How to Collect and Interpret Process Data in Extrusion, Part 2

Originally titled ''

In this second of three installments, the focus is on the crucial element of melt temperature.

Many modern extrusion lines are equipped with data-acquisition systems (DAS). These systems can present extruder operators and process engineers with a large amount of data. There are several critical issues in the proper use of a DAS:

1. Are the correct process variables measured and monitored?
2. Are these process variables measured correctly?
3. Is the data collection rate appropriate?
4. Can operating personnel properly interpret the data? 

Part 1 of this three-installment feature (in January) discussed the proper use of a DAS in extrusion. This article will address issues related specifically to the polymer melt temperature. This is the second of the three vital signs of the extrusion process, as discussed in Part 1, the others being melt pressure and motor load.

Melt temperature is usually measured with an immersion melt-temperature probe. The sensing stem of the probe should be immersed at least 5 mm to provide a good measure of melt temperature. Some melt-temperature probes have a stem with adjustable depth.

These adjustable melt-temperature probes can be used to measure melt temperature at different locations across the depth of the channel. It is good practice to completely retract the sensing stem before shutdown to minimize the chance of damaging the melt-temperature probe when restarting.

Immersion melt-temperature probes have some drawbacks. Since they protrude into the melt stream, they alter the velocities in the melt stream. This, in turn, will change the viscous heating and temperatures of the melt stream. Therefore, the immersion probe will measure a melt temperature that is different from the melt temperature that would occur without a probe immersed in the melt stream.

Another drawback is the fact that an immersion probe will create some degree of stagnation downstream of the probe. This can create problems in thermally sensitive materials like rigid PVC and EVOH. A further drawback is the slow thermal response of an immersion probe. For this reason, an immersion probe cannot detect rapid changes in melt temperature occur- ring over a few seconds or less. This is a serious limitation of immersion probes because rapid melt-temperature changes occur in every extrusion process.

Another method of measuring melt temperature is by measuring infrared radiation (IR). Infrared thermometers are readily available and relatively inexpensive. Inside the extruder or the die, the melt temperature can be measured through a transparent window.

Another way to measure the melt temperature with IR is to use a probe with a sapphire window at its end in contact with the molten polymer. The transparent window is mounted in a threaded stem. The window is made out of sapphire so that it can handle high temperatures (up 600 C/1112 F) and pressures (up to 1500 bar or about 22,000 psi). The probe fits in a standard pressure-transducer hole, making installation quite easy. These probes are commercially available. The threaded probe with the sapphire window can be used in combination with a simple IR thermometer. This allows IR melt-temperature measurement in the extruder barrel or in the die.

IR temperature measurement provides two important benefits: One, the response time is fast, about 10 millisec. Two, the measurement does not disturb the flow, unlike the case with an immersion probe.

Because of the fast response, the IR measurement can detect rapid changes in melt temperature that cannot be detected by an immersion melt-temperature probe.

MELT-TEMPERATURE VARIATION
Melt temperatures in extrusion result from viscous dissipation and heat transfer. In drag flow, the shear rates and viscous dissipation are relatively uniform. However, the heat transfer to the screw and barrel results in large temperature differences within the polymer melt. The largest temperature gradients tend to occur at the barrel surface.

The melt-temperature distribution can be determined by performing a non- isothermal analysis of the flow in the screw channel. In case of a non-Newtonian polymer melt, this is typically done using finite-element analysis (FEA). Figure 1 shows a color contour plot of melt temperatures in the screw channel of a 63-mm single-screw extruder running a 0.2 melt-index HDPE at 100 rpm. The vertical axis has been stretched about eight times to make it easier to examine the temperature distribution.

The top surface is the barrel, the bottom surface is the root of the screw, the left vertical boundary is the pushing flank of the flight, and the right vertical boundary is the trailing flank of the flight. It is non-uniform. The temperatures close to the barrel are relatively low because of the heat transfer to the barrel—the barrel cools down the melt that is in close proximity to the barrel. The temperatures at the pushing flight flank are low because the cool melt close to the barrel moves down along the flight when it reaches the pushing flight flank. 

At the root of the screw, the cool melt from the pushing flight flank moves towards the trailing side of the flight. If there is no heat transfer with the screw, the melt temperatures will gradually increase as the melt moves from the pushing flank to the trailing flank. The melt reaching the trailing flight flank moves up along the flight until it gets close to the barrel. At that point the cycle starts again. The outer region of the screw channel stays relatively cool because of the heat transfer from the melt to the barrel. The situation is quite different for the inner region of the screw channel.

The melt in the inner region remains in this region as it travels down the length of the screw channel—it will not come into close proximity with the barrel. The inner region is thermally insulated from the barrel by a relatively thick layer of polymer melt. As a result, the temperatures in this region rise higher than any other part of the channel—it is a natural hot spot. In Fig. 1, the temperatures in the hot spot are more than 100° C/180° F above the barrel temperature. 

When the screw discharges the melt, this melt will be non-uniform in temperature. A number of studies have examined the variation of melt temperatures in extruders. One such study was made by E. Brown and A. Kelly at Bradford University in England. In this work, a fast-response thermocouple (TC) mesh was used to measure melt-temperature variation using three different extruder screws. The TC mesh was located in the breaker-plate recess of the extruder barrel (Fig. 2).

Melt temperature was measured at multiple locations, as shown in Fig. 3. It is clear from this figure that the melt temperatures across the channel are quite non-uniform, with a low temperature of 190 C/374 F and a high temperature over 215 C/419 F. The melt temperatures vary not only across the melt stream but also over time, as shown in Fig. 4.

