Industrial Distillation Column Research Articles

MODELING AND CONTROL OF A TEMPERATURE-BASED HIGH-PURITY DISTILLATION COLUMN*

Industrial distillation columns typically lack adequate on-line instrumentation required for implementing high performance, model-based control systems which depend on composition measurements. The typical practice is to base product quality control schemes instead on tray temperature measurements since these are usually available on-line. While this strategy has been applied with success on a wide variety of distillation columns, some special considerations may be necessary in dealing with high-purity columns. This paper presents results which address some of the key issues involved with the use of tray temperature measurements as surrogates for composition measurements, and with the development of appropriate input/output models for multi variable control of high-purity columns. For the purpose of illustration, we investigate the performance of several model predictive control schemes based on a linear model, a “high frequency” model, and a nonlinear temperature transformation, along with that of conventional multiloop PID control. Closed-loop control performances are compared for setpoint changes as well as for changes in unmeasurable load disturbances.

Robust control of distillation columns: μ- vs H∞-synthesis

Abstract High-purity distillation columns are known to be difficult to control due to their ill-conditioned and non-linear behaviour. Two approaches for the design of robust controllers yielding high performance are presented. For the first approach, first principles are used to develop an uncertainty model describing the nonlinear dynamics within the entire operating range of an industrial distillation column. This structured uncertainty model is used for μ-synthesis. In a second approach which is based on loop shaping ideas, an H ∞ -controller is designed. This controller performs as well as the μ-controller. The H ∞ -approach offers the advantage that the burden for uncertainty modelling and computation is greatly reduced. However, the GS/T augmentation scheme, which is developed in this paper, must be used for the design of the H ∞ -controller to avoid the inversion of the plant in the controller. The paper concludes with a comparison of the H ∞ - and μ-synthesis methods.

Thermodynamics of semicontinuous mixtures using equations of state with group contributions

In the chemical engineering design of industrial distillation columns data on the vapor-liquid equilibrium of complex mixtures are very important. Because of the expense of the experimental determination of such data, there is interest in their accurate prediction. Multicomponent mixtures such as petroleum or petroleum fractions contain a large number of similar chemical species. The method of continous thermodynamics is suitable for the description of the vapor-liquid equilibria of these systems. Continuous thermodynamics is based on the application of a continuous distribution function for describing the composition of multicomponent mixtures. Coupled with activity coefficient models, this method is shown to be convenient for the prediction of vapor-liquid equilibria. However, applications have been limited to low pressures. Therefore, in this paper a model for the prediction of vapor-liquid equilibria based on a two-parameter cubic equation of state is developed. These parameters are calculated using the mixing rule of Wong and Sandler [1] which allows the inclusion of group contribution models (e.g. UNIFAC, modified UNIFAC or ASOG).

Principal Component Analysis Applied to Process Monitoring of an Industrial Distillation Column

Principal Component Analysis (PCA) provides an efficient monitoring tool to extract relevant information for highly correlated process measurements. PCA handles a relatively large number of process variables by reducing the dimensionality of the problem while still retaining the original variability in the process. Several PCA models are developed for the depropanizer of a heavy gasoline-FCC unit in the Mizushima refinery of the Japan Energy Corporation. An hourly-average model is used to analyze historical data, resulting in an improved understanding of the process. Two models based on hourly-average and minute-by-minute data are employed for monitoring of process performance.

Novel Developments in Process Optimisation Using Predictive Control

An optimising controller that is able to drive a plant from a suboptimal operational condition to its steady-state optimum based on receding horizon optimal long range predictions is presented. The technique has been tested wilh realistic simulations of an industrial distillation column using a rigorous process simulator.

