HYBRID NEURAL MODELLING OF FLUIDISED BED DRYING PROCESS

The aim of this study was to investigate the applicability of hybrid neural models in modelling of drying process. A study aimed at extending a neural network mapping was also carried out. In this approach dimensionless numbers (Re, Ar, H/d) were used as inputs to predict the heat transfer coefficient in a fluidised bed drying process. To produce a data set necessary to train the networks, trials of drying different materials in a fluidised bed were carried out. On the basis of this network, a hybrid model describing the process of drying in a fluidised bed dryer was built. Results obtained were compared not only with available experimental data but also with results obtained using other types of models: a pseudo-dynamic neural model and a classical mathematical model. The analysis of results leads to a conclusion that hybrid models constitute a solid alternative method of process modelling.     

EXTENSION OF THE NEURAL NETWORKS OPERATING RANGE BY THE APPLICATION OF DIMENSIONLESS NUMBERS IN PREDICTION OF HEAT TRANSFER COEFFICIENTS

The paper presents a study aimed at extending the neural network mapping ability. In traditional modelling, operational process parameters (gas/material temperature, air velocity, etc.) are the inputs and outputs to and from the network. In this approach dimensionless numbers (Re, Ar, H/d) were used as inputs to predict the heat transfer coefficient in a fluidised bed drying process. To produce the data set necessary to train the networks, drying trials of different materials in a fluidised bed were carried out.

A series of simulations were performed and several neural networks structures were tested to find an optimal topology of the network. Training data set contained information only about two materials. The networks were tested using data obtained for the third product.

Performance of the network was satisfactory, however further improvement of mapping ability may be expected after filtration of the testing data.     

EXTENSION OF THE NEURAL NETWORKS OPERATING RANGE BY THE APPLICATION OF DIMENSIONLESS NUMBERS IN PREDICTION OF HEAT TRANSFER COEFFICIENTS

ABSTRACT

The paper presents a study aimed at extending the neural network mapping ability. In traditional modelling, operational process parameters (gas/material temperature, air velocity, etc.) are the inputs and outputs to and from the network. In this approach dimensionless numbers (Re, Ar, H/d) were used as inputs to predict the heat transfer coefficient in a fluidised bed drying process. To produce the data set necessary to train the networks, drying trials of different materials in a fluidised bed were carried out.

A series of simulations were performed and several neural networks structures were tested to find an optimal topology of the network. Training data set contained information only about two materials. The networks were tested using data obtained for the third product.

Performance of the network was satisfactory, however further improvement of mapping ability may be expected after filtration of the testing data.     

Moisture Prediction During Paste Drying in a Spouted Bed

The objective of this work was to derive and experimentally verify a hybrid CST/neural network model to determine the moisture content of the powders produced during paste drying in a spouted bed and describe the highly coupled heat and the mass transfer. The model was derived from overall energy and mass balances with effective drying kinetics given by a neural network. Simulations were performed in MatLab and drying experiments for model verification were carried out for different pastes in a conical, semi-pilot-scale spouted bed.     

HYBRID NEURAL APPROACHES FOR MODELLING DRYING PROCESSES FOR PARTICULATE SOLIDS

ABSTRACT

This work presents methods for synthesizing drying process models for particulate solids that combine prior knowledge with artificial neural networks. The inclusion of prior knowledge is investigated by developing two applications with the data from two indirect rotary steam dryers. The first application consisted in the modelling of the drying process of soya meal in a batch indirect rotary dryer, The external and internal mass transfer resistances were associated in the hidden layer of the network to linear and sigmoidal nodes, respectively. The second application consisted in the modelling of the drying process of soya meal in a continuos indirect rotary dryer. The model was constructed using the Semi-parametric Design Approach. The model predicts the evolution of solid moisture content and temperature as a function of the solid position in the dryer. The results show that the hybrid model performs better than the pure “ black box” neural network and default models. They also shows that prior knowledge enhances the extrapolation capabilities of a neural network model,     

