Guy L'Homme

Local measurements of void fraction and liquid holdup in packed columns using X-ray computed tomography

Abstract A ‘tailor made’ computed X-ray tomographic scanner has been developed as a tool for the analysis of the distribution of gas, liquid and solid phases in packed columns. The very good spatial resolution of the scanner has first been assessed by the imaging of objects of known shape and size, called ‘physical’ phantoms. Images have then been realized on a 0.6 m diameter and 2 m height column packed with Cascade Mini-Ring 1A packing elements, which is a random polypropylene packing designed to be used in absorption columns. The solid phase distribution, leading to the void fraction distribution, has been analyzed on cross-section images of the dry packed column (without any liquid flowrate). The measured value of the bed void fraction is equal to the value provided by the manufacturer. The analysis of the axial profile of void fraction shows that the ‘end effect’ may be neglected, whereas the analysis of the radial profile evidences the existence of a non negligible ‘wall effect’. Images have then been carried out on the packed column irrigated by a liquid flowrate ranging between 0 and 6000 l h−1 (0–6.10−3 m s−1). Hold-up values have been measured in different cross-sections of the column and averaged in order to obtain the total hold-up value in the bed. The computed values are in very good agreement with those reported in the literature for similar packing. The dependence of the liquid hold-up on the liquid superficial velocity can be expressed in terms of a power law. The fitted value of the exponent, equal to 0.65, is in the range of exponent values found in correlations of the literature.

Analysis of electrical resistance tomography measurements obtained on a bubble column

Measurements of gas hold-up obtained by using pressure transducers and an optical probe are compared with those obtained by an electrical resistance tomography method. The use of a neural network, the input of which is the electrical measurements, can lead to quantitative results for the gas hold-up distribution. An interesting qualitative diagnosis of the sparger functioning is observed from reconstructed images.

The use of large scale computer assisted tomography for the study of hydrodynamics in trickling filters

Abstract In this paper, a new experimental technique is proposed, based on X-ray tomography, allowing the analysis of the gas—liquid—solid phase distribution in trickling filters. Tomography refers to the cross-sectional imaging of an object from data collected by illuminating this object from many directions. The experimental set-up allows the scanning of transverse sections of rather large filters: up to 0.8 m diameter and 2 m height can be irradiated. Experiments were carried out both without and with liquid flow. In the absence of liquid, the reconstructed images give a view of the non-uniformity of the solid phase distribution. The images obtained are very complex because they result from the superposition of several distributions with different spatial scales. The existence of a first spatial scale is shown which allows separation of the properties belonging to the individual packing elements from those belonging to the bed scale. When liquid is flowing through the packed bed, the distribution of the liquid hold-up is clearly evidenced, despite the very low liquid film thickness.

Electrical resistance tomography sensor simulations: comparison with experiments

This paper compares experimental results (differences of potential), obtained with a 16-electrode electrical resistance tomography (ERT) device using the adjacent-electrode pair measurement strategy, with finite-element method (FEM) simulations in two and three dimensions. In the first part of the paper, ERT measurements made on a homogeneous medium are compared with two- and three-dimensional FEM simulations. It is shown that the dependence of the electrical field on the axial coordinate cannot be neglected. Consequently, only the three-dimensional FEM simulations are able to fit the experimental results. In the second part of the paper, ERT measurements on static physical phantoms immersed in water evidence that ERT is significantly more sensitive to non-symmetrical distributions than to symmetrical ones. This observation is confirmed by three-dimensional FEM simulations performed on numerical phantoms that mimic the physical ones. In the last part of the paper, an attempt to use three-dimensional simulations to determine the characteristics (diameter and gas content) of a gas core created by injecting a gas flow from a single-orifice gas sparger leads to promising results.

Macromixing versus hydrodynamics in trickle flow columns

The hydrodynamics in trickle flow columns is characterized by parameters such as the liquid flow maldistribution, the partial wetting, the liquid holdup, etc which are difficult to determine experimentally on large scale units. The usual way to avoid these hydrodynamic measurements consists in tracer experiments from which macromixing models may be developed. These models provide some information on the liquid flow pattern. Simple macromixing models such as the dispersed plug flow are often used in order to quantify the differences from ideal states of mixing. These models may be used to determine the effect of the hydrodynamics on the performance of the apparatus. Unfortunately, the representation of the mixing pattern remains too simple to give a realistic description of the liquid flow hydrodynamics. Much more elaborated macromixing models have been reported in the literature, claiming to provide an accurate analysis of the relationships between some characteristics of an RTD curve and the hydrodynamic mechanisms occuring in trickle flow columns. Actually, such analysis is often questionable because of the extraordinary complexity of the liquid flow and the relative simplicity of the macromixing models. For this reason, RTD and hydrodynamic models are often developed independently of each other. We try to reconciliate these two approaches by proposing a detailed analysis of macromixing and hydrodynamic mechanisms and by comparing the information inferred from both types of model

A phenomenological description of trickle- beo reactors application to the hydrotreating of petroleum fractions

Abstract A new phenomenological description of the hydrodynamic and mass transfer processes occurring in a trickle-bed reactor is presented. The proposed model, which account for the random and discontinuous nature of the packed bed, is checked with hydrotreating data previously published in the literature. These data may be correlated using a single wetting parameter introduced an physical ground.

Rational description of trickle flow through packed beds part II: Radial spreading of the liquid

Abstract We investigated the liquid spreading in a packed bed under a trickle-flow regime following two approaches. First, we derived analytical relations aflow to be computed, assuming the liquid spreading to be a homogeneous markovian process, the transition probabilities of which are calculated supposing a l algorithm based on a constrained entropy maximization. We show that the usual radial dispersion coefficient is mainly related to the geometrical aspect mean density flow distribution. This coefficient cannot represent the evolution of local quantities such as liquid flow-rate distributions.

LIQUID FLOW MALDISTRIBUTIONS IN TRICKLE-BED REACTORS

(1980). LIQUID FLOW MALDISTRIBUTIONS IN TRICKLE-BED REACTORS. Chemical Engineering Communications: Vol. 7, No. 6, pp. 377-388.

MODELLING OF THE LIQUID DISTRIBUTION IN A TRICKLE FLOW PACKED COLUMN BASED UPON X-RAY TOMOGRAPHY IMAGES

Abstract X-ray tomography is used to characterise the liquid distribution in a 0.6 meter diameter and 2 meter height column filled with gas-liquid absorption polypropylene packing elements (Cascade Mini-Ring 1A, Glitsch Benelux). The liquid holdup distribution is determined for Uquid flowrates ranging from 1000 l/h to 6000 l/h, that correspond to superficial velocities ranging from I0−3 to 6 10−3 m/s, in several cross sections situated at different heights in the packed column. A characteristic length, corresponding to the smallest scale beyond which the packing properties may be supposed statistically homogenous and isotropic, is determined by means of the autocorrelation function method applied to the images of the solid distribution. Both local and global liquid holdup measurements are satisfactorily modelled by means of a partial wetting model based on a probabilistic approach.

Mathematical modelling of the liquid trickling flow through a packed bed using the percolation theory

Abstract Two phase flow through packed beds is described using a new methodology based on two successive observation levels. The local transport processes are introduced in an elementary transport cell at the first level (particle scale). At the second level the liquid flows structures which result from the random cell clustering (bed scale) are described by the percolation theory an algorithm of the liquid flow simulation is presented and an interpretation of the numerical result, is proposed.