©, 1995 Nottingham University Composites Institute
This process is well developed for high volume manufacture of thermoset composites parts based on DMC and fibre reinforced thermoplastics materials. DMC materials can be injection moulded with relative ease to produce parts with nominally random fibre orientations, although in common with compression moulding, the fibre orientations induced during mould filling can be detrimental to mechanical properties.
A variety of DMC compounds are available for injection moulding including epoxies, phenolic and polyimides although those based on polyester are the most important in economic terms. The processing equipment is similar to that use for conventional thermoplastics injection moulding and is based on the use of a screw plunger to transport the preheated charge to the front of the barrel prior to an injection stroke. Since the feedstock material is non-granular hopper feeds cannot be used and the screw must itself be force fed by a separate cylinder. Relatively high mould pressures and temperatures (typically 300 bar, 140oC) are used to provide the short cycle times which are compatible with high volume manufacture. Developments in low shrinkage compounds have enabled several appearance parts for automotive applications to be produced in this way including the tailgates for the Citroen A and BX series cars (France). Other applications include instrument clusters and lamp housings but the largest growth area in recent times has been in under-bonnet (engine) applications. Technical problems associated with the environment such as high temperatures, humidities and operating stress levels have been overcome to enable a wide range of parts including inlet manifolds, oil pumps, water pumps, carburettors, rocker covers and thermostat housings to be introduced. The main limitations of injection moulded DMCs are associated with fibre degradation in the feedscrew, and injection gate and the flow induced orientations described earlier
Following the success of injection moulding for thermoplastics materials, and the development of injection moulding equipment capable of producing items of considerable size, the process has been extended to fibre-filled thermosetting materials. The automotive industry has expressed considerable interest in this development because the surface finish of the injection moulded item seems to be superior to that from compression moulded items. Injection moulded DMC appears to give a better paint finish than compression moulded SMC. Examples are front ends for American cars and the ZMC moulded tailgate for the Citroen AX and BX series cars. In the injection moulding process, heated matched metal tools are manipulated by the press under automatic cyclic control. The supply of material is contained in a screw fed barrel which is forced in to the tool cavity at an appropriate time in the cycle. For polyester dough moulding compounds, hopper feeds are not possible and the screw must itself be force fed by some kind of stuffing box arrangement. The main problems with the process have been in degradation of the material between mixing and the finished article. The main sources of degradation appear to have been reduction of fibre length in the screw, and breaking and filamentisation of the fibres in the tool gate. Improved design screws and rules governing the design of runners and gates have gradually been developed but the properties of finished artefacts are still governed by highly developed fibre orientation phenomena which will be discussed later.
In a sense injection moulding is a development of transfer moulding where there is a continuous supply of material and the tool is arranged for automatic opening, part ejection, and closing before feeding a further supply of material. It is well established for thermoplastic materials where it gives high rates of production and some moulding shops are fully automated. It has become attractive for the production of DMC parts as the difficulties of injecting fibre filled materials have been overcome. Its main attraction is high rates of production but it also gives rise to a very good surface finish.
DMCs require lower injection pressures than thermoplastics in the range 27 to 35 MPa. Nevertheless DMC parts can be rather large and there is a need for very large injection moulding machines with clamping forces up to 2500 tonnes and shot weights up to 10 kg. The DMC injection moulder is similar in principle to a thermoplastics machine except that the barrel needs to be force fed rather than hopper fed because of the nature of the material, and the screw feed needs to be specially designed to minimise degradation of the glass fibres. There is a tendency to cause fibre degradation in the screw, the runner, and the gate. Considerable efforts have been made to develop the process for the automotive industry. The ZMC process from Vetrotex - St. Gobain is a recent example and is further considered in a following section .
