Mixing efficiency and quality: a view from a synthetic rubber producer.

Mixing efficiency and quality: A view from a synthetic rubber producer

There are three contributing factors which, between them, determine both the efficiency of the mixing process and the quality of the mixed compound. These are:

* The quality of the raw materials used;

* The degree of control of the mixing process; and

* The test methods used to assess both raw material quality and the degree of mixing of the compound.

This article will deal with each of these topics, with references to the original literature where appropriate. There are a number of papers and texts (refs. 1-6) which are recommended as preliminary reading in order to gain a fundamental understanding of the mixing process. To the compounder or production engineer, developments in the understanding of the fundamental behavior of elastomers and compounds are only of importance if they can be directly and simply applied to improving industrial-scale rubber mixing. However. In the long term, a clearer understanding of the physical processes occurring in the mixer must enable better control and, therefore, optimization of the process.

Quality of raw materials

There has been a significant change in emphasis towards high quality and consistency of products in the rubber industry during the 1980s. This change was instigated by the end-user, especially the automotive industry, and was transmitted by the processor back to the producers of the raw materials.

The major raw materials of the rubber industry are, in addition to the rubber itself, carbon black, other fillers, oils and various chemicals. For each of the categories the processors have to establish the minimum requirements for suppliers. These minimum requirements are usually stated in such terms as "enough to ensure that suppliers have a quality system to assure that all materials supplied conform to all quality requirements" (ref. 7). This, in effect, means that the suppliers must have in place a quality system covering raw materials, processes and also production.

Many rubber processors today have in place a supplier accreditation system, or audit, whereby raw material suppliers' plants, processes, laboratories and quality systems can be assessed against objective standards. This type of close communication between suppler and processor generates confidence that the guaranteed quality is being provided and, eventually, leads to reduction, even elimination, of incoming raw material testing and minimization of inventories. It also enables development of specifications and standards that are both right for the customer, and feasible for the producer.

Specifications usually include minimum and/or maximum property limits, both for the raw elastomer and for specified compounds. There are also certain implicit, but unstated, properties that are understood by both producer and customer to be part of the agreement. An obvious example is that the product should be free of contamination by foreign materials.

The purpose of specifications is frequently misunderstood. They do not, in themselves, guarantee that the rubber will mix, mold, cure or perform satisfactorily in the customer's process or product. In many cases the test recipe bears little or no resemblance to the compound that the customer produces from the rubber. Equally, the tests that are performed on the raw rubber or compound, and which feature in the specification, often bear little resemblance to the actual processing or service requirements. The object or purpose of specifications, as applied to elastomers, is to guarantee to the customer that he receives that same product shipment by shipment. Thus the specification is, in effect, a benchmark or template. If the rubber meets the agreed specification then it can be expected to perform in processing and application a consistent manner. The specification is thus a measure of the consistency of the product.

The width of a specification range for a particular property is also important, and it must reflect the customer's needs and the manufacturers capability to produce and supply.

Modern quality thinking looks at specifications in a new context. The midpoint should express the most desirable value for a parameter critical to the customer. This is an expression of the product design. Any deviation from this value that is not test error represents a loss, and, as proposed by Taguchi, this loss is a continuous function. Specifications imply a precipitous loss at the specification limits which intuitively does seem to be illogical.

The objective should be to reduce variability continuously within the specifications. Consistent with this thinking it is now common practice to express the ratio of the specification range to the actual process variability as the "process capability." As variability improves within the limits the specifications limits really become goal posts for the monitoring of never ending improvement in variability. In our industry this concept is complicated by the relatively large error inherent in most test methods.

In the production of synthetic rubber to meet the requirements of the customer's supplier accreditation system, it is necessary to prove that the manufacturing process is stable, capable and centered on the targeted properties. In other words, the producer has to be able to make the required product consistently and economically. The best technique for improving the process control in synthetic rubber production is statistical process control (ref. 8), which has been described as a technique to make perceptive observers of all those associated with the process: operators, technical staff, management. A synthetic rubber producer's approach to SPC is well described by Francis (ref. 9) and, therefore, it is quoted directly.

