Prediction of the Temperature Field in Pipelines with Bayesian Filters and Non-intrusive Measurements

One of the greatest challenges for the production of petroleum in deepwater is flow assurance. In fact, knowledge about the transient cool down behavior of the produced fluid is necessary to prevent the formation of hydrates and solid deposits during shutdown periods, which could result in a pipeline blockage and could result in large financial losses. In a typical subsea petroleum production system, the information provided by its monitoring system, regarding the temperature field is limited. One approach to predict the produced fluid temperature field in a pipeline system is to use Bayesian filters. In this paper, we compare the Kalman filter and the particle filter as applied to a problem of practical interest for the petroleum industry. Uncertainties in the state evolution and measurement models are taken into account by assuming that the errors involved are additive, normally distributed and with known means and covariance matrices.