(3hu) Innovating Future-Generation Separation Processes through Systems Engineering

Chemical engineering is at an important crossroad with numerous challenges and opportunities for process systems engineering research. By developing advanced mathematical and computational tools, and coupling them with state-of-the-art experimental approaches as well as novel conceptual process design methodologies, my research group will provide industrial practitioners with solutions and insights to address systems engineering issues related to separation processes, which currently account for 40-70% of capital and operating costs as well as most of the carbon footprint associated with a chemical facility.

My research interests consist of three distinct but interconnected components: computational, experimental, and conceptual (see Figure). Computational research utilizes mathematical modeling, process optimization, and artificial intelligence to systematically synthesize optimal process flowsheets and their optimal operating conditions for a given separation task. Experimental research uses leading-edge online process analytical technologies to generate valuable experimental data that can be used for predictive analytics as well as model training and evaluation. In addition, by building and testing pilot-scale process equipment prototypes, experimental research also serves to validate the feasibility and attractiveness of novel conceptual design ideas. Finally, conceptual process design gathers the computational and experimental research findings, generalizes them into a set of guiding principles, and combines these principles with engineering ingenuity and creativity to generate systematic conceptual design methodologies in the context of process intensification, modularization, continuous manufacturing, etc.

Computational Research

Computational research lays the theoretical and computational foundations for both experimental research and conceptual process design research. Mathematical Modeling is fundamental to the success of other computation research thrusts. I am interested in constructing insightful shortcut models to fundamentally understand and characterize different separation processes:

1. Developing structurally simple and numerically stable differential-algebraic equation (DAE) based models for rate-based separation processes (e.g., crystallization, chromatography, pervaporation, membrane based separations).

2. Constructing a unified shortcut model for equilibrium-staged multicomponent separation processes (e.g., distillation, absorption and stripping, extraction) and exploring its mathematical properties that makes it easily implementable in a global optimization framework.

Process Optimization formulates and solves optimization problems related to synthesis and design of multicomponent separation process systems that optimize certain objective, such as annual profit, energy consumption, carbon footprint, etc. These optimization problems will embed the shortcut models developed under the modeling thrust. Even though many of these problem formulations are known to be NP-hard, my group aims at developing efficient and robust optimization algorithms that can solve these problems quickly. A strong emphasis will be put onto ensuring global optimality of the solution. We will derive and implement novel, formulation-tailored solution strategies, including new relaxation and discretization techniques as well as new algorithms, to solve the following problems:

1. Deterministic global (mixed-integer) dynamic optimization with nonlinear ODEs for batch crystallization and multicomponent membrane cascade synthesis and optimization.

2. Superstructure synthesis and optimization of hybrid separation process network consisting of various types of unit operations.

Artificial Intelligence supports and facilitates both mathematical modeling and process optimization thrusts by introducing leading-edge breakthroughs in data analytics and machine learning. Machine learning will be particularly useful for solving problems that are not amenable to precise mathematical description. I will develop hybrid AI platforms that combine first-principle models and leading-edge data-driven machine learning techniques. I am keen to solve the following challenges in separation:

1. Optimal feedback control of batch crystallizer by implementing state-of-the-art reinforcement learning algorithms (e.g., Q-learning) and online process analytical technologies.

2. Predicting azeotropic behavior of complex multicomponent systems via supervised learning.

Experimental Research

Experimental Research guides and facilitates both computational and conceptual design components. It generates valuable first-hand experimental data that can be utilized by mathematical modeling thrust for assessing model accuracy, by process optimization thrust for data-driven optimization, and by artificial intelligence thrust for predictive analytics, model training, and feedback control. Meanwhile, by building and testing prototypes and performing scale-down analysis, experimental research also serves to evaluate the feasibility and attractiveness of novel conceptual design ideas. Inspired by this, I plan to study the following problems:

1. Kinetic parameter estimation for batch crystallization using in situ process analytical technologies (e.g., FBRM and ATR-FTIR) coupled with EasyMax^TM/OptiMax^TM chemical synthesis systems.

2. Exploring retrofit opportunities to facilitate the transition from batch to continuous manufacturing by experimentally validating innovative conceptual design ideas and building pilot-scale prototypes (e.g., fully operable dividing wall column, distillation-adsorption column).

