In this paper, Artificial Neural Network (ANN) was used for modeling the nonlinear structure of a debutanizer column in a refinery gas process plant. The actual input-output data of the system were measured in order to be used for system identification based on root mean square error (RMSE) minimization approach. It was shown that the designed recurrent neural network is able to precisely predict and track the response of the actual system. The comparison between the results of this paper and those of the most recent published studies as NARX model indicates the significance of the proposed approach.

In this paper, an exergy analysis approach is proposed for optimal design of distillation column by using simulated annealing algorithm. First, the simulation of a distillation column was performed by using the shortcut results and irreversibility in each tray was obtained. The area beneath the exergy loss profile was used as Irreversibility Index in the whole column. Then, First Optimization Algorithm (simulated annealing, SA) was implemented to Grassroots (Number of tray (N) vs. Reflux Ratio (RR)) and Retrofit (Nof vs. Feed splitting) cases, respectively. Next, SA was used to find the maximum recovery in a simple column by seven different variables (Feed Temperature, Feed Pressure, Reflux Rate, Number of theoretical stage, Feed Trays (Feed Splitting, three variables)) simultaneously. During the search for maximum recovery, it was tried to find a better Irreversibility Index. In the second part, SA optimization algorithm was used for a complex column with one pump-around and feed splitter to find a better condition, which means to find the best location for pump-around and feed trays in Distillation column. The main objective in SA was to maximize the recovery of the desired component and to find a better minimum Irreversibility Index. This method was implemented in de-ethanizer; in the first optimization without using pump-around with seven degrees of freedom, Recovery growth was 5.1% and reduction in irreversibility index was 3%. At the best Irreversibility Index, growth of recovery was about 3.7% and irreversibility index reduction was 25%. In the second optimization with pump-around or eight different variables, in the best condition, Recovery Growth was 6.2% but had a very high Irreversibility Index. At the best irreversibility index, recovery reduction was 17% but reduction in irreversibility index was about 21% comparing with initial point. As a result, it is shown that, regarding recovery and Irreversibility, pump-around shouldnt be used in a column. Without using pumparound, a better condition, considering both factors, can be achieved.

Typical production objectives in distillation process require the delivery of products whose compositions meet certain specifications. The distillation control system, therefore, must hold product compositions as near the set points as possible in faces of upset. In this project, inferential model predictive control, that utilizes an artificial neural network estimator and model predictive controller, is developed for an industrial multicomponent distillation column. First, composition control by direct measuring composition is used. This method because of large sampling delay has a poor performance. The selection of the temperature measurement points is done for indirect control of the column. The use of temperature loop leads to an offset in the composition, due to the fact that the temperature set-point must be changed when feed disturbances occurred. An artificial neural network estimator is designed to estimate the product compositions from tray temperature measurements. A model predictive controller is used to control column composition based on composition estimates. The performance of the developed inferential model predictive control system is tested for set-point tracking and load rejection.

In general, the objective of distillation control is to maintain the desired products quality. In this paper, the performances of different one point control strategies for an aromatic distillation column have been compared through dynamic simulation. These methods are: a) Composition control using measured composition directly. This method suffers from large sampling delay of measuring devices. b) Composition control by controlling the temperature of a specific tray. In this strategy, the composition-temperature relationship is used to find the temperature set point corresponding to the desired composition. Since composition-temperature relation depends on feed condition, an artificial neural network has been proposed which receives the feed specifications and provides the setpoint of the temperature control loop. c) Using temperature measurements for predicting the composition and controlling the composition based on predicted values of composition (inferential contol). Simulation results indicate that controlling the 8th tray temperature and using an artificial neural network for calculating corresponding tray temperature setpoint, has the best performance. Due to negligible pressure drop along the column, controlling the tray temperature difference does not improve the control loop performance.

The simulation of distillation columns is an essential step in design, optimization, and rating. In this paper, a new procedure has been proposed for the initial estimation of column profiles based on modified Kremser’s group method for simple and/or complex columns. The effect of this initialization algorithm on simulation procedure has been studied through two examples. The results show significant improvement in convergence and considerable saving in execution time up to 30%. The proposed method can be used in problems which require numerous simulation runs to find the optimum conditions such as distillation column sequencing and the optimum design of distillation columns. The proposed method could also be used as an approximating simulation method for simple and complex distillation columns considering constant relative volatility and constant molar overflow assumptions.

Sieve trays are widely used in the gas- liquid contactors such as distillation and absorption towers. In this article, a three-dimensional, two phase CFD model using Euler-Euler framework was developed to simulate a distillation tower with two sieve trays. Hydrodynamic simulation of air and water system in different rates of gas phase was carried out and velocity distribution parameters, clear liquid height and froth height were calculated and compared to the experimental data and the literature simulation result. Liquid velocity distributions on the two trays were found to be in relatively good agreement with experimental data. It was found that heights of accumulated liquid in the down-comers are not equal.

Distillation control is a challenging undertaking given the inherent nonlinearity of the process severe coupling present for dual-composition control and the sensitivity of disturbances. Among various distillation operations management of the high-purity column poses a difficult control situation due to a number of characteristics of these systems. Including strong directionality. Ill-conditioning and strongly nonlinear behaviour. In this paper a diagonal PID controller is designed and analysed for a high-purity distillation column by computing the structured singular value  m introduced by Doyle (1982) For this purpose a structured uncertainty model has been developed which describes the dynamics of the column for the entire operating range. The achievable manageable performance is also defined in terms of the H ¥-norm of the weighted sensitivity function.