European manufacturers are willing to give far less detail than General Motors or Owens Corning. At the 1983 SPI Meeting and at other places, Vetrotex Saint-Gobain have reported on the ZMC process which appears to have been developed collaboratively with Citroen for the Citroen BX and AX cars. The Citroen cars include some large GRP panels. The bonnet or hood appears to be conventionally manufactured by SMC compression moulding, but the rear door or tailgate has been manufactured by an injection moulding process which they call ZMC. ZMC is a complete manufacturing system or process including the injection machine and a specially prepared compound to obtain optimal characteristics with tools designed for the process. Apparently the project was started in 1979 when the first mention occurred of injection moulding of large BMC parts occurred in the USA. The large parts moulded in the USA had a high production rate, reduced finishing, and good surface appearance although they had relatively poor properties. A number of front ends for American cars were made by this process. It should be noted that the Saint-Gobain group includes a machine builder and a mould maker as well as material producers. The process appears to be based on a fairly conventional large injection moulding machine with a closing force of 2300 tonnes. The special element appears to be the feed hopper, screw and plunger assembly which are designed to minimise degradation of the fibres and to provide automatic sealing of the mould during cure. Compound is supplied in large cans which appear to be capable of pressurisation and contains sufficient material for 2 hours' operation. The opening into the feed screw has a width at least equal to the diameter of the screw and a length 6 times the diameter of the screw. The screw stops automatically when the plunger is fully retracted, i.e. the barrel is full. These two features minimised the shear of the compound. Feeding the injection chamber and the mould is done through large channels. The backward movement of the feed and injection system is made without friction between the screw and the barrel since there is a separate inside barrel. Compound was developed for the process by evaluation of several polyester resins with thermoplastic additives in different proportions, several fillers, calcium carbonate, kaolin, and microspheres, in different proportions, and glass fibres with different sizing systems, different lengths and different percentage by weight. Other additives included elastomers, viscosity reducers and thermoplastic fibres. Only a limited number of properties are given, but it appears that a 20% glass content ZMC material retains the strength of an injection moulding grade of BMC containing 25% fibres, whilst maintaining a modulus not far short of that of a 25% glass content SMC. The impact strength appears to be mid-way between the BMC and the SMC. The proof of the process is in the fact that the rear door so produced passed acceptance tests at Citroen.
A case study by M J Owen
DMC's are used for a wide variety of mouldings. Originally the surface finish and colour were unacceptable for appearance parts but many components for electrical switch gear and the like were manufactured. Gradually as shrinkage control has become understood it has become practicable to mould a large number of appearance parts for electrical and domestic appliances. The reasonable mechanical strength with good electrical properties brings about dramatic part consolidation and hence reduces the total cost of the finished goods. There has been growing interest in DMC for automotive parts. Some of these are electrical, e.g. instrument clusters, lamp housings etc. but there is a growing interest in engine components to work in more demanding conditions. These conditions involve elevated temperature, aggressive environments, assembly stresses, service loads, etc.
Intake manifolds, oil pumps, water pumps, carburettors, rocker covers are some of the components under development. Most of them will be injection moulded. The following paragraphs give an account of some of the work involved in developing a prototype petrol engine intake manifold including relatively fundamental work leading to an understanding of the relationship between flow fibre orientation and mechanical properties.
In 1967, BIP Ltd. drew attention to the fact that the tensile properties obtained in DMC compression mouldings were extremely variable and sometimes only 10% of the published flexural strength of DMC test pieces produced by a standard method. It was thought at the time that some way of reducing scatter even without increasing the average tensile strength in moulded artefacts could lead to an immediate doubling of the market for DMC. The problem is illustrated by some results obtained by Pye in a 1968 student project . BIP Ltd. provided a stock of 8 inch square compression moulded test plaques 1/8 inch thick. These were cut into 1/2 inch wide strips and subjected to tensile testing. The broken ends of each strip were further tested until they were too short to hold in the testing machine grips. As a result of this, maps of the tensile strength and position and orientation of the fracture surface were obtained. These were found to be roughly reproducible from one plaque to another. Furthermore, burning off the resin to leave the glass fibres and fillers showed that the fibre orientation pattern was also generally similar from one plaque to another. These test plaques had been moulded by a straightforward procedure of weighing the compound, rolling it roughly into a ball and placing the ball in the centre of the mould. With hindsight it seems surprising that nobody had realised up to that time that the reproducible flow pattern produces a predictable pattern of fibre orientation and a consequent distribution of strength. The charge rapidly becomes a circular disc which then expands to touch the sides of the cavity. As the circumference stretches the fibres predominantly become circumferentially orientated. The four corners of the mould then fill as a converging flow resulting in fibre alignment along the diagonals - virtually a weld line. Subsequently a research student, K. Whybrew, demonstrated this behaviour using a perspex mould . The strength distributions observed by Pye are easily explained from the flow behaviour. The centre of the moulding remains approximately randomly orientated. Near the mid-sides of the mould the fibres are substantially parallel with the side and hence there is high strength parallel to the sides and low strength perpendicular to the sides. Near the corners where there is a weld line it can be shown that the fibres are parallel to the weld line and rarely cross the weld line. Hence this is a region of considerable weakness.
The behaviour of the compound in this simple test plaque indicates that there will be flow induced fibre orientation in larger and more complex mouldings whether they are produced by open cavity compression moulding or by closed cavity transfer or injection moulding. The moulding technique of course will govern the flow regime.