"Polysar's initial approach to SPC was to view it primarily as a control technique. A skilled observer can tell when a significant event has hit the process, known as a `special cause' in SPC language. This can signal an operator when to intervene in the process. Although continuous processes have some unique properties, this is a powerful aid to process control.

"At first it was believed that the chronic special causes were already visible and well known to those operating the processes. Work would, therefore, already be going on to remove them. The primary application of SPC then should be for short term process control.

"We have now come to view SPC in a broader perspective. Detailed statistical studies have been made of 11 different processes in our rubber operations in North America. Significant events (or special causes) were present in all of them. This information, generated by data from the processes themselves, is often of a type that could be used to improve the processes or products if it were properly observed. What was significant was that in at least half the cases this information had previously been invisible to those operating the process.

"This does not mean that SPC should be done off-line solely by technical people. However, in the operator-use of SPC, emphasis should be placed on asking operators to identify the cause of instability and suggest how this can be eradicated. This same knowledge can aid the operator in process control.

"SPC and advanced process control (computer control) should not be looked upon as alternatives but as marriage partners. Attaining stability of the process should greatly aid in the benefits achieved by more sophisticated control methods. The experience of most companies is that bringing stability to processes can yield huge improvements. This type of quality improvement is not necessarily capital intensive.

"Assuming that the product is properly designed, the variability in critical process parameters is a very important criterion for our customers. Any deviations from the targets mean a loss to our customer, either in the performance of his factory or in the quality of his product. (Getting the right product design in the first place is another subject and one that can be dealt with very effectively by joint experimental design products with our customers)."

The variations caused by the raw materials and the manufacturing process for synthetic rubber combine to determine the product variations, which, in the processor's eyes, is his raw material variation. The problems involved in controlling this variation were described by Korosec (ref. 10) thus:

"Samples of the final product are tested on a regular basis, according to a predetermined sampling scheme, in order to obtain a measure of the total observed variation. The data are summarized in terms of an average (x), sample standard deviation (s) and sample size (n) each time a particular grade is manufactured (i.e. for each production run). A practical problem with this method arises when one production run must be designed to satisfy several customers, each with a different requirement, thereby causing the manufacturer to set different aiming points. In order to obtain an accurate estimate of the true product variation for the grade, the results surrounding each aiming point need to be analyzed separately, i.e. x, s and n for each separate target value within each run. Because of the large amount of data, it can become a difficult and time-consuming task to collate the results. However, microcomputers set up in each laboratory will facilitate this work. Also, it would be very beneficial if the manufacturer could convince the processors to agree to one common target point for each grade. This would help to eliminate a large part of the problem that many rubber manufacturers now face in reducing product variation.

"A time period is chosen that allows for sufficient data for each grade to be collated. Based on the results for each targeted value for the grade, a weighted average and pooled variance is calculated. The resulting average and standard deviation, [x.sub.o] and [s.sub.o], become the estimates of the final observed finished product."

Control of the mixing process

The advances and cost reductions that have been made in electronics, and especially in microprocessors, are at last being applied in rubber mixing. However, microprocessors are used much more in industry in general than in the rubber industry. Howgate and Cottey (ref. 11) consider that there are four reasons for this:

* A lack of quantitative understanding of the interrelationship between processing parameters and product quality.

* The high degree of dexterity, skill or judgement required in many processes making automation difficult and expensive.

* A reluctance to invest in products that seem to be out-dated almost as soon as they become available.

* Difficulty in interpreting and developing the typical claims and jargon of microprocessor equipment manufacturers.

However, the degree of consistency which is now being required of the product of mixing can only be consistently met if the mixing process itself is controlled and consistent. It is difficult to see how this can be achieved other than by microprocessor control. There are a number of systems available (refs. 12-16) that provide automated handling, weighing and charging of all raw materials to the mixer and control the mixing process with a microprocessor. These monitor and/or control temperature, energy input, number of rotations of the rotors, and mixing time. The developments in this field were described in a series of papers in an ACS Rubber Division symposium on applications of microprocessors in the rubber industry in 1983 (ref. 17).

