0ver the past decade the rubber industry has faced an ever increasing demand for improved product performance, in terms of function, the severity of the service environment, and the service life. This challenge has been met by improved product design procedures, by improved materials, and by improved manufacturing. There have been important evolutionary changes in manufacture which have resulted in improved product quality and uniformity, probably shared equally between technical and organisational changes.
It is not possible to point to a critical few processing developments which have characterised the past decade. The rising tide of automation and computer‑controlled systems has affected all areas of rubber processing, bringing new opportunities for the identification and elimination of uncontrolled variables. Computer methods have also brought powerful techniques for process optimisation within the grasp of the technologist. In contrast, we have only recently begun to meet the challenges set by the quantitative design of processing equipment in a useful and realistic way, despite a long history of academic work on process modelling.
Process Compound Properties and Measurement Methods
Components of a Rubber System
Rubber compounds are complex, chemically active, viscoelastic materials and undergo both transient and permanent changes of properties during flow in practical processes. Many are blends of two or more elastomers and virtually all contain particulate fillers (carbon black, silica, china clay, calcium carbonate etc.) and a chemical crosslinking system, usually based on sulphur. In addition, lubricants, plasticisers and organic process aids are used to modify properties. A typical rubber compound is a microcomposite with, on average, twelve components. It is these characteristics which make the modelling of both material behaviour and rubber processes such difficult tasks and hamper the diagnosis of processing problems.
Efficient manufacturing of rubber products requires that shaping processes are completed before the onset of crosslinking, but that crosslinking should then proceed as quickly as possible. During crosslinking (alternatively named cure or vulcanisation) the material behaviour changes from predominantly viscous to predominantly elastic and shaping becomes impossible. The advent of reliable and rapid cure testing equipment (curemeters) over 25 years ago provided one of the most important tools for monitoring and controlling rubber compound properties. However, near isothermal curemeters, which provide a sensitive measure of crosslinking behaviour, undistorted by sample temperature rise effects, are a relatively recent innovation.
In contrast to cure testing, which is a routine operation throughout the rubber industry, measurement of the flow behaviour of rubber compounds is surprisingly undeveloped. The Mooney viscometer is widely used but only provides an empirical, single-point measurement under conditions far removed from those encountered in many practical processes. Various designs of extrusion and rotational rheometers have been introduced but have failed to establish themselves as routine industrial tools. The main reasons are the technical difficulty of selecting appropriate testing conditions, time-consuming procedures and, probably the most important, difficulty of interpreting the results and relating them to process performance. Some information on flow behaviour can be gleaned from a standard curemeter, prior to the onset of crosslinking, in the form of cyclic dynamic viscoelastic properties. Following this line of development, a commercial instrument, the Monsanto rubber processability analyser, has been produced which allows the dependence of properties on both strain amplitude and frequency to be explored in a test which is simple to perform and capable of being pre-programmed. Only time will tell if this instrument is accepted widely by the rubber industry.
Design of Mixing Processes
Process Variables and Uniqueness
All rubber product manufacture begins with a mixing process, and the behaviour of a rubber compound, both in downstream processes and in the final product, is influenced strongly by the treatment it receives in this process. Rubber mixing is dominated by batch processes, so there is an opportunity to vary the mixing treatment over a wide range through the manipulation of operational variables, in contrast to continuous mixing, where the mixing treatment is much more strongly influenced by the mixer geometry. This gives the processor the essential capability of mixing a wide range of compounds with a single mixing system. Despite this, it has proved to be very difficult to obtain similar properties from batches of rubber compound mixed in machines with different geometries. The reasons for these differences are fairly clear. Each mixer design will have different thermodynamic characteristics and will impart a different stress and strain history to the materials being mixed. There are also scale effects associated with mixers of similar design but different size, so that positive steps have to be taken to ensure that rubber compounds developed in laboratory mixers can be transferred to production without substantial changes of properties occurring. Laboratory mixing systems set up to give a good simulation of production mixing are still a rarity.
Quantifying the Process
The expertise, based on experience, to tune internal rotor design for a specific range of rubber compounds exists in a number of companies, but quantitative methods to predict the effect of mixer geometry and mixing conditions on property development of the rubber compound and thus provide viable design tools, are only just starting to emerge. This is a problem of considerable academic activity and industrial interest. Meanwhile, engineering ingenuity and empirical expertise continue to produce evolutionary advances in mixing machinery.
Design of Shaping Processes
Rubber vs. Thermoplastics
Moving downstream, a similar picture appears for extrusion and moulding processes. Both are highly evolved, but it is only recently that effective quantitative methods for design and simulation have begun to emerge. In extrusion, it is possible to borrow methodology from thermoplastics extrusion. Rubber extrusion is, very approximately, equivalent to thermoplastics melt extrusion. Hence the basic models for screw design are similar. However, the detailed behaviour of a rubber compound in an extruder screw is well removed from that of a thermoplastic melt. Viscosity is much higher and it is more elastic sometimes exhibiting substantial wall slip and thixotropy, particularly in the feed zone. In addition, the complex geometries needed to introduce a mixing action in rubber extruders, for effective heat transfer and minimisation of hot spots, create a further challenge for the mathematical modeller.