HYBRID NEURAL APPROACHES FOR MODELLING DRYING PROCESSES FOR PARTICULATE SOLIDS

This work presents methods for synthesizing drying process models for particulate solids that combine prior knowledge with artificial neural networks. The inclusion of prior knowledge is investigated by developing two applications with the data from two indirect rotary steam dryers. The first application consisted in the modelling of the drying process of soya meal in a batch indirect rotary dryer, The external and internal mass transfer resistances were associated in the hidden layer of the network to linear and sigmoidal nodes, respectively. The second application consisted in the modelling of the drying process of soya meal in a continuos indirect rotary dryer. The model was constructed using the Semi-parametric Design Approach. The model predicts the evolution of solid moisture content and temperature as a function of the solid position in the dryer. The results show that the hybrid model performs better than the pure “ black box” neural network and default models. They also shows that prior knowledge enhances the extrapolation capabilities of a neural network model,     

Investigation into gas-solid heat transfer in a cryogenic vibrated fluidised bed

Experiments were carried out in a cryogenic vibrated fluidised bed to investigate the heat transfer between gas and rubber particles obtained from discarded tyres. The effects of parameters such as bed layer thickness and gas flow rate on the gas-solid heat transfer were investigated, and a heat transfer correlation obtained by regressing the experimental data. Theoretical analysis based on radial thermal conductivity indicated that higher heat transfer efficiency could be obtained by the use of a fluidised bed rather than a fixed bed or a moving bed, especially for rubber particles having low thermal conductivity under cryogenic conditions. A numerical modelling was developed, based on assumptions of the movement of the particles and the vibrating bed plate, using a unique method of regarding particles as the source term in the energy equation. Computational results from the modelling showed good agreement with the experimental data.     

基于LSTM神经网络的流化床干燥器内生物质颗粒湿度预测

生物质作为一种储量丰富、环境友好且易于获取的可再生能源，日渐成为能源研究利用领域的热点。生物质湿度是影响生物质利用效率的关键因素，因此干燥是生物质利用之前的必要步骤。流化床由于其良好的传热传质特性，在干燥过程中得到了广泛的应用。为了实时监测生物质颗粒的干燥过程，利用弧形静电传感器阵列，结合用于时间序列建模的长短期记忆（LSTM）神经网络，实现了流化床干燥器内生物质颗粒湿度的预测。在实验室规模的流化床干燥器上进行了多工况实验获取训练和测试数据，通过模型参数优化确定了LSTM模型。通过与标准循环神经网络（RNN）模型的预测结果的对比表明，LSTM神经网络模型的平均相对误差较小，能够较为准确地预测流化床干燥器内生物质颗粒的湿度。     

Batch top-spray fluid bed coating: Scale-up insight using dynamic heat- and mass-transfer modelling

A mathematical model was developed for batch top-spray fluid bed coating processes based on Ronsse et al. [2007a, b. Combined population balance and thermodynamic modelling of the batch top-spray fluidised bed coating process. Part I—model development and validation. Journal of Food Engineering 78, 296-307; Combined population balance and thermodynamic modelling of the batch top-spray fluidised bed coating process. Part II—model and process analysis. Journal of Food Engineering 78, 308-322]. The model is based on one-dimensional discretisation of the fluid bed into a number of well-mixed control volumes. In each control volume, dynamic heat and mass balances were set up allowing the simulation of the contents of water vapour, water on core particles and deposited coating mass as well as fluidisation gas, particle and chamber wall temperature. The model was used to test different scale-up principles by comparing simulation results with experimental temperature and humidity data obtained from inorganic salt coating of placebo cores in three pilot fluid bed scales being a 0.5 kg small-scale (GEA Aeromatic-Fielder Strea-1), 4 kg medium-scale (GEA Niro MP-1) and 24 kg large-scale (GEA MP-2/3). Results show good agreement between simulated and experimental outlet fluidisation air temperature and humidity as well as bed temperature. Simulations reveal that vertical temperature and humidity gradients increase significantly with increasing scale and that in fluid beds as the simulated 900 kg (RICA-TEC Anhydro) production-scale, the gradients become too large to use the simple combined drying force/relative droplet size scale-up approach without also increasing the inlet fluidisation air temperature significantly. Instead, scale-up in terms of combinations of the viscous Stokes theory with simulated particle liquid layer profiles (obtained with the model) is suggested. In this way, the given fluid bed scale may be optimised in terms of low agglomeration tendency for a given process intensity across scale.     

