Jim Frankland is a mechanical engineer who has been involved in all types of extrusion procesing for more than 40 years. He is now president of Frankland Plastics Consulting, LLC. Contact email@example.com or (724) 651-9196.
Screw Surging, Part III: Unfilled Discharge Section
Click Image to Enlarge
Partially filled screw discharge sections can result in erratic output. A screw backfills for a certain distance to develop the pressure necessary to overcome the resistance caused by dies and other elements beyond the barrel. If the backfill is not adequate, the screw output will be unstable, and even the smallest variation in output entering the discharge section will cause the screw to vary in its fill length.
In the last two columns we discussed surging due to feed-related issues and those related to melting limitations. There is a third type, caused when the discharge end of the screw is only partially filled.
This type of surge happens most frequently with two-stage screws when the second stage has far more capacity than the first stage. It can also happen when a screw is limited in feeding or melting, causing a partially filled metering section. For example, a poorly designed barrier screw can restrict the output of the barrier or melting section, leaving the metering section partially filled. Feed restrictions that occur when using high percentages of low-bulk-density regrind can leave even a well-designed screw with a partially filled discharge section. Very low discharge pressures also make such situations more likely.
Since there is always a die and/or other apparatus on the end of the barrel in a production situation, the screw necessarily backfills for a certain distance to develop the pressure necessary to overcome the resistance caused by these elements. The amount of backfill depends on the pressure, viscosity of the polymer, output rate, and screw design.
If the backfill is not adequate, the screw output will be unstable and even the smallest variation in output entering the discharge section will cause the screw to vary in its fill length. This sets up an oscillation, where the screw increases or decreases its backfill causing a change in pressure-generating capability, resulting in a temporary change in output. That, in turn, causes a change in fill length, and unless some change is made to increase the fill length, the output will have an almost perfectly harmonic surge forever.
This situation has frustrated many processors over the years because of its persistence. It does not show up as a variation in motor load, and its pressure cycle is usually a perfect sine wave. This is often seen today because of the wide use of melt pumps to reduce the pressure at the screw discharge. This reduces the fill length necessary to overcome the resistance of the forward apparatus and can send the screw into a harmonic surge that no amount of temperature adjustment will correct.
The solutions are first and foremost to increase the discharge pressure on the extruder. This can be done by adding screen packs when a melt pump is not used. If using a melt pump, simply increase the suction pressure. Alternatively, adjustments can be made to the screw design to better fill the discharge area for more fill length.
The fill length is pretty easy for a screw designer to calculate. All that’s required is the discharge pressure and the viscosity of the polymer at the discharge, along with the screw design. Often these data can be reasonably estimated to make the calculations, particularly on a new application where there is no actual data. By experience I have found that a fill length of less than two turns of the discharge section will result in some instability.
Dead Screw Talking
Screws don’t really die; they just pass away to the scrap yard. But you can learn a lot by performing an “autopsy” on your screw before you heave it. Examining the carcasses of worn or poor-performing screws can often uncover information that is difficult to determine with the screw still in service. I call this process “forensic screw design.”
WHAT THE ‘CARCASS’ SAYS
Screws are seldom removed when they are performing at the expected performance level. So become a forensic examiner yourself and carefully examine every screw when it is removed to determine the telltale signs of its demise. The information in the “carcass” will usually provide strong clues as to why there were performance issues and how they might be corrected.
Areas of high wear on the screw flights can reveal either a design issue or an alignment problem. If the flight has a burr on the trailing edge, it means the screw has a high, localized, and unbalanced pressure that is forcing the screw aggressively to one side of the barrel. This is causing the flight to gall with the barrel, and the flight is being distorted by the mechanical pressure. This usually occurs in an area that is plugging with solids, causing very high pressure on one side of the screw.
The solution would be more melting area in the design to relieve the plugging and pressure point, or a design that balanced the pressure better. These extreme pressure points can show up as surging output as these areas plug and unplug, as I mentioned in an my May column on surging.
If the flight has a burr on both sides, it usually indicates that the barrel is not properly aligned or is bent, and the screw flight is simply being crushed as it is forced to rotate in a space that is not concentric with the screw. Failure of the flight’s hard-surfacing material is always suspected to be due to a poor weld but may in fact be due to either severe abrasion against the barrel from a localized high-pressure area or flexing of the screw due to misalignment.
Wear on the flights or on a mixer at the screw tip indicates a bent barrel. Many times this is due to a large, unsupported weight on the end of the barrel or misalignment of the front support. Wear on the drive end before the start of the flights indicates that the feed throat is not aligned properly with the drive quill. If the wear extends farther on to the early flights, then both the barrel and feed throat may not be aligned with the drive quill.
Areas that have a blue tint to the screw metal indicate the screw is being subjected to a very high temperature in that area, probably exceeding 750 F, which will result in a broad temperature gradient in your extrudate. Again, that indicates that area is plugging with solids and being exposed to very high shear stress or is being rubbed against the barrel with such force that it is causing high frictional heat.
Heavy buildup of polymer on areas of the screw can indicate several things. It could suggest an area of insufficient polymer flow, allowing material to stagnate and degrade. Or it could indicate an extremely hot spot that causes polymer degradation. This could be caused by the screw design or by problem with a heater/thermocouple on the barrel. Areas showing several different colors of material buildup also indicate areas of low or stagnant flow. These can be corrected by increasing the flow rate in these areas—usually by reducing the channel volume. This is often an issue in the melt channel of barrier screws where the channel is too deep or narrow.
Flights or other elements that appear “washed out” indicate an attack by hard contaminants or fillers in the polymer. If the wash-out is localized, the wear rate can often be reduced by allowing for lower flow rates in that area. At extremely high velocities, contaminants or fillers work like an abrasive paste, cutting their own clearance.
Where Does Shear Heating Occur? Here’s How to Find Out
Click Image to Enlarge
The highest shear stress and resultant polymer heating takes place just inside the inner surface of the barrel (red). Based on the heat flow shown for the screw/barrel section, it is apparent that a lot of energy that is being converted near the barrel wall can easily flow into the barrel, since it is a much better conductor of heat than the polymer.
Screws can be somewhat mystifying from outside the barrel because of their seemingly complex geometry and their interacting functions of melting, conveying, pressurizing,and mixing. One of the least understood yet most important concepts is viscous dissipation, which is the shearing or stretching of the polymer between the rotating screw and stationary barrel, causing heat to develop in the material. In single screws, as much as 90% of the drive power is used in this way to heat and melt the polymer.
The highest shear stress and resultant polymer heating takes place just inside the inner surface of the barrel, as shown by the red area in the accompanying illustration. If you consider the heat flow shown for the screw/barrel section as illustrated, it is apparent that a lot of energy that is being converted to heat near the barrel wall can easily flow into the barrel, since it is a hundred times better conductor of heat than the polymer.
One way to get some clues about the location and magnitude of this shear heating is to temporarily turn off the barrel cooling and observe the change in zone temperatures. Some barrel-cooling systems are so effective they completely mask the heating effects that may be occurring in the system. The analysis would seem to be somewhat complicated by the viscosity decreasing as the polymer traverses the length of the barrel. However, that does not affect the validity of the test, as we are looking for zones that appear to be a “hot spot” and not absolute numbers.
If we observe a hot spot what does that tell us? Well, it says that a lot of mechanical energy is being converted to heat at that location. With some knowledge of screw design, we can determine if high shear stress would be predicted in that area and use that information to troubleshoot the performance of the screw.
One example would be a hot spot in the zone near the end of the feed section, the early part of the compression section, or the start of the barrier section. That usually indicates an area that is restricting the polymer flow from the feed section and could resulte from either over-feeding or compression that is too abrupt. That could cause a surge, a disruption in the melting pattern, and consume a lot of power that was mostly going into the barrel.
Another possible cause could be a mixing section that is being plugged with unmelted polymer. If that mixer is at or near the end of the screw, it can cause a non-uniform melt temperature in addition to adding significant overall temperature to the melt.
A third potential culprit would be a hot spot near the end of a barrier section, indicating that a lot of solids were still remaining to be melted at the end of the barrier section, necessitating a change in screw design.
In addition to helping analyze screw performance, knowing where the shear heating is occurring in your extruder can guide you to the best temperature profile. Since one of the main purposes of the extruder is to raise the temperature of the polymer, it makes sense to work with the equipment rather than against it.
Allowing the extruder to put the heat in (or take it out) where the screw is designed to do that will generally result in more stable operation and more efficient power usage. That’s not to say that such analysis will solve all your processing problems, but it is another tool to use in analyzing screw performance and extruder control.
The Truth About Barrel Heating
Barrel heaters have very little impact on melting, an aspect of polymer processing that is misunderstood by quite a few. I recently discussed processing with someone who has been in extrusion for almost 30 years, and he asked me if he would need greater wattage-capacity heaters for more output. Even he did not understand that most melting occurs through shear heating and is the result of the drive power being converted to heat via the screw.
The main purposes of the heaters are to melt the polymer that remains in the barrel at cold startup, to assist in forming the initial melt, and to “trim” the barrel temperatures for specific purposes such as improving feed rate.
Typical barrel-heating capacities are based on watts/in.² of barrel outside surface area, and they have little relationship to extruder output. The typical values used are 25-35 W/in.², depending on how quickly the extruder manufacturer wants to be able to reach startup temperatures. Smaller extruders have more barrel-heating capacity than larger ones because they have more barrel surface area in relation to their output.
