Showing papers by "Raghunathan Rengaswamy published in 2011"

Multivariable optimization studies of cathode catalyst layer of a polymer electrolyte membrane fuel cell

Abstract: The amount of current generated in a polymer electrolyte membrane fuel cell (PEMFC) depends strongly on the local conditions in a cathode such as available oxygen, surface area available for the reactions, amount of ionomer, and amount of electro-catalyst. In the present work, design parameters of a cathode catalyst layer are optimized to achieve the maximum current density at a given operating voltage. The decision variables are chosen such that they can be realized experimentally. To understand the effect of the model fidelity on the decision variables, optimization is performed with a single phase model and a two-phase model with and without membrane. Other objective functions such as maximization of current generation per catalyst loading, minimization of catalyst layer cost per power and minimization of cell cost per power are also considered to study the effects of the objective functions on the decision variables.

Plant Friendly Input Design: Convex Relaxation and Quality

Abstract: A common practice in a system identification exercise is to perturb the system of interest and use the resulting data to build a model. The problem of interest in this contribution is to synthesize an input signal that is maximally informative for generating good quality models while being “plant friendly,” i.e., least hostile to plant operation. In this contribution, limits on input move sizes are the plant friendly specifications. The resulting optimization problem is nonlinear and nonconvex. Hence, the original plant friendly input design problem is relaxed which results in a convex optimization problem. We formulate a SemiDefinite Programme using the theory of generalized Tchebysheff inequalities to derive tight bounds on the quality of relaxation. Simulations show that the relaxation results in more plant friendly input signals.

Drop Traffic in Microfluidic Ladder Networks with Fore-Aft Structural Asymmetry

Abstract: We investigate the dynamics of pairs of drops in microfluidic ladder networks with slanted bypasses, which break the fore-aft structural symmetry. Our analytical results indicate that unlike symmetric ladder networks, structural asymmetry introduced by a single slanted bypass can be used to modulate the relative drop spacing, enabling them to contract, synchronize, expand, or even flip at the ladder exit. Our experiments confirm all these behaviors predicted by theory. Numerical analysis further shows that while ladder networks containing several identical bypasses are limited to nearly linear transformation of input delay between drops, mixed combination of bypasses can cause significant non-linear transformation enabling coding and decoding of input delays.

Optimal Plant Friendly Input Design for System Identification

Abstract: The primary objective in solving optimal input design problems is to obtain maximally informative inputs to be used as perturbation signals in system identification experiments. In plant-friendly identification, the designer has to respect constraints on experiment time, input and output amplitudes or input move sizes. This work focuses on plant friendly input design with constraints on input move size and output power. We present a convex relaxation to the problem of designing an informative input subject to input move size and output power constraints. The problem is finitely parametrized using ideas from Tchebycheff systems and reformulated as a SemiDefinite Programme.

Computationally Efficient Identification of Global ARX Parameters With Guaranteed Stability

Abstract: Identification of stable parametric models from input-output data of a process (stable) is an essential task in system identification For a stable process, the identified parametric model may be unstable due to one or more of the following reasons: 1) presence of noise in the measurements, 2) plant disturbances, 3) finite sample effects 4) over/under modeling of the process and 5) nonlinear distortions Therefore, it is essential to impose stability conditions on the parameters during model estimation In this technical note, we develop a computationally efficient approach for the identification of global ARX parameters with guaranteed stability The computational advantage of the proposed approach is derived from the fact that a series of computationally tractable quadratic programming (QP) problems are solved to identify the globally optimal parameters The importance of identifying globally optimal stable model parameters is high lighted through illustrative examples; this does not seem to have been discussed much in the literature

Receding Nonlinear Kalman (RNK) Filter for Nonlinear Constrained State Estimation

