Cluster Recognition and Early Warning Modeling for Rotating Stall of Gas Turbine (iSPEC 2020)

Author(s):  
Bochao Xu
NDT World ◽  
2021 ◽  
pp. 58-61
Author(s):  
Aleksey Popov ◽  
Aleksandr Romanov

A large number of aviation events are associated with the surge of gas turbine engines. The article analyzes the existing systems for diagnostics of the surge of gas turbine engines. An analysis of the acoustic signal of a properly operating gas turbine engine was carried out, at which a close theoretical distribution of random values was determined, which corresponds to the studied distribution of the amplitudes of the acoustic signal. An invariant has been developed that makes it possible to evaluate the development of rotating stall when analyzing the acoustic signal of gas turbine engines. A method is proposed for diagnosing the pre-surge state of gas turbine engines, which is based on processing an acoustic signal using invariant dependencies for random processes. A hardware-software complex has been developed using the developed acoustic method for diagnosing the pre-surge state of gas turbine engines.


Author(s):  
Thomas Palme´ ◽  
Peter Breuhaus ◽  
Mohsen Assadi ◽  
Albert Klein ◽  
Minkyo Kim

This study investigates the application of nonlinear Principal Component Analysis (PCA), implemented through the use of Auto-Associative Neural Network (AANN), for early warning of impending gas turbine failure. The study is based on a real operational data set that includes a compressor failure. The analyzed data set consists of measured operational parameters whose identity are unknown, hence this study presents a purely data driven approach to the problem of early warning. In this case study, the use of AANNs for early detection of abnormal engine behavior could have provided the operator with a warning a few days prior to the fully developed failure, which resulted in a forced shut-down and extensive maintenance. Furthermore, a comparison is made between the nonlinear PCA by AANNs and the standard PCA model, which is an inherently linear method. The result shows that the AANN provides a more reliable detection of the failure by a higher residual generation during failure mode as well as fewer false indications prior to the failure. Consequently, this study shows that nonlinear PCA as performed with AANNs can be a valuable data driven tool for early warning of gas turbine failure.


Author(s):  
Zdzislaw Mazur ◽  
Alejandro Herna´ndez-Rossette ◽  
Jesu´s Porcayo-Caldero´n

A compressor blade failure was experienced at the 69 MW gas turbine of a combined cycle (C.C.) unit after four years operation since the last overhaul (January 2005). The unit accumulated 27,000 service hours and 97 start-ups since the last overhaul. This unit consists of four gas turbine stages and 19 compressor stages and operates at 3600 rpm. In 2006, the unit was equipped with a fogging system at the compressor air inlet duct to increment unit power output during high ambient temperature days (hot days). These fog water nozzles were installed upstream of the compressor inlet air filter without any water filter/catcher before the water spray nozzles. Three unit failure events occurred at small periods, which caused forced outage. The first failure occurred in December 2008, a second event in March 2009 and the third event in May 2009. Visual examination carried out after the first failure event indicated that the compressor vanes (diaphragms) had cracks in their airfoils initiating at blade tenons welded to the diaphragm outer shroud at stages 3, 8, 9, 10 and 11. Also, many stationary vanes and moving blades at each stage of the compressor showed foreign object damage (FOD) and fractures at the airfoil. Visual examination performed for the second failure event after 60 unit operation hours indicated that many compressor vanes (diaphragms) and moving blades had FOD at the airfoil. This was attributed to fractures of the fogging system water spray nozzle, which were then induced to the compressor flow path channel at high velocity causing the above-mentioned damage. Visual examination completed upon the third failure event after two unit startup attempts indicated damage of compressor stationary vanes and moving blades principally at stages 12 to 16, and also stages 17 to 19. The damage consisted of airfoil fracture in stationary vanes and moving blades, FOD, moving blade tip rubbing, and bending of stationary vanes, moving blades and diaphragm shrouds. A laboratory evaluation of stationary vane tenon fracture indicated a high cycle fatigue (HCF) failure mechanism, and crack initiation was accelerated by corrosion picks on blade surfaces due to high humidity air generated by the fogging system. Stationary vane damage was caused by a rotating stall phenomenon, which generates vibratory stresses in stationary vanes and moving blades during unit start-ups. During the third failure event, stationary vane HCF damage was highly accelerated due to pre-existent partial fractures in tenons generated during previous failure events, which had not been detected by non-destructive tests. Stationary vane and moving blade failure was also influenced by high tenon brittleness in stationary vanes and moving blades generated during manufacture by welding (diaphragms) and repair welding (moving blades) without adequate post-weld heat treatment (stress relieving). A compressor stationary vane and moving blade failure evaluation was completed. This investigation included cracked blade metallographic analysis, unit operation parameter analysis, history-of-events analysis, and crack initiation and propagation analysis. This paper provides an overview of the compressor failure investigation, which led to identification of the HCF failure mechanism generated by rotating stall during unit start-ups, highly accelerated by corrosion generated by the fogging system and influenced by high stationary vane and moving blade brittleness as the primary contribution to the observed failure.


2017 ◽  
Vol 176 ◽  
pp. 207-217 ◽  
Author(s):  
A.E. Mikhailov ◽  
A.B. Mikhailova ◽  
Yu.M. Akhmetov ◽  
D.A. Akhmedzyanov

Author(s):  
Brian Kestner ◽  
Tim Lieuwen ◽  
Chris Hill ◽  
Leonard Angello ◽  
Josh Barron ◽  
...  

