Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
24~ Symposium on Naval Hydrodynamics Fukuoka, JAPAN, 8-13 July 2002 Quantitative Visualization (QViz) Hydrodynamic Measurement Technique of Multiphase Unsteady Surfaces Deborah Furey and Dr. Thomas C. Fu NSWCCD 9500 MacArthur Boulevard West Bethesda, Maryland 20817 Abstract A non-intrusive optical technique, Quantitative Visualization (QViz), has been developed to measure the free surface deformations occurring in the bow and near wake regions of surface ships. These regions are generally difficult to quantify due to the multiphase aspect of the flow as well as their very unsteady nature. However, the unsteady surfaces, droplets and bubbles in these regions are effective scatterers and allow for optical imaging of the extreme deformations in the surface. With a laser sheet and digital camera, the QViz system illuminates the surface of interest and collects digital images representing cross sections in time of the spray envelope occurring in the bow region and surface profiles in the steep wake regions. Variations in illumination and image collection were investigated for several free surface and multiphase flow conditions in order to characterize this technique. Data analysis programs have been developed to extract the surface boundary data, both time averaged profiles and unsteady quantities. The results show that the technique is suitable for multiphase conditions where the boundary is not rapidly changing over large distances relative to the field of view. Comparisons were done between conventional fingerprobe measurements and the QViz technique. Introduction Quantifying surface profiles in the near wake and bow regions of surface ships has not been easily achievable using established methods. Other imaging techniques have been used to evaluate flow characteristics in regions of limited accessibility and complex free surface flow regimes [2,3,4,51. The flow in these regions is unsteady and has discontinuous surfaces making standard methods such as sonic probes, finger probes or dye traces unusable. Quantitative visualization, QViz, provides a method to visualize and quantify these regions of high unsteadiness and non-uniformity. The fluid in the near wake and bow regions of surface vessels is commonly two phase, characterized by both droplets and bubbles generated from spray sheets forming off the bow and splashing caused by impacting fluid. Though this makes using established methods difficult, it provides ideal surfaces for optical scattering and quantitative imaging. Using a light sheet and digital camera, the QViz system illuminates the surface of interest with a specific input laser power and records an instantaneous digital image. This paper will discuss the examination of several technique variables and how they effected the resulting images. The results being presented include comparisons with finger probe measurements, parametric study of QViz variables and data collected from a surface ship model tests. For measurement verification, comparisons were done between the surface displacements measured using QViz and a conventional fingerprobe method. These results were analyzed for amplitude comparisons as well as frequency content. In addition, in an effort to characterize the QViz method, a parametric investigation into the effect of system settings and surface roughness on the resulting images was conducted. Variations in illumination were achieved through changing incoming laser intensity and changing the viewing direction for a given light sheet geometry. Additional variations include changing CCD exposure times and aperture settings. An initial set of QViz data was collected using a surface ship model at NSWCCD to investigate the feasibility of this method. The images in the bow region represent an instantaneous cross section in the spray envelop at a specific axial location and can be used to extrapolate the entire average spray envelop. The images in the near field stern region provide instantaneous wake profiles and can be used to evaluate a time averaged wake profile. Quantitative Visualization Method The QViz system consists of a continuous wave laser and optics to create a steerable light sheet. The light sheet and collection optics are mounted at specific orientations for the flow and conditions being investigated. For the laboratory set up as well as for the model tests on Carriage 5 at NSWCCD, the light sheet is directed into the region of interest from above the water line and oriented perpendicular to the flat free surface. A digital camera is oriented at a known angle relative to the light sheet and free surface being imaged. For the parametric investigation, the camera and incident optics were oriented at angles ranging between 0 and 90 degrees relative to the free surface and to each other for the different surface features being imaged. Figure 1 illustrates the set up in the miniature water basin tests which is similar to the set up on the Carriage 5 test.
