Next Steps in Machine Control for Rotational Molding

(Published in Rotation Magazine, Volume VII, Issue 3, Fall 1998) 

Introduction

In recent times, there have been many areas where rotational molding has taken significant leaps forward. The drive to make it a broadly used process with appeal to designers and quality conscious end-users alike has involved everyone from molders to material suppliers and from mold makers to machinery manufacturers.

A common request, particularly where ISO9000 is in use, relates to better tracking of processing conditions for individual parts; “How can you certify that every part has been cured to a consistent level?”, or perhaps, “Can you ensure that the parts have a consistent crystalline structure?”. Customers who are familiar with detailed reports from molding processes such as injection molding and blow molding have increasingly rigorous demands for repeatability and consistency from rotational molding. And rightly so.

Rotational molding is broadly perceived as a low-tech process with little or no control beyond the hard-earned experience of shop-floor staff. There are four major aspects that immediately strike a potential new user:

1.      Simple molds with lower engineering finishes than injection or blow molding – typically with oxidized surfaces where the mold has cycled through cooling chambers using water.

2.      Higher levels of operator involvement in demolding with a high reliance on their skill for final part quality - production rates, and therefore costs, are not as easily defined by machine settings as, for example, injection molding.

3.      Higher shrink and warpage levels than most other processes – high dependency on release application and maintenance. Customers demanding tight tolerances can be frequently frustrated.

4.      Low levels of feedback from machinery relating to actual part quality and cure levels - direct measurements of parts are complicated by the combined effects of biaxial rotation and cycling between a high temperature oven and a water cooled environment.

However, this is changing. Recently, there have been improvements in all four of the above areas:

1.      Mold makers can now offer CNC machined tooling that reduces clamping forces and improves parting line flash. Machines have improved cooling efficiency with reduced dependency on water in many instances.

2.      Demolding systems that use pneumatic clamping for quick mold opening and automatic handling of simple parts using robots are in use in a number of molding facilities. As both experience and demand grow this should expand into mainstream molding.

3.      Pressurized systems that maintain part dimensions and improve surface finish by forcing parts against the mold surface using low internal pressure can significantly improve tolerance levels.

4.      Recent developments such as the Rotolog^TM system and the use of continuous InfraRed Sensing* have enabled rotomolders to observe processes occurring inside the mold and relate them to data which can provide data for quality control purposes.

The paper discusses the use of IRT sensing for process monitoring and the potential it offers in taking machine controls to a higher level.

Overview

The Rotolog^TM** system, introduced in 1991, allowed molders to ‘see’ inside the mold for the first time. Temperature profiles such as those in Figure 1 could be correlated to the stages occurring during melting and fusion of the polymer as parts were formed. The temperatures could be further related to the properties of the material such that the level of cure could be reasonably predicted. The system utilizes a radio transmitter housed in an insulated vessel to protect it from heat and cooling systems that can typically withstand several molding cycles before needing to be cooled. It is an ideal diagnostic tool for developing an understanding of the molding profile for new parts and quickly establishing machine operating parameters. It is also a powerful tool for troubleshooting molding problems and comparing heat transfer across tools for balancing of wall thickness distribution.

Figure 1 Typical Data Collected during Molding of a Polyethylene Part using Rotolog^TM       

For continuous process feedback, development work has been carried out using infrared thermometry (IRT). This method of gathering data allows temperatures to be gathered directly at the surface of the mold as it rotates in both the oven and cooling environments. This information is gathered using static sensors that pass the information directly back to the machine for analysis continuously during molding operations. The objective for this development work has been to provide production personnel with data that allows repeatability of production cycles.

Infrared sensing is extensively used in many processes and has been reported on recently for rotational molding [1,2]. Spinning molds introduce an extra level of complexity in handling the data that can be gathered. This is dealt with by treating the data collected as a ‘map’ of the surface of a mold or mold spider. Data is collected at high rates in order to provide as smooth a profile as possible – it is then processed to provide a single curve for use as a process control signal for the machine.

System Configuration

The tests described here were conducted on a Ferry 220 carousel machine equipped with a single oven and two cooling stations. The processor included details for the latest RotobaseÒ software from Ferry Industries for data collection and management. Two IRT sensors were used – one mounted above the oven chamber and a second above the second cooling station. This allowed collection of data at the most critical stages of the cycle. The sensors were set to gather data at 8 times per second when used with a shopfloor based PC and up to 20 times per second when directly connected to the PLC within the Ferry 220 machine. These devices have been under continuous operational testing for approximately 18 months at Remcon Plastics.

Most of the testing has been conducted on a standalone PC that gathered data without interfering with the operation of the machine. This allowed mock recording of cycle end-points to test the consistency and repeatability of the data. It also allowed the effect of variables to be tested without changes to the machine settings. Recent programming changes have allowed the software to operate directly on the PLC of the machine without the need for the independent PC. The software automatically scans the data from the sensors and relates it to set-points established for a particular mold.

Typical Data

Typical data collected for both stations is shown in Figures 2 and 3. This data has been processed through three stages: screening the incoming data for background effects, eliminating rotation effects and then finally smoothing for output. The curves represent the surface temperature of the mold and do not give the same sensitivity of data as the internal data that can be gathered using Rotolog^TM. However, the data does allow set-points to be clearly defined for repeatable production cycles.

