Research on advanced PIV algorithms and their aplication in measuring complex flow fields

摘要

This repot focuses on the development and validation of a bootstrap filter based particle image tracking algorithm, an intelligent PIV image processing platform and highefficient TR-PIV algorithms.A detailed analysis of the mainstream particle imageinterrogation algorithms are firstly carried out by summarising their principles, procedures and pointing out their advantages and disadvantages.One of the commonproblems of these methods is the deployment of the constant velocity assumption forparticle motion between two image frames, which will result in serious errors whenapplied in high velocity gradient flows.Therefore, a simple nonlinear dynamic modelwhich takes particle acceleration into account is proposed and a sequential Monte Carlomethod-bootstrap filter is employed to strengthen the particle image tracking performance.This new method is validated by using numerically generated and real PIVimages.It is proved that the Bootstrap Filter Tracking method is very reliable and robust for processing high velocity gradient, low particle density flows, and it can yieldup to 0.8 times more valid particle image pairs than the Super-PIV and Kalman FilterTracking methods.
　　 The second work of the report is the development of a software package which encapsulates six widely used interrogation algorithms.The influence of particle imagedensity on the measurement accuracy of the pattern match schemes are evaluated withartificial particle images.The measurement accuracy of the six algorithms are thenstudied by using particle images generated from synthetic Rankine vortex flow as wellas real vortex and turbulent jet flows.Finally, an automated particle image processingprocedure is proposed and verified by the artificial and real particle images.
　　 Aiming to reduce the computational cost of time-series TR-PIV measurements, thethird part of this report presents improvements made for two widely used PIV algorithms, namely the multi-grid and iterative image deformation cross correlation.Thehistorical displacement field and its variation were employed to determine the windowoffset and image deformation calculation in the above-mentioned algorithms, respectively.Performance of the improved algorithms was extensively evaluated by usingsynthetic images of artificial Rankine vortex flow and wake flow behind square cylinderdetermined by CFD as well as real time-series TR-PIV measurements.The comparison between improved TR-PIV methods and traditional ones show that the improvedalgorithms can save up to 50％ computational time while keep measurement accuracysame as the traditional iterative PIV algorithms.By taking displacement variation intoaccount, the improved methods can successfully handle unsteady flows where localdisplacements vary more than 20％ between the image frames.
　　 The last section of this report applies advanced PIV algorithms to measure fluid flowaround a bio-inspired airfoil with corrugated surfaces and its smooth counterpart atchord Reynolds number Re =2000.The global characteristics of the fluid flow aroundtwo airfoils were analyzed by ensemble-averaged velocity field, distribution of reverseflow intermittency, and time-series flow visualizations.Through PIV experimentalstudies, it was found that at AOA =0~, no significant variation of the global flow patterns was recognized for both configurations.The statistical results of reverse flowintermittency results demonstrated that the protruding peaks of the corrugated airfoildelay flow separation atAOA =40.At large AOAs (80 and 120), however, the flow wasmassively separated in both configurations; the combination of large separation bubbleabove the corrugated airfoil and small recirculation zones in the upstream upper valleyresults in earlier separation of the flow.

目录

Abstract

Contents

Acknowledgement

1 Improve Measurement Accuracy for High Gradient Flows

1．1 Review of PIV Algorithms

1．2 Review of PTV Algorithms

1．3 The Bootstrap Filter Tracking Method

1．3．1 Dynamic System Model

1．3．2 Outline of the Bootstrap Filter Tracker

1．4 Simulation Study

1．4．1 Artificial Image

1．4．2 Effect of Particle Image Density

1．4．3 Effect of Out-of-Plane Motion

1．4．4 Computational Time

1．5 Real Image Test

2 Development of an Intelligent PIV Image Processing Package-SmartPIV

2．1 Influence of the Particle Image Density

2．1．1 Influence of Particle Image Density on Performance of the Pattern Match Algorithms

2．1．2 Influence of Particle Image Density on Particle Image Identification

2．2 Comparative Study on the Six Algorithms

2．2．1 Data Preparation

2．2．2 Results

2．3 The SmartPIV Image Processing Method

2．3．1 Principle

2．3．2 Application

2．4 Conclusion

3 Improve Effciency for TR-PIV Algorithms by Using Historical Displacement Information

3．1 Introduction

3．2 Concept of the improved TR-PIV algorithms

3．3 Test of the improved algorithms using synthetic TR-PIV images

3．3．1 Rankine vortex flow

3．3．2 Wake behind a square cylinder(CFD)

3．4 Test of the improved algorithms using experimental data

3．5 Conclusion

4 Case Study-PIV Measurement of Flow Around a Two-dimensional Corrugated Airfoil

4．1 Experimental setup and procedures

4．2 Experimental results and discussions

4．2．1 AOA=00

4．2．2 AOA=40

4．2．3 AOA=80

4．2．4 AOA=120

4．3 Conclusion

5 Conclusions and Future Work

5．1 Conclusions

5．2 Future Work

Appendix A Publications

Appendix B Research Funds

Bibliography