Set up in the Miniature Water Basin Mirror Lens ' Laser Laser Sheet ~ Camern Figure 1 12 dewing Angle For these tests, a Sony DV Cam was used which collect images at 480 x 640 pixel resolution. The laser was a Spectra Physics, 5 W Argon laser. The resulting digital images represent that portion of the flow within the laser sheet and provide an instantaneous cross section of the fluid. The images are processed to evaluate the specific quantity of interest; surface displacement, cross-sectional area of the spray in the bow region or the displacements In one tree surface occurring In the wake region. Processing methods investigated both instantaneous image analysis as well as averaging techniques to determine the free surface and spray envelop boundaries. . . Initial developments of the QViz analysis averaged several images together, effectively providing a time- averaged cross section. From this image the averaged cross-sectional area and surface profile are determined. Below (figure 2) is a sample image of the bow wave spray region from the model tests. The ship model is on the right side of the image. The illuminated region represents the bow spray envelope at a particular axial location. Figure 2: Spray region illuminated by a laser sheet in the bow wave of a planing ship model. Station 7, x/1=0.7. The images were saved in a raw binary format and processed using Fortran codes to extract the boundary information. Image filtering and thresholding techniques 24th Symposium on Naval Hydrodynamics Fukuoka, JAPAN, 8-13 July 2002 were used to evaluate the average profiles while gradient methods were required to evaluate the unsteady boundaries. The number of images required to establish a steady average varied for the different test conditions due to the nature of the flow generated. The quantities were converted to spatial coordinates using calibration conversion routines. Technique Development: Calibrations: The purpose of developing QViz is to spatially quantify the surface profile in the bow and stern regions of surface ship models. This requires calibration of the CCD camera for each laser position and camera setting. The cameras were calibrated using a calibration target marked with a 2 inch by 2 inch grid, which was imaged for each plane of investigation and camera orientation. This calibration grid was used to map the CCD pixel information to xyz information. The grid was stabilized in position and referenced to the free surface using a plumb bob. The pixel information is transferred to the grid coordinates and then transferred to the model coordinates using the grid positional information. Average Images: The images collected using the laser sheet of the free surface profiles and bow spray envelopes show illuminated regions which are to be quantified spatially using the CCD calibrations mentioned above. However, the profiles in the regions of interest, the bow and the stern, are very unsteady. With this consideration, it was decided to do preliminary image processing to extract average profiles from the images. The digital video was downloaded using an AVID digital video editing system to a series of binary black and white images. The binary files were read into a Fortran program which averages together a specified number of images. Figure 2 shows an image from the model tests conducted on Carriage 5. The imaging plane is at model station 7, x/1=0.7 and the test speed was 30 kts.. A minimum of 20-30 images is required for this condition to achieve a steady average for a set threshold pixel limiting value (refer to Figure 51. From the average image, the pixels which corresponded to the boundary of the illuminated regions are identified and mapped back to spatial coordinates using the calibration data. It was found that for the different running conditions as well as camera positions and field of view settings, a different number of images were required to get a steady average. (Discussed in the next section). Analysis Methods: Thresholding Two analysis methods were used to extract the data from the video images. The main purpose was to identify the edge of the illuminated region. Looking at the average images, the first approach was to use a certain threshold pixel value, which corresponds, to the edge of the bright 2
