Controle preditivo com estimação bayesiana e monitoramento da temperatura do óleo em um sistema de tubulações multicamadas

Abstract

Ensuring oil flow has become a subject of study since the extraction of oil has hit wells in ultra-deep waters. Among the challenges, the deposition of solids in the transport ducts causing a partial or total blockage in the pipelines and consequently unexpected expenses. The pipe-in-pipe (PIP) system is a developed technology that combines the use of thermal insulation and active heating in the pipelines. Optimum temperature monitoring and control of the PIP system ensures that oil flow occurs without obstruction. Thus, this dissertation proposes a temperature control system usinga model predictive controlassociated with the particle filter (PF-MPC) to prevent the fall of temperature in the PIP.This control scheme combines the reduction of uncertainty of measurement by particle filter (PF) with great handling the flow of heat generated at the active heating required to avoid cooling the PIP. First it was studied the robustness only particle filter in reduction of ruin. Subsequently, the PF-MPC controller was implemented the PIP system.The results showed that the particle filter provided estimates of the temperature, and that the PF controller-MPC presented good results and satisfactory performance in temperature control, highlighting your potential as a tool of control.