Computer Modelling of Elasticity
The elasticity of rubber compounds is also a complicating factor in die design. Computer-aided die design packages based on viscous flow analysis are commercially available. These enable the internal geometry of a die to be determined so that flow velocities around the periphery of the die exit are approximately uniform and extrudate distortion is minimised. While such packages are powerful and useful tools, they do not enable the shape of the extrudate to be predicted with any accuracy. The viscoelastic recovery which determines die or extrudate swell is influenced strongly by the deformation history of the rubber compound during its flow through the extruder head and die and, in highly elastic compounds, there may be a residual memory of flow in the channels of the extruder screw. The rubber industry uses short flow paths in extruder heads and dies to minimise pressure drop, temperature rise and the power needed to drive the extruder. Thus, there is little time for viscoelastic memory effects to fade before the material emerges from the die. Computer programs which are capable of simulating some of the viscoelastic memory effects in die flow exist but, at present, they require a high level of mathematical ability and very powerful computers. The design of a user-friendly package capable of introducing viscoelastic effects into the design of practical dies and capable of being run on a conventional workstation is undoubtedly a major challenge. An associated challenge is to provide guidance on flow measurement methods easily accessible to the rubber industry which can be used to characterise the behaviour of rubber compounds for input to the design package.
Computer Designed Moulds
Due to the importance of injection moulding to the thermoplastics industry, the development of sophisticated computer-aided mould design packages has been rapid. The major commercially available packages are now based on finite element methods. This parallel development has been of benefit for rubber injection mould design but there are substantial differences in material behaviour and in the typical mould geometries which limit the commonality of design methods for the two classes of material. Rubber mouldings are thick walled in comparison with thermoplastic mouldings - flow is thoroughly three dimensional and there is good evidence that the filling of a mould with rubber occurs predominantly by jetting, rather than by the spreading flow for thermoplastics. The fact that rubber undergoes crosslinking is another key difference, but, in modelling terms, not a major one. Clearly, there is substantial scope for development of existing computer-aided mould design methods for rubber injection moulding.
Although injection moulding has received the majority of attention in recent years, compression moulding is still a very important process for the rubber industry and, in the manufacture of thin walled or small cross-section products, has technical advantages. Due to the long and restricted flow paths in injection moulds for such products, substantial molecular orientation can occur, with a detrimental effect on dimensional control and product performance. In contrast, compression moulding involves short flow paths. The major challenge is one of innovative process design, to engineer into the compression moulding process a level of automation which will make it comparable in productivity and consistency with injection moulding.
Automation, Process Monitoring, Control and Optimisation
In contrast with other process industries, the level of automation, instrumentation and control in the rubber industry is not high. Substantial opportunities exist to apply existing methods to rubber product manufacturing and substantial progress is being made to accomplish this. This area is dominated by technology transfer, rather than innovation.
The multivariable nature of rubber compounds and rubber processes has prompted the adoption of statistical experiment design and optimisation packages by a number of companies. These enable the simultaneous effect of a number of variables on a particular property of the system being studied to be predicted. The main attribute of such packages is that they can be used effectively in situations where the system being studied is imperfectly understood and the fundamental models of the system are inadequate. Their main disadvantage is that they demand substantial experimental data and can only model the system being studied within the boundaries set by the experiments performed. In contrast, fundamental models, based on the physics and chemistry of a system, provide tools which can be used for predictions well outside the boundaries of experimental data, to new rubber compounds and to new processing equipment geometries and conditions. While statistical experiment design packages will continue to be a very important tool for optimisation, further progress in the development of fundamental, predictive models is very important for future advances in materials and processes.
The framework for development of rubber processes is set by the demands of customers for improved product performance and consistency, and by the unique nature of the industry, in which the manufacturer often formulates a different rubber compound for each product. It is this in-house control of rubber compound composition which is, at the same time, one of the industry's main strengths and one of its main problems. It is a strength because it provides the manufacturer with great flexibility to meet the customers demands. It is a problem because commercially viable processing equipment and computer-aided design methods have to work with a very wide range of materials behaviour. In the former case it is difficult to optimise equipment design and in the latter, the properties of each rubber compound must be characterised for input to the computer package. However, there is substantial scope for advances in the environment described above. There is a continuing evolution of processing equipment design, which can be accelerated by an improved understanding of the fundamental physical and chemical behaviour of rubber compounds under processing conditions and by improved computer-aided design tools. An improved understanding of the effects of rubber compound ingredients on their processing behaviour can also lead to predictive materials models, which will avoid the need for extensive testing every time a formulation is developed or changed. The new armoury of methods for improvement in rubber processing can be summarised as:
• Computer aided formulation
• Finite element analysis for mixer design
• Computer controlled processability testing
• Computer modelling and evaluation of flow, cure and heat transfer properties
• Finite element analysis for extruder and moulding machine design and operation
• Finite element analysis for die and mould design.
These are only a few examples of extensive possibilities. It is only recently that computer-aided methods have advanced to the point where they have begun to have advantages over the accumulated know-how of the rubber industry. They are now the most powerful tools available for accomplishing much needed improvements in processing technology. Research and development opportunities in the near future look very exciting and challenging.