illuminated region. This required manual evaluation to identify the best pixel value and would result in a clearly identified region of interest (ROI). The analysis program would then search down each column to identify the pixel corresponding to the edge of the ROI . These pixel values are then mapped back to the XYZ coordinates using the calibration grid information acquired at the beginning of each run. This method gave good results for cases with high frequency unsteadiness Figure 3: Average image at Station 7, all= 0.7 for model tests on Carriage 5 at DTMB. Figure 4. Thresholded image for Station 7, x/1= 0.7. Threshold pixel value of 240. 24~ Symposium on Naval Hydrodynamics Fukuoka, JAPAN, 8-13 July 2002 of low amplitude, and the average was established in few frames. For cases where the flow was largely unsteady and rough, 60 images were required to establish an average cross sectional image. Figure 3 shows the average image established for the model tests at Station 7, x/1=0.7, 30 knots. This method was considered to be speculative and too dependent on manual input. Using thresholding provides clean edges, however, manual input may cause inaccurate edge definitions. See figure 4 70 - ~ 65 - ant _ 60- fit, < 55- hi _ 50 all In, 45- ii; 40- in 35-) 25 o 5 10 15 20 Number of Images 25 30 Fig 5: Area as function of number of images averaged together. For the30 kt case, station 7, x/1=0.7, it requires approximately 20 images at least. The threshold value was investigated to determine the correct value for determining the correct pixel value for establishing the boundary for the region of interest. The effect of pixel value on area was one indicator. Figure 6 illustrates the change in area as a function of threshold value. The desired threshold value will be where there is no change in area for a change in threshold value. Figures 6a shows that there is a region near a pixel value of 230 and 250 where the area is least sensitive to threshold value. Figure 6b illustrates the derivative directly. At a pixel value of 240 the derivative is a minimum, approaches zero, and correlates to the pixel value corresponding to the boundary for that image. 3
200 ~ arc 150 ~ n 311 0 50 100 150 200 250 300 Threshold Figure 6a: Area vs threshold for 2 cameras imaging station 7, x/l= 0.7. Gradient Method As an alternative to the thresholding method, a gradient technique was developed to utilize the local intensity information in the image. It had been found that there was a significant intensity variation across the image for some test conditions making edge detection more difficult. Therefore, a localized search method was developed to search columnwise for the boundaries in each x location in the image. This approach eliminated the intensity variation problem and provided a suitable approach for a detailed unsteady analysis. The gradient search method interrogates the image columnwise for changes in pixel intensity. The boundary is considered to be located where the gradient is the largest. The average image was calculated using 30, 60, or 120 images depending on the unsteadiness of the flow field. The average boundary was then identified using the gradient search technique and defined the ROI for the cross section. This was considered the average boundary for that cross section. For extracting unsteady information, a comparison was made between the average boundary and the instantaneous boundaries determined from each image. The set of images used to establish the average profile are interrogated individually. Each image is filtered to eliminate noise and then analyzed using the gradient method in the established region of interest. The data was stored as a function of column location, or equivalently x location relative to the model. The RMS was then calculated as a function of location. (Refer to Figure 17) This analysis revealed that the RMS is relatively small and the boundary is considered to be well established using the average image. Real Time Image Analysis The QViz analysis initially developed analyzed images downloaded from video tape to discrete files and were then post processed to compute the desired information. This procedure requires significant time to allow for the images to be saved, digitized, and processed 0 -0.524 ~ -1 5 ·o -2 5 , -3.5 - -4 -4.5 ;) Threshold Figure 6b: Change in area for change in threshold. individually. This original analysis indicated that the averaging and gradient techniques were reasonable and were implemented into a real time system. With the results using the gradient edge detection, a real time analysis technique was developed using National Instruments Labview software. The image analysis software was designed to average several images together and evaluate the surface profile or to interrogate a small region, image by image, to track a point in the flow for varying conditions. The development of the real-time system greatly increases the practicality of using QViz for looking at unsteady flows as well as greatly reducing the data reduction time. A non-real- time system can still be used, but a substantial effort is still necessary