Gestión de cronogramas para la actualización del sistema SCADA de la GRB en SOLTEC S.A

Abstract

This project documents the planning and analytical development undertaken for the modernization of the E-TERRA SCADA system at the Barrancabermeja Refinery (GRB). The primary objective was to integrate the architectural transition from a legacy network topology, together with the overall system upgrade processes, into a predictive management framework powered by Artificial Intelligence (AI). The methodology followed an applied engineering approach structured in sequential phases: vulnerability assessment through an On-Site Technical Survey (TSS), baseline parameterization and Earned Value Management (EVM) using Microsoft Project, and iterative software development. At the physical layer, the results consolidated the design of a Redundant Dual-Star topology, leveraging dark fiber links to comply with IEC 61850 standards and the Parallel Redundancy Protocol (PRP). At the management layer, an independent graphical user interface was developed in Python to directly consume the Anthropic API. The system extracts metrics from isolated Microsoft Project (.mpp) files through automation routines (pywin32) and processes operational performance indicators (SPI/CPI) to generate automated diagnostic reports (Flash Reports). Validation in controlled environments demonstrated that delegating analytical processing to Large Language Models (LLMs) enables Human–AI Symbiosis. It is concluded that this ecosystem streamlines strategic decision-making, identifies operational governance gaps, and proactively mitigates execution risks in critical infrastructure projects.