Most of the systems that have been installed concentrate on the materials handling as much as on control of the mixing process itself. Early systems were controlled using punched cards which have instructions on weighing and addition of raw materials and on dump criteria. Basically, these systems reduce the chance of human error. Modern systems do all of the above, but also include a programmable controller. One important advantage of these is that they can provide a printed record of each batch. Details of the capabilities of the commercially available systems can be obtained from the manufacturers. A paper in the Rubber Division's most recent meeting (ref. 18) shows how such a system can aid in controlling mixing.

Power profiles - recordings of the power consumed during mixing in an internal mixer - have been used for some time, and their importance to the developing science of rubber mixing has been recognized by a number of authors (refs. 19-28). The use of power profiles by the compounder has become so widespread that several equipment suppliers have produced instruments which will monitor the power required for mixing and provide process control action. In fact, most "off-the-shelf" programmable controllers can provide suitable control action.

However, all of these devices require that the compounder has a reasonable understanding of the relationship between the power consumed by the mixer and a desirable process control strategy. This can be a reasonable requirement if the compounder has to consider only a limited number of formulations, or if runs are significantly long for each of the mixes. The compounder may use the technique known as evolutionary operation, EVOP, to determine a control strategy and an appropriate instrument to effect that control action.

The use of statistical tools, such as SPC, to control the mixing process is not easy; however, they can be used to monitor the process and the product of mixing (ref. 29). Recording and analyzing the data will establish whether the process is in (statistical) control and therefore predictable.

The statistics involved in SPC are elementary, little more than an understanding of the basic mathematics of the Gaussian distribution. The actual techniques used, the application of the statistics, are not new. X and R charts have existed for at least 50 years. The belated realization that such simple techniques work is new. What is also new is the added realization that quality is not the concern of the quality control department alone; that quality is, in fact, the concern of everyone.

Reproducibility of test methods

Processability of elastomers, whether in mixing, extrusion, molding or curing, is and always has been a problem for the processor. The concept itself is difficult to define. In fact, rubber processability is a subjective concept in that it depends on what you want to do with the material, and in what equipment, whether it processes well or badly. Furthermore, there has been a lack of simple tests that can measure processability. Generally, if the test is simple it does not correlate directly with processability, and, if it does correlate, it is not simple. A third problem is that even when almost complete information is available on the elastomer or compound, in terms of its viscosity, elasticity, ultimate elongation, etc., the relationship of these parameters to processability in a mixer or extruder, or to flow in a mold, is not clearly understood.

The basic problems of process control in the manufacture of rubber products, and the need for means of assessing the processability of rubbers, can be summarized by the following three points.

* There is need for material characterization at a number of stages in the production cycle, especially (1) incoming material and (2) mixed compound.

* The property measured must relate to a significant response of the material to processing conditions (compound), or to its molecular characteristics (raw rubber).

* The method of measurement must be consistent with the practical constraints of manufacturing operations. This means that the test procedure must be rapid and amenable to execution and interpretation by semi-skilled or non-technical staff, though for development work more complex techniques can be considered.

In recent years, there have been a number of developments that have increased our understanding of the physical processes that take place during the mixing, milling, extrusion and injection molding of elastomers. At the same time, techniques and rapid-acting instruments have been developed for measuring various aspects of processability (refs. 26, 30 and 31). The combination of these developments should lead to increased efficiency in materials, energy and in equipment utilization.

At present, there are only very tenuous links between raw polymer properties and either processing or end-product properties. However, it can hardly be doubted that polymer consistency and quality have an overriding influence on both processing behavior and end product properties. The advances that are being made in our understanding of basic molecular parameters, rheological properties and elasticity, and also recent advances in electronics, microcomputers and instrumentation, will inevitably mean that such links will become firmer.

Processability of raw rubber in an internal mixer depends on polymer type, molecular weight distribution, degree and type of branching, physical form, catalyst and emulsifier residues, stabilizers and other additives. In addition, filler type, filler concentration, polymer-filler interaction, and the levels and types of oils and process aids affect the process.

