8th Annual Amundson Lecture: Nonlinear Mixed-Effects Models for Parameter Estimation
  • 1:25 p.m. April 15, 2021
  • Virtual
  • Dr. Daniel Hickman
  • Senior R&D Fellow
  • Dow

Dr. Daniel A. Hickman, Senior R&D Fellow in the Engineering and Process Science department of Dow’s Core Research & Development organization, will deliver this seminar as the 8th annual Amundson Lecture in CEMS.

When asked about his standards for hiring during an interview in 1995, Neal Amundson said that he sought candidates who did “not use the alibi ‘this is good enough for engineering.’ The general point was it had to be right.” Amundson transformed the field of chemical engineering, especially chemical reaction engineering, by the application of mathematics, initiating what some have called “the second revolution in chemical engineering.” This lecture, in the context of Amundson’s legacy in the field of reaction engineering, will argue that we have not completed this revolution; in at least one category of problems, we have failed to properly apply mathematics.

Reaction kinetic data are frequently obtained from batch experiments, where quantities such as species concentrations are measured at various times during the batch, resulting in time series data. Investigators then fit kinetic parameters to multiple sets of these time series data, often with the use of nonlinear least squares procedures. However, the application of standard least squares methods implicitly assumes that each individual measurement is statistically independent. Using experimental data obtained with a batch recycle reactor, this lecture will demonstrate that assumptions of independent, normally distributed errors for longitudinal measurements from multiple batch experiments yield estimates that provide statistically inferior explanations of the observed data relative to models that incorporate batch-to-batch variation. To circumvent analogous difficulties outside of the fields of chemical reaction engineering and chemical kinetics, numerous investigators have applied nonlinear mixed-effects models. Mixed-effects models for batch reactor longitudinal experiments offer a better explanation of the observed data by reducing or eliminating bias in the estimated parameters and providing confidence intervals that are more realistic. To echo Amundson, “This is the way you do the problem right.”

Seminars are open to alumni, friends of the Department, and the general public.

Contact Information

Department of Chemical Engineering and Materials Science

421 Washington Ave. SE, Minneapolis, MN 55455-0132

P: 612-625-1313 | F: 612-626-7246

Contact Us

Connect on Social Media

© Regents of the University of Minnesota. All rights reserved. The University of Minnesota is an equal opportunity educator and employer. Privacy Statement