Smart Mold: Real-Time In-Cavity Data Acquisition Hamidreza Karbasi, PhD, P. Eng. School of Engineering Conestoga College ITAL Kitchener, Ontario Henry Reiser BSc, MASc, CMfgE, FBCASI School of Engineering Conestoga College ITAL Kitchener, Ontario KEYWORDS Smart Mold, Data Acquisition, Plastic Injection, Process Control. ABSTRACT Continuous improvements in product quality and production cost savings are crucial to maintaining a competitive edge in the injection molding industry. To reduce high-run production costs, automation can be employed to reduce the cycle time of molding, part inspection and verification, and machine setup. A fundamental step in the automation and optimization of any plastic injection process is to precisely design, measure and monitor the injection molding process such that key process variables are observable and controllable. This research paper investigated the measurement of two key process variables during the production process: in-mold/cavity pressure and temperature. Results of the in-mold pressure and temperature profiles demonstrated excellent affinity and high repeatability with the typical trends for their class of polymers. The in-cavity pressure profiles corresponded well with the filling, packing, and holding points of the machine pressure profile. Based on the results of this research, a new process-control strategy for plastic injection machines is proposed. Introduction To obtain the desired critical quality attributes in molded products, the injection molding process must be consciously instrumented and designed such that the key process variables can be measured and used in process control. As shown in Figure 1, typically, a full control block diagram of an injection molding process involves three nested loops. The first loop (the most inner) is the one which controls the machine parameters such as speed, pressure, and temperature. The second loop (the middle loop) feeds back the process variables such as in-mold temperature and pressure and eventually, the third loop (the outer loop) takes care of part quality feedback. The inner loop has been well developed by injection machine manufacturers. The middle loop is in the early stages of commercialization and research on control algorithms and sensor technologies is ongoing. Research and implementation of the third level control (quality control) started in the last decade and still faces many difficulties and challenges such as on-line feedback of quality related variables (part weight, size, and surface finishing).
Machine Inputs Machine Actuators Machine Control Process Control Set Point Control Machine Feedback Process Variables Feedback Injection Process Quality Feedback Part Attributes FIGURE 1. Block diagram of three injection molding control loops. This paper focused on the middle loop and the measurement of two process variables: in-mold temperature and pressure. Pressure sensing of a polymer melt poses some significant challenges relative to pressure sensing of compressed air or hydraulic fluid. High pressures, corrosion at high temperatures, exposure to abrasive materials, and long term variation in readings as a result of temperature variation are some of the significant challenges faced. travel, filling and packing and holding pressure and time. As a screw injects the plastic melt, it usually passes through 5 different segments. The operator is able to adjust the length and position of each segment as well as the desired screw speed at each segment. The cavity pressure within an injection mold can deliver very precise information about the filling phase, the pack phase and the holding pressure phase. The different phases in the cavity pressure profile and the typical correlation between the cavity pressure profile and the injection pressure are depicted in Figure 2. Temperature sensing of the polymer melt poses many of the same significant challenges as pressure sensing, although temperature sensors typically have no moving parts and may be more mechanically robust and less expensive than pressure sensors. However, temperature sensors are not generally as accurate as pressure sensors. The reason is that melt temperature sensors are necessarily embedded in the mold steel, so there is a significant heat transfer from the sensor head to the surrounding steel. Accordingly, temperature sensors may have a significant lag phase and steady-state error in the measurement of melt temperature. The remainder of this paper will first review the injection molding process, and the experimental design used in this research including the mold, injection machine, and LabView setup will then be explained. Finally, the experimental results will be discussed, and a conclusion and future directions for research will be presented. Injection Molding Background Review of Typical Process Used To produce a good quality injected part, some machine parameters must be tuned and adjusted. The main parameters are: mold and barrel temperatures, velocity profile, screw FIGURE 2. Different phases in cavity pressure profile for amorphous materials. At the filling phase (1), polymer melt enters the cavity. As soon as the flow-front reaches the sensor (2), the pressure is registered. The pressure should rise in a near-linear gradient parallel to the duration of the filling time. Once the cavity is filled (3) the melt is compacted during the compression phase to ensure the reproduction of the contours of the mold cavity. The holding pressure phase follows after the maximum cavity pressure has been reached (4). The holding pressure phase compensates for the high thermal contraction of the polymer material, i.e., the reduction of its volume following the cooling down process, by introducing more material. Up to 10% of the part volume is being pushed into the cavity during this production phase. The fact that the molded part starts to cool down and solidify near the cavity wall inhibits the pressure transfer. The melt flow from the area in front of the screw to the cavity is slowed down as the viscosity of the material increases and the flow channel becomes more constricted in the process (gate freezing). The progressive solidification of the melt in the gate area (5) and the progressive thermal contraction causes the pressure within the cavity to drop to ambient levels (6).
