Alison Cozad

Keywords

Optimization, simulation, modeling, machine learning, design of experiments, data analysis, black-box optimization

Education

Carnegie Mellon University. Pittsburgh, PA
PhD candidate, Department of Chemical Engineering
Expected graduation date: May 2014
Advisor: Professor Nick Sahinidis

University of Minnesota. Minneapolis, MN
Bachelors of Chemical Engineering
Graduation date: May 2009
Emphasis: Numerical and Computational Modeling

Software

ALAMO

Automated Learning of Algebraic Models for Optimization

A software suite developed to generate algebraic models of black-box systems (simulations, experiments, etc.). These surrogate models are developed with the following goals:
  • Accurately represented the black-box output variables over the full design space of input variables
  • To identify a simple, low-complexity surrogates of a system with a, previously, unknown functional form
  • Accomplish the above goals using a minimal data set of, potentially, costly black-box function evaluations

For more information, please visit the ALAMO webpage.

Publications and presentations

  1. A. Cozad and N. V. Sahinidis, Derivative-free optimization enhanced-surrogate models for energy systems optimization, Invited talk at the INFORMS Annual Meeting, Charlotte, North Carolina, November, 2011.
  2. A. Cozad, N. V. Sahinidis, and D. C. Miller, Learning surrogate models of processes from experiments or simulations, Talk at the Annual AIChE Meeting, Minneapolis, Minnesota, October, 2011.
  3. A. Cozad, Y. Chang, N. V. Sahinidis, and D. C. Miller, Optimization of carbon capture systems using surrogate models of simulated processes, Talk at the Annual AIChE Meeting, Minneapolis, Minnesota, October, 2011.
  4. D. C. Miller, Y. Chang, A. Cozad, H. Kim, A. Lee, P. Vouzis, N. V. S. N. M. Konda, A. J. Simon, N. Sahinidis, L. Yang, and I. E. Grossmann, Synthesis of optimal adsorptive carbon capture processes, Talk at Annual AIChE Meeting, Minneapolis, Minnesota, October, 2011.
  5. A. Cozad and N. V. Sahinidis, Using derivative-free algorithms to identify surrogate models of energy systems, Invited talk at the SIAM Conference on Computational Science and Engineering(CSE11), Reno, Nevada, March, 2011.
  6. A. Cozad, N. V. Sahinidis, and D. C. Miller, Optimization of power plant simulations with integrated carbon capture systems using black-box algorithms, Talk at the Annual AIChE Meeting, Salt Lake City, Utah, November, 2010.

    Professional Experience

    Manufacturing Engineering Intern. Boston Scientific
    May 2009 – August 2009
    Worked on a yield improvement project by tracing product flaws of catheters for stents throughout the process to locate instrumentation, equipment, and process steps for improvement

    Undergraduate Research. University of Minnesota, Department of Chemical Engineering and Material Science
    September 2008 – May 2009
    Optimized a two-phase process to synthesize hydroxymethylfurfural (HMF) using Aspen HYSYS to model and preform sensitivity analysis.

    Chemical Engineering Intern. Ecolab Inc.
    June 2008 – August 2008
    Redefined, developed, and/or altered several test methods to statistically verify test precision with nonlinear categorical inputs, preform analytical tests, and better represent real-world situations.

    Chemical Engineering Intern. Seagate Technology
    May 2007 – May 2008
    Main accomplishment: Used six sigma methodology to improve the key output variable by 40% by building and validating a transfer function to model the effect of process and vendor inputs on output process metrics in chemical-mechanical hard drive lapping.

    Technical Aide. 3M
    October 2006 – May 2007
    Improved desired characteristics of automotive adhesive including tensile strength, hardness and cure time by running an iterative design of experiments to improve upon the current adhesive recipe.

    Engineering Intern. Donaldson Company, Inc.
    May 2006 – September 2006
    Lead a project designed to restructure and reorganize a product code system to be used on a global level.