to process the large number of images required to examine unsteady flows. The real time system was developed using Labview software and image processing toolbox. The reduction software was written for 2 types of analysis. One analyzes the entire image, and requires averaging a minimum of 10 images together. The boundary is evaluated using the gradient search method discussed previously. An alternative program interrogates at a fixed location in the image, this allows for investigating time varying flows with a moving free surface boundary. Finger Probe-QViz Comparison using Real Time analysis An experimental, in-situ comparison of a finger probe and QViz system was performed to provide a comparison between an established measurement technique and the light sheet method. A mechanical, servo-motor-controlled finger probe has been used to quantify the free surface wave heights in the far field flow regions around towed models. Since the probe operation is based on the periodic sensing of the water surface at a fixed location, as the free surface becomes more turbulent, as in the near field bow and stern regions, the variance of the "captured" wave height increases. For QViz technique validation a comparison was made between data gathered using the conventional fingerprobe interrogation technique and the QViz method for various flow conditions. The results were 4
mixed but did verify that the QViz method could be used to extract mean surface data in slowly varying flows or flows with a small dynamic range. For the finger probe/QViz comparison, both systems were set up in the Miniature water basin at NSWCCD. The basin is 40 feet long and the water depth was set at 19 inches. The basin is equipped with a manually controlled paddle type wave maker. Four different wave settings were used. For changing the surface roughness, a breakwater 18 inches high was placed upstream of the test section. This caused the waves to break and the surface profiles to be irregular. The finger probe and QViz systems were set up for tracking the free surface for various waves. The QViz software was set up to interrogate a narrow region in the image, or a single x location in the surface, comparable to the finger probe comparison of a fixed location. This region of interest (ROI) is defined in the Labview software. The fingerprobe interrogated the free surface at a rate of 10 Hz. The QViz data is collected at 30 and does not incorporate any interpolation or smoothing. The variables investigated include viewing angle, laser power intensity, wave frequency and breaking/non- breaking conditions. For the local QViz measurement, only 8 columns of data are tracked. For these tests, the ROI was positioned in column 316 to column 324, and included rows 160 to 2 1.5 on 1 C: 0-5 ~ ,= O .. -0.5 Q i. -1 = -1.5 -2 ~~ (INS) Figure 7a: QViz data for 11 degree viewing angle and laser power set at 2.1 watts. Wave peak to peak amplitude is 0.15 inches. Wave setting 1. row 460 of the digitized images. This ROI was chosen to be in the brightly lit region of the laser sheet, and was 4 inches upstream from the finger probe. To ensure a true representation of the surface was acquired, the median of the values extracted was used to represent the water surface. This would eliminate any erroneous extreme values caused by data drop outs or extraneous bright pixels resulting from erroneous reflections. This proved to be a good technique to reduce the erratic profiles for cases where the surface was primarily smooth. However, when there were splashes, i.e. breaking conditions, there were many spurious data points. Also, for these measurements, the data was taken frame by frame, no frame averaging was done. Figures 7a to 7d show the non breaking wave profiles collected using QViz and figures 8a to 8d show the profiles collected using the finger probe. The different wave settings are referred to as wave setting 1,2,3 and 4 respectively. It is expected the data will have some inconsistencies because the wave maker is set manually and data was not collected simultaneously for all data runs for the finger probe and QViz. The results show the amplitudes and periods are comparable. This suggests the calibration and transfer functions are reliable. Figures 9a to 9d and 10a to 10d show the frequency content of the data collected by the QViz and fingerprobe. The frequency analysis was done using a MATLAB tE l algorithm. Again, the frequencies are comparable. 2- 1.5 - °0 1 . 0.5- i _ As O . -0.5 - l Q E -1 - -1.5 - z -2 - Time (second) Figure 7b: QViz data for 11 degree viewing angle and laser power set at 2.1 watts. Wave peak to peak amplitude is 0.6 inches. Wave setting 2. 5