Figure 2 Typical Data Collected during Molding of a Polyethylene Part using IRT Sensing  (Aluminum Mold - 0.200” (5.1mm) Part)

Figure 3 Typical Data Collected during Molding of a Polyethylene Part using IRT Sensing (Aluminum Mold - 0.125” (3.2mm) Part)  

Figure 2 shows an external temperature profile for an aluminum mold 3/8” (9.5mm) thick with a 0.200” (5.1mm) thick polyethylene part with an oven temperature of 600°F (315°C). The peak temperature at the end of the oven cycle was just over 500°F (260°C) after 15.5 minutes. Note that the cycle settings for this part included a 5 minute cooling step in the first cooling station which meant that the temperature of the mold had fallen to 315°F (157°C) before entering the cooling station. Fan cooling was used to cool the surface temperature of the mold down to 180°F (82°C) before water was applied to the mold surface.

Figure 3 shows an external temperature profile for a thinner part. This again was an aluminum mold approximately 3/8” (9.5mm) thick. The part was 0.125” (3.1mm) polyethylene with an oven temperature of 600°F (315°C). The peak temperature at the end of the oven cycle was 432°F (222°C) after 10.5 minutes. This part was delayed in the first cooling station due to demolding problems at the unload station with parts on the arm two ahead of this mold. By the time it entered the cooling station it was already at 185°F (84°F). Water could have been applied to the mold at this point but since the machine has a preprogrammed cycle and no way of knowing if the mold is cool or not, the normal cooling cycle was applied. The part could have been removed from the cooler some 10 minutes sooner than it eventually was.

Consistency of Data

Consistency of data from cycle to cycle is most critical for this system since it’s intended function is to allow the machine to control the molding cycle using pre-determined temperature targets. Figures 4 and 5 show two sets of data gathered for two groups of molds and overlaid to show the variability in data that can be observed.

Figure 4  IRT Data Gathered for 7 Consecutive Cycles for Parts 0.200” (5.1mm) thick in an Aluminum Mold 36” x 28” x 10” (0.9 x 0.7 x 0.25m) at 600°F (315°C)

Figure 4 shows a series of seven cycles for a polyethylene part of 0.200” (5.1mm) wall thickness in an aluminum mold overlaid to show the repeatability of the data from cycle to cycle. Some variation can be seen in the oven cycle with a bandwidth of approximately 15°F (8°C) – this is most likely due to the background radiation from the oven burner system. However, the band is consistent. At the start of the oven cycle, the range of mold temperatures is from 158 - 175°F (70 - 79°C), a range of 17°F (10°C). At the end of the oven cycle the range of mold temperature is 493 - 507°F (256 - 264°C), a range of 14°F (8°C). During the cooling cycle, the curves are even more consistent. This mold had a five minute cycle step in the first cooling station after which time the range of temperature at which the mold entered the second cooling station was 322 -328°F (161 - 164°C). There were no delays holding the mold up. At the end of the cooling cycles, the range of mold temperatures upon exiting the cooling station was 112 - 120°F (44 - 49°C).

Figure 5  IRT Data Gathered for 8 Consecutive Cycles for Parts 0.120” (3.1mm) thick in an Aluminum Mold 28” x 18” x 16” (0.7 x 0.5 x 0.4m) at 600°F (315°C)  

Figure 5 shows a series of eight consecutive cycles for another set of molds running on the machine at the same time as those in Figure 4. This part had a 0.120” (3.0mm) wall thickness and was also running in an aluminum mold. A similar pattern can be observed in the oven curves for this mold. The band across the oven temperatures is approximately 18°F (10°C). At the start of the oven cycle, the range of entry temperatures is 160 - 178°F (71 - 81°C), a range of 18°F (10°C). At the end of the oven cycle, the range of exit temperatures is 429 - 448°F (220 - 231°C), a range of 19°F (10°C). This mold, however, had a much shorter cycle than those on the other arms and as a result experienced longer delays prior to entering the cooling station. This means that the range of entry temperatures at the start of the cooling cycle is 183 - 282°F (84 - 139°C), a range of 99°F (37°C). One cycle in particular was held up for an excessive time due to demolding problems. It would have been possible to activate the water cooling cycle immediately had the control system been fully operational. The exit temperature range on the cooler was 79 - 104°F (26 - 40°C), a range of 25°F (14°C).

IRT Data vs Directly Measured Data

Consistency of data is important from a production stand-point. The relationship between the external data and the internal temperature of the part is also critical in determining relationships for part quality control. Figure 6 shows how IRT data measured for a group of four aluminum molds compares to temperatures measured directly using Rotolog^TM. Three thermocouples were used to record the environment surrounding the mold, the external surface of the mold and the internal air temperature within the mold. Figure 6 shows a good correlation between the external surface temperature and the IRT data. The emmissivity setting for the sensors was set at 0.95 which is close to the measured surface emmissivity for the aluminum molds tested in this study.

