REGULAR PAPER

T. Gono • T. Syuto • T. Yamagata •

N. Fujisawa

Time-resolved scanning stereo PIV measurementof three-dimensional velocity field of highly buoyantjet

Received: 10 August 2011 / Accepted: 26 February 2012 / Published online: 26 April 2012� The Visualization Society of Japan 2012

Abstract The time-resolved scanning stereo PIV system, which consists of a CW Nd:YAG laser, two high-speed CMOS cameras, a Galvano mirror and a pulse controller, is developed for time-sequential three-dimensional velocity field measurement of liquid flow. This experimental technique is validated by themeasurement of a laminar non-buoyant jet, and is applied to the 3D velocity measurement of a highlybuoyant jet at Froude number 0.3 and Reynolds number 200, which satisfy the inflow condition due to theunstable density gradient. The measurement of statistical properties of velocity field shows the growth ofturbulence intensities near the nozzle exit, which indicates the presence of inflow motion near the nozzleexit due to the unstable density gradient. The vorticity contours about the horizontal axis shows the presenceof counter-rotating vortices on both sides of the jet centerline, while the vorticity contour about the verticalaxis indicates the presence of spiral growth of a pair of vortices along the jet centerline. The growth of thespiral structure along the jet centerline is recognized in the iso-vorticity contours obtained from the time-resolved measurement of 3D velocity field.

Keywords Time-resolved measurement � Scanning stereo PIV � Three-dimensional flow � Buoyant jet �Inflow

1 Introduction

The buoyant jet is a fundamental thermal flow phenomenon driven by a buoyant fluid issuing from a nozzle.This flow configuration appears when the fluid issues into different density of fluid. Such flow has been animportant topic in the field of fluid and thermal engineering, because it is closely related to the industrialproblems, such as the thermal flow discharge, the coolant duct failure in nuclear power plant, the fire-safetyproblems in building engineering, and so on.

The preliminary studies on the buoyant jet have been reviewed by Chen and Rodi (1980) and List (1982),which cover the laminar to turbulent transition, the flow development and the decays of center-line velocityand density, the radial distributions of velocity and density, and so on. Later, turbulence measurements inthe buoyant jet are carried out by Papanicolaou and List (1988), Murota et al. (1989) and Shabbir andGeorge (1994). In these studies, the turbulence measurements are carried out by pointwise measurementtechniques, such as a hot-wire/film or a laser Doppler velocimetry combined with laser-induced fluorescence(LIF). These measurements allow the understanding of the momentum and heat transfer mechanism in the

T. Gono � T. Syuto � T. Yamagata (&) � N. FujisawaVisualization Research Center, Niigata University, 8050 Nino-cho, Ikarashi, Nishi-ku, Niigata 950-2181, JapanE-mail: yamagata@eng.niigata-u.ac.jp

N. FujisawaE-mail: fujisawa@eng.niigata-u.ac.jp

J Vis (2012) 15:231–240DOI 10.1007/s12650-012-0129-y

buoyant jet. Later, the planar measurements of velocity and temperature fields in the buoyant jet are carriedout by using particle image velocimetry (PIV) (Pham et al. 2005) and by combined PIV and LIF (Funataniet al. 2004 and Watanabe et al. 2005). Furthermore, the turbulence structure of the buoyant jet is studiednumerically by using the direct numerical simulation (Prourde et al. 2008). These studies demonstrate theexistence of the expulsion and contraction phases during the evolution of the buoyant jet, which is closelyrelated to the larger spreading rate of the buoyant jet in comparison with that of the non-buoyant jet.

