Session: 

Paper Number: 151736

Reliable Exit Temperature Profile Prediction Using Engine Test Data and Neural Networks 

The reliable exit temperature profile prediction of gas turbine engine development is extremely significant for the optimal turbine design at the conceptual design stage. The exit temperature profile is unstable and variable to the burning time and combustor design. In the early stage of expandable and low-cost engine development, the first prototype core engine is rapidly designed, manufactured, integrated and tested for obtaining the engine performance and component test data with the limited pressure, thermal and vibration sensors at 0 degree, 120 degree and 240 degree at the difference depth locations. It causes the lacking temperature information according the exit of combustor cross section, therefore the deep Neural Networks is proposed to be implemented for training the first prototype test data with the time series, engine RPM, engine performance and different locations of exit temperature sensors. The trained Neural Networks is used to predict the entire combustor exit temperature profile reliably compared to the empirical and CFD analysis results.

The deep Neural Network model is investigated with different algorithms such as sgdm, rmsprop and adam and the number of neurons in the hidden layers for finding the proper learning model of the first prototype exit temperature test data sets while maintaining the accuracy and training time of model. The prediction of first engine exit temperature profile is presented and compared to the limited sensors test points and more reliable the CFD prediction results. The calibrated factors between prediction model and CFD results is proposed to use for enhancing the next version of expandable turbine engine development for the exit temperature profile prediction.

Keywords: Exit temperature profile (ETP), Neural Networks (NN), Test data, Deep learning, CFD

Presenting Author: Tuan Anh Pham Viettel Aerospace Institute - Viettel Group

Presenting Author Biography: Aerodynamic design engineer at Viettel Aerospace Institute. Graduated from Kharkov Polytechnic University in 2017. Graduated from Kharkov Polytechnic University in 2019 with a master's degree in turbine engine design and manufacturing. Research areas include design and optimization of compressor and turbine aerodynamics. Modeling of gas turbine engines.

Authors:

Tuan Anh Pham Viettel Aerospace Institute - Viettel Group
Duy Lanh Chu Viettel Aerospace Institute - Viettel Group
Quang Hai Nguyen Viettel Aerospace Institute - Viettel Group
Phi Minh Nguyen Viettel Aerospace Institute - Viettel Group
Xuan Hung Vu Viettel Aerospace Institute - Viettel Group
Xuan Long Bui Viettel Aerospace Institute - Viettel Group
Cong Anh Pham Viettel Aerospace Institute - Viettel Group
Huy Hoang Nguyen Viettel Aerospace Institute - Viettel Group