1,721,095 research outputs found
The applicability of the pseudo-fluid model to the settling velocity of a foreign particle in a suspension
No full textKinetics study of CO2 absorption in potassium carbonate solution promoted by diethylenetriamine
No full textIn this work, characterization and kinetics of CO2 absorption in potassium carbonate (K2CO3) solution promoted by diethylenetriamine (DETA) were investigated. Kinetics measurements were performed using a stirred cell reactor in the temperature range of 303.15–323.15 K and total concentration up to 2.5 kmol m−3. The density, viscosity, physical solubility, CO2 diffusivity and absorption rate of CO2 in the solution were determined. The reaction kinetics between CO2 and K2CO3 + DETA solution were examined. Pseudo-first order kinetic constants were also predicted by zwitterion mechanism. It was revealed that the addition of small amounts of DETA to K2CO3 results in a significant enhancement in CO2 absorption rate. The reaction order and activation energy were found to be 1.6 and 35.6 kJ mol−1, respectively. In terms of reaction rate constant, DETA showed a better performance compared to the other promoters such as MEA, EAE, proline, arginine, taurine, histidine and alanine
A CFD-DEM study of monocomponent and same size binary-solid beds at incipient fluidization
No full textThis work analyses in detail the transition process from the fixed to the fluidized state for homogenous and binary mixtures of particles differing in density, with different initial configurations. The phenomenon is studied in cylindrical columns fed with water, with the aid of CFD-DEM numerical simulations. With regard to the homogenous case, the well-established knowledge is substantially confirmed. The transition to the fluidized state strongly depends on the initial solid configuration in the case of same density binary-solid mixture. For the segregated case, the experimental behaviour reported by Di Felice and Scapinello in 2010 is closely reproduced, whereas for the initially perfectly mixed case, the contemporary presence of a fixed and a fluidized region is observed. A simple model that predicts the extent of these two regions as a function of the fluid velocity is proposed
Prediction of silicon content in the hot metal using Bayesian networks and probabilistic reasoning
No full textThe blast furnace is the principal method of producing cast iron. In the production of cast iron, the control of silicon is vital because this impurity is harmful to almost all steels. Artificial neural networks with Bayesian regularization are more robust than traditional back-propagation networks and can reduce or eliminate the need for tedious cross-validation. Bayesian regularization is a mathematical process that converts a nonlinear regression into a "well-posed" statistical problem in the manner of ridge regression. The main objective of this work was to develop an artificial neural network to predict silicon content in hot metal by varying the number of neurons in the hidden layer by 10, 20, 25, 30, 40, 50, 75, and 100 neurons. The results show that all neural networks converged and presented reliable results, neural networks with 20, 25, and 30 neurons showed the best overall results. However, In short, Bayesian neural networks can be used in practice because the actual values correlate excellently with the values calculated by the neural network
A novel committee machine to predict the quantity of impurities in hot metal produced in blast furnace
No full textIn recent years, interest in artificial intelligence and the integration of Industry 4.0 technologies to improve and monitor steel production conditions has increased. Every day, several models are proposed to simulate industrial processes. In this sense, committee machines are presented as competitive alternatives to solve this task. A committee machine uses a multiple classification system in which the responses from multiple ANNs are combined into a single response. In this context, the purpose of this work was to develop committee machines using 108 independent artificial neural networks to predict the amount of impurities (silicon, phosphorus, and sulfur) in the production of cast iron in a blast furnace. It is concluded that neural networks operating in committee mode may be used in practice as a prediction and control tool due to the low RMSE values and high mathematical correlation between the database values and the values calculated by the committee machine
Experimental and numerical study of two-phase flow mixing in gas–liquid external-loop airlift reactor