In recent years, an attention has been focused on the structure of a highly buoyant jet at low Froudenumber (Syuto et al. 2010), where the inflow occurs due to the presence of unstable density gradient in thenear field of the buoyant jet (Epstein 1988; Maeda et al. 2012). It is found from the scanning planar PIVmeasurement that the inflow occurs near the corners of the square nozzle exit, which results in the gen-eration of inflow defined by the local downward flow into the nozzle against the bulk flow through thenozzle. The occurrence of the inflow promotes the laminar to turbulent transition of the flow through thenozzle, so that the highly buoyant jet becomes turbulent state even if the Reynolds number is lower than thatexpected in the non-buoyant jet (Anwar 1972; Ungate et al. 1975). Although the scanning planar PIVmeasurement allows the study of flow structure in the buoyant jet, the out-of-plane velocity component,which is the same order of magnitude as the in-plane velocity components, are not measured (Syuto et al.2010). In order to understand the three-dimensional flow structure in more detail, the three velocity com-ponents in 3D volume of interests have to be measured in the highly buoyant jet. The scanning stereo PIVsystem is the one that can measure the full velocity field in the 3D volume of interests (Hori and Sakakibara2004; Fujisawa et al. 2005, 2008; Burgmann et al. 2008). Such measurement allows the study of instan-taneous flow structure of the flow, but has not been extended to the time-resolved measurement.

In the present paper, the time-resolved scanning stereo PIV system is newly developed to visualize the3D velocity field of a highly buoyant liquid jet at low Froude number. The measurement allows theunderstanding of the buoyant-jet structure, which satisfies the inflow condition due to the unstable densitygradient in the near field of the buoyant jet. An attention is focused on the time-sequential visualization ofthe 3D structure in the near field of the highly buoyant jet.

2 Experimental apparatus and procedure

2.1 Experimental apparatus

The measurement of the 3D velocity field of a highly buoyant jet is carried out using an experimentalapparatus as shown in Fig. 1. The experimental apparatus is made of acrylic material for the flow visuali-zation purposes. A large volume of hot water is prepared in a tank with temperature controller. The hot waterflows into the test tank filled with cold water through the square duct having 420 mm length with a side lengthd = 20 mm. Note that the square duct is insulated by the Styrofoam to minimize the heat loss. The test tankhas a horizontal dimension 400 9 400 mm2 and a height 460 mm. The buoyant jet is issued vertically into

Fig. 1 Experimental apparatus

232 T. Gono et al.

the stagnant cold-water environment in the test tank. The temperatures of the hot and cold water are measuredby thermocouples placed upstream of the straight duct and in the test tank, respectively. The bulk velocitythrough the duct is measured by a digital flow meter. In the present study, temperature of the hot water was setto 303 K and that of the cold water was 288 K, so that the temperature difference between them was 15 K.The bulk velocity through the duct was set to 8 mm/s in this experiment. Therefore, the present experiment

was carried out at the source Froude number Fr0 ¼ V0=ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

dgðqc � qhÞ=qh

p� �

¼ 0:3 and the Reynolds number

Re (= V0d/m) = 200, which satisfy the inflow condition due to the unstable density gradient in the near field ofhighly buoyant jet. Here g, gravitational acceleration; V0, bulk velocity at nozzle exit; q, density of fluid; m,kinematic viscosity of hot water; and subscripts c and h denote cold and hot state, respectively.

2.2 Time-resolved scanning stereo PIV system

The 3D measurement of the velocity field in the highly buoyant jet is carried out by using the time-resolvedscanning stereo PIV system, which is newly developed for the present study. The schematic illustration ofthe measurement system is shown in Fig. 2. It consists of a CW Nd:YAG laser of 8 W power, a Galvanomirror and two high-speed CMOS cameras located in stereoscopic configuration with off-axis angle of 33�on both sides. It should be mentioned that the laser, the Galvano mirror and the cameras are synchronized bya pulse controller. A volume illumination was provided by scanning the light sheet with a Galvano mirrordriven by a pulse motor. The mirror was located at 600 mm from the target volume of interest to provide thelight sheet almost parallel to the x direction (Fig. 1). Note that the thickness of the light sheet was about2 mm and the maximum deflection angle of the light sheet in the target volume was ±2�. The target volumeof interest was 40 mm 9 60 mm 9 40 mm in x, y, z directions, respectively, where x and z are horizontalcoordinates and y is a vertical coordinate, as illustrated in Fig. 2. The synchronous observation of the targetvolume was made by two high-speed CMOS cameras having a spatial resolution of 1,016 9 1,016 pixelswith 10 bits in grey level.