The melt at TC4 (radial location -10 in Fig. 3) drops from 215 C/419 F to 180 C/356 F within 4 sec and then increases again to almost 215 C/419 F within 2 sec. This is rapid and substantial melt-temperature variation. Interestingly, this melt-temperature variation cannot be detected with an immersion melt-temperature probe because of the slow response time.

In this study the melt temperatures were also measured with an infrared thermometer. The results are shown in Fig 5. This figure shows temperatures at three screw speeds: 50, 70, and 90 rpm. The temperature variation increases with screw speed. The variation in TC mesh temperatures is substantially smaller than in IR temperatures. Also, the IR temperatures show much faster variation as a result of the short response time of the IR measurement.

Figure 6 shows infrared melt-temperature measurements plotted vs. time for three extruder screws running at 90 rpm. The melt-temperature variation for the tapered and stepped screw is quite large, 15-20° C (27-36° F). The melt-temperature variation for the barrier screw is smaller, about 5-7° C (9-13° F) and displays a more regular pattern. The time between peaks corresponds closely with the screw speed—90 rpm corresponds to 0.67 sec/revolution. This means that the screw speed can be determined from the IR melt-temperature measurement.

The reduced melt-temperature variation in the barrier screw may be due to the barrier geometry. However, this screw was also equipped with a fluted mixing element in the metering section. Therefore, it is also possible that the reduced melt-temperature variation is due to the presence of the fluted mixer. The tapered screw and the stepped screw had no mixing elements; therefore, these screws can be expected to result in larger melt-temperature variation. 

These results clearly demonstrate that short-term melt-temperature variations occur in polymer extrusion. In fact, these variations are inherent in the extrusion process. They occur because polymers have very low thermal conductivity. As a result, conductive heat transfer is very slow. The time that is required for the melt temperatures to become uniform by conduction is very long—it can be measured in hours for production-size extruders. 

Typical residence times for the molten polymer in extruders are around 20-40 sec. The residence times are usually much shorter, by orders of magnitude, than the times required to achieve uniform melt temperatures by conduction. Therefore, melt temperatures in extrusion will be non-uniform in most typical extrusion operations—especially when the extruder operates at high screw speed and when the melt temperatures are higher than the barrel temperature. These melt-temperature non-uniformities are more pronounced on larger extruders where heat-transfer distances are greater.

When the temperatures of the melt exiting the extruder are non- uniform, the dimensions of the extrudate will become non-uniform as the extrudate cools down and solidifies. These dimensional variations result from non-uniform shrinkage. The manifestation of this problem in pipe extrusion is waviness of the internal diameter. The calibration process fixes the outside diameter of the pipe, but the inside diameter will vary if the shrinkage is non-uniform. 

ID waviness is a common problem in pipe extrusion, but the cause of this problem is not widely recognized. One approach to reduce the problem is the use of a large-inventory die. Such a die will have long melt-residence times and this will reduce the melt-temperature variation at the die exit. However, large-inventory dies have several drawbacks:

• They are expensive.
• They take up a large amount of space.
• They take a long time to heat up.
• Long residence times increase polymer degradation.
• They increase changeover times.
• They result in more scrap.

A much more effective solution is to enhance the mixing capability of the extruder. With proper thermal mixing, large-inventory dies are not necessary and the ID waviness can be largely eliminated.

DATA COLLECTION RATE AND FFT
From the information presented earlier, it is clear that rapid data collection is necessary for certain process variables such as melt pressure, melt temperature, and motor load. In order to capture short-term process variation, these variables have to be measured at least 50 times/sec. Unfortunately, this capability does not exist on many extrusion lines. 

A useful feature of a DAS is Fast Fourier Transform (FFT). This is a technique that determines the base frequencies making up a complex signal. FFT is very useful in troubleshooting because 

signals are often complex and difficult to interpret. This is because the extrusion process is influenced by 50 to 100 different factors. At any one time, 10 to 20 factors may have a measurable effect on the extrusion process. For that reason, process variables such as melt pressure, melt temperature, motor load, extrudate dimensions, etc. often show a complex pattern.

TRAINING
It is not enough to have good instrumentation and data-acquisition capability. It is equally important to provide training so that operators and process engineers can take full advantage of the capabilities of the DAS. This is an essential element in a successful extrusion operation. Some companies skip this critical step and, as a result experience only limited benefits from a good DAS.

Good instrumentation is critically important in extrusion. Without good instrumentation and data-acquisition capability, it is hard to:

• Know what is happening in the extruder.
• Control the extrusion process.
• Optimize the extrusion process.
• Know when something is wrong.
• Troubleshoot the extrusion process.
• Have a profitable extrusion operation.

ABOUT THE AUTHOR: Dr. Chris Rauwendaal is a well-known author, lecturer, researcher, entrepreneur, and consultant in the field of extrusion. He holds numerous patents and has written more than 200 articles and seven books related to extrusion, mixing, injection molding, and statistical process control. A Fellow of the Society of Plastics Engineers (SPE), he is the developer of the CRD, VIP, and ASM mixing technologies that utilize strong elongational flow to improve mixing in extrusion and molding. Rauwendaal also developed the HHT (high heat transfer) extruder screw designed to improve cooling in foam tandem and other extrusion operations. In 1990 he founded and is still president of Rauwendaal Extrusion Engineering. Contact: (530) 269-1082; chris@rauwendaal.com; rauwendaal.com

Related Content

How to Collect & Interpret Process Data in Extrusion: Part 1

The first installment of this three-part series focuses on proper use of a data- acquisition system.