Simulations of Structures in Packed Columns and Validation by X-ray Tomography

Abstract

Packing simulations of generic, nonspherical pellets were performed and compared with experimental data sets obtained using X-ray computerized tomography (CT). Two modified versions of what was previously a purely geometrical, digitally based packing algorithm were implemented. Both are aimed at incorporating the effects of particle interaction forces, one utilizing the distinct element method (DigiDEM) and the other an intermediate solution (collision-guided packing or DigiCGP). This article summarizes the models and the simulations performed using these two modified versions of DigiPac and, for model validation purposes, compares the predicted results with the corresponding X-ray tomographic scans of packed columns, in terms of bulk density, local packing density profiles, and pellet orientation distributions. For packed beds of relatively large and identical pellets, the simulation results indicate that particle-particle and particle-wall interactions cannot be ignored if realistic packing structures are to be obtained by simulation and even a simplistic treatment of these interactions can produce significantly more realistic packing structure than none at all.

Validation of a novel open-source work-flow for the simulation of packed-bed reactors

“…In this way we will perform CFD simulations only in the central region of the packing, in order to consider just the bulk structure of a packed bed reactor, while still considering the (lateral) wall effects. This same rationale was used in the creation of grain packings in [43]. In order to compare our findings with experimental results found in the literature, the geometrical features (ratio between grain diameter and container diameter) and operating conditions (Reynolds numbers) of the simulations of this full packing exactly replicate those found in [50].…”

“…These specifications were chosen so as to precisely replicate the experimental setup used for comparison [43]. In these experiments the orientation of the cylinders was measured and the frequency of distribution of the angle formed by the axis of the cylinders and that of the container evaluated.…”

“…This is particularly important because these exotic shapes, such as trilobes, are becoming more and more popular [27][28][29][30][31][32][33][34][35][36][37][38][39][40]. Other interesting alternatives are based on the use of Monte Carlo methods [41][42][43] that solve some of the above mentioned issues.…”

Advances in shape measurement in the digital world

“…In favourable conditions, what used to be days of hard labour can now be reduced to a few mouse clicks and minutes of machine time! If particle shape is known, but extraction of the shape is required for other analyses (e.g., orientation distribution of certain shape in a packing, as in Figure 4b), a Markov-chain Monte Carlo template matching technique may be used (Caulkin et al, 2009). In packing of multiple shapes, template matching may be performed multiple times, each targeted at a particular shape.…”

“…More examples can be cited, including effects of particle shape on conductivity ( Computer models, particularly those that are designed to handle different particle shapes, can be used for both passive and proactive applications. One example is optimisation of geometry of pellets used in packed column reactors (Caulkin et al, 2009). To be cost effective and time efficient, such uses usually require that the link between shape and quality, by whatever measure is used, is known or can be established by computer models rather than having to rely on making and testing new physical shapes.…”