“Process modeling and AI-supported twinning are key enablers for the circular economy of metals.”

Nikolaus Borowski: Process modeling and digital process twinning are key enablers of the circular economy and relevant for all metals-producing plants. For the recycling of non-ferrous metals, the design of plants for multi-metal recycling is particularly complex due to the large number of elements that must be handled. We aim to recover not only base metals like copper or nickel but also additional metals like tin, zinc, lead, as well as platinum group metals, or PGMs. With process simulations, we couple the metallurgical processing of primary and secondary resources with the product design of our metallurgical equipment. 

Ali Akouch: The main objective is to minimize the dissipation of materials and maintain material quality. However, we also strive to minimize energy losses to ensure minimal exergy dissipation. Digital twins facilitate the integration of the thermodynamic and physical foundations that enable us to evaluate different energy mixes.

Nikolaus Borowski: Digital process twins provide the basis for assessing the most effective metallurgical process with the most efficient equipment. By simulating different scenarios, they help predict outcomes and optimize process parameters. In the first step, we compare different process routes to define the optimum process for a given input material. Then, our simulations allow us to go into much more detail and design the process in such a way that we maximize output and minimize waste by creating valuable by-products.

Nikolaus Borowski: Together with KGHM in Poland, we created a detailed digital process twin of their plant including over 100 calculation units/reactors, linked by close to 600 flows, containing over 40 elements and a significant number of compounds that can arise when processing both concentrates and secondary materials as well as e-waste. Detailed modeling of the various reactors of the Legnica plant was simulated using Gibbs free energy minimization and included various non-idealities in multi-compartments/units (up to five) for some reactors to capture as many of the industrial realities as possible. Various optimization aspects were investigated on this basis. Above all, this permits a physics-based approach to footprinting and, in turn, the allocation of the footprint of a multi-material recycling plant based on exergy. This is a major new milestone in the use of a simulation tool to derive the right conclusions for future investments in plant-optimization projects.

Ali Akouch: We have developed BlueControl, an advanced real-time predictive process control system. BlueControl is the first AI-supported, accurate process control system and a game changer in the non-ferrous metals industry. It is a real-time calculation engine for designing, analyzing, controlling, and optimizing metallurgical processes “on the fly.” It is based on a rigorous simulation, which together with real-time data creates datasets that are integrated by deep-learning applications within level 2 and 3 process control architectures. We have integrated thermodynamic information in the deep-learning results to ensure predictions are always consistent with the fundamental laws of physics.

BlueControl has a wide variety of applications. For example, it can be used to determine the best feed mix to lower the carbon footprint at maximum production, or to optimize the final product quality while using the most economical feed input materials and reducing tap-to-tap times. Other applications include the prediction of operating points to extend the furnace life or the optimization of the gas cleaning process.

Ali Akouch: BlueControl uses two core applications: simulation and optimization. Simulation is the heart of BlueControl and functions as a controlling procedure. Take the copper scrap refining process in a tilting refining furnace (TRF): it calculates the anode copper mass and impurities’ concentration at the end of the refining oxidation and reduction stages. The model calculations produce multiple results, including the dynamic liquid copper weight and its composition, dynamic slag weight and its composition, dynamic off-gas composition, and the solid weight. Using the HMI, the operator can visualize and analyze the results of the simulated refining process.

BlueControl is embedded in a modular optimization algorithm to optimize copper refining efficiency. Based on the copper scrap weight and its composition, feed temperature, oxidation gas rate, and reduction gas rate, the optimization algorithm finds the optimum flux combination to maximize the amount of copper and minimize the impurities at the end of the refining process (purest copper anode). This improves production efficiency, yield, quality, and throughput, and reduces production – and we are talking about many millions of euros per year.

Ali Akouch: A first prototype of BlueControl has already been successfully tested at a customer. We are excited about developing this technology further to deliver next-generation metallurgical plants with enhanced efficiency and effectiveness. We still have various applications in the pipeline for the non-ferrous metals sector. However, the potential applications extend beyond non-ferrous metals to the iron and steel industry, submerged arc furnaces, and open bath furnaces used in ferroalloy production and iron making. Our intention is to offer BlueControl as a performance-based service model.