There is a practical limit to how much wattage to put on the barrel. Too much wattage will tend to upset the desirable steady-state principle that’s critical to most extrusion operations by imposing an overwhelming heat flux. Also, excessive wattage can cause the polymer near the barrel wall to burn before the material down in the screw flights becomes fully heated.
The shear-heating energy that comes from the drive is similar to stirring a very viscous fluid rapidly with a paddle. It’s going to take a lot of muscle and energy on the part of the person stirring. Where does all that energy go? It’s converted to heat in the fluid and is termed “viscous dissipation.” To visualize this, think of bending a wire back and forth until it breaks. The area near the break will be hot because the mechanical energy used to break it ended up as heat in the wire. The more viscous the fluid the more energy it takes to stir it and the more heat that will be generated.
It’s the same in extrusion: The more viscous the polymer, the more energy it takes to turn the screw, resulting in more heat being transferred to the polymer. That’s one of the main reasons that different polymers require different screw designs and different amounts of drive power. It’s also why melt temperatures are different for different polymers. For example a 2.0 MI polyethylene is going to have a lower melt temperature than a 0.2 MI PE with the same conditions and same screw because of viscous dissipation. Finally, consider that the viscosity of some polymers is reduced significantly with increasing shear stress. That accounts for the more rapid rise in melt temperature with screw speed for a polymer that does not significantly “shear thin” than for one that does.
The most effective adjustment the screw designer has to control melt temperature is specific output or output per screw revolution. Essentially the less overall stirring and the less stirring time the polymer experiences, the lower the melt temperature. That explains, for example, why a long L/D extruder is not suitable for low output levels. However the screw designer always has to work within the limits of the available extruder, its drive power (torque), and a usable melt condition with no unmelt. This sometimes makes it impossible to obtain the desired result on every extruder.
Benchmark Your Extruders For Quicker Processing Fixes
I often visit operations where the extruders are operating well below the equipment’s original capability. Generally the output has degraded over a period of time due to a series of “one-time” production upsets that never got completely corrected. Not recognizing these and resolving them quickly can cost a lot of money. This issue should have the attention of management on a daily basis.
Once the decrease in output is recognized, one of the easiest ways to determine the cause is to reset the extruder to a “benchmark.” A benchmark should be established for each extruder and each product, and the written record should be kept for comparison anytime output falls below the norm. A good benchmark analysis should include every aspect of the extruder that can be observed and documented, such as which zones are calling for heat or cooling frequently, the temperature of the feed throat, the melt quality and stability, specific output, drive load, etc.
Even though some extruders are equipped with elaborate data loggers, these devices do not monitor all aspects of the extrusion operation. Establishing a true benchmark goes beyond simply printing out a “snapshot” in time from the data logger and requires some additional visual observation by an experienced person.
Since the moving parts of the extruder are made almost entirely of steel, it is not susceptible to sudden changes. I‘ve actually heard some say—half kidding, I hope—that their extruders have a “personality” and occasionally act up “out of spite.” Trust me on this: Extruders don’t have mood swings, and there’s a logical and scientific explanation behind every output change.
Some wear may occur over extended periods of time, but in most cases that takes years before it starts to affect output rate. Consequently there is no way to explain a sudden loss of output except that something else has changed. If the screw is turning at the speed shown on the benchmark, then it should have the output shown on the benchmark. It should also have the same melt temperature, amp draw, head pressure, and barrel heat/cool response and every other observable characteristic. By referring to the benchmark data, you don’t need to guess what the output used to be or how the extruder was set up.
Using the benchmarking approach saves a great deal of time in troubleshooting, and it can be done without a lot of physical work. Simply resetting everything to the benchmark conditions will often quickly reveal the problem area. In this analysis we’re not after how it works, but only what has caused the change in performance.
When I say, “Reset everything that can be possibly reset to the benchmark settings,” I mean everything, including barrel temperatures, downstream equipment temperatures, suction pressure or head pressure, polymer mix (virgin/regrind/additives), feedthroat cooling, screw cooling, screen packs, etc.
Once that’s done, look for any differences in melt temperature, melt quality, amp load or drive load, stability (surging) and temperature-zone response to get clues as to how they could result in a decrease in output.
For example, if you find the drive amps or motor load are lower than the benchmark level at the benchmark screw speed, you might first assume that not enough material is entering the screw or that the material has changed. Look for any of the following causes:
•Something stuck to the screw in the feed section.
•Obstruction in or above the feed throat.
•Change in flow of screw or feed-throat coolant.
•Change in additive package or regrind.
•Change in a zone’s heating or cooling cycle (particularly in the first zone), indicating a loose thermocouple, burned-out heater element, or stuck cooling valve.
•Lower temperature of the feed material.
This is only a partial list but covers some of the more likely causes of a change in drive amps or load. The point is to go back to a documented condition with every setting and then see what has changed to get clues as to what’s really causing the problem.
To correct a processing issue, some operators start changing every setting until they get the line to run again, even if they succeed at a much lower output. However, that chaotic approach makes everything so convoluted that it becomes difficult to see the real problem, as the new settings may be “fighting” one another and completely mask the original problem. Benchmarking often simplifies the search for a solution.
How Fillers Impact Extrusion Processing
The addition of fillers can significantly change the processing properties of a polymer. Refer to general melting-rate calculations for viscous dissipation in single-screw extruders (shear melting) and you’ll see this clearly. Viscous dissipation is increased with increasing viscosity, density, and thermal conductivity, and by decreasing specific heat.
Adding fillers to the polymer mix affects all of these parameters. The generation of heat in the polymer itself depends on its thermal properties, viscosity, and the conductive heat transfer to or from the extruder. By adding most fillers, the amount of energy required for viscous dissipation is greatly reduced in relation to the mass output.
Polymers are poor conductors of heat compared with mineral or glass fillers. Moreover, polymers require more energy to raise their temperature (specific heat). Additionally, most fillers have a significantly greater density than the common polymers, giving them more “thermal bulk.” Finally, addition of fillers usually increases the compound’s viscosity as compared with the neat polymer.
MORE VISCOSITY=MORE SHEAR STRESS
Increasing viscosity increases the shear stress in the extruder, which is the basis of the mechanical-to-thermal energy conversion in a single-screw extruder. Stated simply, much of the effect of fillers in a single screw extruder can be described by a combination thermal property called diffusivity.
Diffusivity is a good indicator of the ability to raise the temperature and the ability to transfer heat.
Diffusivity=Thermal Conductivity/(Specific Heat x Density)
The units in the SI system are:
m2/sec=W/(m °K) ÷ J/(kg °K) x (kg/m3)
Information on materials’ thermal properties can be found on the internet and a relatively accurate average of each thermal property for the polymer and fillers can be determined by simple proportioning of the thermal properties to the mix percentage.
As an example, a polypropylene filled with 40% calcium carbonate would have a diffusivity approximately 3.5 times greater than that of neat PP. This means it will take much less energy from the extruder drive and will transfer energy much faster than the neat PP in relation to the mass of the output. This means lower motor load and quicker, more uniform melting than neat PP at the same output (lb/hr).
Higher diffusivity has implications throughout the whole process. As noted, the extruder drive load will be lower at the same output and the melt temperature will be more uniform. Cooling will be easier for the same reason. Heat will go in faster and easier and will come out faster and easier.
MORE SENSITIVE PROCESSING
This all sounds favorable at first glance, but the ability to raise or lower the temperature more easily and quickly can make the process more sensitive. Compensation has to be made in the operating conditions, often requiring some equipment changes, when running highly filled polymers. Since heat transfer is greatly improved, the barrel-temperature profile becomes more sensitive, for example.
The flow through adapters and dies also becomes more sensitive; the greatly increased diffusivity of the polymer will cause it to change temperature and resultant viscosity much more quickly when passing through these areas. This can cause variations in flow, particularly when there is frequent and significant cycling of the heat zones.
This is particularly important in multiple-exit die systems or where the flow is spread out, such as in large sheet or film dies, and in profile dies having large variation in cross-section.
The characteristic of filled polymers to heat and cool more easily can result in variations in product quality such as distortion, surface finish, and geometry, requiring changes in tooling, cooling rate, and line speed when changing from neat polymer.
Estimate the magnitude of these effects when running filled polymers by the use of the diffusivity term as a method of categorizing different levels and types of fillers
How Much L/D Do You Really Need?
In the early 1960s, extruders typically had a length/diameter ratio of 20:1, and a machine with a 24:1 L/D was considered long. Since then, extruders have gotten longer, with the 30:1 to 36:1 L/D becoming the industry “standard.” Some extruders even exceed 40:1 L/D for special purposes like double venting, compounding, or high-speed processing.
What benefits does the additional length provide? Mostly increased output and improved homogenization.
Since the feed section stays approximately the same length, regardless of the L/D, the rest of the screw is devoted to melting and pumping. The deeper the screw channels, or the higher the specific output (lb/rpm), the more length you need to complete melting and develop the pressure necessary to push the polymer out the die. As designers reached these limits, extruders were built longer to handle the economic requirements to pump out more and better product.
