FPM for the Simulation of Flow Problems during Glass Production

 A directly connected problem is the adjusting of the temperature at the cylinder walls, where the melt cools fastest. If the solidification starts too early, this is usually due to an insufficient heating of the walls.

The series of images on the left side shows the filling process of a hollow cylinder at an acceptable wall temperature. There are no early solidification processes; instead, an almost quasi-stationary inlet and spreading behavior of the molten glass can be recognized. The series of images on the right side, however, shows a filling process where the wall temperature has evidently been selected too low. The solidification of the melt starts before it has completely spread in the cylinder ring. In this case, the production parameters - especially the wall temperature - ought to be changed in order to reach an acceptable result.

During glass production, streaks often occur in the glass melts. These are undesired pollutions of the glass originating at the walls of the glass tank or in channels and pipes where the glass flows. The streaks have different physical properties compared to the glass melt. In particular, density and viscosity of the streaks and the glass are different.

From a technical point of view, it is extremely difficult to avoid the formation of streaks completely. In comparison, the homogenization of already existing streaks in the molten glass through stirrers appears to be essentially more easy. However, since molten glass is an extremely viscous medium, the optimization of the stirrers is especially important because a mixing of the glass without turbulence can hardly be realized.

Effectivity of a Stirrer by Simulation

The basic idea of this project, commissioned by the company Schott Glas in Mainz, is the verification of the stirrer's effectivity by simulations and the testing of variations of the stirrer geometry, if necessary. We consider the flow of a two-phase medium (glass and streak) and develop a numerical model for the dynamics of the streak within the stirrer. Principally, the application of FPM is very much appropriate here, because it is able to describe moving phases and geometries inherently by its Lagrangian formulation.

One example of an instationary flow problem with free surfaces is a gob feeder (TC-25), as it is used during the production of traditional tube screens. The example of the TC-25 demonstrates the manifold possibilities provided by the Finite Pointset Method (FPM) developed at the ITWM. During a continuous production process, a gob feeder dispenses of one drop of molten glass at a time within one cycle. The gob is fed into a mold, where it is molded in the form of a screen by a punch during a further step.

The gob to be generated must meet certain requirements with respect to size and temperature. Besides, before such a technological process is put into operation, it must be guaranteed that these requirements will be continuously met during the entire production cycle. However, a gob forming process according to the TC-25 pattern is very sensitive, and there are several parameters which can have considerable effects on the forming process. Design can therefore be supported decisively by simulations.

Shaping Process TC-25

The TC-25 device consists of a tank which is open at the top and contains the glass melt. A tube inlet above the surface of the melt guarantees a continuous feeding of the glass melt. At the bottom of the tank there is a spout (outlet). A »needle« which is periodically moving up and down above the spout is responsible for the portioning of the mass flow of the glass.

The upwards movement of the needle leads to a suction effect which prevents the glass melt from flowing through the spout. A cavity is filled with glass melt, which is then pushed through the spout by the successive downwards movement of the needle and thus takes the form of a gob. Each one of the gobs which are periodically created in such a way is cut off and drops into the mold, where it is finally molded into a screen. The needle itself is covered by a revolving tube which rotates around its own axis once per cycle, thus ensuring a thorough mixing of the glass melt.

The ITWM has taken part in a very extensive benchmarking with respect to the TC-25, which has resulted in the proof that such a process can be simulated by FPM without problems.