MODEL PREDICTIVE CONTROL OF SMELT DISSOLVING TANK TTA USING FT-NIR LIQUOR COMPOSITION ANALYSER, 2010 TAPPI/PAPTAC International Chemical Recovery Conference

Please Note: This document will be available in PDF format in the "My Electronic Documents" link on the home page once your order has been completed. Please make sure you have the latest version of Acrobat Reader. Click on the Acrobat Reader icon to check for the latest version, it’s FREE. To print a hardcopy of a PDF file correctly you must have a postscript printer. If you are not sure if your printer is a postscript printer please refer to your owner’s manual.

Costs in the recovery cycle are minimized when both the effective alkali and the causticizing efficiency of the white liquor are as high as possible. This, in turn, requires tight control of the total titratable alkali (TTA) of the clarified green liquor feeding the slaker, which is in large part determined by the strength of the raw green liquor exiting the smelt dissolving tank. Common practice is to control density in the dissolver and TTA downstream via a trim addition either before or after the clarifier. The TTA “measurement” is normally calculated from density. Density meters react to all the ions in liquor, not just those associated with TTA and, even when carefully installed and maintained, are subject to drift due to interference from suspended solids and the steady build-up of scale. Process analysers, such as the FT-NIR liquor analyser, provide complete liquor composition analysis, and circumvent most of the problems associated with density measurement through the use of a sampling system that removes contaminates and prevents scale through backwashing. A drawback, however, is that the measurements are discontinuous. Therefore, in this report, state space methods are used to combine discrete-time FT-NIR and continuous-time density measurements in a TTA soft sensor. The resulting state estimates are then used in a model predictive control (MPC) strategy for TTA regulation. Results from implementation at a kraft mill are reported.