scholarly journals Machine Learning Based Sensitivity Analysis of Aeroelastic Stability Parameters in a Compressor Cascade

2021 ◽  
Vol 6 (3) ◽  
pp. 39
Author(s):  
Marco Rauseo ◽  
Mehdi Vahdati ◽  
Fanzhou Zhao
Keyword(s):  
Sensitivity Analysis ◽  
Surrogate Model ◽  
Structural Parameters ◽  
Operating Conditions ◽  
Data Driven ◽  
Compressor Cascade ◽  
Stability Parameters ◽  
Aerodynamic Damping ◽  
Data Driven Approach ◽  
Input Variables

Aeroelastic instabilities such as flutter have a crucial role in limiting the operating range and reliability of turbomachinery. This paper offers an alternative approach to aeroelastic analysis, where the sensitivity of aerodynamic damping with respect to main flow and structural parameters is quantified through a surrogate-model-based investigation. The parameters are chosen based on previous studies and are represented by a uniform distribution within applicable intervals. The surrogate model is an artificial neural network, trained and tested to achieve an error within 1% of the test data. The quantity of interest is aerodynamic damping and the datasets are obtained from a linearised aeroelastic solver. The sensitivity of aerodynamic damping with respect to the input variables is obtained by calculating normalised gradients from the surrogate model at specific operating conditions. The results show a quantitative comparison of sensitivity across the different input parameters. The outcome of the sensitivity analysis is then used to decide the most appropriate action to take in order to induce stability in unstable operating conditions. The work is a preliminary study, carried out on a simplified two dimensional compressor cascade and it is aimed at proving the validity of a data-driven approach in studying the aeroelastic behaviour of turbomachinery. To the best of the authors’ knowledge, this is the first time a data-driven flutter model has been investigated. The initial results are encouraging, indicating that this approach is worth pursuing in the future. The presented framework can be used as a redesign tool to enhance the flutter stability of an existing blade.

Analytical Maps of Aerodynamic Damping as a Function of Operating Condition for a Compressor Profile

2006 ◽  
Author(s):  
Paul J. Petrie-Repar ◽  
Andrew McGhee ◽  
Peter A. Jacobs ◽  
Rowan Gollan
Keyword(s):  
Inviscid Flow ◽  
Operating Conditions ◽  
Navier Stokes ◽  
Compressor Cascade ◽  
Operating Condition ◽  
Flow Solver ◽  
Aerodynamic Damping ◽  
Wide Range ◽  
Contour Plots ◽  
Standard Configuration

In this paper, analytical maps of aerodynamic damping for a two-dimensional compressor cascade (Standard Configuration 10) are presented. The maps are shown as contour plots of the aerodynamic damping as a function of operating condition. The aerodynamic dampings were calculated by a linearized Navier-Stokes flow solver. The flutter boundaries over a wide range of operating conditions are clearly shown on the damping maps and were found to be strongly dependent on the mode frequency and the mode shape. Extremely low values of negative aerodynamic damping were predicted for some off-design operating conditions where flow separation occurred. A damping map was also constructed based on inviscid flow simulations. There were differences in the viscous and inviscid flutter boundaries particularly at off-design inflow angles. The extremely low values of negative aerodynamic damping were only predicted by the viscous simulations and not the inviscid simulations.

Learning from one’s errors: A data-driven approach for mimicking an ensemble of hydrological model residuals

2021 ◽  
Author(s):  
John M. Quilty ◽  
Anna E. Sikorska-Senoner
Keyword(s):  
Hydrological Model ◽  
Model Performance ◽  
Data Driven ◽  
Gradient Boosting ◽  
Hydrological Models ◽  
Model Errors ◽  
Water Resources Systems ◽  
Extreme Gradient Boosting ◽  
Data Driven Approach ◽  
Input Variables

