A Kalman estimation based oil–water two-phase flow measurement with CRCC. (June 2015)

• Record Type: Journal Article
• Authors: Tan, Chao; Dai, Wei; Yeung, Hoi; Dong, Feng
• Is Part Of: International journal of multiphase flow. Volume 72(2015)
• Page Start: 306
• Page End: 317
• Publication Date: 2015-06
• Subjects: Oil–water two-phase flow -- Conductance ring coupled cone -- Sensor fusion -- Kalman filter -- Conductance ring sensor -- Long-waist cone; Multiphase flow -- Periodicals; Écoulement polyphasique -- Périodiques; Multiphase flow; Periodicals; 620.1064
• DOI: 10.1016/j.ijmultiphaseflow.2014.06.014
• Languages: English
• ISSNs: 0301-9322

Abstract

Highlights: A Conductance Ring Coupled Cone (CRCC) is used for oil–water two-phase flow rate measurement. Measurement data from heterogeneous sensors are fused by Kalman estimation method. Random walk model and differential pressure model are adopted in Kalman estimation. Kalman estimation model is optimized with Expectation–Maximization algorithm. Results are compared with model based estimation and conductance ring array based estimation. Abstract: Measuring the individual phase flow rates of oil–water two-phase flow is an important issue in process industries. A Conductance Ring Coupled Cone (CRCC) meter is presented for this task, and conductance ring array (CRA) was installed in front of CRCC for results comparison. A CRCC provides multiple outputs regarding the flow rate and water holdup by implementing a pair of conductance rings in an annular flow channel formed by a long-waist cone and pipe wall. Three outputs, contracting differential pressure Δ p tp, overall differential pressure δ p tp and dimensionless conductance V a ∗, are fused by a Kalman filter to estimate individual flow rate of oil ( W o ) and water ( W w ) simultaneously. A transition model with state vector [ W o, W w ] T is established based on the random walk assumption of two-phase flow, and an observation model is established with Δ p tp and δ p tp based on the separated flow assumption. The parameters of these two models are optimized with the Expectation–Maximization (EM) algorithm under each experimental condition. The present method treats the two-phase flow a dynamic process with a dynamic state estimation method. The estimation results demonstrate a dynamic response of CRCC to the change of flow conditions. The overall accuracy of the oil flow rate estimation is 2.3 % and for the water flow rate is 4.8 % . The comparison of the proposed method with CRCC model based dynamic/static estimation, as well as the CRA model based estimation, indicates that the proposed sensor structure improved the estimation accuracy by measuring the water holdup in annular channel. The dynamic estimation can also improve the estimate accuracy by incorporating dynamic fluctuation and unknown correlations with inferential parameters through optimizations. More accurate and reliable estimates can be obtained by optimizing the transition and observation models with profound multiphase physical flow models.