However, there are actual limits on increasing output as L/D is increased. Usually these limits are due to the inability of the feed section to deliver more polymer. On smaller-diameter screws, that limit often is determined by screw strength. On small screws you can only go so deep in screw channels before the screw is overtorqued and fails. On larger extruders, the efficiency of feeding decreases as the channels get deeper until there is no further increase in output.
Two-stage screws benefit more with increasing L/D because about 4-6 D is consumed in the vent area, which contributes very little to melting or pressurization. For most applications, you’ll need a 30:1 two-stage screw to match the output of a 24:1 single-stage screw (see accompanying illustration).
Are there any disadvantages associated with longer extruders? Some polymers melt much easier and faster than others. Also, some processes typically have low head pressures, while others have much higher discharge pressures. Inherent viscosity differs a lot between polymers, and some shear-thin significantly while others do not (i.e., are more “Newtonian”). As a result screw performance is optimized at a variety of L/Ds rather than at any one standard L/D.
A screw that is too long for the overall processing situation can actually limit output. The limitation generally shows up as excessively high melt temperature that can cause polymer breakdown, color shift, loss of additive effectiveness, and plate out, to name a few issues. For a polymer that melts easily, the melting length should ideally be shorter, as excessively long transitions can actually reduce melting rate. The same is true of pressure development, as widely used melt pumps greatly reduce the need for long metering sections to handle the discharge pressure.
As a result, the tendency to buy longer and longer L/D extruders can actually penalize overall performance. Custom extrusion houses may simply have to live with this reality because they never know what they will be running next year, and a longer extruder has more inherent flexibility than a shorter one. But if you have a dedicated process, there can be self-imposed limits with an extruder that is too long.
Just like selecting the extruder size and drive combination, the L/D should be carefully evaluated. Everyone wants the most usable output from their extruder, but if the material comes out too hot or too degraded then the singular focus on rate is actually reducing the extruder’s capability. Data such as diffusivity, power-law coefficients, melting points, head pressure, viscosity, and crystallinity should be part of the evaluation process.
The Melting Precision of Barrier Screws
Barrier screws have a number of advantages, but the most important is the ability of the designer to pinpoint the location where melting is to be completed. In conventional screws, the typical melting pattern occurs in the compression section. The channel depth is gradually reduced in this region, forcing unmelted polymer outward, where it is subjected to high shear forces that cause it to melt.
However, in conventional screws the compression section must stop at some point to allow polymer to move forward to the metering section to create the desired output. As a consequence of this transition, some of the unmelted material is not subjected to high shear and is instead conveyed into the metering section unmelted. At that point any unmelt acts like an ice cube in water, drifting along and melting very slowly.
Polymers are poor conductors of heat. If the screw is long enough, polymer will eventually be completely melted by the heat conducted from the surrounding fully melted material. But if the screw is not long enough, some material will emerge from it as either unmelted or in a state where its temperature and viscosity characteristics are inconsistent with the surrounding polymer. This will result in erratic die flow.
Calculating the amount of unmelt escaping into the metering section is complex and difficult to ascertain with complete accuracy, since it varies with screw speed. The typical solution is to equip the screw with a high-shear mixer, such as the widely used Maddock-style mixer, located somewhere in the metering section. This device forces the polymer to pass through very tight clearances, imparting high shear to complete melting.
Although it effectively serves its purpose of completely melting the material, these types of mixing devices have several disadvantages. First, forcing unmelted material through a tight clearance requires pressure, and that reduces the screw output and increases melt temperature.
Secondly, such a mixer does not correct the non-homogeneous melt-temperature problem, since the already melted polymer also gets subjected to additional shear, raising its temperature along with that of the unmelted polymer. This is particularly true if the mixer is located at or near the end of the screw so that no further homogenization and residence time is available.
A properly designed barrier screw, on the other hand, allows for complete elimination of unmelt while not restricting output. The unmelted polymer can be contained in the solids channel until it is melted, with the melted polymer contained in a second melt channel. This allows for the completion of melting to be “designed in.”
Additionally, the thermal history of the entire melt is more uniform since it was all melted the same way. And since the melting was completed well before the screw discharge, the melt is thermally homogeneous to encourage uniform die flow. This design also eliminates the need for a mixer in most cases.
‘Wedging’ Can Cause Severe Screw Wear
Click Image to Enlarge
As polymer moves forward in the tapered channel in the melting section, the melting rate must be great enough to keep up with the screw’s mass-flow rate or the channel will be completely plugged with solid polymer at some point (top). This causes the pressure to build rapidly at the point of the wedge, resulting in a force that pushes the screw to the opposite side of the barrel.
In a column I wrote for the July 2011 issue, I discussed the value of conducting “forensic” examination of worn screws when they are removed for clues to design flaws. Severe wear areas are one of the most important things to look for when a screw is pulled, as they are often an indication of a design problem.
One of the most common causes of high wear is the effect of “wedging.” Wedging occurs when the transport rate of the screw exceeds the melting rate in the melting section of the screw. The melting section is the transition area between the feed and metering sections. This would apply to a conventional screw or a barrier screw.
Most extrusion screws are machined with what is called an “involute taper” in the melting section. An involute taper means the channel is a flat spiral rather than a true taper (or conical) spiral. As the polymer moves forward in the tapered channel, the melting rate needs to be great enough to keep up with the screw’s mass-flow rate or the channel will be completely plugged with solid polymer at some point. Since the screw channel is a spiral tapered ribbon, this plug will look like a wedge wound around a shaft.
It’s not hard to visualize that when the channel plugs with solids, it happens at a specific location and not along the entire channel. This causes the pressure to build rapidly at the point of the wedge, resulting in a force that pushes the screw to the opposite side of the barrel. This wedge is powered by the full force of the drive motor, which has enormous torque due to the gear reduction. Therefore, the force generated by the wedge is also enormous. The side force is so great that it will cause galling of the screw flights against the barrel on the opposite side of the screw. This results in what has been termed “adhesive failure” of the screw’s hard surfacing as it welds to the barrel under extreme pressure, and then is torn off as the screw rotates.
Crystalline polymers exhibit much more tendency to “wedge” than amorphous polymers. This is primarily because crystalline polymers do not soften appreciably until they reach their melting temperature. Also, they require additional energy at the melting temperature to break down the partially crystalline structure, meaning the temperature has to rise well above the actual melting point before they are able to absorb enough energy to break down the crystallinity and flow freely. This additional energy requirement is called the heat of fusion.
Amorphous polymers do not have any crystalline structure and simply soften gradually as the temperature increases. Consequently they are often soft enough to flow away from the area of a plug, reducing the pressure buildup, the resulting side force, and screw wear caused by adhesive failure or galling.
Some screws are made with what are called conical tapers, which tend to plug in a ring around the screw rather than in a more localized spot on one side. This still results in high pressure at the point of the plug, but since it tends to surround the screw, the pressure tends to keep the screw centered in the barrel, thereby avoiding the high side force and adhesive wear.
So why aren’t all screws made with conical tapers? The answer is that the machining cost is much higher, and the wedging effect can usually be mitigated by changes in the design of the screw, changes in operating conditions, or improved screw/barrel materials that resist galling. However the conical taper confirms the effect of the unbalanced wedge by distributing the localized pressure buildup. While more expensive, it is used on many very large screws due to their initial cost and cost of replacement.
The pressures that can be developed almost instantaneously by wedging are somewhat amazing. That’s because crystalline polymers are essentially incompressible and have a very high modulus right up to the point where they melt and begin to flow. Analysis of the worn areas on screws using simple beam-bending analysis and ignoring the strengthening effect of the screw flight indicates that the forces often exceed 100,000 lb on larger screws. This is enough force to bend the screw over a relatively short span, forcing it into the barrel even after the screw has worn many times the normal clearance between the screw and barrel. Depending on the area of the wedge it is likely that pressure builds to as much as 20,000 psi or even more.
The pressure is relieved almost as quickly as it builds. The high pressure accelerates the melt film elimination and therefore melting increases, allowing the plug to advance. However, it may plug again in a short distance, repeating the effect. This wear is usually found somewhere from the middle to the end of the compression section and often extends into the metering section because of the bending of the screw. Interestingly it can also be identified by a burr on the trailing side of the flight caused by the galling screw material being dragged across the flight as the screw rotates.
“Forensic” examination of screws can accurately identify wedging so that a cost-effective solution can be developed. Don’t discard or send screws in for rebuilding without a thorough examination.
Well, It Worked on Brand X
I see requests for screw designs or screw evaluations with essentially no other information than that which appears in the headline above. A screw is an important part of an extruder, but it is just that—a part. No matter what the screw design, or who your designer is, it cannot perform correctly without information about the surrounding parts of the extruder, what your process requirements are, and the characteristics of the polymer you will be processing.
What is the output of the screw design in the accompanying illustration? Even given all the dimensions, its output capacity is unknown until a lot more components are in place and a lot more information is available. Obviously, before a screw can do anything, it has to be enclosed in a close-fitting barrel, have a drive to rotate it with sufficient torque, heat to get the process started, and an apparatus to deliver polymer to the feed end. The differences in extruder hardware alone account for hundreds of screw designs. If the screw were a stand-alone device, only a few designs would be required—one for HIPs, one for PET, PC, and so forth.
