Karen Valverde Pontes

Process analysis and optimization mapping through design of experiments and its application to a polymerization process

The technique of experimental design is used on an ethylene polymerization process model in order to map the feasible optimal region as preliminary information for process optimization. Through the use of this statistical tool, together with a detailed deterministic model validated with industrial data, it is possible to identify the most relevant variables to be considered as degrees of freedom for the optimization and also to acquire significant process knowledge, which is valuable not only for future explicit optimization but also for current operational practice. The responses evaluated by the experimental design approach include the objective function and the constraints of the optimization, which also consider the polymer properties. A Plackett-Burman design with 16 trials is first carried out in order to identify the most important inlet variables. This reduces the number of decision variables, hence the complexity of the optimization model. In order to carry out a deeper investigation of the process, complete factorial designs are further implemented. They provide valuable process knowledge because interaction effects, including highly non-linear interactions between the variables, are treated methodically and are easily observed.

A model-based approach to quality monitoring of a polymerization process without online measurement of product specifications

Quality monitoring and support for decision-making in a complex industrial process.Comprehensive approach to select and estimate parameters of a complex model.Dynamic monitoring of a process without on-line measurement of quality variables.Control of product quality with impact on productivity and customer satisfaction.A novel orthogonalization algorithm able to ensure estimability. This paper presents a model-based approach for on-line monitoring of difficult-to-measure quality variables with the description of their dynamic behavior. The strategy comprises the use of a Virtual On-line Analyzer (VOA), based on an empirical model, whose development is supported on the one hand by experimental data, and on the other hand by a complex phenomenological model. A systematic approach to the selection and estimation of parameters of the phenomenological model from experimental data is also presented in order to adjust this model to the behavior of the industrial process. Regarding the experimental data, this complete model has additional information (such as the dynamic behavior of the process, embedded in the knowledge used in its formulation) and it is able to provide synthetic data for VOA training which comprises a NARX (Non-linear AutoRegressive with Exogeneous inputs) neural network model. A real case study comprising quality monitoring in a polymerization system is investigated. The available measurements of the main polymer properties are performed at a low frequency and are not able to represent the dynamic behavior and support decision-making at the operational level, especially during online grade transitions (changes in product specifications). The results show the ability of the neural model to predict the quality variables with a high frequency (small sampling period), enabling and supporting decision-making for quality monitoring of the polymer in real time.

A Survey of Equations of State for Polymers

The thermodynamics of polymeric systems play an important role in the polymer industry and are often a key factor in polymer production, processing and material development, especially for the design of advanced polymeric materials. Many polymeric products are produced with a solvent or diluent (or a mixture of them) and often with other low molecular weight compounds (plasticizers, among others). A problem which often arises is how to remove the low molecular weight constituent(s) from the final product (polymer). The solution to this problem involves, among other tasks, solving the vapor-liquid equilibrium (VLE) and/or the vapor-liquid-liquid equilibrium (VLLE) problem. Other applications of polymer thermodynamics directly involve the polymerization processes. For example, several processes such as the production of PET (polyethylene terephthalate) are carried out in two-phase (vapor-liquid) reactors. Phase equilibrium compostions of the reacting components will determine their phase concentrations and thus the outcome of the polymerization reaction. Another example is the case of LDPE (Low Density PolyEthylene) made in autoclave reactors where it may be desirable to perform the polymerization reaction nearby but outside the two-liquid phase region, but close to it, which makes accurate liquid-liquid equilibrium (LLE) information at high pressure essential. During PE (polyethylene) or PP (polypropylene) industrial processing, for example, deposition of the polymer on the reactor surface, heat exchangers and flash drums frequently occurs and this can cause clogging in pipelines. Modeling solid-liquid equilibrium (SLE) is a useful basis from which to gain a better understanding of these industrial polymer problems and thus to avoid their occurrence.

Fault Detection and Diagnosis in dynamic systems using Weightless Neural Networks

Abstract This work examines Fault Detection and Diagnosis (FDD) based on Weightless Neural Networks (WNN) with applications in univariate and multivariate dynamic systems. WNN use neurons based on RAM (Random Access Memory) devices. These networks use fast and flexible learning algorithms, which provide accurate and consistent results, without the need for residual generation or network retraining, and therefore they have great potential use for pattern recognition and classification (Ludermir, Carvalho, Braga, de Souto, 1999). The proposed system firstly executes the selection of attributes (in the multivariable case) and does the time series mapping of the data. In the intermediate stage, the WNN performs the detection and diagnosis per class. The network outputs are then passed through a clustering filter in the final stage of the system, if a diagnosis per fault groups is necessary. The system was tested with two case studies: one was an actual application for the temperature monitoring of a sales gas compressor in a natural gas processing unit; and the other one uses simulated data for an industrial plant, known in the literature as “Tennessee Eastman Process”. The results show the efficiency of the proposed systems for FDD with classification accuracies of up to 98.78% and 99.47% for the respective applications.

Modeling and Simulation of Ethylene and 1-Butene Copolymerization in Solution with a Ziegler-Natta Catalyst

A comprehensive mathematical model for the ethylene/1-butene polymerization in solution with Ziegler-Natta catalyst is developed. The process comprises a series of continuously stirred and tubular reactors in which polyethylene resins with different properties may be produced. The mechanistic model considers the moments of the bivariant molecular weight distribution in order to ascertain the average molecular weight and polydispersity. The polymer quality is verified through the melt index, density and stress exponent, which is a measure of the polydispersity. The model developed investigates the stationary and dynamic behavior of the process following step changes in feed conditions. It allows for the prediction of non-linear and inverse responses, encouraging the use of the model for optimization and control purposes.

