Awards
Amos Ben-Zvi received the Best Paper in Conference Award at the 2010 International Conference on Modeling Identification and Control in Okayama, Japan in July 2010 for his paper coauthored by Kirsten Aschbacher entitled: "Identifiability and global estimability tools for dynamical systems".
M.Sc. student Julien Lauzon-Gauthier received 3rd prize at the MITACS/CORS 2010 Annual Conference for his poster entitled “Multivariate Monitoring of a Baked Carbon Anode Manufacturing Process used in Primary Aluminum Smelting”. Julien is currently supported by a FQRNT scholarship from the province of Quebec.
Kim McAuley received the Queen's University Award for Excellence in Graduate Student Supervsion in October 2009. She also received the MITACS Mentorship Award of Excellence at the Second Canada-France Congress in June 2008, after being nominated by her graduate students.
Tom Harris received the D.G. Fisher Award from the Systems and Control Division of the Canadian Society for Chemical Engineering in 2007 for his seminal work on controller performance assessment. His expertise in performance assessment is benefitting graduate students working on MITACS-sponsored research.
New Core Investigators
Three core investigators were added to the project for 2010-2012 so that our group could expand the range of academic research activities and industrial collaboration in our MITACS-funded work. Carl Duchesne and his students at Laval University have been working an aluminum smelting problem sponsored by Alcoa. Amos Ben-Zvi and Jong Min Lee at the University of Alberta worked on modeling and optimization of a variety of chemical and biochemical processes with their students. Recently, Jong Min Lee took up a postion at Seoul National University in Korea, and Amos Ben-Zvi left academia to take up an industrial position at Cenovus in Calgary. We will miss their contributions to our project.
Industrial Partners, Funding and Contributions by Graduate Students
- Abbott Point of Care and Jim McLellan and Kim McAuley at Queen's University signed a research contract in 2011 for a two-year modeling project on single-use sensors for measuring analyte concentrations in blood, valued at $21500. A portion of this money will be used as MPRIME matching funds and some will be used to support MITACS ACCELERATE internships for new Master's student Xu Liang Li.
- Shell Global Solutions provided $17500 in funding for research on parameter estimation and mathematical modeling at Queen's. During the summer of 2011, M.Sc. student Kevin McLean spent three months at Shell developing improved mathematical models for ethylene glycol production using new catalyst systems. In September, Kevin returned to Queen's to defend his Master's thesis. He will start his new job at Hatch in Mississauga in December, 2011. Hatch has also been an industrial sponsor for our research.
- IPL Inc. provided $30000 over two years, starting in 2011, to support the research of Carl Duchesne at Laval University on modeling and control of product quality during the production of composite thermoplastic pipe. Ph.D. student Massoud Barvaz is working on this project.
- ALCOA provided $16000 in funding for the research of Carl Duchesne at Laval University, in support of Master's student Julien Gauthier. Julien collected data at the company and used it to develop a multivariate monitoring system for ALCOA's prebaked anode manufacturing process. Discussions are underway between Prof. Duchesne's group and ALCOA for a larger research project related to aluminum manufacture.
- Novelis awarded $7500 in 2009 to fund research on advanced diagnostics for aluminum rolling processes, and an additional $15000 has been awarded for 2010. Ph.D. student, Hui Yuan, who is cosupervised by investigators Tom Harris and James McLellan, spends part of his time at the Novelis Global Technology Centre in Kingston, collecting data and developing nonlinear models that will be used to assess the performance of high-speed aluminum rolling processes.
- $15000 in funding was awarded by the E.I. du Pont Canada Company to support a research project on mathematical modeling of the company's new Cerenol production process. Cerenol is a new class of polyether polyol products that are made from bio-sourced 1,3- propanediol rather than from petroleum-derived chemicals. A research agreement was finalized during the summer of 2009, and a new Master's student, Jessica Cui started in September 2009. She is cosupervised by Dr. Tuyu Xie from DuPont and Kim McAuley from Queen's.
- $10000 was awarded by BASF to support a Master's student, John Woloszyn, who is developing mathematical models of BASF's expanded polystyrene production processes. John's model, which incorporates the most recent mechanistic information from the polymerization literature, predicts key product quality and production rate variables. Parameter estimation and estimability analysis is underway, using literature data and data supplied by the company. John's research is supervised by Kim McAuley at Queen's, with assistance from Dr. Pascal Hesse at BASF.
- $15000 was awarded by Abbott Point of Care to support a Master's student, Stephen Snyder, who developed dynamic models to predict the stability of two medical cartridges used for point-of-care blood analysis. Stephen, who has spent considerable time at Abbott's technology centre in Ottawa, is the recipient of an NSERC-MITACS IPS. He presented his research at the Abbott Statistics/Quality Engineering Symposium in Chicago in October, 2009. Stephen's research is cosupervised by James McLellan and Kim McAuley at Queen's, with industrial advice from Tamara McCaw and Eric Brouwer at Abbott. Stephen recently received an offer of full-time employment from the company and will begin his work as an Intermediate Research Scientist, starting November 8, 2010. Abbott Scientists are preparing a new proposal to fund additional students through the ACCELERATE internship program.