In the present experiment, the frame rate of the CMOS camera was set to 1,000 frames/s, which is mainlydue to the limitation of the laser power. The camera lens used in this study has a focal length 60 mm withf-number 4, which results in 3 mm in the depth of field. In order to carry out the 3D velocity measurement,the scanning period in z direction was set to 68 ms, which corresponds to the displacement of 0.6 mm withthe mean velocity to travel during the scanning period.

Figure 3 describes the timing chart of the scanning light sheet in the target volume. In order to capturesequential two images for the PIV analysis in the same z plane, a time interval Dt between the two imageswas set to 8 ms, which corresponds to the maximum pixel displacement of 4 pixels in the image. In thepresent experiment, the scanning distance was set to 40 mm ranging from z = -20 mm to 20 mm, whichcorresponds to 2 times of nozzle exit dimension d = 20 mm. In the first scanning motion, 7 sequentialimages were captured at every 1 ms by scanning the light sheet from z = -20 mm to -12 mm. Then, the

Fig. 2 Top view of measurement system

Time-resolved scanning stereo PIV measurement 233

light sheet position returns quickly to the original position in the next 1 ms and another 14 images weresequentially captured at every 1 ms by the scanning the light sheet from z = -20 mm to -3 mm in thesecond scanning motion. The scanning motion allows the measurement of 7 pairs of PIV images at the samez positions with a time interval of 8 ms. Similarly, 8th–14th pairs of PIV images are obtained from secondand third scanning motion of the light sheet with the same time interval. By repeating such scanning motionup to the other side of the target volume z = 20 mm, 31 pairs of PIV images with a time interval of 8 ms areobtained during the scanning motion ranging from z = -20 to 20 mm. Note that the scanning period is68 ms to travel from z = -20 to 20 mm, which is followed by the reverse traverse for 5 ms, and then thenext scanning motion starts at 73 ms.

Three velocity components in a planar cross-section of the light sheet were measured by stereoscopicarrangement of two CMOS cameras tilted with respect to the lens axis, which satisfies the Scheimpflugcondition (Raffel et al. 1998). The captured images contain a non-uniform magnification problem at the air–liquid interface, because the cameras are placed inclining to the surface of the test tank. In order toovercome the difficulty, 3D calibration technique (Soloff et al. 1997; Funatani and Fujisawa 2002) wasapplied to the present experiment. This calibration technique does not require the knowledge of the systemgeometry, such as the separation distance between the cameras, the object distance, the off-axis angle of thecameras, the lens distortion and the refraction at the interface, which may be difficult to measure. Therefore,the errors introduced by the geometrical quantities can be minimized in the calibration. Instead of this, itrequires the calibration data at several different z positions.

The 3D calibration technique requires a target plate for the calibration, which contains 15 9 25 whiteline rulings having 5 mm interval and 0.5 mm width arranged on a black background. This target plate wasplaced inside the test section to coincide exactly the same situation as the 3D measurement of the buoyantjet. The target plate was placed by using a traverse mechanism at 11 different positions separated by every5 mm interval in parallel to x axis to cover the whole volume of interest. Using these calibration data, themapping function is determined, which expresses the relationship between the 3D object image and the 2Dcamera image. It should be noted that the mapping function was approximated by the polynomial expressionhaving a cubic dependence in x and y and a quadratic dependence on z, and the coefficients in the expressionare determined by a least square method. This polynomial expression is the same form as used for stereoPIV measurements (Soloff et al. 1997; Funatani and Fujisawa 2002). The interrogation between thesequential two images was carried out using the cross-correlation algorithm with the Gaussian sub-pixelinterpolation technique. In the present study, the interrogation area was set to 41 9 41 pixels with 50 %overlap. Then, the particle displacements in each image were combined to produce the 3D velocity vectorsby solving a system of equations for stereo cameras with a least square method. It should be mentioned thatthe three velocity components on the light sheet plane, which is tilted slightly from the x axis, are convertedinto 3D velocity components in x, y and z directions by interpolating the measured velocity data. However,the influence of tilted angle on the 3D velocity measurement was found to be small in the range of the angleused in this experiment.

