PDG Pressure Estimation in Offshore Oil Well: Extended Kalman Filter vs. Artificial Neural Networks

Andressa Apio1, Jônathan Dambros1, Fabio Cesar Diehl2, Marcelo Farenzena1, Jorge Otávio Trierweiler3

  • 1Federal University of Rio Grande do Sul
  • 2CENPES, Petrobras
  • 3Federal University of Rio Grande do Sul (UFRGS)

Details

11:40 - 12:00 | Thu 25 Apr | Fauna | ThA2.4

Session: Control and Optimization for Oil and Gas Production 1

Abstract

The Permanent Downhole Gauge (PDG) pressure measurement is of great importance for offshore oil well modeling and control since it is measured close to the bottom hole. The PDG is installed in a remote undersea environment, which makes expensive the maintenance in case of fault. For this reason, PDG measurements are frequently unavailable. To overcome this limitation, the PDG pressure can be estimated using other available measurements. The estimation is not a simple task since, depending on process operational conditions, the multiphase flow might present limit cycles. In this work, Artificial Neural Network (ANN) and Extended Kalman Filter (EKF) are proposed as potential techniques for the PDG pressure estimation. The comparison of the results shows that ANN returns precise estimation for a short-time window after the failure, but fails when a different process operating condition is applied, while EKF returns good estimation in all the cases.