Computer control in rubber processing.

Automated control of polymer processing machinery has become an integral part of the challenging need to meet improved product quality and increased productivity. This increased awareness has become more pronounced with processing machinery manufacturers, for which the term process control, when applicable to the rubber mixing field, takes on a new meaning.

The rubber mixing field has often been referred to as an art as opposed to a science. This expression reflects an industry that was forced to mature with a minimum of test equipment and a maximum of human experience. Even in this field, process control has been utilized ever since the first machine was designed and patented in 1916 by Fernley H. Banbury. In these earliest applications, the mixer was operated using time as the major control factor. Time and other important factors such as observation and sound which, although crude, were used to produce an acceptable product. The ram telltale rod was used to assess the correct batch weight, and the expulsion and ingestion of air around the ram clearances would be indicative of good mixing and batch completion.

In combination with the increasing use of the mixer, and with the inherent gain in human experience, users began to add vulcanizing agents to the mixes which necessitated the need for an accurate measure of stock temperature. With the introduction of the thermocouple within the mixing chamber, time and temperature were to become the major control parameters. Further developments were made as a result of the changing philosophies of major companies in building plants near the regions of rubber supply and demand centers. This was aided by an expanding world economy. With the changes in climatic conditions between one location and another, increased raw material suppliers on both a local and a global scale, and the inherent differences in feedstock characteristics, further developments to produce uniform stock on a consistent basis yielded power consumption/energy/work as a third major control parameter. Other enhancements used to produce a uniform product included: water temperature control, variable speed drive trains and infinitely variable ram pressure.

Added to the above developments, significant strides have also been made in the fields of metallurgy as well as manufacturing techniques. Mixers are now tailor-made to exacting standards for specific applications. Rotor development is a dominant subject in the quest for improvement. Banbury mixers are now equipped with four-wing, ST rotors which utilize rotor tip cooling for enhanced heat transfer characteristics. During the past 20 years, significant strides in the electronics industry have impacted the operation of the present day mixer. The evolution of the electronics industry has resulted in mass storage media, speed of execution, reliable and smaller hardware and cost effective control systems. The modern day control of the mixer incorporates the very latest of these sophisticated electronic developments.

Experimental

Batches of rubber were mixed in a production environment using an F-50 mixer with four-wing, tip-cooled ST rotors and a computerized control system. All ingredients were commercial materials and were used as received. The dynamic shear modulus G' was measured using a high response servohydraulic test machine known as a Dynafast. This machine subjects a double-lap shear specimen to a sinusoidal shear motion at frequencies from 1 to 40 Hertz and from strains of [+ or -]1% to 25% amplitude and measures the resulting force. From the two signals, motion and force, and from the geometry of the sample, the moduli were calculated. Fourier analysis was used to extract the fundamental frequency component from both the measured motion and the force signals. The reported value for G' was measured at 10 hertz and 10% shear strain.

Discussion

Mixer controls and instrumentation: A manufacturer's point of view

When discussing process control of batch mixing equipment, the words uniformity and consistency are frequently mentioned. The characterization of any processing operation involves two basic components. The first is an understanding of the fixed mechanical configuration of the equipment, and the second is the set of variable operating conditions of the equipment. While it remains possible to alter the fixed mechanical configuration of the mixer to change its performance, in an endeavor to attain the product uniformity and consistency goals, it is both costly and time consuming for the user. Developments in the mechanical configuration area have been left to the equipment manufacturers to attain these goals. It is the operation of the process variables which become targeted for meeting the defined goals.

As an OEM supplier of mixers and control systems, Farrel's control concepts are based upon two fundamental guidelines. The first is to retain a simplified concept for ensuring the mechanical repeatability of a given mixing operation to ensure that mixing cycles are duplicated for all batches. The second is to provide the operator with additional inputs concerning process variables, which can be utilized in meeting the overall objectives of mixing to a given uniformity and consistency.

