Many Co-Optimization of Fuels & Engines (Co-Optima) accomplishments have been made possible by the team’s development of new capabilities, numerical algorithms, and computational tools. The following data and tools can be accessed online by the wider research community.
A machine-Learning derived, Fast, Accurate Bond dissociation Enthalpy Tool (ALFABET)
ALFABET makes it possible for researchers to identify the most promising fuels for lower emissions and greater engine efficiency in seconds rather than days. Using ALFABET to determine bond dissociation enthalpy (BDE)—the energy required to break a chemical bond between atoms in organic compounds)—allows researchers to predict chemical reactions and determine suitability for certain uses including biofuels. This tool’s predictive model, developed through machine learning, closely matches the accuracy of traditional density-functional theory (DFT) calculations in a fraction of the time. Read the feature story to learn more about ALFABET.
The Co-Optimizer software tool makes it possible to assess candidate blendstocks in relation to tradeoffs involving a number of complex variables, including production scale and economics, life-cycle emissions, and infrastructure compatibility. Using the Co-Optima boosted spark-ignition merit function to identify blendstocks with the requisite properties to maximize engine efficiency when blended into petroleum base fuels, the tool uses Co-Optima-developed blending models to identify fully-blended fuels that meet current fuel quality specifications. User-supplied constraints then identify a smaller subset of solutions that can be compared over a wide range of market introduction scenarios.
A machine learning framework for predicting a variety of fuel properties, including cetane number and yield sooting index, based on molecular structure and using artificial neural networks. Pre-compiled databases for each of the properties are ready for use and extensive documentation outlining how to construct the models is available. Developers continually add a variety of software enhancements to decrease the amount of time required to construct models while increasing the accuracy of the models.
The continuously updated Fuel Properties Database focuses on bio-based fuel blendstocks (both pure components and mixtures) under investigation by the Co-Optima team and is populated with data from literature, as well as measured and/or predicted data. It contains data on more than 400 bio-based fuel blendstocks, as well as on gasoline and gasoline surrogates designed for such blending (reformulated blendstock for oxygenate blending, or RBOB).
The RetSynth (retrosynthesis) tool can be used to rapidly identify and evaluate the viability of pathways for producing bio-based molecules of interest to Co-Optima. Given a target molecule and a biomass-derived precursor and/or organism as input, RetSynth outputs the available biological, chemical, and hybrid production pathways, including a list of genes, reaction conditions, and theoretical yields for the target molecule. For biological pathways, RetSynth can also rank the optimal routes with the smallest number of steps.
Researchers integrated the yield sooting index (YSI) computational method into a tool that rapidly estimates the sooting tendency of fuel blendstocks, allowing the interactive development of potential new blendstocks that meet YSI targets. Experimental data on sooting tendency are continually added to the YSI database to broaden the scope of the compounds analyzed and improve prediction accuracy.