Perdicting the performance of copper-alloy mold cores.
Many mold builders, mold engineers, and professionals involved in mold specification are reluctant to use copper alloys to enhance mold cooling because of uncertainties about how to apply them. A series of questions have been raised about performance characteristics and design guidelines. This study represents the first of a series of comprehensive investigations to establish guidelines for the application of copper-based alloys.
Until recently, it was difficult to do a three-dimensional (3-D) cooling analysis of a mold containing copper-alloy inserts. A two-dimensional (2-D), steady-state, finite-element cross-sectional analysis could be used for their evaluation. However, it required the time investment to build a completely different model. Moreover, the model and boundary conditions had to be modeled and chosen carefully to obtain useful results. For these reasons, few cooling analyses were conducted. Mold engineers were, therefore, forced to guess at the effect of installing higher thermal conductivity core and cavity inserts. Even if a cooling analysis was run to identify existing hot spots, further 2-D analysis was rarely performed to investigate the [TABULAR DATA FOR TABLE 1 OMITTED] effect of installing high thermal conductivity inserts in hot spots.
In the past several years, this situation has been improved with the introduction of 3-D analysis for molds containing more than one tooling material. A 3-D analysis allows the comparison of various tooling-material inserts for the reduction of hot spots and improvement in the temperature distribution in the mold. It is a well-accepted precept that a correctly designed mold, which does not vary in temperature, will substantially contribute to part integrity and will reduce, if not eliminate, warpage. Therefore, the fundamental question to be answered by this work was how accurately could 3-D boundary-element cooling software predict actual mold-temperature distributions. Assessment of this question required both analysis predictions and production trials.
This experiment consisted of five treatment groups and a replicate comparing 420 SS, H13, BeCu (C17200 and C17510), and NiSiCrCu (C18000) cores, running 10 [degrees] C cooling water under turbulent conditions. The mold that was used produced a 33-mm bottle cap with a snap ring that was stripped off the core using an ejector ring. A Huntsman polypropylene copolymer was used.
Cooling analyses were run using Moldflow's MF/COOL Version 4.1 3-D software with the process settings that occurred in the experimental design. Inputs to the cooling analysis included the 3-D finite-element model [ILLUSTRATION FOR FIGURE 1 OMITTED], mold-material properties, heat load, water conditions, and cycle-time components. The required property values for the mold materials were provided by the suppliers of the alloys (Table 1). Heat load for the cooling analysis was obtained by a finite-element filling analysis, using the average mold temperature, melt temperature, and injection time that occurred in the molding trials. The water conditions included the flow rates and water temperatures for each of the three circuits in the tool. Components of the cycle included injection time, pack/hold time, cooling time, and clamp-open time. The mold-process parameters are listed below.
Fill time = 0.3 sec
Coolant temperature = 10 [degrees] C
Pack/hold pressure = 2758 kPa
Water flow rates; line sizes:
Core = 14.76 l/min; 1/4 NPT 11 mm Cavity = 8.93 l/min; 1/4 NPT 11 mm Core plate = 8.93 l/min; 1/8 NPT 8 mm
Trials were conducted on a 85-ton hydraulic toggle injection molding machine (built 1992) with a 35-mm injection screw, and a 5-oz shot, and equipped with EL controls and an AEC hopper loader (1993). Mold water was regulated by an AEC mold-temperature controller (1993) and a 3-ton Thermal Care chiller (1995). Circuit water was monitored on all zones for flow rate, input and output temperatures, and pressure loss through the circuit. Direct machine data were acquired by an RJG Technologies Dart system that monitored cavity pressure, screw position, and hydraulic pressure. Part weight was determined at press side using an Acculab electronic balance with a resolution of 0.01 gram. Critical weights during gate-seal and cooling-time analyses were verified by a Sartorius analytical balance with a resolution of 10 micrograms.
Laboratory temperature and humidity were continuously monitored. Core temperatures were measured with a J-type thermocouple installed to within approximately 3 mm of the core surface. The thermocouple was inserted into a tightly fitting well and embedded in a heat-transfer compound to eliminate any air gaps. Surface-temperature readings in the core and cavity were taken using a K-type right-angle thermocouple.
Experimental Design and Procedure
For each core-cavity combination, a gate-seal experiment was run, starting at 3.0 sec and working down until gate seal was identified by a shift in part weight. Weights were taken at press side with the pack/hold time decreasing in 0.2-sec increments until a drop was detected. At this point, the last stable weight was repeated and the time was reduced by 0.1-sec increments until a drop in part weight reoccurred. Hold was then set to the shortest time that produced a stable part.
TABLE 2. Predicted Core and Cavity Temperatures, [degrees] C. Insert Core Cavity Material Min. Max. Range Min. Max. Range 420 SS 29.8 48.6 18.8 19.4 30.0 10.6 H-13 28.6 45.4 16.8 21.8 29.8 8.0 C17200 20.4 25.5 5.1 20.8 30.2 9.4 C17510 17.2 20.4 3.2 20.5 30.4 9.9 C18000 17.3 20.6 3.3 20.6 30.6 10.0
Cooling times were established following determination of gate seal. Previous experimentation on this tool suggested that the maximum cooling time required to mold parts was approximately 5.5 sec. Thus, 5.5 sec was chosen as a logical starting point to determine the required cooling time. It was observed that solidification of the snap ring, or undercut, governed the cooling time. The last area on this undercut to solidify was directly across from the gate at the weld line. Prior to complete solidification, the force of ejection would cause a small deformation that appeared as a small buckle on the exterior surface of the cap. At the point at which sufficient solidification had occurred, the part could be ejected without deformation. Cooling times were reduced in 0.2-sec intervals from 5.5 sec to 1.5 sec. The absence of buckling in the parts was then verified after cooling to determine the last point when enough solidification had occurred. The last stable cooling time was run and decreased by 0.1 sec to identify the exact point of cooling.
