Bayesian Network Approach for Gas Path Fault Diagnosis

2004 ◽  
Vol 128 (1) ◽  
pp. 64-72 ◽  
Author(s):  
C. Romessis ◽  
K. Mathioudakis
Keyword(s):  
Fault Diagnosis ◽  
Diagnostic Procedure ◽  
Deterministic Model ◽  
Probabilistic Approach ◽  
Engine Performance ◽  
Performance Model ◽  
Jet Engine ◽  
Turbine Performance ◽  
Gas Path Fault ◽  
Set Up

A method for solving the gas path analysis problem of jet engine diagnostics based on a probabilistic approach is presented. The method is materialized through the use of a Bayesian Belief Network (BBN). Building a BBN for gas turbine performance fault diagnosis requires information of a stochastic nature expressing the probability of whether a series of events occurred or not. This information can be extracted by a deterministic model and does not depend on hard to find flight data of different faulty operations of the engine. The diagnostic problem and the overall diagnostic procedure are first described. A detailed description of the way the diagnostic procedure is set-up, with focus on building the BBN from an engine performance model, follows. The case of a turbofan engine is used to evaluate the effectiveness of the method. Several simulated and benchmark fault case scenarios have been considered for this reason. The examined cases demonstrate that the proposed BBN-based diagnostic method composes a powerful tool. This work also shows that building a diagnostic tool, based on information provided by an engine performance model, is feasible and can be efficient as well.

Bayesian Network Approach for Gas Path Fault Diagnosis

2004 ◽  
Author(s):  
C. Romessis ◽  
K. Mathioudakis
Keyword(s):  
Fault Diagnosis ◽  
Diagnostic Procedure ◽  
Deterministic Model ◽  
Probabilistic Approach ◽  
Engine Performance ◽  
Performance Model ◽  
Jet Engine ◽  
Turbine Performance ◽  
Gas Path Fault ◽  
Set Up

A method for solving the gas path analysis problem of jet engine diagnostics based on a probabilistic approach is presented. The method is materialized through the use of a Bayesian Belief Network (BBN). Building a BBN for gas turbine performance fault diagnosis requires information of a stochastic nature expressing the probability of whether a series of events occurred or not. This information can be extracted by a deterministic model and does not depend on hard to find flight data of different faulty operations of the engine. The diagnostic problem and the overall diagnostic procedure are first described. A detailed description of the way the diagnostic procedure is set-up, with focus on building the BBN from an engine performance model, follows. The case of a turbofan engine is used to evaluate the effectiveness of the method. Several simulated and benchmark fault case scenarios have been considered for this reason. The examined cases demonstrate that the proposed BBN-based diagnostic method composes a powerful tool. This work also shows that building a diagnostic tool, based on information provided by an engine performance model, is feasible and can be efficient as well.

Uncertainty Quantification of a Jet Engine Performance Model Under Scarce Data Availability

2021 ◽  
Author(s):  
Norbert Ludwig ◽  
Giulia Antinori ◽  
Marco Daub ◽  
Fabian Duddeck
Keyword(s):  
Uncertainty Quantification ◽  
Probabilistic Approach ◽  
Engine Performance ◽  
Theoretical Background ◽  
Performance Model ◽  
Data Availability ◽  
Data Set ◽  
Jet Engine ◽  
Probabilistic Uncertainty ◽  
Flow Capacity

Abstract The development of jet engine components requires a detailed quantification of different uncertainty sources to improve the quality and robustness of the design. In order to get a better understanding of the entire interdisciplinary jet engine design process, the detailed uncertainty quantification of the performance parameters is of vital importance. This paper demonstrates a new approach how to represent the uncertainties in a jet engine performance model caused by the manufacturing and assembly process. In earlier research studies, the manufacturing process was modeled with a probabilistic approach, i.e. by assuming a multivariate normal distribution for the corresponding parameters. Within the scope of this paper, the uncertainty of the components’ flow capacity and efficiency is quantified based on a limited set of data. Due to the extreme scarcity of the data set, it is proposed to use methods from the field of non-probabilistic uncertainty quantification. In this paper, three different approaches to derive the variation of the components’ flow capacity and efficiency are compared with each other. In contrast to probabilistic methodologies, all approaches are able to represent the lack of data without making any additional assumptions regarding the underlying type of distribution. As a result, each of the methodologies describes the uncertain parameters by probability-boxes. After clarifying the theoretical background, the results obtained from the different approaches are discussed in detail. It figured out that the propagation method for probability-boxes plays a crucial role for the uncertainty quantification.

