Bulletin of the American Physical Society
64th Annual Meeting of the APS Division of Fluid Dynamics
Volume 56, Number 18
Sunday–Tuesday, November 20–22, 2011; Baltimore, Maryland
Session H26: Focus Session: PIV Uncertainty |
Hide Abstracts |
Chair: Pavlos P. Vlachos, Virginia Tech Room: 329 |
Monday, November 21, 2011 10:30AM - 10:43AM |
H26.00001: Uncertainty of Spatial and Temporal velocity for High-speed PIV Koji Okamoto High-speed PIV system can provide huge amount of images for PIV analysis. With frame straddling technique, the dynamic range of velocity could be improved with four image analysis technique. This technique sacrifices the temporal directional resolution to improve the dynamic ranges. Because total amount of the image information is constant, if we improve a spatial resolution, the temporal resolutions will decrease. There should be a sufficient conversion of spatial and temporal resolutions. The relation between sampling frequency and flow characteristic frequency in both spatial and temporal direction will be discussed. [Preview Abstract] |
Monday, November 21, 2011 10:43AM - 10:56AM |
H26.00002: Random uncertainty estimates of PIV measurements using correlation statistics Steve Wereley In recent years researchers have made efforts to extract scalar and even tensor quantities from the characteristics of the correlation peak. These have included temperature (Wereley, 2002, 2010), velocity distribution (Wereley 2006, Westerweel $\sim $2006), particle hydrodynamic size (Wereley 2008) and even Reynolds stresses (Kaehler, $\sim $2008). It is also possible to extract random uncertainty estimates from the correlation peak characteristics---provided enough is known about the flow. Obtaining uncertainty estimates in this way would provide local estimates of uncertainty rather than the current global rules of thumb that are relied upon. [Preview Abstract] |
Monday, November 21, 2011 10:56AM - 11:09AM |
H26.00003: Estimation of Uncertainty Bounds for Individual PIV Measurements John Charonko, Pavlos Vlachos Numerous studies have established firmly that particle image velocimetry (PIV) is a robust method for non-invasive, quantitative measurements of fluid velocity, and that when carefully conducted, typical measurements can accurately detect displacements in digital images to within a tenth of a pixel or less. However, previously, these estimates have only been able to provide guidance on the expected error for an average measurement under specific image quality and flow conditions. This work demonstrates a new method for estimating the uncertainty bounds to within a given confidence interval for a specific, individual measurement. We show that for a phase-only, generalized cross-correlation the ratio of primary to secondary peak heights correlated strongly with the range of observed error values for each individual displacement measurement, regardless of flow condition or image quality. Using an analytical model of the relationship derived from synthetic data sets, the uncertainty bounds at a 95\% confidence interval are then computed for several artificial and experimental flow fields, and the true errors are shown to match the predicted uncertainties. While this method is not able to predict the true error for a given measurement, knowledge of the uncertainty level for a PIV experiment will provide great benefits in engineering design studies and CFD validation efforts. [Preview Abstract] |
Monday, November 21, 2011 11:09AM - 11:22AM |
H26.00004: Resolution limit of Digital Particle Image and Particle Tracking Velocimetry Christian J. Kaehler, Sven Scharnowski, Christian Cierpka This work analyzes the spatial resolution that can be achieved by DPIV in dependency on the tracer particles and the imaging and recording system. While the in-plane resolution of DPIV is given by the interrogation window size, ensemble-correlation seams to increase the achievable resolution up to a single pixel. However, it is shown that the resolution limit of single-pixel ensemble-correlation is determined by the particle image size which is dependent on the diameter of the particles, the magnification, the f--number of the imaging system, and optical aberrations. As the minimum detectable particle image size is given by the pixel dimension of the camera sensor in DPIV, this quantity is also important for a systematic analysis of the resolution limit. It is shown that the optimal magnification that results in the best possible spatial resolution can be estimated from the particle size, the lens properties, and the pixel size of the camera. Thus, the provided information in this paper allows to optimize the camera and objective lens choices as well as the working distance for a given setup. Furthermore, it is shown that the resolution limit of DPIV can be resolved by using PTV evaluation techniques. [Preview Abstract] |