10 15 20 25 nme(seconds) Figure 7c: QViz data for 11 degree viewing angle and laser power set at 2.1 watts. Wave peak to peak amplitude is 2 inches. Wave setting 3. 2.0 1.5 10 0.5 -1.0 - a. -1.5- D -2.0 - ~ 20 5 10 15 20 rime (seconds) Figure 7d: QViz data for 11 degree viewing angle and laser power set at 2.1 watts. Wave peak to peak amplitude is 3 inches. Wave setting 4. 25 30 35 nme (seconds) Figure 8a: Fingerprobe data for wave peak to peak amplitude of 0.14 inches. Wave setting 1. 20 1.5 1.0 ', Q5 ~ DO .' ~.5 E -1.0 -1.5 -2.0 r~ne(se=~s) Figure 8b: Fingerprobe data for wave peak to peak amplitude of 0.64 inches. Wave setting 2. 0 5 10 15 Time (seconds) Figure 8c: Fingerprobe data for wave peak to peak amplitude of 1.8 inches. Wave setting 3. 2.0 - ~ 1.5 - . al 1.0 c 0.5- _ ~ ,= 0.0- . Q -0.5 - -1 .0 - -1.5 - -2.O 5 10 Time (seconds) . . 15 20 Figure 8d: Fingerprobe data for wave peak to peak amplitude of 2.8 inches. Wave setting 4. 6
figure 9a: QViz frequency analysis for wave 1. figure 9c: QViz frequency analysis for wave 3 figure 9b: QViz frequency analysis for wave 2. figure 9d: QViz frequency analysis for wave 4 figure lea: finger probe wave setting 1 figure lob: finger probe wave setting 2 7
figure lOc: finger probe wave setting 3 From the frequency analysis, the timing is consistent between the two methods. However, combining the amplitude information from the scaled profiles and frequency information, it would be expected that the power in the respective frequency components should be closer. This discrepancy is not yet understood. Intensity of the incoming laser sheet was also investigated for the different wave settings, viewing 5 7 9 11 13 15 17 19 Rome (seconds) Figure lla: QViz data for a 56 degree viewing angle, laser power at 0.2 watts. Wave setting 2. ~ '~ 1 5 -2 5 Figure lib: QViz data for a 36 degree viewing angle, laser power at 1.1 watts. Wave setting 2. figure led: finger probe wave setting 4. angles and breaking conditions. Below, figures lla to 1 1 d show how the wave profiles are effected by incoming laser intensity. It is evident that for low laser powers there are more erroneous measurements due to low contrast in the image. As the power increases, the wave profiles have less scatter. At the highest laser setting, the profile is smooth throughout the sample time. 1.25 1 c 0.5 O -0.5 -1 -1 .5 -2 15 17 19 21 23 25 Time (seconds) Figure 1 lo: QViz data for a 36 degree viewing angle, laser power at 0.6 watts. Wave setting 2. 1.5 Tic 1 I 0~5 O _ -0.5 -1 -1 .5 -2 15 20 25 Time (seconds) Figure 1 id: QViz data for a 36 degree viewing angle, laser power at 2.1 watts. Wave setting 2. 8
A breakwater was introduced to cause a roughening of the surface. Figures 12a to 12 d show the QViz profiles for the breaking and non breaking conditions. For low amplitude low frequency conditions, QViz technique was able to track the surface profile reliably for both the breaking (Figure 12b) and non breaking (Figure 12a) conditions with some data drop outs and erroneous A 0 5 10 15 Time (seconds) Figure 12a: QViz data at an 11 degree viewing angle, 2 watts with no break water for wave setting 2. 1.5 y 0.5 Tic OS 10 14 16 18 20 Time (seconds) 22 24 Figure 12b: QViz data at an 11 degree viewing angle, 1.1 watts with break water for wave setting 2. The variation in viewing angle did not produce any unforeseen results. The scaling changed for each case causing a reduction in resolution for higher viewing angles. This can be improved by using a smaller field of view for higher angles giving improved resolution. From the fingerprobe comparison to QViz, it is evident that the QViz method is capable of tracking the surface. For surfaces with a fairly steady boundary, the optical technique can measure the boundary reliably. However, if cases where the boundary has a largely dynamic surface, it is not as reliable. For using QViz in regions inaccessible to the fingerprobe, the flow must have some steady character, or a steady average of the boundary must be established for the QViz technique to be viable. The results of this study will guide the choice of techniques for use in various flow field regimes. Camera Settings The characterization of the QViz system required investigating the effects of illumination parameters as well as image collection parameters. Results showed that largely bubbly and unsteady free surfaces result in points in the breaking case. For the higher amplitude higher frequency waves, figures 12c and 12d, QViz is unable to track the surface for the breaking condition. This was due to the large number of scatterers splashing in the test section which caused the edge detection to resolve on random droplets in the ROI. tar cO OS, ~c'°S,2 ~ 2 ! I 10 15 20 Time (seconds) Figure 12c: QViz data at a 36 degree viewing angle, 2.1 watts with no break water for wave 4. 