To optimize screw performance, you first need to know what the performance target is in terms of output, melt temperature, head pressure, melt homogeneity, and stability. Those goals are compared against the available screw diameter, L/D, drive power, screw speed, polymer characteristics, and head pressure.
And there are many other considerations: Is the screw vented? Does it have a grooved feed throat, supplemental feed device, melt pump? What is being fed to the screw—pellets, powder, flake, or fluff—and how? The form of the feed material has to be considered to determine whether the screw can match the target output. Is the feed opening sufficiently large, and are there any restrictions prior to the throat, such as stuffers, crammers or feeders? Is screw cooling required? If there are additives going directly into the feed throat, what is their effect on feeding? For example, heavy mineral fillers often must be starve-fed to prevent them from segregating, which in turn affects the output and melting.
Under what level of head pressure is the screw going to operate? Head pressure affects output, melt temperature, and melt homogeneity. How do you estimate the head pressure on a new application without a melt pump? What is the desired melt temperature? The screw may be intended primarily for LDPE blown film or for LDPE extrusion coating; these are essentially the same polymer, but they have two widely different melt temperatures, requiring two different screw designs.
These are only a few of the considerations that have to be evaluated before a screw design is even started. Many years ago I tried to make a data sheet that would cover all of the considerations that have to go into a screw design but gave up when it exceeded 50 items. I figured no processor would ever take the time to fill it out. That leaves most of the research to the screw designer.
Once the design is under way, what type of screw would give the best result: conventional flighted design, barrier design, wave-type screw, or a decompression design? Does the screw need a supplemental mixer for distributing additives, blending different polymers, or to complete melting?
With those considerations resolved, the properties of the polymer are to be considered next. The physical, thermal, and rheological properties for that particular resin must be applied to each section of the screw to get a balanced design that will meet the target performance. This data is not always easily available and often requires hours of research. Occasionally the target cannot be met and must be reduced or the extruder changed in some way.
Without every bit of this information the design cannot be optimized for the application and will contain a lot of guesswork. Often the research is limited to, “It worked for Brand X, it will probably work for Brand Y.” Be cautious of screw designs for which no information is requested except for the screw’s envelope dimensions and polymer type. You’ll be getting someone else’s screw.
Water, Oil, Air, or None?
What type of barrel cooling is best for your extruder? Barrel cooling effectiveness is largely related to the characteristics of the cooling medium. That is, its heat capacity and, in the case of water, its heat of vaporization. A comparison of the three cooling media and those key characteristics appears in the accompanying table.
There is no one “best” barrel cooling medium, but there seems to be a shift in the industry toward air cooling. This has been prompted by the development of highly effective air-cooling systems that are many times more effective than the old systems, where air was simply passed over conventional heaters.
Water cooling has the advantage of being able to remove an enormous amount of heat very quickly. That’s because water has a high heat capacity and changes to steam immediately upon entering the heater coils. Water has a heat capacity of 1 Btu/lb-°F, but when it flashes to steam its heat of vaporization is 1050 Btu/lb. Although very effective at removing heat, water systems can destabilize the extruder even when using the most sophisticated PID controls with process-tuning functions.
Also as a result of their superior cooling effectiveness, water systems offer the operator the ability to overcool the barrel in individual zones, which can actually disrupt the melting pattern. Overcooling negatively affects the energy efficiency of the extruder. In severe cases, 30% of the drive energy can be removed by the cooling system. Not only is this a significant loss of efficiency, but the water then requires cooling for reintroduction into the process, requiring the use of even more energy.
Oil is a less severe cooling medium than water, as most heat-transfer oils have less than half the heat capacity of water. But more importantly, these cooling oils do not vaporize at processing temperatures, which helps to eliminate instability while reducing energy losses. However oil systems are more expensive than either water or air systems and add another medium that needs monitoring and maintenance.
Air cooling is very gentle compared with water, as the heat capacity of air is about one quarter that of water. However, with modern air-cooling jacket designs, their efficiency approaches that of water cooling. Air cooling also eliminates the piping and valving needed for both water or oil systems and results in a more streamlined extruder design. Naturally leaks are eliminated as well.
The disadvantage is that the hot air exiting the heaters is discharged into the process area. This can be an advantage in cold weather but not so desirable in hot weather unless the air can be vented outside.
IT’S NOT AN OVEN
Barrel cooling in general is overused by operating people, many of whom perceive the barrel as an oven—if you set the temperature right, you won’t burn the cookies. However, extruders don’t work like that; the primary heat source is inside the barrel from mechanical working rather than outside from the heaters. By raising the barrel temperatures you tend to keep the heat in the barrel, and by lowering them you tend to remove a percentage of the internally generated heat.
Removal of heat often can be advantageous after the polymer is melted, for reducing the melt temperature. However the effect is greatly overestimated by many operators. Removal of the internally generated heat can be a disadvantage in the feeding and melting areas of the extruder, where the purpose of the extruder is to melt the polymer by generating heat through viscous dissipation or shear.
All cooling systems have their pros and cons, but using no cooling at all puts an onus on the screw design that can only be met at a very limited range of conditions. With any screw design the melt temperature will rise with increasing speed and head pressure. Also, the melting area elongates as the speed increases. This can cause operator concerns that they try to counter with barrel cooling. When used judiciously, the loss of stability and energy efficiency can be minimal with barrel cooling.
Salt and Screw Cooling
Salt and screw cooling have something in common. A little improves the flavor but too much spoils the soup. The principle behind screw cooling is that by lowering the temperature of the screw the coefficient of friction between the polymer and the screw surface is reduced so the solids can slide on the screw.
This follows the traditional solids-feeding mechanism in which the polymer must stick to the barrel and slip on the screw, otherwise it will rotate with the screw and not move forward. In the case of polymers with low melting or softening points, such as EVA or polystyrene, screw cooling may be required for stable operation because of their tendency to stick to the screw at relatively low temperatures. This is also true of polymers that are typically preheated before being fed to the extruder.
So if a little cooling is good then it stands to reason that a lot is better, right? No. Excessive screw cooling can result in a skin of polymer building up on the screw root past where solids feeding has ended. This is not trivial, as it tends to reduce output and melting rate. In barrier screws, it also can restrict the flow in the melt channel, causing surging and reduced output and difficulty in purging. Moreover, it extracts energy from the early portion of the melting zone, where the purpose of the extruder is to add energy to the polymer.
During normal solids feeding in a single screw, the incoming polymer is typically at room temperature. It serves to cool the screw, keeping it below the melting or softening point of the polymer. When the screw is stopped with the barrel at operating temperature, this does not occur. Instead, heat from the discharge end of the screw is conducted back to the feed section until the screw temperature exceeds the melting or softening temperature of the polymer. This results in polymer sticking to the screw. Upon resumption of processing, the incoming polymer may not be able to dislodge the material stuck to the screw. This is commonly called a “screw collar.”
Most operators have encountered this situation and don’t want it to happen again; in many cases the screw has to be pulled and cleaned to completely resolve the problem. So operators tend to respond by using far more cooling than necessary. Some even believe this helps reduce the melt temperature.
When polymer is stuck to the screw in the solids-feeding area the screw will typically surge, output will be reduced, and air will often be entrapped in the extrudate. When these symptoms occur, the best fix is to feed chunks of polymer into the feed throat to try to dislodge the material. The size and amount of the chunks is proportional to the size of the screw, but in general use chunks as big as the screw can accommodate.
A 4.5-in. extruder, for example, might require a water bucket full of chunks to dislodge a collar. In many cases this works, but in severe cases the screw will have to be pulled and cleaned. Even small amounts of material stuck to the screw can be disruptive to processing.
The collar is usually not visible in the feed throat, as that is separately cooled. It most often forms just forward of the feed throat at the start of the barrel. New screws seem to have a greater propensity for collars when first installed; they have not built up a coating of waxes and other additives that reduce the tendency for polymer to bond to the screw.
The screw-cooling bore should stop at the end of the solids feeding zone. That location is a little hard to predict because all polymers melt differently depending on their thermal and frictional properties. Additionally, the onset of melting moves down the screw as speed increases. However, five diameters (5D) from the start of the flights is sufficient for most applications. In barrier screws (see illustration) the bore should stop well short of the barrier section. When it extends into the barrier section it can result in the problems noted.
As a rule of thumb, the water exiting the screw should be noticeably warmer than the water entering or you are using too much cooling water
What’s the Deal With High-Speed Extruders?
How fast is fast? If we are talking about extruder screw speed, how about 1800 rpm? That’s how fast some extruders are running today. With all we have heard about the advantages of running slowly for better control, does such blazing speed make sense?
There is certainly something to it, but results have been kept as guarded secrets by most processors using these machines. But high-speed extruders are not new. The first one I saw was in the 1980s, developed by George Kruder of HPM Corp., an industry pioneer. He built an extruder that ran at about 800 rpm using his “wave” screw design. It ran well, but its use was limited pretty much to color compounding.
Most of today’s high-speed extruders tend to have direct drive with no gearbox, and they are typically 40-50:1 L/D to provide sufficient melting area for the additional output. Outputs are truly remarkable and are almost directly proportional to the speed. For example, a 2.5-in. machine could put out 3000 lb/hr for HIPS or 2000 lb/hr for PP.
