Recently, a mineral processing plant utilizing the SABC milling process has implemented an advanced control system with significant results. Despite the need for manual intervention during unusual conditions such as roller replacements or blockages, the advanced control system has successfully automated most of the milling process. Operators have highly praised the system, emphasizing that a favored tool by them is seldom ineffective. While advanced control systems are widely used in large petrochemical industries, the non-ferrous metals sector often opts for expert systems instead. In our plant, we have two milling lines: one using an expert system and the other utilizing advanced control, creating a competitive atmosphere, albeit friendly.
Instrumentation Dependence
Expert systems in milling require additional instrumentation, such as ore size analyzers, semi-autogenous grinding fill analyzers, and cyclone overflow particle size analyzers. These instruments are prerequisites for the operation of expert systems. Delays in instrument delivery and installation have postponed the commissioning of the expert system. In contrast, the advanced control projects focus on automating the knowledge of operators. If the operators can manage the process, their expertise can be translated into an automated system. Advanced control emphasizes maximizing the potential of existing equipment and focuses on automating operational knowledge.
System Division of Labor
Expert systems aim for second-level control and optimization through an OPC server, which puts a strain on communication. Advanced control systems also operate via an OPC but require communication and execution only every 30 seconds. Advanced controls aim to perform rapid control actions at the DCS (Distributed Control System) level, reducing the complexity of higher-level optimizations and focusing solely on human operational knowledge.
Control Algorithms
By definition, expert systems utilize rule-based control algorithms, whereas advanced controls are model-based. Rule-based algorithms currently do not have the wide industrial application that model predictive control algorithms do. This limitation is more related to the non-uniformity of rules, scarcity of skilled personnel, complex maintenance by clients, and variable engineering conditions rather than mere technological superiority. An algorithm that cannot grow within an industry is challenging to adapt to other sectors. Expert systems are rarely considered in large-scale petrochemical settings, whereas advanced controls have penetrated more diverse fields.
Engineering Details
Providing a solution that truly addresses client needs involves more than just supplying tools. A robust solution must consider the integration of the control system, communication stability, adaptability of the solution, ease of maintenance, handling of exceptions, and staff training. Many companies develop advanced control software, but engineering applications are often fraught with challenges. Successful engineering applications rely not just on algorithms and software but also on the engineering capabilities of the developers. If fundamental technical details are not adequately addressed, it is akin to building skyscrapers on sand.