Date & time
Wednesday, May 8, 2024
3 p.m. – 4 p.m.
3 p.m. – 4 p.m.
Workshops & seminars
Share this event:
New Optimization and Reinforcement Learning Strategies for Optimal Process Flowsheet Design and Operations Management in Chemical Engineering and Manufacturing Plants
Speaker(s)
Dr. Luis Ricardez-Sandoval
Cost
This event is free
Organization
Contact
Where
John Molson Building
1450 Guy
Room s1.235
Wheel chair accessible
Yes
The aim of this talk is to present the recent efforts made to advance key tasks in Process Systems Engineering (PSE) using new optimization and machine learning methods developed in our group.
In the first part of the talk, we will introduce a new optimization method that can efficiently solve problems involving naturally ordered discrete decisions: Discrete-Steepest Descend Algorithm (DSDA). This method has been used to solve a wide variety of problems that emerge in Chemical Engineering, e.g., optimal design of catalytic columns, simultaneous scheduling and dynamic optimization problems, etc. In this talk, we will present a novel nested hybrid framework that can be used to assess the optimal design of complex process flowsheets using some of capabilities available in Chemical Engineering Modelling Software combined with the DSDA.
Next, we will present a novel reinforcement learning (RL) strategy to perform optimal process flowsheet for chemical systems. The innovative feature in our methodology is that the attractive chemical process flowsheets can be found under the assumption that no a priori knowledge of the process structure/flowsheet may be available to the user. A case study that links our framework with Aspen Plus will be presented to demonstrate the potential of our RL strategy as a promising tool to find new chemical process flowsheets.
In the last part of the talk, we will present a RL strategy that can be used to schedule flow-shop and job-shop batch plants under time-dependent uncertainty. Case studies featuring small and large-scale plants will be presented to demonstrate the potential of the proposed RL strategy to address process scheduling in large-scale chemical engineering manufacturing plants.
Dr. Luis Ricardez-Sandoval
About the speaker
Dr. Louis Ricardez-Sandoval is an Associate Professor in the Department of Chemical Engineering at the University of Waterloo (UW). Dr. Ricardez-Sandoval holds a Canada Research Chair (Tier II) in Multiscale Modelling and Process Systems and leads the development of methods for optimal design and operations management under uncertainty, the development of novel CO2 capture and conversion technologies aimed at reducing the carbon footprint, and computer-aided design of novel catalyst materials.
Dr. Ricardez-Sandoval has published more than 200 journal articles, 50 full-length peer-reviewed conference papers, 3 book chapters and 1 book. Dr. Ricardez-Sandoval (h-index: 48) has 10 publications that each have been cited more than 100 times and has published numerous publications on optimal process integration, modelling and optimization of conventional and emerging CO2 capture technologies, atomistic and molecular design of novel catalyst materials for CO2 conversion, chemical looping combustion (CLC) technologies, and the implementation of machine learning (ML) methods for the optimal design and manufacture of nano-scale and macro-scale systems and materials.
His novel contributions in optimal process integration, multiscale modelling, process systems and CO2 capture and conversion technologies have been widely recognized by delivering multiple plenary and keynote talks at international conferences, leading the organization of top-tier conferences (e.g. International Program Chair: 2022 DYCOPS-CAB IFAC Symposium) and receiving multiple research-related awards, e.g. NSERC Discovery Accelerator Supplement (2017) and Ontario’s Early Researchers Award (2015). Dr. Ricardez-Sandoval is Associate Editor for Digital Chemical Engineering and the Canadian Journal of Chemical Engineering.