2.3 Uncertainty of velocity measurement

Figure 4a, b shows examples of the tracer image captured in the in-focus plane (z = 0 mm) and that in theout-of-focus plane (z = 12 mm), respectively, which are taken from a camera in stereoscopic arrangement.

Fig. 3 Timing chart of scanning light sheet

234 T. Gono et al.

The image size is 150 9 150 pixels, which corresponds to 9 mm 9 9 mm in the actual space. Note that theplane at z = 12 mm is near the edge of the volume of interest. The PIV analysis of the present tracer imagesshowed no erroneous vectors, when they are analyzed by the correlation area 41 9 41 pixels with 50 %overlap. The detail examination of the tracer images shows that the tracers of 2 pixels in diameter are almostuniformly distributed in the focal plane, but those in the out-of-focus plane are blurred in horizontaldirection. Therefore, the particle size is found more than three times larger in the out-of-focus plane.Nevertheless, all the tracer images showed very few erroneous vectors less than 1 % of the total velocityvectors in the PIV analysis irrespective of the plane positions. It is expected that the blurred tracers does notfail the PIV analysis.

In order to confirm the accuracy of stereo PIV measurement, an experiment was carried out to confirmthe displacement accuracy of an artificial particle image drawing on a plate, which was traversed in 3Ddirections x, y and z by using the traverse mechanism. The actual displacement was evaluated by PIVanalysis, and the result was compared with the exact displacement to confirm the accuracy of stereo PIV.Note that the accuracy of the traverse mechanism is 5 lm. It was found that the rms errors of measurementin x, y, and z directions are found as 0.6, 0.6 and 2.7 %, respectively, in the in-focus plane (z = 0 mm), andthey are 0.8, 0.7 and 3.1 %, respectively, in the out-of-focus plane (z = 12 mm). Therefore, the presentresults suggest that the same level of uncertainty of measurement is found in the whole volume of interest.

3 Results and discussions

3.1 Measurement of nozzle exit velocities in non-buoyant jet

Figure 5a shows the scanning stereo PIV measurement of vertical mean velocity contour near the nozzle exit(y/d = 0.1) of the non-buoyant jet at the Reynolds number Re = 200, which is obtained from 100 real-izations taken at every 1 s. The vertical mean velocity contour shows the symmetric velocity profile about

Fig. 4 Tracer images in focal plane and out-of-focal plane (150 9 150 pixels). a z = 0 mm. b z = 12 mm

Fig. 5 Vertical velocity contour of non-buoyant jet near nozzle exit (y/d = 0.1). a Measurement. b Analysis

Time-resolved scanning stereo PIV measurement 235

the x and z axes through the nozzle center. The result is compared with the analytical laminar velocity profilethrough a square pipe (Knudsen and Katz 1979), which is shown in Fig. 5b. Both velocity contours are inclose agreement with each other, indicating that the flow through the square pipe is laminar in the non-buoyant case. Note that the standard deviation of the measurement with respect to the analytical result is5.0 % in the vertical velocity. On the other hand, the measurements of the horizontal velocity contoursindicated that the standard deviation of the horizontal velocity magnitude in the x–z plane was only a fewpercent of the maximum vertical velocity, which is the same level of magnitude as the rms error of stereoPIV measurement. Thus, the horizontal velocity magnitude near the nozzle exit is considered to be negli-gibly small in agreement with the analytical result. These results suggest the validity of the present scanningstereo PIV measurement.

