fr

Post-doctoral position : from chemical reaction engineering to machine learning – A new methodology...

Post-doctoral position : from chemical reaction engineering to machine learning – A new methodology...

France 04 oct. 2022
IFP Energies Nouvelles

IFP Energies Nouvelles

Université étatique, Parcourir ses opportunités similaires

DÉTAILS OPPORTUNITÉ

Université étatique
Région
Pays hôte
Date limite
04 oct. 2022
Niveau d'études
Type d'opportunité
Spécialités
Financement d'opportunité
Financement complet
Pays éligibles
Cette opportunité est destiné à tous les pays
Région éligible
Toutes les régions

Post-doctoral position in the research division of process design and modelling.

In order to ensure a reliable extrapolation of a chemical process, for instance to produce renewable fuels from biomass feedstocks, the reactor modeling is an essential step. However, a model is generally developed only after a new catalyst has been validated and the performance of a future catalyst cannot be predicted.

The objective of this post-doctoral position is to integrate catalyst descriptors into reactor performance predictive models. This way, the development phase of reactor modelling could be accelerated integrating a priori information inherent to the catalyst. Also, the development of new catalysts could by driven rationalizing the impact of physico-chemical properties of the catalysts on their performances at industrial reactor scale. The proposed field of application is the family of bi-functional catalysts, which will concern in the near future the processes of fuel production from the conversion of biomass or recycled waste and plastics.

The performance of chemical processes is mainly determined by the operational conditions (temperature, pressure, LHSV, load) which can obscure the impact of catalyst properties in a simple data analysis. The originality of this work is to use kinetic models to overcome this problem. By developing a bank of kinetic models for each catalyst formulation, the reactivity of the catalysts can be evaluated independently of the operational conditions. In this way, the machine learning approach can be fruitfully applied, having a raw data set to analyze the relationships between physical properties and catalyst performance, independent of variation in feedstock and process conditions.

The research work is expected to result in publications in scientific journals.

The development is expected to be done in R or Python.


Autres organisations


Ajouter votre question

Choisissez votre destination d'études


Choisissez le pays que vous souhaitez le visiter pour étudier gratuitement, travailler ou faire du bénévolat

Vous trouverez aussi