Monday, November 21, 2011 11:22AM - 11:35AM |
H26.00005: Characteristics of seeding particles for PIV/PTV analysis Tal Hadad, Alexander Liberzon, Anne Bernhaim, Roi Gurka PIV and PTV are non-intrusive state-of-the-art techniques widely used for flow measurements. Seeding particles are required to be used as tracers to the flow. The accuracy of the velocity measurements is limited by the ability of the tracer particles to adequately follow the instantaneous motion of the continuous phase. In order to follow the flow effectively, the particles should satisfy numerous requirements: size, sphericity, density, high refractive index, concentration and chemical inert. Since seeding particles for liquids are commonly polymer-based particles we probe the influence of their surface coating on the results obtained from optical measurements. Using a canonical lid-driven cavity flow we measured the velocity field using PIV and PTV and compared the results (velocity and acceleration) obtained with the same particles with and without chemical treatment of surfactants. Probability density functions of the results using particles before and after treatment are compared statistically utilizing the two-sample Kolmogorov-Smirnov tests. Although the mean values exhibit similar trends, fluctuations and velocity derivatives show some discrepancy in respect to the chemical treatment. The obtained results show a variance of up to 5{\%} between the values obtained for using washed and un-washed particles, for both PIV and PTV experiments with some influence related to the size of the particles. [Preview Abstract] |
Monday, November 21, 2011 11:35AM - 11:48AM |
H26.00006: Practical estimation of DPIV uncertainty using pseudo-image pairs Michael McPhail, Matthew Weldon, Michael Krane, Arnold Fontaine, Howard Petrie, John Buchanan, Donald Lorentz, Richard Bauer While the sources of uncertainty in PIV measurement have long been understood, quantifying the contribution of each source has proved elusive. Here we present a straightforward approach using image pairs formed from a real image and that same image, subject to a uniform non-integer displacement. This procedure thus naturally incorporates contributions of image quality, optical setup, and pixel resolution to uncertainty. The pseudo-image pairs for a range of displacements are used to estimate an ensemble of displacement vectors, which is then used to estimate the bias and random errors in displacement. This method is particularly useful for identifying pixel locking bias. Applications of this approach to both fully-developed turbulent pipe flow and impinging jet flow will be presented. [Preview Abstract] |
Monday, November 21, 2011 11:48AM - 12:01PM |
H26.00007: Uncertainty in Velocity Fluctuations for Two-Component PIV Measurements Brandon Wilson, Jeff Harris, Barton Smith The subpixel displacement estimator in PIV generates random fluctuations, in the velocity measurement. This noise causes velocity fluctuations to be overestimated. We quantify this overestimate of the fluctuation level due to four error sources: particle displacement, particle image size, particle density and flow gradients. A jet experiment that is designed to segregate error sources and demonstrate their effects by comparison of PIV and hot wire measurements is used. While the influence of the error sources studied on the mean velocity is negligible, increases in velocity fluctuation levels are observed for all error sources when compared to hot wire results, particularly particle displacement and flow gradients. Uncertainties in the velocity fluctuation levels are one-directional because the error sources considered can only increase the measured fluctuations. Sources that attenuate the fluctuation level (i.e. particle lag or volumetric averaging) are not studied. The amount by which the velocity fluctuation level is overestimated is consistent with previous theoretical studies and with recent studies using synthetic images. [Preview Abstract] |
