The Automated Precipitation Control Optimiser (APCO) is an Expert System that delivers the process engineer’s ‘Holy Grail’ of Bayer process control: rapid, science-based optimisation of input parameters to maximise production of quality product. APCO embeds 25 years of fundamental research from the joint industry-funded AMIRA P521-521D projects within a dynamic precipitation model to predict liquor yield, the particle size distribution and the soda content of the product. The Optimiser can be customised to any equipment configuration, including bypasses, split seeding, classification and inter-stage cooling. It dynamically models the responses to changes in flow, temperature, supersaturation, seeding rates etc., and has the potential to be integrated into existing on-line process control systems.
In this paper we describe the development and main features of APCO, and demonstrate its performance with a generic plant design that includes all the elements of a typical modern Bayer precipitation circuit. We demonstrate the power of the system with a surprising example of how a key parameter may track adversely after a process change before eventually resolving positively. With innovative use of surrogate models, APCO is superior to existing simulators and a must for every process engineer’s toolbox.