Measurements can be made on the raw materials in order to determine how well they will mix, and on the product of mixing, to see how well it has in fact been mixed.

Veith (ref. 32) states that poor test quality, or lack of agreement between laboratories in the rubber industry is a serious problem. ASTM Committee D11 and ISO/TC-45 are actively addressing the problem of precision of test methods. Veith's paper gives a very clear account of the problems involved and emphasizes the difference between accuracy and precision.

"The two measurement concepts of accuracy and precision are often used interchangeably - they should not be. Accuracy describes how well measured values agree with a reference or "true" value: High accuracy implies good (close) agreement. Precision describes how well measured values agree with each other: High precision implies good (close) agreement. The difference between the measured mean value and the reference value is the bias. A large bias implies an inaccurate measurement process, which may or may not be precise. A large systematic between-lab bias, with its value unique to each lab, is the root cause of poor interlab precision. Eliminating these large biases is a difficult job."

Brown (ref. 33) stated that ignoring real material variations - which with proper organization should be small - there are three sources of test differences:

* Calibration of equipment;

* Care (or training) of the operators;

* Quality of the written standard test method; and suggested measures for minimizing the contribution to variation from each of them. He quoted Niemiec's (ref. 34) conclusions on the Mooney test:

"By standardizing the critical elements of the Mooney viscometer, by strict adherence to the proper operating procedures, and by reasonable care and maintenance, reliable measurements suitable for process control and specifications can be achieved. It is perfectly feasible to expect measurement levels obtained on different manufacturers' models in different laboratories to agree within [+ or -]1.5 Mooney points with 95 percent confidence."

Korosec (ref. 10) noted that most finished product tests in the rubber industry are destructive tests that are empirical in nature, and that standard reference materials (formerly handled by the U.S. National Bureau of Standards) are no longer readily available. To correct this she suggested that each company needs to establish its own bank of internal standard reference materials. In such a program the reference material has to be tested daily and the results plotted on an SPC X and R chart. The arithmetic mean and standard deviation of three months of the daily results for the reference material represent a good estimate of the average and standard deviations for the test. In a company with several production sites the reference materials should be used in each one to determine whether there are significant biases between plant laboratories. Of course, each laboratory should take steps to eliminate, or at least minimize, variations from the three sources identified by Brown. Even then, it is not unusual to find that there is a bias between two laboratories within the same company.

Wong (ref. 35) described a bias adjustment technique designed to minimize problems due to test bias, especially between supplier and customer, i.e. between the synthetic rubber manufacturer and the rubber processor. He said:

"There are two complementary and concurrent methods to improve quality information:

* Reducing test variations;

* Managing test variations. When the test variation is reduced, the data itself becomes more consistent. Statistical techniques, such as bias adjustment, are methods to extract better information from the current data. They do not improve the consistency of the data. While this approach is very different from improving the test itself, the goal is the same - to have better information so that better decisions can be made.

"The objective of bias adjustment is to make the best of a bad situation, while a concerted effort continues to be expended to improve the situation. With an accurate test, bias adjustment will be proven to be redundant - yet, as long as we still have an inaccurate test, bias adjustment can help in managing test variation.

"It is important to emphasize that bias adjustment is not intended to substitute for test improvement. In fact, bias adjustment can emphasize the need for test improvement and it can also facilitate test improvement when applied correctly."

Conclusions

This article has reviewed the three contributing factors to mixing efficiency and quality - the raw materials, the mixing process, and the test methods. The synthetic rubber producer has a role to play in all three areas.

Obviously the quality of the raw material (the rubber) is his primary concern. This is maximized by control of the raw materials used in the manufacture of the rubber, control of the manufacturing process and control of the tests.

Control of the mixing process is, just as obviously, the concern primarily of the processor. However, the rubber manufacturer needs to understand his customers' problems and this can only be achieved by investigating the mixing process and its control.

