Showing papers by "Raghunathan Rengaswamy published in 2010"

PEMFC Fault Diagnosis, Modeling, and Mitigation

Abstract: This paper introduces fault diagnosis and separation, mitigation, and modeling of a proton exchange membrane fuel cell (PEMFC). Experimental tests of a single PEMFC were performed during this study. Flooding and drying faults were implemented to be detected from the cell voltage and impedance response of the cell. The impedance response at low frequency was used to identify the cause of the fault. The slope of the magnitude and/or the negative phase response of the cell impedance at low frequency were observed to allow separation of a fault. A cell impedance model based on resistive capacitive (C model) and resistive constant-phase-element (CPE model) circuits is developed. The CPE model has a better approximation of the cell impedance. However, the C model is easy to implement since it is well known in most simulation tools (MATLAB/Simulink or PSpice). A power electronic control is designed and tested to mitigate the faults. Pulsing the cell current at low frequency was seen to increase the cell power by 8% during drying.

System Identification and Nonlinear Model Predictive Control of a Solid Oxide Fuel Cell

Abstract: Solid oxide fuel cells (SOFCs) are high temperature fuel cells with a strong potential for stationary power house applications. However, considerable challenges need to be overcome to connect these cells to the power grid. The fluctuating grid demand has to be met without sacrificing the cell efficiency and causing structural/material damage to the system. This requirement coupled with fast and highly nonlinear transients of the transport variables results in a challenging control problem. This paper is on synthesizing a controller that can address some of these challenges. For using in the model predictive controller (MPC), input−output models are identified from the data generated by a detailed dynamic model. A traditional SISO control and a novel MIMO control are considered here. In the SISO control problem, power is the controlled variable (CV) and H2 flow is the manipulated variable (MV). In the MIMO control problem, power and the utilization factor (UF) of the fuel are the CVs while voltage and the ...

Achieving resilience in critical infrastructures: A case study for a nuclear power plant cooling loop

Abstract: Engineered systems are increasingly equipped with sensing and actuating equipment making the operation ans supervisory task increasingly difficult to handle by means of human interaction alone. In particular, the detection, identification and accommodation of abnormal, potentially harmful, events has been a long-standing challenge. Many scientists in different scientific areas have attacked this problem which has resulted in a plethora of techniques for both Fault Detection and Identification (FDI) and advanced control, each with their strengths and weaknesses. Because of the diverse nature of adopted theory and paradigms and because of a historical separation of FDI specialists and control theoreticians, it remains a challenge to establish automated systems able to handle exceptional events with minimal human intervention. As such, a project has been set up to enable full integration of diverse FDI methods as well as optimal coupling of FDI modules and control modules in the closed-loop supervisory control system. In this contribution, we introduce the basic paradigms of our approach, a strategic plan to achieve this goal as well as some preliminary results.

Control of proton exchange membrane fuel cells using data driven state space models

Abstract: Proton exchange membrane fuel cells (PEMFCs) are increasingly being researched upon due to their potential toward sustainable energy generation. Toward improved productivity of PEMFCs, it is important to develop systematic approaches for optimization and control of their operations. PEMFCs pose interesting challenges toward these tasks due to their complex behavior such as nonlinearity and spatial variations. While first principles model based approaches could be used, a more mathematically attractive and cost-effective alternative is to use empirical modeling approaches for representing the system dynamics toward optimization and control. In this paper, we propose to use a novel, innovation form of state space models that facilitate the development of advanced control algorithms such as linear quadratic Gaussian (LQG) and model predictive control (MPC), and provide improved disturbance rejection necessary for these applications. We demonstrate the applications of such model based algorithms via simulations involving a distributed along-the-channel model of the PEMFC, and also present experimental validation on a PEMFC setup.

Development of a Closed Form Nonlinear Predictive Control Law Based on a Class of Wiener Models

Abstract: Nonlinear model predictive control (NMPC) is increasingly being used for controlling microscale and system-on-chip devices, which exhibit complex and very fast dynamics. For effective control of such systems it is necessary to develop computationally efficient approaches for solving the NMPC problem. In this work, a Wiener type model has been used for capturing dynamics of multivariable nonlinear systems with fading memory. The resulting discrete nonlinear state space model is used to generate multistep predictions and formulate an unconstrained NMPC problem. A closed form solution to this problem is constructed analytically using the theory of solutions of quadratic operator polynomials. The effectiveness of the resulting quadratic control law is demonstrated by conducting simulation studies on a proton exchange membrane fuel cell (PEMFC) system, which exhibits fast dynamics and input multiplicity behavior. The quadratic control law is expected to control the PEMFC at its optimum (singular) operating poi...

Resilient control in view of valve stiction: extension of a Kalman-based FTC scheme

Abstract: In this contribution we propose an active Fault Tolerant Control (FTC) strategy which enables the isolation and identification of valve stiction and valve blocking, in addition to the additive faults like sensor and actuator biases. The developed method is an extension of the original method proposed by Prakash et al . (2002). This method is based on the Kalman filter and is developed under the assumption that the monitored system is Linear Time Invariant (LTI). It has been shown to work well for additive faults such as sensor and actuator biases. Within this method the fault isolation and identification task is based on the Generalized Likelihood Ratio (GLR) test by which the most plausible fault type in a library of faults is selected following estimation of fault parameters.