A synopsis of our research is given below. If this research interests you, check out current openings in our group. You may contact us if you have any questions or are interested in working with us.
Introduction and Motivation
Our research spans two core areas of Chemical Engineering: Reaction Engineering and Process Systems Engineering. The crux of research can be summarized as multi-scale modelling, design and control of multifunctional microreactors for energy applications. Multiscale modeling links molecular phenomena with macroscopic variables and properties, to guide rational design of materials and catalysts. A better understanding of the underlying physical phenomena allows us to exploit spatial and temporal patterning of flow, temperature and catalyst states for improved productivity. Finally, we are interested in using optimisation and control tools to operate devices at their optimal efficiency and productivity.
Schematic of a Fuel Processing / Fuel Cell System
The above figure shows a schematic of a fuel processing fuel cell system used in portable and decentralised power generation applications. The main reaction for generating hydrogen is steam reforming of a hydrocarbon or alcohol fuel. A burner provides heat required for the endothermic reforming. Water gas shift (WGS) and preferential oxidation (PROX) units are required to reduce the amount of CO in the fuel cell inlet stream. The hydrogen is electrochemically oxidised in the fuel cell, accompanied by generation of electricity.
Current Research Areas
Multiscale Modeling and Control of Integrated Fuel Processing Fuel Cell System
The main challenges in modeling and control of the Fuel Processing Fuel Cell system (shown in the schematic) arise due to: (a) increased coupling between various subsystems; (b) distributed parameter nature of the models; (c) complex dynamics; and (d) transient operation due to frequent load changes and startup and shutdown. The objective is to build dynamic, computationally efficient and comprehensive models for the entire integrated plant. The strong coupling between the subsystems and the time scales of operation makes the system amenable to "plant-wide" control. The main objective is not just to look at individual units, but, more importantly, to focus on interaction and coupling between the various units in the entire system.
Process Intensification in Thermally Coupled Microdevices
Process Intensification refers to reduction of process strategy by combining several unit operations into a single multifunctional device. Strategies for efficient thermal coupling between the endothermic and exothermic processes will be investigated. Advances in Micro Electro Mechanical Systems (MEMS) technology allows us to carefully design microreactors to achieve complex spatio-temporal patterns and provide efficient coupling between various units. We use a hierarchical modelling approach, wherein different models with varying complexity and accuracy are deployed for this purpose.
Transient Microreactor Analysis
Microreactors for portable power generation are often used in a transient, non-steady operation during start-up / shutdown and due to frequent load changes. Moreover, dynamic system analysis is useful to study incipient system instabilities, which need to be avoided for safe and efficient device operation. Finally, our interest in transient analysis stems from the possibility of accessing "novel" catalyst states through forced unsteady state operation (such as, reverse flow).
Approximate Dynamic Programming (ADP)-based control of Fuel Cells
The objective of Dynamic Programming (DP) is to find the optimal control law that relates control actions to the information state of the system. The optimal control policy is computed offline. The advantage of DP over conventional control techniques is lower online computational burden and systematic handling of process uncertainties. The computation of optimal control policy increases exponentially with state dimension. ADP aims to alleviate this so-called curse of dimensionality by computing the control policy in an approximate way using simulations and iterative improvement of suboptimal control policies. We aim to develop improved risk-averse model-based as well as model-free ADP techniques for online optimal control.