A Hyper-Ellipsoid Approach for Inverse Lack-of-Knowledge Uncertainty Quantification

2021 ◽  
Vol 7 (2) ◽  
Author(s):  
Norbert Ludwig ◽  
Fabian Duddeck ◽  
Marco Daub
Keyword(s):  
Uncertainty Quantification ◽  
Engine Performance ◽  
Computation Time ◽  
Solution Procedure ◽  
Performance Model ◽  
High Dimensional ◽  
System Response ◽  
Test Case ◽  
Jet Engine ◽  
Output Space

Abstract This paper presents a novel methodology to solve an inverse uncertainty quantification problem where only the variation of the system response is provided by a small set of experimental data. Furthermore, the method is extended for cases where the uncertainty of the response quantities is given by an incomplete set of statistical moments. For both cases, the uncertainty on the output space is represented by a minimum volume enclosing ellipsoid (MVEE). The actual inverse uncertainty quantification is conducted by identifying also a hyper-ellipsoid for the input parameters, which has an image on the output space that matches the MVEE as close as possible. Hence, the newly introduced approach is a contribution to the field of nonprobabilistic uncertainty quantification methods. Compared to literature, the new approach has often superior accuracy and especially an improved efficiency for high-dimensional problems. The method is validated first by an analytical test case and subsequently applied to a jet engine performance model, where this type of inverse uncertainty quantification has to be solved to allow for a consistent and integrated solution procedure. In both cases, the results are compared with an inverse method where the variability on the input side is quantified by a multidimensional interval. It can be shown that the hyper-ellipsoid approach is superior with respect to the computation time in high-dimensional problems encountered not only in jet engine design.

Performance Model “Zooming” for In-Depth Component Fault Diagnosis

2010 ◽  
Author(s):  
N. Aretakis ◽  
I. Roumeliotis ◽  
K. Mathioudakis
Keyword(s):  
Fault Diagnosis ◽  
Measurement Data ◽  
Engine Performance ◽  
Main Tool ◽  
Performance Model ◽  
Diagnostic Methods ◽  
Performance Modelling ◽  
Engine Model ◽  
Path Component ◽  
Modelling Techniques

A method giving the possibility for a more detailed gas path component fault diagnosis, by exploiting the “zooming” feature of current performance modelling techniques, is presented. A diagnostic engine performance model is the main tool that points to the faulty engine component. A diagnostic component model is then used to identify the fault. The method is demonstrated on the case of compressor faults. A 1-D model based on the “stage stacking” approach is used to “zoom” into the compressors, supporting a 0-D engine model. A first level diagnosis determines the deviation of overall compressor performance parameters, while “zooming” calculations allow a localization of the faulty stages of a multistage compressor. The possibility to derive more detailed information with no additional measurement data is established, by incorporation of empirical knowledge on the type of faults that are usually encountered in practice. Although the approach is based on known individual diagnostic methods, it is demonstrated that the integrated formulation provides not only higher effectiveness but also additional fault identification capabilities.

Performance Model “Zooming” for In-Depth Component Fault Diagnosis

2010 ◽  
Vol 133 (3) ◽  
Author(s):  
N. Aretakis ◽  
I. Roumeliotis ◽  
K. Mathioudakis
Keyword(s):  
Fault Diagnosis ◽  
Measurement Data ◽  
Engine Performance ◽  
Main Tool ◽  
Performance Model ◽  
Diagnostic Methods ◽  
Engine Model ◽  
Path Component ◽  
1D Model ◽  
Compressor Performance

A method giving the possibility for a more detailed gas path component fault diagnosis by exploiting the “zooming” feature of current performance modeling techniques is presented. A diagnostic engine performance model is the main tool that points to the faulty engine component. A diagnostic component model is then used to identify the fault. The method is demonstrated on the case of compressor faults. A 1D model based on the “stage stacking” approach is used to “zoom” into the compressors, supporting a 0D engine model. A first level diagnosis determines the deviation of overall compressor performance parameters while zooming calculations allow a localization of the faulty stages of a multistage compressor. The possibility to derive more detailed information with no additional measurement data is established by the incorporation of empirical knowledge on the type of faults that are usually encountered in practice. Although the approach is based on known individual diagnostic methods, it is demonstrated that the integrated formulation provides not only higher effectiveness but also additional fault identification capabilities.

Instructing the Principles of Gas Turbine Performance Monitoring and Diagnostics by Means of Interactive Computer Models

2000 ◽  
Author(s):  
K. Mathioudakis ◽  
A. Stamatis ◽  
A. Tsalavoutas ◽  
N. Aretakis
Keyword(s):  
Gas Turbine ◽  
Gas Turbines ◽  
Performance Monitoring ◽  
Measurement Data ◽  
Engine Performance ◽  
Performance Model ◽  
Operating Conditions ◽  
Engine Operation ◽  
Turbine Performance ◽  
Engine Components

The paper discusses how the principles employed for monitoring the performance of gas turbines in industrial duty can be explained by using suitable Gas Turbine performance models. A particular performance model that can be used for educational purposes is presented. The model allows the presentation of basic rules of gas turbine engine behavior and helps understanding different aspects of its operation. It is equipped with a graphics interface, so it can present engine operating point data in a number of different ways: operating line, operating points of the components, variation of particular quantities with operating conditions etc. Its novel feature, compared to existing simulation programs, is that it can be used for studying cases of faulty engine operation. Faults can be implanted into different engine components and their impact on engine performance studied. The notion of fault signatures on measured quantities is clearly demonstrated. On the other hand, the model has a diagnostic capability, allowing the introduction of measurement data from faulty engines and providing a diagnosis, namely a picture of how the performance of engine components has deviated from nominal condition, and how this information gives the possibility for fault identification.