<p>Despite significant efforts to improve the calibration of hydrological models, when applied to real-world case studies, model errors (residuals) remain. These residuals impair flow estimates and can lead to unreliable design, management, and operation of water resources systems. Since these residuals are auto-correlated, they should be treated with appropriate methods that do not require limiting assumptions (e.g., that the residuals follow a Gaussian distribution).</p><p>This study introduces a novel data-driven framework to account for residuals of hydrological models. Our framework relies on a conceptual-data-driven approach (CDDA) that integrates two models, i.e., a hydrological model (HM) with a data-driven (i.e., machine learning) model (DDM), to simulate an ensemble of residuals from the HM. In the first part of the CDDA, a HM is used to generate an ensemble of streamflow simulations for different parameter sets. Afterwards, residuals associated with each simulation are computed and a DDM developed to predict the residuals. Finally, the original streamflow simulations are coupled with the DDM predictions to produce the CDDA output, an improved ensemble of streamflow simulations. The proposed CDDA is a useful approach since it respects hydrological processes via the HM and it profits from the DDM’s ability to simulate the complex (nonlinear) relationship between residuals and input variables.</p><p>To explore the utility of CDDA, we focus principally on identifying the best DDM and input variables to mimic HM residuals. For this purpose, we have explored eight different DDM variants and multiple input variables (observed precipitation, air temperature, and streamflow) at different lag times prior to the simulation day. Based on a case study involving three Swiss catchments, the proposed CDDA framework is shown to be very promising at improving ensemble streamflow simulations, reducing the mean continuous ranked probability score by 16-29 % when compared to the standalone HM. It was found that eXtreme Gradient Boosting (XGB) and Random Forests (RF), each using 29 input variables, were the strongest predictors of the HM residuals. However, similar performance could be achieved by selecting only the six most important (of the original 29) input variables and re-training the XGB and RF models.</p><p>Additional experimentation shows that by converting CDDA to a stochastic framework (i.e., to account for important uncertainty sources), significant gains in model performance can be achieved.</p>

scholarly journals Data-Driven Prediction of Vessel Propulsion Power Using Support Vector Regression with Onboard Measurement and Ocean Data

Sensors ◽  
2020 ◽  
Vol 20 (6) ◽  
pp. 1588 ◽  
Author(s):  
Donghyun Kim ◽  
Sangbong Lee ◽  
Jihwan Lee
Keyword(s):  
Big Data ◽  
Support Vector Regression ◽  
Operating Conditions ◽  
Data Driven ◽  
Superior Performance ◽  
Support Vector ◽  
Power Prediction ◽  
Static Water ◽  
Data Driven Approach ◽  
Propulsion Power

The fluctuation of the oil price and the growing requirement to reduce greenhouse gas emissions have forced ship builders and shipping companies to improve the energy efficiency of the vessels. The accurate prediction of the required propulsion power at various operating condition is essential to evaluate the energy-saving potential of a vessel. Currently, a new ship is expected to use the ISO15016 method in estimating added resistance induced by external environmental factors in power prediction. However, since ISO15016 usually assumes static water conditions, it may result in low accuracy when it is applied to various operating conditions. Moreover, it is time consuming to apply the ISO15016 method because it is computationally expensive and requires many input data. To overcome this limitation, we propose a data-driven approach to predict the propulsion power of a vessel. In this study, support vector regression (SVR) is used to learn from big data obtained from onboard measurement and the National Oceanic and Atmospheric Administration (NOAA) database. As a result, we show that our data-driven approach shows superior performance compared to the ISO15016 method if the big data of the solid line are secured.

Application of Orthogonal Experimental Design on Reliability and Sensitivity Analysis

2011 ◽  
Vol 211-212 ◽  
pp. 651-655 ◽  
Author(s):  
Qin Shu He ◽  
Xi Nen Liu ◽  
Shi Fu Xiao
Keyword(s):  
Sensitivity Analysis ◽  
Experimental Design ◽  
Structural Parameters ◽  
Orthogonal Experimental Design ◽  
Random Input ◽  
Range Analysis ◽  
First Order ◽  
Reliability Method ◽  
Structural Responses ◽  
Input Variables

In the present paper, the effects of four structural parameters at three levels on the reliability and sensitivity of structure are investigated. Sensitivity of parameters is achieved by the range analysis and the significance of parameters is achieved by the variance analysis. A response surface based on orthogonal experimental design and finite element calculations is elaborated so that the relation between the random input variables and structural responses could be established. The First-Order Reliability Method (FORM) as an approximated method is used here to assess the reliability. Comparing with the results of Monte Carlo simulations by ANSYS for a numerical example, the effect of sensitivity analysis has been proved, while the precision of the reliability and sensitivity should be improved in the future.