2 1 . c 0.5 c ~' O .~ -0.5 1 -1.5 -2 5 10 15 Time (seconds) 20 25 Figure 12d: QViz data at a 36 degree viewing angle, 2 watts with breakwater for wave setting 4. greater recorded image intensity than for less disturbed free surfaces for the same incident laser sheet power Also, as expected, the images become less sharp for longer exposure settings. The surface is clearly imaged due to the large number of scatterers as the surface becomes roughened. However, for the longer exposure times it is evident that there are also reflections from internal scatterers as well, obscuring the surface boundary and making it impossible for the QViz analysis to decipher the real surface. For high power settings, it is better to use short exposures when imaging roughened surfaces to capture scatterers within the ROI. Longer exposures with lower power settings are suitable for smooth boundary surfaces with fewer scatterers. Model Testing Results One set of experiments utilized the QViz method in the bow and stern regions of a surface ship model towed on Carriage 5 at the David Taylor Model Basin [1,6,71. These regions of the flow are unsteady and are not accessible by the conventional methods, i.e finger probes. However, using an optical technique, allows for 9
access to these regions. For the initial ship model data, QViz was utilized to evaluate the average profiles of the mean cross-sectional area of the spray envelope in the bow region and a mean surface profiles for the stern region. Sample data sets will be included here. Figures 15a and l5b show the average image and the corresponding average profile extracted from the QViz analysis for station 7, x/l= 0.7. The averaged images show which portions of the spray envelope are the most steady, and give an indication of the spatial variability of the bow spray. Figures 16a and 16b show the resulting average stern profiles for the model test results. Figure 15a: Averaged image for the spray region illuminated by a laser sheet in the bow region of a planning ship model at station 7, xlI'0.7. Average Cross Section Bow Envelope X (inches) Figure 15b: Boundary envelope for average image. Station 7, x/L=0.7. Figure 16a: Station 13, x/1=1.3 of model test at 30 kts. A frame by frame analysis was done and the instantaneous profiles compared to the average profile. This provided a measure of the RMS about the average, refer to Figure 17. The unsteady analysis showed that the variation in the boundary when compared image to image was small enough and that the average image was a sufficient measure of the flow profile. Spoon 13 Y (inC~) Figure 16b: Station 13 QViz profile at 30 kts 15 - 10 - 07 ~ O- _ -5 - -10 - -15 . _ Y (inches) Figure 17: Average profile of stern profile at x/l=13 with RMS error bars. 10
Conclusions Quantitative Visualization provides a viable technique to quantify regions of a surface ship wake, which would otherwise be inaccessible through established means. From the initial data set and investigations into the effects of system components and surface characteristics, surface profiles can be quantified using the QViz method. Analyzing time averaged images using the methods developed provides a realistic measure of the mean boundary defined by the spray in the bow region and wake region of surface ship models. It was found that QViz is not viable for largely dynamic surfaces, where there are a large number of discrete scatterers. However QViz is suitable for slowly varying flows or evaluating boundaries with high frequency low amplitude changes as well as steep surface slopes where conventional techniques are less suited. Further, the use of QViz can potentially provide a viable way to validate computational predictions in flow regimes where conventional data is not available. References: 1. Stahl, R.G., "Ship Model Size Selection, Faciliteis and Notes on Experimental Techniques", CRDKNSWC/HD-1448-01, May 1995. 2. Goldstein and Smits, "Flow Vizualization of the 3-D Time Evolving Structure of a Turbulent Boundary Layer", Phys Fluids, 6, p. 577, 1994. 3. Rockewell, Lin, Cetiner, Downes and Yang, " Quantitative Imaging of the Wake of a Cylinder in a Steady Current and Free Surface Waves", Journal of Fluids and Structures, Vol 15, No. 314, Apr 2001, p. 427-443 Logory, Hirsa and Anthony, 'Interaction of Wake Turbulence with a Free Surface", Phy Fluids, p. 805, 1996. 5. Huany, Kawall, Keffer, and Ferre, "On the Entrainment Process in Plane Turbulent Wakes", , Phy Fluids ,7, 1130, 1995. 6. Zselecszky, J.J., "Resistance and Seakeeping Model Tests of a Hard Chine Planing Hull", U.S. Naval Academy, Report EW-17-96, June 1996. Ratcliffe, T. Mutnick, I, and Rice, J. "Resistance Characteristics of a Planing Hull as Respresented by Model 5572 Towed in Calm Water and In Regular Waves", NSWCCD-50- TR-2001/025. 11
