Toward an enhanced Bayesian estimation framework for multiphase flow soft-sensing

TL;DR

This paper enhances a Bayesian soft-sensing framework for multiphase flow by introducing automatic variance estimation methods, improving the accuracy of flow rate predictions based on sensor data.

Contribution

It proposes two automatic approaches for optimizing variance estimation in a Bayesian multiphase flow soft-sensing framework, replacing manual parameter tuning.

Findings

Framework with automatic variance estimation matches sensor measurements.

Improved flow rate estimation accuracy demonstrated in numerical example.

Automatic methods outperform manual variance selection.

Abstract

In this work the authors study the multiphase flow soft-sensing problem based on a previously established framework. There are three functional modules in this framework, namely, a transient well flow model that describes the response of certain physical variables in a well, for instance, temperature, velocity and pressure, to the flow rates entering and leaving the well zones; a Markov jump process that is designed to capture the potential abrupt changes in the flow rates; and an estimation method that is adopted to estimate the underlying flow rates based on the measurements from the physical sensors installed in the well. In the previous studies, the variances of the flow rates in the Markov jump process are chosen manually. To fill this gap, in the current work two automatic approaches are proposed in order to optimize the variance estimation. Through a numerical example, we show that, when the estimation framework is used in conjunction with these two proposed variance-estimation approaches, it can achieve reasonable performance in terms of matching both the measurements of the physical sensors and the true underlying flow rates.