3.2 Characteristics of mean and fluctuating velocities of highly buoyant jet

Figure 6a–c show the instantaneous velocity contours of the highly buoyant jet issuing from a square nozzleat Fr = 0.3 and Re = 200, which are measured by the present stereo PIV. The color scale in Figs. 6a, bshows the magnitude of horizontal and vertical components of in-plane velocity, respectively, while thecolor bar in Fig. 6c indicates that of out-of-plane velocity. The length of the velocity vector in Fig. 6 isproportional to the velocity magnitude in the 3D domain. The instantaneous velocity contours in Figs. 6a–cshows the presence of periodical structure of the highly buoyant jet, which indicates the zigzag growth ofbuoyant jet in vertical direction. It is seen that the buoyant jet grows vertically by entraining the surroundingfluid into the jet and expanding the jet outward into the surrounding fluid. These fluid motions suggest thepresence of contraction and expulsion phases during the evolution of the buoyant jet, which are defined bythe inward and outward motion of the buoyant jet in the plane normal to the jet axis (Pham et al. 2005).However, the contraction and expulsion phases in the horizontal plane are not observed here, suggesting thedeviation of the flow structure in the near field of the highly buoyant jet from that in the far field of buoyantjet. The detailed observation of the instantaneous velocity (b) indicates the presence of downward velocitynear the nozzle, which shows the occurrence of inflow due to the unstable density gradient near the nozzleexit. Note that the inflow is defined by the local downward flow into the nozzle against the bulk flow throughthe nozzle (Syuto et al. 2010).

The statistical properties of the highly buoyant jet are obtained from 1,000 realizations of instantaneousvelocity vectors taken at every 1 s. Figure 7a, b, c shows the mean velocity contours of U, V and W,respectively, for the highly buoyant jet in the vertical plane through the jet centerline. The horizontalvelocity U normal to the jet centerline (a) shows the entrainment magnitude of the surrounding fluid into thejet. The result indicates that the large magnitude of the inward velocity is generated near the nozzle exit, andit decreases with an increase in the distance from the nozzle exit. The vertical velocity contour (b) showsthat the maximum vertical velocity increases in a short distance up to y/d = 2 from the nozzle exit and itdecreases gradually with an increase in the distance from the nozzle in the far field. The width of the highlybuoyant jet normal to the jet centerline increases with the distance from the nozzle exit, which suggests the

Fig. 6 Instantaneous velocity contours of highly buoyant jet. a u, b v, c w

236 T. Gono et al.

spreading of the momentum normal to the jet axis. The out-of-plane velocity contour shows almost zerovelocity in the average, representing the small magnitude of the mean axial rotational velocity in the highlybuoyant jet (c).

Figure 8a, b, c shows the turbulence intensity contours u0rms, v0rms and w0rms, respectively, in the highlybuoyant jet. It is seen that the turbulence intensity v0rms is higher than the other components u0rms and w0rms

near the nozzle, though they come closer in the far field. Therefore, the turbulence intensity near the nozzleis highly non-isotropic in magnitude, which suggests that the vertical velocity fluctuation is generated in thenozzle and the turbulence energy is transferred to the other components of velocity fluctuations. It should bementioned that the turbulence intensity contour v0rms agrees with that of previous measurement by the planarPIV at Fr = 0.58 and Re = 200 (Syuto et al. 2010), where the inflow is observed in the near field by theflow visualization. These results indicate that the enhanced turbulent mixing near the nozzle exit is causedby the occurrence of inflow due to the unstable density gradient in the near field of the highly buoyant jet.

3.3 Iso-velocity contours

Figure 9a, b shows the 3D reconstruction of instantaneous iso-velocity contours of vertical (a) and radialvelocities (b), respectively, of the highly buoyant jet at Froude number Fr = 0.3 and Reynolds numberRe = 200. Note that the radial velocity is defined by vr ¼

ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

u2 þ w2p

: The results are shown for the iso-velocity contours of upward velocity (v/V0 = 3.0) and downward velocity (v/V0 = -0.2). These velocity

Fig. 7 Mean velocity contours of highly buoyant jet. a U, b V, c W

Fig. 8 Turbulence intensity contours of highly buoyant jet. a u0rms, b v0rms and c w0rms,