Amorphous and semi-crystalline thermoplastics are two different polymer materials which display different compression behaviors even though their viscosity is identical. As shown in Figure 2, during the holding pressure phase of injection molding processes involving amorphous polymers such as polystyrene (PS), acrylonitrile-butadienestyrene (ABS), styrene-acrylo-nitrile (SAN), polymethyl methacrylate (PMMA), polycarbonate (PC), and polyvinyl chloride (PVC), the mold cavity pressure drops to ambient pressure levels parallel to the declining part temperature. This drop in pressure is due to the increase in viscosity and the corresponding deterioration of the pressure transfer from the area in front of the screw. As seen in Figure 3, due to an initially sufficient pressure transfer during the injection molding processes involving semi-crystalline materials such as polyethylene (PE), polypropylene (PP), polyamide (PA), and polyoxymethylene (POM), there is almost no change in the mold cavity pressure during the period after the compression phase and at the on-set of the crystalline melting point. After that, however, the significant volume contraction during crystallization brings about a sudden drop in the pressure. The duration of the holding pressure phase depends on factors such as the wall thickness of the molded part, the degree of crystallization, and the processing parameters. The crystalline melting point of semi-crystalline materials for example, is dependent on the prevailing cavity pressure. Further comparison between amorphous and semi-crystalline materials indicates that a quantity of semi-crystalline material needs to be introduced to the cavity at the beginning of the holding pressure phase to achieve a pressure build-up. When the melt cools down during the holding pressure phase, more semi-crystalline material must be introduced into the cavity to compensate for the volume contraction and to prevent voids in the finished part. Experimental Set-up Mold Setup An ASTM plaque mold, Figure 4, was selected for testing the sensors. This mold makes 4 different parts for various mechanical tests on new materials. Runner shut-offs were used such that only the specimen (A) was molded. Figure 4 clearly illustrates placement of the sensors. The temperature sensor is located at the front of the part, near the start of the fill, and the pressure sensor was positioned on the rear ejector pin, near the end of the fill. FIGURE 3. Cavity pressure profile for semicrystalline materials. FIGURE 4. ASTM plaque mold. Injection Molding Machine An 80 ton horizontal machine was used for the trial. This machine and the PC used for data acquisition are illustrated in Figure 5. Figure 6 illustrates the mold set-up with integrated sensors on the machine. LabView Setup A LabView setup was used to facilitate data acquisition and sensor calibration for this project. The LabView hardware and software layout, and their inter-connections to the in-
cavity sensors, are shown in Figure 7. On the software side, a PC with LabView (7.1) read the sensors and wrote the data into an Excel file via a graphical program in LabView. On the hardware side, a data acquisition card, PCI 6221, interfaced the PC/LabView software to an I/O board, SC-2075. The I/O board collected analog signals from sensor amplifiers through its channels 0 and 1. A 24VDC power supply fed the Kistler amplifiers, and the 2209A1 and 5041E1 temperature and pressure amplifiers, respectively. Finally, the outputs of the sensors were wired to the amplifier inputs. L Real-Time In-Cavity Pressure and Temperature Measurements: Graphical Programming, DAQ card (PCI 6221), and I/O board SC-2075 Pressure Reading Temp. Reading 68-pin port Analog Input, CH0 Temperature Amplifier Connector # 3 R LabView Graphical Programming, # 1 Conversion and Calibration Conversion and Calibration PCI 6221, # 2 Analog Input, CH1 Pressure Amplifier Power Supply # 5 # 7 # 6 Monitor and Write to file Monitor and Write to file SC-2075 # 4 Temp. sensor Pressure Sensor # 8 # 9 Cavity List of Components: 1- PC with LabView 7.1 software 2- DAQ card (PCI 6221), installed on the MB of PC 3-68-pin connector cable (supplied with I/O board) 4- I/O accessory board (SC-2075) 5- Kistler 2209A1 amplifier for temperature sensor 6- Kistler 5041E1 charge amplifier, thumbwheel 7-24 VDC power supply 8- Kistler 6193A04 temperature sensor, type K 9- Kistler 9221A quartz pressure sensor FIGURE 7. LabView hardware and software layout to acquire data from the cavity sensors. Results & Discussion FIGURE 5. PC for data acquisition adjacent the horizontal molding machine. FIGURE 6. Mold set up with integrated sensors on the machine. As discussed previously, there are a number of machine parameters which must be set to produce a good quality injection plastic part. The expert setup mode is referred to as those parameters which are adjusted by an expert operator at their optimum values through trial and error. The machine parameters are supposed to differ from the optimum values in a non-expert setup mode. In this project, the two parameters of packing and holding pressures are deviated from their optimum values to simulate a non-expert mode while the rest of parameters are kept at their optimum values. This approach enabled the study of the effects of machine parameters on in-cavity pressure and temperature using controlled conditions. The injection molding was run at the aforementioned modes six times. Figure 8 shows the samples produced from white Polypropylene, left, and black Zytel, right.
