Froth Pipeline Water Content Estimation and Control

Abstract

This paper presents a successful application of soft sensor based control for froth pipeline water content in oil sands industry. Water content is a key process quality variable for inter-site froth transportation through pipeline, which should be controlled within a specific range by adding hot processing water into the pipeline to maintain a reliable operating condition. A dynamic soft sensor incorporated with on-line bias update is developed to estimate the water content in real-time. Based on the soft sensor, a control strategy is consequently implemented to control the froth pipeline water content. Significant improvement of water content control is obtained from the successful application to a froth pipeline, which demonstrates the effectiveness, reliability and advantage of the soft sensor and control proposed.

Extracting dynamic features with switching models for process data analytics and application in soft sensing

“…It should be noted that the constraint (3) is implemented for the continuous or relatively steady process, providing the benefit as a regulator. While it is necessary to assume significant transition, such as batch process, this constraint could be removed, and the switching structure is not limited to (3). Thus, an illustration of this extraction model can be presented in Figure 2: (1) S 1:T is connected with the transition model to capture common driving forces among different modes; (2) multiple emission models M ð1:KÞ map the latent feature to observations X 1:T and reflect the mode changes.…”

“…Typically, online estimation of quality variables can be improved by analyzing process data and developing soft sensors. [1][2][3] Similar to the hardware instruments, however, knowledge based soft sensors can fail frequently due to harsh operation conditions and unattainable assumptions. Data-driven approaches have proven to be a good alternative for soft sensor development.…”

An Adaptive Soft Sensor for On‐Line Monitoring the Mass Conversion in the Emulsion Copolymerization of the Continuous SBR Process

“…[23,24] Concerning the timing, it is common to update the bias when a new laboratory report is available. [18,19,25,26] This strategy leads to implement unnecessary bias updates (e.g., when the process is in a steady state and the prediction error changes due to usual sensor noises or laboratory errors). Furthermore, frequent bias updates can degrade the performance of closed-loop control strategies that use the output of the SS as a feedback signal.…”