Abstract: State estimation is an important problem in process operations. For linear dynamical systems, Kalman Filter (KF) results in optimal estimates. Chemical engineering problems are characterized by nonlinear models and constraints on the states. Nonlinearities in these models are handled effectively by the Extended Kalman Filter (EKF), whereas constraints pose more serious problems. Several constrained estimation problems where the EKF approach fails have been reported in the literature. To address this issue, receding horizon approaches such as the Moving Horizon Estimation (MHE) have been proposed. The MHE approach has been shown to provide the most reliable estimates in several example problems; albeit at a high computational price. Unlike the KF, the MHE formulation does not use an explicit predictor-corrector approach. In this paper, we study the following questions in nonlinear constrained state estimation: (i) can the EKF be extended to include a receding horizon in a simple intuitive fashion? (ii) are there any performance gains over an EKF due to a receding horizon? and, (iii) are there any computational gains over the standard MHE through such an extension? A Receding Nonlinear Kalman (RNK) Filter formulation is proposed to answer these questions. The RNK formulation follows a predictor-corrector approach and uses linearization of the state space model for covariance calculation much like the EKF approach. We demonstrate through examples that inclusion of a receding horizon improves performance over the standard EKF approach. We also discuss the computational properties of RNK in comparison with MHE.

Multiple root cause analysis of linear oscillatory closed-loop single-input single-output (SISO) systems

Abstract: In general, oscillatory variables indicate poor performance of control loops. Therefore, diagnosis of the causes for oscillations in control loops is vital for maintaining the product quality within desired limits. In a linear closed-loop SISO system, oscillations can occur due to one or more of the following reasons: (i) poor controller tuning, (ii) control valve stiction and, (iii) external oscillatory disturbances. Several offline data-driven methods have been developed to address the diagnosis problem by focusing on only one of the causes for oscillations. In this work, an algorithm for identification of multiple root causes for oscillations in closed-loop systems is presented. The proposed approach comprises of: (i) Hammerstein based stiction detection algorithm, (ii) amplitude based discrimination algorithm using Hilbert Huang (HH) spectrum for identification of controller and disturbance caused oscillations and, (iii) an algorithm for analyzing the model obtained from Hammerstein approach. A decision algorithm based on the information obtained from the above three components is used for determination of multiple causes for oscillations in linear SISO systems.

Application of empirical mode decomposition in the field of polymer physics

Abstract: Noisy data has always been a problem to the experimental community. Effective removal of noise from data is important for better understanding and interpretation of experimental results. Over the years, several methods have evolved for filtering the noise present in the data. Fast Fourier transform (FFT) based filters are widely used because they provide precise information about the frequency content of the experimental data, which is used for filtering of noise. However, FFT assumes that the experimental data is stationary. This means that: (i) the deterministic part of the experimental data obtained from a system is at steady state without any transients and has frequency components which do not vary with respect to time and (ii) noise corrupting the experimental data is wide sense stationary, that is, mean and variance of the noise does not statistically vary with respect to time. Several approaches, for example, short time Fourier transform (STFT) and wavelet transform-based filters, have been developed to handle transient data corrupted with nonstationary noise (mean and variance of noise varies with respect to time) data. Both these approaches provide time and frequency information about the data (time at which a particular frequency is present in the signal). However, these filtering approaches have the following drawbacks: (i) STFT requires identification of an optimal window length within which the data is stationary, which is difficult and (ii) there are theoretical limits on simultaneous time and frequency resolution. Hence, filtering of noise is compromised. Recently, empirical mode decomposition (EMD) has been used in several applications to decompose a given nonstationary data segment into several characteristic oscillatory components called intrinsic mode functions (IMFs). Fourier transform of these IMFs identifies the frequency content in the signal, which can be used for removal of noisy IMFs and reconstruction of the filtered signal. In this work, we propose an algorithm for effective filtering of noise using an EMD-based FFT approach for applications in polymer physics. The advantages of the proposed approach are: (i) it uses the precise frequency information provided by the FFT and, therefore, efficiently filters a wide variety of noise and (ii) the EMD approach can effectively obtain IMFs from both nonstationary as well as nonlinear experimental data. The utility of the proposed approach is illustrated using an analytical model and also through two typical laboratory experiments in polymer physics wherein the material response is nonstationary; standard filtering approaches are often inappropriate in such cases. © 2010 Wiley Periodicals, Inc. J Polym Sci Part B: Polym Phys, 2011