FIGURE 8. Parts produced from Polypropylene (left) and Zytel (right). Polypropylene Tests with the Polypropylene material were conducted six times in both expert and nonexpert modes and in-cavity pressure and temperature data were collected. Figure 9 through Figure 14 provide the data collected. The results of the expert mode are illustrated by dashed lines (expert 1 to 3) while the non-expert mode results are illustrated by continuous lines (non-expert 4 to 6). In Figure 9 and Figure 10, the pressure and temperature profiles for complete molding time are shown. The cycle-time and injection-time were 27.4sec and 0.48sec, respectively, according to machine setup. For the non-expert mode the packing and holding pressures were degraded by 15% from 500psi and 275psi to 425psi and 200psi, respectively. The mold and barrel temperatures were kept at 90ºF (32.2ºC) and 374ºF (190ºC) respectively. By the end of the molding time, the part was ejected by ejector pins and small pressure ripples were recorded by the pressure sensor (see Figure 9.) In Figure 10, the non-expert temperature profile shows a small upward shift in comparison to the expert profile. This is due to a decrease in heat transfer from the melt to the mold, as increased pressure led to higher heat transfer. Higher heat transfer means less cooling time, and as a result, savings in molding time and cost. In Figure 11 and Figure 12, the first 11sec. of the molding time is illustrated for better visualization of pressure and temperature profiles. Comparing the trend of expert pressure profiles in Figure 11 with the pattern in Figure 3, two differences are noted. First, the packing pressure goes to a pick and then drops to the holding pressure. In the pattern they are almost the same and this is due to the set-up of the machine for different packing and holding pressure by the operator. Second, the pressure is sustained in an almost constant level from 6 th to 8 th sec. The reason for this can be the screw recovering or feed time which occurs during this time. This creates a back pressure by charging the material into the barrel for the next shot. Other than these two differences, the trend of the pressure profile is quite similar with the pattern of other points such as solidification of the melt in the gate area, and reaching ambient pressure levels within the cavity, caused by progressive thermal contraction. Figure 13 and Figure 14 show the pressure and temperature profiles in the first second of molding. Comparison between the expert pressure profiles in Figure 14 and the pattern in Figure 3 reveals an excellent trend match at the start of the filling phase, the flow-front reaching the sensor, completing the filling phase, and the maximum cavity pressure point. These similarities can also be seen, if we compare these points with points 1 through 4 in Figure 2, respectively. Generally, the graphs in this section demonstrate excellent consistency and high repeatability for expert and non-expert results. The in-cavity pressure and temperature profiles indicate good agreement with machine expert and non-expert settings. FIGURE 9. Pressure profile for complete molding time of Polypro (expert and non-expert modes).
FIGURE 10. Temperature profile for complete molding time of Polypro (expert and non-expert modes). FIGURE 13. First second of pressure profile for Polypro (expert and non-expert modes). FIGURE 14. First second of temperature profile for Polypro (expert and non-expert modes). FIGURE 11. First 11 sec of pressure profile for Polypro (expert and non-expert modes). FIGURE 12. First 11 sec of temperature profile for Polypro (expert and non-expert modes). Process Control Strategy Generally established molding principles suggest true process capability relies on consistent delivery of four key plastic variables inside the mold cavity: plastic temperature, flowrate, pressure gradient, and cooling rate and time. When these four plastic variables are matched, the same part will be created, regardless of the machine the mold is run in. They also suggest that the effects of the four aforementioned variables can be seen in the incavity pressure profile. Therefore, the template for an ideal process can be created based on the continuous monitoring and recording of incavity pressure, and can be used to adjust other machine s parameters. Although installing a high number of pressure sensors might not be economical, for a normal size mold and existing pressure sensor technology, it is completely achievable.
