Bridging fundamental and processes for optimal control and robust optimization: A case study in oil transportation and CO2 capture in offshore platforms

Short summary of the project

This doctoral research aims to develop a self-sufficient and flexible advanced system encompassing optimal control that bridges the molecular scale with the operational scale. The focus is to create a system that continuously adapts and improves, seeking maximum efficiency at all process stages. The study will be approached in three main dimensions: i) fundamental scale, ii) processes and operations, and iii) integration and collaboration.

Included in the project are two specific case studies. The first will examine the oil transportation process. The second study will focus on CO₂ capture.

Detailed project 

The methodological strategy of this project is to address the issue in its essential aspects: i) fundamental, ii) process and operation, and iii) integration and cooperation. At the fundamental scale, it is necessary to enable the prediction of oil-water mixture properties for the oil transportation case study and the prediction of adsorption capacity based on the characteristics of adsorbents.

In this manner, AI models can be used to predict thermodynamic, kinetic, physicochemical, or other properties, starting from the molecular structure and selecting those with characteristics that ensure significant process potential and efficiency. An advantage of this methodology is that it supports traditional experimental methods and helps reduce the number of experiments, costs, and time demands involved in the project.

These AI tools should link molecular properties with the physicochemical characteristics of the substances involved in the focused process. These models will be tested and incorporated into the virtual process for validation in a simulated environment. This aspect's challenge is obtaining sufficient data without the need for extensive experiments.

The second aspect (ii) deals with the design of a reconfigurable process adaptable according to socioeconomic parameters. This aspect will be responsible for proposing the interconnection between different tools, including dynamics and artificial intelligence models, economic and process optimization, control, and uncertainty, to generate a unique and integrated process at the molecular scale. The challenge of this stage is to design a tool capable of using the information and characteristics generated at the fundamental scale during the reconfiguration and optimization of the process. That is, it is to design optimization strategies with molecular characteristics, for example, as decision variables, and relate them to the general objectives of the process. Another challenge of aspect (ii) is the hypothesis under models and optimization problems because the hypotheses imposed on the models and optimizers can generate underspecified models or make an optimization unfeasible. In a scenario like this, the solution is to reassess these hypotheses, rebuild the models or optimizations, or possibly use hybrid models. Finally, it is important to consider the computational effort necessary to solve models and optimize them together. However, using parallelization tools and other techniques, the solution is to use more efficient algorithms that extract the maximum potential from the available hardware.

Regarding aspect iii, dealing with a production system separated from a global system context becomes impossible. Thus, the process cannot operate in an isolated context but needs to consider its role in the globalized environment; for example, if the plant identifies an increase in demand, it will communicate with its supplier systems to ensure they have the necessary conditions to operate under the new conditions. Therefore, integration and cooperation are essential to enable the ideas discussed here.

More specifically, the last aspect (iii) addresses the connection of industrial processes to the entire ecosystem in which it is inserted. In this context, the chemical industry is positioned in a globalized context of intense renewal. Process performance is affected by plant operation, consumer demands, and environmental, social, political, and financial issues. The current social dynamics require an industry capable of responding quickly to external changes to defend itself from cyber threats (i.e., the flexibility to reconfigure itself), constantly optimizing and controlling its processes. These demands are beginning to surpass the human capacity to handle and respond quickly to all these situations simultaneously. On the other hand, delegating these tasks to technologies will lead to a safer, environmentally friendly, and efficient industrial operation capable of quickly adapting the process to the constantly changing environment, pointing to the urgent need to transform production chains to be more agile in adapting to anticipated uncertainties.