Pulping operation, simulated

The possibility of ordering in native celluloses used in papermaking, and the significance of such ordering to the paper properties were explored in an experiment designed to simulate the physical environment during a typical commercial pulping operation (3). 

The overall system that we will analyze comprises the unbleached Kraft pulp line, chemicals and energy recovery zones of a specific paper mill (Melville and Williams, 1977). We will employ a somewhat simplified but still realistic representation of the plant, originally developed in a series of research projects at Purdue University (Adler and Goodson, 1972 Foster et al., 1973 Melville and Williams, 1977). The records of simulated operation data, used to support the application of our learning architecture, were generated by a reimplementation, with only minor changes, of steady-state models (for each individual module and the system as a... 

For the particular test cases, we have simulated a digester that produces pulp out of softwood. The digester was split into 50 zones, thus producing a set of 950 nonlinear ODEs. The simulation was performed using the ode23 solver of MATLAB. The simulation time corresponding to one-hour real operation time was 30s in a Pentium IV 1400 MHz processor, which, given the complexity of the system, is a rather short time. 

This work is part of the smart enterprise division in Tumut Visy Pulp and Paper, and addresses the advanced control and operation of the mill. In this context, a robust dynamic model is developed for the recovery boiler, and validated over a wide range of operating conditions. In the steady state case, energy and mass balances were carried out over the different sections of the process. In the dynamic case, initially, heat and mass transfer across the bed are coupled with moisture evaporation, black liquor pyrolysis, char combustion and gasification, gas-phase. The influence of model parameters kinetic constants and operational variables on process dynamics are studied by numerical simulation. The model was developed/implemented in a Visual C++ environment. 

As part of the smart enterprise division at Visy pulp and paper mill this work is related to the development of an advanced operation/control strategy for the plant recovery boiler. The development/implementation of a steady state/dynamic model was initially conducted, in this work and was reported here. Model validation, were found to be within 6% error with respect to actual plant data (for the steady-state case) and preliminary dynamic simulations show consistent trends. Work is underway to extend the capabilities of the dynamic model to two/three dimensional case for more realistic considerations as well as to incorporate the upper furnace and steam generation sections. The developed model will then be used to investigate the implementation of advanced control strategies that could achieve different operation objectives and maintain desired paper quality for the mill. 

---