Conceptual Process Design

Finally, Conceptual Process Design takes all the research findings previously discovered and combines them with engineering ingenuity and creativity to come up with new, intensified, energy-efficient, and cost-effective process flowsheet concepts. Recent trends in process intensification, modularization, and continuous manufacturing are also considered when formulating and refining these conceptual design methodologies. These design methodologies and guidelines should be easily convertible into a set of mathematical rules and constraints to enable systematic synthesis of new and attractive flowsheet options with the click of a button. Meanwhile, these methodologies and high-level concepts also generate new design ideas, which further lead to future computational and experimental research directions. My group will develop a software platform that systematically synthesizes revolutionary intensified separation process flowsheets in an automated way. Combining the valuable knowledge and insights generated from previous research efforts will equip me with the right methodology and tools to tackle this exciting challenge.

Graduate Research Highlights

My graduate research has been focused on systems engineering related to multicomponent distillation. Distillation is the most important separation process that accounts for 90-95% of all separations in the chemical industries. Even slight improvements can tremendously impact the landscape of the chemical economy world. The goal of my graduate research was to develop mathematical modeling and global optimization approaches as well as systematic process intensification strategies to synthesize compact, easy-to-operate, energy-efficient, and cost-effective distillation systems. Towards this goal, I explored several aspects of multicomponent distillation:

1. I independently solved a longstanding challenge in chemical engineering of developing a shortcut method to determine the minimum reflux condition for any multi-feed, multi-product distillation column separating ideal multicomponent mixtures. The classic Underwood's method turns out to be a special case of this approach.

2. I developed the first enumeration based global optimization algorithm to identify optimal distillation configurations that can save up to 50% of total cost or total exergy loss compared to conventional schemes from the immense configuration search space. For the first time in literature, global optimality is guaranteed.

3. I proposed a systematic and comprehensive multi-layer approach for process intensification in multicomponent distillation. For the first time, industrial practitioners now have an easy-to-follow recipe to generate an array of completely new and attractive highly intensified configuration designs that further enhance operability, improve efficiency, and reduce total costs.

Teaching Interests

I am interested in teaching undergraduate and graduate-level courses on separation processes, process and product design, numerical methods, and optimization theory. My professional experience in agrochemical business also equips me to teach courses on industrial process development and process safety. Apart from teaching existing courses, I look forward to designing a new course at the interface between process design and process optimization by introducing some of the modern optimization algorithms and tools that are widely used today in the context of real-world applications, such as crude oil blending, pharmaceutical process scheduling, NGL supply chain optimization, etc. In addition, I am enthusiastic about collaborating with colleagues in experimental areas to develop an interdisciplinary course that explores the use of design of experiments, data analytics, and machine learning across multiple disciplines, such as advanced functional materials synthesis and design, bioinformatics and drug discovery, etc. Lastly, I would like to offer a graduate-level course in linear operator theory that discusses some of the advanced topics in linear algebra and functional analysis, both of which establish the theoretical foundation for modern mathematical programming and machine learning.

Selected Publications

Jiang Z, Agrawal R. Process intensification in multicomponent distillation: A review of recent advancements. Chemical Engineering Research and Design. 2019;147:122–145. (Invited review article in the special issue featuring 2018 International Conference on Distillation & Absorption)

Jiang Z, Chen Z, Huff J, Shenvi A, Tawarmalani M, Agrawal R. Global minimization of total exergy loss of multicomponent distillation configurations. AIChE Journal. 2019;65(11):e16737

Jiang Z, Mathew TJ, Huff J, Nallasivam U, Tawarmalani M, Agrawal R. Global optimization of multicomponent distillation configurations: Global minimization of total cost for multicomponent mixture separations. Computers & Chemical Engineering. 2019;126:249–262

Jiang Z, Madenoor Ramapriya G, Tawarmalani M, Agrawal R. Minimum energy of multicomponent distillation systems using minimum additional heat and mass integration sections. AIChE Journal. 2018;64(9):3410–3418

Selected Working Papers

Jiang Z, Tawarmalani M, Agrawal R. An accurate minimum reflux calculation method for multi-feed, multi-product distillation columns distilling ideal multicomponent mixtures: Mathematical model, optimization algorithm, and case studies. Under review at Chemical Engineering Science.

Jiang Z. An accurate shortcut model for designing multicomponent azeotropic distillation systems.

Jiang Z. Optimal control of batch crystallization by reinforcement learning.

Selected Honors and Awards

People’s Choice – Mission Possible Award, Corteva Agriscience, 2019

AIChE Separations Division Graduate Student Research Award, 2018

College of Science and Engineering Merit Scholarship, University of Minnesota, 2013

Charles A. Mann Award, Department of Chemical Engineering and Material Science, University of Minnesota, 2012