The actual performance of modern high-speed extruders goes against some of the long-held concepts of extrusion. For example, the linearity of output with screw speed is surprising. The screw is turning so fast it might be expected to block the flow of polymer into the feed flights, but that’s not the case. Because of the rate of solids feeding, trapped air can be an issue that must be dealt with by venting. Ability to vent might be expected to be compromised because of the short time that melt is exposed to the vent, but again that does not seem to be the case.
Extruders typically show greater and greater barrel temperature override as the screw speed is increased, but the special screw designs developed for these machines have made it possible for the heat to be absorbed into the polymer with little or no barrel override at very high speeds. It seems likely that the size of high-speed extruders could be limited because heat-transfer issues increase as the screw size increases. I have not heard of any larger than 90 mm being successful at those high speeds.
Polymers that exhibit greater non-Newtonian behavior (more shear thinning) have been the most successful to date. Polymers having more Newtonian behavior are much more difficult to maintain at realistic melt temperatures, and their use on high speed extruders is still in the R&D stage.
There are obvious advantages and some potential disadvantages to the high-speed extruders. The oldest of these machines have been in full service for just three to five years and again, the results have been kept pretty much under wraps. So the disadvantages have not been quantified.
Some of the advantages are:
•Reduced capital investment based on output.
•Reduced floor-space requirement.
•Improved electrical efficiency due to no gearbox losses and less surface area for heat dissipation.
•Generally improved polymer properties due to shorter residence time at elevated temperature.
•Reduced purging and changeover time due to lower volume in the screw.
Some of the potential disadvantages that I can envision are:
•Rapid screw/barrel wear.
•More complicated, precise, and polymer-specific screw designs.
•Reduced thrust-bearing life and increased thrust-bearing cost because of the higher speed.
•Need for specially designed motors (or use of 1:1 reducers with conventional motors).
Floor space is seldom a reason to consider high-speed extruders, as they are nearly as long as a larger extruder of equal output, though they are narrower. Capital cost savings have not been much of a factor to date; these machines require special motors, much larger thrust bearings, and special screws. Their value is in better mixing, greater energy efficiency, and lower polymer residence time. In short, high-speed extruders have proven benefits but they bring a number of potential problems with them. As R&D work continues, many of these issues will be eliminated or at least minimized in the future.
Get Smarter on Extruder Sizes
Knowing that customer requirements change continuously and output improvements are always being pursued, operating people nowadays tend to recommend a much larger extruder than what’s really necessary for the job and the capabilities of the downstream equipment. While getting an oversized extruder provides “just-in-case” flexibility, it can also result in much higher daily operating costs. Let’s take a look at why.
Both AC and DC drives have a higher power usage per pound of output at low speeds. DC drives are even more expensive to operate than AC drives at low speeds because of their additional disadvantage of poor power factor at low speed. The larger extruder increases thermal losses to the environment because the latter are proportional to the heated surface area of the extruder. This can be useful in cold weather to heat the plant, but it is a major cost factor in warm weather, particularly if it requires use of air conditioning.
The electrical requirements for heat-up and temperature maintenance are also proportional to the extruder’s mass. Heat-up time can double from one extruder size to the next. Although not usually a major cost factor, larger extruders also require more floor space. More importantly, they also require larger-capacity utilities with higher expenses associated with installation and maintenance.
Another important cost factor is the extra time it takes to start up and change over a larger extruder. This extra time results in lost production and generation of larger amounts of scrap. Even routine changeovers requiring purging are more expensive, as much more purge material is required to change materials or colors.
Naturally the capital investment expense goes up with machine size, and that has to be amortized into the production cost. Moving up one extruder size roughly doubles the installed capital investment. At equivalent outputs, larger extruders have more residence time and have more material inside, increasing the likelihood of degradation as well as the amount of degraded polymer.
Simply put, extruder size needs to be considered in the daily economics of producing extruded products. For example, operating a 4.5-in. extruder at 1000 lb/hr and a 6-in. extruder at the same output can result in a 50-100% increase in total operating cost, depending on the particulars of the equipment, electrical cost, number of changeovers, and polymers being processed.
Deciding on the right extruder size is also important when deploying the extruders you already have in place. Many times, I have seen little or no thought going into the selection of the most economical extruder. Recently I saw thin-gauge, narrow PET sheet being run on a 6-in., 34:1 extruder with a 500-hp drive. The screw speed was 36 rpm with a maximum of 125 rpm. Sitting nearby was an older 4.5-in, 30:1 extruder with 200 hp that would have done fine. But the processor preferred the new line because it was easier to operate. Here the hourly operating cost was approximately twice what it could/should have been.
Picking the right size extruder is pretty much a function of the output at a reasonable screw speed. The proper L/D ratio depends on the requirements for mixing, venting, and pressure generation. Horsepower is a function of output and the polymer being processed and can be easily calculated. However, this can be a rather complex matrix of choices, and decisions should be left to experts.
My advice is to develop an hourly operating cost for each extrusion line and try to fit production to the least expensive line as a part of the scheduling procedure.
Mission (Nearly) Impossible: Estimating Extrusion Melt Temperature
Click Image to Enlarge
The drive is the main source of energy to the extruder. Power is lost through parts of the cooling system, the environment, and mechanical losses in the drive and gear reducer. Accounting for all this energy movement is nearly impossible without test data from the extruder in question. And without the data, it’s impossible to accurately calculate melt temperature.
Processors often ask screw designers or manufacturers to estimate the output and melt temperature of a new design in the works. Both bits of information are important to the success of the new design. The projected output of a new screw can usually be estimated fairly accurately using a limited amount of information supplied by the purchaser.
The melt-temperature estimate, however, is another story. It’s far more complicated to determine and generally can only be approximated based on comparisons of similarly designed screws processing the same polymer. The problem with estimating melt temperature accurately is not related to the screw, but to the lack of information on the extruder in which the screw will be installed.
Remember this principle: The screw is not a complete processing system. It interacts with the rest of the extruder, and unless extensive data is collected on the exact extruder in question, melt-temperature estimates are just that—estimates.
Melt temperature will vary depending on a long list of variables. Some of these are: barrel wear; head pressure; heater placement; heating and cooling capacity of barrel heaters; barrel liner material/thickness; thermocouple placement; temperature controllers; polymer thermal, physical, and particle properties; screw speed; feed-throat and screw cooling; accuracy of output estimate; alignment; incoming polymer temperature; and operating environment.
The extruder is a complex thermal device that takes energy from the drive and converts it to heat to melt the polymer through a combination of friction, viscous dissipation, and conducted heat. Smaller additional amounts of power are required for solids and melt conveying. Power entering the system through the barrel heaters is generally a minor source compared with the drive. Power is lost through parts of the cooling system, the environment, and through mechanical losses in the drive and gear reducer. This is depicted in the accompanying illustration.
Accounting for this energy movement is nearly impossible without test data from the extruder in question. And it’s impossible to calculate the melt temperature accurately without such data.
Moreover, output and melt temperature vary with fluctuations in the polymer and head pressure. There is always at least some error introduced in the output calculation due to variations in the polymer, which affect the head pressure and melt temperature. Polymers can vary considerably among suppliers, even with the same general specification. Head pressure becomes a variable that is a “catch 22.” Even if you have logged head-pressure data at a specific output on a past run, don’t count on it being the same again. Head pressure will vary with any difference in the melt temperature.
Pellet geometry or the use of regrind will change output rates, thereby affecting melt temperature. Further, regrind can have substantial changes in rheology compared with virgin. So, to even estimate an accurate melt temperature requires data on the specific polymer being extruded, including any regrind. Just calculating where melting starts in a particular extruder is difficult without substantial test data. Naturally, that information is necessary to even begin a calculation of the final melt temperature.
Designers can utilize tools to perform more accurate melt-temperature calculations. Just keep in mind that such an analysis can cost more than the screw. It’s best to supply screw designers with the current screw design and complete operating conditions so that they can make the most accurate estimates. By modeling the performance of the current screw, an experienced designer can factor in many of the variables without collecting all the data.
The Power-Law Coefficient
The geometry of the screw makes shear heating hard to understand. Envision a flightless shaft turning in a tube. Visualize a very viscous fluid filling the space between the shaft and tube, and a slight bit of pressure at the right end. Since this imaginary material is very viscous it will take take a lot of work to turn the shaft. The slight pressure moves the viscous fluid from right to left.
The work by the shaft is converted to heat in the viscous fluid, known as viscous dissipation. The same thing happens with melted polymer in an actual extruder. The melt continues being heated by the rotation of the screw for as long as it remains in the barrel. The heat either stays in the polymer until it exits the extruder, or is removed from the system by passing through the barrel and cooling system.
It’s hard to cool polymers because they are poor conductors of heat. Polymers are, in fact, excellent insulators—their heat-transfer rate is less than 1% of that of steel.
Meanwhile, the rotation of the screw is adding heat continuously to the melt, offsetting what is extracted. Minor changes in barrel temperature have a very small effect on the overall withdrawal of heat from the system. As screw speed is increased, addition of heat from viscous dissipation usually overrides the ability of the cooling system to absorb heat, and the melt temperature continues to increase even with maximum cooling being applied.
Screw size is a factor as well. As screw size increases, the amount of polymer vs. the surface area of the barrel increases exponentially, so the effectiveness of barrel cooling is reduced.