Control of test procedures is the concern of both the rubber manufacturer and the processor. Cooperation in improving and standardizing test methods through such organizations as ISO and ASTM is one part of this. Another important factor is cooperation between the two, directly, in establishing and managing test bias.

References [1.] Palmgren, H., (a) Europ. Rubber J., 156, 1974, 30; reprinted in (b) Rubber Chem. Technol., 48, 1975, 462. [2.] Funt, J.M., Mixing of rubbers, RAPRA Publications, Shawbury, U.K., 1977. [3.] Morrell, S.H., Prog. Rubber Technol., 41, 1978, 97. [4.] Johnson, P.S., chapter 3 of "Basic compounding and processing of rubber," ed. H. Long, Rubber Division, ACS, Washington, 1985. [5.] Johnson, P.S., chapter 7 of "Developments in rubber technology, volume 4," ed. A. Whelan and K.S. Lee, Elsevier Applied Science, N.Y., 1987. [6.] Freakley, P.K., chapter 3 of "Rubber processing and production organization," Plenum Press, N.Y., 1985. [7.] A paraphrase of Firestone's "General policy for supplier quality requirements," Firestone World Tire Group, Akron, Ohio, U.S.A., 1984. [8.] Wong, H., Paper 12, ACS Rubber Division meeting, Cincinnati, Ohio, October 1988. [9.] Francis, A.E., "Network - A quarterly for Polysar's technical community," p. 188, no. 2, 1989. [10.] Korosec, C.M., Rubber World, April 1988, p. 18. [11.] Howgate, P.G. and Cottey, D.P., Prog. Rubber Technol., 43, 1980, 99. [12.] Schmid, H.M., Rubber World, February 1984, 33. [13.] Rapetski, W., Elastomerics, September 1981, 21. [14.] Acquarulo, L.A. and Notte, A.J., Elastomerics, November 1983, 17. [15.] Muller, D. and Maire, U., Rubber World, February 1984, 25. [16.] Freakley, P.K. and Mattews, B.R., Rubber Chem. Technol., 60, 1987, 618. [17.] Symposium, Applications of microprocessors in the rubber industry, ACS Rubber Division, 123rd Meeting, Toronto, May 1983. [18.] Gark, K., paper in symposium on processing, ACS Rubber Division, Detroit, October 1989. [19.] Gessler, A.M., Hess, W.M. and Medalia, A.I., Plastics and Rubber Processing, 3, 1978, 109. [20.] Ebell, P.C. and Hemsley, D.A., Rubber Chem. Technol., 54, 1981, 698. [21.] Hess, W.M., Swor, R.A. and Micek, J.E., Rubber Chem. Technol., 57, 1984, 959. [22.] Dizon, E.S., Micek, E.J. and Scott, C.E., J. Elastomers Plast., 8, 1976, 474. [23.] Dizon, E.S., Rubber Chem. Technol., 49, 1976, 12. [24.] O'Connor, G.E. and Putnam, J.B., Rubber Chem. Technol., 51, 1978, 799. [25.] Dolezal, T.P. and Johnson, P.S., Rubber Chem. Technol., 53, 1980, 252. [26.] Johnson, P.S., Kautsch u. Gummi Kunst., 33, 1980, 725. [27.] Wolff, S., Rubber World, June 1984, 28. [28.] Dizon, E.S. and Papazian, L.A., Rubber Chem. Technol., 51, 1978, 799. [29.] Helmers, H., Eur. Rubber J., 171, 27, 1989. [30.] Norman, R.H. and Johnson, P.S., Rubber Chem. Technol., 54, 1981. [31.] Sezna, J., (i) Polym. Test., 8, (3) 161, 1988; (ii) Rubber World, 199 (4), 21 (1989). [32.] Veith, A.G., Polym. Test., 7, (4), 239, 1987. [33.] Brown, R., Europ. Rubber J., 171, Jan./Feb. 1989, 25. [34.] Niemiec, S., Polym. Test., 1 (3), 1980. [35.] Wong, H., Paper #74. ACS Rubber Division meeting, Dallas, Texas, April 1988.