Author ORCID Identifier
Semester
Fall
Date of Graduation
2024
Document Type
Dissertation
Degree Type
PhD
College
Statler College of Engineering and Mineral Resources
Department
Chemical and Biomedical Engineering
Committee Chair
Debangsu Bhattacharyya
Committee Member
Fernando V. Lima
Committee Member
Stephen E. Zitney
Committee Member
Sarika Khushalani Solanki
Committee Member
Benjamin Omell
Abstract
The integration of renewable energy sources into the electric grid requires the current fleet of fossil-fired power plants to operate more flexibly, leading to frequent load changes that stress critical high temperature boiler components that were primarily designed for base load operations. The more complex operating scenarios can modify component damage, compromising reliability and increasing operational costs. To ensure efficient plant operations, adaptive condition monitoring tools generalizable for different plant configurations are essential for recognizing the impacts of load-following, improving safety, and preventing outages. This dissertation work is primarily focused on the development of advanced modeling and estimation techniques frameworks which can be used for condition monitoring of critical components in advanced and integrated energy systems. These tools can eventually help these plants make efficient plans for preventive maintenance, avoid undesired forced outages and develop advanced process control strategies for improved flexibility without compromising safety and reliability.
Condition monitoring tools using large amounts of measurement data have been extensively developed and widely reported in the existing literature for rotary equipment items such as gas and steam turbines. However, condition monitoring of static equipment items such as the boiler components is challenging due to the difficulty in estimating equipment health by simply using the measurement data. Furthermore, due to harsh operating conditions in high temperature regions that are most vulnerable to failure, especially in the combined cycle and coal-fired plants boiler systems, there can be no or limited number of sensors. In addition, it is impractical to measure certain variables of interest. For example, even though the through-wall temperature profile of a tube in the high temperature superheater section of a boiler can provide important indication of scale formation or fouling, it is impractical to estimate the through-wall temperature profile. In order to address these challenges, this dissertation work made the following contributions: (i) development of rigorous, detailed dynamic lumped and distributed parameter first-principles (FP) dynamic models for advanced process control and condition monitoring of boiler components; (ii) development of advanced and adaptive estimator algorithms for using these complex models for condition monitoring of operating plants using limited measurement data; (iii) development of hybrid grey-box modeling frameworks that combines FP physics-based and data-driven artificial intelligence (AI) models for a comprehensive health monitoring framework; (iv) development of novel constrained estimation algorithms to satisfy mass and energy balances for complex dynamic systems.
Dynamic modeling and estimation in operating plants can be valuable for process monitoring and control applications. However, most industrial processes are highly nonlinear, with complex dynamics and limited measurement data available only for certain variables. In order to develop adaptive and advanced monitoring tools requires detailed models characterizing the complete system geometry and its real-time behavior in different operating scenarios. Models developed using operational data have simple online adaptation formulation but lack the predictive capabilities required for complicated tasks. Ordinary differential equation (ODEs) or partial differential equation (PDEs) based models can represent the dynamic behavior of energy systems quite well but fail to capture the effect of unknown system parameters arising due to flow distribution, heat transfer coefficient, or other thermodynamic constraints. Hence, a differential algebraic equation (DAE) model should be used to model such systems accurately. In this dissertation work we have developed a distributed DAE model to characterize high-temperature boiler components like superheaters/reheaters to study the impact of load-following on their performance. The proposed models were based on mass and energy balances for the operating fluids, i.e., steam and flue gas, with due consideration for tube wall dynamics included in them. The tubes modeled were assumed thick with conduction across them to obtain through-wall temperature profiles, which any sensors cannot measure. Rigorous properties models for heat transfer calculations and geometrical characterization were also considered for model development. The developed models were validated using industrial operating data from various power plants.
Furthermore, novel hybrid grey-box modeling approaches were developed by combining the rigorous-physics-based model with a data-driven black box model for modeling systems with unknown phenomena or poorly understood physics. The physics models were based on the distributed DAE model, while the data-driven black box models were developed using machine learning (ML) techniques. The proposed grey-box systems have been utilized in a health monitoring framework to predict oxide scale formation and tube life consumption in boiler systems. The dynamic models developed were utilized for state and parameter estimation of operating plants using Kalman filter (KF) based estimation techniques. Since the developed DAE and grey-box models provide suboptimal results with the standard extended Kalman filter (EKF) algorithm, a modified EKF has been utilized for DAE systems. The modified EKF can handle uncertainties in both differential and algebraic equations with due consideration of noise in both differential and algebraic states. Furthermore, constrained state estimation algorithms were developed that can satisfy mass and energy balances. The proposed estimators have been validated using literature and industrial data while keeping computation time low for online implementation.
Recommended Citation
Saini, Vivek, "Development of Dynamic Modeling and Estimation Techniques for Condition Monitoring of Advanced Energy Systems" (2024). Graduate Theses, Dissertations, and Problem Reports. 12673.
https://researchrepository.wvu.edu/etd/12673
Included in
Complex Fluids Commons, Process Control and Systems Commons, Transport Phenomena Commons
