In a significant stride towards enhancing internal combustion (IC) engine performance and reducing emissions, researchers have developed a novel approach to estimate oxyhydrogen (HHO) gas production using machine learning algorithms. The study, led by Mohammad Amin Adoul from the Division of Operation and Maintenance Engineering at Luleå University of Technology in Sweden, was recently published in the journal “Energy Conversion and Management: X” (formerly known as Energy Conversion and Management: Reports).
The research focuses on a wet-type flat-plate electrolyser that uses sodium hydroxide (NaOH) as an electrolyte. The team investigated the interdependent effects of voltage, current, and NaOH concentration on HHO gas generation rate and system efficiency. Their findings reveal that moderate current and voltage levels, combined with higher NaOH concentrations, yield a maximum gas production rate of 0.5 liters per minute while maintaining energy efficiency.
Adoul explains, “We found that as the current increases, the rate of HHO gas production also increases. The maximum production of 0.5 L/min was achieved with a current of 30 amperes.” This discovery is crucial for optimizing gas output and efficiency while minimizing resource usage.
The study goes a step further by employing machine learning algorithms to predict the performance of the HHO gas system. The team used experimental data to train five different gradient boosting algorithms, including categorical boosting (CatBoost), extreme gradient boosting (XGBoost), light gradient boosting machine (LightGBM), adaptive boosting (AdaBoost), and gradient boosting (GB). Among these, CatBoost achieved the highest accuracy, with R2 values of 0.9903 for hydrogen production and 0.9583 for efficiency on test data.
The implications of this research are far-reaching for the energy sector. By optimizing the production of HHO gas, which can be used as a supplementary fuel in IC engines, this technology has the potential to improve engine performance and reduce harmful emissions. The use of machine learning algorithms to predict system performance can also significantly reduce experimental burdens and costs, making the technology more accessible and scalable.
Adoul envisions a future where this technology could be integrated into various industries. “The potential applications are vast,” he says. “From automotive to industrial processes, the optimized production of HHO gas can lead to more efficient and environmentally friendly operations.”
As the world continues to seek sustainable energy solutions, this research offers a promising avenue for enhancing existing technologies and reducing their environmental impact. The integration of machine learning with traditional engineering approaches demonstrates the power of interdisciplinary collaboration in driving innovation and shaping the future of the energy sector.