Batch duplication, excluding the obvious control of all auxiliary and peripheral equipment, includes the mixing-related functions of the control of the ram (up/down/float) and its regulated air supply, the discharge door mechanism, the tempered water supply, the control of motor speed and the control of feeding equipment. These are all functions which, although their discrete set-point values may change between differing formulations, are generally considered to be repetitive for all mixing cycles. However, it is when they occur (and possibly at what levels), that provides clues to meeting uniformity and consistency targets. Some of the inputs which are available to the operator on current control systems include: Solids, powder and liquid feed; ram pressure and position; machine and mix temperature; discharge control, power draw and rotor speed. These inputs represent additional information, on which decisions for future mixes could be based to meet uniformity and consistency criteria.

Ram position/pressure and batch weight

Historically, the industry has maintained ram pressure at a fixed or constant level throughout the mixing cycle. This has usually been set to a high enough level as to facilitate rapid ingestion of rubber bales and other ingredients. In order to attain ever increasing productivity requirements, mixers over the years have been installed with an ever increasing size of the air cylinders. In this respect, where a facility might have the same size mixer - but differing size cylinders - it is pertinent to talk of ram pressure in terms of an effective batch pressure.

The topics of how an internal mixer works and why it mixes have been the subjects of numerous investigations at both industrial and academic institutions around the world. Although innumerable studies have described the mixing phenomenon in terms of dispersive and distributive criteria, based upon mathematical models formulated from recordings of the pressure and temperature profiles or rheological assessments, all the studies are in overall agreement to maximize the material flow characteristics in order to attain product uniformity. In evaluating this phenomenon of flow dynamics, the variables of batch weight and the effective batch pressure have been shown to be significantly influential in dictating the extent of mixing. Taken to extremes, an overly large batch weight prevents flow attributes by minimizing the free volume within the mixing chamber. On the other hand, too low a batch weight does not allow for the build up of high shear stress conditions to facilitate the dispersive criteria. A similar analogy is usually drawn for ram (effective batch) pressure. It is apparent that both of these variables are important when attempting to meet the goals of uniformity and consistency and should be optimized.

Modern day control systems have been tailored to provide the operator with a means for optimizing mixing parameters by allowing automatic adjustments in the ram pressure settings/evels. Together with the developments of the mixer manufacturers to facilitate the rapid ingestion of the materials, instrumentation in the form of a precise ram position measurement is now installed as standard equipment. This instrumentation controls the ram pressure in a number of ways following the initial ingestion phase (which can now be readily determined by the ram reaching a predefined point of travel). Using predetermined or empirical data, an automatic adjustment to a lower preset value can be made. The lower preset value can also be based on the product's viscosity. A soft formulation, generating lower shear characteristics compared to a harder formulation, requires less ram pressure to maintain the flow dynamics within the mixing chamber. Consequently, softer formulations require lower effective batch pressure compared to harder formulations.

Figures 1 and 2 show test findings from innumerable investigations upon different Mooney viscosity formulations studied within the F80 mixer equipped with the latest ST rotors. The information in these figures represents a starting point upon which to build an experimental data base. Figure 1 is a summary of the effective batch pressure requirements for formulations of differing Mooney values, and shows the differences in effective pressure requirements between one step and final mix formulations as opposed to masterbatch and remill stages. With this information, figure 2 is utilized to determine the fill factor level which represents an optimum value for maintaining the desired shear and flow characteristics

[CHART OMITTED]

Another way in which a control system can adjust the effective batch pressure level, following the ingestion phase, is to allow for continuous changes to the pressure value rather than a single set-point. Many users have based this feedback logic on an idealized ram position trace, resulting from observations that the mix formulation, as mixing progresses, undergoes a change in the density. As the formulation increases in density (specific gravity), the free volume within the mixing chamber increases - and this is reflected by a steady lowering of the mean ram position level. Control logic can now be used to adjust the effective batch pressure such that the ram position trace (with mixing time) provides an equivalent operation for all the batches. The pressure level is prevented from exceeding a point where the constituents are forced into the mixing chamber in the early part of the cycle, and pressure reduction during the latter stages allows for the upward movement of the ram to maintain the desired flow characteristics.

