National Science Foundation     |     Directorate for Engineering  (ENG)
Division of Chemical, Bioengineering, Environmental, & Transport Systems  (CBET)
 
CBET Award Achievements (Nuggets)
Notable Accomplishments from CBET Awards
 
 
Transition from Edisonian to a Computer-Aided Approach: Using Mathematical Models to Optimize Catalyst and Reactor Design
 
Dionisios Vlachos  –  University of Delaware

Background:  Catalysts and chemical reactors form the backbone of chemical, petrochemical, pharmaceutical, environmental, and energy sectors.  Development of mathematical models that can lead to optimized, rational discovery of novel catalytic materials and efficient design of reactors can have an unprecedented impact on numerous industries, the US economy, the environment and the society.  This research team transformed design practices from the current trial-and-error (Edisonian) to a scientifically-founded strategy (computer-aided).  This research also set the foundation for multiscale model-based product engineering.

Results:  The research team obtained the following results:
 
      (1)  Development of novel high throughput models for design of experiments  - -  The conceptual framework developed uses a fundamental, multiscale model to predict optimal experimental conditions for model assessment and refinement.  One starts with a detailed chemical kinetic model developed via a hierarchical multiscale modeling approach.  The computational methodology involves generation of a large library (of hundred of thousands) of operating condition sets via a Monte Carlo search method, followed by a reactor calculation and a sensitivity analysis at each point.  An informatics approach is carried out to identify patterns within this library.  These patterns are then tested for statistical validity, and experiments are conducted using these patterns.  The model is refined (if necessary) with respect to its parameters and/or its structure.  The entire procedure is repeated until the deviation between model and experiments in the entire experimental parameter space is low.  It has been found that identification of optimal regions (new approach) for experiment design is superior to picking a single optimal point (traditional approach).  The outcome is globally validated models that can be used for reactor and catalyst design.
 
      (2)  Transient data collection and model analysis  - -  A computational singular perturbation framework was developed that maps experimentally measured quantities with the related time scales of a complex reaction network.  This novel approach connects for the first time responses seen experimentally at various time scales to intrinsic rate constants of the network and provides transient models suitable for reactor dynamics.
 
      (3)  Fusing data and models for the preferential oxidation of carbon monoxide  - -  The above frameworks were applied to the preferential oxidation (PROX) of carbon monoxide (CO) for the purification of hydrogen for fuel cells.  In the PROX reaction, aside from the catalyst activity toward CO oxidation, the selectivity toward burning hydrogen is a major concern.  A rich set of experimental data was designed for the PROX chemistry using multiscale models.  This data was employed to refine the detailed chemistry model (Figure below).  The model in turn was used to search for better catalytic materials.
 
      (4)  Development of high throughput models for design of novel catalytic materials  - -  Using validated models, an optimization problem was formulated for the first time for identifying best catalytic material properties using a high throughput search.  The approach was tested for the PROX reaction.  It appears that conversion and selectivity may be optimized with different materials properties (Figure below).  This finding points to considerable new material designs.

Dion Vlachos 1
 
Figure.  Elements of the overall approach of this NSF-funded research:
(aQuantum mechanical simulations to refine kinetic parameters.
(bReactor model.
(cReactions from a detailed reaction mechanism for the PROX reaction.
(dPrediction of optimal properties of catalytic materials for the PROX reaction using high throughput modeling.
(eHigh resolution micrograph of the catalyst.
(fPicture of a microreactor for conducting experiments.
(gComputational fluid dynamics (CFD) simulation of microreactors to account for transport-chemistry interactions.
 
Credit for Images 1 & 2:  Dion Vlachos, University of Delaware
 
This project addresses the NSF Strategic Outcome Goals, as described in the NSF Strategic Plan 2006-2011, as follows:
 
Primary Strategic Outcome Goal:      (1) Discovery:  NSF emphasizes interdisciplinary research and education to address complex and emergent problems.  This proposal develops new tools in catalysis and reaction engineering, which encompass methods from the systems community (optimization), catalysis and reaction engineering, bioinformatics, and multiscale mathematics to solve complex problems in catalytic reactors.  The developed multiscale-based methodology provides a rational approach to design experiments and products in multiple disciplines, including among others chemical processing, biotechnology, and materials.
                                                                   (1) Discovery Category:
                                                                          - Engineering Research

 
Secondary Strategic Outcome Goal:  (2) Learning:  The research team has:

*  Trained one graduate student on a spectrum of topics, including multiscale chemical kinetic model development and use, systems approaches for design of experiments, informatics approaches to data mining, catalyst synthesis, microreactor fabrication, kinetics experimentation, and integration of data with models.

