Motivation
Conventional steel production comprises a multitude of process steps, including coke- and sinter-making, hot metal production, steel conversion and refinement, casting, and rolling. In some of these steps, the potential for further decarbonization is limited, thus new C-lean processes are being developed and industrially validated. Such a decarbonisation process requires not only technological changes but also huge investments, and new plants will be operated for decades to regain these costs. However, the main boundary conditions for sustainable production, i.e., the availability and prices of the different possible energy carriers with low carbon footprint, are impossible to predict accurately, thus the sustainability of the operation and investments cannot be reliably assessed by static scenario analysis. Therefore, adequate tools are urgently needed for operation planning under fluctuating market conditions and decision-making on options to invest in more sustainable process steps to suppress carbon emissions during all phases of the green transition.
The AgiFlex project aims at developing a general-purpose agent-based computational tool to optimize the sustainability of steelmaking, considering external and internal constraints.
The project applies digital twins for current and future production steps, with transfer functions expressing the relations between input and output material, energy, and CO2 emissions. The agents, which estimate the environmental and economic impact of the units, evolve based on a distributed-agent based optimization algorithm that determines the best operation strategy or the best investment scenario with respect to the transition towards carbon-neutral operation of the plant.
Mission
AgiFlex develops a highly innovative computational tool which is capable to model and optimise complete integrated steelworks to curtail CO2 emissions and fossil fuel and reductant usage by Smart Carbon Usage and Process Integration by partial replacement of coal by hydrogen and advanced management of the energy streams and process gases.
The main focus of the project in on the optimisation of intermediate states with mixed production chains including the Blast Furnace – Basic Oxygen Furnace route and the Direct Reduction-Electric Arc Furnace route.
The breakthrough innovation of AgiFlex is the development of a holistic agent-based tool for process and energy integration that finds the promising operation points and configurations of the steel plant with respect to efficient carbon use. In this approach, the flowsheet of the plant is not a priori fixed. This is a very important feature since state-of-the-art approaches based on fixed flowsheets cannot explore the space of plant configurations, particularly due to new possible points of operation of steel plants under transition where traditional and new unit processes interplay, e.g., by efficient use of gases. The new processes to consider in future steel plants include direct reduction Direct Reduction furnaces, Electric Arc Furnaces and smelters, units of gas production (e.g., water electrolysers) and conversion (e.g., syngas reformers and CO2 electrolysers), and external energy sources (such as electricity, natural gas, and biomass).
Objectives
Methods
AgiFlex applies smart agent-based simulation and optimisation to tackle the challenging problem of minimizing the environmental impact of steelmaking during the transition phase of this industrial sector to sustainable operation.
After an initial phase specifying the requirements on the agents, time and scalability issues and the optimisation strategy, digital twins are developed for steelmaking and ancillary units, gas production and conversion and external energy flows. This will create a repository of units that can be included or excluded in scenario analyses of future steel plants with the goal of minimizing the environmental impact.
The first test stage assesses the performance by comparison with plant data of the industrial partners in a validation phase, where discrepancies or problems detected trigger revisions and improvements of the system. Using the validated states as basepoint scenarios, novel steelmaking and gas processing agents are incorporated for first assessments of the system performance under future scenarios.
The effects of short-term changes in the boundary conditions are also investigated, exploring the potential offered by adaptation to the market. This step can reveal completely new, promising low emission plant configuration and operation strategies, and the external conditions under which these are feasible. The potential offered by flexible operation is identified to mitigate environmental impact and the economic benefits by demand site response.
