Author: randallv

Title: Optimizing the Performance of Analytical Chemistry Instrumentation

Presenter: Srividya Sekar

Date: April 6th

Time: 12:00 pm -2:00 pm

Location: WH 425A or virtually on Teams

Teams Live Info: Click Here to Join Live Event

Supervisors: Dr. Victoria Chen, Dr. Jay Rosenberger

Committee members: Dr. Kevin Schug, Dr. Shouyi Wang, Dr. Chen Kan

Abstract: Optimal parameter settings play an important role in the efficiency of the analytical chemistry instrumentation. The research focuses on developing a globally optimal surrogate optimization methodology to obtain parameter settings for a run of the instrument. It uses a smooth Quintic Multi Adaptive Regression Splines (QMARS) as the surrogate model and a Mixed Integer Quadratically Constrained Program (MIQCP) to globally optimize the metamodel. The algorithm will provide an abundance of knowledge about the analytical chemistry instrument, reduce the sample preparation time and the trial-and-error runs needed to achieve the optimal and efficient extraction of the analysis.

Title: Integrating Multiscale Modeling and Machine Learning – Design, Analysis and Manufacturing of Advanced Materials and Structures

Presenter: Dr. Xin Liu

Date: February 22, 2021

Time: 1:15 pm-2:15 pm

Teams Live Info:

Join on your computer or mobile app

Click here to join the meeting

Abstract: The superior performances of advanced materials and structures are mainly achieved through engineering the microstructure at different scales. This seminar will introduce a novel physics-based data-driven multiscale modeling approach to connecting the microstructures to the material properties and structural performances. The first part of this seminar will introduce the basic idea of the multiscale modeling method called mechanics of structure genome (MSG) and its application to textile composite structures. The accuracy and efficiency of the MSG models will be demonstrated by comparing with conventional finite element models. Moreover, the neural network models were trained to further accelerate the multiscale modeling. The second part of the seminar will introduce the on-going research of developing multiscale and multiphysics models to predict the process-structure-property-performance relation. The multiscale modeling was carried out to predict the effective thermal conductivity. In addition, a two-step homogenization approach was developed to enable in-situ monitoring and performance prediction considering the manufacturing-induced geometry imperfections. The developed approach can be used for the in-process decision making for additively manufactured materials with complex geometry shapes (e.g., metamaterials).

Bio: Dr. Xin Liu is an Assistant Professor in the Industrial, Manufacturing, & Systems Engineering Department at UTA. He is also a member in the Institute for Predictive Performance Methodologies at UTA Research Institute. Dr. Liu obtained his PhD in 2020 and Master of Engineering in 2016 from Purdue University in Aeronautical and Astronautical Engineering. His expertise is in data-driven multiscale modeling of composite materials and structures. He has authored/co-authored 20+ journal papers and refereed conference papers. He also developed 3 computer codes for multiscale modeling of composites.  He received the American Society for Composites (ASC) Ph.D. Research Scholarship Award in 2018.

The IMSE seminar is returning the Monday after Thanksgiving. Specifically, our own Ph.D. student Nahal Sakhavand is giving her dissertation defense as a seminar at noon on Monday, November 30 on Microsoft Teams.  Information and a link to Ms. Sakhavand’s presentation are below. 

All students and faculty are encouraged to attend.

Title: New Algorithms for Stochastic Power Systems Planning and Operations Problems 
Presenter: Nahal Sakhavand
Date: Monday, November 30, 2020
Time: noon

Teams Info:

Join on your computer or mobile app

Click here to join the meeting

Or call in (audio only)

+1 817-502-2418,,608465888#   United States, Fort Worth

Phone Conference ID: 608 465 888# 

Abstract: In this paper, we present a two-stage stochastic programming and simulation-based framework for tackling large-scale planning and operational problems that arise in power systems with significant renewable generation. Traditional algorithms such as the L-shaped method used to solve the sample average approximation of the true problem suffer from computational difficulties when the number of scenarios or the size of the subproblem increases. To address this, we summarize a cutting plane method that uses sampling internally within optimization to select only a random subset of subproblems to solve in any iteration.  In addition, we present a demonstration of the design and analysis of computer experiments approach (DACE) to the stochastic unit commitment-economic dispatch problem. We use a multivariate adaptive regression splines algorithm to approximate the second stage of the problem with an endeavor to provide more computationally tractable solutions over the traditional L-shaped method. We conduct the experiments on a modified IEEE118 test system and assess the quality of the solutions obtained from both the DACE and L-shaped method in a replicated procedure. The results obtained from this approach attest to the significant improvement in the computational performance over the L-shaped method. 

Biographical Sketch: Nahal Sakhavand is a Ph.D. candidate in the Industrial, Manufacturing, and Systems Engineering department at UTA. She received her M.S. in Operations Research from SMU and in Industrial Engineering from UTA. She holds a B.S in Applied Mathematics from the National University of Iran and in Industrial Engineering from the University of Tehran North branch. Her research interests are in large-scale stochastic optimization, data analytics, statistical modeling, and their applications. 

IMSE’s Center on Stochastic Modeling, Optimization, & Statistics (COSMOS) has developed a COVID-19 online tool, called COSMOS COVID-19 Linear Programming (CC19LP), to assist county-level decision-makers in planning the reopening of their communities.  The CC19LP tool allows decision-makers to explore the two primary conflicting objectives, namely (1) maintaining a low COVID-19 fatality rate and (2) enabling recovery of the U.S. economy via reopening.  The recent rise in COVID-19 cases due to reopening, despite significant increases in contact tracing efforts, has created urgency in finding alternate approaches to controlling the impact of the pandemic in the U.S.  Other than contact tracing, the major control policy that decision-makers have implemented in the U.S. is the complete lockdown of communities.  While the lockdown policy can successfully lower the fatality rate, it is severely detrimental to the U.S. economy, and subsequently does not achieve balance in the two objectives.  Reopening strategies were specified to alleviate stress on the economy, but no quantitative analysis was employed to confirm that communities were prepared to reopen.  Simply imploring with the U.S. public to be responsible has not worked, and non-compliance continues to be a major issue.

CC19LP employs a key contact partitioning structure that focuses compliance on a smaller percentage of the population that has direct connections with individuals that are at higher risk for severe illness.  COSMOS researchers have created a questionnaire to assist individuals in identifying if they are a key contact.  Finally, COSMOS Director Victoria Chen posted a 15-minute podcast with the Institute for Operations Research and the Management Sciences (INFORMS) that motivates the development of CC19LP.  This podcast, the key contact partitioning questionnaire, and the CC19LP online tool are all accessible from the COSMOS COVID-19 project page: https://cosmos.uta.edu/projects/covid-19/.

Dr. G. Don Taylor is a distinguished alumni here at UTA, having received both M.S and B.S degrees for the IMSE department. Dr. G. Don Taylor is now Vice Provost for Learning Systems Innovation and Effectiveness, and Charles O. Gordon Professor in ISE at Virginia Tech. He is also a Fellow and a Past-President and Member of the Board of Trustees of the Institute of Industrial and Systems Engineers (IISE).

Amongst many accomplishments, Dr. G. Don Taylor has served as Principal Investigator or Co-Principal Investigator on more than 60 externally funded projects. His research has led to the publication of 10 edited books, more than 75 journal articles and book chapters, more than 120 conference papers and technical reports.