 Figure 6  IRT Data Gathered for Four Aluminum Molds 12” x 12” x 12” (0.3m x 0.3m x 0.3m) 0.150” (3.8mm) Thick Parts 600°F (315°C) Oven – 1.5 minute Delay into Cooler  

The relationship between external surface temperature and the internal air temperature profile is obviously a complex dynamic one. The thickness of the mold and the part will dominate the time it takes for heat to transfer during the oven and cooling cycles. In particular, delays in cycle allow the part and the mold to equilibriate in temperature. This means that the effect of air or water cooling for a given length of time is different depending not only upon the actual temperature that the mold surface enters the cooling bay but also on the gradient present through the mold/part combined wall. More work is being carried out to investigate this.

Figure 7 plots data for the internal temperature vs the external surface temperature for a series of molding cycles with forced delays. The cycles were set to allow measurements to be taken for parts that had rotated in the first cooling station for 10, 15 and 20 minutes. The graphs shown are the temperatures measured just as the water was applied to the mold (the application of water to the mold surface causes a dramatic temperature drop that is not immediately conducted through to the internal air – comparisons after this point will require more analysis). The graph shows a trending down of both the external and internal temperatures. The curves are diverging but can clearly be seen to have a relationship that can be tracked. This type of data requires further study but will assist in the development of logic for the machine control system.

 Figure 7  Comparison of IRT Surface Temperature vs Internal Air Temperature for a series of molding with Forced Delays between Oven and Cooler Stations  

Effect of Water Spray

The infrared systems used here operate at wavelengths between 4.5 and 9mm. Atmospheric absorption, particularly by water, can occur between 5 and 8mm. This means that the sensor and target (mold) should be as close as possible to prevent the signal from being degraded. During these trials, water was applied in both a fine mist and also as a water shower. The molds are no more than 12 feet (3.6m) from the sensor at any time. Observations have not shown a reduced signal or masking of the temperatures measured. The temperature curves follow those measured by direct means as shown in Figure 6. 

Effect of Emmissivity

Emmissivity is a measure of how effectively a surface can emit and absorb infrared radiation. It has a value between 0 and 1. Dull, rough surfaces have high values and emit radiation well – hard, shiny surfaces on the other hand have lower values and are less effective at emitting radiation. Measurements taken on a range of molds in use at Remcon Plastics using a reference plate of known emmissivity showed that steel molds with dull, oxidized surfaces had values in the range of 0.958 – 0.984. Aluminum molds also had dull surfaces and were in the range 0.932 – 0.969. Stainless steel had the hardest and shiniest surface and had a range of 0.768 – 0.905.

Since emmissivity will vary from mold to mold, and indeed may vary over time, the system is not aiming to produce absolute values of temperature for every mold. Instead, target values are established for each individual mold. This means that if the sensor is set for an emmissivity of 0.95, a reading of 400°F on a dull steel mold and on a stainless steel mold will in fact mean that the surfaces are at different temperatures. However, the objective is to allow repeatability so that the shape of the curve and cycle to cycle consistency are more important than absolute values.

Conclusions

A number of conclusions can be drawn from this work:

1.      Data can be gathered using infrared sensors from the surface of moving molds in a fashion that allows control data to be available continuously during rotational molding.

2.      IRT surface data can be used to correlate with temperatures inside the mold so that relationships between the properties of the material and other physical properties such as size can be developed.

3.      Unique targets and set-points can be established for molds such that the operation of the machine no longer depends upon timers but upon actual mold temperatures.

4.      The effect of delays in cycle, changes in ambient temperature and mold changes can be minimized by using surface temperature measurements to control part cure levels.

5.      More consistent demolding temperatures can help reduce variation in size of parts.

Work to date has shown great potential for an IRT system to allow molders to improve repeatability in terms of cycle characteristics and end-points. Beyond this basic application, the system offers the possibility of developing intuitive machines that can act directly upon the measurements taken at the mold surface. Controlling the rate of heating to achieve a pre-determined heating rate, for example, or adjusting the level of water and air used to control the rate of cooling applied to a part are now possible. Giving a machine information on the state of each arm relative to the others may allow the machine to better balance cycle progress by speeding up or slowing down individual arms. The possibility of intuitive controls is a tantalizing one.

Trials with the system are on-going and future plans are being considered on how to incorporate it most effectively into machines for the next level of testing. For more details on the IRT system for rotational molding, contact Ferry Industries Inc. of Stow, Ohio. 

References

1.      Nugent, P.J., Little, E., and Peev, G., “The Use of Non-Contact Temperature Sensing in Extending Process Control for Rotational Molding”. ANTEC 97, Toronto, April1997.

2.      Nugent, P.J., “A Semi-Automatic Rotational Molding Machine?”, Association of Rotational Molders, Spring Conference, Maui, Hawaii, March 1998.

*    - Patent Applied For

**  - Rotolog^TM  is a trademark of Rotosystems Ltd., Northern Ireland – for more 

         information on this system, contact Ferry Industries, Inc.

[ Home ] [ Up ]

Send mail to email@paulnugent.com with questions or comments about this web site.

Copyright © 2001-2011 Paul Nugent
Last modified: November 1, 2011