Whybrew showed from simple model regimes that an expanding flow tends to orientate the fibres parallel to the flow front and a contracting flow tends to orientate the fibres parallel to the direction of flow. By studying some prototype transfer moulded wheel mouldings which could be gated at different positions, he was able to show that the properties obtained for the same finished shape varied according to the position of gating and hence whether or not in a given area expanding or contracting flow took place. This was demonstrated by a sequence of colour slides showing grease filling a perspex model of the wheel. The options are to gate the model at the centre and fill radially outwards thus producing a weld line at the mid- point in the rim between each pair of adjacent spokes. Alternatively the model can be gated at the rim when there is only one weld line in the rim and a complex weld line near the centre of the model. The point is that the properties obtained and the performance of the finished artefact will be controlled by the moulding regime.
In the previous example of transfer or injection moulding, the formation of weld lines is fairly readily understood. Less easy to understand is the behaviour of material behind the flow front. Experiments with perspex models of transfer moulds using fibre-filled grease show that in the entrance gate and converging regions of flow the fibres tend to become orientated parallel to the flow. In the regions of expanding flow, they rapidly become orientated parallel to the expanding flow front. When the expanding flow front touches the sides of the mould, slip planes form and material is cut off is stagnant regions. These slip planes have much the same fibre orientation and strength properties as weld lines. Detailed studies of grease and fibre models, specially prepared DMC mouldings and commercial moulding enabled Thomas and Found to develop methods for predicting fibre orientation in complex transfer mouldings through prediction of flow fronts.
Injection moulding DMC is attractive in order to obtain high rates of production and accurately made parts. Front end mouldings for U.S. cars have been developed and the European manufacturers have been interested for several years culminating in the production of the Citroen BX rear door (ZMC process). Ford Motor Co. Ltd. developed a prototype petrol engine intake manifold for the Kent series engine which was reported by Rowbotham and Suthurst in 1980
The intake manifold is typical of a range of parts which are currently gravity die cast with sand cores in aluminium. Aluminium prices are expected to continue to rise in the foreseeable future. Provided that plastic components can be produced sufficiently accurately to do away with machining operations, there should be a saving in cost against the price of a finished aluminium part. The intake manifold was selected as a development component because of its size, complexity of shape, assumed loading, temperature gradients, and aggressive environments (water/glycol mixture, and petrol vapour/air mixture). The manifold selected was to suit the 1300 cc Kent engine which was the main power unit for Escort cars produced up to 1981. In the early stages it was expected that a manifold could be produced which go on to those models as a running change, and therefore the GRP parts would have to be a direct substitute for the existing aluminium part. All the fixings and attachments and general dimensions were unchangeable. Only minor changes to the external shape could be contemplated. The engine is a cross flow engine i.e. with the intake manifold on one side and the exhaust manifold on the other side of the head. It is a 4- cylinder engine with a single down-draught carburettor and even the aluminium manifold has a water chamber below the carburettor to provide a hot spot intended to ensure complete evaporation of the fuel. It was probably not appreciated by materials engineers that the aluminium manifold is a hot tube operating at about 60oC. Without this feature the carburettor alone is inadequate. The water heated hot spot is able to maintain this temperature because the aluminium is a very good thermal conductor.
The first step in the development process was to devise a feasible method of manufacture. The aluminium manifold had an external shape suitable for gravity die or sand casting and therefore there is a parting line which permits a solid pattern to form the mould cavity. However for aluminium casting a sand core is used which can be knocked out when the casting has solidified. The core boxes also have a parting line. The problem for DMC is what kind of core to use. Consideration of various low melting point alloys, especially tin-bismuth alloys suggested that a low melting point alloy core could be made which would be solid and strong enough at the DMC moulding temperature and yet which could be melted out at approximately 15oC higher temperature. It was envisaged that the cores would be made on a conventional die-casting machine and that the productivity of such machines would permit one die-casting machine to serve several DMC moulding machines.
A compression moulding tool was constructed for which the cavity would look very similar to a conventional mould for the aluminium part. Solid low melting point alloy cores were made in core boxes very similar to the sand core boxes. Successful prototypes were made by this technique but a number of problems were immediately thrown up. When trying to melt out the ore in an air oven, the exposed metal surface was so small that it took several hours to melt out the core and this in turn led to severe oxidation and loss of the low melting point alloy. Secondly, there was distortion of the manifold bolting face and also sufficient irregularity from the core prints that machining would be required. Thirdly, there was unacceptable porosity and occasional cracking. After much discussion it was concluded that certain changes were required. The tool was converted to a transfer moulding tool and an injection point selected on the basis of previous work on flow, and fibre orientation. A prediction of the filling pattern was subsequently confirmed from the first few transfer moulded manifolds. In order to provide a fair finish to the manifold face, the cores were cast on to a steel plate and the steel plate in turn inserted in the mould. Furthermore a system was devised for casting hollow cores in the manner in which, for example, kettle spouts are cast. This is patented technology. The combination of hollow cores and melting out in a tank of oil maintained at about 15oC above the moulding temperature permits the cores to be melted out in a few minutes. This avoids deterioration of the low melting point alloy and avoids significant damage to the manifolds themselves. Fry's Metals have carried out much further work on low melting point alloys for injection moulding cores. The Dunlop tennis racket is one result.