Machine temperature control

Mixers are now utilized with a water tempering system to further provide stringent control of the mixing cycle. Although the application of a cooling media formed part of Fernely Banbury's original patent, significant advances in the finite element modeling field have resulted in the redesign of the cooling passages. Cooling passages are larger and closer to required surfaces, resulting in improved heat transfer characteristics without the loss of structural integrity. This has been achieved for a mixer's three major working components: the mixer sides, discharge door and rotors.

A closed system water tempering unit provides stringent control of the metal components and virtually eliminates the first batch effect observations. The selection of water temperature settings is becoming more a science in today's industrial environment, and provides an atmosphere of enhanced mixing characteristics. Although different for the different classes of rubbers (e.g. butyl, nitrile, natural), the selection of the temperature settings is also made to reflect the changing coefficients of friction between the rubber and the metal surface. Enhanced mixing traits can be attained by careful control of the metal surfaces such that there is a tendency for the material to stick to the mixer sides, and yet slip on the rotor surfaces. Such aids to the mixing process have been shown to significantly increase overall productivity gains, in addition to minimizing power requirements. An additional use of enhancing slip-stick characteristics is to eliminate previous observations of hang-up with certain formulations.

Actions initiation basis

The principal parameters of time, temperature and energy - all gained through experience - are retained in most modern day control systems to initiate not only the discharge criteria, but also other actions that are required during the mixing cycle. Examples of these include: load, sweep and automatic ram lift. Time, originally explained as the total mix time since the start of the mixing cycle, is now commonly defined as ram down mixing time. As such, it represents a timer denoting the interval that the formulation was maintained within the mixing chamber under an applied (air) pressure. This is of significance when evaluating mixer efficiency and calculating power consumption.

The temperature of the stock is most commonly measured in today's mixer from two thermocouples. Both thermocouples are located through the apex of the discharge door. Thermocouple developments have resulted in more robust designs with improved response time. The location of the thermocouples within the door, in addition to their protrusion depth, represents another of the fixed mechanical configuration developments by machinery manufacturers. Usage of two thermocouples now provides the control system with an ability to average stock temperatures and provide a back-up in case of failure.

The energy term reflects the amount of mechanical work performed by the rotors in incorporating, dispersing and distributing the formulation constituents into a homogeneous product. In theory, the coversion of mechanical energy into heat by passing the material over the rotor tips should be uniform for a given product. It represents, after all, the enthalpy plot for the product. However, changes in feedstock characteristics and operating conditions such as batch weight and mixer temperature level can be readily ascertained by this value.

In addition to the above three parameters, others are also used to a lesser extent within the industry. In some instances, a measure of the amperage draw or power has formed a basis for discharge criteria (once some of the above three parameters have been fulfilled). This is an intuitive attempt to denote the mix viscosity based upon prior experience. The amperage draw is a reflection of the torque required to rotate the rotors at a given speed and as such, the level of torque changes with the rotor speed. However, changes in the reading are also reflected by changes in batch weight, stock temperature, ram position and applied ram pressure. Other reported discharge conditions have utilized the number of rotor revolutions as a count following a specified action (e.g. the addition of the cure package) on which to initiate the action. However, once again, minor changes to the operating system can upset the targeted requirement.

