本研究主要在提升PIV量測應用於背向階梯流場(backward- facing step)的可靠度。由於PIV量測需要在流場中加入追蹤粒子來反映流體運動，因此原始流體的運動是否會因為粒子的加入而改變，是PIV量測首先必須面對的問題。
本實驗的流場中包含許多複雜的流體運動，尤其是存在著迴流區及剪力流層等速度梯度大的區域。在這些速度變化急遽的地方，追蹤粒子很容易會因為慣性力脫離原本的流體運動，或是堆積在渦流內發生碰撞，造成追蹤粒子與流場運動不一致。此外，追蹤粒子在空間分佈上變的很不均勻，這些現象都會造成PIV量測上嚴重的誤差。因此，在不改變原始流場特性及現有設備的情形下，本研究針對在迴流區下游的截面，以影像後處理的方式來克服上述這些問題。除了影像後處理外，另外還搭配田口式參數分析來找出最佳的參數組合。最後以熱線測速儀在不同雷諾數下( Reθ = 505 及808 )不同截面( x = 9.87 h及12.92 h) 的結果作為比較的基準，判斷估算出來的參數組合是否確實為最佳參數組合，並具有一致性。
實驗結果顯示，影像後處理與田口式參數分析確實能夠有效改善PIV量測的結果，但是所求出的最佳參數組合在不同的雷諾數下缺乏一致性。此外，對不同的速度特性也有不一樣的表現。對於這樣的結果，本研究推測可能是因為所選擇的參數之間交互效應太強，亦或者是對此流場最關鍵的參數並沒有被考慮到。
另外，在重新修正參數水準及選取有效數據的範圍後，Reθ = 505的最佳參數組合表現仍不如預期，但是在Reθ = 808的實驗條件下，最佳化後的結果確實比原始實驗設定的組合要好。兩者表現不一的原因是因為初始動量厚度的增加，使得Reθ = 808在相同位置較接近完全發展紊流，而Reθ = 505則因為仍受到大尺度渦流的影響，造成追蹤粒子分佈不均，進而影響數據的可靠度。此結果同時也說明了慎選可靠數據的範圍能有效提升田口氏最佳參數分析的正確性。

The object of this research is to promote the reliability of PIV measurements when applied to backward-facing step flow field. The velocities of fluid are determined by the motion of the tracer particles which are introduced into the flow. However, do the tracer particles influence the original flow field or follow the fluid motion faithfully? This seeding problem is the major issue to be faced in the PIV measurements.
There are many complicated fluid motions exhibited in the backward-facing step flow field, especially in the large velocity gradient regions such as the recirculation zone and the shear layer. In these regions, parts of the tracer particles cannot follow the fluid motion exactly due to their inertial forces or collide with each other in the vortex. The distribution of the tracer particles is non-uniform in some regions of the flow field. This phenomenon sometimes results in serious errors of PIV measurements. This research aims at solving above mentioned problems with the aides of image post-processing and Taguchi method. The data measured by a two-component hot-wire anemometer are used as the comparison basis. Two cases with Reynolds numbers of Reθ = 505 and 808 are conducted in the study.
The results show that the image post-processing and Taguchi method are two effective means to improve PIV measurements. However the optimal sets behave non-identically between the two different Reynolds numbers and different velocity properties ( e.g., mean or fluctuation values). The strong interactions between chosen parameters or the miss of key parameters in the study might cause this problem.
After adjusting the parameters and re-defining the range of valid data, the optimal set of Reθ = 505 still behaves worse than the original sets, whereas the other optimal set ( Reθ = 808 ) behaves properly. The different performances between them might be attributed to different initial momentum thicknesses formed in these two cases. The flow field reaches its fully states sooner due to thicker momentum thickness ( Reθ = 808 ) while the thinner case is still influenced by the large scale vortex, which in turn result in the non-uniform distribution of tracer particles. It also demonstrates that the necessity of the care definition of valid data range in the use of Taguchi method.

[1] V. de Brederode, P. Bradshaw, 1972, “Three-
dimensional flow in nominally two-dimensional
separation bubbles”, Imperial College of
Science and Technology, Technical Aero Report No. 72-
19.

[2] F. Scarano, and M. L. Riethmuller,1999, “Iterative
multigrid approach in PIV image processing with
discrete window offset,” Experiments in fluids,
Vol.26, pp.513–523.

