Session: 05-07 Advanced Diagnostics & Data Analytics I

Paper Number: 128650

128650 - Predicting Gas Turbine Power Output in Single Shaft Combined Cycle Power Plants: A Physics-Based and Data Analytics Approach 

Accurate measurement of gas turbine power output is essential for original equipment manufacturers (OEMs) to evaluate their gas turbines' performance. In multi-shaft combined cycle power plants, gas and steam turbine power are measured separately, enabling heat and mass balance analysis within individual turbine boundaries. However, in single shaft combined cycle power plants, only the total combined cycle power is measured, complicating separate gas and steam turbine power output predictions. Precise gas turbine power output prediction is critical to avoid improper operations, such as over-firing or under-firing, which can adversely impact life of hot gas path components and/or lead to power loss.

The paper's first section presents a method for predicting single shaft gas turbine power using the first law of thermodynamics. To validate this approach, data from a multi-shaft combined cycle power plant is used. The objective is to confirm the first law method using limited data, including compressor inlet temperature, fuel flow, fuel lower heating value, combined cycle power output, and gas turbine power output. This article explores data preparation, validation, and methodology limitations for predicting gas turbine output in single-shaft combined cycle power plants.

The second section examines sensor noise elimination through advanced filtration techniques and principal component analysis of critical measured parameters. The paper investigates output prediction using Just-In-Time Gaussian Process Regression (JIT-GPR) and JIT-GPR with Deep Feedforward Neural Network (DFNN) methods for process regression. Lastly, power output prediction using system identification techniques is discussed.

In conclusion, this paper offers valuable insights into the significance of forecasting gas turbine power output in single-shaft gas turbine combined cycle power plants. This approach underscores the importance of using soft-sensors to understand engine state during its normal operation in near real-time environment, thereby creating opportunities for gas turbine performance optimization.

 

Presenting Author: Pugalenthi Nandagopal Siemens India LTD

Presenting Author Biography: Gas Turbine Performance engineer with 17+ years of experience in thermodynamics & performance domain.

Authors:

Pugalenthi Nandagopal Siemens India LTD
Manjunath More Siemens India Ltd
Kirti Sharma Indian Institute Of Technology–Ropar (IIT–Ropar)
Uttam Saroj Indian Institute Of Technology–Ropar (IIT–Ropar)
Gurdev Singh Siemens Energy