Hydrogenation of avermectins catalyzed by RhCl(Ph3P)3. Choosing a model for predicting production plant performance

CRISTALDI M., CABRERA M., MARTÍNEZ E., GRAU R.

Mar del Plata - Buenos Aires - Argentina

Congreso; VI CONGRESO ARGENTINO DE INGENIERIA QUIMICA - CAIQ2010; 2010

Asociación Argentina de Ingenieros Químicos

Trabajo aceptado The production of Ivermectin (Iv) is a leading case of regiospecific homogeneous catalysis found in the fine-chemical industry. Commercial Iv is a mixture containing more than 80% Ivermectin B1a (Iva) and less than 20% Ivermectin B1b (Ivb), which is widely used in veterinary medicine and against particular parasitic human infestations. The synthesis of Iv requires the regiospecific hydrogenation of the cis carbon-carbon double bond at 22(23) position of homologous Avermectins B1a (Ava) and B1b (Avb), without affecting the other four carbon-carbon double bonds at the 3(4), 8(9), 10(11) and 14(15) positions of the cyclic lactone moiety. In current practice, the hydrogenation process is catalyzed by Rh or Ir precursors modified with tertiary phosphines, the most usual catalyst is chlorotris(triphenylphosphine)rhodium(I) (Wilkinson’s catalyst). Since Iv became a generic drug in the middle 1990s, an increasing competition has depressed its price for years, while the Rh price continued to climb. In this scenario, batch process design and optimization for operational profitability improvement has gained increasing importance for Iv industrial production. Complex reaction networks could be invoked to describe this reacting system; however, very refined descriptions would involve too many adjustable parameters for performing unambiguous estimates within a framework of overall kinetic measurements. Consequently, in this work modeling purpose is not to build extremely complex models to fully explain the underlying reaction mechanism. The modeling goal here is instead  to construct the simplest model capable of providing a reliable prediction of the kinetic behavior once the constrained industrial scenario has been defined. From such particular viewpoint, the focus is on determining the most important reaction pathways to reduce both model complexity and uncertainty sources until the simplest kinetic model with the best parametrization is found. Sensitivity analysis (SA) is a tool easy to use and integrate with the classical modeling procedure. More specifically, the variance-based global SA (GSA) is a reliable method for quantifying the relative importance of all parameters and their interactions on the model output when parameter values are changed simultaneously into the parametric uncertainty domain. Thus, model parameters can be ranked based on their relative relevante for explaining model estimation variance after GSA is applied. Afterwards, those parameters identified as less important can be fixed to any value within the upper and lower values of their confidence bounds or even eliminated from model resulting in a significant reduction of model uncertainty sources.