Overview of estimation methods for industrial dynamic systems

Abstract

Measurement technology is advancing in the oil and gas industry. Factors such as wireless transmitters, reduced cost of measurement technology, and increased regulations that require active monitoring tend to increase the number of available measurements. There is a clear opportunity to distill the recent flood of measurements into relevant and actionable information. Common methods to do this include a filtered bias update, Implicit Dynamic Feedback, Kalman Filtering, and Moving Horizon Estimation. The purpose of these techniques is to validate measurements and align imperfect mathematical models to the actual process. Additionally, they can determine a best-estimate of the current state of the process and any potential disturbances. These methods allow potential improvements in earlier detection of disturbances, process equipment faults, and improved state estimates for optimization and control.

Automatic Generation of a Simulation-Based Digital Twin of an Industrial Process Plant

“…Model adaptation is based on applying model optimization methods for reducing model residuals. These methods focus on adjusting model parameters utilizing recent plant measurements, obtained from historical process data [58]. In highly complex systems, model optimization results may fail to converge at the required cycle time [59].…”

“…Consequently, these methods must be executed offline. In contrast, dynamic estimation methods, such as Kalman filtering [60] or implicit dynamic feedback [58], can be used for real-time model parameter estimation. Dynamic estimation, also known as online calibration [10], can be seen as a feedback process targeted to bring the model results to the real-time process measurements.…”

“…In the process industry domain, applications which fulfil some of the characteristics of DTs have been proposed in recent decades. They follow implementation approaches for capturing the process dynamics that differ from those followed in the manufacturing domain, as discrete-event modelling cannot be applied for prediction of continuous systems [43], [58]. The commercially-available DTs for process plants [20], [21], are based on DDMs for representing the plant dynamics.…”

“…OMBAs [10], [58] are online simulation systems that are continuously adjusted by dynamic estimation methods to ensure that their result will correspond to the current process state. These systems, also known as symbiotic simulation systems [106]- [108], have been developed for applications ranging from virtual sensors [109] and prediction systems [5] to model-based control [110].…”

A Continuous Formulation for Logical Decisions in Differential Algebraic Systems using Mathematical Programs with Complementarity Constraints

“…This is done using a DAE solution package known as APMonitor [40][41][42]. This software package allows a user to define a model using both differential and algebraic equations [43,44]. The software performs the collocation to convert the differential equations to algebraic equations and the problem is converted to a set of nonlinear algebraic equations.…”