This engine study varies injector-hole diameter from 0.26–0.34 mm across diesel and ammonia–biodiesel blends, then trains machine-learning models to predict performance and emissions from 100 operating points.
Key findings
- A20 with a 0.26-mm injector delivered 6.39 kW indicated power, 35.3% brake thermal efficiency and 3 g/kWh NOx. The authors attribute this to finer atomisation and mixing. Machine-learning accuracy reached about 99% within the dataset.
Why this matters globally
Injector and blend optimisation may improve ammonia–biodiesel use in existing engines, but real decarbonisation depends on fuel production, energy inputs and pollutants not captured by a narrow engine test.
Thai researcher contribution
Kittitat Yinyuan, jointly affiliated with Shinawatra University and institutions in India, contributed Thai-affiliated expertise to the international engine study.
Limitations to consider
Only 100 points from one engine and operating range were used. Data splitting and preprocessing within validation require scrutiny for leakage; 99% accuracy may reflect interpolation rather than transfer to another engine. Ammonia slip, N₂O, particulates, durability, safety, cost and life-cycle carbon were not established.