In this paper the problem of designing a two-point linear temperature controller for the regulation of the product compositions in binary distillation columns is addressed. The combination of feedforward, feedback, and internal model control concepts yields a control design methodology with (i) criteria to choose the (decentralized and one-or two-way) decoupling structure and the temperature measurements location, (ii) a linear control system that consists of a static interaction compensator and a pair of decoupled proportional-integral controllers with feedtemperature-based setpoint adjustment, (iii) an implementation that requires the static dependencies of the measured tray temperatures on the feed temperature, the slopes of the operating lines, and the temperature-gradient-to-holdup quotients in the measurement trays, and (iv) a conventional-like tuning scheme that resembles the ones employed in linear firstorder controllers and filters. The linear controller recovers the behavior of a material-balancebased feedforward-feedback nonlinear controller. The proposed design is applied to two representative examples through simulations, in the presence of holdup dynamics, nonideal thermodynamics, and actuator errors, as well as measurement delays and lags.

Although the two-input-two-output (2I-2O) dual composition (DC) and two-point temperature (TPT) distillation control problems have been studied extensively, the development of two-input-four-output (2I-4O) composition-temperature (CT) control schemes, driven by two composition and two temperature measurements, lags far behind. The difficulty of the 2I-4O CT control problem is due to the lack of systematic means to screen the large number of combined structural-algorithmic design degrees of freedom. With contradictory results, the problem has been recently addressed with linear and nonlinear approaches. Based on a constructive control framework, in this work, a rather simple solution to the combined 2I-4O CT control problem is obtained. The resulting scheme consists of linear PI components and static interaction compensators, has a systematic construction procedure with reduced model dependency, and has easy-to-apply tuning guidelines. The proposed approach is applied to three representative examples through simulations, yielding behaviors that improve those of previous linear and nonlinear cascade control schemes.

No abstract

The problem of designing an on-line optimizing robust output-feedback (OF) controller for binary batch distillation columns with temperature measurements is addressed. The combination of optimality, passivity and detectability notions leads to an estimator-based material balance controller that, for an ample range of load compositions maximizes the profit and yields the prespecified product purity. The scheme has: (i) a criterion to choose the sensor number and locations, (ii) an event controller that decides the total reflux, start-up and partial reflux withdrawal periods, and (iii) a tracking controller that steers the batch distillation column motion along an on-line optimized path. The approach is applied to a previously studied example.

The intent of this note is to show that the incorporation of a temperature measurement in the stripping section improves the performance of distillate cascade controller. The proposed controller regulates the distillate composition by manipulating the reflux flow rate, based on a primary component, which is driven by the distillate composition measurement, and a secondary component, which is driven by two temperature measurements located at the most sensitive trays of the stripping and rectifying sections. Compared to the standard single-temperature cascade control scheme, the proposed two-temperature cascade controller has better behavior, because of the improvement of the feed-forward-like disturbance rejection capability of the secondary control component. Numerical simulations are used to illustrate the performance of the control scheme in the face of feed flow and composition disturbances.