Optimization of parameters for models based on the concept of group contribution applied to the calculation of viscosity of non-ideal mixtures

Author(s)

Camila Nardi Pinto

• Document type: Doctoral Thesis
• Press: Pirassununga.
• Institution: Universidade de São Paulo (USP). Faculdade de Zootecnica e Engenharia de Alimentos (FZE/BT)
• Defense date: 2019-08-12

Examining board members

Cintia Bernardo Gonçalves; Gustavo Voltani von Atzingen; Rodrigo Corrêa Basso; Alessandra Lopes de Oliveira; Jose Antonio Rabi; Rogers Ribeiro

• Advisor: Cintia Bernardo Gonçalves

Abstract

The predictive models of viscosity UNIFAC-VISCO and GC-UNIMOD are based on the theory of group contribution, i.e., they depend on determining interaction parameters to enable its application. The objective of this work was to create a groundbreaking database (with several organic functions) for a new optimization of interaction parameters for aforementioned models. Database and models were implemented in SQL and Python, respectively. Group interaction parameters obtained for UNIFAC-VISCO showed average relative deviations (ARD) lower than 4.27 and 6.88 % when used to predict essential oils mixtures and real ester mixtures, respectively. However, to predict vegetable oil model systems, the parameters had ARD of 25.33 %. For GC-UNIMOD best fitting parameters showed ARDs lower than 3.41 and 4.64 % when predicting essential oils mixtures and real ester mixtures, respectively. UNIFAC-VISCO and GC-UNIMOD models can be satisfactorily used to predict viscosity of mixtures of essential oils, alcohols, esters and vegetable oils, with overall parameters of GC-UNIMOD model being the most versatile as well as with good predictive capacity. An ANN (artificial neural network) was created as an alternative for viscosity prediction of fatty systems. The network presented MAPE (Mean Absolute Percentage Error) of 11.99 % for test data, proving to be an excellent alternative for predicting viscosities. A graphical interface, available on internet, was developed to provide access to the scientific community to tools made and also enable predictions with the group interaction parameters obtained for the models studied. (AU)

• FAPESP's process: 15/16781-7 - Optimization of parameters for models based in group contribution concept applied to calculation of viscosity for non-ideal mixtures
• Grantee: Camila Nardi Pinto
• Support Opportunities: Scholarships in Brazil - Doctorate