Effect of disturbances in optimizing control: Steady‐state open‐loop backoff problem

AbstractOne of the key components in the operation of chemical plants is the ability to operate over a range of conditions while satisfying performance specifications. A method for determining the necessary open‐loop backoff from a steady‐state nominal optimum is introduced to ensure that disturbances cause no constraint violation. This approach consists of defining a joint optimization‐flexibility problem that can be solved within an optimization framework based on an iterative procedure. By formulating the optimization problem in economic terms, the backoff in the objective function is a measurement of the open‐loop economic penalty that is necessary to be paid to achieve feasible operation over the disturbance range of interest (considering the worst combination). An upper bound on the economic potential available for closed‐loop control can then be established, providing a valid reference for ranking different control schemes. Three examples presented illustrate the application of this approach: (1) a simple linear example, (2) a system of two CSTRs; and (3) an industrial distillation column.

Purdue Control Modules: A flexible set of software modules for an undergraduate process dynamics and control laboratory

Purdue Control Modules PCM are introduced as a set of programs written in the MATLAB/SIMULINK environment, which are meant to be used in the instruction of an undergraduate course on process dynamics and control. The modules are based on fundamental process models of industrial unit operations distillation column and heated-tube furnace, and incorporate a realistic graphical user interface to emulate an industrial control environment. © 1996 John Wiley & Sons, Inc.

Robust PID control for an industrial distillation column

State-space controllers are difficult to implement in a distributed control system. Therefore a control design based on the common PID control structures is desirable. This article presents the /spl mu/-optimal tuning of advanced PID control structures as well as a simple feedforward control design for an industrial binary high purity distillation column. The design is based on an optimization approach. The resulting controllers ensure stability and a high performance for the entire operating range of the distillation column. A summary of the practical experiences complements the theoretical results. >

Verification of a Distillation Column Benchmark

This paper presents the validation of an elaborate model of a semi industrial scale distillation column equipped with an indirect heat pump. The model is derived for ordinary operation within the designed operating window. The paper describes the basis assumptions of the model. From experimental data model parameters are found and the model is validated against static and dynamic experimental data. The performance of the model is generally good, however, it does not catch some of the more complex behaviour of the process.

Introduction to the Process Identification Workshop at the 1992 Canadian Chemical Engineering Conference

Abstract This set of papers documents the results of the Workshop on Process Identification held at the 42nd Canadian Chemical Engineering Conference in October 1992. Three university and two industrial groups participated in the workshop, where they were asked to identify process models for an industrial distillation column. The goal of the workshop was to permit valid comparisons among the identification methods currently in use in industry, and those proposed by academic researchers. This first paper outlines the workshop problem itself.

Fuzzy-neural-net-based inferential control for a high-purity distillation column

Many industrial processes are difficult to control because the product quality cannot be measured rapidly and reliably. One solution to this problem is inferential control, which uses an inferential estimator to infer unmeasurable process outputs from secondary measurements, and controls these outputs. This contribution proposes a new approach for designing a Fuzzy-Neural-Net (FNN)-based inferential estimator. The FNN is constructed by distributed multi-networks, whose classification, online running and learning are based upon fuzzy set theory. Application of this method to an industrial high purity distillation column shows that the FNN based inferential control is far superior to conventional composition control.

Model based control of a high purity distillation column

AbstractIn this paper, the problem of dual product composition control of an industrial high purity distillation column, a deisohexanizer (DIH), is addressed using a Generic Model Control framework. A dynamic simulation of the DIH was performed for preliminary studies of the performance of different controller strategies/algorithms. The performance of Generic Model Control incorporating different process models was studied. Process models are presented ranging from simple first order approximations to mechanistic short cut distillation models where a tradeoff between model complexity and model adaptivity is investigated. The different controllers were implemented and compared using a dynamic simulation of an industrial deisohexanizer (DIH) to select the best condidate controller. A controller using a nonlinear process model emerged as the best controller and was implemented on the actual process, resulting in improved performance over the original controller. Simulation results and industrial plant data are presented.

004 Improved operation of an industrial distillation column through advanced control: J.B. Lear, G.W. Barton, R. Di. Mascio, J.A. Romagnoli, pp 15–18

004 Improved operation of an industrial distillation column through advanced control: J.B. Lear, G.W. Barton, R. Di. Mascio, J.A. Romagnoli, pp 15–18

Model-based control using mechanistic, nonlinear models

The need for optimal control is increasing due to more emphasis on minimizing energy use and making products with less variability. We have seen significant energy savings and decreases in product variability using mechanistic, nonlinear models for control of industrial distillation columns. Steady state and dynamic models of this type could play an important role in the future of model predictive control because this type controller works well when process gains change significantly from changes in operating conditions or from nonstationary behavior. This approach also does not require process perturbation tests.

