Pressure drop over a packed bed of coke particle mixtures

Abstract

Coke particles are a solid fuel used in blast furnaces, and the packing of these particles inside the enclosed vessel is a typical example of a packed bed reactor. The type of coke used in a blast furnace impacts the packing and bed voidage of the blast furnace. When gas is injected into the blast furnace to reduce the iron oxide to liquid iron, with the help of the carbon in the coke, this characteristic packing is critical in understanding and improving the reaction behaviour of this complex multiphase reactor. The injected gas would experience a pressure drop while passing through the packed bed of coke particles, impacting the overall process and the blast furnace's efficiency. Many factors influence the packing of particles in a reactor or blast furnace, for example, how the particles are handled, the characteristics of the reactor geometry and the attributes of the particle. The particle itself is the most unpredictable due to various parameters that can affect the packing density. The factors include the particle shape and size, roughness, the particle size distribution used in the reactor and the particle coefficient of restitution. Particles form a packing structure when inserted into a reactor and depend on particle characteristics and how individual particles are stacked upon each other. Non-spherical particles form a random packing structure when forming a packed bed. The three main effects particles have on each other during packing are the wedging, wall and loosening effects. The packing structure leads to forming bed voidage, a critical parameter that is used to estimate the pressure drop in a packed bed. Flow through a packed bed has been investigated before and is classified into two categories, the discrete particle model and the pipe flow analogy. The pipe flow analogy models can be used for spherical particles, with the capillary tube model being the most used pipe flow analogy model. The capillary model defines the flow through the packed bed as a pack of straight capillaries which are equally sized. The discrete particle model is used for non-spherical particles and defines the flow through the packed bed as individual particles with their boundary layers. The study focused on predicting the effect of non-spherical particle characteristics on fixed bed morphology to model the pressure drop based on a fundamental approach. This can be done by determining the particle characteristics that influence packed bed voidage of arbitrary‐shaped coke particle mixtures and evaluating the effect of particle and packing characteristics of mixtures based on the measured and modelled pressure drop in a fixed bed. It was found that present bed voidage models that rely only on the pipe diameter to particle diameter or particle diameter to particle diameter are insufficient to model the packing; the model can be improved by including mono-size batch voids as a parameter. The main parameter that

influences bed voidage is the average particle diameter of the bed. The modified Kwan model was the best-fitting models for binary mixtures and the regressed D-optimal Design model was the best-fitting models for ternary mixtures of coke particles but were regressed from models found in literature to fit the coke particle characteristics. Most pressure drop models are developed from the basis of the friction factor but were inefficient in predicting pressure drop for the size range and mixtures of coke particles. Statistical models were created due to no existing model being able to accurately predict the pressure drop for coke particles. It was found that the standard polynomial statistical method was the most accurate for predicting pressure drop across a packed bed of coke particles. Comparing the mostly used literature pressure drop model to the statistical models it was found that the pressure drop to fluid velocity correlation for the Ergun model is given as ΔP ∝ U2 compared to the correlation for the statistical models given as ΔP ∝ U1.441. From using statistical models to predict pressure drop it would be beneficial to integrate particle parameters to the model like sphericity, voidage and average particle diameter. Particle size distribution was seen to have the most significant influence on pressure drop in mixtures compared to sphericity and bed voidage.