Mixer controls and instrumentation: A user's point of view

With instrumentation and controls of the nature just described, the end user has the ability to tailor more complicated mix procedures than would be feasible with a manual method of mixing. Each step of the procedure can be initiated on time, temperature, energy, or any combination of the three. While an operator would have difficulty in monitoring all three conditions separately or in combination and performing the appropriate operations at exactly the right time, the computer control system has no problem accomplishing such a task. This frees the operator to concentrate on the ingredient addition and sweep operations. In repetitive operations, a mixer operator will make mistakes occasionally even in a simple mix procedure. Computerized control allows the procedure to become more complex in order to optimize mixing while making the operator's job simpler and less confusing

We have found that variable speed rotors are extremely important in developing a robust mix procedure. As little as 5 rpm can significantly affect the consistency of the mix. To optimize a particular mixing step, rotor speed can be adjusted. Table 1 shows the results of mixing the same compound to a specific temperature (105[degrees]C), using a simple upside-down mix procedure with the curatives added as a second addition. Both mixes have the rotor speed set at 35 rpm for the initial charging of ingredients. The first mix uses 30 rpm for ingredient incorporation, 30 rpm for curative incorporation, and 35 rpm between the final sweep and the dump. This mix required 141 seconds of ram-down time to complete the mix. The second mix has 5 rpm added to both of the incorporation steps and between the final sweep and dump to make it 35, 35 and 40 rpm, respectively. This mix required only 63 seconds of ram-down time, less than one half that of the previous batch. With the rotors under computer control, there is no concern that the operator will forget to make the specified change during a given step. In addition to changing rotor speed with each step, it is also possible to alter the ram pressure on any given step. These adjustments are representative of the many which are impractical with a machine operated in the mannual mode.

[TABULAR DATA OMITTED]

There are several conditions where readjustments in the rotor speed can facilitate optimized mixing cycles. When adding dry fillers, the batch loses some of its cohesiveness until the powders are incorporated. Similarly, when adding plasticizers alone, the batch quits mixing efficiently until the plasticizer has been incorporated and ceases to merely lubricate the outside of the rubber. For either case, one can prevent the batch from going to sleep by increasing the rotor speed until a temperature rise signals the computer that incorporation is complete, at which point the rotor speed can be reduced to a normal range to prevent excessive heat build-up.

Table 2 shows an example of a formulation and mix procedure for a soft compound which is highly loaded with oil. The black and polymer were mixed together first to achieve dispersion, before the viscosity had been greatly reduced by the oil addition. While most of the mix is controlled to temperature, a maximum of 120 seconds was added to the first mixing stage as a secondary criterion. This was done in the event the temperature failed to rise because the mixer had such a low fill factor at this stage. The ram pressure was reduced after the plasticizer addition to prevent forcing it around the ram. Once the oil was incorporated, the ram pressure was increased, and the rotor speed was increased further to aid in dispersion. The ram was then raised one final time to give the batch a chance to turn over prior to dumping. While the mix was controlled by temperature, one can analyze the data to determine reasonable time and energy constraints on each individual step. Table 3 shows that it is not uncommon for batches mixed at widely separated times to be practically identical throughout the entire mix profile when the final time and temperature are similar.

Table 2

                         Formulation
                                                  PHR
               Polybutadiene                     100.0
               N300 series carbon black           55.0
               Napthenic oil                      60.0
               Stabilizers and curatives          15.0
                              Mix specification
Step  Time   Temp           Energy  Rotor  Ram    Instructions
 No.  (sec)   ([degrees]C)  (KWH)   (RPM)  (MPa)
  1      0     41            0.00    50.0   .00   Load black,
                                                  polymer
  2    120    87.8           0.00    45.0   .27   Load oil and all
                                                  other
  3      0    93.3           0.00    50.0   .07   Sweep
  4      0   110.0           0.00    60.0   .27   Ram raise
  5      0   118.3           0.00    40.0   .27   Dump

[TABULAR DATA OMITTED]

