Modeling of injection zone in blast furnaces
Chih-Chia Huang defended his PhD thesis at the department of Mechanical Engineering on September 19th.
Blast furnaces (BF) have been essential for ironmaking, producing over 70% of the world's iron. For his PhD research Chih-Chia Huang focused on modern BFs where pulverized coal is injected to reduce coke consumption, stabilize furnace operation, provide heat, and decrease carbon footprint. A region known as the ‘raceway’ is key for iron production efficiency. However, direct measurement within the raceway is challenging due to extreme conditions. A combined model using Computational Fluid Dynamics (CFD) and the Discrete Element Method (DEM) is developed to understand the underlying physics. The findings contribute to the potential improvement of blast furnace designs, enhancing efficiency, and reducing the carbon footprint of the ironmaking industry.
Until now, blast furnaces (BF) have been the most crucial ironmaking process. More than 70% of the iron consumed worldwide is produced via BF. In modern blast furnaces, pulverized coal is injected with hot blast air. The momentum thrust brought by this high-speed air pushes coke particles away and creates a void region, called ‘raceway’, nearby the injection zone. The introduction of pulverized coal not only provides heat and reducing agents but also complements the usage of coke. The combustion performance of the gas-coke-coal flows in the raceway is thus crucial for the efficiency of iron production. Direct measurement of such performance is almost impossible due to the harsh conditions in the injection zone, i.e. high temperature and pressure. This project resorts to numerical modeling and tries to unveil the complex underlying physics in the raceway.
Understanding blast furnace dynamics
At the injection level of blast furnaces, coke particles' collision and friction and their interaction with the blast air determine the dynamics of the raceway. Inside the raceway, the combustion of pulverized coal produces carbon dioxide that will further react with the coke to produce carbon monoxide, the most essential ingredient in the blast furnace. A model that uses Computational Fluid Dynamics (CFD) coupled with the Discrete Element Method (DEM) is developed as it is able to capture all the mentioned physics. CFD is a versatile method as it can simulate from compressible to incompressible flows with added features like combustion or turbulence. DEM treats particles as soft spheres and can resolve particles' motion, collision, rotation, and friction. The combination of CFD and DEM incorporates forces, like drag force, to bridge the gas and solid phases.
Enhancing computational efficiency
Despite the ever-increasing computational capacities brought by the fast advancement of semiconductors, modeling industrial multiphase particulate systems using CFD-DEM is still challenging due to the involvement of billions of particles and complex reactions. To tackle these two challenges, Huang first developed a model that incorporates the Coarse-Graining (CG) method, which reduces the computational cost by replacing particles with fewer representative parcels, with the Flamelet Generated Manifold (FGM) method, which reduces the computational cost by looking up the combustion information from a pre-built database instead of calculating during the simulations. Overall, the computational cost is reduced by a factor of 2.3 when the CG factor of 2 is used, and it produces an accurate prediction of particle dynamics and oxygen concentration.
A new approach
During the validation study, it was noticed that the state-of-the-art diffusion-based smoothing technique is not designed for polydisperse particulate systems. It distributes the multiphase exchange information from the location of particles homogeneously regardless of particle size. If, for instance, the heat transfer from particle to gas is identical for a large and a small particle, the heat will be smoothed to the same distance for both particle sizes. Their energy contribution will then be indistinguishable. Huang proposed a particle-size dependent smoothing scheme, Non-Constant Diffusivity Smoothing (NCDS) approach, for polydisperse Euler-Lagrange systems. This approach distributes the exchanged information based on the local Sauter diameter such that the area of influence per particle varies by size.
Improvement blast furnace design
Huang applied the developed model to the injection zone of an industrial-scale blast furnace. He observed that the physical and chemical raceway, defined by the boundary where oxygen levels drop below 5%, do not overlap, indicating ineffective oxygen utilization. Altering the lance angle didn't improve oxygen or pulverized coal utilization due to poor mixing. The momentum of high-speed blast air primarily shapes the physical raceway, while pulverized coal expansion doesn't impact its size or shape. Additionally, particle size segregation and a crescent-shaped jet plume were observed due to the lance's bluff-body effect. These insights can improve blast furnace design and enhance pulverized coal combustion efficiency, reducing the ironmaking industry's carbon footprint.
Title of PhD thesis: . Supervisors: Niels Deen, Jeroen van Oijen and Yali Tang.