Comparison of Theoretical and Experimental Data for an Oscillating Transonic Compressor Cascade

1999 ◽  
Author(s):  
Volker Carstens ◽  
Stefan Schmitt
Keyword(s):  
Experimental Data ◽  
Transonic Flow ◽  
Operating Conditions ◽  
Experimental Investigations ◽  
Test Case ◽  
Flow Conditions ◽  
Compressor Cascade ◽  
Blade Passage ◽  
Aerodynamic Damping ◽  
Transonic Compressor

Numerical and experimental results are compared for a compressor cascade performing harmonic oscillations in transonic flow. The flow field was calculated by a Q3D Navier Stokes code, the basic features of which are the use of an upwind flux difference scheme for the convective terms, the implementation of an effective one-equation turbulence model and the use of deforming multi-block grids. The experimental investigations were performed in an annular cascade windtunnel where unsteady blade pressures were measured for two different operating conditions of the cascade. The present data were all obtained for tuned torsional modes where the blades performed pitching oscillations with the same frequency and amplitude, but with a constant interblade phase angle. In the first test case the steady flow around the blades was purely subsonic. For the second test case the compressor cascade was run under transonic flow conditions where a normal shock in the front part of the blades’ suction side is followed by a blade passage shock. It becomes apparent that under subsonic flow conditions the predicted aerodynamic damping coefficients are in resonable agreement with the experimental data, although the numerical pressure amplitudes are much higher than the measured ones. In transonic flow significant discrepancies between computed and experimentally determined pressure amplitudes are observed, whereas the accuracy of the pressure phase prediction is comparable to the subsonic test case. Another important result of these investigations is that oscillations of the blade passage shock lead to strong variations of the local aerodynamic damping of the blades, but do not significantly change the global damping coefficient of the tested compressor cascade.

scholarly journals A Comparison Study of Data-Driven Anomaly Detection Approaches for Industrial Chillers

2021 ◽  
Vol 1 ◽  
Author(s):  
Constantin Falk ◽  
Ron Van de Sand ◽  
Sandra Corasaniti ◽  
Jörg Reiff-Stephan
Keyword(s):  
Operating Conditions ◽  
Data Driven ◽  
Sensor Data ◽  
Comparison Study ◽  
True Negative ◽  
The Past ◽  
Use Of Data ◽  
Data Driven Approach ◽  
System States ◽  
System Operating

Faults in industrial chiller systems can lead to higher energy consumption, increasing wear of system components and shorten equipment life. While they gradually cause anomalous system operating conditions, modern automatic fault detection models aim to detect them at low severity by using real-time sensor data. Many scientific contributions addressed this topic in the past and presented data-driven approaches to detect faulty system states. Although many promising results were presented to date, there is lack of suitable comparison studies that show the effectiveness of the proposed models by use of data stemming from different chiller systems. Therefore this study aims at detecting a suitable data-driven approach to detect faults reliable in different domains of industrial chillers. Thus, a unified procedure is developed, to train all algorithms in an identical way with same data-basis. Since most of the reviewed papers used only one dataset for training and testing, the selected approaches are trained and validated on two different datasets from real refrigeration systems. The data-driven approaches are evaluated based on their accuracy and true negative rate, from which the most suitable approach is derived as a conclusion.

Extended-Range Prediction with Low-Dimensional, Stochastic-Dynamic Models: A Data-driven Approach

2012 ◽  
Author(s):  
Michael Ghil ◽  
Mickael D. Chekroun ◽  
Dmitri Kondrashov ◽  
Michael K. Tippett ◽  
Andrew Robertson ◽  
...  
Keyword(s):  
Dynamic Models ◽  
Data Driven ◽  
Stochastic Dynamic ◽  
Extended Range ◽  
Data Driven Approach ◽  
Low Dimensional
Sign in / Sign up

Export Citation Format

Share Document