Machine learning models for properties prediction and design of fuel/biofuel blends.

AYLEN MERLO RAIMONDI; SELVA PEREDA; ANIBAL BLANCO; MARIANA GONZALEZ PRIETO

Kumamoto

Encuentro; AY2022 JST Sakura Science Exchange Program, "Cross-Cultural and Multidisciplinary Exchanges on Alternative Energies for SDGs"; 2023

Kumamoto University

Biofuels are an essential component of the future low-carbon global energy system, playing an important role in the decarbonization of heavy transportation sectors.Additionally, novel fuels are expected to perform efficiently on current and next-generation engines. Several properties characterize the quality of a fuel, many closely related to phase equilibria and can be predicted with proper thermodynamic models. On the other hand, the octane number (ON) of a fuel is a non-equilibrium property that measures the fuel ignition quality and its tendency to resist knocking. The ON strongly depends on the molecular structure of the species in the fuel, as well as on the interaction among them, and its experimental determination is expensive and time consuming.In this work, a new model to predict ON is presented and integrated into a fuel/biofuel blend design software, which already takes into account volatility properties of the fuel, such as Reid vapor pressure, distillation curve and the PIANOX classification. Many ON models proposed in the open literature were evaluated concluding that none of them is capable of accurately predicting ON of pure compounds and mixtures simultaneously.Since machine learning (ML) models have shown to be flexible enough to describe the complex antagonistic and synergistic effect of multicomponent mixtures, we propose a new artificial neural network (ANN) model for the prediction of ON of pure components and mixtures. The ANN inputs are the functional groups that shape the species and other physicochemical descriptors such as species molar volume, normal boiling point, heat of vaporization, among others. The proposed approach also aims to identify which descriptors, and to which extent, they control the fuel ON.