[3] B. F. Armaly, F. Durst, J. C. F. Pereira, B.
Schönung , 1983,“Experimental and theoretical
investigation of backward-facing step flow”, Journal
of Fluid Mechanics, Vol. 127, pp. 473-496.

[4] 黃興閎，2004，背向階梯流場之剪流層非穩態特性之研究，清
華大學碩士論文

[5] David M. Driver, H. Lee Seegmiller, Joe G. Marvin,
1987, “Time-dependent behavior of a reattaching shear
layer”, AIAA, Vol.25, No.7, pp.914-919.

[6] J.K.Eaton, J.P.Johnston,1981 ,Sep., “A review of
research on subsonic turbulent flow reattachment”,
AIAA, Vol.19, No.9, pp.1093-1100.

[7] D. M. Kuehn, 1980, March, ”Some effects of adverse
pressure gradient on the incompressible reattaching
flow over a rearward -facing step”, AIAA, Vol.18,
pp.343-344.

[8] E.W. Adams, J.P. Johnston, 1988, ”Effects of the
separating shear layer on the reattachment flow
structure Part 1: Pressure and turbulence
quantities”, Experiments in fluids, Vol.6, pp.493-499.

[9] S. Pronchick, S. Kline, 1983, “An experimental
investigation of the structure of a turbulent
reattaching flow behind a backward-facing step”,
Stanford University, Mechanical Engineering Dept.,
Rept, MD-42.

[10] R. L. Simpson, 1989, “Turbulent boundary layer
separation”, Annual review of fluid mechanics,
Vol.21, pp.205-324.

[11] J. Westerweel, 1997, “Fundamentals of digital
particle image velocimetry”, Measurements science
and technology, Vol. 8, pp.1379-1392.

[12] Clayton Crowe, Martin Sommerfeld, Yutaka Tsuji, 1998,
“Multiphase flows with droplets and particles”, CRC
Press.[13] R. D. Keane, R. J. Adrian, 1990,
“Optimization of particle image
velocimeters. Part I: Double pulsed systems”,
Measurements science and technology, Vol.1, pp.1202-
1215.

[14] R.D. Keane, R. J. Adrian, 1992, “Theory of cross-
correlation analysis of PIV images”, Applied
scientific research, Vol.49, pp.191-215.

[15] J. Westerweel, D. Dabiri, M. Gharib, 1997, ” The
effect of a discrete window offset on the accuracy of
cross-correlation analysis of digital PIV
recordings”, Experiments in Fluids, Vol. 23, pp.20-
28.

[16] P. Bradshaw, F.Y.F. Wong, 1972, ”The reattachment
and relaxation of a turbulent shear layer”, Journal
of fluid mechanics., Vol.52, part1, pp.113-135.

[17] Integrated Design Tools, Inc., 2006 Oct., “MotionPro
X Timing Hub:Cross-platform user manual:
Synchronization unit”.

[18] C. Kähler, B. Sammler, J. Kompenhans, 2002,
“Generation and control of tracer particles for
optical flow investigation in air”,Experiment in
Fluids, Vol. 33, No. 6, pp.736-742.

[19] Integrated Design Tools, Inc., 2006 Oct.,
“MotionPro X Camera:cross-platform user manual high-
speed CMOS digital camera”.

[20] Chiuan-Ting Li, Keh-Chin Chang, Muh-Rong Wang, 23-25
Sept. 2007, “Parametric study on PIV measurement of
turbulent flow in planar mixing layer”, 2nd
International Symposium on Advanced
Fluid/Solid Science and Technology in Experimental
Mechanics, Osaka, Japan.

[21] Integrated Design Tools, Inc., 2007 March.,
“proVISION-XS: User manual-Particle image
velocimetry”.

[22] 井上成喜、八木伸行等人著，吳上立、林宏墩譯，C語言數位
影像處理，全華科技圖書股份有限公司，2006年3月

[23] R. J. Adrian, 1 November 1986, ”An image shifting
technique to resolve directional ambiguity in double-
pulsed velocimetry” ,Vol.25, No.21, Applied optics.

[24] M. Raffel, C. Willert, J. Kompenhans, 1998,
“Particle image velocimetry”, Springer.

[25] H. Nobach, C. Tropea, 2005, “Improvements to PIV
image analysis by Recognizing the velocity
gradient”, Experiments in fluids, Vol. 39, pp.612-
620.