Time-resolved scanning stereo PIV measurement 237

contours indicate that the vertical velocity field of the buoyant jet (a) varies with the distance from thenozzle exit due to the presence of large-scale structure accompanied by the inflow. It is seen that the upwardvelocity contour grows in volume with an increase in the distance from the nozzle exit, and it diminishesgradually in the far field (y/d [ 2). This result indicates that the velocity field of the buoyant jet isaccelerated by the buoyancy along the jet axis in the near field, which is followed by a deceleration in the farfield. It should be mentioned that the downward velocity prevails not only near the nozzle exit but also in theregion apart from the nozzle exit, which is distributed around the high-velocity contour. This result alsoindicates that the flow structure in the near field of the highly buoyant jet is greatly influenced by thedownward velocity. The 3D reconstruction of the radial velocity contour (b) shows the inward (vr/V0 =-0.5) and the outward velocity (vr/V0 = 0.5), which are wrapping around the jet axis in spiral manner. Theradial velocity contour indicates the presence of a large-scale structure in the buoyant jet, which is generatednear the nozzle exit and grows along the jet centerline. The inward velocity is induced by the entrainment ofthe surrounding fluid into the buoyant jet. It should be mentioned that the inward and outward velocities aresimultaneously observed in all horizontal planes, so that the contraction and expulsion phases (Pham et al.2005) are not clearly observed in the same horizontal plane.

3.4 Vorticity contour

Figure 10a, b, c shows the 3D reconstruction of the iso-vorticity contours n = ± 5, g = ± 5, f = ± 5 ofhighly buoyant jet, respectively, where n, g, f are the x, y, z components of the vorticity vectors. Each

Fig. 9 Iso-velocity contours of instantaneous velocity fields. a Vertical velocity contour. b Radial velocity contour

Fig. 10 Three-dimensional vorticity contours of highly buoyant jet. a n, b g, c f

238 T. Gono et al.

component of vorticity is non-dimensionalized by the nozzle exit dimension d and the bulk velocity V0 at thenozzle exit. The iso-vorticity contours of n and f show positive and negative values on the vertical planethrough the jet axis, while the iso-vorticity contour of g indicates the spiral growth of the positive andnegative contours along the jet centerline. These features of the growth of iso-vorticity contours representthe structure of turbulence near the nozzle exit. The iso-vorticity contours of n and f show the presence ofcounter-rotating vortices about the jet centerline. On the other hand, the spiral structure of the iso-vorticitycontour of g shows that the counter-rotating vortices grow along the jet centerline wrapping with each otherin a spiral manner. It is seen that the spiral structure is initiated from the nozzle exit and grows along the jetcenterline. These results suggest that the 3D spiral structure is generated from the nozzle exit, where thedownward fluid motion and the generation of disturbance are expected due to the unstable density gradientof highly buoyant jet.

3.5 Time-variation of vorticity contour

Figure 11a–h shows the time-variation of iso-vorticity contour of g, which are measured by the present time-resolved scanning stereo PIV. It should be mentioned that the time-resolved scanning PIV measurement iscarried out at every 73-ms interval and the results are shown at every 146-ms interval. All the vorticitycontours show that the positive and negative vorticities are wrapping spirally along the jet centerline, as wasobserved in Fig. 10. It can be seen that the 3D structure is initiated from the nozzle exit and develops intospiral vorticity along the jet centerline with an increase in time.

4 Conclusions

The time-resolved scanning stereo PIV system is developed to measure the time-variation of three com-ponents of velocity in the 3D flow field. This system consists of the CW Nd:YAG laser, the two high-speed

Fig. 11 Time-resolved measurement of iso-vorticity contour about y-axis. a t = 0 s, b t = 0.15 s, c t = 0.3 s, d t = 0.45 s,e t = 0.6 s, f t = 0.75 s, g t = 0.9 s, h t = 1.05 s

Time-resolved scanning stereo PIV measurement 239

CMOS cameras, the Galvano mirror and the pulse controller. The validation of this system was carried outby measuring the 3D velocity field of the non-buoyant jet at low Reynolds number. This new experimentalsystem is applied to the measurement of three-velocity field in the near field of highly buoyant jet, whichsatisfies the inflow condition due to the unstable density gradient. The measurement of mean and fluctuatingvelocity field reveals that the turbulence intensities are increased near the nozzle exit due to the presence ofinflow by the unstable density gradient. The inward velocity is generated by the entrainment motion of thesurrounding fluid into the jet. The examination of the vorticity contours about the horizontal axis indicatesthe presence of counter-rotating vortices on both sides of the jet centerline, while the vorticity about the jetaxis shows the growth of spiral vortices along the jet centerline of the highly buoyant jet. The growth of thespiral vorticity contour along the jet centerline was observed by the time-resolved observation of the 3Dstructure.