μ-Optimal Control of an Industrial Binary Distillation Column

A μ-optimal controller design using μ-K iteration for an industrial distillation column is presented. The design includes a temperature measurement near the feed tray to improve disturbance rejection. It is based on an uncertainty model which represents column nonlinearity. This uncertainty model includes a linear combination of two models for different operating points. Compared to the multiplicative uncertainty at the plant output usually used, a less conservative controller design is achieved. The advantage of this controller design over the commonly used diagonal PI controller is demonstrated by simulations for the whole operating range of the distillation column. A report on the experiences with the practical implementation will be presented orally.

Improved Operation of an Industrial Distillation Column Through Advanced control

This paper considers the potential improvement in operation of an industrial C3-splitter column. A simplified model was developed to determine the optimal operation of the column and requirements for closed-loop control. Model predictive control was shown to provide a substantial improvement over a conventionally tuned multiloop controller. Controlled column behaviour comparable to that possible with model predictive control was shown to be achievable using a multiloop controller tuned specifically so as to incorporate disturbance rejection capabilities.

Control design for an industrial distillation column

Control design of industrial distillation columns is a well-known problem. The diversity of columns makes a general approach impractible. Non-linearities, multivariable character and possible time delays of individual columns are the source of the control design problems. This paper discusses a process anlaysis and control design procedure for industrial binary high-purity distillation columns. Use is being made of both mechanistic modelling and process identification. It is shown that the procedure, as applied, permits rapid evaluation of several types of control strategies, ranging from PID control and physical decoupling to mathematical decoupling and model-based control. It was proven that multivariable control leads to improved control performance for the application considered. Furthermore it is shown that by model-based process analysis and basic control design, complex control problems can be significantly reduced.

Order reduction of rigorous dynamic models for distillation columns

A method for order reduction of rigorous dynamic models of distillation columns is presented. The approach is a modification and extension of the compartmental models presented in 1986 by Benallou et al Formulating the algebraic equations and the differential equations for total holdup for the compartments, but keeping the differential equations for concentration dynamics for each tray, Murphree tray efficiencies can be used and response accuracy increased. The resulting models of substantially lower order show good dynamic and steady-state conformity with the complete rigorous model. The example of an industrial binary distillation column will demonstrate the effectiveness of this approach.

Experimental validation of a packed bed distillation model

AbstractA mathematical model of a packed bed distillation column is presented and applied to a binary system in a full scale industrial distillation column with structured packings. The model is reduced by orthogonal collocation to sets of ordinary differential and algebraic equations. A linearized large state space model is then obtained and further reduced by the optimal Hankel norm method in order to obtain a model of manageable size. The model is studied under three different operating conditions in order to cover the feasible operating ranges during the experiments. The experiments carried out with a monoethanolamine‐diethanolamine system show a good agreement with the model implying that a control design based on the model is satisfactory. It is also shown that the control configuration to be used can be easily analyzed by the model, since different input and output structures can be studied. Furthermore, models based on one‐film and two‐film theories are compared in the frequency domain and show that, due to its simplicity, the one‐film theory should be used in control design.

Control design for an industrial high-purity distillation column using mechanistic modelling and process identification

Control design of industrial distillation columns is a well-known problem. The diversity of columns makes a general approach impractible. The non-linearities, multivariable character and possible time delays of individual columns are the source of the control design problems. This paper discusses a process analysis and control design procedure for industrial binary high-purity distillation columns. Use is being made of both mechanistic modelling and process identification. It is shown that the procedure, as applied, permits rapid evaluation of several types of control strategies, ranging from PID control and physical decoupling to mathematical decoupling and model-based control. It was proven that multivariable control leads to improved control performance for the application considered. Furthermore it is shown that by model-based process analysis and basic control design, complex control problems can be significantly reduced.