Wall effects in particle-laden flows.
[Tesi di dottorato]
|Tipologia del documento:
Tesi di dottorato
||Wall effects in particle-laden flows
||30 Novembre 2011
|Numero di pagine:
||Università degli Studi di Napoli Federico II
|Scuola di dottorato:
|Ciclo di dottorato:
|Coordinatore del Corso di dottorato:
||30 Novembre 2011
|Numero di pagine:
||wall effects; entrained-flow gasification; multiphase reactors; particle-slag interactions;, multilevel modeling;
multiphase flow; particle-laden flow; turbophoresis;
|Settori scientifico-disciplinari del MIUR:
||Area 09 - Ingegneria industriale e dell'informazione > ING-IND/25 - Impianti chimici
||13 Dic 2011 10:58
||30 Apr 2014 19:48
The aim of this study is to investigate about the complex phenomenology associated with the interaction of a particle-laden turbulent flow with the slag-covered wall of an entrained-flow gasifier.
Recent observations, indeed, highlighted that this phenomenology can have an impact on the global gasifier performance greater than that expected from previous analyses.
The design of new generation of entrained-flow coal gasifiers aims at favoring ash migration/deposition onto the reactor walls, whence the molten ash (slag) flows and is eventually drained separately at the bottom of the gasifier.
In terms of efficiency, the oxidation of the volatile compounds released around the particles depends upon its mixing with the fresh oxidant mixture.
Therefore combustion efficiency is influenced by the spatial distribution of the particle phase, with an homogeneous distributions favoring a better mixing. From the observation that a significant number of coal particles can spent most of the time in the gasifier close to the slag layer, where usually their concentration largely increase, leads to the need to understand the effective conditions experienced before complete conversion.
An experimental evidence of a picture for the fate of coal particles has been recently assessed by analyzing the chemical composition of samples of coarse slag and slag fines generated in the ELCOGAS entrained-flow gasifier located in Puertollano, Ciudad Real (Spain). Quantitative SEM-EDX analysis of the coarse slag revealed the presence of small marks with a significant carbon content as high as 48.8%-54.2%. This fact can be explained by assuming the entrapment of not fully burned coal particles into the slag.
The results of the SEM analysis performed on whole slag fines particles showed that the carbon content was larger than the value obtained from the inspection of coarse slag particles.
This is particularly evident for porous particles where C-content ranged between 82.3% and 86.5%.
A considerable amount of unreacted coal is therefore entrapped into the slag matrix.
From this observations emerges that a certain level of spatial non homogeneity of the solid phase distribution exists. In a recently published study by Montagnaro and Salatino (2010), these data have been interpreted by assuming that different regimes of particles-slag interaction can occur: either char entrapment inside the melt or carbon-coverage of the slag may occur, depending on properties like char density, particle diameter and impact velocity, slag viscosity, interfacial particle-slag tension.
Occurrence of char entrapment prevents further progress of combustion/gasification.
On the contrary, if char particles reaching the wall adhere to the slag layer's surface without being fully engulfed, the progress of combustion/gasification is still permitted. The observed high rate of coal conversion can actually be explained only if this second regime establishes on the slag surface.
The addressed considerations highlights the technological need to build up methods for the prediction of the mechanism particles clustering and segregation in condition representative of coal particle flying and converting into a gasifier.
Actually a comprehensive numerical simulation of the whole range of spatial and temporal chemical and turbulent time scales involved in a full scale gasifier, is still unfeasible due to the high computational cost: the scales of turbulence involved in the gasification processes range from sub-micron scale up to the integral scale of a gasifier reactor chamber (of the order of tens of meters). To overcome this difficulty, the approach proposed in this study is based on the development of a multilevel approach..
In a first level, the motion of particles representing classes of partially converted coal in a 3-dimensional representation of the gasifier is modeled with a Computational Fluid Dynamic (CFD) approach..
Turbulence of the flow field is described adopting the Reynolds Averaged Navier Stokes (RANS) approach, while particle motion is resolved with a Lagrangian Particle Tracking (LPT) approach. The use of the RANS method for the gas phase coupled with the LPT for the solid phase in this analysis is twofold.
Firstly it has been used to address the behavior of coarse and fine coal particles trajectories when subjected to a swirl motion which induced a turbulent field. This model, while avoiding the great complexity and computational effort required by comprehensive numerical CFD models of gasifiers already proposed in the literature, is sufficient to characterize the range of conditions, in terms of momentum possessed and direction, that the different particles show when approaching the gasifier walls.
The second aspect concerns the identification of regions where different mechanisms for the coal clustering becomes foreseeable: distinct regions close to the wall have been identified: finer particles could be mainly responsible of particle layering near the solid walls as they, after their first impinging on the wall, assumes a pathway parallel to the wall; in contrast, larger particles continue to bounce over the walls along the whole length of the gasifier.
The identification of these two different regions and the characterization of particle classes representative of partially burned coal particles, was the basis for the proper setup of numerical simulations based on a Large Eddy Simulation (LES) approach in two completely different configurations. This level aims at a detailed investigation of the mechanisms of slag-particle interaction. The first is a plane particle-laden channel flow, that well represents the main features of the gasifier regions where particles move parallel to the wall. The second is a periodic particle laden curved channel flow, that best represent regions close to the wall but dominated by the external swirling flow.
For these two configurations the particle interaction with the slag has been treated as a rebound on a not perfectly elastic wall. A parametric study has been conducted obtaining results for different particle sizes (representing different particle inertia) and different momentum restitution in the particle-wall impact.
Numerical multiphase simulations are based on the Eulerian-Lagrangian approach implemented in the OpenFOAM CFD framework.
Both RANS and LES turbulence models are implemented for the gas phase. The equations of particles motion were solved via a Lagrangian particle tracking algorithm with the TrackToFace method. Simulations were performed involving a number of particles from 10^5 to 10^6, a level of detail that allowed to obtain a clear picture of the multiphase flow behavior responsible for char deposition phenomena.
Numerical simulation results with the LES approach do confirm the establishment of a region near the wall slag layer (the dense-dispersed phase leading to the formation of the slag fines), in which particles impacting the slag accumulate to an extent depending on the system fluid-dynamics and on parameters such as particles Stokes number and restitution coefficient.
However, particle concentration near the wall in all the simulated cases does not appear perfectly steady not evenly spatially distributed.
Interestingly, the segregation of char particles near the wall is more evident for the curved channel flow geometry and is enhanced for coarser particles, making evident the role played by the effective impact with the slag not recovered by the simpler models adopted in the RANS simulations.
Actions (login required)