Long term follow up of control performance in TMP processes

In this paper we will focus on three important issues; Process understanding –in relation to control of nonlinear processes; Natural decoupling by using internal state measurement devices – The idea is to use rudimentary controllers with reliable runnability before entering more sophisticated control concepts; Long term follow up procedures of process control investments – An important issue when identifying the economic potential of future investments. As an example we will use a new control system for Thermo Mechanical Pulp (TMP) refiners, one of the most demanding non-linear processes in the pulp and paper industry. It is based on a cascaded control structure where the internal states, in this case the refining zone temperature profiles, are controlled in the inner loop whereas the outer loop handles pulp properties. The characteristics of the temperature profile dynamics makes it possible to introduce a decoupling scheme where the anti-diagonal elements in the transfer function matrix describing the process, can be eliminated naturally. Hence, the idea is to control this complex and nonlinear process with serially linked refiners using basic control loops to improve the runnability. At the time of writing, an MPC-concept for overall TMP-plant optimization is discussed. The system is built to handle several pulp propertied simultaneously but in this study mean fiber length (MFL) is the target variable. The process is followed during about 200 days in manual mode control and 200 days in automatic mode in order to evaluate the control performance at different operating conditions. A straight-forward high-pass filtering technique in combination with a threshold selection procedure is introduced for allowing comparison of the data sets from the process. It is shown that the standard deviations in the pulp property variables freeness (CSF) and MFL were decreased about 40 and 60%, respectively. The reduction in variability of shives was approximately 25% when running the process control system and a significant reduction in the motor load standard deviations was also achieved. On top of this process stabilization, an increased production was obtained at the same time as the control system runnability was raised from 50 % to 98 %, levels that are far from commercial MPCcontrol concepts in TMP refining control available on the market.