Fortunately runaway temperatures are not the norm because polymer viscosity is decreased by both increasing temperature and shear rate (screw speed). As a result, most extrusion processes operate in a relatively narrow range of melt temperatures for a given polymer. That said, different polymers have different sensitivities to either shear rate or temperature.
Polymers can be described by the power-law coefficient, which is a simple relationship derived from the shear-rate/viscosity curves at different temperatures. It describes the viscosity in most of the processing range of the extruder.
Polymers have a consistency index (m) and a power-law coefficient (n) that describe its general viscosity behavior with respect to changing temperature (T) and shear rate (Ÿ). The consistency index is primarily the relationship between the polymer’s viscosity and temperature. Although useful in extrapolating the viscosity for design calculations, the consistency index is somewhat considered a dependent variable.
The power-law coefficient, however, has significant effect on heat generation, which in turn impacts the final temperature and resultant viscosity. This is important to understanding what can be expected for a polymer’s final melt temperature using a particular screw design. Polymers that have lower power-law coefficients will see a greater change in viscosity with changing shear. As the viscosity drops, the amount of energy required to rotate that portion of the screw will decrease and additional heat generated from the screw will be reduced. The accompanying table shows the consistency index and power-law coefficient for several polymer types, and the temperature range in which the values were established.
HDPE, for example, would be expected to reduce its viscosity by a factor of three compared with the viscosity reduction for nylon 66 over the same change in shear rate or screw speed. This means that the additional shear heating of HDPE would be less than for nylon 66 as screw speed increases.
By referring to the power-law coefficient, the effect of barrel override in the metering section of many screws can be explained and anticipated. Polymers with a high power-law coefficient would likely generate more heat in the melt with increasing shear rate through viscous dissipation. This adds another complexity to optimum extruder selection, wherein the ideal L/D can be influenced by the polymer’s power-law coefficient. In general, polymers with high power-law coefficients should be processed at lower screw speeds and on shorter L/D screws than those with low coefficients.
New Frontier for Single Screw R&D: Mixing & Melting by Extensional Shear
There are many types of single-screw mixers on the market, and they can be generally categorized as either distributive or dispersive, all having some degree of each. Distributive mixing is more “macro” mixing—that is, the components of the mix are equally distributed. Dispersive mixing tends to be more on the “micro” level, where high stresses are imparted to the lesser component through viscous shear of the surrounding polymer to obtain high levels of integration. For dispersive mixing, there is a second mixing mechanism to be considered besides simple shear. It is extensional shear, which generally has been associated only with twin-screw extruders.
However, some have recently promoted the concept in single-screw mixing. I was originally skeptical about the idea because I felt there had to be two moving forces to develop such mixing.
However, after meeting with Keith Luker of Randcastle Extrusion, Cedar Grove, N.J., I became convinced it can be done with a single moving force (screw) by changing the geometry of the flow field perpendicular to the direction of flow. An example of extensional shear is depicted in the accompanying illustration of fiber extrusion. Note simple shear in the die orifice and extensional shear where the cross-section of the fiber is being continually reduced through drawdown.
This is important because the viscosity of polymers in extension is approximately three times the viscosity in tangential or simple shear (Trouton’s ratio). The higher the viscosity of a material, the more effective it is to mix another material into it. By mixing in the extensional mode the higher (3X) viscosity greatly increases the dispersive effect, breaking down the lesser component with three times the shear stress. Additionally, extensional shear does not reduce the viscosity by shear thinning like simple shear does, so that the viscosity remains high even at high shear rates. This would be particularly effective for polymers exhibiting high non-Newtonian (shear-thinning) behavior.
Luker has taken this concept further with single-screw “mixers” that are intended to be both primary melting devices and intensive mixers). He uses extensional shear to increase the viscous dissipation through the viscosity multiplier. His tests show melting can occur in very short distances when extensional shear is employed in the design. This replicates the extensional shear melting that occurs in twin screws, where all solids can be melted in as little as one screw diameter.
Design of these mixers requires a lot of thought for application to single screws—there has to be a big change in shape of the melt flow in order to develop a shear field of any intensity perpendicular to the main flow field. It’s hard to visualize a screw geometry where the polymer is stretched in one dimension while compressed in another direction (as in the fiber drawdown illustration) without a pressure drop. And, extensional shear has to happen simultaneously with the simple shear occurring in the screw channel from the screw’s rotation relative to the barrel.
At this point, I can visualize doing this only with a pressure drop to change the melt flow’s shape independent of the simple shear from screw rotation. Both Luker and Chris Rauwendaal of Rauwendaal Extrusion Engineering, Auburn, Calif., employ a pressure drop in their designs to accomplish this. Pressure drops limit output and generate excessive temperature from pressure flow in the screw channels, so the addition of these devices will require changes in the entire screw design to be effective.
But properly designed extensional shear devices have the potential to reduce screw length while increasing the effectiveness of mixing and melting. If all polymer in the screw can be subjected to equal levels of extensional mixing, the potential for dispersive mixing is 3n times that of a traditional screw with the same shear rate (“n” being the number of passes through an extensional shear field).
In retrospect, I see evidence of elongational shear flow in many older mixers and screw designs, which may explain how some are surprisingly efficient at mixing and/or melting in such a small area. Maximizing elongational shear flow in single-screw extrusion may be the most significant step since the barrier screw.
The Evolving Barrier Screw
Click Image to Enlarge
Of the various barrier screw designs, the Dray/Lawrence screw shown in Fig. 2 offers a number of advantages over other types. For one thing, use of an essentially constant channel width from the feed section through the entire barrier section eliminates the continuous distortion of the solid bed necessary with the crossing barrier design (Fig. 1). However, no one solution fits all situations. The so-called “hybrid” design in Fig. 3 has proven effective for processing softer polymers and/or resins with lower energy requirements.
The use of barrier screws for plastics processing is now a well-established practice for both smooth- and grooved-bore extruders. At the same time there has been a gradual transition from the earlier Maillefer designs used in Europe with the crossing barrier design (Fig. 1) to the Dray/Lawrence design using the parallel barrier design with an increased helix angle to accommodate wider channels (Fig. 2).
The crossing barrier design starts with a full-width solids channel and ends with a channel having no width. The parallel design with an increased helix angle has been found to have several potential advantages over the crossing barrier type.
First, the use of an essentially constant-width channel from the feed section through the entire barrier section eliminates the continuous distortion of the solid bed that is necessary with the crossing barrier design. The pressure required to facilitate that distortion in the crossing barrier can be carried back into the solids-feeding section, reducing the feed efficiency and output.
Also, the depth of the solids channel in a crossing barrier often must be kept essentially constant in order to maintain sufficient transport area to meet the output requirement. This changes the melting pattern in a number of ways. Melting in tapered-depth channels is generally faster than in constant-depth channels because of higher shear rates in the melt film. Constant channel depth makes distortion of the solids bed, rather than reduced channel depth, the mechanism to force unmelted resin against the barrel wall. The continuous distortion increases the likelihood of a breakup of the solid bed, which would reduce the melting rate—particularly with hard polymers, which do not have a strong solid bed. Finally, melting rate is closely related to the area in which the unmelted polymer is in contact with the barrel.
By increasing the helix angle in the Dray/Lawrence type design, the channel is wider, allowing for introduction of the barrier and melt channel without restricting the width of the solids channel and the solid bed. Conversely, the crossing barrier design reduces the area exposed to the barrel in the solids channel of the barrier section to less than half as much as a parallel barrier of the same length. The only way to overcome that restriction is a much longer barrier section, which is usually not possible except with exceptionally long L/D extruders.
Screws having a parallel barrier but no increase in helix angle (Fig. 3), fall somewhere in between the Dray/Lawrence and Maillefer concept with respect to melting rate per unit length. However, they have worked well in applications for softer polymers and/or polymers having a low energy requirement. Their narrower channels can provide lower melt temperatures in many instances.
This is not a blanket indictment of either the crossing barrier design or the parallel barrier design without the increased helix angle. Many extruders are limited in torque, have sufficient L/D, and do not require a maximized melting rate because of feed limitations. As is usually the case with polymer processing, there is no single design to fit all processing situations.
The Limits Of Compression Ratio
You have probably heard the terms 2:1, 3:1, and so on in describing single screws. These numbers refer to the compression ratio (C/R) of the screw, which is the ratio of the feed depth to the metering depth. Most people use C/R as a method to select the proper feed depth of the screw, but it significantly impacts the melting rate as well.
While C/R should not be overlooked, there are a lot of other parameters to consider when designing the optimum feed and melting sections of a screw. The C/R does not tell you the channel depth or section length, just the ratio between them. Would you expect a screw with feed and metering depths of 0.900 in. and 0.300 in., respectively, to perform the same as one of the same size with depths of 0.450 in. and 0.150 in.? Or would you expect a screw with five feed flights to perform the same as one with 10 flights at the same depth? What if there is a pitch change between the feed section and metering section? These configurations can all have a 3:1 C/R, but for certain they will perform differently.
Feeding properties are hard to quantify and measure, so most screw designers necessarily use a largely empirical approach when making decisions. However, the best designers utilize a number of empirical relationships relating to the frictional characteristics of the individual polymer along with the particle characteristics of the individual application, such as bulk density, packing, and uniformity.