All of the 0.1-sec-increment parts were reweighed on the analytical balance to make certain that press-side measurements were in fact correct. The process was returned to the minimum cooling cycle that produced an unbuckled part. One hundred shots were then produced. Every other shot was saved for later verification and measurement. Core and cavity surface temperatures were then rechecked before the next run.
Figure 2 shows a 1.9-sec range in the minimum cooling times required when the slowest steel core was compared with the fastest copper-alloy core. Fixed-cycle times were 4.5 sec for all core materials [ILLUSTRATION FOR FIGURE 3 OMITTED]. Thus, actual cycle times ranged from 9.7 sec for 420 SS to 7.8 sec for C18000. These results were explained well by the cooling analysis.
Experiments subsequent to the tests detailed in this article have demonstrated that differences of 0.2 sec or less for copper-alloy cores are not transferable to other configurations of 33-mm caps. In other words, the differences between C17200 and C17510, as well as those differences between C17510 and C18000, should be regarded as insignificant. However, a significant difference was shown between C17200 and C18000.
MF/COOL-predicted surface temperatures, which ranged across the entire core [ILLUSTRATION FOR FIGURE 4 OMITTED] and cavity surfaces, are compared with actual readings taken during the experiment in Tables 2 and 3. The experimental readings were taken just subsurface (which would slightly underpredict surface temperature). Moreover, readings were taken at the surface in the center of the core and cavity, as described earlier. It should be noted that the center of the core was predicted to be the coolest point and that the center of the cavity was generally predicted to be the warmest point on the mold. Thus, one would expect actual numbers to fall at the low end of each of the predicted scales.
Cavity temperatures also were predicted to change as progressively more efficient copper-core cooling decreased the cycle times. Therefore, one would expect different cavity temperatures to occur, dependent upon the thermal conductivity of the core. Predicted median cavity temperatures (Table 2) and their variances from the actual temperatures (Table 3) during running of the various cores were as follows: 420 SS core, median = 24.7 [degrees] C, variance = 0.4 [degrees] C; H13 core, 25.8 [degrees] C and 0.3 [degrees] C; C17200 core, 25.5 [degrees] C and 0.6 [degrees] C; C17510 core, 25.4 [degrees] C and 1.0 [degrees] C; and C18000 core, 25.6 [degrees] C and -0.3 [degrees] C.
The copper-alloy cores showed a much more uniform temperature distribution with a lower net running temperature than the steel cores. The higher thermal conductivity of the alloy cores, in turn, produced about a 20% reduction in cycle time.
The predicted core and cavity temperatures from the cooling analysis were found to correlate well with the actual temperatures, with the predicted temperature ranges closely overlaying those of the actual measurements. Actual core temperatures fell in the lower portions of the predicted temperature ranges. This was anticipated, as the lowest temperatures were projected to be at the centers of the core surfaces, the same area where the measurements were taken. The only exception was that the C17510 and C18000 cores showed statistically identical core temperatures in the analysis while showing a one-degree difference in actual core temperatures. The agreement between the predicted and actual temperature ranges for the cavity was excellent, showing a maximum variance of only 1.0 [degrees] C.
The three copper-alloy cores demonstrated substantial improvement over the steel corex Subsequent tests of 33-mm-cap core and cavity combinations have supported the indistinguishability of C17510 and C18000, as shown in the cooling analyses. The agreement of the cooling analysis with laboratory data appears to be dependent upon what attributes are being measured and when in the cycle measurements are taken.
TABLE 3. Actual Core and Cavity Temperatures, [degrees] C. Insert Core Material Min. Max. Range Cavity 420 SS 26.6 28.4 1.8 24.3 H-13 26.1 27.8 1.7 25.5 C17200 19.4 21.7 2.3 24.9 C17510 19.4 20.6 1.2 24.4 C18000 17.8 20.0 2.2 25.9
The authors extend their thanks to the Copper Development Association Inc. for providing necessary support for this research. Performance Alloys & Services Inc. constructed all necessary mold components. Ampco Metals, Brush Wellman Inc., Copper and Brass Sales, and NGK Berylco provided alloys for the mold. We are also deeply appreciative of the guidance on the project provided by Robert Dealey, Cliff Moberg, and Dr. Dale Peters.
This research was completed with the investment of more than 650 hours of work by the following undergraduate students: Tom Bayer, Mike Carlson, John Frei, Todd Harroun, Jerome Hund, Tim Perkins, Joe Rewa, Frank Rinderspacher, Dennis Rosten, Rick Truza, Ryan Wejrowski, and Chong Wong.
1. C. Austin, Moldflow Design Principles, Moldflow, Kilsyth, Australia (1991).
2. M. Vander Kooi, P. Engelmann, M. Monfore, and S. Ramrattan, SPE ANTEC Tech. Papers, 42, 719 (1996).
|Printer friendly Cite/link Email Feedback|
|Title Annotation:||new 3-D mold-cooling software|
|Author:||Engelmann, Paul; Dawkins, Eric; Shoemaker, Jay; Monfore, Michael|
|Date:||Oct 1, 1996|
|Previous Article:||"3R" unscrewing mold technology.|
|Next Article:||Robert D. Schad - one to one.|