14th International Symposium on Particle Image Velocimetry – ISPIV 2021
August 1–5, 2021

A stereoscopic PIV system for the Princeton Superpipe
L. Ding∗, E. Limacher, I. E. Gunady, A. Piqué, M. Hultmark and A. J. Smits

Department of Mechanical and Aerospace Engineering
Princeton University, Princeton, New Jersey, USA
∗Corresponding author: liuyangd@princeton.edu

Abstract
Herein, we describe the design and testing of a stereoscopic PIV system uniquely adapted for the high
pressure environment of the Princeton Superpipe. The Superpipe is a recirculating pipe facility that utilizes
compressed air as the working fluid to attain very high Reynolds numbers. Commercial piping is used as the
pressure vessel to hold pressure up to 220 bars, and a test pipe is enclosed inside with a development length of
200 diameters that ensures a fully-developed condition at the test section. The highest achievable Reynolds
number (based on the bulk velocity and the pipe diameter) is 35×106, corresponding to a maximum friction
Reynolds number of 5×105.

The unprecedented range of Reynolds number has enabled a number of new insights in the behav-
ior of high Reynolds number wall-bounded turbulence (Zagarola and Smits, 1998; Hultmark et al., 2013).
However, past measurements in the Superpipe have been primarily restricted to single-component, one- or
two-point statistics of fully-developed pipe flows. The present work aims to expand the capability of the Su-
perpipe to study turbulent coherent structures and multi-point statistics by means of a new stereoscopic PIV
system. The high pressure environment and the confined space inside the pressure vessel pose challenges to
both imaging and seeding, the solutions to which will be discussed.

1 Imaging
The imaging system has been designed so that only mirrors are contained in the high-pressure environment,
and all cameras are located external to the test section. This configuration offers a significant benefit – its
performance is nearly unaffected when the refractive index of compressed air changes with pressure. More
specifically, the object plane is chosen to be the one determined by the axes of the test pipe and the access
port of the test section (see figure 1). A laser light sheet is delivered into the test pipe through the sight
windows on the sealing flange and the internal test pipe. An optical relay assembly consisting of two near-
axis and two off-axis mirrors is mounted directly to the test pipe and used to relay particle images to cameras
placed outside the pressure vessel through the flange sight window. As indicated by the simulated light rays
in figure 1, the mirrors on each side of the light sheet provide one view of the object plane, so the two pairs
of mirrors constitute a stereoscopic system.

By means of optical ray tracing, the imaging system is optimized in terms of the following aspects: (i)
the angle between the two views in stereoscopic imaging is 90◦ to minimize the uncertainty in the azimuthal
velocity component (Lawson and Wu, 1997); (ii) the field of view encompasses the region between the wall
and the centerline of the pipe and spans at least one radius (R = 63.8 mm) in the pipe axial direction; (iii)
the chief ray emanating from the center of the field of view (y = R/2 with y being the wall-normal distance)
is perpendicular to the sight window on the sealing flange (#3 in figure 1) to minimize astigmatism. With
the ray tracing analysis, we have also evaluated the images of point sources in the object plane. The image
sizes across the field of view are smaller than 10 µm, as compared to the diffraction-limited spot size of 17
µm at the same magnification and f -number.

2 Seeding
The design of the flow seeding system must address the following challenges: (i) it must allow for the
generation and injection of seeding particles at or near the desired test pressures; (ii) it must not foul the
internal surfaces of the Superpipe, which is difficult to clean; and (iii) it must produce a narrow distribution
of particle sizes with a mean diameter around 1µm.

Our chosen design consists of a sealed, pressure-rated housing that serves as a liquid reservoir, with
several Laskin nozzles submerged in the liquid. Compressed air at higher pressure is driven through the
nozzles to generate an aerosol, which is then carried into the Superpipe through a connecting tube. We have



Figure 1: The imaging system shown in a cross-
section of the access port, with simulated light rays
indicating the imaging paths. 1: test pipe with a
thin sight window. 2: mirrors (two near-axis ones
and two off-axis ones). 3: sealing flange with a
sight window. 4: lens group of a 135-mm prime
lens. 5: laser sheet.

0 200 400 600 800 1000 1200

0

1

2

3

4

5

Figure 2: Dp versus ∆P for all test cases run. Filled
circles denote purely seeded cases; empty squares
represent cases with dry air mixed with seeded air.
Yellow-filled red stars are results from Legrand
et al. (2017) with no dry air added.

opted for propylene glycol as the seeding liquid because it is inexpensive with a low vapor pressure – a
desirable property keeping the surfaces of the test section free of liquid residue.

A Dantec PDA system was used to characterize the size distribution of droplets produced by the seeder.
The modal (peak) diameter Dp is plotted in figure 2 versus the pressure difference, ∆P = P1 −P2, where
P1 is the supply pressure, and P2 is the housing pressure. These results are for a nozzle with two 0.51mm-
diameter holes; other configurations were also tested, but are omitted for brevity. In general, increasing the
nozzle pressure difference tends to decrease droplet sizes, and adding dry air for a given configuration and
∆P tends to reduce the droplet size further, presumably due to evaporation. To confidently get a droplet size
distribution with a modal diameter of 2 µm or less, one should maintain a pressure difference across the
nozzle of about 500 psid, if purely seeded air is desired. Mixing dry air is likely desirable to achieve seeding
densities more appropriate to PIV. However, it bears noting that these results and the above suggestion are
sensitive to other system design parameters that are not fully understood. Legrand et al. (2017), for example,
managed to obtain smaller modal diameters than us at low pressure differences (yellow-filled red stars on
figure 2).

3 System Integration
Having designed, assembled and tested various subsystems, we have moved on to begin testing their inte-
gration. We will present preliminary results showing acquired particle images in the test pipe at ambient
pressure, and we will discuss recently addressed and ongoing challenges such as background reflection and
droplet condensation.

Acknowledgements
This work is supported by ONR DURIP Grant No. N00014-19-1-2301 and ONR Grant No. N00014-17-1-
2309. IEG was supported by the Department of Defense (DoD) through the National Defense Science and
Engineering Graduate Fellowship (NDSEG) Program.

References
Hultmark M, Vallikivi M, Bailey SCC, and Smits A (2013) Logarithmic scaling of turbulence in smooth-and

rough-wall pipe flow. Journal of Fluid Mechanics 728:376–395

Lawson N and Wu J (1997) Three-dimensional particle image velocimetry: error analysis of stereoscopic
techniques. Measurement Science and Technology 8:894–900

Legrand M, Nogueira J, Rodriguez PA, Lecuona A, and Jimenez R (2017) Generation and droplet size
distribution of tracer particles for PIV measurements in air, using propylene glycol/water solution. Exper-
imental Thermal and Fluid Science 81:1–8

Zagarola MV and Smits AJ (1998) Mean-flow scaling of turbulent pipe flow. Journal of Fluid Mechanics
373:33–79


	Imaging
	Seeding
	System Integration