Main Applications of NCO Tracking

Measurement-based Optimization of a Batch Distillation Column


Distillation is utilized in industry to separate liquid mixtures containing components with different volatilities. Batch distillation offers high flexibility and low investment costs, especially for small quantities. The type and the amount of products can be adjusted quickly according to market conditions. The optimization objective for batch distillation operations is often to maximize the amount of product at final time, while satisfying product purity requirements. When temperature-sensitive mixtures are distilled, path constraints have to be considered. The reflux ratio, the reboiler heat duty, the condenser cooling duty and the column pressure are typically the main input variables.
The optimal solution for the distillation of a binary mixture consists of three arcs: (i) a full-reflux phase to heat up the column and increase the composition in the condenser, (ii) a distillate-withdrawal phase, where distillate is accumulated in a product tank, and (iii) a zero-reflux phase to remove the high-purity product left in the condenser. Open-loop implementation of an optimal reflux policy typically results in infeasible and/or suboptimal operation in the presence of uncertainties such as inaccurate vapor-liquid equilibrium data. To ensure the required product purity, conservative operation strategies have to be implemented, which reduces performance. This project has examined the use of on-line and batch-end measurements to reduce this conservatism. Experiments were carried out on a laboratory-scale batch distillation column in collaboration with the University of Applied Sciences in Fribourg (EIF).



Measurement-based Optimization of a Batch Reactor under Safety Constraints

This experimental project has investigated the adjustment of the feed rate to maximize the yield of a second-order reaction (2-butanol + propionic anhydride) taking place in an isothermal semi-batch reactor. For exothermic reactions, a cooling failure can lead to a runaway situation with severe consequences if no preventive measures are taken. In this context, a feed strategy has been developed to optimize productivity under safety constraints by enforcing constraints on the maximal temperature reached under cooling failure and the amount of heat produced. The optimal solution in this case corresponds to riding on the safety constraints. For its implementation, estimates of the temperature under cooling failure and the heat generation were obtained from spectroscopic and calorimetric measurements. Optimal operation of the experimental reactor was then obtained by riding along the active constraints.

Measurement-based Optimization of a Polymerization Process


Emulsion polymerization is a very subtle polymerization process. Despite its relative complexity, batch emulsion polymerization is the method of choice whenever specifications on conversion, particle size and molecular weight distribution are stringent. A typical dynamic optimization problem for batch polymerization reactions is to minimize batch time, while respecting path and terminal constraints. Path constraints are typically bounds on temperature or heat generation, while terminal constraints involve requirements on both conversion and average molecular weight. The reactor temperature is the most appropriate manipulated variable as it strongly affects the molecular weight distribution and the reaction rates. Moreover, the initiator concentration can be used as an additional manipulated variable, especially towards the end of the reaction if a stringent constraint on conversion need to be met.
The optimal solution consists of using initially the maximal allowed temperature and then reducing it to a value that exhibits a compromise between molecular weight and conversion objectives. The temperature profile can be parameterized using two parameters, the switching time between the two temperature levels and the constant temperature value in the second arc. Using measurements of the final conversion and molecular weight, these two input parameters can be adapted toward optimality on a run-to- run manner.
The aforementioned methodology was investigated for several emulsion polymerization reactions that include copolymerization of styrene/alpha-methylstyrene and the copolymerization of acrylamide with quaternary ammonium cationic monomers. The optimization problems considered the presence of constraints, uncertainty regarding the rate parameters and noisy measurements. The project was carried out experimentally in collaboration with Aqua+Tech, and led to a 20% increase in productivity for a 1-ton industrial batch reactor.

Optimization of Filamentous Fungi Fermenter


This project, performed in collaboration with Novozymes in Denmark and the Université Catholique de Louvain in Belgium, has addressed the optimization of filamentous fungi fermentation. A mechanistic model was built from industrial data, with special emphasis given to the phenomenon of oxygen limitation, which is particularly important in this reactor due to the filamentous nature of the bio-organisms.
Numerical optimization indicated that the initial substrate concentration could be reduced considerably. The optimal solution consists of two arcs, namely a batch phase followed by a feeding phase that can be approximated as an arc of constant dissolved oxygen. As a result, a cascade controller for the dissolved oxygen concentration was implemented, with the reference for dissolved oxygen being adjusted on a run-to-run manner.