This paper summarizes an analysis of data obtained from an instrumented compressor of an operational, heavy duty industrial gas turbine; the goal of the aforementioned analysis is to understand some of the fundamental drivers, which may lead to compressor blade vibration. Methodologies are needed to (1) understand the fundamental drivers of compressor blade vibration, (2) quantify the severity of “events,” which accelerate the likelihood of failure and reduce the remaining life of the blade, and (3) proactively detect when these issues are occurring so that the operator can take corrective action. The motivation for this analysis lies in understanding the correlations between different sensors, which may be used to measure the fundamental drivers and blade vibrations. In this study, a variety of dynamic data was acquired from an operating engine, including acoustic pressure, bearing vibration, tip timing, and traditional gas path measurements. The acoustic pressure sensors were installed on the first four compressor stages, while the tip timing was installed on the first stage only. These data show the presence of rotating stall instabilities in the front stages of the compressor, occurring during every startup and shutdown, and manifesting itself as increased amplitude oscillations in the dynamic pressure measurements, which are manifested in blade and bearing vibrations. The data that lead to these observations were acquired during several startup and shutdown events, and clearly show that the amplitude of these instabilities and the rpm at which they occur can vary substantially.


Author(s):  
Meinhard T. Schobeiri

Gas turbines in general and aircraft engines in particular undergo frequently dynamic operations. These operations include the routine start-up, load change and shut downs to cover their operation envelope. The frequency of the dynamic operation depends on the size of the engines and the field of application. Engines for commuter aircrafts and particularly helicopter engines operate more often in an off-design mode compared to large commercial aircraft engines and power generation gas turbines. During these routine operations, the compressor mass flow, the pressure ratio, the combustion chamber fuel and air mass flow as well as turbine mass flow change. These changes affect the engine aerodynamic performance and its efficiency. To avoid the inception of rotating stall and surge, high performance gas turbines are equipped with mechanisms that adjust the stator stagger angles thus aligning the stator exit flow angle to the rotor inlet angle, which reduces an excessive incidence. The reduction of incidence angle not only preserves the stable operation of the compressor but it also prevents the compressor efficiency from deterioration. The existence of an inherent positive pressure gradient may cause the boundary layer separation on compressor blades leading to the rotating stall and surge. Such condition, however, does not exist in a turbine, and therefore, there has been no compelling reason to apply the blade adjusting method to the turbine component. For the first time, the impact of turbine blade stagger angle adjustment on the gas turbine efficiency during the operation is shown in this paper. Given a statistically distributed load condition, the extensive dynamic simulation reported in this paper shows how the efficiency can be positively affected through proper blade adjustment. For the time dependent operation, the code GETRAN developed by the author was enhanced to include the turbine blade adjustment as a function of time. To conduct the dynamic simulation with turbine stator stagger angle adjustment during a dynamic operation, the full geometry of the Brown Boveri GT-9 gas turbine was utilized. Starting from the reference stagger angle, it is varied within an incidence range of ± 3 degree. Detailed simulation results show the substantial efficiency improvement through stator stagger blade adjustment.


Author(s):  
Santhosh Kasram ◽  
Sajath Kumar Manoharan ◽  
Mahesh P. Padwale ◽  
G. P. Ravishankar

Abstract The challenges faced during starting of an aircraft gas turbine engine using a Jet Fuel Starter (JFS) at high altitude airbase are discussed in this paper. Autonomous ground starts at high altitude airbase in soaked sub-zero temperature condition without any external ground support assistance is a challenge. Generally, the start cycle (sub-idle speed) at sub-zero temperatures of a gas turbine engine at high altitudes is influenced by several factors. Drag loads are estimated due to change in lube oil viscosity of engine gearbox and accessory gear box that affects available torque margin of a starter. These estimated loads are superimposed on starter characteristics to identify the available margins for successful starts. The cold start is particularly severe, since it increases the tip clearance between rotor and casing of the engine due to difference in its thermal growth. Higher tip clearances significantly degrade compressor surge margin and results in rotating stall. Inconsistent engine starts were resolved by adopting alternative methods without any change in hardware. This paper presents set of methods used to overcome inconsistent engine starts at high altitude cold weather conditions.


Author(s):  
Abdesselam Debbah ◽  
Hamid Kherfane

Abstract In gas turbine process, the axial compressor is subjected to aerodynamic instabilities because of rotating stall and surge associated with bifurcation nonlinear behaviour. This paper presents a Genetic Algorithm and Particle Swarm Optimization (GA/PSO) of robust sliding mode controller in order to deal with this transaction between compressor characteristics, uncertainties and bifurcation behaviour. Firstly, robust theory based equivalent sliding mode control is developed via linear matrix inequality approach to achieve a robust sliding surface, then the GA/PSO optimization is introduced to find the optimal switching controller parameters with the aim of driving the variable speed axial compressor (VSAC) to the optimal operating point with minimum control effort. Since the impossibility of finding the model uncertainties and system characteristics, the adaptive design widely considered to be the most used strategy to deal with these problems. Simulation tests were conducted to confirm the effectiveness of the proposed controllers.


Sign in / Sign up

Export Citation Format

Share Document