James B. Rawlings Research Group

Paul A. Larsen B.S. Ch.E., Brigham Young University, 2002 Ph.D., University of Wisconsin-Madison, 2007 palarsen[AT]bevo[DOT]che[DOT]wisc[DOT]edu

Current Work

Crystallization is a key process for the pharmaceutical and chemical industries. However, the ability to model, monitor, and control this process is severely limited by the inability to reliably measure important process variables, such as crystal size and shape. The purpose of this research is to demonstrate that these variables can be measured using visual data. Data extracted from images will be used not only as a reliable sensor for process monitoring and control, but also to enhance understanding of the nucleation and kinetics of crystallization, thus enabling construction of better models.

Specifically, the questions this research will answer are the following:

• How will the three-dimensional characteristics of a crystal be inferred from two-dimensional images?
• In what way will the massive amount of data provided by an image be sorted and analyzed?
• How will this data be used in a feedback control loop?
• How will this data be used in creating better models for monitoring and control?

Successful monitoring and control using visual data has been demonstrated using the sodium chlorate (NaClO₃) system. Impurity-free solutions of NaClO₃ result in cubic NaClO₃ crystals. However, solutions of NaClO₃ having at least 50 ppm sodium dithionate result in tetrahedral NaClO₃ crystals. Using visual data (such as the images shown below), a controller was able to determine the concentration of sodium dithionate that was required to maintain the percentage of cubic crystals (as opposed to tetrahedral) below 40%, as demonstrated in the figure below.

Examples of visual data used to control crystal shape. The image on the left shows tetrahedral crystals resulting from an injection of impurities while the image on the right shows cubic crystals resulting from a decrease in impurity level.

Feedback control of particle shape using visual data.

Publications

1

Paul A. Larsen.
Computer vision and statistical estimation tools for in situ, imaging-based monitoring of particulate populations.
PhD thesis, University of Wisconsin-Madison, July 2007.

2

Paul A. Larsen and Jambes B. Rawlings.
High-resolution imaging-based PSD measurement for industrial crystallization.
Submitted for publication in AIChE J., June 2007.

3

Paul A. Larsen and James B. Rawlings.
Derivations for maximum likelihood estimation of particle size distribution using in situ video imaging.
Technical Report 2007-01, TWMCC, Available at http://www.che.utexas.edu/twmcc/, March 2007.

4

Paul A. Larsen and James B. Rawlings.
Maximum likelihood estimation of particle size distribution for high-aspect-ratio particles using in situ video imaging.
Submitted for publication in Technometrics, April 2007.

5

Paul A. Larsen, James B. Rawlings, and Nicola J. Ferrier.
Model-based object recognition to measure crystal size and shape distributions from in situ video images.
Chem. Eng. Sci., 62:1430-1441, 2007.

6

Paul A. Larsen and James B. Rawlings.
Assessing the reliability of particle size distribution measurements obtained by image analysis.
Submitted to Particle and Particle Systems Characterization, June 2007.

7

Paul A. Larsen and James B. Rawlings.
Assessing the reliability of crystal size distribution measurements obtained by in situ video microscopy and image analysis.
In AIChE Annual Meeting, San Francisco, California, November 2006.

8

Paul A. Larsen and James B. Rawlings.
Crystal size and shape control using in situ microscopy and model-based object recognition.
Association for Crystallization Technology, 14th Larson Workshop, Princeton, New Jersey, October 8-11, 2006.

9

Paul A. Larsen, James B. Rawlings, and Nicola J. Ferrier.
Crystal size and shape monitoring using high-speed, in situ video imaging and model-based recognition.
In Proceedings of the fifth World Congress on Particle Technology, Orlando, Florida, April 23-28 2006.

10

Paul A. Larsen, James B. Rawlings, and Nicola J. Ferrier.
An algorithm for analyzing noisy, in situ images of high-aspect-aspect ratio crystals to monitor particle size distribution.
Chem. Eng. Sci., 61(16):5236-5248, 2006.

11

Paul A. Larsen, Daniel B. Patience, and James B. Rawlings.
Industrial crystallization process control.
IEEE Ctl. Sys. Mag., 26(4):70-80, August 2006.

Personal Web Page: jbrwww.che.wisc.edu/~palarsen