Papers by Argimiro R. Secchi
Scientific Reports, Sep 23, 2022
Governments weigh social distancing policies to avoid overload in the health system without signi... more Governments weigh social distancing policies to avoid overload in the health system without significant economic impact. The mutability of the SARS-CoV-2 virus, vaccination coverage, and mobility restriction measures change epidemic dynamics over time. A model-based control strategy requires reliable predictions to be efficient on a long-term basis. In this paper, a SEIR-based model is proposed considering dynamic feedback estimation. State and parameter estimations are performed on state estimators using augmented states. Three methods were implemented: constrained extended Kalman filter (CEKF), CEKF and smoother (CEKF & S), and moving horizon estimator (MHE). The parameters estimation was based on vaccine efficacy studies regarding transmissibility, severity of the disease, and lethality. Social distancing was assumed as a measured disturbance calculated using Google mobility data. Data from six federative units from Brazil were used to evaluate the proposed strategy. State and parameter estimations were performed from 1 October 2020 to 1 July 2021, during which Zeta and Gamma variants emerged. Simulation results showed that lethality increased between 11 and 30% for Zeta mutations and between 44 and 107% for Gamma mutations. In addition, transmissibility increased between 10 and 37% for the Zeta variant and between 43 and 119% for the Gamma variant. Furthermore, parameter estimation indicated temporal underreporting changes in hospitalized and deceased individuals. Overall, the estimation strategy showed to be suitable for dynamic feedback as simulation results presented an efficient detection and dynamic characterization of circulating variants. The first official cases of COVID-19 were dated in December 2019 in Wuhan, China. Its spread worldwide in later months resulted in a pandemic classification from the World Health Organization (WHO) on 11 March 2020 1 . The first official case in Brazil was reported two weeks earlier, on 26 February 2020, from a man returning to São Paulo from Italy 2 . Social distancing presents effective mitigation over virus spread 3 ; however, it generates negative impacts on the economy and on the mental health of the population 4 . Vaccination is a control action that progressively reduces virus transmissibility aiming for disease elimination defined by zero community infections 5 . Nonetheless, vaccination coverage is delimited by the vaccine acceptance rate, which makes its goal unfeasible even if vaccination provides 100% efficacy against transmission. The SARS-CoV-2 virus is highly mutable, with thousands of variants documented since its origin in December 2019 6 . Mutations might change system dynamics; thus, model updating is required for reliable predictions. Genomic surveillance of SARS-CoV-2 virus in Brazil indicated four predominant circulating variants from
Brazilian Journal of Chemical Engineering, Mar 1, 2011
Brazilian Journal of Chemical Engineering, Dec 1, 2016
Computers & Chemical Engineering, Mar 1, 2016
Equation-oriented simulators have some advantages over the modular sequential ones, but improveme... more Equation-oriented simulators have some advantages over the modular sequential ones, but improvements are still necessary to deal with nonlinearities, while preserving the robustness of the solver. Linear approximations and/or surrogate models can be used in place of nonlinear models, but the loss of predictive accuracy may be a drawback. An alternative to circumvent this problem is the use of grid-based look-up tables for interpolating responses from rigorous models. This methodology was integrated in an equation-oriented simulator (EMSO). A case study involving the production of bioethanol from sugarcane is used to demonstrate the robustness of this approach. Look-up tables replaced the models of two distillation column trains and of the cellulose hydrolysis reactor. These models were included into the global process and an optimization problem aiming at the maximum production of ethanol was successfully solved by a PSO algorithm varying the solid mass fraction in the hydrolysis reactor.
Industrial & Engineering Chemistry Research
Brazilian Journal of Chemical Engineering
Processes
This paper presents a literature review of reinforcement learning (RL) and its applications to pr... more This paper presents a literature review of reinforcement learning (RL) and its applications to process control and optimization. These applications were evaluated from a new perspective on simulation-based offline training and process demonstrations, policy deployment with transfer learning (TL) and the challenges of integrating it by proposing a feasible approach to online process control. The study elucidates how learning from demonstrations can be accomplished through imitation learning (IL) and reinforcement learning, and presents a hyperparameter-optimization framework to obtain a feasible algorithm and deep neural network (DNN). The study details a batch process control experiment using the deep-deterministic-policy-gradient (DDPG) algorithm modified with adversarial imitation learning.