These vary by polymer, resin manufacturer, use of regrind, presence of additives or fillers, and feedthroat design. Screw designers will not rely on a single fixed ratio like C/R for each polymer, but one that varies to suit as many characteristics of the feed material and the feed-throat design as can be quantified.
The other aspect of C/R is that it affects the melting performance. That’s because C/R does not describe the rate of compression. For example, a 3:1 C/R will have a different compression rate if the compression section is five turns as opposed to 10. Different polymers—and even different grades of the same polymer—can require different rates of compression in order to optimize melting without plugging or breakup of the melting pattern.
The C/R alone does not provide that information. In addition, some polymers will experience a reduction in viscosity more than others under similar conditions of shear. This can require a different metering depth to control melt temperature, even though a target C/R has been met for the feed capacity. Consequently, the ideal feed and metering depth are seldom based simply on C/R.
Today many screws utilize a barrier section, which completely eliminates the need for any relationship between the feed depth and the metering depth. Since melting is controlled in the barrier section, the metering section can be designed to accommodate other aspects of the extrusion process, such as head pressure, melt temperature, and mixing, without affecting either the feed rate or the melting performance.
The use of “standard” C/Rs for certain polymers has resulted in many poorly designed screws. The shortcoming is usually in matching the melting rate to the polymer and output. A bad match can result in high screw wear, surging, poor melt quality, and poor mixing.
The Cause of Catastrophic Screw Wear
Click Image to Enlarge
FIG 1: Areas of the screw that are full will develop pressure from the wedge action because the solid polymer does not slide easily on the barrel wall. That results in unbalanced force on the screw, which is pushed in the opposite direction against the barrel wall.
I frequently come across catastrophic wear when I’m examining screws used in a recycling process. Recycling presents some issues not generally found in traditional extrusion operations that rely on pelletized polymers, even when these conventional processes use high percentages of regrind. That’s because the film, fiber, foam, or bottle scrap fed into recycling extruders often has low bulk density and non-free-flowing characteristics. Whether the extruder is fed by gravity or with a side feeder, crammer, or stuffer, there is always some inconsistency in the feed uniformity due to the erratic characteristics of these feedstocks.
Most processors intuitively know how much force a screw jack or even a bolt can exert, but they do not realize that the same principle applies to the extruder screw. The screw is actually an inclined plane or wedge that’s wound spirally around a cylinder in a helical form. This design creates an action just like a wedge and multiplies the force of the screw drive. For a standard flight pitch, the multiplier is about four times the torque of the drive, not taking friction into account.
The localized forces acting in a single-screw extruder are enormous when the polymer is still in solid form. When the screw is fed inconsistently, it is alternately full and partially full at different locations until compaction is fully completed farther down the screw. Areas that are full will develop pressure from the wedge action, because solid polymer does not slide easily on the barrel wall. That results in pressure in the polymer and an unbalanced force on the screw. This pushes the screw in in the opposite direction and presses it against the barrel wall with enormous force because of the small resisting area of the screw flight (see Fig 1).
Although these forces cannot be observed from outside the extruder, there is no question they exist. Otherwise, how could you explain a screw that is worn over a short distance and yet essentially unworn before and after that area. In order to wear in that pattern, the screw has to bend over that distance. I recently investigated a case in which a 6-in. screw was experiencing catastrophic wear in only three diameters near the end of the feed section (Fig. 2). I calculated the side force necessary to bend a shaft having that root diameter enough to permit that amount of deflection: It was about 60,000 lb. And that was assuming no bending support from the screw flights, which do greatly strengthen the screw in bending—meaning the actual force was much more.
There are two types of “wedging.” There is a distinct difference between the wear caused by erratic feeding and the wear due to insufficient melting capability. Wedging due to inadequate feeding is generally contained to the feed section and first turn of the compression section. On the other hand, wedging due to inadequate melting capacity can cause the compression section (barrier section) to temporarily plug with solids and cause an unbalanced force on the screw. That form of wedging is usually contained in the latter half of the compression section. Melt-limited wedging causes accelerated wear but generally not catastrophic wear, because the screw is filled and there is some support from the opposite side, depending on the degree of melting.
Some operating practices—such as manually dumping bales of film scrap into the hopper—really accelerate the wear as the screw alternatively runs full and empty, causing large moving side forces. Letting the screw run empty and then suddenly filling it at full speed will have the same effect. Less severe but still very problematic are the use of feed-assist devices such as crammers, stuffers, and side feeders that are not designed and/or operated properly.
If catastrophic wear is occurring in the feed section or early part of the compression section, consider stabilizing the feed rate into the screw. This may require additional polymer grinding or changes in design of the screw or any apparatus that assists in the feeding recycled polymer into the screw.
As a guideline, when the feed rate into the screw is erratic there will be evidence in continuous variation in the motor amps with a proportional change in head pressure at the same frequency. The forces are so great that the use of premium screw and barrel materials will only have a minor effect on the wear rate.
Understanding Screw Breakage
Click Image to Enlarge
Torque failures always initiate from the outermost portion of the screw—or the flight surface—and crack toward the center. As a result there are radial cracks in the flight surface around the break. Additionally the screw “winds up” like a spring during a torque failure, and the flight pitch around the break is permanently reduced.
Extrusion screws, even those smaller than 2 in. diam., very seldom break. But when they do, what are the causes? Your first guess might be that the screw was compromised due to over-torquing, but in reality bending causes more breakages than over-torquing.
Over-torque failures almost always occur in the feed section, which makes sense because that is screw section most susceptible to torque. Torque failures have characteristics that are easy to identify. They always initiate from the outermost portion of the screw—the flight surface—and crack toward the center. As a result you’ll see radial cracks in the flight surface around the break. Additionally the screw will “wind up” like a spring, and the flight pitch around the break will be permanently reduced (see illustration). As the screw winds up and cracks it also grows in diameter, making it very difficult to remove from the barrel.
There is no exact calculation for screw strength in torsion because of the complex or indeterminate shape of the cross-sectional area. But it can be estimated by assuming that just the root diameter carries the torque load, even though the flight adds to the screw’s torsional strength.
I have found that a safe design can be calculated using simple torque formulas and assuming a maximum allowable shear stress of 50,000 psi. That’s a little over half the yield strength of 4140 heat-treated steel, from which the vast majority of feed screws are made in the U.S.
The formula is: Ss=16T/πD³ where:
T= (hp x 63025)/max screw rpm=in.-lb
D= Diameter of feed-section root=in.
Ss= Shear stress (<50,000)=lb/in.²
This formula applies to a screw without a cooling bore. If a cooling bore is used the following formula works:
Ss=16 TD/π(D4-d4) (d=bore diameter).
The cooling bore has almost no effect on torsional strength because the stress is maximum at the outer surface and zero at the center or neutral axis where the cooling bore is located.
Screw bending or fatigue failures do not typically show radial cracks. These failures generally reveal themselves by a single crack that goes all the way through the screw. In addition, these failures don’t result in the “wind-up” effect or a pitch reduction.
Bending occurs either from the side forces the polymer exerts on the screw during melting, or from misalignment of the barrel. Bending by either mechanism is termed “reversed bending.” Think of the rapid failure caused when you break a wire between your fingers by flexing it back and forth.
In order for the internal barrel pressures to cause a screw break, the screw/barrel must first be worn enough to provide the clearance necessary for the bending to occur. This may take a long time, but once sufficient clearance is present, breakage quickly follows. I have covered this type of localized wear, called wedging, in previous columns (see links to these articles at the top right-hand side of this page).
The screw is also under torsional load from the drive, adding to the overall stress, since both bending and torsional stresses are at their maximum on the outer surface of the screw. With the combination of the reversible bending stress and torsional stress, failure will occur in very few cycles after the critical stress is reached at the surface. That is typically at about 40% of the normal tensile strength, meaning reversed bending reduces the strength to about half the steel’s specification.
Bending failures due to barrel alignment are similar but can occur sooner because it’s not necessary to wait for the screw and barrel to wear. Interestingly, bending failures usually occur in the strongest sections of the screw, because that’s where it takes the highest force to bend it, with the resulting highest stress at the flight surface. When a screw breaks in one of its thickest or strongest sections, it’s extremely unlikely that it was an over-torque situation alone.
The Scale-Up Conundrum
When designing an extruder screw, scale-up is one of those head-scratching areas. There are many methods used to determine scale-up based on output, but most designers do not scale the melting requirement along with the output. Instead, they tend to use a simple ratio for the flight depths and stick with the same number of flights in each section.
There are so many variables to consider when scaling up, and they all change at the same time, making manual calculations difficult to do. The interdependence of the variables doesn’t simplify scale-up all that much, even with computer simulation.
Cost and time issues have made it necessary for designers to develop a simpler method than full computer simulation for much of screw design and manufacture. There are some shortcuts that apply—assumptions about the distribution of energy going into the polymer, and a simplification of the classic melting theory into a bulk energy transfer that can approximate the melting rate with reasonable accuracy and in a reasonable time frame. But it still requires a good understanding of how the energy is distributed and of polymer rheology to get usable answers. Melting capacity requires knowledge of the polymer thermal properties, the mass flow, screw geometry, shear rates, the consistency index, and the power-law coefficient.