EFFECT OF TEMPERING ON SUBSEQUENT DRYING OF PADDY USING FLUIDISATION TECHNIQUE

Drying and tempering models for paddy drying by a fluidised bed technique have been developed to describe the moisture movement inside a single paddy kernel. The grain shape was considered as a finite cylinder. The internal diffusion is an important contribution to control the drying rate of paddy. The dependence of effective diffusion coefficient on drying temperature can be adequately explained based on Arrhenius form. The parameters of this equation were evaluated in the range of temperatures between 110°C and 170°C by using the regression analysis with 189 experimental drying data. As compared with no tempering, the faster drying rate can be obtained by tempering treatment between drying stages. The effect of degrees of tempering on determining the moisture reduction in the second stage has also been explored. According to the simulation results, a prediction equation of the required tempering time for the tempering index of 0.95 has been established in which the drying air temperature, initial moisture content and drying time are taken into account. The tempering time for 35 min is recommended for the continuous fluidised bed dryers being operated in rice mills.     

A comparative hydrodynamic study of two types of spouted bed reactor designs

In order to properly design and scale up spouted beds, one needs to predict the minimum spouting velocity of specific systems having different bed dimensions, and properties of particle and spouting gas. Because of inherent complexity of predicting minimum spouting velocity, the prevailing approach has been to use empirical correlations, a number of which are available in the literature. Central jet distributors are commonly used in the experimental studies reported in the literature. Circular slit distributor is a new concept in which air is supplied to the bed of particles through a circular slit. This paper presents results of an experimental study on the hydrodynamics of central jet and circular slit distributors. In this paper a fully connected feed-forward neural network model was used to predict the minimum spouting velocity of central jet and circular slit spouted beds. A neural network model was also developed to predict minimum fluidization velocity. The actual experimental data obtained from published literature and from the experiments carried out in this study were used for training and validating the models. The minimum spouting and fluidization velocities predicted using the neural network models developed in this study show a better approximation to the actual experimental values than those obtained from correlations available in the open literature. The position of flow regime of circular slit spouted bed was also established relative to the flow regimes of central jet spouted bed and fluidized bed.     

Control of a Batch Polymerization System Using Hybrid Neural Network ‐ First Principle Model

In this work, the utilization of neural network in hybrid with first principle models for modelling and control of a batch polymerization process was investigated. Following the steps of the methodology, hybrid neural network (HNN) forward models and HNN inverse model of the process were first developed and then the performance of the model in direct inverse control strategy and internal model control (IMC) strategy was investigated. For comparison purposes, the performance of conventional neural network and PID controller in control was compared with the proposed HNN. The results show that HNN is able to control perfectly for both set points tracking and disturbance rejection studies.     

基于气固流态化原理的油页岩干燥动力学

为了考察气固流化床干燥器能否使油页岩含水质量分数达到要求,以柳树河油页岩颗粒为原料,研究进口气体温度和颗粒直径对油页岩干燥性能的影响,采用薄层干燥模型,对油页岩干燥实验数据进行模拟,确定油页岩干燥方程和干燥速率方程,建立油页岩干燥速率特征常数和有效扩散系数之间的关联式。研究结果表明:薄层干燥模型中修正Page模型Ⅰ适合描述油页岩的干燥过程;油页岩在流化床内干燥过程主要发生在降速干燥阶段,进口气体温度越高,油页岩颗粒直径越小,所需干燥时间越短,进口气体温度为350℃时,使2.4 mm油页岩含水质量分数低于5%,所需干燥时间为2.5 min。     

Modelling of batch fluidised bed drying of pharmaceutical granules

This paper presents a mathematical model based on a three-phase theory, which is used to describe the mass and heat transfer between the gas and solids phases in a batch fluidised bed dryer. In the model, it is assumed that the dilute phase (i.e., bubble) is plug flow while the interstitial gas and the solid particles are considered as being perfectly mixed. The thermal conductivity of wet particles is modelled using a serial and parallel circuit. The moisture diffusion in wet particles was simulated using a numerical finite volume method. Applying a simplified lumped model to a single solid particle, the heat and mass transfer between the interstitial gas and solid phase is taken into account during the whole drying process as three drying rate periods: warming-up, constant rate and falling-rate. The effects of the process parameters, such as particle size, gas velocity, inlet gas temperature and relative humidity, on the moisture content of solids in the bed have been studied by numerical computation using this model. The results are in good agreement with experimental data of heat and mass transfer in fluidised bed dryers. The model will be employed for online simulation of a fluidised bed dryer and for online control.     

