To address the dynamic operational demands of the gas processing sector, which is continuously evolving due to gas market opening, increase in natural gas production, and the growing challenge of upstream-midstream integration in a competitive environment, this work presents the Integrated-Gas-Scheduling-System, IntegraGAS. The proposed methodology innovates by using first principles rigorous process simulation coupled with a scheduling tool for short/medium/long-term, enabling gas plants to swiftly adapt to varying operational conditions and meet the requirements of this new market. IntegraGAS was implemented in Petrobras and has significantly enhanced scheduling efficiency, reducing execution time by up to 99.2% and avoiding approx. US$ 2.3 million in annual labor costs, optimizing resource utilization. By integrating Excel for the frontend, Aspen HYSYS for process simulation, VBA for automation, and Microsoft PowerBI for real-time data visualization, IntegraGAS improves decision-making, regulatory compliance, and operational agility. Its key functionalities include alerts for operational limit violations, automated mass and energy balance calculations, optimized gas allocation, integration with maintenance shutdown plans and KPI monitoring. With an intuitive interface and a robust architecture driven by digital transformation, IntegraGAS eliminates manual inefficiencies, ensures seamless coordination among stakeholders, and enables rapid responses to market dynamics.

The objective of this work is to propose a simulation strategy for production planning that is compatible with the dynamism of natural gas processing, especially under an open-market arrangement, in which several scheduling simulations must be performed within short time horizons. In such contexts, traditional first-principles-based approaches, although accurate, require prohibitive computational times, motivating the need for an alternative simulation strategy. This work thus proposes a data-driven model built with the aid of machine learning and applied in a case study with historical data from the largest gas processing site in Brazil: Cabiúnas Petrobras asset. Main plant flowrates were selected: 18 targets and 44 input candidates—1282 observations from three and a half years of operation. Principal Component Analysis was used for order reduction, keeping the 22 main principal components. A forward neural network (2 hidden layers and 225 neurons per layer) was built from training/test sets randomly selected and optimized hyperparameters—learning rate (0.001533) and batch size (8). Training converged in roughly 200 epochs (Adam optimizer), with early stop triggered by the validation set. A mean absolute error of 0.0017 (test set) and R² = 0.72 were found, a promising result considering plant complexity and data simplicity. Results showed a particularly good fit for lighter products (sales gas and natural gas liquid), also indicating an opportunity for further work by including inputs related to liquid fractionation.

The objective of this work is to propose a simulation strategy for production planning that is compatible with the dynamism of natural gas processing, especially under open-market arrangement, in which several scheduling simulations must be performed within short time horizons. In such contexts, traditional first-principles-based ap-proaches, although accurate, require prohibitive computational times, motivating the need for an alternative simulation strategy. This work thus proposes a data-driven model built with the aid of machine learning and applied in a case study with historical data from the largest gas processing site in Brazil: Cabiúnas Petrobras asset. Main plant flowrates were selected: 18 targets and 44 input candidates – 1282 observations from three and a half years of operation. Principal Component Analysis was used for order reduction, keeping the 22 main principal components. A forward neural network (2 hidden layers and 225 neurons per layer) was built from training/test sets randomly selected and optimized hyperparameters – learning rate (0.001533) and batch size (8). Training converged in roughly 200 epochs (Adam optimizer), with early stop triggered by validation set. A mean absolute error of 0.0017 (test set) and R2=0.72 were found, a promising result considering plant complexity and data simplicity. Results showed particularly good fit for lighter products (sales gas, natural gas liquid), also indicating an opportunity for further work by including inputs related to liquid fractionation.