior of the Pressure Die Casting Process,” ASME J. Manuf. Sci. Eng., 122共1兲, pp. 90–99. 关3兴 Prabhu, K. N., and Cambell, J., 1999, “Investigation of Casting/Chill Interfacial Heat Transfer During Solidification of Al-11% Si Alloy by Inverse Modelling and Real-Time X-Ray Imaging,” Int. J. Cast Metal Res., 12共3兲, pp. 137–143. 关4兴 Adeleke, J., 1998, “A Steptype Technique of Natural Regularization for Determining Interfacial Conditions in the Diecasting Process,” Scand. J. Metall., 27共6兲, pp. 240–245. 关5兴 Das, S., and Paul, A. J., 1993, “Determination of Interfacial Heat-Transfer Coefficients in Casting and Quenching Using a Solution Technique for Inverse Problems Based on the Boundary Element Method,” Metall. Trans. B, 24共6兲, pp. 1077–1086. 关6兴 Clark, L. D., Rosindale, I., Davey, K., and Hinduja, S., 2000, “Predicting Heat Extraction Due to Boiling in the Cooling Channels During the Pressure Die Casting Process,” Proc. Inst. Mech. Eng., Part C: J. Mech. Eng. Sci., 214共3兲, pp. 465–482. 关7兴 Clark, L. D., and Davey, K., 2001, “Novel Cooling Channel Shapes in Pressure Die Casting,” Int. J. Numer. Methods Eng., 50共10兲, pp. 2411–2440. 关8兴 Atkinson, H. V., and Rickinson, B. A., 1991, Hot Isostatic Processing 共The Adam Hilger Series on Manufacturing Processes and Materials兲, Kluwer, Dordrecht. 关9兴 Pawlowski, L., 1994, The Science and Engineering of Thermal Spray Coatings, Wiley, New York.

Journal of Manufacturing Science and Engineering

关10兴 Davey, K., and Hinduja, S., 1990, “Modelling the Pressure Diecasting Process With the Boundary Element Method: Steady State Approximation,” Int. J. Numer. Methods Eng., 30共7兲, pp. 1275–1299. 关11兴 Davey, K., and Hinduja, S., 1990, “Modelling the Transient Thermal Behavior of the Pressure Diecasting Process With the Boundary Element Method,” Appl. Math. Model., 14共8兲, pp. 394–409. 关12兴 Davey, K., Bounds, S., Rosindale, I., and Alonso Rasgado, M. T., 2002, “A Coarse Preconditioner for Multi-Domain Boundary Element Equations,” Comput. Struct., 80共7–8兲, pp. 643–658. 关13兴 Brebbia, C. A., 1978, The Boundary Element Method for Engineers, Pentech Press, London. 关14兴 Clark, L. D., Davey, K., Rosindale, I., and Hinduja, S., 2002, “Determination of Heat Transfer Coefficients Using a 1-D Flow Model Applied to Irregular Shaped Cooling Channels in Pressure Die Casting,” ASME J. Manuf. Sci. Eng., 122共4兲, pp. 678–690. 关15兴 Rizzo, F. J., and Shippy, D. J., 1977, “An Advanced Boundary Integral Equation Method for Three-Dimensional Thermoelasticity,” Int. J. Numer. Methods Eng., 11, pp. 1753–1768. 关16兴 Milroy, J., Hinduja, S., and Davey, K., 1997, “The Elastostatic ThreeDimensional Boundary Element Method: Analytical Integration for Linear Isoparametric Triangular Elements,” Appl. Math. Model., 21共12兲, pp. 763–782. 关17兴 Davey, K., and Clark, L. D., 2003, “Sensitivity and Optimization for Shape and Non-Linear Boundary Conditions in Thermal Boundary Elements,” Int. J. Numer. Methods Eng., 56共4兲, pp. 553–587. 关18兴 Alonso Rasgado, M. T., Davey, K., Hinduja, S., and Clark, L. D., 2006, “Boundary Element Stress Analysis for Copper Based Dies,” Comput. Struct., 84, pp. 254–267.

NOVEMBER 2006, Vol. 128 / 859

Downloaded 12 Jun 2008 to 202.63.117.72. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm