Performance and Emission Characteristics of CI Engine Fueled with Lemongrass Biodiesel-Diesel Blends and Prediction Using Ml Techniques

Abstract

 The increased fears regarding depletion of fossil fuels and pollution of the environment have amplified the hunt toward sustainable and renewable sources of energy like biodiesel. A non-edible biomass source, and underutilized resource, lemongrass (Cymbopogon flexuosus) oil, has been considered as a potential feedstock source of biodiesel because of its high oil content, and because of its good combustion characteristics. In this paper, the authors will attempt to measure the performance and emission features of a single-cylinder compression ignition (CI) engine operated on different blends (B10 to B100) of lemongrass biodiesel and fossil diesel. The biodiesel was synthesized under optimal conditions of transesterification process with heterogeneous catalyst and its blends were tested across different engine loading conditions. The important engine performance parameters such as Brake Thermal Efficiency (BTE), Specific Energy Consumption (SEC), and Hydrocarbon (HC) emissions were recorded. Also, an Artificial Neural Network (ANN) based machine learning model was designed to provide an estimate of biodiesel yield as a function of catalyst concentration. The outcomes indicated that B10 had the best BTE (33.81%) at full load, whereas B20 had the least SEC (10.3 MJ/kWh), which means that the combustion was efficient. B50 had the lowest HC values (32 ppm) confirming the ecological superiority of mid-range blends. ANN model demonstrated very good predictive accuracy with deviation between +0.5 and -0.5 percent of observed real yield values. To sum up, blends of lemongrass biodiesel with diesel, especially in lower and mid blends, are good candidates to be used in CI engines with improved performance and minimal emission, and ML tools present an excellent predictive tool that can be used to optimize biodiesel production.