Monday, November 21, 2011 12:01PM - 12:14PM |
H26.00008: Evaluation of two-component PIV uncertainty for flow in porous media Vishal Patil, James Liburdy The measurement of flow in porous media is challenging due to accessibility and large range of flow passage scales. The use of PIV requires matching of refractive indices of fluid and solid phases. Slight mismatches are shown to cause significant tracking errors. In gaining PIV data at discrete planar locations along the optical axis, variations occur in the imaging magnification, and if not taken into consideration may lead to increased error. This paper addresses three forms of error in PIV measurements as they pertain to porous media flow: tracking error, bias error due to displacement gradients and perspective error. As applied to a porous bed of spherical beads, the local magnification is evaluated based on measured variations of detected bead diameter along the optical axis. The variation of magnification through the bed is then used to evaluate perspective error. The error due to displacement gradients was evaluated from correlation peak width. The bias error was also evaluated by reducing the interrogation window size and estimating the RMS difference between the two velocity estimates. The bias error evaluated using these two methods compared well. Results are shown for a flow at a pore Reynolds number of 4. Perspective errors are shown to be most significant, with total errors up to 6{\%}. [Preview Abstract] |
Monday, November 21, 2011 12:14PM - 12:27PM |
H26.00009: PIV estimates of dissipation: their accuracy and uncertainty Edwin Cowen Particle image velocimetry (PIV) allows the determination of the rate of dissipation of turbulent kinetic energy, $\epsilon$, over a two-dimensional region in space. As $\epsilon$ ranges over orders-of-magnitude its uncertainty is relatively high and we often consider a factor of 2 to be acceptable. Researchers have approached the determination of $\epsilon$ in many ways, including spectra-based estimates, structure function-based estimates, and the direct calculation from the available fluctuating gradients. In this presentation the accuracy and uncertainty of each of these approaches is reviewed and optimal methods for the accurate determination of $\epsilon$ with the narrowest uncertainty bounds are presented. [Preview Abstract] |
Monday, November 21, 2011 12:27PM - 12:40PM |
H26.00010: When RANS Goes Wrong: Using PIV to Assess Ergodicity W. Ethan Eagle Ergodic flow, where the spatial, temporal and ensemble averages, $\bar{u}$$_V$, $\bar{u}$$_T$ , $\bar{u}$$_N$ are taken to be equal, is a widely used assumption in turbulent flows and is useful in applications from Taylor's frozen turbulence approximation to the Reynolds decomposition $u$ = $\bar{u}$ + u$'$. Interestingly, ergodicity is only routinely (albeit tediously) verified by CFD grid convergence studies, while experimental assessments of ergodicity that require different spatial averages $\bar{u}$$_V$ (an impossibility for a given probe dimension), are not undertaken. In many cases, the unverified existence of experimental ergodicity results in spurious comparisons between computed and measured mean quantities. To investigate flow ergodicity, mean flow quantities were computed from 5 sets of 100 Stereo-PIV images of a supersonic turbulent boundary layer. To evaluate a pseudo-temporal ergodicity, $\bar{u}$$_t$, each set was recorded at a different laser pulse timing, $d$$_t$=\{500ns-900ns\} Spatial ergodicity, $\bar{u}$$_V$, was assessed using 4 interrogation window cell sizes \{2$^n$x2$^n$ n=3,4,5,6.\} Results suggest measuring, verifying, and reporting ergodicity enhances understanding of PIV optimization and uncertainty quantification programs and is essential when PIV experiments are compared to RANS (or LES) in validation and verification exercises. [Preview Abstract] |
Follow Us |
Engage
Become an APS Member |
My APS
Renew Membership |
Information for |
About APSThe American Physical Society (APS) is a non-profit membership organization working to advance the knowledge of physics. |
© 2024 American Physical Society
| All rights reserved | Terms of Use
| Contact Us
Headquarters
1 Physics Ellipse, College Park, MD 20740-3844
(301) 209-3200
Editorial Office
100 Motor Pkwy, Suite 110, Hauppauge, NY 11788
(631) 591-4000
Office of Public Affairs
529 14th St NW, Suite 1050, Washington, D.C. 20045-2001
(202) 662-8700