DISCUSSION T. Waniewski Sur Science Applications International Corporation, USA The QViz system described in this paper appears to be a simple yet practical technique for making measurements of unsteady surface locations; however, a description of its accuracy and precision would help others to interpret future measurements. The comparison of QViz and finger probe measurements taken in the miniature model basin presented herein (Figures 7, 8, 9, and 10) is a good start and could be expanded in several ways. For example, the authors state that the QViz and finger probe data was "not collected simultaneously for all data runs," but do not indicate which data was collected simultaneously and which was not. Since the distance between the two instruments is known, it seems possible and useful to adjust the phase of the simultaneous measurements accordingly and then compare them. In addition, data from different runs is presented in the figures, yet the authors do not indicate the repeatability of the flows created by the four "wave settings" of the manually controlled paddle type wave maker. If the repeatability of the flows could be demonstrated, it would be easier to study the repeatability of the QViz measurement technique. This paper also presents QViz measurements in the bow and stern regions of a towed surface ship model. Free surface data collected in these regions will be helpful in validating numerical models. Can QViz resolve a multi-valued free surface; for example, both the upper and lower surfaces of a plunging breaking bow wave jet?
DISCUSSION T. Waniewski Sur Science Applications International Corporation, USA The QViz system described in this paper appears to be a simple yet practical technique for making measurements of unsteady surface locations; however, a description of its accuracy and precision would help others to interpret future measurements. The comparison of QViz and finger probe measurements taken in the miniature model basin presented herein (Figures 7, 8, 9, and 10) is a good start and could be expanded in several ways. For example, the authors state that the QViz and finger probe data was "not collected simultaneously for all data runs," but do not indicate which data was collected simultaneously and which was not. Since the distance between the two instruments is known, it seems possible and useful to adjust the phase of the simultaneous measurements accordingly and then compare them. In addition, data from different runs is presented in the figures, yet the authors do not indicate the repeatability of the flows created by the four "wave settings" of the manually controlled paddle type wave maker. If the repeatability of the flows could be demonstrated, it would be easier to study the repeatability of the QViz measurement technique. This paper also presents QViz measurements in the bow and stern regions of a towed surface ship model. Free surface data collected in these regions will be helpful in validating numerical models. Can QViz resolve a multi-valued free surface; for example, both the upper and lower surfaces of a plunging breaking bow wave jet? AUTHORS' REPLY Addressing the first question phase comparison of the data, concerning the the fingerprobe data was collected on every eighth run for the QViz data. The camera settings and laser settings had to be varied for each wave condition making a much larger run matrix for examining the Qviz parameter effects. Often, the fingerprobe data was collected for the first laser setting which was a very low light condition and not a good data run for the QViz making that type of comparison difficult for this data set but your suggestion could be implemented for future test set ups. The second question was about repeatability. The test was set up in the miniature water basin at NSWC which is instrumented with a manually controlled wave maker. This type of wavemaker made it difficult to have the exact same setting for each run. We had incremental markings on a continuous pot type control making the exact wave setting difficult to match. So the results give wave periods of comparable periods and amplitudes but the conditions could not be matched from run to run for this type of wavemaker. The last question asked if QViz can resolve multi-valued surfaces. Yes, the images collected by the QViz system can resolve both the upper and lower surfaces for many conditions. The ability to resolve multi-valued surfaces depends on the flow conditions, the laser settings and optics.
