G. Marchetti

RIGOROUS SIMULATION AND MODEL PREDICTIVE CONTROL OF A CRUDE DISTILLATION UNIT

Abstract This paper describes the application of a widely-used commercial multivariable predictive controller to a rigorously simulated crude distillation process. After describing the main process and controller features, it is shown how the two simulation and control environments can be interfaced together. A number of simulation results of typical product quality changes and crude switches are presented. The final goal of this paper is to demonstrate how rigorous dynamic simulators can be effectively used to reduce the costs of Advanced Process Control projects by shortening model identification, controller design and commissioning phases.

MULTIVARIABLE SUBSPACE IDENTIFICATION AND PREDICTIVE CONTROL OF A HEAT-INTEGRATED SUPERFRACTIONATOR RIGOROUS MODEL

Abstract The control of a heat-integrated superfractionator is addressed in this paper. A multivariable subspace identification method is proposed to overcome the difficulties associated to the large time constants of the process and identify a linear process model, upon which a constrained predictive controller is developed. The effectiveness of the proposed identification and control algorithm is shown by means of closed-loop rigorous dynamic simulation results.

RELAY AUTOTUNING OF MULTIVARIABLE SYSTEMS: APPLICATION TO AN EXPERIMENTAL PILOT-SCALE DISTILLATION COLUMN

Abstract This paper describes the application of multivariable sequential relay autotuning techniques to a pilot scale distillation system under industrial DCS monitoring and control environment. Implementation aspects arising in any realistic industrial environment are fully discussed such as the architecture for on-line communication and information extraction as well as valve hysteresis and backlash compensation. The performance of alternative tuning strategies is compared for both set-point changes and disturbances rejection.

Sequential Identification and Autotuning by Relay Techniques of Decentralised Controllers for MIMO Processes

Abstract Three different relay techniques for on-line identification of standard decentralised PI controllers for MIMO processes are compared to analyse the trade-off between increased complexity and improved performance. The first uses a standard relay and adopts an enhanced tuning rules (SY) without requiring any additional knowledge; the second uses a modified relay (TCR) to guarantee some stability margin; the third allows a parametric identification by means of three tests with additional relay (ATV + ). Two indexes are defined in order to quantify the duration of experimental test and the achievable performance. A quantitative analysis of results for typical chemical processes shows that the problem cannot be always solved by the more conservative SY tuning. The TCR technique gives acceptable results by requiring identification times which are generally less than twice longer with respect to the base case. The larger knowledge obtained by the ATV procedure, is not compensated by improvements in the achievable performance when the controller structured is limited to PI.

A comparison of Relay Techniques for Identification and Autotuning of Chemical Processes

Abstract A comparison is presented between two improved relay techniques, the Two Channel Relay and the ATV+ identification, recently introduced to obtain more information about the process and therefore to improve the closed loop performance. The techniques are briefly described and the comparison is stated in terms of duration of experimental tests versus achievable performance. From the analysis of a large number of cases, the ATV+ method shows to give better results, because a parametric model is identified and a more accurate tuning technique can be employed; the corresponding experimental times are about twice longer.

Closed-Loop Identification and Control of Multivariable Chemical Processes: a Case Study

Closed-loop identification methods based on relay techniques have been analyzed and compared with open loop methods for an application to a multivariable nonlinear chemical process (a distillation unit: C 3 — C 4 splitter). Classical relay techniques and a recently proposed method for identification and control design (ATV+) are applied in a sequential procedure which leads to the design of decentralised controllers.

Adaptive control of an experimental distillation column by relay methods

Abstract The control of a non linear pilot-plant distillation column in a wide range of operating conditions is performed by identification of process models in few operati.ng points and adopting time varying controllers. Relay feedback tests are used to bUIld simple two-parameter models able to describe the high frequency response of the system and to evaluate the relative effect of different variables. PI control with parameters varying in time as a function of the temperature is able to give satisfactory performance in the required range of operation. The proposed methodology shows to be fast and effective way of facing the problem of non linearity in the case under study.

Accuracy of Autotuning Identification Methods and Achievable Closed Loop Performance

ABSTRACT The paper compares two identification methods used to build models for control purposes to be used for autotuning, namely relay with additional delays and sinusoidal inputs. The two techniques allows to obtain more (2 or 3) points in the Nyquist plane and from them a model can be obtained by a least square method. The accuracy in identification which is obtainable by the two methods is evaluated and possible improvement in closed loop performance are illustrated. As expected, the identification by means of sinusoids is able to give higher accuracy, with respect to the inherently approximate relay techniques, when at least 3 tests can be performed; such tests however are more time consuming and more sensitive to measurements perturbations. Full advantages of the improved identification can be taken by adopting a model based controller.