Modelling of A Fluidized Bed Drier Using Artificial Neural Network

ABSTRACT

Proper modelling of a fluidized bed drier (FBD) is important to design model based control strategies. A FBD is a non-linear multivariable system with non-minimum phase characteristics. Due to the complexities in FBD conventional modelling techniques are cumbersome. Artificial neural network (ANN) with its inherent ability to “learn” and “absorb” non-linearities, presents itself as a convenient tool for modelling such systems.

In this work, an ANN model for continuous drying FBD is presented. A three layer fully connected feedfordward network with three inputs and two outputs is used. Backpropagation learning algorithm is employed to train the network. The training data is obtained from computer simulation of a FBD model from published literature. The trained network is evaluated using randomly generated data as input and observed to predict the behaviour of FBD adequately.     

Intelligent fitting of minimum spout‐fluidised velocity in spout‐fluidised bed

The experiments were carried on to study the minimum spout‐fluidised velocity in the spout‐fluidised bed. It was found that the minimum spout‐fluidised velocity increased with the rise of static bed height, spout nozzle diameter, particle density, particle diameter, fluidised gas velocity but decreased with the rise of carrier gas density. Based on the experiments, least square support vector machine (LS‐SVM) was established to predict the minimum spout‐fluidised velocity, and adaptive genetic algorithm and cross‐validation algorithm were used to determine the parameters in LS‐SVM. The prediction performance of LS‐SVM is better than that of the empirical correlations and neural network.     

Optimal quality control of baker's yeast drying in large scale batch fluidized bed

Optimal quality control of drying process of baker's yeast in large scale batch fluidized bed dryer is presented using neural network based models and modified genetic algorithm (GA). The objective of this study is to determine optimal conditions to maximize product quality while minimizing energy consumption. For this purpose, the drying process and quality models based on neural network with delay units are combined for predicting the dry matter, product temperature, change in dry matter and the quality loss while minimizing energy consumption and this model is then used for optimal quality control. A stochastic method based optimization structure is designed in order to solve the optimization problem whose the objective function is discontinuous, non-differentiable, complex and highly non-linear. The results obtained by optimal quality control based on modified GA showed that the performance of the existing industrial scale drying process was improved. The constructed optimal quality control structure is very convenient for the production process applications and may be applied without too much modification.     

Modelling of a woodchip refiner using artificial neural network

Neural networks can be an attractive alternative to mathematical modelling of complex and poorly understood processes if input/output data can easily be obtained. Woodchip refining falls into this category. The mechanism of the refining process is still being studied and no thorough models have yet been developed. A feed-forward neural network is proposed for modelling of woodchip refiners. The outputs predicted by the neural network are compared with industrial refiner data. It is also shown that a modified neural network structure can be used to optimize refiner operation and product quality. The advantages and disadvantages of neural network model application in simulation and optimization of industrial processes are discussed.     

EFFECT OF DRYING ON FLUIDISATION BEHAVIOUR OF PACKED BEDS OF SHREDDED COCONUT

Drying of shredded coconut is usually carried out commercially in order to facilitate storage over reasonable periods of time and to obtain advantages of reduction of weight and volume in transport and packaging. Fluidised bed drying of materials is generally accepted to be an efficient method of drying. Experiments were carried out to investigate the behaviour of fluidisation of shreddet coconut at various moisture contents. The pressure drop was measured across random packings of shredded coconut. It is seen that shredded coconut does not fluidise easily at moisture contents greater than 0.55 (55 weight percent moisture). It is also seen that fluidisation can be easily achieved by drying shredded coconut to moisture contents between approximately 0.25 to 0.55. The pressure drop characteristics within this region is seen to closely resemble the theoretical behaviour of a fluidised bed. It is also seen that particles of shredded coconut in beds of moisture contents less than 0.25 tend to undergo pneumatic transport if efforts are made to fluidise such beds.