Reinaldo Giudici

Neural network based approach for optimization of industrial chemical processes

Abstract Process optimization involves the minimization (or maximization) of an objective function, that can be established from a technical and/or economic viewpoint. In general, the decision variables are subject to constraints such as valid ranges (max and min limits) as well as constraints related to safety considerations and those that arise from the process model equations. Usually in chemical engineering problems, both the objective function and the constraints are non-linear. Computational methods of non-linear programming with constraints usually have to cope with problems such as numerical evaluation of derivatives (Jacobian, Hessian) and feasibility issues. The basic idea of the optimization method using neural network (NN) is to replace the model equations or plant data by an equivalent NN, and use this NN to carry on a grid search on the region of interest. As an additional benefit, the full mapping of the objective function allows one to identify multiple optima easily, an important feature not presented by conventional optimization methods. Moreover, the constraints are easily treated afterwards since the points with violated constraints can be recognized and classified (according to weak or hard constraints). This approach was applied in some industrial chemical process: the process of nylon-6,6 polymerization in a twin-screw extruder reactor and an acetic anhydride plant.

Polymerization Kinetics of Pre-heated Composite

Temperature affects the polymerization behavior of dimethacrylate-based materials. This study describes the influence of pre-polymerization temperature and exposure duration on polymerization kinetics of a commercial dental photo-activated composite at the top and at 2-mm depth. We used the temperature-controlled stage of a diamond-attenuated-total-reflectance unit to pre-set composite temperature between 3° and 60°C. Composite was light-exposed by a conventional quartz-tungsten-halogen curing unit for 5, 10, 20, or 40 sec. Real-time conversion, maximum conversion rate (R p max), time to achieve R p max, and conversion at R p max were calculated from infrared spectra. Composite pre-warming enhanced maximal polymerization rate and overall monomer conversion (top significantly greater than 2 mm). Time when R p max occurred did not change with temperature, but occurred sooner at the top than at 2-mm depth. Conversion at R p max increased with temperature, allowing more of the reaction to occur prior to vitrification than at room temperature.

Deactivation of steam reforming catalysts by sintering: experiments and simulation

Sintering of coprecipitated Ni/Al2O3 catalysts was studied in a microreactor at 500 and 800°C under steam reforming reaction. Changes in microstructure (crystalline phases, average metal crystallite sizes and size distributions) were investigated by temperature programmed reduction (TPR), X-ray diffraction (XRD) and transmission electron microscopy (TEM). Differences between particle size distributions determined by XRD and TEM are discussed in view of their specific limitations. Low temperature calcination of unpromoted catalysts may have a beneficial stabilizing effect upon reduction. The rate of sintering of small nickel crystallites increases as a result of the combined action of steam and high temperature, particularly for narrow initial crystallite size distributions and high initial dispersions. This behavior suggests a shift, from crystallite migration in the initial stages of heat treatment, to atomic migration in the latter stages. The lanthanum promoter seems to inhibit metal crystallite growth under reduction and reaction conditions. Predictions of a developed Monte Carlo based model are in good qualitative agreement with experimental observations.

Modeling and experimental studies of emulsion copolymerization systems. I. Experimental results

A systematic experimental and modeling study of several emulsion copolymerization systems has been performed, and will be reported in a series of papers. Ten binary and three ternary copolymerizations involving styrene, methyl methacrylate, butyl acrylate, butadiene, vinyl acetate, acrylic acid, and ethylene were studied varying polymerization temperature, monomer composition, water to monomer ratio, initiator and emulsifier concentrations. Conversion, particle size, copolymer composition, and gel content were measured at several reaction times. The goal of this series of papers is to assess our quantitative understanding of emulsion copolymerization expressed in the form of a comprehensive mathematical model applied to monomers widely used in industry. In this first paper of the series, a global comparison of the experimental results is made. It is observed that the gel content is higher in systems containing butyl acrylate and butadiene, and smaller in systems containing methyl methacrylate. Larger particle numbers are obtained for lattices containing acrylic acid and butadiene. It is also shown that, for most of the systems, integration of the simple Mayo–Lewis equation is adequate to explain the drift in copolymer composition observed experimentally.

Optimization of a cocktail of initiators for suspension polymerization of vinyl chloride in batch reactors

Abstract The industrial manufacturing of poly(vinyl chloride) (PVC) by suspension polymerization is carried out in batch reactors. The productivity increase of these processes is directly related to the reduction of the time required to complete each batch. The reactor cooling system is designed such that it is capable to compensate for the maximum rate of heat release by the exothermic polymerization, so that the reactor cooling capacity is underutilized. An attractive mode of operation is to run the industrial process isothermally and to use a mixture (“cocktail”) of different initiators, which is able to spread the polymerization rate over the batch time. In this work, the optimal formulation of an initiator mixture is studied as a dynamic optimization problem, which makes use of a representative mathematical model for the batch suspension vinyl chloride polymerization process. Results show that by choosing the optimal amounts of initiators in the cocktail, a significant reduction of the total processing time for a given polymer specification can be obtained, as compared to the case in which only one initiator is optimally chosen.