No full textThe experimental and numerical study of two-phase gas–liquid flows in the external-loop airlift reactor is presented. The effect of riser-to-downcomer cross-sectional area ratio on the mixing hydrodynamic efficiency, gas holdup, and circulation liquid velocity is investigated. The transverse dimensions of vertical columns are 2.5 cm, 5 cm and 10 cm, the height is 2 m long. Low values of gas flow rate are tested (from 0.1 m3/h to 0.6 m3/h). The test rig with an open wide channel connecting the riser and the downcomer in the up, which provides the conditions for complete air release from water phase, is used. A numerical study of gas–liquid flow in the reactor's riser is also performed using the CFD solver Star-ccm+, which uses a three-dimensional mathematical model of two-phase flow based on the two-fluid Euler–Euler method. Both phases are calculated by solving the steady-state Reynolds-Averaged Navier–Stokes (RANS) conservation equations with the high Reynolds number k–ε turbulence model. The aerodynamic drag, shear-lift and added mass forces, and the turbulent dispersion of the flow are considered. The breakup-coalescence effect is also considered by solving the population balance equation. The numerical tool is used to examine the flow regime map of the experimentally measured flows. The flow regimes (bubble and bubble-to-slug) and transitions from the bubble to bubble-to-slug were identified experimentally and numerically using non-dimensional parameters. The formation of “bubble tracks” or chains of bubbles in the riser were experimentally observed at the flow regime which has been identified as bubble-to-slug. Experiments also have shown that the bubbles are carried away by the flow from the riser and the open top channel to the downcomer when the Reynolds number of the flow becomes bigger than 25,000. The liquid circulation rate remains constant with further increase in the gas flow rate. The gas phase holdup shows the lowest values as well as the liquid circulation rate has highest values when the ratio of column's cross-sectional dimensions is 1. Opposite, the highest values of gas holdup as well as the lowest liquid circulation rates is observed when the ratio is 0.25. The Wang correlation, for the drag coefficient, shows the best fit with the experimental data if the flow regime is bubble. The Schiller–Naumann correlation over-predicts values. A simple analytical model built on the balance of friction losses for a two-phase gas–liquid flow shows very good agreement with the experimental values of the gas phase holdup in the riser and the liquid circulation rate. No fitting parameters are used because of simple design of the experimental rig and low gas flow rate values
On the influence of contact models on friction forces in discrete element method simulations
No full textThe discrete element method (DEM) is an important tool to simulate granular systems with high accuracy. Depending on the application, it is often unclear which model is more appropriate for the calculation of collision forces: the Hertzian approach is generally considered more accurate, but it makes the simulation significantly slower than the Hookean one. In this work, these two approaches are compared in two different situations: the stress distribution of static particles in a cylindrical column (DEM) and the onset of water fluidisation for a completely segregated mixture (CFD-DEM). In both cases, particle contact forces are of great relevance to determine the output. It is found that in the first case the Hookean approach does not produce the expected asymptotical stress trend and does not even respond satisfyingly to friction mobilisation. Conversely, in the fluidisation simulations the results are virtually identical, pointing out that the more complex Hertzian approach may be unnecessary in that case
A discrete element method study of solids stress in cylindrical columns using mfix
No full textFriction phenomena play a key role in discrete element method (DEM) modeling. To an-alyze this aspect, we employed the open-source program MFiX to perform DEM simulations of cylindrical vertical columns filled with solid particles. These are still associated with and described by the pioneering model by the German engineer H.A. Janssen. By adapting the program’s code, we were able to gather numerous insights on the stress distribution within the solids. The column was filled with different amounts of solids and, after the system had stabilized, we assessed the pressure in the vertical and radial directions and the distribution of the friction force for all particles. An analysis of the bottom pressure for varying particle loads allowed us to infer that the program can correctly predict the expected asymptotical behavior. After a detailed assessment of the behavior of a single system, we performed a sensitivity analysis taking into account several of the variables employed in the simulations. The friction coefficient and filling rate seem to affect the final behavior the most. The program appears suitable to describe friction phenomena in such a static system
Artificial Neural Networks for Prediction of Hot Metal Production in a Blast Furnace
No full textA blast furnace is a chemical reactor used in the steel industry to produce molten iron or hot metal. The size of this reactor is variable and can be more than 30 m high. It is coated with metal on the outside and refractory material on the inside. The reactor operates at high temperature and pressure. It is fed with coke, iron ore and fluxes in the upper part and air and auxiliary fuels are injected in the lower part. The rising gases react with the descending solids and melt the material. The production of pig iron also produces slag, which is normally used to make cement. This scientific article reports on the successful application of artificial neural networks in pig iron production. The neural network was modelled in MATLAB using 23 operational variables with 100 neurons. The validation of the mathematical model was carried out through statistical tests in the MINITAB software, which ensure the necessary statistical certainty for the validation of its application on an industrial scale
- …