*  Trained one postdoctoral fellow in this research and in teaching and prepared him for a faculty position in this area (currently an Assistant Professor at the University of Alberta).

*  Trained two undergraduate students in research in this area (one of them will attend Caltech for graduate studies and the other is applying to graduate schools).

*  Introduced tools to NSF REU (Research Experience for Undergraduates) students.

*  Trained an exchange graduate student from Milan, Italy, in this area and mentored him for a faculty position (currently a Humboldt fellow who will start as a faculty member at the University of Milan, Italy).

*  Developed web-based software for thermodynamic consistency of chemical reaction mechanisms and detailed reaction models.

*  Wrote review articles on microreactors and on the role of catalysis and reaction engineering in solving the energy problem.

*  Gave lectures in outside the discipline meetings, such as a workshop organized by the applied mathematics community (Workshop on ‘Mathematical and Computational Methods for Accelerated Molecular and Stochastic Simulations’, Institute for Computational and Applied Mathematics (IACM), FORTH, Greece), the SIAM meeting in Montreal, a Stochastic modeling meeting in Stockholm and a Marie Curie training program in Patras, Greece with nearly 100 students and postdocs from around the globe.
 
                                                                   (2) Learning Categories:
                                                                          - Undergraduate Education and Undergraduate Student Research
                                                                          - Graduate Education and Graduate Student Research
                                                                          - Postdoctoral Education, including International Postdoctoral Fellowships
                                                                          - International Research Experiences for Undergrad & Graduate Students
                                                                          - Broadening Participation to Improve Workforce Development
                                                                          - Professional and Career Development

 
Secondary Strategic Outcome Goal:  (3) Research Infrastructure:  This proposal develops new tools in computational science and engineering to drive rational, model-driven discovery.  It provides for the first time a framework to enable rational design of experiments and materials when there is an inherent coupling of phenomena across length and time scales.  The developed methodology designs a handful of diverse experiments to maximize the information content extracted from them and provides a platform for the discovery of new materials.
 
                                                                   (3) Research Infrastructure Categories:
                                                                          - Networking and Computational Resources Infrastructure and Services
                                                                          - Research Resources (minor facilities, infrastructure and instrumentation,
                                                                                field stations, museum collections, etc.)

In terms of Intellectual Merit, this work is notable.  This research enables model-based design of experiments with the objective of maximizing the information content and the fidelity of reaction and reactor models in the entire experimental parameter space.  In turn, this framework enables multiscale model-based optimal reactor and catalyst design.  The proposed framework combines experimentation and first-principles computational methods to deliver mechanistic insights and a rational approach toward rational catalyst and reactor design.

In terms of Broader Impacts, this work is notable because the methods developed in this research have a far-reaching impact to the entire field of chemical engineering in:
(1) using first-principles models,
(2) designing experiments in order to validate and/or improve models, and
(3) systematically practice product design.
The outcome of the developed framework can impact the entire chemical, petrochemical, pharmaceutical, environmental, and energy sectors.  The dissemination of multiscale methods to the mathematics and materials disciplines and the multidisciplinary educational activities (e.g., teaching by the PI in workshops of students with a very diverse background) enable rapid cross-fertilization of concepts and educational material.

This research is Transformative.  A multiscale model-based paradigm has been developed that can transform the way future experiments are conducted for extraction of kinetic information and can radically alter future catalyst design practices from the current trial-and-error to a scientifically-founded strategy.  This research sets the foundations for multiscale model-based product engineering.

This research represents Broadening Participation.  Professor Vlachos actively recruits students from underrepresented groups using the college infrastructure as well as personal contacts with faculty at schools with a large fraction of students from underrepresented groups.  In addition, he often engages undergraduate students from the state of Delaware, which is an EPSCoR state.

Existing or potential Societal Benefits of this research:  Catalysis and reaction engineering constitute the heart of chemical and petrochemical industry.  Development of more active and selective catalysts can have a profound effect on these industries and thus on the US economy, and can result in substantial reduction of waste with benefits for the environment, energy savings, and the society.  As a result, this research addresses the needs of the Energy Independence and Security Act of 2007.


 
Program Director:
 
 
 
Maria Burka
CBET Program Director - Process and Reaction Engineering
     
NSF Award Number:   0651043
     
Award Title:
 
  Hierarchical multiscale model-based process engineering
     
PI Name:   Dionisios Vlachos
     
Institution Name:   University of Delaware
     
Program Element Code:   1403
     
NSF Investments:
 
  American Competitiveness Initiative (ACI)
     
CBET Nugget:

  FY 2009


Top of Page


This Nugget was Updated on 30 November 2009.