Process control charts (also referred to as Shewhart charts) are graphical and analytical tools for deciding whether a process is in a state of statistical quality control. The concepts underlying the control chart are that the natural variability in any process can be quantified with a set of control limits, and that variation exceeding these limits signals a special cause of variation. The X axis for the control charts shown in figures 3 through 6 are successive batches in a time sequence covering approximately ten months. The first process control chart (figure 3) shows that the mix temperature which is the only criteria for ending the mix has natural process limits ([+ or -]3 standard deviations) which span a range of [+ or -]1.5[degrees]C. Allowing ram-down time to vary without constraints produces natural process limits of [+ or -]25 seconds (figure 4) and the energy input (figure 5) shows natural process limits of [+ or -]1 kwh. The consistency of the mixed rubber is measured using the dynamic shear modulus G', a test which is far more sensitive than durometer or tensile modulus. Our normal procedure is to mix batches and then to blend those batches to achieve modulus control of [+ or -]8%. The control chart (figure 6) shows that G' for this compound is in control with natural process limits ([+ or -]3 standard deviations) as mixed of [+ or -]4.8%.

[CHART OMITTED]

Checking XY plots of energy versus G' and time versus G' shows that there is no correlation between G' and either time or energy over the range of the mixes which were made. Therefore, we could place upper limits on the drop conditions on either one or both of the conditions of time and energy. For example, the mix specification could be modified by placing limits of a maximum of 120 seconds on the ram-down time and a maximum of 8 kwh on the energy, in order to further control mixing conditions. Each step in the mix procedure can be individually analyzed and tolerances added for time and energy using the same techniques. Incremental improvements such as these become part of the process of continuous improvement as enough data become available for statistical analysis.

While there has been a great deal of improvement in mixers and control systems in recent years with the advent of the computer age, there are still shortcomings. Because mixing is typically a batch process, and the testing to determine the acceptability of the batch takes place long after the batch is out of the mixer, true statistical process control (SPC) is currently impossible. The control systems of today are designed to repeat a series of actions at preset conditions until a target condition is achieved. While computers excel at duplicating steps precisely, current systems are not capable of monitoring the mix and making changes to the rotor speed and ram pressure in order to keep the overall variation to a minimum. In short, the next logical improvement would be a computer control system which can learn and then initiate actions based on that learning.

Future developments - Heuristic control of the mixing cycle

The above defined batch process, once the process has been optimized, depends upon the ability to duplicate the operating conditions between one batch and the next to meet corresponding results between the batches. Although much of the industry currently uses this duplication approach to maximize on consistency, one limitation to this approach lies in the potential for variations in the feed material supply. With the control systems to-date, these have relied upon off-line characterization of the mixed compound to be indicative of meeting either an on spec or off spec criteria. An ideal control system would be capable of providing uniform and consistent results irrespective of the feed material inconsistency.

This idealized approach can be achieved by making the mixer a viscometer. Process control can be based not only on the conventional parameters of time, temperature or energy, but also on viscosity. In this regard, instrumentation manufacturers are rapidly gaining ground in providing devices that measure major rheological characteristics of the mix, while the batch is still mixing.

The continuous mixing industry, in particular plastics processing, has had this capability for quite soem time, whereas the rubber mixing field has remained in its developmental phase. This is a result of the former operating under so-called stabilized controls of the process variables. Real time process control, a requirement for the batch application, is not as straightforward in batch mixing as it is continuous systems since the controlled parameters never usually stabilize. Therefore, it is their rates of change - rather than their steady state values - which need to be forced to comply with some predetermined, or adaptive, criteria. Anticipatory logic has to be used to steer the appropriate parameter-time profile to achieve a desired response.

With the principal variables for steering the direction of the mixing cycle being rotor speed and ram pressure, one manner in which this anticipatory logic can be formulated is to store information which denotes how the rates of change (of energy incorporation and temperature rise) varies as a function of these variables. Once stored within the processor memory, real time feedback control can be employed to meet the criteria of time, temperature and/or energy. It is envisaged that energy can be used as a true process variable between batches such that the differences between feedstock characteristics will be eliminated from the discharged product. Each batch can be mixed to differing levels of energy but still be mixed to the same time and temperature - the precise target energy end-point dictated by the rheological instrumentation.

