(199h) Mechanism Identification of Population Balance Models With Limited Data

The formation of metal spherical particles by reaction precipitation is a very fast process (of the order of seconds), thus hindering the capability of sampling to generate kinetic data. For such systems UV-vis optical kinetic is collected using a stopped flow reactor for fast kinetics [1]. The dynamic optical response of these systems is characterized by broad bands, making their analysis difficult to interpret in terms of particles sizes. To overcome this problem, the data is analyzed in terms a new strategy called simulated dynamic optical response (SDOR) [2]. In this work the capability of combining this strategy with global optimization to identify mechanism and parameters is studied systematically. The identified model is compared with a more ideal case of having a complete particle size distribution as a function of time for the same system, but in this case, it is assumed that the data is only available in limited time intervals. In theory this data can be collected by a sequence of freeze quench experiments.  The problem of limited/noisy data is studied using Bayes analysis developed for these systems including its synergy with the global optimization approach.

[1] Irizarry R., Burwell L., and M. S. Leon-Velazquez (2011) Preparation and formation mechanism of silver particles with spherical open structures, Ind. Eng. Chem. Res.,50, 8023–8033

[2] Irizarry R. (2010) Simulated dynamic optical response strategy for model identification of metal colloid synthesis, Ind. Eng. Chem. Res., 49, 5588–5602.