Making Industrial Processes More Cost Efficient
Several researchers have modeled various aspects of the fibre drawing process in many different ways. There is also a lot of information on the chemistry of fibres. However, for industrial control purposes, a complete picture of the whole process, also taking into account the fibre characteristics, is desirable. In other words, it is not sufficient to accurately control fibre diameter neglecting the strength and attenuation characteristics. Most of the models reported in literature so far are physical models based on first principles, with several simplifications and assumptions. Empirical approaches have also been considered in some papers. The empirical models usually resemble reality more closely than the physical models, and such models can be used for control purposes. Figure 1 shows the ideal situation where one or several separate empirical models for several variables would also take into account natural laws like conservation of mass, heat and momentum as far as possible.
While dynamic models are useful in the ramp up stages, steady state models are sufficient for control of fibre characteristics during normal production runs where short term dynamics do not play a significant role. This paper reports only a small subset of the desired set of models as depicted in Figure 1. Two such component models, which are steady state models, and software incorporating the models is reported in this work. The software has features for efficient use of the models and for determining optimal process conditions. In future, the software will also be able to learn these and other component models on its own from production or experimental data.