Journal of Molecular Liquids
Abstract The cellulose dissolution is an essential pretreatment process for the chemical conversi... more Abstract The cellulose dissolution is an essential pretreatment process for the chemical conversion of lignocellulosic biomass into biofuels. Here, the dissolution of a 36-chain Iβ cellulose model with a hexagonal cross-section (M36HCS) in water is analyzed by Molecular Dynamics (MD) with CHARMM36/TIP3P (C36/TIP3P) force field using gradual heating at 25 MPa. Our simulations showed that the dissolution of M36HCS starts to occur between 560 K and 600 K, which agrees with experimental observations. In our system, conditions near the critical point of water reveal that translational and rotational entropies decrease, while the low hydration level increases vibrational entropy. This investigation theoretically shows that C36/TIP3P adequately reproduces the dissolution of M36HCS even in high-pressure water as corroborated in reactors.
Journal of Chromatography A
13th International Symposium on Process Systems Engineering (PSE 2018), 2018
Abstract With recent advances in industrial automation, data acquisition, and successful applicat... more Abstract With recent advances in industrial automation, data acquisition, and successful applications of Machine Learning methods to real-life problems, data-based methods can be expected to grow in use within the process control community in the near future. Model-based control methods rely on accurate models of the process to be effective. However, such models may be laborious to obtain and, even when available, the optimization problem underlying the online control problem may be too computationally demanding. Furthermore, the process degradation with time imposes that the model should be periodically updated to stay reliable. One way to address these drawbacks is through the merging of Reinforcement Learning (RL) techniques into the classical process control framework. In this work, a methodology to tackle the control of nonlinear chemical processes with RL techniques is proposed and tested on the wellknown benchmark problem of the non-isothermal CSTR with the Van de Vusse reaction. The controller proposed herein is based on the implementation of a policy that associates each state of the process to a certain control action. This policy is directly deduced from a measure of the expected performance gain, given by a value function dependent on the states and actions. In other words, in a given state, the action that provides the highest expected performance gain is chosen and implemented. The value function is approximated by a neural network that can be trained with pre-simulated data and adapted online with the continuous inclusion of new process data through the implementation of an RL algorithm. The results show that the proposed adaptive RLbased controller successfully manages to control and optimize the Van de Vusse reactor against unmeasured disturbances.
Macromolecules, 2021
To blend and process different high molar mass polymers into multiphase materials, predicting/con... more To blend and process different high molar mass polymers into multiphase materials, predicting/controlling their morphology and rheological behavior is paramount. Herein, a segmental repulsive poten...
Journal of Petroleum Science and Engineering, 2021
Abstract The natural gas produced in primary separation is passed through a compression system to... more Abstract The natural gas produced in primary separation is passed through a compression system to be pressurized and conditioned before being sent to its final destination. The operation of that system needs to be safe and efficient to avoid equipment damage and reduce energy consumption. The stable and secure operation of the equipment in a compression system is provided, many times, by a classical regulatory control layer. In this work, we present a Model Predictive Control (MPC) strategy to provide setpoints for the regulatory control layer of a gas compression system, aiming to avoid excessive energy consumption, decrease the plant variability, and guarantee a stable and safe operation against load disturbances. The proposed method is tested in a digital twin of a typical industrial unit using a Dead-Time Compensator Generalized Predictive Controller (DTC-GPC). Some disturbances in the feed flow rate of gas were considered as case studies. The controller responded satisfactorily to these disturbances keeping the plant operation stable and returning the controlled variables in the desired operating range after a short time.
The Canadian Journal of Chemical Engineering, 2020
Fluid Phase Equilibria, 2019
Industrial & Engineering Chemistry Research, 2019
Uploads
Papers by Argimiro R. Secchi