- $20000 was awarded by DSM in the Netherlands to support Master's student, Angelica Bitton, who is developing a fundamental model to predict diffusion of solvent and residual monomers from rubber particles in the degassing portion of DSM's EPDM production process. This model will enable DSM to optimize their process, so that high product throughputs can be achieved while reducing the concentrations of residual monomers and solvents in their products. Angelica spent the summer of 2008 at DSM, collecting process data and performing experiments aimed at determining key solubility and diffusional parameters for use in her model. She returned DSM in 2010 to work with company scientists to update the values of thermodynamic parameters that are required in the model. Angelica is supervised by Kim McAuley at Queen's and receives advice from Dr. Luigi D'Agnillo at DSM.
- Ph.D. student, Roy Wu, completed an internship at Shell Global Solutions in Houston in May 2008, where he successfully estimated selected parameters in an industrial ethylene-oxide reactor model used for on-line process monitoring. Roy used the estimability analysis tools developed by our research group to determine which parameter values could be updated using the available data, and which should be left at their initial values. In February 2009, Roy Wu and Kim McAuley gave a presentation about these techniques at Shell's Research Centre in Houston. Roy was cosupervised at Queen's by Kim McAuley and Tom Harris. Ethylene oxide facilities at several locations around the world benefitted from the updated model, including a plant in Scotford, near Edmonton. Eugene Theobald at Shell was a key mentor for Roy's research. Eugene is currently preparing a proposal so that Master's student Kevin McLean will also be able to spend time at Shell, using advanced parameter estimation techniques to estimate parameters in chemical reactor models.
- Ph.D. student, Duncan Thompson, extended our estimability analysis methods to account for uncertainties in initial parameter values and measured responses. He used industrial data supplied by INEOS (formerly BP Chemicals) to estimate parameters in a model that predicts molecular weight distributions and copolymer compositions in high-density polyethylene products. Duncan augmented the sensitivity matrix used for estimability analysis and used it to determine sequential V-optimal experiments aimed at improving model predictions. The final year of Duncan's research was supported by MITACS, with matching funds from Hatch and Matrikon. Recently, Duncan's estimability techniques have aided parameter estimation in models of polymer gel dosimeters and in models of biochemical reactors developed by Ph.D. student Jennifer Littlejohn's from Andrew Daugulis's research group at Queen's. Duncan Thompson completed his Ph.D. in May 2009 and is now a control engineer at Imperial Oil in Dartmouth, NS.
- Raluca Apostu, a Ph.D. student from Michael Mackey's group at McGill visited Queen's in October 2009 to obtain advice on using estimablity analysis for estimating parameters in a dynamic disease model. The main problem with the estimability analysis algorithm developed by Duncan Thompson is that it is difficult to determine the appropriate number of parameters to estimate from the ranked list. We are delighted that Roy Wu recently developed a novel model-selection criterion to address this issue in an elegant and computationally efficient way.
- Ph.D. student, Saeed Varziri, developed an Approximate Maximum Likelihood Estimation (AMLE) technique for parameters in fundamental nonlinear differential equation models. Some of the advantages of this technique over traditional parameter identification methods are that it can accommodate: imperfect models and stochastic process disturbances; unmeasured states and nonstationary disturbances; measurements at irregular sampling times; unknown initial condition. The AMLE methodology, which uses basis functions to discretize differential equation models was tested using data from a lab-scale nylon polymerization reactor. Further testing using a more complex model with additional data arising from degradation reactions is required. Saeed Varziri defended his Ph.D. in June 2008 and is a control engineer at Nova Chemicals in Joffre, AB. Saeed was cosupervised by James McLellan and Kim McAuley at Queen's, with advice from Jim Ramsay at McGill. His research was supported by Cybernetica, DuPont, Hatch, Matrikon, SAS and MITACS. Saeed's research resulted in four journal articles and numerous presentations.
- Jim Ramsay and his students developed a novel Parameter Cascading Approach for estimating parameters in differential equations. This new methodology has been applied for estimation of parameters in fundamental models of several different chemical and biological systems. An article describing the new methodology was read before the Royal Statistical Society in 2007. Postdoctoral Fellow Giles Hooker and Ph.D. students David Campbell and Jiguo Cao who worked on this project are now Assistant Professors at Cornell University and at Simon Fraser University. The parameter cascading approach can also be applied to a wide range of situations involving large numbers of nuisance parameters. Cao and Ramsay (2009) applied the methodology to the linear mixed-effects model problem which is an important extension of regression analysis. The approach performed as well as pre-existing methods in all situations tested and, moreover, was markedly superior in high-dimensional situations. The Parameter Cascading Approach has recently been extended to accommodate models with constrained states, with application to a nylon chemical reactor model in a recent article by David Campbell, Giles Hooker and Kim McAuley that was submitted to Technometrics in fall 2009. Furthermore, David Campbell and R. Steele developed a novel Bayesian approach called Smooth Functional Tempering (SFT) that speeds up convergence of estimation compared to other Bayesian methods, works with unobserved states, and overcomes prior information that is inconsistent with the data. In the presence of multiple parameter regions that provide useful inference, SFT locates these multiple regions and determines their relative importance
|