References

Anwar HO (1972) Appearance of unstable buoyant jet. J Hydraul Div 7:1143–1156Burgmann S, Dannemann J, Schroeder W (2008) Time-resolved and volumetric PIV measurements of a transitional separation

bubble on an SD7003 airfoil. Exp Fluids 44:609–622Chen CJ, Rodi W (1980) Vertical turbulent buoyant jets: a review of experimental data. Pergamon Press, OxfordEpstein M (1988) Buoyancy-driven exchange flow through small openings in horizontal partitions. J Heat Transf 110:885–893Fujisawa N, Funatani S, Katoh N (2005) Scanning liquid-crystal thermometry and stereo velocimetry for simultaneous three-

dimensional measurement of temperature and velocity field in a turbulent Rayleigh–Bernard convection. Exp Fluids38:291–303

Fujisawa N, Funatani S, Watanabe Y (2008) Simultaneous imaging techniques for temperature and velocity fields in thermalfluid flows. J Vis 11:247–255

Funatani S, Fujisawa N (2002) Simultaneous measurement of temperature and three velocity components in planar crosssection by liquid-crystal thermometry combined with stereoscopic particle image velocimetry. Meas Sci Technol13:1197–1205

Funatani S, Fujisawa N, Ikeda H (2004) Simultaneous measurement of temperature and velocity using two-colour LIFcombined with PIV with a colour CCD camera and its application to the turbulent buoyant plume. Meas Sci Technol15:983–990

Hori T, Sakakibara J (2004) High-speed scanning stereoscopic PIV for 3D vorticity measurement in liquids. Meas Sci Technol15:1067–1078

Knudsen JD, Katz DL (1979) Fluid dynamics and heat transfer. R.E. Krieger, Huntington, NY 101List EJ (1982) Turbulent jets and plumes. Ann Rev Fluid Mech 14:189–212Maeda T, Fujisawa N, Shutoh T, Yamagata T (2012) Experimental and numerical study on onset of inflow in near field of

buoyant jet at low Froude number. J Vis 15 (in press)Murota A, Nakatsuji K, Tamai M (1989) Experimental study on turbulence structure in turbulent plane forced plume. J Jpn Soc

Civ Eng 405:79–87Papanicolaou PN, List EJ (1988) Investigations of round vertical turbulent buoyant jets. J Fluid Mech 195:341–391Pham MV, Plourde F, Kim SD (2005) Three-dimensional characterization of a pure thermal plume. ASME J Heat Transf

127:624–636Prourde F, Pham MV, Kim SD, Balachandar S (2008) Direct numerical simulations of a rapidly expanding thermal plume:

structure and entrainment interaction. J Fluid Mech 604:99–123Raffel M, Willert C, Kompenhans J (1998) Particle image velocimetry. Springer, Heidelberg, p 174Shabbir A, George WK (1994) Experiments on a round turbulent buoyant plume. J Fluid Mech 275:1–32Soloff SM, Adrian RJ, Liu ZC (1997) Distortion compensation for generalized stereoscopic particle image velocimetry. Meas

Sci Technol 8:1441–1454Syuto T, Fujisawa N, Takasugi T, Yamagata T (2010) Three-dimensional flow visualization and velocity measurement in near

field of strongly buoyant jet. J Vis 13:203–211Ungate CD, Harleman DRF, Jirka GB (1975) Stability and mixing of submerged turbulent jets at low Reynolds numbers.

Energy Laboratory Report, MIT-EL 75-014Watanabe Y, Hashizume Y, Fujisawa N (2005) Simultaneous flow visualization and PIV measurement of turbulent buoyant

plume. J Vis 8:293–294

240 T. Gono et al.