A new methodology of thermodynamic diagnosis, using the thermoeconomic method together with an artificial neural network (ANN): A case study of an externally fired gas turbine (EFGT)

• Published in: Energy (Oxford) Vol. 123; pp. 20 - 35
• Main Authors: Orozco, Dimas José Rúa, Venturini, Osvaldo José, Escobar Palacio, José Carlos, del Olmo, Oscar Almazán
• Format: Journal Article
• Language: English
• Published: Oxford Elsevier Ltd 15-03-2017; Elsevier BV
• Subjects: Artificial neural networks; Computer programs; Computer simulation; Devices; Diagnosis; Diagnostic systems; Efficiency; Exergetic operator; Exergy; Fuel impact; Gas turbine engines; Gas turbines; Intrinsic malfunction; Malfunctions; Neural networks; Power efficiency; Residual gas; Routines; Simulation; Software; Thermodynamic diagnosis; Thermodynamics; Transition structure; Fuel impact; Thermodynamic diagnosis; Exergetic operator; Intrinsic malfunction; Transition structure
• ISSN: 0360-5442; 1873-6785
• DOI: 10.1016/j.energy.2016.12.099

Thermodynamic diagnostics aim to identify and act upon thermal system devices exhibiting abnormal behaviour (malfunctions) in order to later, through the use of maintenance routines, return the devices to their optimum operating condition. Several methods have been developed to solve the problem of thermal system thermodynamic diagnosis, all of which are designed to identify components exhibiting malfunctions and their effect on cycle power output and efficiency. Individually these methods have both advantages and disadvantages, with the complementary use of two or more providing better results. In this paper a diagnostic system is proposed for externally fired gas turbines (EFGT), using the thermoeconomic method in conjunction with artificial neural networks to identify malfunctioning components (intrinsic malfunctions) and their fuel impact. The concepts “Exergetic Operator” and “Transition Structure” are also presented. An EFGT was simulated using the commercial software GateCycle™ 5.51, aiming to reach a power of 99.80 kW (design point) using wood carbonisation residual gas as fuel. An ANN was developed with the commercial software MATLAB®.
•A thermoeconomic method of diagnosis is developed to separate the fuel impact into its three components.•To identify an intrinsic malfunction it is necessary to consider that it can be represented by two or more components.•The Fuel Impact Formula together with ANN and two new concepts: Exergetic Operator and Transition Structure are used.•The equipment exhibiting anomalies is separated into many components: equipment without anomalies and Exergetic Operators.•The transition structure is created substituting all components by their corresponding equipment plus exergetic operators.