Parameter estimation with estimability analysis applied to an industrial scale polymerization process

Abstract This paper aims to estimate the parameters of a complex model representing an industrial scale polymerization process. The estimability analysis of the parameters prior to estimation allows simplifying the optimization problem but it is usually neglected in literature when industrial data is used for estimation. In this case, though, the estimability analysis would be even more important since usually less data is available, they are associated with a higher uncertainty and the experiments might not be designed as in laboratory or pilot plant. The orthogonalization method reduced from 68 to 29 the number of parameters of the model. Polymer properties, which are measured offline with low frequency, as well as process temperatures and flow rates are used for validating the model. Small deviations, up to 5%, between model prediction and experimental data indicate the quality of fit of the model and the importance of carrying out first an estimability analysis.

Chemical, physico-chemical and sensory characterization of mixed açai (Euterpe oleracea) and cocoa´s honey (Theobroma cacao) jellies

Four formulations of mixed acai (Euterpe oleracea) (A) and cocoa´s honey (Theobroma cacao) (CH) jellies were prepared according to the following proportions: T1 (40% A:60% CH), T2 (50% A:50% CH), T3 (60% A: 40% CH) and T4 (100% A - control). All formulations were prepared using a rate 60:40 (w/w) of sucrose and pulp, plus 0.5% pectin and the products reached to average of 65% soluble solids content. The jellies were analyzed by chemical and physicochemical (titratable acidity, pH, soluble solid content, dry matter, total protein, lipids, vitamin C and calories) and sensory characteristics; also were evaluated levels of P, K, Ca, Mg, Fe, Zn, Cu and Mn. It was used a hedonic scale of 7 points to evaluate the attributes: overall impression, spreadability, brightness, flavor, texture and color, and also was verified the purchase intention score. The titratable acidity and pH ranged from 0.46 to 0.64% and 3.35 to 3.64, respectively, that are within the range found at most fruit jellies. The soluble solids content ranged between 65.2 and 65.5 oBrix. The sensory acceptance results showed that all treatments (T1, T2, T3 and T4) presented means of sensory attributes above 4, demonstrating good acceptance of the product, but the treatment T1 presented the higher scores for the evaluated attributes. Cocoa´s honey added a positive influence on the attributes of color, texture and spreadability.

Optimal Design of a Dividing Wall Column for The Separation of Aromatic Mixtures using the Response Surface Method

Abstract Dividing wall columns (DWC) are the industrial implementation of the Petlyuk configuration in a single shell. The optimal design of dividing wall columns is a multivariable, mixed-integer and non-linear problem with a non-convex optimization function presenting several local optima. The response surface method (RSM) associated with the steepest descent method offers a practical approach to optimize the DWC design. In the present study, a central composite design is carried out to obtain and validate the response surface for a DWC performing an aromatic ternary separation. The resulting aromatic DWC system presented a better optimal design compared with previous work and energy savings by 44% compared with a conventional two-column configuration (C2C) typically used for aromatic separation. The method is able to systematically solve the problem while minimizing simulation effort and optimizing the DWC design parameters.

A Quasi-Virtual Online Analyser based on an Artificial Neural Networks and Offline Measurements to predict purities of raffinate/extract in Simulated Moving Bed processes

Abstract The quality control and optimization of Simulated Moving Bed processes are still a challenge. Among the main reasons for that, the real time measurement of its main properties can be highlighted. Further developments in this field are necessary in order to allow the development of better control and optimization systems of these units. In the present work, a system composed by two Artificial Neural Networks working concomitantly with an offline measurement system is proposed, named Quasi-Virtual Analyser (Q-VOA) system. The development of the Q-VOA is presented and the system is simulated in order to evaluate its efficiency. The methodology used to select the input variables for the Q-VOA is another contribution of this work. The results show that the Q-VOA is capable of reducing the system errors and keep the prediction closer to the process true responses, when compared with the simple VOA system, which is based solely on model predictions. Furthermore, the results show the efficiency of the measurement system even under the presence of non-measured perturbations.

Biodegradable thermoplastic starch of peach palm (Bactris gasipaes kunth) fruit: Production and characterisation

ABSTRACT The objective of this study was to produce and characterise biodegradable thermoplastic starch (TPS) derived from the fruit of pejibaye palm (Bactris gasipaes Kunth) plasticised with glycerol and sorbitol. The plasticised starch yielded a strength (σ) of 1.3 ± 0.2 MPa, deformation (ε) of 9.4 ± 1.6%, and Young’s modulus (E) of 191 ± 72.0 MPa. The thermal analysis showed a 61.14% mass loss over the temperature range 290–388°C and an endothermic peak at 130°C. X-ray diffractograms of the TPS revealed peaks corresponding to crystallites of type Vh at 19.4° and type V at 22.6°. Morphological studies, by scanning electron microscopy, showed the plasticised starch had a homogeneous surface, without phase separation and without cracks, with few granules not gelatinised. The analysis of its biodegradation by soil burial tests showed a total mass loss of 84.4 ± 4.4%, after 18 weeks. Biodegradable TPS was successfully obtained from the pejibaye palm fruit, presenting resistant to traction and to the thermal degradation.