Theoretical and experimental study on performance improvement of diesel engines at different altitudes by adaptive regulation method of the two-stage turbocharging system

• Published in: Energy (Oxford) Vol. 281; p. 128291
• Main Authors: Lu, Kangbo, Shi, Lei, Zhang, Huiyan, Chen, Ziqiang, Deng, Kangyao
• Format: Journal Article
• Language: English
• Published: Elsevier Ltd 15.10.2023
• Subjects: Altitude; Control strategy; Diesel engine; Prediction model; Turbocharging system; Diesel engine; Turbocharging system; Control strategy; Prediction model; Altitude
• ISSN: 0360-5442
• DOI: 10.1016/j.energy.2023.128291

The regulation of the two-stage turbocharging system according to the operation of the diesel engine is key to maximizing its potential for power recovery of an engine at high altitudes. This paper targets air intake recovery of diesel engines at different altitudes, the effects of diesel engine operating conditions, in-cylinder combustion, mechanical efficiency, exhaust temperature, turbocharging system operating parameters, and altitudes on the demand characteristics of the equivalent turbine area (ETA) were analyzed by a thermodynamic model, and a multifactor coupled prediction model for the ETA was established, which can predict the regulation of the ETA at different altitudes and operation conditions based on plain and a small amount of plateau data. To improve the accuracy of the model, the error source of the model was analyzed, and the neural network was adapted to predict major errors. Results indicate that the prediction model had high accuracy and the maximum predicted error of ETA is less than 3%. The control strategy obtained by prediction model can improve air supply and efficiency of turbocharging system, the power and economy of the diesel engine also improved at different altitudes. •The equivalent turbine area prediction model of the turbocharging system at different altitudes is proposed.•The neural network was adopted to predict turbocharger efficiency.•The improvement of the diesel engine performance by prediction model was investigated.