When increasing screw diameter, the output increases at a rate greater than the square of the diameter because of increasing channel depth with increasing diameter (assuming constant head pressure). However, the melting area increases proportionally to the square of the diameter. Simply using the same number of turns in the melting section for a scale-up in screw size results in restricting the melting capacity. The same is true when scaling up for greater L/D, or even when the screw is simply deepened for more output. This can become a problem leading to rapid screw wear, surging, and poor melt quality, particularly with semi-crystalline polymers.
Although melting can occur in all sections of the screw, the bulk of it normally occurs in the compression or barrier section. In conventional screws, some solids typically exit the compression section unmelted due to eventual solids-bed breakup, and that percentage has to be considered based on the polymer characteristics. Determining the correct percentage requires further assumptions and considerable trial-and-error experience. For barrier screws, your best bet is to assume all melting occurs in the solids channel.
Conductive melting is ineffective with polymers except in small-diameter extruders or at very low screw speeds, because polymers are excellent insulators and do not absorb heat readily. Consequently, it greatly simplifies the calculation in the melting section to assume any conducted heat flow in/out between the barrel and the polymer is minimal and constant.
This typically introduces only a small error, except in cases involving exceptionally aggressive barrel cooling. With semi-crystalline polymers, conductive melting is further restrained by the requirement for the breakdown of the crystallinity, or the heat of fusion. Amorphous polymers allow much more latitude in the melting-rate calculations because the polymers will often be soft enough to pass through restrictive areas of the screw below their typical flow temperature.
Most often, the barrel temperatures are lower than the melt film temperature at the barrel wall in the melting section because of the high shear rates in the film and over the flights. This reduces the melting rate due to absorption of energy from the film because of the higher thermal conductivity of the barrel. Consequently, actual melting rates are most often lower than calculated rates, necessitating some over-design.
In the absence of polymer training and data to calculate melting rates, a simple rule for scale-up is to maintain the ratio between the area for melting and the output for a particular polymer with a screw design that works well. If you increase the output 50% you should increase the melting area 50%. This requires only some simple geometric calculations of area and estimation of the new output versus the present one.
However, each polymer has a different ratio between output and melting area, so a separate ratio must be developed for every polymer for a new screw design or for troubleshooting. Consideration must also be given to the width of the screw flight and clearance, as shearing of the polymer in the clearance can account for 10% to 30% of the power entering the polymer. This is treated in the same way as the energy in the channel and is simply added to the power from the channel. Keeping the same proportion of flight width and flight clearance between screw diameters will avoid over- or under-affecting the melting rate.
There’s no single ratio that can simplify the determination of melting capacity of various polymers. But once yours has been established it can be used for scale-up, troubleshooting, and new screw design. It takes lots of time to collect and analyze all the data, but once completed it’s a big time saver.
‘General Purpose’ = No Purpose
So-called general-purpose screws are primarily used in injection molding, which makes sense because there are backpressure adjustments for melt quality in this process, and plasticating rate does not determine the cycle time.
But such screws have little value in extrusion, which is totally dependent on the screw design for both melt quality and output. If there really were a good “general-purpose” extrusion design there would be no further need for screw designers.
Various polymers may have a few similar properties. But in truth, the properties of polymers vary so much that screw designs intended for multiple polymers are usually at best a workable compromise compared with a screw designed for a specific polymer.
Each aspect of screw plasticating depends on many different properties to optimize the feeding, melting, and pumping characteristics, making for an extensive list. Unfortunately most polymer data sheets contain only a few pieces of information pertaining to screw design (such as melt flow and density). As a result, processors have little information relative to the suitability of their screw design to a particular polymer. Even melt flow is just an indication of molecular weight and does not describe the viscosity characteristics of the polymer as it is processed through the screw.
I’m not telling you that every polymer requires a different screw. There are some compromises that can be made, particularly with amorphous polymers, which simply soften as the temperature increases. This allows somewhat dissimilar amorphous polymers to be processed over a wider range of screw geometries.
Crystalline polymers, on the other hand, require more precise amounts of energy at specific locations to melt before they will soften and pass through the screw. As a result they require more specific screw designs, and attempts at using a general-purpose design often leads to significant compromises in performance. Yet some crystalline polymers of the same family of polymers—like acetals, nylons, and terephthalate-based polyesters—can sometimes be processed with good results on the same screw.
All that being said, the suitability of using the same screw with both amorphous or crystalline polymers should be determined by economic evaluation and not by convenience. Since screws are volumetric devices, the output of quality products from different polymers should be approximately the same when adjusted for melt specific gravity, or a different screw design should be considered.
However, some processors have a lot of short production runs, and changing screws frequently seems so time-consuming that they often process different polymers on the same screw despite any inefficiency. That’s generally not the best business decision, as operating the extrusion line at its maximum output and quality level is the essence of processing profitably. Other processors will run an extruder larger than necessary to compensate for the reduced output that occurs with the “one screw fits all” approach.
But changing screws, even large-diameter ones, does not have to be a hassle if the proper equipment, training, and preparation are applied. Devices like quick clamps, screw carts, hoists, powered equipment, and easily moved downstream equipment can make the job go very quickly. Unfortunately, this equipment is not available even in many of the best-managed plants, making screw changes very tedious. As a result, they are avoided even when appreciable savings can be realized in a single day.
To optimize polymer changeovers, an evaluation should be made on each extruder to determine the time and cost to change screws when everything to facilitate the job is available. Sometimes, machine modifications will be required beyond that provided by the OEM to accomplish that goal.
Generally, extrusion lines have not been designed for rapid changeover. Redesign requires study by a competent person thinking “out of the box” to come up with the necessary changes and equipment. Once established, the downtime for screw changeover should be compared with the cost of reduced production efficiency from using a less-than-optimal screw design.
When processors switch from one polymer to another, they usually make other changes in the upstream and downstream equipment/tooling, so the extruder is out of operation anyway. Consequently, little or no production time may be sacrificed by changing the screw as well, and the cost comparison is mostly the additional manpower to change the screw compared with the improvement in production efficiency.
Regrind and Melt Pumps
Processing large percentages of regrind into finished product always causes some problems with output stability. Today this is further complicated by the wide use of regrind typically purchased from various sources in an effort to reduce product cost and improve environmental impact.
Since regrind comes from many different products and is resized by any one of a handful of particle-reduction devices, the bulk density and feeding properties vary much more than with typical in-house regrind.
A typical screw can handle a range of bulk densities, allowing for low levels of regrind with minimal changes in performance. But when the overall bulk-density (virgin plus regrind) variation exceeds 20-25%, the variation becomes impossible for the screw to handle. The screw is a volumetric device and has only limited capability to adjust for wide variation in bulk density being fed. The feed-section channels have a specific volume; if the bulk density changes, the mass (weight) of polymer entering the feed section will go up or down with the bulk density.
Most screws are designed to process pellets, with some flexibility built in to accept some percentages of lower-bulk-density regrind. This is accomplished by using a compression ratio greater than that necessary for an all-pellet feed material. But this is a moving target, requiring knowledge of the bulk density and particle-flow properties of all the different regrind materials.
Make the screw too shallow, and it will not fill when running low-bulk-density regrind; make it too deep, and the screw will overfeed when running pellets. Either one can cause wear, surging, and issues with melt quality and instability. Surprisingly, things can get more complicated when adding a melt pump to the line.
Once the polymer is melted, it has a fixed density regardless of what its original bulk density was. So as the bulk density gets lower with regrind content, the melted resin may only partially fill the screw flights, affecting the output and stability. Since discharge pressure is proportional to output, the discharge pressure varies continuously. When a melt pump is employed, this causes the screw to change speed continuously, as the extruder screw speed is controlled by the pressure entering the melt pump, or suction pressure. As the pressure falls, the screw speed is automatically increased; and, conversely, as the pressure rises the screw speed is decreased to maintain a consistent pressure.
Unfortunately, there is no linkage between the polymer regrind entering the screw and the pressure at the discharge. It takes a certain amount of time for the polymer to traverse the length of the screw, and that varies with the screw speed and screw design. Consequently, the system can become even more unstable with the melt pump, as it is trying to correct the pressure out of sequence with the mass of polymer entering the screw.
If you need a melt pump in your process to control product dimensions and want to use varying regrind levels, you have to deal with this situation. There are several things you can do to minimize the instability. First, you need to “damp down” the response of the screw speed to varying pressure at the discharge. This can be easily accomplished by making a significant increase in the proportional band or gain in the PID controller that functions as the interface between suction pressure and screw speed. This allows rapid changes in bulk density to be overlooked by the controller.
Making a significant change in control parameters is best to start with, and I find using four times the current value to be a good starting point, as it will usually show results. For example, if the proportional band is set at 3000, change it to 12,000. Finer tuning can be done once the system is stabilized. In addition to the proportional band, there are a lot of other adjustments that can be made for further stability, such as band width, ramping, and filtering, all of which are built into modern PID controls.
The second thing you can do is raise the suction pressure. For a given screw design, the level of discharge pressure determines how far back from the tip the screw is filled. With varying feed rates, the fill will change constantly and the length of fill affects the output at that given instant. By raising the pressure, the fill is increased so that greater fill length results in smaller percentage changes in the discharge pressure. Again, these adjustments are best done with significant changes rather than incremental changes, so the result is quickly apparent. For example, an increase in suction pressure should be a minimum of 300 psi, with 500 psi probably being a best first move. Again this can be fine-tuned after stability has been regained.