Stochastic Reconstruction of Complex Heavy Oil Molecules Using an Artificial Neural Network

Abstract

An approach for the stochastic reconstruction of petroleum fractions based on the joint use of artificial neural networks and genetic algorithms was developed. This hybrid approach reduced the time required for optimization of the composition of the petroleum fraction without sacrificing accuracy. A reasonable initial structural parameter set in the optimization space was determined using an artificial neural network. Then, the initial parameter set was optimized using a genetic algorithm. The simulations show that the time savings were between 62 and 74% for the samples used. This development is critical, considering that the characteristic time required for the optimization procedure is hours or even days for stochastic reconstruction. In addition, the standalone use of the artificial neural network step that produces instantaneous results may help where it is necessary to make quick decisions.

Effect of Boiling Point and Density Prediction Methods on Stochastic Reconstruction

“…If the molecule is not found in the database, it is saved in the database and invoked from the database when it is needed again. Further details of the parameters and molecule generation algorithm are given elsewhere. ,, …”

“…Further details of the parameters and molecule generation algorithm are given elsewhere. 21,30,31 3.3.1. Ring Compactness Factor.…”

“…For instance, a quadrature approach was used to reduce the number of molecules formed to reconstruct the mixture . The use of multiple linear regression models and artificial neural networks , were investigated, and a new set of structural identifiers was proposed . Recently, biomass has been reconstructed through an additionally elaborate set of PDFs in order to model the gasification process at the molecular level .…”

Moving from postdictive to predictive kinetics in reaction engineering

“…Note that even for freshly prepared catalysts, the active sites are usually occupied by something, they are not completely vacant, and indeed it is commonly observed that different pretreatment or “activation” of a catalyst leads to different measured conversions.If we are performing reactions on complicated feeds such as biomass, petroleum fractions, or waste plastic, we generally do not know the molecular composition of the feed, though we usually have some partial information. However, there are well‐developed “feed reconstruction” methods 73‐76 for constructing a reasonable estimate of the feed composition. Convenient affordable methods exist to solve most simple kinetic models (e.g., <100 chemical species) in common idealized reactors, in packages such as CHEMKIN 77 and CANTERA 78 . Typically the idealized reactors are 0‐d or 1‐d, sometimes 2‐d, but not three‐dimensional (3‐d).…”

“…If we are performing reactions on complicated feeds such as biomass, petroleum fractions, or waste plastic, we generally do not know the molecular composition of the feed, though we usually have some partial information. However, there are well‐developed “feed reconstruction” methods 73‐76 for constructing a reasonable estimate of the feed composition.…”