Mechanical Pulp Refiner Transfer Function Identification, a Stochastic Approach

In this paper, application of Bivariate Autoregressive Moving Average (ARMA) models via Dynamic Data System (DDS) Methodology are proposed as the stochastic transfer function models for on-line system identification and optimum control of mechanical pulping process. The objective is to define a methodology to minimize the energy consumption. The bivariate ARMA modelings are, therefore, applied to the mechanical process to estimate transfer functions between power delivered to the refining process, disk gap and feed rate. The results indicate that the dynamics of refining process can be described as a first order system. Furthermore, it is shown that there is a coupled relationship between the dynamics of refining process and the dynamics of the refiner structure.