As a manufacturer of batch mixing equipment, Farrel is presently involved with instrumentation suppliers for generating on-line rheological assessments of the batches being mixed. As a precursor to the logic requirements, factorial experimentation designs (which are well known within the rubber mixing field) have been successfully employed in generating polynomial expressions which have been used to alter these rates. These designs are used in favor of screening variable methods as they provide both a statistical evaluation of the measured data in terms of confidence levels, in addition to providing an insight into any synergistic interactions. Although included in this article under the subtitle "future developments," figure 7 represents the surface response denoting how the rate of change of stock temperature varies as a function of rotor speed for an SBR/black masterbatch step. Similar response plots are also generated for the rate of change of temperature with ram pressure, in addition to the rate of change of energy incorporation with both these independent variables.

[CHART OMITTED]

With the instrumentation still under development, the writers were able to test the logic by requesting three consecutive batches be mixed to three differing mixing times for the discharge criteria - while maintaining the same discharge (stock) temperature; 2 minutes and 155[degrees]C, 3 minutes and 155[degrees]C, and finally 4 minutes and 155[degrees]C. In these investigations, this feedback logic was initiated when the stock temperature reached an arbitrary 85[degrees]C (185[degrees]F), a level which coincided with an indicator suggesting all ingredients had been added to the chamber.

Shown in figure 8 are the results of the measured stock temperatures drawn as a function of the mixing time. For the mixing cycles with the two lowest mixing times (2 and 3 minutes), the actual discharges occurred within some 5 seconds of the entered values The rotor speed range selected for this mixer (F40 designation which has a nominal 40 liter free volume capacity), upon whose range the polynomical equations had been derived, was sufficient to provide a steady increment of stock temperature with time. The actual rotor speeds utilized are shown in figure 9. For the final batch, with the discharge criteria of 4 minutes and 155[degrees]C, the actual discharge occurred at some 17 seconds before the stipulated time parameter. If we compare the actual rotor speed determined by the polynomial equations, the logic dictated that the minimum speed allowed (of 35 rpm) was utilized throughout the complete feedback process. This is indicative that the time parameter entered was of too long a duration for this formulation, but the system response was to attempt to attain the target criteria. Under such conditions, the inevitable bells and whistles could be programmed into the software logic to alert the operator of an inconsistency. In these demonstration runs, this represents a 100% increase in mixing time, which corresponds to a doubling of the rate of temperature rise. This would be indicative of a gross inconsistency in the infeed stock characteristics.

[CHART OMITTED]

One other benefit of this logic, and again provided as an example, is that the system becomes capable of eliminating the first batch effect. With the above responses having been generated with the circulating water temperature at some 40[degrees]C (sides), the response of the control system with deliberate settings of 20[degrees]C for a cold machine and 60[degrees]C for a hot machine are shown in figure 10. The results demonstrate that there is a natural tendency for the mixer to utilize lower rotor speeds as the mixer metal components heat up.

[CHART OMITTED]

Conclusion

In this article, the writers have summarized by way of historical evolution how the mixing concepts of time, temperature and energy have formed the basis for the control of the mixing cycle. Together with the developments of the machinery manufacturers and the improvements in rubber chemistry/formulations, the mixer operator has rapidly become deluged with a list of process variables. The operator must select those variables which will result in a product that meets designated specifications. Without the use of the modern day control systems and their capability to mimic operator responses, it is impossible for mixer operators to duplicate actions/decisions on a batch-after-batch basis, let alone batches mixed months apart, as demonstrated in the experimental work here.

With the impending development of additional instrumentations which provide a quality assessment while the batch is still mixing, the operator will be further deluged with inputs requiring an immediate response. Furthermore, these developments will dictate that not every batch is mixed the same way as performed by current control systems. Machinery manufacturers are currently developing the systems of tomorrow which will incorporate adaptive control logic to attain batch to batch consistency and uniformity. Examples of these systems have been included in this presentation to demonstrate how the mixing cycle operations performed since the days of Fernley Banbury in 1916 have changed to the present day requirements. The field of rubber mixing is rapidly becoming a science as opposed to the art with which it has grown up.