Based on the aforementioned principles and the results of in-cavity pressure profiles obtained through this research, we will be able to establish an effective process control that set all machine parameters based on a given set point (template) for in-cavity pressure. For any given part, this template is unique and can be found through the following simple procedure: An expert operator will adjust the machine parameters (expert setup) by trial and error to produce a good quality part for production. When the expert operator is satisfied with the quality of the part, a pressure sensor(s) will continuously record the status of the in-cavity pressure. Since in-cavity pressure profiles can directly or indirectly play a fundamental role in the constant delivery of quality plastic parts, we will consider it as a set point for the set-up of a fully automated machine intended to produce the same part. By developing this set point, we will reduce the control system to a set point tracking system which is simple and easy to implement. By utilizing this control system, the part manufacturers can save a considerable set-up time. In addition, they can continuously control the quality of the parts through the real-time monitoring of in-cavity pressure profile, prevent any problems that may develop during production runs, and reduce the production of defective parts. The consistency and reliability in the measurement of in-cavity pressure profiles in expert mode is a very important finding of this research. It will enable us to find a pressure template for any part employing the same techniques, and using it as a set point in a set point tracking machine control system. Conclusion & Future Work Pressure profiles for Polypropylene polymer demonstrated excellent agreement and high repeatability with the typical trend for its class of polymer. The in-cavity pressure profile corresponded well with the filling, packing, and holding points of the machine pressure profile. Future projects can be divided into two main themes: 1) as a continuation of this research project the first theme investigates improving the existing process controls by diversifying the type, location, and number of sensors, and enhancing or replacing the existing process control algorithms with more advanced ones; 2) the second theme focuses on the exploration and implementation of new technologies. The objectives that can be addressed through the first theme could include: Employment of new sensors to prevent the production of poor quality/scrap parts in a high speed production line by employing multi-cavity tools. The most common problem associated with producing parts with numerous small details (such as ribs and webs) is the short-shot phenomena. Short-shot parts are scrap since their cavity is not fully packed. Fill sensors, such as an optical sensor, ultrasonic sensor, and the Linear Variable Differential Transformer (LVDT) can be employed to detect and monitor mold separation, flow-front advancement during filling stage, and incavity formation of the parts. Short-shot problem could be addressed by installing more in-cavity pressure sensors. However, this is not always practical with existing pressure sensors technology specially when dealing with busy and miniature parts [2, 6]. Development of an I/O-based model for the plastic injection molding process. Considering the highly non-linear behavior of plastics, developing experimental models could be an interesting future project. By employing additional in-cavity pressure and temperature sensors, and testing different materials, we are able to develop an I/Obased model for the molding process in which machine parameters (injection speed, barrel temperature, packing pressure ) are inputs and process variables (cavity pressure, temperature ) and quality variables (part weight and dimensions ) are outputs. This model will be essential to implement any adaptive control algorithms such as model predictive control (MPC) [2]. Establishment of real-time feedback from quality variables. Among machine parameters, process variables, and quality variables, quality variables are the most difficult to measure in real-time and establishing a methodology for their on-line measurement would be an excellent future research project.
The second theme is to focus on exploring and developing new technologies including: Embedding a fiber Brag grating sensor into a cavity in which Satellite 6 or cobalt-based materials will be laser deposited to form a layer around the sensor and improve wear resistance in pre-determined locations. This technology enables us to run tiny fibers anywhere under shallow in-cavity surfaces and measure pressure and temperature continuously across the cavity. This technology will be particularly effective with small and busy molds that conventional pressure sensors can not use because of their sizes. Wireless pressure and temperature sensors can be employed to reduce instrumentation and mold modification costs, utilize sensors at strategic locations and sensors networking, and expedite data management and analysis. Acknowledgments The authors are grateful to Ontario Center of Excellence Material and Manufacturing Ontario (MMO) and Polymer for their financial support. The authors are also thankful to Paul Murawsky, Polymer Technologies Inc., for his help as an expert operator, and Tyler Gross and Graham Beattie, Conestoga College ITAL students, for their support in conducting the literature review and purchasing the sensors. References How to make your molds machine and location independent by utilizing in-cavity pressure data. www.ides.com/articles/cavity_pressure_data.asp Kazmer, David O., Knepper, P., Johnston, S., A review of in-mold pressure and temperature instrumentation. ANTEC Tech Papers 2005, pp. 3300-3304. Kazmer, David O., Polymer injection molding technology for the next millennium, Journal of Injection Molding Technology, 1997. 1(2): p. 81-90. Masuda, N., Sutoh, K., and Yokoi, H., Measurement of pressure distribution in partial thin-wall cavity in ultra-high-speed injection molding, The University of Tokyo, Study group: Ultimate injection molding in 2003 of the foundation for the promotion of industrial science. Kistler website: www.kistler.com, Plastics: 100% Quality Injection Molding TAT Ming Engineering Works LTD., www.catalog.com.hk/tatming/, Quality injection molding. Yokoi, H., Watanabe, J., Masuda, N., PPS-18 Abstract for the 18th Annual Meeting of the Polymer Processing Society, 271, 2002. Zhang, L., Charles, C. B., Gao, R. X, and Kazmer, D. O., "Design Of Ultrasonic Transmitters With Defined Frequency Characteristics For Wireless Pressure Sensing In Injection Molding," IEEE Transactions on Ultrasonics.