(7gw) Data Driven Modeling and Control for Engineering Next-Generation Processes
- Conference: AIChE Annual Meeting
- Year: 2017
- Proceeding: 2017 Annual Meeting
- Group: Meet the Faculty Candidate Poster Session - Sponsored by the Education Division
- Time: Sunday, October 29, 2017 - 1:00pm-3:30pm
Keywords: Dynamic systems and control, multiscale modeling, machine learning for system identification, nonlinear systems analysis
Application areas: bioengineering and biomanufacturing, photovoltaics and semiconductor manufacturing
Process systems engineering has enabled routine high performance and high reliability in modern chemical manufacturing. As the interests of chemical engineers have diversified to new domains including biology, materials science, and alternative energy, new tools are needed to bring fundamental advances to bear. The systems that govern these new domains are complex and nonlinear, which makes conventional systems analysis and process engineering challenging or impossible. In my work, I use theory and computation for understanding complex systems, developing effective dynamic models, and designing new processes and robust control systems so that these new technologies can be deployed. Here, I present two two example of my research:
1. Postdoctoral Project: Data-driven analysis and control of novel photobioreactors
Advised by Yannis G. Kevrekidis at Johns Hopkins University and Princeton University
Recent advances in machine learning, data science, and synthetic biology enable a new biomanufacturing platform where high-resolution process inputs and outputs can be used for online optimization and control. In particular, biosensors and optogenetic âswitchesâ allow for measuring and manipulating the state of a bioreactor with more precision than ever before. In this âphotobioreactionâ process, the reactor state is observed using biosensors measured with an IR detector and actuated with LEDs that turn metabolic pathways âonâ and âoffâ using optogenetics. Abundant state information from biosensors can be used for data-assisted modeling and data-driven control system design. We are developing new machine learning tools for dynamical system identification, and we will use tools from control systems theory (optimal control, model predictive control) to optimize photobioreactor performance. The effectiveness of these methods will be demonstrated through optimization of a process to produce the advanced biofuel isobutanol. We envision that this platform will also be used for myriad processes that produce pharmaceuticals, chemicals, and other products, all from renewable feedstocks.
2. PhD Dissertation: Rapid thermal processing for production of chalcopyrite thin films for solar cells: Design, analysis, and experimental implementation
Advised by Babatunde A. Ogunnaike and Robert W. Birkmire at the University of Delaware
Due to their potential for low cost and high efficiency, thin film solar cells are a strong candidate for utility-scale electricity production. One attractive manufacturing route for thin film solar cells is rapid thermal processing. For my PhD research, I designed a rapid thermal processing (RTP) reactor for producing thin film solar cells and deployed a novel control system using an advanced model-based controller and a nonlinear observer. With the reactor in place, I developed a novel stochastic model to describe thin film deposition in the reacting system. I derived expressions to relate simulation parameters to physical properties, showed that the model could predict film compositionâincluding composition gradients that affect solar cell performanceâas observed in the lab, and used the stochastic nature of the model to characterize lateral heterogeneity in the film. I also showed that the model predictions correspond with experimental data. Finally, I applied statistical design of experiments to probe the effects of process variables on the material properties of the films and optimize the efficiency of solar cells.
2016, University of Delaware, Assistant Instructor
Courses: Engineering Statistics (graduate), Chemical Engineering Laboratory (undergraduate)
2012, University of Delaware, Graduate Teaching Assistant
Course: Process Dynamics and Control (undergraduate)
I plan to treat teaching as a vital component of my position. I have experience as the instructor of computer labs (as a TA) and experimental labs (as an instructor), as well as lecturing experience at the graduate level. Beyond classroom experience, I helped develop an undergraduate laboratory project where students produced biodiesel from soybean oil with bench-scale and pilot-scale reactors.
My teaching philosophy is to to lead the class through experiential learning. After core concepts have been introduced, I believe that most effective way to cement student understanding is through a series of projects, hands-on demonstrations, and âvirtualâ demonstrations. I will use the programming skills I have developed through my research to build educational applications and demonstrations that students can use to reinforce the material.
I am capable of teaching any core chemical engineering course at the undergraduate or graduate level. Given the option, I would choose to teach courses with a substantial mathematical modeling component, such as Process Dynamics and Control, Computational Methods, or Differential Equations for Engineers. I am also interested in developing/teaching elective courses, possibly in fields such as Advanced Process Control and Machine Learning for Chemical Engineers.
2017, AIChE Journal, âHierarchical monitoring of industrial processes for fault detection, fault grade evaluation and fault diagnosis,â L. Luo, R. J. Lovelett, B. A. Ogunnaike. www.dx.doi.org/10.1002/aic.15662
2016, Journal of Process Control, âDesign and experimental implementation of an effective temperature control system for thin film Cu(InGa)Se2 production via rapid thermal processing,â R. J. Lovelett, G. M. Hanket, W. N. Shafarman, R. W. Birkmire, B. A. Ogunnaike. www.dx.doi.org/10.1016/j.jprocont.2016.07.005
2016, AIP Advances, âA stochastic model of solid state thin film deposition: Application to chalcopyrite growth,â R. J. Lovelett, X. Pang, T. M. Roberts, W. N. Shafarman, R. W. Birkmire, B. A. Ogunnaike. www.dx.doi.org/10.1063/1.4948404
2016, University of Delaware Ph.D. Dissertation âRapid thermal processing for production of chalcopyrite thin films for solar cells: Design, analysis, and experimental implementation,â R. J. Lovelett, http://udspace.edu/handle/19716/21450
2016, IEEE Photovoltaic Specialists Conference, âGrowth of Cu(In,Ga)(S,Se)2 films: Unravelling the mysteries by in-situ x-ray imaging,â B. West, M. Stuckelberger, L. Chen, R. J. Lovelett, B. Lai, J. Maser, W. Shafarman, M. Bertoni. www.dx.doi.org/10.1109/PVSC.2016.7749650
2015, IEEE Photovoltaic Specialists Conference, âA stochastic model for Cu(InGa)(SeS)2 absorber growth during selenization/sulfization,â R. J. Lovelett, W. N. Shafarman, R. W. Birkmire, B. A. Ogunnaike. www.dx.doi.org/10.1109/PVSC.2015.7356226