(238a) Property Prediction for Lipids Based Product Design and Analysis | AIChE

(238a) Property Prediction for Lipids Based Product Design and Analysis


Diaz-Tovar, C. A. - Presenter, Technical University of Denmark
Gani, R. - Presenter, Department of Chemical and Biochemical Engineering, Technical University of Denmark (DTU)
Sarup, B. - Presenter, Alfa Laval Copenhagen A/S

Vegetable oils and fats have an important role in human nutrition and in the chemical industry since they are a source of energy and fat-soluble vitamins, and also in the production of renewable sources of energy. Nowadays as the consumer preferences for natural products and healthier food increase along with growing interest in biofuels, the oleochemical industry faces in the upcoming years major challenges in terms of design and development of better products and more sustainable processes to make them. Fortunately, the unique condition of edible oil and fats of being mixtures of different chemicals provides a wide range of industrial uses and applications of these lipid-related compounds that, once they are purified, can be offered in the market as final products (edible oils and fats) or as raw materials for other processes ( i.e. sterols as precursors of steroids). The aim of this work is to present the systematic development of computer-aided methods and tools related to the prediction of necessary physical properties suitable for lipid-based product design. The methods and tools include: a filled lipid-database of collected experimental data from the open literature and confidential data from industry, and, generated data from validated predictive property models; modeling tools for fast adoption-analysis of property prediction models and for fast development of process models relevant in product design and that are not available in process simulators. This was achieved by first identifying and classifying the lipid compounds found in the edible oils and their potential industrial uses and applications. Then creating a list of the physical-chemical properties needed for model-based product design. Next, collecting the available experimental data from different sources were collected for the identified lipid compounds and their corresponding properties. Finally, selecting and adopting the appropriate models to predict the necessary properties, to fill-out the lipid-database and to make it suitable for application with other computer-aided tools (such as commercial process simulators). For pure component property prediction, the GC-based method of Marrero-Gani and its extension to atom-connectivity methods were used to predict the single value properties like the normal melting point temperatures, the normal heats of formation as well as the critical properties. For temperature dependent pure component properties (vapor pressure, liquid heat capacity, viscosity and surface tension), published GC-based models have been adopted to obtain a general temperature dependant GC-based model. For liquid density, the modified Racket equation has been adopted to create a GC-based model fine-tuned for the lipid chemicals. The regression of the adopted GC-based model parameters needed experimental data. PC-SAFT EoS was used to generate pseudo-experimental data when real data was not available. With the development of the lipid-database and the associated property models, it has been possible to analyze the behavior of various lipids based products and design the conditions under which they can be recovered. The presentation will also show results of the recovery and analysis of lipids-based products illustrating the use of the developed property models and the lipids-database.