Further work was required on the development of the moulding compound. Shrinkage was controlled initially using a standard system which incorporated some asbestos. Development turned to the use of other resin systems with thermoplastic additives. In the meantime, a large number of flat tray mouldings were made from which tensile and other test coupons could be obtained. Tensile test coupons were variously conditioned by boiling in water, water/glycol antifreeze, 4 star petrol at reflux temperature, and standard multi-grade lubricating oil for various periods up to 500 hours. From these tests the residual tensile strength at various temperatures was ascertained. None of the moulders' standard compounds had an adequate retention of strength under these conditions. Most of them had been formulated for use in air under normal ambient conditions, and give good flow characteristics in complex mouldings. A major study was commenced to find a suitable compound formulation which would have adequate strength retention good shrinkage control, and would mould easily into the complex shape. Such a compound designated X52 was eventually discovered. To some extent the strength retention requirement was eased by the realisation that in service the aggressive environment is normally on one side of the material only.
Along the way a number of other problems had been discovered. Whilst material development was going on, trial mouldings continued to be made with a view to getting them on to cars for test. Some of the mouldings were obviously cracked and the water box leaked. This necessitated modifications to the entrance gate for the compound to make it easier to break out the moulding from the mould without damaging the wall at that point. When this problem had been overcome a number of mouldings were tried on test cars. They were bolted on using the SAE standard bolting torque for the bolt size. After a few days' service, it was found that the water was leaking from the water box, along the manifold face and entering the ports and so entering the engine. Various theories were proposed. Firstly it was discovered that the bolting torque was rapidly relaxed and this was attributed to creep of the material in compression, especially at the elevated temperature in the manifold. There is little doubt that this is true, but at the same time it can be calculated that the residual bolting torque is really quite adequate for normal clamping. Simulation trails were carried out with a cylinder head in the University of Nottingham. By passing water through the cylinder head and water box alternately at 95oC and 20oC, it was found that approximately 20 to 30 cycles produced a leak at the bolting face close to the water passage.
It was established that this leak was due to the fact that differential thermal expansion of the water box relative to the inlet ports produced a progressive creep phenomenon, eventually causing a leak. It was therefore decided to incorporate into the material development programme a search for a compound which also had as low as possible coefficient of thermal expansion. Fortunately one of the better compounds from the point of view of degradation resistance also had a relatively low coefficient of thermal expansion and obviated this particular problem. What had been overlooked was the fact that dough moulding compounds have low thermal conductivity which caused the engine to operate with cold inlet passages. The evaporating fuel spray from the carburettor is cold and the passages remain at this temperature because they are not heated by thermal conduction along the branches of the manifold from the water heated hot spot.
Eventually all these problems were overcome, and the manifold performed satisfactorily in all but one respect. In spite of the fact that engine performance testing leading to engine mapping is carried out under computer control, there are still subjective tests carried out on cars. One of these is a "stumbling" test. A car is driven at approximately 18 miles per hour in top gear and then the driver puts his foot down hard on the accelerator pedal to give maximum throttle. The car should accelerate without back-firing or stumbling. With the aluminium manifold the Kent engine meets this test providing the engine is adequately warmed up. The DMC manifold did not meet this requirement when the car was handled by an experienced tester.
The development work described above took several years and the DMC manifold was never introduced on the Kent series engine. The technology and the experience have in fact been passed to engine component designers and there is now activity in developing mouldings for carburettors, diesel engine intake manifolds, water pumps, oil pumps, timing covers, etc. and it can be confidently expected that there will be DMC engine parts in virtually all future engines. When petrol distribution problems in intake manifolds are better understood there will be intake manifolds.
A number of lessons have been learnt. Close collaboration is required between materials experts, moulders, and component engineers. Everybody must understand the relationships between material selection, processing, properties, and design. A long accepted metal component which works reasonably well may never have been subjected to close scrutiny. However, when the component is made in the new material it may come under close scrutiny. Initial ideas may lead to quite arbitrary requirements for material properties, and yet important material properties can be overlooked. Attempts to copy an existing metal component in order to introduce a "running change" may lead to unnecessary difficulties. An opportunity to redesign the component may make the application relatively straightforward in the new material.