Open Rotor Engine Cycle Modeling and Performance Assessment

2020 ◽  
Author(s):  
Ziyu Zhang ◽  
Li Zhou ◽  
Xiaobo Zhang ◽  
Zhanxue Wang
Keyword(s):  
Engine Performance ◽  
Performance Model ◽  
Civil Aviation ◽  
Environment Impact ◽  
Turbine Performance ◽  
New Concepts ◽  
Open Rotor ◽  
Potential Impact ◽  
And Performance ◽  
Bypass Ratio

Abstract Aiming to enable dramatic reductions in the environment impact and fuel consumption of future civil aviation, NASA and European related research institutions are committed to developing new concepts and technologies in which counter rotating open rotor (CROR) concept can achieve this objective. In order to evaluate its potential impact, an open rotor engine performance model needs to be established. This paper presents the modeling method of an open rotor engine with the geared counter rotating open rotor (GOR) as object, and implements it in an in-house modular program of gas turbine performance prediction. In addition, the steady-state performance of the model is analyzed, and the model accuracy is verified based on the existing data. On this basis, the performance of open rotor engine and high bypass ratio turbofan engine is compared and results show that the counter rotating open rotor engine has obvious fuel saving advantages.

Generation of Physically Based Analysis Factors to Improve Synthesis Models of Jet Engines

2005 ◽  
Author(s):  
Xavier Canalias ◽  
Frank Ko¨pf ◽  
Peter Sahm
Keyword(s):  
Engine Performance ◽  
Good Choice ◽  
Jet Engines ◽  
Analysis Method ◽  
Jet Engine ◽  
Analysis By Synthesis ◽  
Physically Based ◽  
The Common ◽  
Set Up ◽  
Definition Of

The ANalysis by SYNthesis (ANSYN) technique is a standard performance analysis method widely used in industry. It is currently used to evaluate engine performance from tests and to derive correcting factors (i.e. ANSYN factors) that modify certain parameters of the components’ characteristics in order to allow the reproduction of real engine behaviour by means of a synthesis calculation. Once they have been determined, these ANSYN factors can be implemented into the common synthesis tools to accurately predict the behaviour of the engine in non-tested conditions. The definition of how the engine synthesis will be modified to reproduce the measured cycle is called the “matching scheme”. It is the choice of the parameters that will not be modified, the ones that will be scaled and the ones that will be used as a reference. A good choice of the ANSYN factors and of the magnitudes used to tabulate them will make possible their physical interpretation and analysis, which can lead to the identification of inaccurate assumptions and phenomena that had not been accurately taken into account in the previous performance models. This understanding of their origin is a prerequisite for any improvement of the models and could lead to an enhanced process for the development of non-dimensional jet engine characteristics. In this study, a matching scheme for the RR-BR710 engine has been set up, implemented and applied. This method is an alternative to some of the currently used in the Rolls-Royce Deutschland ANSYN tools. The main goal of the present work has been the definition of a matching scheme that leads to the obtention of physically meaningful ANSYN factors. This allows their subsequent analysis and interpretation and can provide useful information on engine component operation and on the phenomena responsible for the observed deviations from predicted engine behaviour.

Effect of Manufacturing Variability on Turbine Engine Performance: A Probabilistic Study

2013 ◽  
Author(s):  
Michael Gorelik ◽  
Jacob Obayomi ◽  
Jack Slovisky ◽  
Dan Frias ◽  
Howie Swanson ◽  
...  
Keyword(s):  
Performance Metrics ◽  
Model Development ◽  
Engine Performance ◽  
Performance Model ◽  
Probabilistic Methods ◽  
System Level ◽  
Turbine Engine ◽  
Probabilistic Study ◽  
Blade Geometry

While turbine engine Original Equipment Manufacturers (OEMs) accumulated significant experience in the application of probabilistic methods (PM) and uncertainty quantification (UQ) methods to specific technical disciplines and engine components, experience with system-level PM applications has been limited. To demonstrate the feasibility and benefits of an integrated PM-based system, a numerical case study has been developed around the Honeywell turbine engine application. The case study uses experimental observations of engine performance such as horsepower and fuel flow from a population of engines. Due to manufacturing variability, there are unit-to-unit and supplier-to-supplier variations in compressor blade geometry. Blade inspection data are available for the characterization of these geometric variations, and CFD analysis can be linked to the engine performance model, so that the effect of blade geometry variation on system-level performance characteristics can be quantified. Other elements of the case study included the use of engine performance and blade geometry data to perform Bayesian updating of the model inputs, such as efficiency adders and turbine tip clearances. A probabilistic engine performance model was developed, system-level sensitivity analysis performed, and the predicted distribution of engine performance metrics was calibrated against the observed distributions. This paper describes the model development approach and key simulation results. The benefits of using PM and UQ methods in the system-level framework are discussed. This case study was developed under Defense Advanced Research Projects Agency (DARPA) funding which is gratefully acknowledged.

Sign in / Sign up

Export Citation Format

Share Document