Due to the mostly uneven movements of the active movement devices in hollow glass processing machines together with the non-linear behaviour of the drives used, time-variable inertia forces and moments occur there, which can lead to undesirable loads and vibrations. In order to counteract this with suitable problem solutions, the behaviour of the entire system must be simulated as well as possible. In addition to the dynamic behavior of the drives, the modeling required for this must also take into account the dynamic behavior of the non-uniformly translating mechanisms and the retroactive processes.
Today, there are basically three different processes in hollow glass processing machines: the traditional blow/blow process, the conventional press/blow process and the advanced narrow-neck press/blow process. Figure 1 shows the press/blow moulding process as an example. After the droplet feed through the funnel into the preform, the beak is formed upside down by the level mechanism pressing from below, whereby the mouth of the container is given its final shape by the mouth form. The preform base and the two mould halves are then opened and the finished mould is transferred to the finished mould side of the station. After the tank has been rewarmed and the associated tube lengthening has been completed, the tank is blown out. Finally, the container is placed on the conveyor belt by the gripper mechanism and transported to the cooling furnace.
In order to carry out the processing described above, the preform halves, the funnel, the preform bottom, the press stamp and the neck finish must be brought together in coordinated and precise movements on the preform side (Fig. 2). On the finished mould side, this applies to the finished mould halves, the base plate and the die head. In terms of time, the two forming processes on the preform and finished mold sides are linked with each other by the transfer mechanism, which swivels the mould from the preform to the finished mold side. A stable and repeatable glass forming process is the key to a high quality glass container. This process stability and repeatability concerns above all the station mechanisms, whose movements must be optimally coordinated with each other at the highest possible cycle rate and high accuracy.
In traditional machines, most movement devices are pneumatically driven. For this reason, a simulation model was developed to describe the dynamic behaviour of pneumatic cylinder drives with a non-uniformly geared transmission downstream[Cs96/1]. The model of the compressed air cylinder shown in Fig. 3 shows all essential parameters that influence the temporal behaviour of the pressure in the two cylinder chambers. This not only takes into account the inflow and outflow behaviour of supply and exhaust air, but also the possible leakage of the two cylinder chambers. Pneumatic end position cushioning can also be simulated by defining inlet and outlet cross sections depending on the piston position. Alternatively, it is also possible to include hydraulic shock absorbers with stepped throttle holes in the simulation.
As a typical example of an active movement device, the kinematic diagram of the preform closing mechanism is shown in Figure 4. It is used to bring the two halves of the mould, which determine the outer contour of the cooling chamber, together and compress them with a certain force. This can be used to counteract the process forces that arise when the container is formed. It is a twelve-unit transmission gear whose kinematic structure is a parallel and series connection of four-unit crank gears. The gear unit is divided into two partial gear units, the lower partial gear unit with the pneumatic drive cylinder, consisting of the moving links 1,2,3,4 and 8, and the upper partial gear unit or the clamping unit, consisting of links 6,7,10 and 11, which are coupled together by the so-called locking shafts 5 and 9, which are pivoted in the fixed points A0 and C0, and which are mounted in the fixed points A0 and C0, respectively.
Measurements were taken to verify the modeling described above by recording the pressure in the two cylinder chambers as well as the movement of the preform clamping arms and the mold clamping force. Based on these measurements, calculations were carried out which were based partly on precisely determinable parameters such as dimensions and masses and partly on estimated parameters such as friction forces. In an iterative process, a very good agreement between simulation and practice could be achieved by comparing measurement and calculation, as shown in Figure 5.
A further example of a very important active movement device in terms of process technology is the level mechanism. In addition, the retroactive effect of the glass forming process must also be taken into account, which is not exactly straightforward. Figure 6 shows schematically the start of pressing (a), the actual pressing process (b) and the end of pressing (c). With this model, the forming process can be approximated by considering an annular hydraulic throttle with variable length and variable cross-section. The necessary equations can then be derived from the force equilibrium for a infinitesimal small ring element of this throttle, taking pressure and shear forces into account. This was based on a Newtonian behaviour of the glass melt. The derivation shows that the force required for glass deformation is determined by three quantities.
In this, (T) describes the temperature-dependent viscosity of the glass, (x) a position-dependent factor resulting solely from the geometric model shown in Figure 6, and the level velocity.
Based on the results of the simulation model for the pneumatically operated level mechanism, a mechanism with servo-electric drive has now been designed. In terms of control technology, a combined position, speed and force regulator is used. For the force controller system, the glass pressure force to be controlled is calculated by means of an observer model using the motor current signal and the acceleration signal obtained by appropriate dual differentiation of the position signal. The controller design was based on the simulation model, whereby the pneumatic drive model was replaced by a model of the servo-electric drive. The special friction situation due to the use of a spindle-nut system for converting the drive rotational movement into a linear output movement was also taken into account. Subsequently, measurements were carried out on an experimental model in glass production with the appropriate controller settings. Figure 7 shows a corresponding comparison of simulation and measurement. The very good agreement speaks for the reliability of the model.