Evidences of correlation between polymer particle size and Raman scattering

This work describes evidences of correlation between polymer particle size and Raman scattering and shows that it is possible to use Raman scattering to monitor the evolution of average particle size during emulsion polymerization reactions. The main focus is the estimation of the average polymer particle diameter from spectra collected in a short acquisition time and consequently low signal-to-noise ratio. Finally, a multivariate linear model, (Partial Least Square-PLS), is fitted from the reaction data and a good linearity between spectra and average polymer particle diameter is found. It is shown that despite varying monomer and polymer concentrations it is possible to monitor average particle sizes during emulsion polymerization reactions using Raman spectroscopy.

A Simulation Study on Random Scission of Branched Chains

ABSTRACTThis paper deals with the mathematical modelling of the random scission of branched chains. The objective is to predict the reaction products as well as the molecular weigth distribution for processes such as the chemical modification of polyolefins by free radical mechanism. In comparison with linear chains, where the number of scission possibilities is clear and scission products are linear, for branched chains the problem is much more complex. Two modelling approaches for branched chains were used. The first one, a kinetic, deterministic approach assumes a simple, regular morphology for all chains in order to compute its effect on the process. The second, more realistic approach uses Monte Carlo methods, and is able to keep track of the structure of each chain during the scission process. Stars and combs were used as idealized structures in the simulations with both approaches. The two approaches are applied, compared and discussed. The simulated results show that the chain morphology plays an ...

Neural network based approach for optimisation applied to an industrial nylon-6,6 polymerisation process

The basic idea of the proposed optimisation method is to replace the model equations by an equivalent neural network (NN) that mimics the phenomenological model, and use this NN to carry out a grid search, mapping all the region of interest. The proposed optimisation approach was applied to the industrial process of nylon-6,6 polymerisation in a twin-screw extruder reactor. This corresponds to the finishing stage of an industrial polymerisation plant. A qualitative optimisation procedure is used taking in account safe operation conditions, wear and tear of the equipment, product quality and energy consumption. The chosen operational variables are then checked with the phenomenological model. This approach provides more comprehensive information for the engineers analysis than the conventional non-linear programming procedure.

Optimization of semicontinuous emulsion polymerization reactions by IDP procedure with variable time intervals

Abstract Optimization of a semicontinuous emulsion polymerization reaction is carried out using a mathematical model derived from first principles to represent the process. Iterative Dynamic Programming (IDP), being a straightforward procedure that allows the implementation of optimization constraints, is an attractive tool to optimize highly complex and nonlinear equation systems such as those that are used to represent the emulsion polymerization processes. In order to minimize the reaction time of a semicontinuous reaction, time intervals with varying lengths were used allowing the establishment of the optimal operation conditions where the final time is not specified. The optimization procedure is applied to two emulsion polymerization systems: (a) butyl acrylate/vinyl acetate copolymerization with controlled composition; (b) vinyl acetate homopolymerization with specified molecular weight. Results show that significant reductions of reaction time are possible and that their extent depends on the reactors heat removal capacity and on the formulated constraints.

Modeling and experimental studies of emulsion copolymerization systems. II. Styrenics

Using a model previously published, predictions for evolution of conversion and average particle diameter in batch experiments are compared against experimental data for four emulsion copolymerizations of styrene with the following monomers: (1) methyl methacrylate, (2) butyl acrylate, (3) butadiene, and (4) acrylic acid. For each copolymerization system the experiments covered simultaneous variations in five variables: initiator and surfactant concentrations, water to monomer ratio, monomer composition, and temperature. It is shown that after data fitting for unknown or uncertain parameters, the model is capable of explaining quantitatively the experimental observations for conversion evolution and only qualitatively the particle size evolution data. This points out to the possible contribution of particle nucleation mechanisms other than the micellar one, which is the only mechanism included in the model. Some of the adjustable parameter values were found to depend on the copolymer composition. The only case in which the model does not perform well is in the prediction of the effect of initiator concentration on the copolymerization rate for butadiene-rich formulations. It is also found that the model predictions are very sensitive to the value of the diffusion coefficients of monomeric radicals in the copolymer particle, which are not readily available in the literature. It is concluded that it is important to independently measure these parameters in order to enhance the predictive power of models. It is also concluded that the model can be useful for practical applications.