Assessment of CFD phase models for simulating iron combustion in retrofitted coal combustion chambers

Kurzfassung

The idea of using metals as energy carriers, by first oxidising metal particles to release energy and then reducing the oxides formed to begin the cycle anew, has recently gained significant traction [1-4]. Among the numerous metals present in the Earth’s crust, iron is a particularly promising candidate, partially due to its low costs and high volumes of production, but also due to its peculiar mode of oxidation [5]. As a result, interest in iron oxidation has increased over the past few years [6,7]. Implementing this idea at the largest scale would require retrofitting coal-fired power plants to instead run on iron, an option currently undergoing scrutiny [8]. Alongside experimental investigations, researchers are also employing numerical methods for iron particle combustion [9,10]. Computational Fluid Dynamics (CFD) calculations can resolve the flow mechanics, kinetics and thermodynamics of the iron combustion process within a single simulation. CFD will undoubtedly play a large role in simulating coal combustion chambers retrofitted to burn iron powder; however, it has its limitations and assessing the usefulness of the various possibilities present within CFD requires examining them carefully. In this work, we will compare and contrast various CFD phase models, gauging their suitability for iron combustion simulations. After a brief overview of both the Eulerian and the Lagrangian flow models, we will detail how these may be coupled for particulate flow. We will initially tackle the Euler-Lagrange approach and show that while it handles single particle and small particle number (N_p) simulations well, the complexity increases rapidly at the N_p scales present within a combustion chamber. Next, we will turn our attention to the Euler-Euler approach. Traditionally, this approach is used to model flow with large N_p, with the caveat that the field representation may leave inter-particle physical effects unresolved. We will discuss the order of magnitude of these effects and the consequences of their exclusion on the overall particle behaviour. We will also detail the outcome of the field representation on the overall complexity of the simulation and its run-time. Furthermore, we will show that although the approach itself is mature (in a scientific sense), the same cannot be said for all implementations of it, with the OpenFOAM v11 implementation in particular lacking certain features that would otherwise make it the most suited towards simulating iron combustion at large scales. Lastly, we will take the potentially surprising step of looking at single-phase Eulerian flow. As with the other approaches, we will discuss both its advantages and its drawbacks. However, we will also detail why we deem it the best approach for simulating coal combustion chambers retrofitted to burn iron, and shall in support of this conclusion present initial results obtained in this way for a Bunsen burner [11] mesh. With this, we hope to lay the foundation towards more work being carried out on this topic and provide an impetus towards exploring this non-traditional usage of a single-phase flow model.